1 // SPDX-License-Identifier: GPL-2.0
3 * Basic worker thread pool for io_uring
5 * Copyright (C) 2019 Jens Axboe
8 #include <linux/kernel.h>
9 #include <linux/init.h>
10 #include <linux/errno.h>
11 #include <linux/sched/signal.h>
13 #include <linux/mmu_context.h>
14 #include <linux/sched/mm.h>
15 #include <linux/percpu.h>
16 #include <linux/slab.h>
17 #include <linux/kthread.h>
18 #include <linux/rculist_nulls.h>
22 #define WORKER_IDLE_TIMEOUT (5 * HZ)
25 IO_WORKER_F_UP = 1, /* up and active */
26 IO_WORKER_F_RUNNING = 2, /* account as running */
27 IO_WORKER_F_FREE = 4, /* worker on free list */
28 IO_WORKER_F_EXITING = 8, /* worker exiting */
29 IO_WORKER_F_FIXED = 16, /* static idle worker */
30 IO_WORKER_F_BOUND = 32, /* is doing bounded work */
34 IO_WQ_BIT_EXIT = 0, /* wq exiting */
35 IO_WQ_BIT_CANCEL = 1, /* cancel work on list */
36 IO_WQ_BIT_ERROR = 2, /* error on setup */
40 IO_WQE_FLAG_STALLED = 1, /* stalled on hash */
44 * One for each thread in a wqe pool
49 struct hlist_nulls_node nulls_node;
50 struct list_head all_list;
51 struct task_struct *task;
54 struct io_wq_work *cur_work;
59 const struct cred *cur_creds;
60 const struct cred *saved_creds;
61 struct files_struct *restore_files;
64 #if BITS_PER_LONG == 64
65 #define IO_WQ_HASH_ORDER 6
67 #define IO_WQ_HASH_ORDER 5
82 * Per-node worker thread pool
87 struct io_wq_work_list work_list;
88 unsigned long hash_map;
90 } ____cacheline_aligned_in_smp;
93 struct io_wqe_acct acct[2];
95 struct hlist_nulls_head free_list;
96 struct list_head all_list;
105 struct io_wqe **wqes;
108 get_work_fn *get_work;
109 put_work_fn *put_work;
111 struct task_struct *manager;
112 struct user_struct *user;
114 struct completion done;
119 static bool io_worker_get(struct io_worker *worker)
121 return refcount_inc_not_zero(&worker->ref);
124 static void io_worker_release(struct io_worker *worker)
126 if (refcount_dec_and_test(&worker->ref))
127 wake_up_process(worker->task);
131 * Note: drops the wqe->lock if returning true! The caller must re-acquire
132 * the lock in that case. Some callers need to restart handling if this
133 * happens, so we can't just re-acquire the lock on behalf of the caller.
135 static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
137 bool dropped_lock = false;
139 if (worker->saved_creds) {
140 revert_creds(worker->saved_creds);
141 worker->cur_creds = worker->saved_creds = NULL;
144 if (current->files != worker->restore_files) {
145 __acquire(&wqe->lock);
146 spin_unlock_irq(&wqe->lock);
150 current->files = worker->restore_files;
151 task_unlock(current);
155 * If we have an active mm, we need to drop the wq lock before unusing
156 * it. If we do, return true and let the caller retry the idle loop.
160 __acquire(&wqe->lock);
161 spin_unlock_irq(&wqe->lock);
164 __set_current_state(TASK_RUNNING);
166 unuse_mm(worker->mm);
174 static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
175 struct io_wq_work *work)
177 if (work->flags & IO_WQ_WORK_UNBOUND)
178 return &wqe->acct[IO_WQ_ACCT_UNBOUND];
180 return &wqe->acct[IO_WQ_ACCT_BOUND];
183 static inline struct io_wqe_acct *io_wqe_get_acct(struct io_wqe *wqe,
184 struct io_worker *worker)
186 if (worker->flags & IO_WORKER_F_BOUND)
187 return &wqe->acct[IO_WQ_ACCT_BOUND];
189 return &wqe->acct[IO_WQ_ACCT_UNBOUND];
192 static void io_worker_exit(struct io_worker *worker)
194 struct io_wqe *wqe = worker->wqe;
195 struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
199 * If we're not at zero, someone else is holding a brief reference
200 * to the worker. Wait for that to go away.
202 set_current_state(TASK_INTERRUPTIBLE);
203 if (!refcount_dec_and_test(&worker->ref))
205 __set_current_state(TASK_RUNNING);
208 current->flags &= ~PF_IO_WORKER;
209 if (worker->flags & IO_WORKER_F_RUNNING)
210 atomic_dec(&acct->nr_running);
211 if (!(worker->flags & IO_WORKER_F_BOUND))
212 atomic_dec(&wqe->wq->user->processes);
216 spin_lock_irq(&wqe->lock);
217 hlist_nulls_del_rcu(&worker->nulls_node);
218 list_del_rcu(&worker->all_list);
219 if (__io_worker_unuse(wqe, worker)) {
220 __release(&wqe->lock);
221 spin_lock_irq(&wqe->lock);
224 nr_workers = wqe->acct[IO_WQ_ACCT_BOUND].nr_workers +
225 wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers;
226 spin_unlock_irq(&wqe->lock);
228 /* all workers gone, wq exit can proceed */
229 if (!nr_workers && refcount_dec_and_test(&wqe->wq->refs))
230 complete(&wqe->wq->done);
232 kfree_rcu(worker, rcu);
235 static inline bool io_wqe_run_queue(struct io_wqe *wqe)
236 __must_hold(wqe->lock)
238 if (!wq_list_empty(&wqe->work_list) &&
239 !(wqe->flags & IO_WQE_FLAG_STALLED))
245 * Check head of free list for an available worker. If one isn't available,
246 * caller must wake up the wq manager to create one.
248 static bool io_wqe_activate_free_worker(struct io_wqe *wqe)
251 struct hlist_nulls_node *n;
252 struct io_worker *worker;
254 n = rcu_dereference(hlist_nulls_first_rcu(&wqe->free_list));
258 worker = hlist_nulls_entry(n, struct io_worker, nulls_node);
259 if (io_worker_get(worker)) {
260 wake_up_process(worker->task);
261 io_worker_release(worker);
269 * We need a worker. If we find a free one, we're good. If not, and we're
270 * below the max number of workers, wake up the manager to create one.
272 static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct)
277 * Most likely an attempt to queue unbounded work on an io_wq that
278 * wasn't setup with any unbounded workers.
280 WARN_ON_ONCE(!acct->max_workers);
283 ret = io_wqe_activate_free_worker(wqe);
286 if (!ret && acct->nr_workers < acct->max_workers)
287 wake_up_process(wqe->wq->manager);
290 static void io_wqe_inc_running(struct io_wqe *wqe, struct io_worker *worker)
292 struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
294 atomic_inc(&acct->nr_running);
297 static void io_wqe_dec_running(struct io_wqe *wqe, struct io_worker *worker)
298 __must_hold(wqe->lock)
300 struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
302 if (atomic_dec_and_test(&acct->nr_running) && io_wqe_run_queue(wqe))
303 io_wqe_wake_worker(wqe, acct);
306 static void io_worker_start(struct io_wqe *wqe, struct io_worker *worker)
308 allow_kernel_signal(SIGINT);
310 current->flags |= PF_IO_WORKER;
312 worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
313 worker->restore_files = current->files;
314 io_wqe_inc_running(wqe, worker);
318 * Worker will start processing some work. Move it to the busy list, if
319 * it's currently on the freelist
321 static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
322 struct io_wq_work *work)
323 __must_hold(wqe->lock)
325 bool worker_bound, work_bound;
327 if (worker->flags & IO_WORKER_F_FREE) {
328 worker->flags &= ~IO_WORKER_F_FREE;
329 hlist_nulls_del_init_rcu(&worker->nulls_node);
333 * If worker is moving from bound to unbound (or vice versa), then
334 * ensure we update the running accounting.
336 worker_bound = (worker->flags & IO_WORKER_F_BOUND) != 0;
337 work_bound = (work->flags & IO_WQ_WORK_UNBOUND) == 0;
338 if (worker_bound != work_bound) {
339 io_wqe_dec_running(wqe, worker);
341 worker->flags |= IO_WORKER_F_BOUND;
342 wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers--;
343 wqe->acct[IO_WQ_ACCT_BOUND].nr_workers++;
344 atomic_dec(&wqe->wq->user->processes);
346 worker->flags &= ~IO_WORKER_F_BOUND;
347 wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers++;
348 wqe->acct[IO_WQ_ACCT_BOUND].nr_workers--;
349 atomic_inc(&wqe->wq->user->processes);
351 io_wqe_inc_running(wqe, worker);
356 * No work, worker going to sleep. Move to freelist, and unuse mm if we
357 * have one attached. Dropping the mm may potentially sleep, so we drop
358 * the lock in that case and return success. Since the caller has to
359 * retry the loop in that case (we changed task state), we don't regrab
360 * the lock if we return success.
362 static bool __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
363 __must_hold(wqe->lock)
365 if (!(worker->flags & IO_WORKER_F_FREE)) {
366 worker->flags |= IO_WORKER_F_FREE;
367 hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
370 return __io_worker_unuse(wqe, worker);
373 static struct io_wq_work *io_get_next_work(struct io_wqe *wqe, unsigned *hash)
374 __must_hold(wqe->lock)
376 struct io_wq_work_node *node, *prev;
377 struct io_wq_work *work;
379 wq_list_for_each(node, prev, &wqe->work_list) {
380 work = container_of(node, struct io_wq_work, list);
382 /* not hashed, can run anytime */
383 if (!(work->flags & IO_WQ_WORK_HASHED)) {
384 wq_node_del(&wqe->work_list, node, prev);
388 /* hashed, can run if not already running */
389 *hash = work->flags >> IO_WQ_HASH_SHIFT;
390 if (!(wqe->hash_map & BIT_ULL(*hash))) {
391 wqe->hash_map |= BIT_ULL(*hash);
392 wq_node_del(&wqe->work_list, node, prev);
400 static void io_wq_switch_mm(struct io_worker *worker, struct io_wq_work *work)
403 unuse_mm(worker->mm);
411 if (mmget_not_zero(work->mm)) {
415 worker->mm = work->mm;
416 /* hang on to this mm */
421 /* failed grabbing mm, ensure work gets cancelled */
422 work->flags |= IO_WQ_WORK_CANCEL;
425 static void io_wq_switch_creds(struct io_worker *worker,
426 struct io_wq_work *work)
428 const struct cred *old_creds = override_creds(work->creds);
430 worker->cur_creds = work->creds;
431 if (worker->saved_creds)
432 put_cred(old_creds); /* creds set by previous switch */
434 worker->saved_creds = old_creds;
437 static void io_worker_handle_work(struct io_worker *worker)
438 __releases(wqe->lock)
440 struct io_wq_work *work, *old_work = NULL, *put_work = NULL;
441 struct io_wqe *wqe = worker->wqe;
442 struct io_wq *wq = wqe->wq;
448 * If we got some work, mark us as busy. If we didn't, but
449 * the list isn't empty, it means we stalled on hashed work.
450 * Mark us stalled so we don't keep looking for work when we
451 * can't make progress, any work completion or insertion will
452 * clear the stalled flag.
454 work = io_get_next_work(wqe, &hash);
456 __io_worker_busy(wqe, worker, work);
457 else if (!wq_list_empty(&wqe->work_list))
458 wqe->flags |= IO_WQE_FLAG_STALLED;
460 spin_unlock_irq(&wqe->lock);
461 if (put_work && wq->put_work)
462 wq->put_work(old_work);
466 /* flush any pending signals before assigning new work */
467 if (signal_pending(current))
468 flush_signals(current);
472 spin_lock_irq(&worker->lock);
473 worker->cur_work = work;
474 spin_unlock_irq(&worker->lock);
476 if (work->flags & IO_WQ_WORK_CB)
479 if (work->files && current->files != work->files) {
481 current->files = work->files;
482 task_unlock(current);
484 if (work->mm != worker->mm)
485 io_wq_switch_mm(worker, work);
486 if (worker->cur_creds != work->creds)
487 io_wq_switch_creds(worker, work);
489 * OK to set IO_WQ_WORK_CANCEL even for uncancellable work,
490 * the worker function will do the right thing.
492 if (test_bit(IO_WQ_BIT_CANCEL, &wq->state))
493 work->flags |= IO_WQ_WORK_CANCEL;
495 work->flags |= IO_WQ_WORK_HAS_MM;
497 if (wq->get_work && !(work->flags & IO_WQ_WORK_INTERNAL)) {
505 spin_lock_irq(&worker->lock);
506 worker->cur_work = NULL;
507 spin_unlock_irq(&worker->lock);
509 spin_lock_irq(&wqe->lock);
512 wqe->hash_map &= ~BIT_ULL(hash);
513 wqe->flags &= ~IO_WQE_FLAG_STALLED;
515 if (work && work != old_work) {
516 spin_unlock_irq(&wqe->lock);
518 if (put_work && wq->put_work) {
519 wq->put_work(put_work);
523 /* dependent work not hashed */
530 static inline void io_worker_spin_for_work(struct io_wqe *wqe)
535 if (io_wqe_run_queue(wqe))
543 static int io_wqe_worker(void *data)
545 struct io_worker *worker = data;
546 struct io_wqe *wqe = worker->wqe;
547 struct io_wq *wq = wqe->wq;
550 io_worker_start(wqe, worker);
553 while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
554 set_current_state(TASK_INTERRUPTIBLE);
557 io_worker_spin_for_work(wqe);
558 spin_lock_irq(&wqe->lock);
559 if (io_wqe_run_queue(wqe)) {
560 __set_current_state(TASK_RUNNING);
561 io_worker_handle_work(worker);
566 /* drops the lock on success, retry */
567 if (__io_worker_idle(wqe, worker)) {
568 __release(&wqe->lock);
571 spin_unlock_irq(&wqe->lock);
572 if (signal_pending(current))
573 flush_signals(current);
574 if (schedule_timeout(WORKER_IDLE_TIMEOUT))
576 /* timed out, exit unless we're the fixed worker */
577 if (test_bit(IO_WQ_BIT_EXIT, &wq->state) ||
578 !(worker->flags & IO_WORKER_F_FIXED))
582 if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
583 spin_lock_irq(&wqe->lock);
584 if (!wq_list_empty(&wqe->work_list))
585 io_worker_handle_work(worker);
587 spin_unlock_irq(&wqe->lock);
590 io_worker_exit(worker);
595 * Called when a worker is scheduled in. Mark us as currently running.
597 void io_wq_worker_running(struct task_struct *tsk)
599 struct io_worker *worker = kthread_data(tsk);
600 struct io_wqe *wqe = worker->wqe;
602 if (!(worker->flags & IO_WORKER_F_UP))
604 if (worker->flags & IO_WORKER_F_RUNNING)
606 worker->flags |= IO_WORKER_F_RUNNING;
607 io_wqe_inc_running(wqe, worker);
611 * Called when worker is going to sleep. If there are no workers currently
612 * running and we have work pending, wake up a free one or have the manager
615 void io_wq_worker_sleeping(struct task_struct *tsk)
617 struct io_worker *worker = kthread_data(tsk);
618 struct io_wqe *wqe = worker->wqe;
620 if (!(worker->flags & IO_WORKER_F_UP))
622 if (!(worker->flags & IO_WORKER_F_RUNNING))
625 worker->flags &= ~IO_WORKER_F_RUNNING;
627 spin_lock_irq(&wqe->lock);
628 io_wqe_dec_running(wqe, worker);
629 spin_unlock_irq(&wqe->lock);
632 static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
634 struct io_wqe_acct *acct =&wqe->acct[index];
635 struct io_worker *worker;
637 worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node);
641 refcount_set(&worker->ref, 1);
642 worker->nulls_node.pprev = NULL;
644 spin_lock_init(&worker->lock);
646 worker->task = kthread_create_on_node(io_wqe_worker, worker, wqe->node,
647 "io_wqe_worker-%d/%d", index, wqe->node);
648 if (IS_ERR(worker->task)) {
653 spin_lock_irq(&wqe->lock);
654 hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
655 list_add_tail_rcu(&worker->all_list, &wqe->all_list);
656 worker->flags |= IO_WORKER_F_FREE;
657 if (index == IO_WQ_ACCT_BOUND)
658 worker->flags |= IO_WORKER_F_BOUND;
659 if (!acct->nr_workers && (worker->flags & IO_WORKER_F_BOUND))
660 worker->flags |= IO_WORKER_F_FIXED;
662 spin_unlock_irq(&wqe->lock);
664 if (index == IO_WQ_ACCT_UNBOUND)
665 atomic_inc(&wq->user->processes);
667 wake_up_process(worker->task);
671 static inline bool io_wqe_need_worker(struct io_wqe *wqe, int index)
672 __must_hold(wqe->lock)
674 struct io_wqe_acct *acct = &wqe->acct[index];
676 /* if we have available workers or no work, no need */
677 if (!hlist_nulls_empty(&wqe->free_list) || !io_wqe_run_queue(wqe))
679 return acct->nr_workers < acct->max_workers;
683 * Manager thread. Tasked with creating new workers, if we need them.
685 static int io_wq_manager(void *data)
687 struct io_wq *wq = data;
688 int workers_to_create = num_possible_nodes();
691 /* create fixed workers */
692 refcount_set(&wq->refs, workers_to_create);
693 for_each_node(node) {
694 if (!create_io_worker(wq, wq->wqes[node], IO_WQ_ACCT_BOUND))
701 while (!kthread_should_stop()) {
702 for_each_node(node) {
703 struct io_wqe *wqe = wq->wqes[node];
704 bool fork_worker[2] = { false, false };
706 spin_lock_irq(&wqe->lock);
707 if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
708 fork_worker[IO_WQ_ACCT_BOUND] = true;
709 if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
710 fork_worker[IO_WQ_ACCT_UNBOUND] = true;
711 spin_unlock_irq(&wqe->lock);
712 if (fork_worker[IO_WQ_ACCT_BOUND])
713 create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
714 if (fork_worker[IO_WQ_ACCT_UNBOUND])
715 create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
717 set_current_state(TASK_INTERRUPTIBLE);
718 schedule_timeout(HZ);
723 set_bit(IO_WQ_BIT_ERROR, &wq->state);
724 set_bit(IO_WQ_BIT_EXIT, &wq->state);
725 if (refcount_sub_and_test(workers_to_create, &wq->refs))
730 static bool io_wq_can_queue(struct io_wqe *wqe, struct io_wqe_acct *acct,
731 struct io_wq_work *work)
735 if (!(work->flags & IO_WQ_WORK_UNBOUND))
737 if (atomic_read(&acct->nr_running))
741 free_worker = !hlist_nulls_empty(&wqe->free_list);
746 if (atomic_read(&wqe->wq->user->processes) >= acct->max_workers &&
747 !(capable(CAP_SYS_RESOURCE) || capable(CAP_SYS_ADMIN)))
753 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
755 struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
760 * Do early check to see if we need a new unbound worker, and if we do,
761 * if we're allowed to do so. This isn't 100% accurate as there's a
762 * gap between this check and incrementing the value, but that's OK.
763 * It's close enough to not be an issue, fork() has the same delay.
765 if (unlikely(!io_wq_can_queue(wqe, acct, work))) {
766 work->flags |= IO_WQ_WORK_CANCEL;
771 work_flags = work->flags;
772 spin_lock_irqsave(&wqe->lock, flags);
773 wq_list_add_tail(&work->list, &wqe->work_list);
774 wqe->flags &= ~IO_WQE_FLAG_STALLED;
775 spin_unlock_irqrestore(&wqe->lock, flags);
777 if ((work_flags & IO_WQ_WORK_CONCURRENT) ||
778 !atomic_read(&acct->nr_running))
779 io_wqe_wake_worker(wqe, acct);
782 void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work)
784 struct io_wqe *wqe = wq->wqes[numa_node_id()];
786 io_wqe_enqueue(wqe, work);
790 * Enqueue work, hashed by some key. Work items that hash to the same value
791 * will not be done in parallel. Used to limit concurrent writes, generally
794 void io_wq_enqueue_hashed(struct io_wq *wq, struct io_wq_work *work, void *val)
796 struct io_wqe *wqe = wq->wqes[numa_node_id()];
800 bit = hash_ptr(val, IO_WQ_HASH_ORDER);
801 work->flags |= (IO_WQ_WORK_HASHED | (bit << IO_WQ_HASH_SHIFT));
802 io_wqe_enqueue(wqe, work);
805 static bool io_wqe_worker_send_sig(struct io_worker *worker, void *data)
807 send_sig(SIGINT, worker->task, 1);
812 * Iterate the passed in list and call the specific function for each
813 * worker that isn't exiting
815 static bool io_wq_for_each_worker(struct io_wqe *wqe,
816 bool (*func)(struct io_worker *, void *),
819 struct io_worker *worker;
822 list_for_each_entry_rcu(worker, &wqe->all_list, all_list) {
823 if (io_worker_get(worker)) {
824 ret = func(worker, data);
825 io_worker_release(worker);
834 void io_wq_cancel_all(struct io_wq *wq)
838 set_bit(IO_WQ_BIT_CANCEL, &wq->state);
841 for_each_node(node) {
842 struct io_wqe *wqe = wq->wqes[node];
844 io_wq_for_each_worker(wqe, io_wqe_worker_send_sig, NULL);
849 struct io_cb_cancel_data {
851 work_cancel_fn *cancel;
855 static bool io_work_cancel(struct io_worker *worker, void *cancel_data)
857 struct io_cb_cancel_data *data = cancel_data;
862 * Hold the lock to avoid ->cur_work going out of scope, caller
863 * may dereference the passed in work.
865 spin_lock_irqsave(&worker->lock, flags);
866 if (worker->cur_work &&
867 !(worker->cur_work->flags & IO_WQ_WORK_NO_CANCEL) &&
868 data->cancel(worker->cur_work, data->caller_data)) {
869 send_sig(SIGINT, worker->task, 1);
872 spin_unlock_irqrestore(&worker->lock, flags);
877 static enum io_wq_cancel io_wqe_cancel_cb_work(struct io_wqe *wqe,
878 work_cancel_fn *cancel,
881 struct io_cb_cancel_data data = {
884 .caller_data = cancel_data,
886 struct io_wq_work_node *node, *prev;
887 struct io_wq_work *work;
891 spin_lock_irqsave(&wqe->lock, flags);
892 wq_list_for_each(node, prev, &wqe->work_list) {
893 work = container_of(node, struct io_wq_work, list);
895 if (cancel(work, cancel_data)) {
896 wq_node_del(&wqe->work_list, node, prev);
901 spin_unlock_irqrestore(&wqe->lock, flags);
904 work->flags |= IO_WQ_WORK_CANCEL;
906 return IO_WQ_CANCEL_OK;
910 found = io_wq_for_each_worker(wqe, io_work_cancel, &data);
912 return found ? IO_WQ_CANCEL_RUNNING : IO_WQ_CANCEL_NOTFOUND;
915 enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
918 enum io_wq_cancel ret = IO_WQ_CANCEL_NOTFOUND;
921 for_each_node(node) {
922 struct io_wqe *wqe = wq->wqes[node];
924 ret = io_wqe_cancel_cb_work(wqe, cancel, data);
925 if (ret != IO_WQ_CANCEL_NOTFOUND)
932 static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
934 struct io_wq_work *work = data;
938 if (worker->cur_work != work)
941 spin_lock_irqsave(&worker->lock, flags);
942 if (worker->cur_work == work &&
943 !(worker->cur_work->flags & IO_WQ_WORK_NO_CANCEL)) {
944 send_sig(SIGINT, worker->task, 1);
947 spin_unlock_irqrestore(&worker->lock, flags);
952 static enum io_wq_cancel io_wqe_cancel_work(struct io_wqe *wqe,
953 struct io_wq_work *cwork)
955 struct io_wq_work_node *node, *prev;
956 struct io_wq_work *work;
960 cwork->flags |= IO_WQ_WORK_CANCEL;
963 * First check pending list, if we're lucky we can just remove it
964 * from there. CANCEL_OK means that the work is returned as-new,
965 * no completion will be posted for it.
967 spin_lock_irqsave(&wqe->lock, flags);
968 wq_list_for_each(node, prev, &wqe->work_list) {
969 work = container_of(node, struct io_wq_work, list);
972 wq_node_del(&wqe->work_list, node, prev);
977 spin_unlock_irqrestore(&wqe->lock, flags);
980 work->flags |= IO_WQ_WORK_CANCEL;
982 return IO_WQ_CANCEL_OK;
986 * Now check if a free (going busy) or busy worker has the work
987 * currently running. If we find it there, we'll return CANCEL_RUNNING
988 * as an indication that we attempt to signal cancellation. The
989 * completion will run normally in this case.
992 found = io_wq_for_each_worker(wqe, io_wq_worker_cancel, cwork);
994 return found ? IO_WQ_CANCEL_RUNNING : IO_WQ_CANCEL_NOTFOUND;
997 enum io_wq_cancel io_wq_cancel_work(struct io_wq *wq, struct io_wq_work *cwork)
999 enum io_wq_cancel ret = IO_WQ_CANCEL_NOTFOUND;
1002 for_each_node(node) {
1003 struct io_wqe *wqe = wq->wqes[node];
1005 ret = io_wqe_cancel_work(wqe, cwork);
1006 if (ret != IO_WQ_CANCEL_NOTFOUND)
1013 struct io_wq_flush_data {
1014 struct io_wq_work work;
1015 struct completion done;
1018 static void io_wq_flush_func(struct io_wq_work **workptr)
1020 struct io_wq_work *work = *workptr;
1021 struct io_wq_flush_data *data;
1023 data = container_of(work, struct io_wq_flush_data, work);
1024 complete(&data->done);
1028 * Doesn't wait for previously queued work to finish. When this completes,
1029 * it just means that previously queued work was started.
1031 void io_wq_flush(struct io_wq *wq)
1033 struct io_wq_flush_data data;
1036 for_each_node(node) {
1037 struct io_wqe *wqe = wq->wqes[node];
1039 init_completion(&data.done);
1040 INIT_IO_WORK(&data.work, io_wq_flush_func);
1041 data.work.flags |= IO_WQ_WORK_INTERNAL;
1042 io_wqe_enqueue(wqe, &data.work);
1043 wait_for_completion(&data.done);
1047 struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
1049 int ret = -ENOMEM, node;
1052 wq = kzalloc(sizeof(*wq), GFP_KERNEL);
1054 return ERR_PTR(-ENOMEM);
1056 wq->wqes = kcalloc(nr_node_ids, sizeof(struct io_wqe *), GFP_KERNEL);
1059 return ERR_PTR(-ENOMEM);
1062 wq->get_work = data->get_work;
1063 wq->put_work = data->put_work;
1065 /* caller must already hold a reference to this */
1066 wq->user = data->user;
1068 for_each_node(node) {
1071 wqe = kzalloc_node(sizeof(struct io_wqe), GFP_KERNEL, node);
1074 wq->wqes[node] = wqe;
1076 wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
1077 atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
1079 wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
1080 task_rlimit(current, RLIMIT_NPROC);
1082 atomic_set(&wqe->acct[IO_WQ_ACCT_UNBOUND].nr_running, 0);
1085 spin_lock_init(&wqe->lock);
1086 INIT_WQ_LIST(&wqe->work_list);
1087 INIT_HLIST_NULLS_HEAD(&wqe->free_list, 0);
1088 INIT_LIST_HEAD(&wqe->all_list);
1091 init_completion(&wq->done);
1093 wq->manager = kthread_create(io_wq_manager, wq, "io_wq_manager");
1094 if (!IS_ERR(wq->manager)) {
1095 wake_up_process(wq->manager);
1096 wait_for_completion(&wq->done);
1097 if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
1101 refcount_set(&wq->use_refs, 1);
1102 reinit_completion(&wq->done);
1106 ret = PTR_ERR(wq->manager);
1107 complete(&wq->done);
1110 kfree(wq->wqes[node]);
1113 return ERR_PTR(ret);
1116 bool io_wq_get(struct io_wq *wq, struct io_wq_data *data)
1118 if (data->get_work != wq->get_work || data->put_work != wq->put_work)
1121 return refcount_inc_not_zero(&wq->use_refs);
1124 static bool io_wq_worker_wake(struct io_worker *worker, void *data)
1126 wake_up_process(worker->task);
1130 static void __io_wq_destroy(struct io_wq *wq)
1134 set_bit(IO_WQ_BIT_EXIT, &wq->state);
1136 kthread_stop(wq->manager);
1140 io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
1143 wait_for_completion(&wq->done);
1146 kfree(wq->wqes[node]);
1151 void io_wq_destroy(struct io_wq *wq)
1153 if (refcount_dec_and_test(&wq->use_refs))
1154 __io_wq_destroy(wq);