[mirror_ubuntu-hirsute-kernel.git] / tools / perf / bench / futex-hash.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (C) 2013  Davidlohr Bueso <davidlohr@hp.com>
 *
 * futex-hash: Stress the hell out of the Linux kernel futex uaddr hashing.
 *
 * This program is particularly useful for measuring the kernel's futex hash
 * table/function implementation. In order for it to make sense, use with as
 * many threads and futexes as possible.
 */

/* For the CLR_() macros */
#include <string.h>
#include <pthread.h>

#include <errno.h>
#include <signal.h>
#include <stdlib.h>
#include <linux/compiler.h>
#include <linux/kernel.h>
#include <linux/zalloc.h>
#include <sys/time.h>
#include <internal/cpumap.h>
#include <perf/cpumap.h>

#include "../util/stat.h"
#include <subcmd/parse-options.h>
#include "bench.h"
#include "futex.h"

#include <err.h>

static unsigned int nthreads = 0;
static unsigned int nsecs    = 10;
/* amount of futexes per thread */
static unsigned int nfutexes = 1024;
static bool fshared = false, done = false, silent = false;
static int futex_flag = 0;

struct timeval start, end, runtime;
static pthread_mutex_t thread_lock;
static unsigned int threads_starting;
static struct stats throughput_stats;
static pthread_cond_t thread_parent, thread_worker;

struct worker {
	int tid;
	u_int32_t *futex;
	pthread_t thread;
	unsigned long ops;
};

static const struct option options[] = {
	OPT_UINTEGER('t', "threads", &nthreads, "Specify amount of threads"),
	OPT_UINTEGER('r', "runtime", &nsecs,    "Specify runtime (in seconds)"),
	OPT_UINTEGER('f', "futexes", &nfutexes, "Specify amount of futexes per threads"),
	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_END()
};

static const char * const bench_futex_hash_usage[] = {
	"perf bench futex hash <options>",
	NULL
};

static void *workerfn(void *arg)
{
	int ret;
	struct worker *w = (struct worker *) arg;
	unsigned int i;
	unsigned long ops = w->ops; /* avoid cacheline bouncing */

	pthread_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);

	do {
		for (i = 0; i < nfutexes; i++, ops++) {
			/*
			 * We want the futex calls to fail in order to stress
			 * the hashing of uaddr and not measure other steps,
			 * such as internal waitqueue handling, thus enlarging
			 * the critical region protected by hb->lock.
			 */
			ret = futex_wait(&w->futex[i], 1234, NULL, futex_flag);
			if (!silent &&
			    (!ret || errno != EAGAIN || errno != EWOULDBLOCK))
				warn("Non-expected futex return call");
		}
	}  while (!done);

	w->ops = ops;
	return NULL;
}

static void toggle_done(int sig __maybe_unused,
			siginfo_t *info __maybe_unused,
			void *uc __maybe_unused)
{
	/* inform all threads that we're done for the day */
	done = true;
	gettimeofday(&end, NULL);
	timersub(&end, &start, &runtime);
}

static void print_summary(void)
{
	unsigned long avg = avg_stats(&throughput_stats);
	double stddev = stddev_stats(&throughput_stats);

	printf("%sAveraged %ld operations/sec (+- %.2f%%), total secs = %d\n",
	       !silent ? "\n" : "", avg, rel_stddev_stats(stddev, avg),
	       (int) runtime.tv_sec);
}

int bench_futex_hash(int argc, const char **argv)
{
	int ret = 0;
	cpu_set_t cpuset;
	struct sigaction act;
	unsigned int i;
	pthread_attr_t thread_attr;
	struct worker *worker = NULL;
	struct perf_cpu_map *cpu;

	argc = parse_options(argc, argv, options, bench_futex_hash_usage, 0);
	if (argc) {
		usage_with_options(bench_futex_hash_usage, options);
		exit(EXIT_FAILURE);
	}

	cpu = perf_cpu_map__new(NULL);
	if (!cpu)
		goto errmem;

	sigfillset(&act.sa_mask);
	act.sa_sigaction = toggle_done;
	sigaction(SIGINT, &act, NULL);

	if (!nthreads) /* default to the number of CPUs */
		nthreads = cpu->nr;

	worker = calloc(nthreads, sizeof(*worker));
	if (!worker)
		goto errmem;

	if (!fshared)
		futex_flag = FUTEX_PRIVATE_FLAG;

	printf("Run summary [PID %d]: %d threads, each operating on %d [%s] futexes for %d secs.\n\n",
	       getpid(), nthreads, nfutexes, fshared ? "shared":"private", nsecs);

	init_stats(&throughput_stats);
	pthread_mutex_init(&thread_lock, NULL);
	pthread_cond_init(&thread_parent, NULL);
	pthread_cond_init(&thread_worker, NULL);

	threads_starting = nthreads;
	pthread_attr_init(&thread_attr);
	gettimeofday(&start, NULL);
	for (i = 0; i < nthreads; i++) {
		worker[i].tid = i;
		worker[i].futex = calloc(nfutexes, sizeof(*worker[i].futex));
		if (!worker[i].futex)
			goto errmem;

		CPU_ZERO(&cpuset);
		CPU_SET(cpu->map[i % cpu->nr], &cpuset);

		ret = pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset);
		if (ret)
			err(EXIT_FAILURE, "pthread_attr_setaffinity_np");

		ret = pthread_create(&worker[i].thread, &thread_attr, workerfn,
				     (void *)(struct worker *) &worker[i]);
		if (ret)
			err(EXIT_FAILURE, "pthread_create");

	}
	pthread_attr_destroy(&thread_attr);

	pthread_mutex_lock(&thread_lock);
	while (threads_starting)
		pthread_cond_wait(&thread_parent, &thread_lock);
	pthread_cond_broadcast(&thread_worker);
	pthread_mutex_unlock(&thread_lock);

	sleep(nsecs);
	toggle_done(0, NULL, NULL);

	for (i = 0; i < nthreads; i++) {
		ret = pthread_join(worker[i].thread, NULL);
		if (ret)
			err(EXIT_FAILURE, "pthread_join");
	}

	/* cleanup & report results */
	pthread_cond_destroy(&thread_parent);
	pthread_cond_destroy(&thread_worker);
	pthread_mutex_destroy(&thread_lock);

	for (i = 0; i < nthreads; i++) {
		unsigned long t = worker[i].ops/runtime.tv_sec;
		update_stats(&throughput_stats, t);
		if (!silent) {
			if (nfutexes == 1)
				printf("[thread %2d] futex: %p [ %ld ops/sec ]\n",
				       worker[i].tid, &worker[i].futex[0], t);
			else
				printf("[thread %2d] futexes: %p ... %p [ %ld ops/sec ]\n",
				       worker[i].tid, &worker[i].futex[0],
				       &worker[i].futex[nfutexes-1], t);
		}

		zfree(&worker[i].futex);
	}

	print_summary();

	free(worker);
	free(cpu);
	return ret;
errmem:
	err(EXIT_FAILURE, "calloc");
}
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
a0439711 DB	2	/*
	3	* Copyright (C) 2013 Davidlohr Bueso <davidlohr@hp.com>
	4	*
	5	* futex-hash: Stress the hell out of the Linux kernel futex uaddr hashing.
	6	*
	7	* This program is particularly useful for measuring the kernel's futex hash
	8	* table/function implementation. In order for it to make sense, use with as
	9	* many threads and futexes as possible.
	10	*/
	11
8a158589	12	/* For the CLR_() macros */
a0f213e1	13	#include <string.h>
8a158589 ACM	14	#include <pthread.h>
8a158589 ACM	15
9c304f6c ACM	16	#include <errno.h>
	17	#include <signal.h>
	18	#include <stdlib.h>
86695f59	19	#include <linux/compiler.h>
9c304f6c	20	#include <linux/kernel.h>
d8f9da24	21	#include <linux/zalloc.h>
9c304f6c	22	#include <sys/time.h>
87ffb6c6	23	#include <internal/cpumap.h>
9c3516d1	24	#include <perf/cpumap.h>
9c304f6c	25
a0439711	26	#include "../util/stat.h"
4b6ab94e	27	#include <subcmd/parse-options.h>
a0439711 DB	28	#include "bench.h"
	29	#include "futex.h"
	30
	31	#include <err.h>
a0439711 DB	32
	33	static unsigned int nthreads = 0;
	34	static unsigned int nsecs = 10;
	35	/* amount of futexes per thread */
	36	static unsigned int nfutexes = 1024;
	37	static bool fshared = false, done = false, silent = false;
86c87e13	38	static int futex_flag = 0;
a0439711 DB	39
	40	struct timeval start, end, runtime;
	41	static pthread_mutex_t thread_lock;
	42	static unsigned int threads_starting;
	43	static struct stats throughput_stats;
	44	static pthread_cond_t thread_parent, thread_worker;
	45
	46	struct worker {
	47	int tid;
	48	u_int32_t *futex;
	49	pthread_t thread;
	50	unsigned long ops;
e2e1680f	51	};
a0439711 DB	52
	53	static const struct option options[] = {
	54	OPT_UINTEGER('t', "threads", &nthreads, "Specify amount of threads"),
	55	OPT_UINTEGER('r', "runtime", &nsecs, "Specify runtime (in seconds)"),
	56	OPT_UINTEGER('f', "futexes", &nfutexes, "Specify amount of futexes per threads"),
	57	OPT_BOOLEAN( 's', "silent", &silent, "Silent mode: do not display data/details"),
	58	OPT_BOOLEAN( 'S', "shared", &fshared, "Use shared futexes instead of private ones"),
	59	OPT_END()
	60	};
	61
	62	static const char * const bench_futex_hash_usage[] = {
	63	"perf bench futex hash <options>",
	64	NULL
	65	};
	66
	67	static void workerfn(void arg)
	68	{
	69	int ret;
a0439711	70	struct worker w = (struct worker ) arg;
e2e1680f DB	71	unsigned int i;
e2e1680f DB	72	unsigned long ops = w->ops; /* avoid cacheline bouncing */
a0439711 DB	73
	74	pthread_mutex_lock(&thread_lock);
	75	threads_starting--;
	76	if (!threads_starting)
	77	pthread_cond_signal(&thread_parent);
	78	pthread_cond_wait(&thread_worker, &thread_lock);
	79	pthread_mutex_unlock(&thread_lock);
	80
	81	do {
e2e1680f	82	for (i = 0; i < nfutexes; i++, ops++) {
a0439711 DB	83	/*
	84	* We want the futex calls to fail in order to stress
	85	* the hashing of uaddr and not measure other steps,
	86	* such as internal waitqueue handling, thus enlarging
	87	* the critical region protected by hb->lock.
	88	*/
86c87e13	89	ret = futex_wait(&w->futex[i], 1234, NULL, futex_flag);
a0439711 DB	90	if (!silent &&
	91	(!ret \|\| errno != EAGAIN \|\| errno != EWOULDBLOCK))
	92	warn("Non-expected futex return call");
	93	}
	94	} while (!done);
	95
e2e1680f	96	w->ops = ops;
a0439711 DB	97	return NULL;
	98	}
	99
	100	static void toggle_done(int sig __maybe_unused,
	101	siginfo_t *info __maybe_unused,
	102	void *uc __maybe_unused)
	103	{
	104	/* inform all threads that we're done for the day */
	105	done = true;
	106	gettimeofday(&end, NULL);
	107	timersub(&end, &start, &runtime);
	108	}
	109
	110	static void print_summary(void)
	111	{
	112	unsigned long avg = avg_stats(&throughput_stats);
	113	double stddev = stddev_stats(&throughput_stats);
	114
	115	printf("%sAveraged %ld operations/sec (+- %.2f%%), total secs = %d\n",
	116	!silent ? "\n" : "", avg, rel_stddev_stats(stddev, avg),
	117	(int) runtime.tv_sec);
	118	}
	119
b0ad8ea6	120	int bench_futex_hash(int argc, const char **argv)
a0439711 DB	121	{
a0439711 DB	122	int ret = 0;
3b2323c2	123	cpu_set_t cpuset;
a0439711	124	struct sigaction act;
3b2323c2	125	unsigned int i;
a0439711 DB	126	pthread_attr_t thread_attr;
a0439711 DB	127	struct worker *worker = NULL;
f854839b	128	struct perf_cpu_map *cpu;
a0439711 DB	129
	130	argc = parse_options(argc, argv, options, bench_futex_hash_usage, 0);
	131	if (argc) {
	132	usage_with_options(bench_futex_hash_usage, options);
	133	exit(EXIT_FAILURE);
	134	}
	135
9c3516d1	136	cpu = perf_cpu_map__new(NULL);
3b2323c2 DB	137	if (!cpu)
3b2323c2 DB	138	goto errmem;
a0439711 DB	139
	140	sigfillset(&act.sa_mask);
	141	act.sa_sigaction = toggle_done;
	142	sigaction(SIGINT, &act, NULL);
	143
	144	if (!nthreads) /* default to the number of CPUs */
3b2323c2	145	nthreads = cpu->nr;
a0439711 DB	146
	147	worker = calloc(nthreads, sizeof(*worker));
	148	if (!worker)
	149	goto errmem;
	150
86c87e13 DB	151	if (!fshared)
	152	futex_flag = FUTEX_PRIVATE_FLAG;
	153
a0439711 DB	154	printf("Run summary [PID %d]: %d threads, each operating on %d [%s] futexes for %d secs.\n\n",
	155	getpid(), nthreads, nfutexes, fshared ? "shared":"private", nsecs);
	156
	157	init_stats(&throughput_stats);
	158	pthread_mutex_init(&thread_lock, NULL);
	159	pthread_cond_init(&thread_parent, NULL);
	160	pthread_cond_init(&thread_worker, NULL);
	161
	162	threads_starting = nthreads;
	163	pthread_attr_init(&thread_attr);
	164	gettimeofday(&start, NULL);
	165	for (i = 0; i < nthreads; i++) {
	166	worker[i].tid = i;
	167	worker[i].futex = calloc(nfutexes, sizeof(*worker[i].futex));
	168	if (!worker[i].futex)
	169	goto errmem;
	170
3b2323c2 DB	171	CPU_ZERO(&cpuset);
3b2323c2 DB	172	CPU_SET(cpu->map[i % cpu->nr], &cpuset);
a0439711	173
3b2323c2	174	ret = pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset);
a0439711 DB	175	if (ret)
	176	err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
	177
	178	ret = pthread_create(&worker[i].thread, &thread_attr, workerfn,
	179	(void )(struct worker ) &worker[i]);
	180	if (ret)
	181	err(EXIT_FAILURE, "pthread_create");
	182
	183	}
	184	pthread_attr_destroy(&thread_attr);
	185
	186	pthread_mutex_lock(&thread_lock);
	187	while (threads_starting)
	188	pthread_cond_wait(&thread_parent, &thread_lock);
	189	pthread_cond_broadcast(&thread_worker);
	190	pthread_mutex_unlock(&thread_lock);
	191
	192	sleep(nsecs);
	193	toggle_done(0, NULL, NULL);
	194
	195	for (i = 0; i < nthreads; i++) {
	196	ret = pthread_join(worker[i].thread, NULL);
	197	if (ret)
	198	err(EXIT_FAILURE, "pthread_join");
	199	}
	200
	201	/* cleanup & report results */
	202	pthread_cond_destroy(&thread_parent);
	203	pthread_cond_destroy(&thread_worker);
	204	pthread_mutex_destroy(&thread_lock);
	205
	206	for (i = 0; i < nthreads; i++) {
	207	unsigned long t = worker[i].ops/runtime.tv_sec;
	208	update_stats(&throughput_stats, t);
	209	if (!silent) {
	210	if (nfutexes == 1)
	211	printf("[thread %2d] futex: %p [ %ld ops/sec ]\n",
	212	worker[i].tid, &worker[i].futex[0], t);
	213	else
	214	printf("[thread %2d] futexes: %p ... %p [ %ld ops/sec ]\n",
	215	worker[i].tid, &worker[i].futex[0],
	216	&worker[i].futex[nfutexes-1], t);
	217	}
	218
d8f9da24	219	zfree(&worker[i].futex);
a0439711 DB	220	}
	221
	222	print_summary();
	223
	224	free(worker);
3b2323c2	225	free(cpu);
a0439711 DB	226	return ret;
	227	errmem:
	228	err(EXIT_FAILURE, "calloc");
	229	}