[fio.git] / workqueue.c

/*
 * Rated submission helpers
 *
 * Copyright (C) 2015 Jens Axboe <axboe@kernel.dk>
 *
 */
#include <unistd.h>

#include "fio.h"
#include "flist.h"
#include "workqueue.h"
#include "lib/getrusage.h"

struct submit_worker {
	pthread_t thread;
	pthread_mutex_t lock;
	pthread_cond_t cond;
	struct flist_head work_list;
	unsigned int flags;
	unsigned int index;
	uint64_t seq;
	struct workqueue *wq;
	struct thread_data td;
};

enum {
	SW_F_IDLE	= 1 << 0,
	SW_F_RUNNING	= 1 << 1,
	SW_F_EXIT	= 1 << 2,
	SW_F_EXITED	= 1 << 3,
	SW_F_ACCOUNTED	= 1 << 4,
	SW_F_ERROR	= 1 << 5,
};

static struct submit_worker *__get_submit_worker(struct workqueue *wq,
						 unsigned int start,
						 unsigned int end,
						 struct submit_worker **best)
{
	struct submit_worker *sw = NULL;

	while (start <= end) {
		sw = &wq->workers[start];
		if (sw->flags & SW_F_IDLE)
			return sw;
		if (!(*best) || sw->seq < (*best)->seq)
			*best = sw;
		start++;
	}

	return NULL;
}

static struct submit_worker *get_submit_worker(struct workqueue *wq)
{
	unsigned int next = wq->next_free_worker;
	struct submit_worker *sw, *best = NULL;

	assert(next < wq->max_workers);

	sw = __get_submit_worker(wq, next, wq->max_workers - 1, &best);
	if (!sw && next)
		sw = __get_submit_worker(wq, 0, next - 1, &best);

	/*
	 * No truly idle found, use best match
	 */
	if (!sw)
		sw = best;

	if (sw->index == wq->next_free_worker) {
		if (sw->index + 1 < wq->max_workers)
			wq->next_free_worker = sw->index + 1;
		else
			wq->next_free_worker = 0;
	}

	return sw;
}

static bool all_sw_idle(struct workqueue *wq)
{
	int i;

	for (i = 0; i < wq->max_workers; i++) {
		struct submit_worker *sw = &wq->workers[i];

		if (!(sw->flags & SW_F_IDLE))
			return false;
	}

	return true;
}

/*
 * Must be serialized wrt workqueue_enqueue() by caller
 */
void workqueue_flush(struct workqueue *wq)
{
	wq->wake_idle = 1;

	while (!all_sw_idle(wq)) {
		pthread_mutex_lock(&wq->flush_lock);
		pthread_cond_wait(&wq->flush_cond, &wq->flush_lock);
		pthread_mutex_unlock(&wq->flush_lock);
	}

	wq->wake_idle = 0;
}

/*
 * Must be serialized by caller. Returns true for queued, false for busy.
 */
bool workqueue_enqueue(struct workqueue *wq, struct workqueue_work *work)
{
	struct submit_worker *sw;

	sw = get_submit_worker(wq);
	if (sw) {
		pthread_mutex_lock(&sw->lock);
		flist_add_tail(&work->list, &sw->work_list);
		sw->seq = ++wq->work_seq;
		sw->flags &= ~SW_F_IDLE;
		pthread_mutex_unlock(&sw->lock);

		pthread_cond_signal(&sw->cond);
		return true;
	}

	return false;
}

static void handle_list(struct submit_worker *sw, struct flist_head *list)
{
	struct workqueue *wq = sw->wq;
	struct workqueue_work *work;

	while (!flist_empty(list)) {
		work = flist_first_entry(list, struct workqueue_work, list);
		flist_del_init(&work->list);
		wq->fn(&sw->td, work);
	}
}

static int init_submit_worker(struct submit_worker *sw)
{
	struct thread_data *parent = sw->wq->td;
	struct thread_data *td = &sw->td;
	int fio_unused ret;

	memcpy(&td->o, &parent->o, sizeof(td->o));
	memcpy(&td->ts, &parent->ts, sizeof(td->ts));
	td->o.uid = td->o.gid = -1U;
	dup_files(td, parent);
	td->eo = parent->eo;
	fio_options_mem_dupe(td);

	if (ioengine_load(td))
		goto err;

	if (td->o.odirect)
		td->io_ops->flags |= FIO_RAWIO;

	td->pid = gettid();

	INIT_FLIST_HEAD(&td->io_log_list);
	INIT_FLIST_HEAD(&td->io_hist_list);
	INIT_FLIST_HEAD(&td->verify_list);
	INIT_FLIST_HEAD(&td->trim_list);
	INIT_FLIST_HEAD(&td->next_rand_list);
	td->io_hist_tree = RB_ROOT;

	td->o.iodepth = 1;
	if (td_io_init(td))
		goto err_io_init;

	fio_gettime(&td->epoch, NULL);
	fio_getrusage(&td->ru_start);
	clear_io_state(td, 1);

	td_set_runstate(td, TD_RUNNING);
	td->flags |= TD_F_CHILD;
	td->parent = parent;
	return 0;

err_io_init:
	close_ioengine(td);
err:
	return 1;
}

#ifdef CONFIG_SFAA
static void sum_val(uint64_t *dst, uint64_t *src)
{
	if (*src) {
		__sync_fetch_and_add(dst, *src);
		*src = 0;
	}
}
#else
static void sum_val(uint64_t *dst, uint64_t *src)
{
	if (*src) {
		*dst += *src;
		*src = 0;
	}
}
#endif

static void pthread_double_unlock(pthread_mutex_t *lock1,
				  pthread_mutex_t *lock2)
{
#ifndef CONFIG_SFAA
	pthread_mutex_unlock(lock1);
	pthread_mutex_unlock(lock2);
#endif
}

static void pthread_double_lock(pthread_mutex_t *lock1, pthread_mutex_t *lock2)
{
#ifndef CONFIG_SFAA
	if (lock1 < lock2) {
		pthread_mutex_lock(lock1);
		pthread_mutex_lock(lock2);
	} else {
		pthread_mutex_lock(lock2);
		pthread_mutex_lock(lock1);
	}
#endif
}

static void sum_ddir(struct thread_data *dst, struct thread_data *src,
		     enum fio_ddir ddir)
{
	pthread_double_lock(&dst->io_wq.stat_lock, &src->io_wq.stat_lock);

	sum_val(&dst->io_bytes[ddir], &src->io_bytes[ddir]);
	sum_val(&dst->io_blocks[ddir], &src->io_blocks[ddir]);
	sum_val(&dst->this_io_blocks[ddir], &src->this_io_blocks[ddir]);
	sum_val(&dst->this_io_bytes[ddir], &src->this_io_bytes[ddir]);
	sum_val(&dst->bytes_done[ddir], &src->bytes_done[ddir]);

	pthread_double_unlock(&dst->io_wq.stat_lock, &src->io_wq.stat_lock);
}

static void update_accounting(struct submit_worker *sw)
{
	struct thread_data *src = &sw->td;
	struct thread_data *dst = sw->wq->td;

	if (td_read(src))
		sum_ddir(dst, src, DDIR_READ);
	if (td_write(src))
		sum_ddir(dst, src, DDIR_WRITE);
	if (td_trim(src))
		sum_ddir(dst, src, DDIR_TRIM);
}

static void *worker_thread(void *data)
{
	struct submit_worker *sw = data;
	struct workqueue *wq = sw->wq;
	struct thread_data *td = &sw->td;
	unsigned int eflags = 0, ret;
	FLIST_HEAD(local_list);

	ret = init_submit_worker(sw);
	pthread_mutex_lock(&sw->lock);
	sw->flags |= SW_F_RUNNING;
	if (ret)
		sw->flags |= SW_F_ERROR;
	pthread_mutex_unlock(&sw->lock);

	pthread_mutex_lock(&wq->flush_lock);
	pthread_cond_signal(&wq->flush_cond);
	pthread_mutex_unlock(&wq->flush_lock);

	if (sw->flags & SW_F_ERROR)
		goto done;

	while (1) {
		pthread_mutex_lock(&sw->lock);

		if (flist_empty(&sw->work_list)) {
			if (sw->flags & SW_F_EXIT) {
				pthread_mutex_unlock(&sw->lock);
				break;
			}

			if (td->io_u_queued || td->cur_depth ||
			    td->io_u_in_flight) {
				int ret;

				pthread_mutex_unlock(&sw->lock);
				ret = io_u_quiesce(td);
				if (ret > 0)
					td->cur_depth -= ret;
				pthread_mutex_lock(&sw->lock);
			}

			/*
			 * We dropped and reaquired the lock, check
			 * state again.
			 */
			if (!flist_empty(&sw->work_list))
				goto handle_work;

			if (sw->flags & SW_F_EXIT) {
				pthread_mutex_unlock(&sw->lock);
				break;
			} else if (!(sw->flags & SW_F_IDLE)) {
				sw->flags |= SW_F_IDLE;
				wq->next_free_worker = sw->index;
				if (wq->wake_idle)
					pthread_cond_signal(&wq->flush_cond);
			}
			update_accounting(sw);
			pthread_cond_wait(&sw->cond, &sw->lock);
		} else {
handle_work:
			flist_splice_init(&sw->work_list, &local_list);
		}
		pthread_mutex_unlock(&sw->lock);
		handle_list(sw, &local_list);
	}

	update_accounting(sw);

done:
	pthread_mutex_lock(&sw->lock);
	sw->flags |= (SW_F_EXITED | eflags);
	pthread_mutex_unlock(&sw->lock);
	return NULL;
}

static void free_worker(struct submit_worker *sw)
{
	struct thread_data *td = &sw->td;

	fio_options_free(td);
	close_and_free_files(td);
	if (td->io_ops)
		close_ioengine(td);
	td_set_runstate(td, TD_EXITED);

	pthread_cond_destroy(&sw->cond);
	pthread_mutex_destroy(&sw->lock);
}

static void shutdown_worker(struct submit_worker *sw, unsigned int *sum_cnt)
{
	struct thread_data *parent = sw->wq->td;

	pthread_join(sw->thread, NULL);
	(*sum_cnt)++;
	sum_thread_stats(&parent->ts, &sw->td.ts, *sum_cnt == 1);
	free_worker(sw);
}

void workqueue_exit(struct workqueue *wq)
{
	unsigned int shutdown, sum_cnt = 0;
	struct submit_worker *sw;
	int i;

	for (i = 0; i < wq->max_workers; i++) {
		sw = &wq->workers[i];

		pthread_mutex_lock(&sw->lock);
		sw->flags |= SW_F_EXIT;
		pthread_cond_signal(&sw->cond);
		pthread_mutex_unlock(&sw->lock);
	}

	do {
		shutdown = 0;
		for (i = 0; i < wq->max_workers; i++) {
			sw = &wq->workers[i];
			if (sw->flags & SW_F_ACCOUNTED)
				continue;
			pthread_mutex_lock(&sw->lock);
			sw->flags |= SW_F_ACCOUNTED;
			pthread_mutex_unlock(&sw->lock);
			shutdown_worker(sw, &sum_cnt);
			shutdown++;
		}
	} while (shutdown && shutdown != wq->max_workers);

	free(wq->workers);
	pthread_mutex_destroy(&wq->flush_lock);
	pthread_cond_destroy(&wq->flush_cond);
	pthread_mutex_destroy(&wq->stat_lock);
}

static int start_worker(struct workqueue *wq, unsigned int index)
{
	struct submit_worker *sw = &wq->workers[index];
	int ret;

	INIT_FLIST_HEAD(&sw->work_list);
	pthread_cond_init(&sw->cond, NULL);
	pthread_mutex_init(&sw->lock, NULL);
	sw->wq = wq;
	sw->index = index;

	ret = pthread_create(&sw->thread, NULL, worker_thread, sw);
	if (!ret) {
		pthread_mutex_lock(&sw->lock);
		sw->flags = SW_F_IDLE;
		pthread_mutex_unlock(&sw->lock);
		return 0;
	}

	free_worker(sw);
	return 1;
}

int workqueue_init(struct thread_data *td, struct workqueue *wq,
		   workqueue_fn *fn, unsigned max_pending)
{
	unsigned int running;
	int i, error;

	wq->max_workers = max_pending;
	wq->td = td;
	wq->fn = fn;
	wq->work_seq = 0;
	wq->next_free_worker = 0;
	pthread_cond_init(&wq->flush_cond, NULL);
	pthread_mutex_init(&wq->flush_lock, NULL);
	pthread_mutex_init(&wq->stat_lock, NULL);

	wq->workers = calloc(wq->max_workers, sizeof(struct submit_worker));

	for (i = 0; i < wq->max_workers; i++)
		if (start_worker(wq, i))
			break;

	wq->max_workers = i;
	if (!wq->max_workers)
		goto err;

	/*
	 * Wait for them all to be started and initialized
	 */
	error = 0;
	do {
		struct submit_worker *sw;

		running = 0;
		pthread_mutex_lock(&wq->flush_lock);
		for (i = 0; i < wq->max_workers; i++) {
			sw = &wq->workers[i];
			pthread_mutex_lock(&sw->lock);
			if (sw->flags & SW_F_RUNNING)
				running++;
			if (sw->flags & SW_F_ERROR)
				error++;
			pthread_mutex_unlock(&sw->lock);
		}

		if (error || running == wq->max_workers) {
			pthread_mutex_unlock(&wq->flush_lock);
			break;
		}

		pthread_cond_wait(&wq->flush_cond, &wq->flush_lock);
		pthread_mutex_unlock(&wq->flush_lock);
	} while (1);

	if (!error)
		return 0;

err:
	log_err("Can't create rate workqueue\n");
	td_verror(td, ESRCH, "workqueue_init");
	workqueue_exit(wq);
	return 1;
}
Commit	Line	Data
a9da8ab2 JA	1	/*
	2	* Rated submission helpers
	3	*
	4	* Copyright (C) 2015 Jens Axboe <axboe@kernel.dk>
	5	*
	6	*/
	7	#include <unistd.h>
	8
	9	#include "fio.h"
a9da8ab2 JA	10	#include "flist.h"
	11	#include "workqueue.h"
	12	#include "lib/getrusage.h"
	13
	14	struct submit_worker {
	15	pthread_t thread;
	16	pthread_mutex_t lock;
	17	pthread_cond_t cond;
	18	struct flist_head work_list;
	19	unsigned int flags;
	20	unsigned int index;
	21	uint64_t seq;
	22	struct workqueue *wq;
	23	struct thread_data td;
	24	};
	25
	26	enum {
	27	SW_F_IDLE = 1 << 0,
	28	SW_F_RUNNING = 1 << 1,
	29	SW_F_EXIT = 1 << 2,
	30	SW_F_EXITED = 1 << 3,
	31	SW_F_ACCOUNTED = 1 << 4,
	32	SW_F_ERROR = 1 << 5,
	33	};
	34
	35	static struct submit_worker __get_submit_worker(struct workqueue wq,
	36	unsigned int start,
	37	unsigned int end,
	38	struct submit_worker **best)
	39	{
	40	struct submit_worker *sw = NULL;
	41
	42	while (start <= end) {
	43	sw = &wq->workers[start];
	44	if (sw->flags & SW_F_IDLE)
	45	return sw;
	46	if (!(best) \|\| sw->seq < (best)->seq)
	47	*best = sw;
	48	start++;
	49	}
	50
	51	return NULL;
	52	}
	53
	54	static struct submit_worker get_submit_worker(struct workqueue wq)
	55	{
	56	unsigned int next = wq->next_free_worker;
	57	struct submit_worker sw, best = NULL;
	58
	59	assert(next < wq->max_workers);
	60
	61	sw = __get_submit_worker(wq, next, wq->max_workers - 1, &best);
	62	if (!sw && next)
	63	sw = __get_submit_worker(wq, 0, next - 1, &best);
	64
	65	/*
	66	* No truly idle found, use best match
	67	*/
	68	if (!sw)
	69	sw = best;
	70
	71	if (sw->index == wq->next_free_worker) {
	72	if (sw->index + 1 < wq->max_workers)
	73	wq->next_free_worker = sw->index + 1;
74	else
75	wq->next_free_worker = 0;
76	}
77
78	return sw;
79	}
80
1391052a	81	static bool all_sw_idle(struct workqueue *wq)
a9da8ab2 JA	82	{
	83	int i;
	84
	85	for (i = 0; i < wq->max_workers; i++) {
	86	struct submit_worker *sw = &wq->workers[i];
	87
	88	if (!(sw->flags & SW_F_IDLE))
1391052a	89	return false;
a9da8ab2 JA	90	}
a9da8ab2 JA	91
1391052a	92	return true;
a9da8ab2 JA	93	}
	94
	95	/*
	96	* Must be serialized wrt workqueue_enqueue() by caller
	97	*/
	98	void workqueue_flush(struct workqueue *wq)
	99	{
	100	wq->wake_idle = 1;
	101
	102	while (!all_sw_idle(wq)) {
	103	pthread_mutex_lock(&wq->flush_lock);
	104	pthread_cond_wait(&wq->flush_cond, &wq->flush_lock);
	105	pthread_mutex_unlock(&wq->flush_lock);
	106	}
	107
	108	wq->wake_idle = 0;
	109	}
	110
	111	/*
b07f6ad1	112	* Must be serialized by caller. Returns true for queued, false for busy.
a9da8ab2	113	*/
88271841	114	bool workqueue_enqueue(struct workqueue wq, struct workqueue_work work)
a9da8ab2 JA	115	{
	116	struct submit_worker *sw;
	117
	118	sw = get_submit_worker(wq);
	119	if (sw) {
a9da8ab2	120	pthread_mutex_lock(&sw->lock);
88271841	121	flist_add_tail(&work->list, &sw->work_list);
a9da8ab2 JA	122	sw->seq = ++wq->work_seq;
	123	sw->flags &= ~SW_F_IDLE;
	124	pthread_mutex_unlock(&sw->lock);
	125
	126	pthread_cond_signal(&sw->cond);
b07f6ad1	127	return true;
a9da8ab2 JA	128	}
a9da8ab2 JA	129
b07f6ad1	130	return false;
a9da8ab2 JA	131	}
	132
	133	static void handle_list(struct submit_worker sw, struct flist_head list)
	134	{
	135	struct workqueue *wq = sw->wq;
88271841	136	struct workqueue_work *work;
a9da8ab2 JA	137
a9da8ab2 JA	138	while (!flist_empty(list)) {
88271841 JA	139	work = flist_first_entry(list, struct workqueue_work, list);
	140	flist_del_init(&work->list);
	141	wq->fn(&sw->td, work);
a9da8ab2 JA	142	}
	143	}
	144
	145	static int init_submit_worker(struct submit_worker *sw)
	146	{
	147	struct thread_data *parent = sw->wq->td;
	148	struct thread_data *td = &sw->td;
	149	int fio_unused ret;
	150
	151	memcpy(&td->o, &parent->o, sizeof(td->o));
	152	memcpy(&td->ts, &parent->ts, sizeof(td->ts));
	153	td->o.uid = td->o.gid = -1U;
	154	dup_files(td, parent);
736a50cd	155	td->eo = parent->eo;
a9da8ab2 JA	156	fio_options_mem_dupe(td);
	157
	158	if (ioengine_load(td))
	159	goto err;
	160
	161	if (td->o.odirect)
	162	td->io_ops->flags \|= FIO_RAWIO;
	163
	164	td->pid = gettid();
	165
	166	INIT_FLIST_HEAD(&td->io_log_list);
	167	INIT_FLIST_HEAD(&td->io_hist_list);
	168	INIT_FLIST_HEAD(&td->verify_list);
	169	INIT_FLIST_HEAD(&td->trim_list);
	170	INIT_FLIST_HEAD(&td->next_rand_list);
	171	td->io_hist_tree = RB_ROOT;
	172
	173	td->o.iodepth = 1;
	174	if (td_io_init(td))
	175	goto err_io_init;
	176
	177	fio_gettime(&td->epoch, NULL);
	178	fio_getrusage(&td->ru_start);
ac28d905	179	clear_io_state(td, 1);
a9da8ab2 JA	180
	181	td_set_runstate(td, TD_RUNNING);
	182	td->flags \|= TD_F_CHILD;
	183	td->parent = parent;
	184	return 0;
	185
	186	err_io_init:
	187	close_ioengine(td);
	188	err:
	189	return 1;
	190	}
	191
f5cd2907	192	#ifdef CONFIG_SFAA
a9da8ab2 JA	193	static void sum_val(uint64_t dst, uint64_t src)
	194	{
	195	if (*src) {
	196	__sync_fetch_and_add(dst, *src);
	197	*src = 0;
	198	}
	199	}
2a274336 JA	200	#else
	201	static void sum_val(uint64_t dst, uint64_t src)
	202	{
	203	if (*src) {
	204	dst += src;
	205	*src = 0;
	206	}
	207	}
	208	#endif
a9da8ab2	209
a6a3469e JA	210	static void pthread_double_unlock(pthread_mutex_t *lock1,
a6a3469e JA	211	pthread_mutex_t *lock2)
a9da8ab2	212	{
f5cd2907	213	#ifndef CONFIG_SFAA
a6a3469e JA	214	pthread_mutex_unlock(lock1);
a6a3469e JA	215	pthread_mutex_unlock(lock2);
2a274336	216	#endif
a6a3469e JA	217	}
	218
	219	static void pthread_double_lock(pthread_mutex_t lock1, pthread_mutex_t lock2)
	220	{
	221	#ifndef CONFIG_SFAA
	222	if (lock1 < lock2) {
	223	pthread_mutex_lock(lock1);
	224	pthread_mutex_lock(lock2);
	225	} else {
	226	pthread_mutex_lock(lock2);
	227	pthread_mutex_lock(lock1);
	228	}
	229	#endif
	230	}
	231
	232	static void sum_ddir(struct thread_data dst, struct thread_data src,
	233	enum fio_ddir ddir)
	234	{
	235	pthread_double_lock(&dst->io_wq.stat_lock, &src->io_wq.stat_lock);
2a274336	236
a9da8ab2 JA	237	sum_val(&dst->io_bytes[ddir], &src->io_bytes[ddir]);
	238	sum_val(&dst->io_blocks[ddir], &src->io_blocks[ddir]);
	239	sum_val(&dst->this_io_blocks[ddir], &src->this_io_blocks[ddir]);
	240	sum_val(&dst->this_io_bytes[ddir], &src->this_io_bytes[ddir]);
	241	sum_val(&dst->bytes_done[ddir], &src->bytes_done[ddir]);
2a274336	242
a6a3469e	243	pthread_double_unlock(&dst->io_wq.stat_lock, &src->io_wq.stat_lock);
a9da8ab2 JA	244	}
	245
	246	static void update_accounting(struct submit_worker *sw)
	247	{
	248	struct thread_data *src = &sw->td;
	249	struct thread_data *dst = sw->wq->td;
	250
	251	if (td_read(src))
	252	sum_ddir(dst, src, DDIR_READ);
	253	if (td_write(src))
	254	sum_ddir(dst, src, DDIR_WRITE);
	255	if (td_trim(src))
	256	sum_ddir(dst, src, DDIR_TRIM);
	257	}
	258
	259	static void worker_thread(void data)
	260	{
	261	struct submit_worker *sw = data;
	262	struct workqueue *wq = sw->wq;
	263	struct thread_data *td = &sw->td;
	264	unsigned int eflags = 0, ret;
	265	FLIST_HEAD(local_list);
	266
	267	ret = init_submit_worker(sw);
	268	pthread_mutex_lock(&sw->lock);
	269	sw->flags \|= SW_F_RUNNING;
	270	if (ret)
	271	sw->flags \|= SW_F_ERROR;
	272	pthread_mutex_unlock(&sw->lock);
	273
	274	pthread_mutex_lock(&wq->flush_lock);
	275	pthread_cond_signal(&wq->flush_cond);
	276	pthread_mutex_unlock(&wq->flush_lock);
	277
	278	if (sw->flags & SW_F_ERROR)
	279	goto done;
	280
	281	while (1) {
	282	pthread_mutex_lock(&sw->lock);
	283
	284	if (flist_empty(&sw->work_list)) {
	285	if (sw->flags & SW_F_EXIT) {
	286	pthread_mutex_unlock(&sw->lock);
	287	break;
	288	}
	289
	290	if (td->io_u_queued \|\| td->cur_depth \|\|
	291	td->io_u_in_flight) {
40649b00 JA	292	int ret;
40649b00 JA	293
a9da8ab2	294	pthread_mutex_unlock(&sw->lock);
40649b00 JA	295	ret = io_u_quiesce(td);
	296	if (ret > 0)
	297	td->cur_depth -= ret;
a9da8ab2 JA	298	pthread_mutex_lock(&sw->lock);
	299	}
	300
	301	/*
	302	* We dropped and reaquired the lock, check
	303	* state again.
	304	*/
	305	if (!flist_empty(&sw->work_list))
	306	goto handle_work;
	307
	308	if (sw->flags & SW_F_EXIT) {
	309	pthread_mutex_unlock(&sw->lock);
	310	break;
	311	} else if (!(sw->flags & SW_F_IDLE)) {
	312	sw->flags \|= SW_F_IDLE;
	313	wq->next_free_worker = sw->index;
	314	if (wq->wake_idle)
	315	pthread_cond_signal(&wq->flush_cond);
	316	}
	317	update_accounting(sw);
	318	pthread_cond_wait(&sw->cond, &sw->lock);
	319	} else {
	320	handle_work:
	321	flist_splice_init(&sw->work_list, &local_list);
	322	}
	323	pthread_mutex_unlock(&sw->lock);
	324	handle_list(sw, &local_list);
	325	}
	326
	327	update_accounting(sw);
	328
	329	done:
	330	pthread_mutex_lock(&sw->lock);
	331	sw->flags \|= (SW_F_EXITED \| eflags);
	332	pthread_mutex_unlock(&sw->lock);
	333	return NULL;
	334	}
	335
	336	static void free_worker(struct submit_worker *sw)
	337	{
	338	struct thread_data *td = &sw->td;
	339
	340	fio_options_free(td);
	341	close_and_free_files(td);
	342	if (td->io_ops)
	343	close_ioengine(td);
	344	td_set_runstate(td, TD_EXITED);
	345
	346	pthread_cond_destroy(&sw->cond);
	347	pthread_mutex_destroy(&sw->lock);
	348	}
	349
	350	static void shutdown_worker(struct submit_worker sw, unsigned int sum_cnt)
	351	{
	352	struct thread_data *parent = sw->wq->td;
	353
	354	pthread_join(sw->thread, NULL);
	355	(*sum_cnt)++;
fd595830	356	sum_thread_stats(&parent->ts, &sw->td.ts, *sum_cnt == 1);
a9da8ab2 JA	357	free_worker(sw);
	358	}
	359
	360	void workqueue_exit(struct workqueue *wq)
	361	{
	362	unsigned int shutdown, sum_cnt = 0;
	363	struct submit_worker *sw;
	364	int i;
	365
	366	for (i = 0; i < wq->max_workers; i++) {
	367	sw = &wq->workers[i];
	368
	369	pthread_mutex_lock(&sw->lock);
	370	sw->flags \|= SW_F_EXIT;
	371	pthread_cond_signal(&sw->cond);
	372	pthread_mutex_unlock(&sw->lock);
	373	}
	374
	375	do {
	376	shutdown = 0;
	377	for (i = 0; i < wq->max_workers; i++) {
	378	sw = &wq->workers[i];
	379	if (sw->flags & SW_F_ACCOUNTED)
	380	continue;
b7d0bbf1	381	pthread_mutex_lock(&sw->lock);
a9da8ab2	382	sw->flags \|= SW_F_ACCOUNTED;
b7d0bbf1	383	pthread_mutex_unlock(&sw->lock);
a9da8ab2 JA	384	shutdown_worker(sw, &sum_cnt);
	385	shutdown++;
	386	}
	387	} while (shutdown && shutdown != wq->max_workers);
	388
	389	free(wq->workers);
	390	pthread_mutex_destroy(&wq->flush_lock);
	391	pthread_cond_destroy(&wq->flush_cond);
2a274336	392	pthread_mutex_destroy(&wq->stat_lock);
a9da8ab2 JA	393	}
	394
	395	static int start_worker(struct workqueue *wq, unsigned int index)
	396	{
	397	struct submit_worker *sw = &wq->workers[index];
	398	int ret;
	399
	400	INIT_FLIST_HEAD(&sw->work_list);
	401	pthread_cond_init(&sw->cond, NULL);
	402	pthread_mutex_init(&sw->lock, NULL);
	403	sw->wq = wq;
	404	sw->index = index;
	405
	406	ret = pthread_create(&sw->thread, NULL, worker_thread, sw);
	407	if (!ret) {
	408	pthread_mutex_lock(&sw->lock);
	409	sw->flags = SW_F_IDLE;
	410	pthread_mutex_unlock(&sw->lock);
	411	return 0;
	412	}
	413
	414	free_worker(sw);
	415	return 1;
	416	}
	417
	418	int workqueue_init(struct thread_data td, struct workqueue wq,
	419	workqueue_fn *fn, unsigned max_pending)
	420	{
	421	unsigned int running;
	422	int i, error;
	423
	424	wq->max_workers = max_pending;
	425	wq->td = td;
	426	wq->fn = fn;
	427	wq->work_seq = 0;
	428	wq->next_free_worker = 0;
	429	pthread_cond_init(&wq->flush_cond, NULL);
	430	pthread_mutex_init(&wq->flush_lock, NULL);
2a274336	431	pthread_mutex_init(&wq->stat_lock, NULL);
a9da8ab2 JA	432
	433	wq->workers = calloc(wq->max_workers, sizeof(struct submit_worker));
	434
	435	for (i = 0; i < wq->max_workers; i++)
	436	if (start_worker(wq, i))
	437	break;
	438
	439	wq->max_workers = i;
bddc8d16 JA	440	if (!wq->max_workers)
bddc8d16 JA	441	goto err;
a9da8ab2 JA	442
	443	/*
	444	* Wait for them all to be started and initialized
	445	*/
	446	error = 0;
	447	do {
	448	struct submit_worker *sw;
	449
	450	running = 0;
	451	pthread_mutex_lock(&wq->flush_lock);
	452	for (i = 0; i < wq->max_workers; i++) {
	453	sw = &wq->workers[i];
	454	pthread_mutex_lock(&sw->lock);
	455	if (sw->flags & SW_F_RUNNING)
	456	running++;
	457	if (sw->flags & SW_F_ERROR)
	458	error++;
	459	pthread_mutex_unlock(&sw->lock);
	460	}
	461
	462	if (error \|\| running == wq->max_workers) {
	463	pthread_mutex_unlock(&wq->flush_lock);
	464	break;
	465	}
	466
	467	pthread_cond_wait(&wq->flush_cond, &wq->flush_lock);
	468	pthread_mutex_unlock(&wq->flush_lock);
	469	} while (1);
	470
bddc8d16 JA	471	if (!error)
bddc8d16 JA	472	return 0;
a9da8ab2	473
bddc8d16 JA	474	err:
	475	log_err("Can't create rate workqueue\n");
	476	td_verror(td, ESRCH, "workqueue_init");
	477	workqueue_exit(wq);
	478	return 1;
a9da8ab2	479	}