/* * self_count.c - example of a simple self monitoring using mmapped page * * Copyright (c) 2009 Google, Inc * Contributed by Stephane Eranian * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies * of the Software, and to permit persons to whom the Software is furnished to do so, * subject to the following conditions: * * The above copyright notice and this permission notice shall be included in all * copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, * INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A * PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE * OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. * * This file is part of libpfm, a performance monitoring support library for * applications on Linux. */ #include #include #include #include #include #include #include #include #include #include #include #include #include "perf_util.h" static const char *gen_events[]={ "cycles", NULL }; static volatile int quit; void sig_handler(int n) { quit = 1; } #if defined(__x86_64__) || defined(__i386__) #ifdef __x86_64__ #define DECLARE_ARGS(val, low, high) unsigned low, high #define EAX_EDX_VAL(val, low, high) ((low) | ((uint64_t )(high) << 32)) #define EAX_EDX_ARGS(val, low, high) "a" (low), "d" (high) #define EAX_EDX_RET(val, low, high) "=a" (low), "=d" (high) #else #define DECLARE_ARGS(val, low, high) unsigned long long val #define EAX_EDX_VAL(val, low, high) (val) #define EAX_EDX_ARGS(val, low, high) "A" (val) #define EAX_EDX_RET(val, low, high) "=A" (val) #endif #define barrier() __asm__ __volatile__("": : :"memory") static inline int rdpmc(struct perf_event_mmap_page *hdr, uint64_t *value) { int counter = hdr->index - 1; DECLARE_ARGS(val, low, high); if (counter < 0) return -1; asm volatile("rdpmc" : EAX_EDX_RET(val, low, high) : "c" (counter)); *value = EAX_EDX_VAL(val, low, high); return 0; } #else /* * Default barrier macro. * Given this is architecture specific, it must be defined when * libpfm is ported to new architecture. The default macro below * simply does nothing. */ #define barrier() {} /* * Default function to read counter directly from user level mode. * Given this is architecture specific, it must be defined when * libpfm is ported to new architecture. The default routine below * simply fails and the caller falls backs to syscall. */ static inline int rdpmc(struct perf_event_mmap_page *hdr, uint64_t *value) { int counter = hdr->index - 1; if (counter < 0) return -1; printf("your architecture does not have a way to read counters from user mode\n"); return -1; } #endif /* * our test code (function cannot be made static otherwise it is optimized away) */ unsigned long fib(unsigned long n) { if (n == 0) return 0; if (n == 1) return 2; return fib(n-1)+fib(n-2); } uint64_t read_count(perf_event_desc_t *fds) { struct perf_event_mmap_page *hdr; uint64_t values[3]; uint64_t count = 0; uint32_t width; unsigned int seq; ssize_t ret; int idx = -1; hdr = fds->buf; width = hdr->pmc_width; do { seq = hdr->lock; barrier(); /* try reading directly from user mode */ if (!rdpmc(hdr, &values[0])) { values[1] = hdr->time_enabled; values[2] = hdr->time_running; ret = 0; } else { idx = -1; ret = read(fds->fd, values, sizeof(values)); if (ret < (ssize_t)sizeof(values)) errx(1, "cannot read values"); printf("using read\n"); break; } barrier(); } while (hdr->lock != seq); printf("raw=0x%"PRIx64 " width=%d ena=%"PRIu64 " run=%"PRIu64" idx=%d\n", values[0], width, values[1], values[2], idx); count = values[0]; count <<= 64 - width; count >>= 64 - width; values[0] = count; return perf_scale(values); } int main(int argc, char **argv) { perf_event_desc_t *fds = NULL; long lret; size_t pgsz; uint64_t val, prev_val; int i, ret, num_fds = 0; lret = sysconf(_SC_PAGESIZE); if (lret < 0) err(1, "cannot get page size"); pgsz = (size_t)lret; /* * Initialize pfm library (required before we can use it) */ ret = pfm_initialize(); if (ret != PFM_SUCCESS) errx(1, "Cannot initialize library: %s", pfm_strerror(ret)); ret = perf_setup_argv_events(argc > 1 ? (const char **)argv+1 : gen_events, &fds, &num_fds); if (ret || !num_fds) errx(1, "cannot setup events"); fds[0].fd = -1; for(i=0; i < num_fds; i++) { /* request timing information necesaary for scaling */ fds[i].hw.read_format = PERF_FORMAT_SCALE; fds[i].hw.disabled = 0; //fds[i].fd = perf_event_open(&fds[i].hw, 0, -1, fds[0].fd, 0); fds[i].fd = perf_event_open(&fds[i].hw, 0, -1, -1, 0); if (fds[i].fd == -1) err(1, "cannot open event %d", i); fds[i].buf = mmap(NULL, pgsz, PROT_READ, MAP_SHARED, fds[i].fd, 0); if (fds[i].buf == MAP_FAILED) err(1, "cannot mmap page"); } signal(SIGALRM, sig_handler); /* * enable all counters attached to this thread */ ioctl(fds[0].fd, PERF_EVENT_IOC_ENABLE, 0); alarm(10); prev_val = 0; for(;quit == 0;) { for (i = 0; i < num_fds; i++) { val = read_count(&fds[i]); /* print evnet deltas */ printf("%20"PRIu64" %s\n", val - prev_val, fds[i].name); prev_val = val; } fib(35); } /* * disable all counters attached to this thread */ ioctl(fds[0].fd, PERF_EVENT_IOC_DISABLE, 0); for (i=0; i < num_fds; i++) { munmap(fds[i].buf, pgsz); close(fds[i].fd); } perf_free_fds(fds, num_fds); /* free libpfm resources cleanly */ pfm_terminate(); return 0; }