/*
* Copyright (c) 1991-1994 by Xerox Corporation. All rights reserved.
* Copyright (c) 1996-1999 by Silicon Graphics. All rights reserved.
* Copyright (c) 2003-2011 Hewlett-Packard Development Company, L.P.
*
*
* THIS MATERIAL IS PROVIDED AS IS, WITH ABSOLUTELY NO WARRANTY EXPRESSED
* OR IMPLIED. ANY USE IS AT YOUR OWN RISK.
*
* Permission is hereby granted to use or copy this program
* for any purpose, provided the above notices are retained on all copies.
* Permission to modify the code and to distribute modified code is granted,
* provided the above notices are retained, and a notice that the code was
* modified is included with the above copyright notice.
*
*/
/* Memory model documented at http://www-106.ibm.com/developerworks/ */
/* eserver/articles/archguide.html and (clearer) */
/* http://www-106.ibm.com/developerworks/eserver/articles/powerpc.html. */
/* There appears to be no implicit ordering between any kind of */
/* independent memory references. */
/* TODO: Implement double-wide operations if available. */
#if (AO_GNUC_PREREQ(4, 8) || AO_CLANG_PREREQ(3, 8)) \
&& !defined(AO_DISABLE_GCC_ATOMICS)
/* Probably, it could be enabled even for earlier gcc/clang versions. */
/* TODO: As of clang-3.8.1, it emits lwsync in AO_load_acquire */
/* (i.e., the code is less efficient than the one given below). */
# include "generic.h"
#else /* AO_DISABLE_GCC_ATOMICS */
/* Architecture enforces some ordering based on control dependence. */
/* I don't know if that could help. */
/* Data-dependent loads are always ordered. */
/* Based on the above references, eieio is intended for use on */
/* uncached memory, which we don't support. It does not order loads */
/* from cached memory. */
#include "../all_aligned_atomic_load_store.h"
#include "../test_and_set_t_is_ao_t.h"
/* There seems to be no byte equivalent of lwarx, so this */
/* may really be what we want, at least in the 32-bit case. */
AO_INLINE void
AO_nop_full(void)
{
__asm__ __volatile__("sync" : : : "memory");
}
#define AO_HAVE_nop_full
/* lwsync apparently works for everything but a StoreLoad barrier. */
AO_INLINE void
AO_lwsync(void)
{
#ifdef __NO_LWSYNC__
__asm__ __volatile__("sync" : : : "memory");
#else
__asm__ __volatile__("lwsync" : : : "memory");
#endif
}
#define AO_nop_write() AO_lwsync()
#define AO_HAVE_nop_write
#define AO_nop_read() AO_lwsync()
#define AO_HAVE_nop_read
#if defined(__powerpc64__) || defined(__ppc64__) || defined(__64BIT__)
/* ppc64 uses ld not lwz */
# define AO_PPC_LD "ld"
# define AO_PPC_LxARX "ldarx"
# define AO_PPC_CMPx "cmpd"
# define AO_PPC_STxCXd "stdcx."
# define AO_PPC_LOAD_CLOBBER "cr0"
#else
# define AO_PPC_LD "lwz"
# define AO_PPC_LxARX "lwarx"
# define AO_PPC_CMPx "cmpw"
# define AO_PPC_STxCXd "stwcx."
# define AO_PPC_LOAD_CLOBBER "cc"
/* FIXME: We should get gcc to allocate one of the condition */
/* registers. I always got "impossible constraint" when I */
/* tried the "y" constraint. */
# define AO_T_IS_INT
#endif
#ifdef _AIX
/* Labels are not supported on AIX. */
/* ppc64 has same size of instructions as 32-bit one. */
# define AO_PPC_L(label) /* empty */
# define AO_PPC_BR_A(labelBF, addr) addr
#else
# define AO_PPC_L(label) label ": "
# define AO_PPC_BR_A(labelBF, addr) labelBF
#endif
/* We explicitly specify load_acquire, since it is important, and can */
/* be implemented relatively cheaply. It could be implemented */
/* with an ordinary load followed by a lwsync. But the general wisdom */
/* seems to be that a data dependent branch followed by an isync is */
/* cheaper. And the documentation is fairly explicit that this also */
/* has acquire semantics. */
AO_INLINE AO_t
AO_load_acquire(const volatile AO_t *addr)
{
AO_t result;
__asm__ __volatile__ (
AO_PPC_LD "%U1%X1 %0,%1\n"
"cmpw %0,%0\n"
"bne- " AO_PPC_BR_A("1f", "$+4") "\n"
AO_PPC_L("1") "isync\n"
: "=r" (result)
: "m"(*addr) : "memory", AO_PPC_LOAD_CLOBBER);
return result;
}
#define AO_HAVE_load_acquire
/* We explicitly specify store_release, since it relies */
/* on the fact that lwsync is also a LoadStore barrier. */
AO_INLINE void
AO_store_release(volatile AO_t *addr, AO_t value)
{
AO_lwsync();
*addr = value;
}
#define AO_HAVE_store_release
#ifndef AO_PREFER_GENERALIZED
/* This is similar to the code in the garbage collector. Deleting */
/* this and having it synthesized from compare_and_swap would probably */
/* only cost us a load immediate instruction. */
AO_INLINE AO_TS_VAL_t
AO_test_and_set(volatile AO_TS_t *addr) {
/* Completely untested. And we should be using smaller objects anyway. */
AO_t oldval;
AO_t temp = 1; /* locked value */
__asm__ __volatile__(
AO_PPC_L("1") AO_PPC_LxARX " %0,0,%1\n"
/* load and reserve */
AO_PPC_CMPx "i %0, 0\n" /* if load is */
"bne " AO_PPC_BR_A("2f", "$+12") "\n"
/* non-zero, return already set */
AO_PPC_STxCXd " %2,0,%1\n" /* else store conditional */
"bne- " AO_PPC_BR_A("1b", "$-16") "\n"
/* retry if lost reservation */
AO_PPC_L("2") "\n" /* oldval is zero if we set */
: "=&r"(oldval)
: "r"(addr), "r"(temp)
: "memory", "cr0");
return (AO_TS_VAL_t)oldval;
}
#define AO_HAVE_test_and_set
AO_INLINE AO_TS_VAL_t
AO_test_and_set_acquire(volatile AO_TS_t *addr) {
AO_TS_VAL_t result = AO_test_and_set(addr);
AO_lwsync();
return result;
}
#define AO_HAVE_test_and_set_acquire
AO_INLINE AO_TS_VAL_t
AO_test_and_set_release(volatile AO_TS_t *addr) {
AO_lwsync();
return AO_test_and_set(addr);
}
#define AO_HAVE_test_and_set_release
AO_INLINE AO_TS_VAL_t
AO_test_and_set_full(volatile AO_TS_t *addr) {
AO_TS_VAL_t result;
AO_lwsync();
result = AO_test_and_set(addr);
AO_lwsync();
return result;
}
#define AO_HAVE_test_and_set_full
#endif /* !AO_PREFER_GENERALIZED */
#ifndef AO_GENERALIZE_ASM_BOOL_CAS
AO_INLINE int
AO_compare_and_swap(volatile AO_t *addr, AO_t old, AO_t new_val)
{
AO_t oldval;
int result = 0;
__asm__ __volatile__(
AO_PPC_L("1") AO_PPC_LxARX " %0,0,%2\n" /* load and reserve */
AO_PPC_CMPx " %0, %4\n" /* if load is not equal to */
"bne " AO_PPC_BR_A("2f", "$+16") "\n" /* old, fail */
AO_PPC_STxCXd " %3,0,%2\n" /* else store conditional */
"bne- " AO_PPC_BR_A("1b", "$-16") "\n"
/* retry if lost reservation */
"li %1,1\n" /* result = 1; */
AO_PPC_L("2") "\n"
: "=&r"(oldval), "=&r"(result)
: "r"(addr), "r"(new_val), "r"(old), "1"(result)
: "memory", "cr0");
return result;
}
# define AO_HAVE_compare_and_swap
AO_INLINE int
AO_compare_and_swap_acquire(volatile AO_t *addr, AO_t old, AO_t new_val)
{
int result = AO_compare_and_swap(addr, old, new_val);
AO_lwsync();
return result;
}
# define AO_HAVE_compare_and_swap_acquire
AO_INLINE int
AO_compare_and_swap_release(volatile AO_t *addr, AO_t old, AO_t new_val)
{
AO_lwsync();
return AO_compare_and_swap(addr, old, new_val);
}
# define AO_HAVE_compare_and_swap_release
AO_INLINE int
AO_compare_and_swap_full(volatile AO_t *addr, AO_t old, AO_t new_val)
{
int result;
AO_lwsync();
result = AO_compare_and_swap(addr, old, new_val);
if (result)
AO_lwsync();
return result;
}
# define AO_HAVE_compare_and_swap_full
#endif /* !AO_GENERALIZE_ASM_BOOL_CAS */
AO_INLINE AO_t
AO_fetch_compare_and_swap(volatile AO_t *addr, AO_t old_val, AO_t new_val)
{
AO_t fetched_val;
__asm__ __volatile__(
AO_PPC_L("1") AO_PPC_LxARX " %0,0,%1\n" /* load and reserve */
AO_PPC_CMPx " %0, %3\n" /* if load is not equal to */
"bne " AO_PPC_BR_A("2f", "$+12") "\n" /* old_val, fail */
AO_PPC_STxCXd " %2,0,%1\n" /* else store conditional */
"bne- " AO_PPC_BR_A("1b", "$-16") "\n"
/* retry if lost reservation */
AO_PPC_L("2") "\n"
: "=&r"(fetched_val)
: "r"(addr), "r"(new_val), "r"(old_val)
: "memory", "cr0");
return fetched_val;
}
#define AO_HAVE_fetch_compare_and_swap
AO_INLINE AO_t
AO_fetch_compare_and_swap_acquire(volatile AO_t *addr, AO_t old_val,
AO_t new_val)
{
AO_t result = AO_fetch_compare_and_swap(addr, old_val, new_val);
AO_lwsync();
return result;
}
#define AO_HAVE_fetch_compare_and_swap_acquire
AO_INLINE AO_t
AO_fetch_compare_and_swap_release(volatile AO_t *addr, AO_t old_val,
AO_t new_val)
{
AO_lwsync();
return AO_fetch_compare_and_swap(addr, old_val, new_val);
}
#define AO_HAVE_fetch_compare_and_swap_release
AO_INLINE AO_t
AO_fetch_compare_and_swap_full(volatile AO_t *addr, AO_t old_val,
AO_t new_val)
{
AO_t result;
AO_lwsync();
result = AO_fetch_compare_and_swap(addr, old_val, new_val);
if (result == old_val)
AO_lwsync();
return result;
}
#define AO_HAVE_fetch_compare_and_swap_full
#ifndef AO_PREFER_GENERALIZED
AO_INLINE AO_t
AO_fetch_and_add(volatile AO_t *addr, AO_t incr) {
AO_t oldval;
AO_t newval;
__asm__ __volatile__(
AO_PPC_L("1") AO_PPC_LxARX " %0,0,%2\n" /* load and reserve */
"add %1,%0,%3\n" /* increment */
AO_PPC_STxCXd " %1,0,%2\n" /* store conditional */
"bne- " AO_PPC_BR_A("1b", "$-12") "\n"
/* retry if lost reservation */
: "=&r"(oldval), "=&r"(newval)
: "r"(addr), "r"(incr)
: "memory", "cr0");
return oldval;
}
#define AO_HAVE_fetch_and_add
AO_INLINE AO_t
AO_fetch_and_add_acquire(volatile AO_t *addr, AO_t incr) {
AO_t result = AO_fetch_and_add(addr, incr);
AO_lwsync();
return result;
}
#define AO_HAVE_fetch_and_add_acquire
AO_INLINE AO_t
AO_fetch_and_add_release(volatile AO_t *addr, AO_t incr) {
AO_lwsync();
return AO_fetch_and_add(addr, incr);
}
#define AO_HAVE_fetch_and_add_release
AO_INLINE AO_t
AO_fetch_and_add_full(volatile AO_t *addr, AO_t incr) {
AO_t result;
AO_lwsync();
result = AO_fetch_and_add(addr, incr);
AO_lwsync();
return result;
}
#define AO_HAVE_fetch_and_add_full
#endif /* !AO_PREFER_GENERALIZED */
#undef AO_PPC_BR_A
#undef AO_PPC_CMPx
#undef AO_PPC_L
#undef AO_PPC_LD
#undef AO_PPC_LOAD_CLOBBER
#undef AO_PPC_LxARX
#undef AO_PPC_STxCXd
#endif /* AO_DISABLE_GCC_ATOMICS */