diff options
Diffstat (limited to '')
-rw-r--r-- | tools/perf/bench/futex-requeue.c | 215 |
1 files changed, 153 insertions, 62 deletions
diff --git a/tools/perf/bench/futex-requeue.c b/tools/perf/bench/futex-requeue.c index 7a15c2e61022..69ad896f556c 100644 --- a/tools/perf/bench/futex-requeue.c +++ b/tools/perf/bench/futex-requeue.c @@ -6,7 +6,8 @@ * on futex2, N at a time. * * This program is particularly useful to measure the latency of nthread - * requeues without waking up any tasks -- thus mimicking a regular futex_wait. + * requeues without waking up any tasks (in the non-pi case) -- thus + * mimicking a regular futex_wait. */ /* For the CLR_() macros */ @@ -14,13 +15,13 @@ #include <pthread.h> #include <signal.h> +#include "../util/mutex.h" #include "../util/stat.h" #include <subcmd/parse-options.h> #include <linux/compiler.h> #include <linux/kernel.h> #include <linux/time64.h> #include <errno.h> -#include <internal/cpumap.h> #include <perf/cpumap.h> #include "bench.h" #include "futex.h" @@ -28,28 +29,35 @@ #include <err.h> #include <stdlib.h> #include <sys/time.h> +#include <sys/mman.h> static u_int32_t futex1 = 0, futex2 = 0; -/* - * How many tasks to requeue at a time. - * Default to 1 in order to make the kernel work more. - */ -static unsigned int nrequeue = 1; - static pthread_t *worker; -static bool done = false, silent = false, fshared = false; -static pthread_mutex_t thread_lock; -static pthread_cond_t thread_parent, thread_worker; +static bool done = false; +static struct mutex thread_lock; +static struct cond thread_parent, thread_worker; static struct stats requeuetime_stats, requeued_stats; -static unsigned int threads_starting, nthreads = 0; +static unsigned int threads_starting; static int futex_flag = 0; +static struct bench_futex_parameters params = { + /* + * How many tasks to requeue at a time. + * Default to 1 in order to make the kernel work more. + */ + .nrequeue = 1, +}; + static const struct option options[] = { - OPT_UINTEGER('t', "threads", &nthreads, "Specify amount of threads"), - OPT_UINTEGER('q', "nrequeue", &nrequeue, "Specify amount of threads to requeue at once"), - OPT_BOOLEAN( 's', "silent", &silent, "Silent mode: do not display data/details"), - OPT_BOOLEAN( 'S', "shared", &fshared, "Use shared futexes instead of private ones"), + OPT_UINTEGER('t', "threads", ¶ms.nthreads, "Specify amount of threads"), + OPT_UINTEGER('q', "nrequeue", ¶ms.nrequeue, "Specify amount of threads to requeue at once"), + OPT_BOOLEAN( 's', "silent", ¶ms.silent, "Silent mode: do not display data/details"), + OPT_BOOLEAN( 'S', "shared", ¶ms.fshared, "Use shared futexes instead of private ones"), + OPT_BOOLEAN( 'm', "mlockall", ¶ms.mlockall, "Lock all current and future memory"), + OPT_BOOLEAN( 'B', "broadcast", ¶ms.broadcast, "Requeue all threads at once"), + OPT_BOOLEAN( 'p', "pi", ¶ms.pi, "Use PI-aware variants of FUTEX_CMP_REQUEUE"), + OPT_END() }; @@ -66,43 +74,83 @@ static void print_summary(void) printf("Requeued %d of %d threads in %.4f ms (+-%.2f%%)\n", requeued_avg, - nthreads, + params.nthreads, requeuetime_avg / USEC_PER_MSEC, rel_stddev_stats(requeuetime_stddev, requeuetime_avg)); } static void *workerfn(void *arg __maybe_unused) { - pthread_mutex_lock(&thread_lock); + int ret; + + mutex_lock(&thread_lock); threads_starting--; if (!threads_starting) - pthread_cond_signal(&thread_parent); - pthread_cond_wait(&thread_worker, &thread_lock); - pthread_mutex_unlock(&thread_lock); + cond_signal(&thread_parent); + cond_wait(&thread_worker, &thread_lock); + mutex_unlock(&thread_lock); + + while (1) { + if (!params.pi) { + ret = futex_wait(&futex1, 0, NULL, futex_flag); + if (!ret) + break; + + if (ret && errno != EAGAIN) { + if (!params.silent) + warnx("futex_wait"); + break; + } + } else { + ret = futex_wait_requeue_pi(&futex1, 0, &futex2, + NULL, futex_flag); + if (!ret) { + /* got the lock at futex2 */ + futex_unlock_pi(&futex2, futex_flag); + break; + } + + if (ret && errno != EAGAIN) { + if (!params.silent) + warnx("futex_wait_requeue_pi"); + break; + } + } + } - futex_wait(&futex1, 0, NULL, futex_flag); return NULL; } static void block_threads(pthread_t *w, pthread_attr_t thread_attr, struct perf_cpu_map *cpu) { - cpu_set_t cpuset; + cpu_set_t *cpuset; unsigned int i; + int nrcpus = perf_cpu_map__nr(cpu); + size_t size; - threads_starting = nthreads; + threads_starting = params.nthreads; + + cpuset = CPU_ALLOC(nrcpus); + BUG_ON(!cpuset); + size = CPU_ALLOC_SIZE(nrcpus); /* create and block all threads */ - for (i = 0; i < nthreads; i++) { - CPU_ZERO(&cpuset); - CPU_SET(cpu->map[i % cpu->nr], &cpuset); + for (i = 0; i < params.nthreads; i++) { + CPU_ZERO_S(size, cpuset); + CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset); - if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset)) + if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) { + CPU_FREE(cpuset); err(EXIT_FAILURE, "pthread_attr_setaffinity_np"); + } - if (pthread_create(&w[i], &thread_attr, workerfn, NULL)) + if (pthread_create(&w[i], &thread_attr, workerfn, NULL)) { + CPU_FREE(cpuset); err(EXIT_FAILURE, "pthread_create"); + } } + CPU_FREE(cpuset); } static void toggle_done(int sig __maybe_unused, @@ -133,55 +181,82 @@ int bench_futex_requeue(int argc, const char **argv) act.sa_sigaction = toggle_done; sigaction(SIGINT, &act, NULL); - if (!nthreads) - nthreads = cpu->nr; + if (params.mlockall) { + if (mlockall(MCL_CURRENT | MCL_FUTURE)) + err(EXIT_FAILURE, "mlockall"); + } + + if (!params.nthreads) + params.nthreads = perf_cpu_map__nr(cpu); - worker = calloc(nthreads, sizeof(*worker)); + worker = calloc(params.nthreads, sizeof(*worker)); if (!worker) err(EXIT_FAILURE, "calloc"); - if (!fshared) + if (!params.fshared) futex_flag = FUTEX_PRIVATE_FLAG; - if (nrequeue > nthreads) - nrequeue = nthreads; + if (params.nrequeue > params.nthreads) + params.nrequeue = params.nthreads; - printf("Run summary [PID %d]: Requeuing %d threads (from [%s] %p to %p), " - "%d at a time.\n\n", getpid(), nthreads, - fshared ? "shared":"private", &futex1, &futex2, nrequeue); + if (params.broadcast) + params.nrequeue = params.nthreads; + + printf("Run summary [PID %d]: Requeuing %d threads (from [%s] %p to %s%p), " + "%d at a time.\n\n", getpid(), params.nthreads, + params.fshared ? "shared":"private", &futex1, + params.pi ? "PI ": "", &futex2, params.nrequeue); init_stats(&requeued_stats); init_stats(&requeuetime_stats); pthread_attr_init(&thread_attr); - pthread_mutex_init(&thread_lock, NULL); - pthread_cond_init(&thread_parent, NULL); - pthread_cond_init(&thread_worker, NULL); + mutex_init(&thread_lock); + cond_init(&thread_parent); + cond_init(&thread_worker); for (j = 0; j < bench_repeat && !done; j++) { - unsigned int nrequeued = 0; + unsigned int nrequeued = 0, wakeups = 0; struct timeval start, end, runtime; /* create, launch & block all threads */ block_threads(worker, thread_attr, cpu); /* make sure all threads are already blocked */ - pthread_mutex_lock(&thread_lock); + mutex_lock(&thread_lock); while (threads_starting) - pthread_cond_wait(&thread_parent, &thread_lock); - pthread_cond_broadcast(&thread_worker); - pthread_mutex_unlock(&thread_lock); + cond_wait(&thread_parent, &thread_lock); + cond_broadcast(&thread_worker); + mutex_unlock(&thread_lock); usleep(100000); /* Ok, all threads are patiently blocked, start requeueing */ gettimeofday(&start, NULL); - while (nrequeued < nthreads) { + while (nrequeued < params.nthreads) { + int r; + /* - * Do not wakeup any tasks blocked on futex1, allowing - * us to really measure futex_wait functionality. + * For the regular non-pi case, do not wakeup any tasks + * blocked on futex1, allowing us to really measure + * futex_wait functionality. For the PI case the first + * waiter is always awoken. */ - nrequeued += futex_cmp_requeue(&futex1, 0, &futex2, 0, - nrequeue, futex_flag); + if (!params.pi) { + r = futex_cmp_requeue(&futex1, 0, &futex2, 0, + params.nrequeue, + futex_flag); + } else { + r = futex_cmp_requeue_pi(&futex1, 0, &futex2, + params.nrequeue, + futex_flag); + wakeups++; /* assume no error */ + } + + if (r < 0) + err(EXIT_FAILURE, "couldn't requeue from %p to %p", + &futex1, &futex2); + + nrequeued += r; } gettimeofday(&end, NULL); @@ -190,17 +265,32 @@ int bench_futex_requeue(int argc, const char **argv) update_stats(&requeued_stats, nrequeued); update_stats(&requeuetime_stats, runtime.tv_usec); - if (!silent) { - printf("[Run %d]: Requeued %d of %d threads in %.4f ms\n", - j + 1, nrequeued, nthreads, runtime.tv_usec / (double)USEC_PER_MSEC); + if (!params.silent) { + if (!params.pi) + printf("[Run %d]: Requeued %d of %d threads in " + "%.4f ms\n", j + 1, nrequeued, + params.nthreads, + runtime.tv_usec / (double)USEC_PER_MSEC); + else { + nrequeued -= wakeups; + printf("[Run %d]: Awoke and Requeued (%d+%d) of " + "%d threads in %.4f ms\n", + j + 1, wakeups, nrequeued, + params.nthreads, + runtime.tv_usec / (double)USEC_PER_MSEC); + } + } - /* everybody should be blocked on futex2, wake'em up */ - nrequeued = futex_wake(&futex2, nrequeued, futex_flag); - if (nthreads != nrequeued) - warnx("couldn't wakeup all tasks (%d/%d)", nrequeued, nthreads); + if (!params.pi) { + /* everybody should be blocked on futex2, wake'em up */ + nrequeued = futex_wake(&futex2, nrequeued, futex_flag); + if (params.nthreads != nrequeued) + warnx("couldn't wakeup all tasks (%d/%d)", + nrequeued, params.nthreads); + } - for (i = 0; i < nthreads; i++) { + for (i = 0; i < params.nthreads; i++) { ret = pthread_join(worker[i], NULL); if (ret) err(EXIT_FAILURE, "pthread_join"); @@ -208,14 +298,15 @@ int bench_futex_requeue(int argc, const char **argv) } /* cleanup & report results */ - pthread_cond_destroy(&thread_parent); - pthread_cond_destroy(&thread_worker); - pthread_mutex_destroy(&thread_lock); + cond_destroy(&thread_parent); + cond_destroy(&thread_worker); + mutex_destroy(&thread_lock); pthread_attr_destroy(&thread_attr); print_summary(); free(worker); + perf_cpu_map__put(cpu); return ret; err: usage_with_options(bench_futex_requeue_usage, options); |