^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) // SPDX-License-Identifier: GPL-2.0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * Copyright (C) 2015 Davidlohr Bueso.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * Block a bunch of threads and let parallel waker threads wakeup an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * equal amount of them. The program output reflects the avg latency
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) * for each individual thread to service its share of work. Ultimately
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) * it can be used to measure futex_wake() changes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) #include "bench.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) #include <linux/compiler.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) #include "../util/debug.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) #ifndef HAVE_PTHREAD_BARRIER
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) int bench_futex_wake_parallel(int argc __maybe_unused, const char **argv __maybe_unused)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) pr_err("%s: pthread_barrier_t unavailable, disabling this test...\n", __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #else /* HAVE_PTHREAD_BARRIER */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) /* For the CLR_() macros */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) #include <string.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) #include <pthread.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) #include <signal.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #include "../util/stat.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) #include <subcmd/parse-options.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) #include <linux/kernel.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) #include <linux/time64.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) #include <errno.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) #include "futex.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) #include <internal/cpumap.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) #include <perf/cpumap.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) #include <err.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) #include <stdlib.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) #include <sys/time.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) struct thread_data {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) pthread_t worker;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) unsigned int nwoken;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) struct timeval runtime;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) static unsigned int nwakes = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) /* all threads will block on the same futex -- hash bucket chaos ;) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) static u_int32_t futex = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) static pthread_t *blocked_worker;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) static bool done = false, silent = false, fshared = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) static unsigned int nblocked_threads = 0, nwaking_threads = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) static pthread_mutex_t thread_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) static pthread_cond_t thread_parent, thread_worker;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) static pthread_barrier_t barrier;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) static struct stats waketime_stats, wakeup_stats;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) static unsigned int threads_starting;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) static int futex_flag = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) static const struct option options[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) OPT_UINTEGER('t', "threads", &nblocked_threads, "Specify amount of threads"),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) OPT_UINTEGER('w', "nwakers", &nwaking_threads, "Specify amount of waking threads"),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) OPT_BOOLEAN( 's', "silent", &silent, "Silent mode: do not display data/details"),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) OPT_BOOLEAN( 'S', "shared", &fshared, "Use shared futexes instead of private ones"),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) OPT_END()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) static const char * const bench_futex_wake_parallel_usage[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) "perf bench futex wake-parallel <options>",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) NULL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) static void *waking_workerfn(void *arg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) struct thread_data *waker = (struct thread_data *) arg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) struct timeval start, end;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) pthread_barrier_wait(&barrier);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) gettimeofday(&start, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) waker->nwoken = futex_wake(&futex, nwakes, futex_flag);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) if (waker->nwoken != nwakes)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) warnx("couldn't wakeup all tasks (%d/%d)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) waker->nwoken, nwakes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) gettimeofday(&end, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) timersub(&end, &start, &waker->runtime);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) pthread_exit(NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) static void wakeup_threads(struct thread_data *td, pthread_attr_t thread_attr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) unsigned int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) pthread_attr_setdetachstate(&thread_attr, PTHREAD_CREATE_JOINABLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) pthread_barrier_init(&barrier, NULL, nwaking_threads + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) /* create and block all threads */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) for (i = 0; i < nwaking_threads; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) * Thread creation order will impact per-thread latency
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) * as it will affect the order to acquire the hb spinlock.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) * For now let the scheduler decide.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) if (pthread_create(&td[i].worker, &thread_attr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) waking_workerfn, (void *)&td[i]))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) err(EXIT_FAILURE, "pthread_create");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) pthread_barrier_wait(&barrier);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) for (i = 0; i < nwaking_threads; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) if (pthread_join(td[i].worker, NULL))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) err(EXIT_FAILURE, "pthread_join");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) pthread_barrier_destroy(&barrier);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) static void *blocked_workerfn(void *arg __maybe_unused)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) pthread_mutex_lock(&thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) threads_starting--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) if (!threads_starting)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) pthread_cond_signal(&thread_parent);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) pthread_cond_wait(&thread_worker, &thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) pthread_mutex_unlock(&thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) while (1) { /* handle spurious wakeups */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) if (futex_wait(&futex, 0, NULL, futex_flag) != EINTR)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) pthread_exit(NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) static void block_threads(pthread_t *w, pthread_attr_t thread_attr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) struct perf_cpu_map *cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) cpu_set_t cpuset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) unsigned int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) threads_starting = nblocked_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) /* create and block all threads */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) for (i = 0; i < nblocked_threads; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) CPU_ZERO(&cpuset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) CPU_SET(cpu->map[i % cpu->nr], &cpuset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) if (pthread_create(&w[i], &thread_attr, blocked_workerfn, NULL))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) err(EXIT_FAILURE, "pthread_create");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) static void print_run(struct thread_data *waking_worker, unsigned int run_num)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) unsigned int i, wakeup_avg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) double waketime_avg, waketime_stddev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) struct stats __waketime_stats, __wakeup_stats;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) init_stats(&__wakeup_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) init_stats(&__waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) for (i = 0; i < nwaking_threads; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) update_stats(&__waketime_stats, waking_worker[i].runtime.tv_usec);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) update_stats(&__wakeup_stats, waking_worker[i].nwoken);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) waketime_avg = avg_stats(&__waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) waketime_stddev = stddev_stats(&__waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) wakeup_avg = avg_stats(&__wakeup_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) printf("[Run %d]: Avg per-thread latency (waking %d/%d threads) "
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) "in %.4f ms (+-%.2f%%)\n", run_num + 1, wakeup_avg,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) nblocked_threads, waketime_avg / USEC_PER_MSEC,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) rel_stddev_stats(waketime_stddev, waketime_avg));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) static void print_summary(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) unsigned int wakeup_avg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) double waketime_avg, waketime_stddev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) waketime_avg = avg_stats(&waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) waketime_stddev = stddev_stats(&waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) wakeup_avg = avg_stats(&wakeup_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) printf("Avg per-thread latency (waking %d/%d threads) in %.4f ms (+-%.2f%%)\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) wakeup_avg,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) nblocked_threads,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) waketime_avg / USEC_PER_MSEC,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) rel_stddev_stats(waketime_stddev, waketime_avg));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) static void do_run_stats(struct thread_data *waking_worker)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) unsigned int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) for (i = 0; i < nwaking_threads; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) update_stats(&waketime_stats, waking_worker[i].runtime.tv_usec);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) update_stats(&wakeup_stats, waking_worker[i].nwoken);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) static void toggle_done(int sig __maybe_unused,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) siginfo_t *info __maybe_unused,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) void *uc __maybe_unused)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) done = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) int bench_futex_wake_parallel(int argc, const char **argv)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) unsigned int i, j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) struct sigaction act;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) pthread_attr_t thread_attr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) struct thread_data *waking_worker;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) struct perf_cpu_map *cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) argc = parse_options(argc, argv, options,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) bench_futex_wake_parallel_usage, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) if (argc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) usage_with_options(bench_futex_wake_parallel_usage, options);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) exit(EXIT_FAILURE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) memset(&act, 0, sizeof(act));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) sigfillset(&act.sa_mask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) act.sa_sigaction = toggle_done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) sigaction(SIGINT, &act, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) cpu = perf_cpu_map__new(NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) if (!cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) err(EXIT_FAILURE, "calloc");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) if (!nblocked_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) nblocked_threads = cpu->nr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) /* some sanity checks */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) if (nwaking_threads > nblocked_threads || !nwaking_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) nwaking_threads = nblocked_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) if (nblocked_threads % nwaking_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) errx(EXIT_FAILURE, "Must be perfectly divisible");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) * Each thread will wakeup nwakes tasks in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) * a single futex_wait call.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) nwakes = nblocked_threads/nwaking_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) blocked_worker = calloc(nblocked_threads, sizeof(*blocked_worker));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) if (!blocked_worker)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) err(EXIT_FAILURE, "calloc");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) if (!fshared)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) futex_flag = FUTEX_PRIVATE_FLAG;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) printf("Run summary [PID %d]: blocking on %d threads (at [%s] "
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) "futex %p), %d threads waking up %d at a time.\n\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) getpid(), nblocked_threads, fshared ? "shared":"private",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) &futex, nwaking_threads, nwakes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) init_stats(&wakeup_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) init_stats(&waketime_stats);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) pthread_attr_init(&thread_attr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) pthread_mutex_init(&thread_lock, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) pthread_cond_init(&thread_parent, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) pthread_cond_init(&thread_worker, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) for (j = 0; j < bench_repeat && !done; j++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) waking_worker = calloc(nwaking_threads, sizeof(*waking_worker));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) if (!waking_worker)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) err(EXIT_FAILURE, "calloc");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) /* create, launch & block all threads */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) block_threads(blocked_worker, thread_attr, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) /* make sure all threads are already blocked */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) pthread_mutex_lock(&thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) while (threads_starting)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) pthread_cond_wait(&thread_parent, &thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) pthread_cond_broadcast(&thread_worker);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) pthread_mutex_unlock(&thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) usleep(100000);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) /* Ok, all threads are patiently blocked, start waking folks up */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) wakeup_threads(waking_worker, thread_attr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) for (i = 0; i < nblocked_threads; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) ret = pthread_join(blocked_worker[i], NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) err(EXIT_FAILURE, "pthread_join");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) do_run_stats(waking_worker);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) if (!silent)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) print_run(waking_worker, j);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) free(waking_worker);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) /* cleanup & report results */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) pthread_cond_destroy(&thread_parent);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) pthread_cond_destroy(&thread_worker);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) pthread_mutex_destroy(&thread_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) pthread_attr_destroy(&thread_attr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) print_summary();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) free(blocked_worker);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) perf_cpu_map__put(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) #endif /* HAVE_PTHREAD_BARRIER */
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags