2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/kthread.h>
20 #include <linux/slab.h>
21 #include <linux/list.h>
22 #include <linux/spinlock.h>
23 #include <linux/freezer.h>
24 #include "async-thread.h"
26 #define WORK_QUEUED_BIT 0
27 #define WORK_DONE_BIT 1
28 #define WORK_ORDER_DONE_BIT 2
29 #define WORK_HIGH_PRIO_BIT 3
32 * container for the kthread task pointer and the list of pending work
33 * One of these is allocated per thread.
35 struct btrfs_worker_thread {
36 /* pool we belong to */
37 struct btrfs_workers *workers;
39 /* list of struct btrfs_work that are waiting for service */
40 struct list_head pending;
41 struct list_head prio_pending;
43 /* list of worker threads from struct btrfs_workers */
44 struct list_head worker_list;
47 struct task_struct *task;
49 /* number of things on the pending list */
52 /* reference counter for this struct */
55 unsigned long sequence;
57 /* protects the pending list. */
60 /* set to non-zero when this thread is already awake and kicking */
63 /* are we currently idle */
68 * btrfs_start_workers uses kthread_run, which can block waiting for memory
69 * for a very long time. It will actually throttle on page writeback,
70 * and so it may not make progress until after our btrfs worker threads
71 * process all of the pending work structs in their queue
73 * This means we can't use btrfs_start_workers from inside a btrfs worker
74 * thread that is used as part of cleaning dirty memory, which pretty much
75 * involves all of the worker threads.
77 * Instead we have a helper queue who never has more than one thread
78 * where we scheduler thread start operations. This worker_start struct
79 * is used to contain the work and hold a pointer to the queue that needs
83 struct btrfs_work work;
84 struct btrfs_workers *queue;
87 static void start_new_worker_func(struct btrfs_work *work)
89 struct worker_start *start;
90 start = container_of(work, struct worker_start, work);
91 btrfs_start_workers(start->queue, 1);
95 static int start_new_worker(struct btrfs_workers *queue)
97 struct worker_start *start;
100 start = kzalloc(sizeof(*start), GFP_NOFS);
104 start->work.func = start_new_worker_func;
105 start->queue = queue;
106 ret = btrfs_queue_worker(queue->atomic_worker_start, &start->work);
113 * helper function to move a thread onto the idle list after it
114 * has finished some requests.
116 static void check_idle_worker(struct btrfs_worker_thread *worker)
118 if (!worker->idle && atomic_read(&worker->num_pending) <
119 worker->workers->idle_thresh / 2) {
121 spin_lock_irqsave(&worker->workers->lock, flags);
124 /* the list may be empty if the worker is just starting */
125 if (!list_empty(&worker->worker_list)) {
126 list_move(&worker->worker_list,
127 &worker->workers->idle_list);
129 spin_unlock_irqrestore(&worker->workers->lock, flags);
134 * helper function to move a thread off the idle list after new
135 * pending work is added.
137 static void check_busy_worker(struct btrfs_worker_thread *worker)
139 if (worker->idle && atomic_read(&worker->num_pending) >=
140 worker->workers->idle_thresh) {
142 spin_lock_irqsave(&worker->workers->lock, flags);
145 if (!list_empty(&worker->worker_list)) {
146 list_move_tail(&worker->worker_list,
147 &worker->workers->worker_list);
149 spin_unlock_irqrestore(&worker->workers->lock, flags);
153 static void check_pending_worker_creates(struct btrfs_worker_thread *worker)
155 struct btrfs_workers *workers = worker->workers;
159 if (!workers->atomic_start_pending)
162 spin_lock_irqsave(&workers->lock, flags);
163 if (!workers->atomic_start_pending)
166 workers->atomic_start_pending = 0;
167 if (workers->num_workers + workers->num_workers_starting >=
168 workers->max_workers)
171 workers->num_workers_starting += 1;
172 spin_unlock_irqrestore(&workers->lock, flags);
173 start_new_worker(workers);
177 spin_unlock_irqrestore(&workers->lock, flags);
180 static noinline int run_ordered_completions(struct btrfs_workers *workers,
181 struct btrfs_work *work)
183 if (!workers->ordered)
186 set_bit(WORK_DONE_BIT, &work->flags);
188 spin_lock(&workers->order_lock);
191 if (!list_empty(&workers->prio_order_list)) {
192 work = list_entry(workers->prio_order_list.next,
193 struct btrfs_work, order_list);
194 } else if (!list_empty(&workers->order_list)) {
195 work = list_entry(workers->order_list.next,
196 struct btrfs_work, order_list);
200 if (!test_bit(WORK_DONE_BIT, &work->flags))
203 /* we are going to call the ordered done function, but
204 * we leave the work item on the list as a barrier so
205 * that later work items that are done don't have their
206 * functions called before this one returns
208 if (test_and_set_bit(WORK_ORDER_DONE_BIT, &work->flags))
211 spin_unlock(&workers->order_lock);
213 work->ordered_func(work);
215 /* now take the lock again and call the freeing code */
216 spin_lock(&workers->order_lock);
217 list_del(&work->order_list);
218 work->ordered_free(work);
221 spin_unlock(&workers->order_lock);
225 static void put_worker(struct btrfs_worker_thread *worker)
227 if (atomic_dec_and_test(&worker->refs))
231 static int try_worker_shutdown(struct btrfs_worker_thread *worker)
235 spin_lock_irq(&worker->lock);
236 spin_lock(&worker->workers->lock);
237 if (worker->workers->num_workers > 1 &&
240 !list_empty(&worker->worker_list) &&
241 list_empty(&worker->prio_pending) &&
242 list_empty(&worker->pending) &&
243 atomic_read(&worker->num_pending) == 0) {
245 list_del_init(&worker->worker_list);
246 worker->workers->num_workers--;
248 spin_unlock(&worker->workers->lock);
249 spin_unlock_irq(&worker->lock);
256 static struct btrfs_work *get_next_work(struct btrfs_worker_thread *worker,
257 struct list_head *prio_head,
258 struct list_head *head)
260 struct btrfs_work *work = NULL;
261 struct list_head *cur = NULL;
263 if(!list_empty(prio_head))
264 cur = prio_head->next;
267 if (!list_empty(&worker->prio_pending))
270 if (!list_empty(head))
277 spin_lock_irq(&worker->lock);
278 list_splice_tail_init(&worker->prio_pending, prio_head);
279 list_splice_tail_init(&worker->pending, head);
281 if (!list_empty(prio_head))
282 cur = prio_head->next;
283 else if (!list_empty(head))
285 spin_unlock_irq(&worker->lock);
291 work = list_entry(cur, struct btrfs_work, list);
298 * main loop for servicing work items
300 static int worker_loop(void *arg)
302 struct btrfs_worker_thread *worker = arg;
303 struct list_head head;
304 struct list_head prio_head;
305 struct btrfs_work *work;
307 INIT_LIST_HEAD(&head);
308 INIT_LIST_HEAD(&prio_head);
315 work = get_next_work(worker, &prio_head, &head);
319 list_del(&work->list);
320 clear_bit(WORK_QUEUED_BIT, &work->flags);
322 work->worker = worker;
326 atomic_dec(&worker->num_pending);
328 * unless this is an ordered work queue,
329 * 'work' was probably freed by func above.
331 run_ordered_completions(worker->workers, work);
333 check_pending_worker_creates(worker);
337 spin_lock_irq(&worker->lock);
338 check_idle_worker(worker);
340 if (freezing(current)) {
342 spin_unlock_irq(&worker->lock);
345 spin_unlock_irq(&worker->lock);
346 if (!kthread_should_stop()) {
349 * we've dropped the lock, did someone else
353 if (!list_empty(&worker->pending) ||
354 !list_empty(&worker->prio_pending))
358 * this short schedule allows more work to
359 * come in without the queue functions
360 * needing to go through wake_up_process()
362 * worker->working is still 1, so nobody
363 * is going to try and wake us up
367 if (!list_empty(&worker->pending) ||
368 !list_empty(&worker->prio_pending))
371 if (kthread_should_stop())
374 /* still no more work?, sleep for real */
375 spin_lock_irq(&worker->lock);
376 set_current_state(TASK_INTERRUPTIBLE);
377 if (!list_empty(&worker->pending) ||
378 !list_empty(&worker->prio_pending)) {
379 spin_unlock_irq(&worker->lock);
380 set_current_state(TASK_RUNNING);
385 * this makes sure we get a wakeup when someone
386 * adds something new to the queue
389 spin_unlock_irq(&worker->lock);
391 if (!kthread_should_stop()) {
392 schedule_timeout(HZ * 120);
393 if (!worker->working &&
394 try_worker_shutdown(worker)) {
399 __set_current_state(TASK_RUNNING);
401 } while (!kthread_should_stop());
406 * this will wait for all the worker threads to shutdown
408 int btrfs_stop_workers(struct btrfs_workers *workers)
410 struct list_head *cur;
411 struct btrfs_worker_thread *worker;
414 spin_lock_irq(&workers->lock);
415 list_splice_init(&workers->idle_list, &workers->worker_list);
416 while (!list_empty(&workers->worker_list)) {
417 cur = workers->worker_list.next;
418 worker = list_entry(cur, struct btrfs_worker_thread,
421 atomic_inc(&worker->refs);
422 workers->num_workers -= 1;
423 if (!list_empty(&worker->worker_list)) {
424 list_del_init(&worker->worker_list);
429 spin_unlock_irq(&workers->lock);
431 kthread_stop(worker->task);
432 spin_lock_irq(&workers->lock);
435 spin_unlock_irq(&workers->lock);
440 * simple init on struct btrfs_workers
442 void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max,
443 struct btrfs_workers *async_helper)
445 workers->num_workers = 0;
446 workers->num_workers_starting = 0;
447 INIT_LIST_HEAD(&workers->worker_list);
448 INIT_LIST_HEAD(&workers->idle_list);
449 INIT_LIST_HEAD(&workers->order_list);
450 INIT_LIST_HEAD(&workers->prio_order_list);
451 spin_lock_init(&workers->lock);
452 spin_lock_init(&workers->order_lock);
453 workers->max_workers = max;
454 workers->idle_thresh = 32;
455 workers->name = name;
456 workers->ordered = 0;
457 workers->atomic_start_pending = 0;
458 workers->atomic_worker_start = async_helper;
462 * starts new worker threads. This does not enforce the max worker
463 * count in case you need to temporarily go past it.
465 static int __btrfs_start_workers(struct btrfs_workers *workers,
468 struct btrfs_worker_thread *worker;
472 for (i = 0; i < num_workers; i++) {
473 worker = kzalloc(sizeof(*worker), GFP_NOFS);
479 INIT_LIST_HEAD(&worker->pending);
480 INIT_LIST_HEAD(&worker->prio_pending);
481 INIT_LIST_HEAD(&worker->worker_list);
482 spin_lock_init(&worker->lock);
484 atomic_set(&worker->num_pending, 0);
485 atomic_set(&worker->refs, 1);
486 worker->workers = workers;
487 worker->task = kthread_run(worker_loop, worker,
488 "btrfs-%s-%d", workers->name,
489 workers->num_workers + i);
490 if (IS_ERR(worker->task)) {
491 ret = PTR_ERR(worker->task);
495 spin_lock_irq(&workers->lock);
496 list_add_tail(&worker->worker_list, &workers->idle_list);
498 workers->num_workers++;
499 workers->num_workers_starting--;
500 WARN_ON(workers->num_workers_starting < 0);
501 spin_unlock_irq(&workers->lock);
505 btrfs_stop_workers(workers);
509 int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
511 spin_lock_irq(&workers->lock);
512 workers->num_workers_starting += num_workers;
513 spin_unlock_irq(&workers->lock);
514 return __btrfs_start_workers(workers, num_workers);
518 * run through the list and find a worker thread that doesn't have a lot
519 * to do right now. This can return null if we aren't yet at the thread
520 * count limit and all of the threads are busy.
522 static struct btrfs_worker_thread *next_worker(struct btrfs_workers *workers)
524 struct btrfs_worker_thread *worker;
525 struct list_head *next;
528 enforce_min = (workers->num_workers + workers->num_workers_starting) <
529 workers->max_workers;
532 * if we find an idle thread, don't move it to the end of the
533 * idle list. This improves the chance that the next submission
534 * will reuse the same thread, and maybe catch it while it is still
537 if (!list_empty(&workers->idle_list)) {
538 next = workers->idle_list.next;
539 worker = list_entry(next, struct btrfs_worker_thread,
543 if (enforce_min || list_empty(&workers->worker_list))
547 * if we pick a busy task, move the task to the end of the list.
548 * hopefully this will keep things somewhat evenly balanced.
549 * Do the move in batches based on the sequence number. This groups
550 * requests submitted at roughly the same time onto the same worker.
552 next = workers->worker_list.next;
553 worker = list_entry(next, struct btrfs_worker_thread, worker_list);
556 if (worker->sequence % workers->idle_thresh == 0)
557 list_move_tail(next, &workers->worker_list);
562 * selects a worker thread to take the next job. This will either find
563 * an idle worker, start a new worker up to the max count, or just return
564 * one of the existing busy workers.
566 static struct btrfs_worker_thread *find_worker(struct btrfs_workers *workers)
568 struct btrfs_worker_thread *worker;
570 struct list_head *fallback;
573 spin_lock_irqsave(&workers->lock, flags);
574 worker = next_worker(workers);
577 if (workers->num_workers + workers->num_workers_starting >=
578 workers->max_workers) {
580 } else if (workers->atomic_worker_start) {
581 workers->atomic_start_pending = 1;
584 workers->num_workers_starting++;
585 spin_unlock_irqrestore(&workers->lock, flags);
586 /* we're below the limit, start another worker */
587 __btrfs_start_workers(workers, 1);
596 * we have failed to find any workers, just
597 * return the first one we can find.
599 if (!list_empty(&workers->worker_list))
600 fallback = workers->worker_list.next;
601 if (!list_empty(&workers->idle_list))
602 fallback = workers->idle_list.next;
604 worker = list_entry(fallback,
605 struct btrfs_worker_thread, worker_list);
608 * this makes sure the worker doesn't exit before it is placed
609 * onto a busy/idle list
611 atomic_inc(&worker->num_pending);
612 spin_unlock_irqrestore(&workers->lock, flags);
617 * btrfs_requeue_work just puts the work item back on the tail of the list
618 * it was taken from. It is intended for use with long running work functions
619 * that make some progress and want to give the cpu up for others.
621 int btrfs_requeue_work(struct btrfs_work *work)
623 struct btrfs_worker_thread *worker = work->worker;
627 if (test_and_set_bit(WORK_QUEUED_BIT, &work->flags))
630 spin_lock_irqsave(&worker->lock, flags);
631 if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
632 list_add_tail(&work->list, &worker->prio_pending);
634 list_add_tail(&work->list, &worker->pending);
635 atomic_inc(&worker->num_pending);
637 /* by definition we're busy, take ourselves off the idle
641 spin_lock(&worker->workers->lock);
643 list_move_tail(&worker->worker_list,
644 &worker->workers->worker_list);
645 spin_unlock(&worker->workers->lock);
647 if (!worker->working) {
653 wake_up_process(worker->task);
654 spin_unlock_irqrestore(&worker->lock, flags);
660 void btrfs_set_work_high_prio(struct btrfs_work *work)
662 set_bit(WORK_HIGH_PRIO_BIT, &work->flags);
666 * places a struct btrfs_work into the pending queue of one of the kthreads
668 int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
670 struct btrfs_worker_thread *worker;
674 /* don't requeue something already on a list */
675 if (test_and_set_bit(WORK_QUEUED_BIT, &work->flags))
678 worker = find_worker(workers);
679 if (workers->ordered) {
681 * you're not allowed to do ordered queues from an
684 spin_lock(&workers->order_lock);
685 if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags)) {
686 list_add_tail(&work->order_list,
687 &workers->prio_order_list);
689 list_add_tail(&work->order_list, &workers->order_list);
691 spin_unlock(&workers->order_lock);
693 INIT_LIST_HEAD(&work->order_list);
696 spin_lock_irqsave(&worker->lock, flags);
698 if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
699 list_add_tail(&work->list, &worker->prio_pending);
701 list_add_tail(&work->list, &worker->pending);
702 check_busy_worker(worker);
705 * avoid calling into wake_up_process if this thread has already
708 if (!worker->working)
713 wake_up_process(worker->task);
714 spin_unlock_irqrestore(&worker->lock, flags);