[fio.git] / gettime.c

/*
 * Clock functions
 */

#include <unistd.h>
#include <math.h>
#include <sys/time.h>
#include <time.h>

#include "fio.h"
#include "smalloc.h"

#include "hash.h"
#include "os/os.h"

#ifdef ARCH_HAVE_CPU_CLOCK
static unsigned long cycles_per_usec;
int tsc_reliable = 0;
#endif

struct tv_valid {
	struct timeval last_tv;
	int last_tv_valid;
	unsigned long last_cycles;
};
static pthread_key_t tv_tls_key;

enum fio_cs fio_clock_source = FIO_PREFERRED_CLOCK_SOURCE;
int fio_clock_source_set = 0;
enum fio_cs fio_clock_source_inited = CS_INVAL;

#ifdef FIO_DEBUG_TIME

#define HASH_BITS	8
#define HASH_SIZE	(1 << HASH_BITS)

static struct flist_head hash[HASH_SIZE];
static int gtod_inited;

struct gtod_log {
	struct flist_head list;
	void *caller;
	unsigned long calls;
};

static struct gtod_log *find_hash(void *caller)
{
	unsigned long h = hash_ptr(caller, HASH_BITS);
	struct flist_head *entry;

	flist_for_each(entry, &hash[h]) {
		struct gtod_log *log = flist_entry(entry, struct gtod_log,
									list);

		if (log->caller == caller)
			return log;
	}

	return NULL;
}

static struct gtod_log *find_log(void *caller)
{
	struct gtod_log *log = find_hash(caller);

	if (!log) {
		unsigned long h;

		log = malloc(sizeof(*log));
		INIT_FLIST_HEAD(&log->list);
		log->caller = caller;
		log->calls = 0;

		h = hash_ptr(caller, HASH_BITS);
		flist_add_tail(&log->list, &hash[h]);
	}

	return log;
}

static void gtod_log_caller(void *caller)
{
	if (gtod_inited) {
		struct gtod_log *log = find_log(caller);

		log->calls++;
	}
}

static void fio_exit fio_dump_gtod(void)
{
	unsigned long total_calls = 0;
	int i;

	for (i = 0; i < HASH_SIZE; i++) {
		struct flist_head *entry;
		struct gtod_log *log;

		flist_for_each(entry, &hash[i]) {
			log = flist_entry(entry, struct gtod_log, list);

			printf("function %p, calls %lu\n", log->caller,
								log->calls);
			total_calls += log->calls;
		}
	}

	printf("Total %lu gettimeofday\n", total_calls);
}

static void fio_init gtod_init(void)
{
	int i;

	for (i = 0; i < HASH_SIZE; i++)
		INIT_FLIST_HEAD(&hash[i]);

	gtod_inited = 1;
}

#endif /* FIO_DEBUG_TIME */

static int fill_clock_gettime(struct timespec *ts)
{
#ifdef FIO_HAVE_CLOCK_MONOTONIC
	return clock_gettime(CLOCK_MONOTONIC, ts);
#else
	return clock_gettime(CLOCK_REALTIME, ts);
#endif
}

#ifdef FIO_DEBUG_TIME
void fio_gettime(struct timeval *tp, void *caller)
#else
void fio_gettime(struct timeval *tp, void fio_unused *caller)
#endif
{
	struct tv_valid *tv;

#ifdef FIO_DEBUG_TIME
	if (!caller)
		caller = __builtin_return_address(0);

	gtod_log_caller(caller);
#endif
	if (fio_tv) {
		memcpy(tp, fio_tv, sizeof(*tp));
		return;
	}

	tv = pthread_getspecific(tv_tls_key);

	switch (fio_clock_source) {
	case CS_GTOD:
		gettimeofday(tp, NULL);
		break;
	case CS_CGETTIME: {
		struct timespec ts;

		if (fill_clock_gettime(&ts) < 0) {
			log_err("fio: clock_gettime fails\n");
			assert(0);
		}

		tp->tv_sec = ts.tv_sec;
		tp->tv_usec = ts.tv_nsec / 1000;
		break;
		}
#ifdef ARCH_HAVE_CPU_CLOCK
	case CS_CPUCLOCK: {
		unsigned long long usecs, t;

		t = get_cpu_clock();
		if (tv && t < tv->last_cycles) {
			dprint(FD_TIME, "CPU clock going back in time\n");
			t = tv->last_cycles;
		} else if (tv)
			tv->last_cycles = t;

		usecs = t / cycles_per_usec;
		tp->tv_sec = usecs / 1000000;
		tp->tv_usec = usecs % 1000000;
		break;
		}
#endif
	default:
		log_err("fio: invalid clock source %d\n", fio_clock_source);
		break;
	}

	/*
	 * If Linux is using the tsc clock on non-synced processors,
	 * sometimes time can appear to drift backwards. Fix that up.
	 */
	if (tv) {
		if (tv->last_tv_valid) {
			if (tp->tv_sec < tv->last_tv.tv_sec)
				tp->tv_sec = tv->last_tv.tv_sec;
			else if (tv->last_tv.tv_sec == tp->tv_sec &&
				 tp->tv_usec < tv->last_tv.tv_usec)
				tp->tv_usec = tv->last_tv.tv_usec;
		}
		tv->last_tv_valid = 1;
		memcpy(&tv->last_tv, tp, sizeof(*tp));
	}
}

#ifdef ARCH_HAVE_CPU_CLOCK
static unsigned long get_cycles_per_usec(void)
{
	struct timespec ts;
	struct timeval s, e;
	unsigned long long c_s, c_e;

	fill_clock_gettime(&ts);
	s.tv_sec = ts.tv_sec;
	s.tv_usec = ts.tv_nsec / 1000;

	c_s = get_cpu_clock();
	do {
		unsigned long long elapsed;

		fill_clock_gettime(&ts);
		e.tv_sec = ts.tv_sec;
		e.tv_usec = ts.tv_nsec / 1000;

		elapsed = utime_since(&s, &e);
		if (elapsed >= 1280) {
			c_e = get_cpu_clock();
			break;
		}
	} while (1);

	return (c_e - c_s + 127) >> 7;
}

#define NR_TIME_ITERS	50

static void calibrate_cpu_clock(void)
{
	double delta, mean, S;
	unsigned long avg, cycles[NR_TIME_ITERS];
	int i, samples;

	cycles[0] = get_cycles_per_usec();
	S = delta = mean = 0.0;
	for (i = 0; i < NR_TIME_ITERS; i++) {
		cycles[i] = get_cycles_per_usec();
		delta = cycles[i] - mean;
		if (delta) {
			mean += delta / (i + 1.0);
			S += delta * (cycles[i] - mean);
		}
	}

	S = sqrt(S / (NR_TIME_ITERS - 1.0));

	samples = avg = 0;
	for (i = 0; i < NR_TIME_ITERS; i++) {
		double this = cycles[i];

		if ((fmax(this, mean) - fmin(this, mean)) > S)
			continue;
		samples++;
		avg += this;
	}

	S /= (double) NR_TIME_ITERS;
	mean /= 10.0;

	for (i = 0; i < NR_TIME_ITERS; i++)
		dprint(FD_TIME, "cycles[%d]=%lu\n", i, cycles[i] / 10);

	avg /= samples;
	avg = (avg + 9) / 10;
	dprint(FD_TIME, "avg: %lu\n", avg);
	dprint(FD_TIME, "mean=%f, S=%f\n", mean, S);

	cycles_per_usec = avg;
}
#else
static void calibrate_cpu_clock(void)
{
}
#endif

void fio_local_clock_init(int is_thread)
{
	struct tv_valid *t;

	t = calloc(sizeof(*t), 1);
	if (pthread_setspecific(tv_tls_key, t))
		log_err("fio: can't set TLS key\n");
}

static void kill_tv_tls_key(void *data)
{
	free(data);
}

void fio_clock_init(void)
{
	if (fio_clock_source == fio_clock_source_inited)
		return;

	if (pthread_key_create(&tv_tls_key, kill_tv_tls_key))
		log_err("fio: can't create TLS key\n");

	fio_clock_source_inited = fio_clock_source;
	calibrate_cpu_clock();

	/*
	 * If the arch sets tsc_reliable != 0, then it must be good enough
	 * to use as THE clock source. For x86 CPUs, this means the TSC
	 * runs at a constant rate and is synced across CPU cores.
	 */
	if (tsc_reliable) {
		if (!fio_clock_source_set)
			fio_clock_source = CS_CPUCLOCK;
	} else if (fio_clock_source == CS_CPUCLOCK)
		log_info("fio: clocksource=cpu may not be reliable\n");
}

unsigned long long utime_since(struct timeval *s, struct timeval *e)
{
	long sec, usec;
	unsigned long long ret;

	sec = e->tv_sec - s->tv_sec;
	usec = e->tv_usec - s->tv_usec;
	if (sec > 0 && usec < 0) {
		sec--;
		usec += 1000000;
	}

	/*
	 * time warp bug on some kernels?
	 */
	if (sec < 0 || (sec == 0 && usec < 0))
		return 0;

	ret = sec * 1000000ULL + usec;

	return ret;
}

unsigned long long utime_since_now(struct timeval *s)
{
	struct timeval t;

	fio_gettime(&t, NULL);
	return utime_since(s, &t);
}

unsigned long mtime_since(struct timeval *s, struct timeval *e)
{
	long sec, usec, ret;

	sec = e->tv_sec - s->tv_sec;
	usec = e->tv_usec - s->tv_usec;
	if (sec > 0 && usec < 0) {
		sec--;
		usec += 1000000;
	}

	if (sec < 0 || (sec == 0 && usec < 0))
		return 0;

	sec *= 1000UL;
	usec /= 1000UL;
	ret = sec + usec;

	return ret;
}

unsigned long mtime_since_now(struct timeval *s)
{
	struct timeval t;
	void *p = __builtin_return_address(0);

	fio_gettime(&t, p);
	return mtime_since(s, &t);
}

unsigned long time_since_now(struct timeval *s)
{
	return mtime_since_now(s) / 1000;
}

#if defined(FIO_HAVE_CPU_AFFINITY) && defined(ARCH_HAVE_CPU_CLOCK)

#define CLOCK_ENTRIES	100000

struct clock_entry {
	unsigned long seq;
	unsigned long tsc;
	unsigned long cpu;
};

struct clock_thread {
	pthread_t thread;
	int cpu;
	pthread_mutex_t lock;
	pthread_mutex_t started;
	uint64_t *seq;
	struct clock_entry *entries;
};

static inline uint64_t atomic64_inc_return(uint64_t *seq)
{
	return 1 + __sync_fetch_and_add(seq, 1);
}

static void *clock_thread_fn(void *data)
{
	struct clock_thread *t = data;
	struct clock_entry *c;
	os_cpu_mask_t cpu_mask;
	int i;

	memset(&cpu_mask, 0, sizeof(cpu_mask));
	fio_cpu_set(&cpu_mask, t->cpu);

	if (fio_setaffinity(gettid(), cpu_mask) == -1) {
		log_err("clock setaffinity failed\n");
		return (void *) 1;
	}

	pthread_mutex_lock(&t->lock);
	pthread_mutex_unlock(&t->started);

	c = &t->entries[0];
	for (i = 0; i < CLOCK_ENTRIES; i++, c++) {
		uint64_t seq, tsc;

		c->cpu = t->cpu;
		do {
			seq = atomic64_inc_return(t->seq);
			tsc = get_cpu_clock();
		} while (seq != *t->seq);

		c->seq = seq;
		c->tsc = tsc;
	}

	log_info("cs: cpu%3d: %lu clocks seen\n", t->cpu, t->entries[CLOCK_ENTRIES - 1].tsc - t->entries[0].tsc);
	return NULL;
}

static int clock_cmp(const void *p1, const void *p2)
{
	const struct clock_entry *c1 = p1;
	const struct clock_entry *c2 = p2;

	if (c1->seq == c2->seq)
		log_err("cs: bug in atomic sequence!\n");

	return c1->seq - c2->seq;
}

int fio_monotonic_clocktest(void)
{
	struct clock_thread *threads;
	unsigned int nr_cpus = cpus_online();
	struct clock_entry *entries;
	unsigned long tentries, failed;
	uint64_t seq = 0;
	int i;

	fio_debug |= 1U << FD_TIME;
	calibrate_cpu_clock();
	fio_debug &= ~(1U << FD_TIME);

	threads = malloc(nr_cpus * sizeof(struct clock_thread));
	tentries = CLOCK_ENTRIES * nr_cpus;
	entries = malloc(tentries * sizeof(struct clock_entry));

	log_info("cs: Testing %u CPUs\n", nr_cpus);

	for (i = 0; i < nr_cpus; i++) {
		struct clock_thread *t = &threads[i];

		t->cpu = i;
		t->seq = &seq;
		t->entries = &entries[i * CLOCK_ENTRIES];
		pthread_mutex_init(&t->lock, NULL);
		pthread_mutex_init(&t->started, NULL);
		pthread_mutex_lock(&t->lock);
		pthread_create(&t->thread, NULL, clock_thread_fn, t);
	}

	for (i = 0; i < nr_cpus; i++) {
		struct clock_thread *t = &threads[i];

		pthread_mutex_lock(&t->started);
	}

	for (i = 0; i < nr_cpus; i++) {
		struct clock_thread *t = &threads[i];

		pthread_mutex_unlock(&t->lock);
	}

	for (failed = i = 0; i < nr_cpus; i++) {
		struct clock_thread *t = &threads[i];
		void *ret;

		pthread_join(t->thread, &ret);
		if (ret)
			failed++;
	}
	free(threads);

	if (failed) {
		log_err("Clocksource test: %u threads failed\n", failed);
		goto err;
	}

	qsort(entries, tentries, sizeof(struct clock_entry), clock_cmp);

	for (failed = i = 0; i < tentries; i++) {
		struct clock_entry *prev, *this = &entries[i];

		if (!i) {
			prev = this;
			continue;
		}

		if (prev->tsc > this->tsc) {
			uint64_t diff = prev->tsc - this->tsc;

			log_info("cs: CPU clock mismatch (diff=%lu):\n", diff);
			log_info("\t CPU%3lu: TSC=%lu, SEQ=%lu\n", prev->cpu, prev->tsc, prev->seq);
			log_info("\t CPU%3lu: TSC=%lu, SEQ=%lu\n", this->cpu, this->tsc, this->seq);
			failed++;
		}

		prev = this;
	}

	if (failed)
		log_info("cs: Failed: %lu\n", failed);
	else
		log_info("cs: Pass!\n");

err:
	free(entries);
	return !!failed;
}

#else /* defined(FIO_HAVE_CPU_AFFINITY) && defined(ARCH_HAVE_CPU_CLOCK) */

int fio_monotonic_clocktest(void)
{
	log_info("cs: current platform does not support CPU clocks\n");
	return 0;
}

#endif
Commit	Line	Data
02bcaa8c	1	/*
f5cc024a	2	* Clock functions
02bcaa8c	3	*/
f5cc024a	4
02bcaa8c	5	#include <unistd.h>
c223da83	6	#include <math.h>
02bcaa8c	7	#include <sys/time.h>
03e20d68	8	#include <time.h>
02bcaa8c JA	9
02bcaa8c JA	10	#include "fio.h"
be4ecfdf	11	#include "smalloc.h"
02bcaa8c JA	12
02bcaa8c JA	13	#include "hash.h"
7d11f871	14	#include "os/os.h"
02bcaa8c	15
09a32402	16	#ifdef ARCH_HAVE_CPU_CLOCK
c223da83	17	static unsigned long cycles_per_usec;
fa80feae	18	int tsc_reliable = 0;
09a32402	19	#endif
5d879392 JA	20
	21	struct tv_valid {
	22	struct timeval last_tv;
	23	int last_tv_valid;
93f0b09a	24	unsigned long last_cycles;
5d879392 JA	25	};
5d879392 JA	26	static pthread_key_t tv_tls_key;
02bcaa8c	27
16de1bf9	28	enum fio_cs fio_clock_source = FIO_PREFERRED_CLOCK_SOURCE;
fa80feae	29	int fio_clock_source_set = 0;
01423eae	30	enum fio_cs fio_clock_source_inited = CS_INVAL;
c223da83	31
02bcaa8c JA	32	#ifdef FIO_DEBUG_TIME
	33
	34	#define HASH_BITS 8
	35	#define HASH_SIZE (1 << HASH_BITS)
	36
01743ee1	37	static struct flist_head hash[HASH_SIZE];
02bcaa8c JA	38	static int gtod_inited;
	39
	40	struct gtod_log {
01743ee1	41	struct flist_head list;
02bcaa8c JA	42	void *caller;
	43	unsigned long calls;
	44	};
	45
	46	static struct gtod_log find_hash(void caller)
	47	{
	48	unsigned long h = hash_ptr(caller, HASH_BITS);
01743ee1	49	struct flist_head *entry;
02bcaa8c	50
01743ee1 JA	51	flist_for_each(entry, &hash[h]) {
	52	struct gtod_log *log = flist_entry(entry, struct gtod_log,
	53	list);
02bcaa8c JA	54
	55	if (log->caller == caller)
	56	return log;
	57	}
	58
	59	return NULL;
	60	}
	61
	62	static struct gtod_log find_log(void caller)
	63	{
	64	struct gtod_log *log = find_hash(caller);
	65
	66	if (!log) {
	67	unsigned long h;
	68
	69	log = malloc(sizeof(*log));
01743ee1	70	INIT_FLIST_HEAD(&log->list);
02bcaa8c JA	71	log->caller = caller;
	72	log->calls = 0;
	73
	74	h = hash_ptr(caller, HASH_BITS);
01743ee1	75	flist_add_tail(&log->list, &hash[h]);
02bcaa8c JA	76	}
	77
	78	return log;
	79	}
	80
	81	static void gtod_log_caller(void *caller)
	82	{
	83	if (gtod_inited) {
	84	struct gtod_log *log = find_log(caller);
	85
	86	log->calls++;
	87	}
	88	}
	89
	90	static void fio_exit fio_dump_gtod(void)
	91	{
	92	unsigned long total_calls = 0;
	93	int i;
	94
	95	for (i = 0; i < HASH_SIZE; i++) {
01743ee1	96	struct flist_head *entry;
02bcaa8c JA	97	struct gtod_log *log;
02bcaa8c JA	98
01743ee1 JA	99	flist_for_each(entry, &hash[i]) {
01743ee1 JA	100	log = flist_entry(entry, struct gtod_log, list);
02bcaa8c	101
5ec10eaa JA	102	printf("function %p, calls %lu\n", log->caller,
5ec10eaa JA	103	log->calls);
02bcaa8c JA	104	total_calls += log->calls;
	105	}
	106	}
	107
	108	printf("Total %lu gettimeofday\n", total_calls);
	109	}
	110
	111	static void fio_init gtod_init(void)
	112	{
	113	int i;
	114
	115	for (i = 0; i < HASH_SIZE; i++)
01743ee1	116	INIT_FLIST_HEAD(&hash[i]);
02bcaa8c JA	117
	118	gtod_inited = 1;
	119	}
	120
	121	#endif /* FIO_DEBUG_TIME */
	122
9ff1c070 JA	123	static int fill_clock_gettime(struct timespec *ts)
	124	{
	125	#ifdef FIO_HAVE_CLOCK_MONOTONIC
	126	return clock_gettime(CLOCK_MONOTONIC, ts);
	127	#else
	128	return clock_gettime(CLOCK_REALTIME, ts);
	129	#endif
	130	}
	131
1e97cce9	132	#ifdef FIO_DEBUG_TIME
02bcaa8c	133	void fio_gettime(struct timeval tp, void caller)
1e97cce9 JA	134	#else
	135	void fio_gettime(struct timeval tp, void fio_unused caller)
	136	#endif
02bcaa8c	137	{
93f0b09a	138	struct tv_valid *tv;
5d879392	139
02bcaa8c JA	140	#ifdef FIO_DEBUG_TIME
	141	if (!caller)
	142	caller = __builtin_return_address(0);
	143
	144	gtod_log_caller(caller);
02bcaa8c	145	#endif
be4ecfdf JA	146	if (fio_tv) {
	147	memcpy(tp, fio_tv, sizeof(*tp));
	148	return;
c223da83 JA	149	}
c223da83 JA	150
93f0b09a JA	151	tv = pthread_getspecific(tv_tls_key);
93f0b09a JA	152
c223da83 JA	153	switch (fio_clock_source) {
c223da83 JA	154	case CS_GTOD:
02bcaa8c	155	gettimeofday(tp, NULL);
c223da83 JA	156	break;
c223da83 JA	157	case CS_CGETTIME: {
02bcaa8c JA	158	struct timespec ts;
02bcaa8c JA	159
9ff1c070	160	if (fill_clock_gettime(&ts) < 0) {
c223da83 JA	161	log_err("fio: clock_gettime fails\n");
c223da83 JA	162	assert(0);
02bcaa8c JA	163	}
	164
	165	tp->tv_sec = ts.tv_sec;
	166	tp->tv_usec = ts.tv_nsec / 1000;
c223da83 JA	167	break;
	168	}
	169	#ifdef ARCH_HAVE_CPU_CLOCK
	170	case CS_CPUCLOCK: {
	171	unsigned long long usecs, t;
	172
	173	t = get_cpu_clock();
93f0b09a	174	if (tv && t < tv->last_cycles) {
c223da83	175	dprint(FD_TIME, "CPU clock going back in time\n");
93f0b09a JA	176	t = tv->last_cycles;
	177	} else if (tv)
	178	tv->last_cycles = t;
c223da83 JA	179
	180	usecs = t / cycles_per_usec;
	181	tp->tv_sec = usecs / 1000000;
	182	tp->tv_usec = usecs % 1000000;
c223da83 JA	183	break;
	184	}
	185	#endif
	186	default:
	187	log_err("fio: invalid clock source %d\n", fio_clock_source);
	188	break;
02bcaa8c	189	}
3e488920 JA	190
	191	/*
	192	* If Linux is using the tsc clock on non-synced processors,
	193	* sometimes time can appear to drift backwards. Fix that up.
	194	*/
93f0b09a JA	195	if (tv) {
	196	if (tv->last_tv_valid) {
	197	if (tp->tv_sec < tv->last_tv.tv_sec)
	198	tp->tv_sec = tv->last_tv.tv_sec;
	199	else if (tv->last_tv.tv_sec == tp->tv_sec &&
	200	tp->tv_usec < tv->last_tv.tv_usec)
	201	tp->tv_usec = tv->last_tv.tv_usec;
5d879392	202	}
93f0b09a JA	203	tv->last_tv_valid = 1;
93f0b09a JA	204	memcpy(&tv->last_tv, tp, sizeof(*tp));
3e488920	205	}
02bcaa8c	206	}
be4ecfdf	207
09a32402	208	#ifdef ARCH_HAVE_CPU_CLOCK
c223da83 JA	209	static unsigned long get_cycles_per_usec(void)
c223da83 JA	210	{
9ff1c070	211	struct timespec ts;
c223da83 JA	212	struct timeval s, e;
	213	unsigned long long c_s, c_e;
	214
9ff1c070 JA	215	fill_clock_gettime(&ts);
	216	s.tv_sec = ts.tv_sec;
	217	s.tv_usec = ts.tv_nsec / 1000;
	218
c223da83 JA	219	c_s = get_cpu_clock();
	220	do {
	221	unsigned long long elapsed;
	222
9ff1c070 JA	223	fill_clock_gettime(&ts);
	224	e.tv_sec = ts.tv_sec;
	225	e.tv_usec = ts.tv_nsec / 1000;
	226
c223da83	227	elapsed = utime_since(&s, &e);
486332e5	228	if (elapsed >= 1280) {
c223da83 JA	229	c_e = get_cpu_clock();
	230	break;
	231	}
	232	} while (1);
	233
c011000d	234	return (c_e - c_s + 127) >> 7;
c223da83 JA	235	}
c223da83 JA	236
fa80feae JA	237	#define NR_TIME_ITERS 50
fa80feae JA	238
09a32402	239	static void calibrate_cpu_clock(void)
c223da83 JA	240	{
c223da83 JA	241	double delta, mean, S;
fa80feae	242	unsigned long avg, cycles[NR_TIME_ITERS];
c223da83 JA	243	int i, samples;
c223da83 JA	244
c223da83 JA	245	cycles[0] = get_cycles_per_usec();
c223da83 JA	246	S = delta = mean = 0.0;
fa80feae	247	for (i = 0; i < NR_TIME_ITERS; i++) {
c223da83 JA	248	cycles[i] = get_cycles_per_usec();
	249	delta = cycles[i] - mean;
	250	if (delta) {
	251	mean += delta / (i + 1.0);
	252	S += delta * (cycles[i] - mean);
	253	}
	254	}
	255
fa80feae	256	S = sqrt(S / (NR_TIME_ITERS - 1.0));
c223da83 JA	257
c223da83 JA	258	samples = avg = 0;
fa80feae	259	for (i = 0; i < NR_TIME_ITERS; i++) {
c223da83 JA	260	double this = cycles[i];
c223da83 JA	261
03e20d68	262	if ((fmax(this, mean) - fmin(this, mean)) > S)
c223da83 JA	263	continue;
	264	samples++;
	265	avg += this;
	266	}
	267
fa80feae	268	S /= (double) NR_TIME_ITERS;
89db727d	269	mean /= 10.0;
c223da83	270
fa80feae	271	for (i = 0; i < NR_TIME_ITERS; i++)
c223da83 JA	272	dprint(FD_TIME, "cycles[%d]=%lu\n", i, cycles[i] / 10);
c223da83 JA	273
d7abad3d JA	274	avg /= samples;
d7abad3d JA	275	avg = (avg + 9) / 10;
c223da83 JA	276	dprint(FD_TIME, "avg: %lu\n", avg);
	277	dprint(FD_TIME, "mean=%f, S=%f\n", mean, S);
	278
	279	cycles_per_usec = avg;
09a32402 JA	280	}
	281	#else
	282	static void calibrate_cpu_clock(void)
	283	{
	284	}
	285	#endif
	286
5d879392 JA	287	void fio_local_clock_init(int is_thread)
	288	{
	289	struct tv_valid *t;
	290
	291	t = calloc(sizeof(*t), 1);
	292	if (pthread_setspecific(tv_tls_key, t))
	293	log_err("fio: can't set TLS key\n");
	294	}
	295
	296	static void kill_tv_tls_key(void *data)
	297	{
	298	free(data);
	299	}
	300
09a32402 JA	301	void fio_clock_init(void)
09a32402 JA	302	{
01423eae JA	303	if (fio_clock_source == fio_clock_source_inited)
	304	return;
	305
5d879392 JA	306	if (pthread_key_create(&tv_tls_key, kill_tv_tls_key))
	307	log_err("fio: can't create TLS key\n");
	308
01423eae	309	fio_clock_source_inited = fio_clock_source;
09a32402	310	calibrate_cpu_clock();
fa80feae JA	311
	312	/*
	313	* If the arch sets tsc_reliable != 0, then it must be good enough
	314	* to use as THE clock source. For x86 CPUs, this means the TSC
	315	* runs at a constant rate and is synced across CPU cores.
	316	*/
	317	if (tsc_reliable) {
	318	if (!fio_clock_source_set)
	319	fio_clock_source = CS_CPUCLOCK;
	320	} else if (fio_clock_source == CS_CPUCLOCK)
	321	log_info("fio: clocksource=cpu may not be reliable\n");
c223da83 JA	322	}
c223da83 JA	323
39ab7da2	324	unsigned long long utime_since(struct timeval s, struct timeval e)
be4ecfdf	325	{
39ab7da2 JA	326	long sec, usec;
	327	unsigned long long ret;
	328
	329	sec = e->tv_sec - s->tv_sec;
	330	usec = e->tv_usec - s->tv_usec;
	331	if (sec > 0 && usec < 0) {
	332	sec--;
	333	usec += 1000000;
	334	}
	335
	336	/*
	337	* time warp bug on some kernels?
	338	*/
	339	if (sec < 0 \|\| (sec == 0 && usec < 0))
	340	return 0;
	341
	342	ret = sec * 1000000ULL + usec;
	343
	344	return ret;
be4ecfdf JA	345	}
be4ecfdf JA	346
39ab7da2	347	unsigned long long utime_since_now(struct timeval *s)
be4ecfdf	348	{
39ab7da2 JA	349	struct timeval t;
	350
	351	fio_gettime(&t, NULL);
	352	return utime_since(s, &t);
be4ecfdf	353	}
783a3eb1	354
39ab7da2	355	unsigned long mtime_since(struct timeval s, struct timeval e)
783a3eb1	356	{
39ab7da2	357	long sec, usec, ret;
783a3eb1	358
39ab7da2 JA	359	sec = e->tv_sec - s->tv_sec;
	360	usec = e->tv_usec - s->tv_usec;
	361	if (sec > 0 && usec < 0) {
	362	sec--;
	363	usec += 1000000;
783a3eb1 JA	364	}
783a3eb1 JA	365
39ab7da2 JA	366	if (sec < 0 \|\| (sec == 0 && usec < 0))
	367	return 0;
	368
	369	sec *= 1000UL;
	370	usec /= 1000UL;
	371	ret = sec + usec;
	372
	373	return ret;
783a3eb1 JA	374	}
783a3eb1 JA	375
39ab7da2	376	unsigned long mtime_since_now(struct timeval *s)
783a3eb1	377	{
39ab7da2 JA	378	struct timeval t;
39ab7da2 JA	379	void *p = __builtin_return_address(0);
783a3eb1	380
39ab7da2 JA	381	fio_gettime(&t, p);
	382	return mtime_since(s, &t);
	383	}
783a3eb1	384
39ab7da2 JA	385	unsigned long time_since_now(struct timeval *s)
	386	{
	387	return mtime_since_now(s) / 1000;
783a3eb1	388	}
7d11f871 JA	389
	390	#if defined(FIO_HAVE_CPU_AFFINITY) && defined(ARCH_HAVE_CPU_CLOCK)
	391
	392	#define CLOCK_ENTRIES 100000
	393
	394	struct clock_entry {
	395	unsigned long seq;
	396	unsigned long tsc;
	397	unsigned long cpu;
	398	};
	399
	400	struct clock_thread {
	401	pthread_t thread;
	402	int cpu;
	403	pthread_mutex_t lock;
	404	pthread_mutex_t started;
	405	uint64_t *seq;
	406	struct clock_entry *entries;
	407	};
	408
	409	static inline uint64_t atomic64_inc_return(uint64_t *seq)
	410	{
	411	return 1 + __sync_fetch_and_add(seq, 1);
	412	}
	413
	414	static void clock_thread_fn(void data)
	415	{
	416	struct clock_thread *t = data;
	417	struct clock_entry *c;
	418	os_cpu_mask_t cpu_mask;
	419	int i;
	420
	421	memset(&cpu_mask, 0, sizeof(cpu_mask));
	422	fio_cpu_set(&cpu_mask, t->cpu);
	423
	424	if (fio_setaffinity(gettid(), cpu_mask) == -1) {
	425	log_err("clock setaffinity failed\n");
	426	return (void *) 1;
	427	}
	428
7d11f871	429	pthread_mutex_lock(&t->lock);
b9b3498e	430	pthread_mutex_unlock(&t->started);
7d11f871 JA	431
	432	c = &t->entries[0];
	433	for (i = 0; i < CLOCK_ENTRIES; i++, c++) {
	434	uint64_t seq, tsc;
	435
	436	c->cpu = t->cpu;
	437	do {
	438	seq = atomic64_inc_return(t->seq);
	439	tsc = get_cpu_clock();
	440	} while (seq != *t->seq);
	441
	442	c->seq = seq;
	443	c->tsc = tsc;
	444	}
	445
	446	log_info("cs: cpu%3d: %lu clocks seen\n", t->cpu, t->entries[CLOCK_ENTRIES - 1].tsc - t->entries[0].tsc);
	447	return NULL;
	448	}
	449
	450	static int clock_cmp(const void p1, const void p2)
	451	{
	452	const struct clock_entry *c1 = p1;
	453	const struct clock_entry *c2 = p2;
	454
b9b3498e JA	455	if (c1->seq == c2->seq)
	456	log_err("cs: bug in atomic sequence!\n");
	457
7d11f871 JA	458	return c1->seq - c2->seq;
	459	}
	460
	461	int fio_monotonic_clocktest(void)
	462	{
	463	struct clock_thread *threads;
	464	unsigned int nr_cpus = cpus_online();
	465	struct clock_entry *entries;
	466	unsigned long tentries, failed;
	467	uint64_t seq = 0;
	468	int i;
	469
4f1d43c2 JA	470	fio_debug \|= 1U << FD_TIME;
	471	calibrate_cpu_clock();
	472	fio_debug &= ~(1U << FD_TIME);
	473
7d11f871 JA	474	threads = malloc(nr_cpus * sizeof(struct clock_thread));
	475	tentries = CLOCK_ENTRIES * nr_cpus;
	476	entries = malloc(tentries * sizeof(struct clock_entry));
	477
	478	log_info("cs: Testing %u CPUs\n", nr_cpus);
	479
	480	for (i = 0; i < nr_cpus; i++) {
	481	struct clock_thread *t = &threads[i];
	482
	483	t->cpu = i;
	484	t->seq = &seq;
	485	t->entries = &entries[i * CLOCK_ENTRIES];
	486	pthread_mutex_init(&t->lock, NULL);
	487	pthread_mutex_init(&t->started, NULL);
	488	pthread_mutex_lock(&t->lock);
	489	pthread_create(&t->thread, NULL, clock_thread_fn, t);
	490	}
	491
	492	for (i = 0; i < nr_cpus; i++) {
	493	struct clock_thread *t = &threads[i];
	494
	495	pthread_mutex_lock(&t->started);
	496	}
	497
	498	for (i = 0; i < nr_cpus; i++) {
	499	struct clock_thread *t = &threads[i];
	500
	501	pthread_mutex_unlock(&t->lock);
	502	}
	503
	504	for (failed = i = 0; i < nr_cpus; i++) {
	505	struct clock_thread *t = &threads[i];
	506	void *ret;
	507
	508	pthread_join(t->thread, &ret);
	509	if (ret)
	510	failed++;
	511	}
	512	free(threads);
	513
	514	if (failed) {
	515	log_err("Clocksource test: %u threads failed\n", failed);
	516	goto err;
	517	}
	518
	519	qsort(entries, tentries, sizeof(struct clock_entry), clock_cmp);
	520
	521	for (failed = i = 0; i < tentries; i++) {
	522	struct clock_entry prev, this = &entries[i];
	523
	524	if (!i) {
	525	prev = this;
	526	continue;
	527	}
	528
	529	if (prev->tsc > this->tsc) {
	530	uint64_t diff = prev->tsc - this->tsc;
	531
	532	log_info("cs: CPU clock mismatch (diff=%lu):\n", diff);
	533	log_info("\t CPU%3lu: TSC=%lu, SEQ=%lu\n", prev->cpu, prev->tsc, prev->seq);
	534	log_info("\t CPU%3lu: TSC=%lu, SEQ=%lu\n", this->cpu, this->tsc, this->seq);
	535	failed++;
	536	}
	537
538	prev = this;
539	}
540
541	if (failed)
542	log_info("cs: Failed: %lu\n", failed);
543	else
544	log_info("cs: Pass!\n");
545
546	err:
547	free(entries);
548	return !!failed;
549	}
550
551	#else /* defined(FIO_HAVE_CPU_AFFINITY) && defined(ARCH_HAVE_CPU_CLOCK) */
552
553	int fio_monotonic_clocktest(void)
554	{
555	log_info("cs: current platform does not support CPU clocks\n");
556	return 0;
557	}
558
559	#endif