summaryrefslogtreecommitdiff
path: root/fs/io-wq.c
blob: 9b32b3c811f5a3a5b4236a8fc0febbf13ac6aede (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
// SPDX-License-Identifier: GPL-2.0
/*
 * Basic worker thread pool for io_uring
 *
 * Copyright (C) 2019 Jens Axboe
 *
 */
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/sched/signal.h>
#include <linux/mm.h>
#include <linux/mmu_context.h>
#include <linux/sched/mm.h>
#include <linux/percpu.h>
#include <linux/slab.h>
#include <linux/kthread.h>
#include <linux/rculist_nulls.h>

#include "io-wq.h"

#define WORKER_IDLE_TIMEOUT	(5 * HZ)

enum {
	IO_WORKER_F_UP		= 1,	/* up and active */
	IO_WORKER_F_RUNNING	= 2,	/* account as running */
	IO_WORKER_F_FREE	= 4,	/* worker on free list */
	IO_WORKER_F_EXITING	= 8,	/* worker exiting */
	IO_WORKER_F_FIXED	= 16,	/* static idle worker */
	IO_WORKER_F_BOUND	= 32,	/* is doing bounded work */
};

enum {
	IO_WQ_BIT_EXIT		= 0,	/* wq exiting */
	IO_WQ_BIT_CANCEL	= 1,	/* cancel work on list */
	IO_WQ_BIT_ERROR		= 2,	/* error on setup */
};

enum {
	IO_WQE_FLAG_STALLED	= 1,	/* stalled on hash */
};

/*
 * One for each thread in a wqe pool
 */
struct io_worker {
	refcount_t ref;
	unsigned flags;
	struct hlist_nulls_node nulls_node;
	struct list_head all_list;
	struct task_struct *task;
	wait_queue_head_t wait;
	struct io_wqe *wqe;

	struct io_wq_work *cur_work;
	spinlock_t lock;

	struct rcu_head rcu;
	struct mm_struct *mm;
	struct files_struct *restore_files;
};

#if BITS_PER_LONG == 64
#define IO_WQ_HASH_ORDER	6
#else
#define IO_WQ_HASH_ORDER	5
#endif

struct io_wqe_acct {
	unsigned nr_workers;
	unsigned max_workers;
	atomic_t nr_running;
};

enum {
	IO_WQ_ACCT_BOUND,
	IO_WQ_ACCT_UNBOUND,
};

/*
 * Per-node worker thread pool
 */
struct io_wqe {
	struct {
		spinlock_t lock;
		struct list_head work_list;
		unsigned long hash_map;
		unsigned flags;
	} ____cacheline_aligned_in_smp;

	int node;
	struct io_wqe_acct acct[2];

	struct hlist_nulls_head free_list;
	struct hlist_nulls_head busy_list;
	struct list_head all_list;

	struct io_wq *wq;
};

/*
 * Per io_wq state
  */
struct io_wq {
	struct io_wqe **wqes;
	unsigned long state;
	unsigned nr_wqes;

	get_work_fn *get_work;
	put_work_fn *put_work;

	struct task_struct *manager;
	struct user_struct *user;
	struct mm_struct *mm;
	refcount_t refs;
	struct completion done;
};

static bool io_worker_get(struct io_worker *worker)
{
	return refcount_inc_not_zero(&worker->ref);
}

static void io_worker_release(struct io_worker *worker)
{
	if (refcount_dec_and_test(&worker->ref))
		wake_up_process(worker->task);
}

/*
 * Note: drops the wqe->lock if returning true! The caller must re-acquire
 * the lock in that case. Some callers need to restart handling if this
 * happens, so we can't just re-acquire the lock on behalf of the caller.
 */
static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
{
	bool dropped_lock = false;

	if (current->files != worker->restore_files) {
		__acquire(&wqe->lock);
		spin_unlock_irq(&wqe->lock);
		dropped_lock = true;

		task_lock(current);
		current->files = worker->restore_files;
		task_unlock(current);
	}

	/*
	 * If we have an active mm, we need to drop the wq lock before unusing
	 * it. If we do, return true and let the caller retry the idle loop.
	 */
	if (worker->mm) {
		if (!dropped_lock) {
			__acquire(&wqe->lock);
			spin_unlock_irq(&wqe->lock);
			dropped_lock = true;
		}
		__set_current_state(TASK_RUNNING);
		set_fs(KERNEL_DS);
		unuse_mm(worker->mm);
		mmput(worker->mm);
		worker->mm = NULL;
	}

	return dropped_lock;
}

static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
						   struct io_wq_work *work)
{
	if (work->flags & IO_WQ_WORK_UNBOUND)
		return &wqe->acct[IO_WQ_ACCT_UNBOUND];

	return &wqe->acct[IO_WQ_ACCT_BOUND];
}

static inline struct io_wqe_acct *io_wqe_get_acct(struct io_wqe *wqe,
						  struct io_worker *worker)
{
	if (worker->flags & IO_WORKER_F_BOUND)
		return &wqe->acct[IO_WQ_ACCT_BOUND];

	return &wqe->acct[IO_WQ_ACCT_UNBOUND];
}

static void io_worker_exit(struct io_worker *worker)
{
	struct io_wqe *wqe = worker->wqe;
	struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);
	unsigned nr_workers;

	/*
	 * If we're not at zero, someone else is holding a brief reference
	 * to the worker. Wait for that to go away.
	 */
	set_current_state(TASK_INTERRUPTIBLE);
	if (!refcount_dec_and_test(&worker->ref))
		schedule();
	__set_current_state(TASK_RUNNING);

	preempt_disable();
	current->flags &= ~PF_IO_WORKER;
	if (worker->flags & IO_WORKER_F_RUNNING)
		atomic_dec(&acct->nr_running);
	if (!(worker->flags & IO_WORKER_F_BOUND))
		atomic_dec(&wqe->wq->user->processes);
	worker->flags = 0;
	preempt_enable();

	spin_lock_irq(&wqe->lock);
	hlist_nulls_del_rcu(&worker->nulls_node);
	list_del_rcu(&worker->all_list);
	if (__io_worker_unuse(wqe, worker)) {
		__release(&wqe->lock);
		spin_lock_irq(&wqe->lock);
	}
	acct->nr_workers--;
	nr_workers = wqe->acct[IO_WQ_ACCT_BOUND].nr_workers +
			wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers;
	spin_unlock_irq(&wqe->lock);

	/* all workers gone, wq exit can proceed */
	if (!nr_workers && refcount_dec_and_test(&wqe->wq->refs))
		complete(&wqe->wq->done);

	kfree_rcu(worker, rcu);
}

static inline bool io_wqe_run_queue(struct io_wqe *wqe)
	__must_hold(wqe->lock)
{
	if (!list_empty(&wqe->work_list) && !(wqe->flags & IO_WQE_FLAG_STALLED))
		return true;
	return false;
}

/*
 * Check head of free list for an available worker. If one isn't available,
 * caller must wake up the wq manager to create one.
 */
static bool io_wqe_activate_free_worker(struct io_wqe *wqe)
	__must_hold(RCU)
{
	struct hlist_nulls_node *n;
	struct io_worker *worker;

	n = rcu_dereference(hlist_nulls_first_rcu(&wqe->free_list));
	if (is_a_nulls(n))
		return false;

	worker = hlist_nulls_entry(n, struct io_worker, nulls_node);
	if (io_worker_get(worker)) {
		wake_up(&worker->wait);
		io_worker_release(worker);
		return true;
	}

	return false;
}

/*
 * We need a worker. If we find a free one, we're good. If not, and we're
 * below the max number of workers, wake up the manager to create one.
 */
static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct)
{
	bool ret;

	/*
	 * Most likely an attempt to queue unbounded work on an io_wq that
	 * wasn't setup with any unbounded workers.
	 */
	WARN_ON_ONCE(!acct->max_workers);

	rcu_read_lock();
	ret = io_wqe_activate_free_worker(wqe);
	rcu_read_unlock();

	if (!ret && acct->nr_workers < acct->max_workers)
		wake_up_process(wqe->wq->manager);
}

static void io_wqe_inc_running(struct io_wqe *wqe, struct io_worker *worker)
{
	struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);

	atomic_inc(&acct->nr_running);
}

static void io_wqe_dec_running(struct io_wqe *wqe, struct io_worker *worker)
	__must_hold(wqe->lock)
{
	struct io_wqe_acct *acct = io_wqe_get_acct(wqe, worker);

	if (atomic_dec_and_test(&acct->nr_running) && io_wqe_run_queue(wqe))
		io_wqe_wake_worker(wqe, acct);
}

static void io_worker_start(struct io_wqe *wqe, struct io_worker *worker)
{
	allow_kernel_signal(SIGINT);

	current->flags |= PF_IO_WORKER;

	worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
	worker->restore_files = current->files;
	io_wqe_inc_running(wqe, worker);
}

/*
 * Worker will start processing some work. Move it to the busy list, if
 * it's currently on the freelist
 */
static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
			     struct io_wq_work *work)
	__must_hold(wqe->lock)
{
	bool worker_bound, work_bound;

	if (worker->flags & IO_WORKER_F_FREE) {
		worker->flags &= ~IO_WORKER_F_FREE;
		hlist_nulls_del_init_rcu(&worker->nulls_node);
		hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->busy_list);
	}

	/*
	 * If worker is moving from bound to unbound (or vice versa), then
	 * ensure we update the running accounting.
	 */
	worker_bound = (worker->flags & IO_WORKER_F_BOUND) != 0;
	work_bound = (work->flags & IO_WQ_WORK_UNBOUND) == 0;
	if (worker_bound != work_bound) {
		io_wqe_dec_running(wqe, worker);
		if (work_bound) {
			worker->flags |= IO_WORKER_F_BOUND;
			wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers--;
			wqe->acct[IO_WQ_ACCT_BOUND].nr_workers++;
			atomic_dec(&wqe->wq->user->processes);
		} else {
			worker->flags &= ~IO_WORKER_F_BOUND;
			wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers++;
			wqe->acct[IO_WQ_ACCT_BOUND].nr_workers--;
			atomic_inc(&wqe->wq->user->processes);
		}
		io_wqe_inc_running(wqe, worker);
	 }
}

/*
 * No work, worker going to sleep. Move to freelist, and unuse mm if we
 * have one attached. Dropping the mm may potentially sleep, so we drop
 * the lock in that case and return success. Since the caller has to
 * retry the loop in that case (we changed task state), we don't regrab
 * the lock if we return success.
 */
static bool __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
	__must_hold(wqe->lock)
{
	if (!(worker->flags & IO_WORKER_F_FREE)) {
		worker->flags |= IO_WORKER_F_FREE;
		hlist_nulls_del_init_rcu(&worker->nulls_node);
		hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
	}

	return __io_worker_unuse(wqe, worker);
}

static struct io_wq_work *io_get_next_work(struct io_wqe *wqe, unsigned *hash)
	__must_hold(wqe->lock)
{
	struct io_wq_work *work;

	list_for_each_entry(work, &wqe->work_list, list) {
		/* not hashed, can run anytime */
		if (!(work->flags & IO_WQ_WORK_HASHED)) {
			list_del(&work->list);
			return work;
		}

		/* hashed, can run if not already running */
		*hash = work->flags >> IO_WQ_HASH_SHIFT;
		if (!(wqe->hash_map & BIT_ULL(*hash))) {
			wqe->hash_map |= BIT_ULL(*hash);
			list_del(&work->list);
			return work;
		}
	}

	return NULL;
}

static void io_worker_handle_work(struct io_worker *worker)
	__releases(wqe->lock)
{
	struct io_wq_work *work, *old_work = NULL, *put_work = NULL;
	struct io_wqe *wqe = worker->wqe;
	struct io_wq *wq = wqe->wq;

	do {
		unsigned hash = -1U;

		/*
		 * If we got some work, mark us as busy. If we didn't, but
		 * the list isn't empty, it means we stalled on hashed work.
		 * Mark us stalled so we don't keep looking for work when we
		 * can't make progress, any work completion or insertion will
		 * clear the stalled flag.
		 */
		work = io_get_next_work(wqe, &hash);
		if (work)
			__io_worker_busy(wqe, worker, work);
		else if (!list_empty(&wqe->work_list))
			wqe->flags |= IO_WQE_FLAG_STALLED;

		spin_unlock_irq(&wqe->lock);
		if (put_work && wq->put_work)
			wq->put_work(old_work);
		if (!work)
			break;
next:
		/* flush any pending signals before assigning new work */
		if (signal_pending(current))
			flush_signals(current);

		spin_lock_irq(&worker->lock);
		worker->cur_work = work;
		spin_unlock_irq(&worker->lock);

		if (work->flags & IO_WQ_WORK_CB)
			work->func(&work);

		if ((work->flags & IO_WQ_WORK_NEEDS_FILES) &&
		    current->files != work->files) {
			task_lock(current);
			current->files = work->files;
			task_unlock(current);
		}
		if ((work->flags & IO_WQ_WORK_NEEDS_USER) && !worker->mm &&
		    wq->mm && mmget_not_zero(wq->mm)) {
			use_mm(wq->mm);
			set_fs(USER_DS);
			worker->mm = wq->mm;
		}
		if (test_bit(IO_WQ_BIT_CANCEL, &wq->state))
			work->flags |= IO_WQ_WORK_CANCEL;
		if (worker->mm)
			work->flags |= IO_WQ_WORK_HAS_MM;

		if (wq->get_work && !(work->flags & IO_WQ_WORK_INTERNAL)) {
			put_work = work;
			wq->get_work(work);
		}

		old_work = work;
		work->func(&work);

		spin_lock_irq(&worker->lock);
		worker->cur_work = NULL;
		spin_unlock_irq(&worker->lock);

		spin_lock_irq(&wqe->lock);

		if (hash != -1U) {
			wqe->hash_map &= ~BIT_ULL(hash);
			wqe->flags &= ~IO_WQE_FLAG_STALLED;
		}
		if (work && work != old_work) {
			spin_unlock_irq(&wqe->lock);

			if (put_work && wq->put_work) {
				wq->put_work(put_work);
				put_work = NULL;
			}

			/* dependent work not hashed */
			hash = -1U;
			goto next;
		}
	} while (1);
}

static int io_wqe_worker(void *data)
{
	struct io_worker *worker = data;
	struct io_wqe *wqe = worker->wqe;
	struct io_wq *wq = wqe->wq;
	DEFINE_WAIT(wait);

	io_worker_start(wqe, worker);

	while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
		prepare_to_wait(&worker->wait, &wait, TASK_INTERRUPTIBLE);

		spin_lock_irq(&wqe->lock);
		if (io_wqe_run_queue(wqe)) {
			__set_current_state(TASK_RUNNING);
			io_worker_handle_work(worker);
			continue;
		}
		/* drops the lock on success, retry */
		if (__io_worker_idle(wqe, worker)) {
			__release(&wqe->lock);
			continue;
		}
		spin_unlock_irq(&wqe->lock);
		if (signal_pending(current))
			flush_signals(current);
		if (schedule_timeout(WORKER_IDLE_TIMEOUT))
			continue;
		/* timed out, exit unless we're the fixed worker */
		if (test_bit(IO_WQ_BIT_EXIT, &wq->state) ||
		    !(worker->flags & IO_WORKER_F_FIXED))
			break;
	}

	finish_wait(&worker->wait, &wait);

	if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
		spin_lock_irq(&wqe->lock);
		if (!list_empty(&wqe->work_list))
			io_worker_handle_work(worker);
		else
			spin_unlock_irq(&wqe->lock);
	}

	io_worker_exit(worker);
	return 0;
}

/*
 * Called when a worker is scheduled in. Mark us as currently running.
 */
void io_wq_worker_running(struct task_struct *tsk)
{
	struct io_worker *worker = kthread_data(tsk);
	struct io_wqe *wqe = worker->wqe;

	if (!(worker->flags & IO_WORKER_F_UP))
		return;
	if (worker->flags & IO_WORKER_F_RUNNING)
		return;
	worker->flags |= IO_WORKER_F_RUNNING;
	io_wqe_inc_running(wqe, worker);
}

/*
 * Called when worker is going to sleep. If there are no workers currently
 * running and we have work pending, wake up a free one or have the manager
 * set one up.
 */
void io_wq_worker_sleeping(struct task_struct *tsk)
{
	struct io_worker *worker = kthread_data(tsk);
	struct io_wqe *wqe = worker->wqe;

	if (!(worker->flags & IO_WORKER_F_UP))
		return;
	if (!(worker->flags & IO_WORKER_F_RUNNING))
		return;

	worker->flags &= ~IO_WORKER_F_RUNNING;

	spin_lock_irq(&wqe->lock);
	io_wqe_dec_running(wqe, worker);
	spin_unlock_irq(&wqe->lock);
}

static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
{
	struct io_wqe_acct *acct =&wqe->acct[index];
	struct io_worker *worker;

	worker = kcalloc_node(1, sizeof(*worker), GFP_KERNEL, wqe->node);
	if (!worker)
		return false;

	refcount_set(&worker->ref, 1);
	worker->nulls_node.pprev = NULL;
	init_waitqueue_head(&worker->wait);
	worker->wqe = wqe;
	spin_lock_init(&worker->lock);

	worker->task = kthread_create_on_node(io_wqe_worker, worker, wqe->node,
				"io_wqe_worker-%d/%d", index, wqe->node);
	if (IS_ERR(worker->task)) {
		kfree(worker);
		return false;
	}

	spin_lock_irq(&wqe->lock);
	hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
	list_add_tail_rcu(&worker->all_list, &wqe->all_list);
	worker->flags |= IO_WORKER_F_FREE;
	if (index == IO_WQ_ACCT_BOUND)
		worker->flags |= IO_WORKER_F_BOUND;
	if (!acct->nr_workers && (worker->flags & IO_WORKER_F_BOUND))
		worker->flags |= IO_WORKER_F_FIXED;
	acct->nr_workers++;
	spin_unlock_irq(&wqe->lock);

	if (index == IO_WQ_ACCT_UNBOUND)
		atomic_inc(&wq->user->processes);

	wake_up_process(worker->task);
	return true;
}

static inline bool io_wqe_need_worker(struct io_wqe *wqe, int index)
	__must_hold(wqe->lock)
{
	struct io_wqe_acct *acct = &wqe->acct[index];

	/* if we have available workers or no work, no need */
	if (!hlist_nulls_empty(&wqe->free_list) || !io_wqe_run_queue(wqe))
		return false;
	return acct->nr_workers < acct->max_workers;
}

/*
 * Manager thread. Tasked with creating new workers, if we need them.
 */
static int io_wq_manager(void *data)
{
	struct io_wq *wq = data;
	int i;

	/* create fixed workers */
	refcount_set(&wq->refs, wq->nr_wqes);
	for (i = 0; i < wq->nr_wqes; i++) {
		if (create_io_worker(wq, wq->wqes[i], IO_WQ_ACCT_BOUND))
			continue;
		goto err;
	}

	complete(&wq->done);

	while (!kthread_should_stop()) {
		for (i = 0; i < wq->nr_wqes; i++) {
			struct io_wqe *wqe = wq->wqes[i];
			bool fork_worker[2] = { false, false };

			spin_lock_irq(&wqe->lock);
			if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
				fork_worker[IO_WQ_ACCT_BOUND] = true;
			if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
				fork_worker[IO_WQ_ACCT_UNBOUND] = true;
			spin_unlock_irq(&wqe->lock);
			if (fork_worker[IO_WQ_ACCT_BOUND])
				create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
			if (fork_worker[IO_WQ_ACCT_UNBOUND])
				create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
		}
		set_current_state(TASK_INTERRUPTIBLE);
		schedule_timeout(HZ);
	}

	return 0;
err:
	set_bit(IO_WQ_BIT_ERROR, &wq->state);
	set_bit(IO_WQ_BIT_EXIT, &wq->state);
	if (refcount_sub_and_test(wq->nr_wqes - i, &wq->refs))
		complete(&wq->done);
	return 0;
}

static bool io_wq_can_queue(struct io_wqe *wqe, struct io_wqe_acct *acct,
			    struct io_wq_work *work)
{
	bool free_worker;

	if (!(work->flags & IO_WQ_WORK_UNBOUND))
		return true;
	if (atomic_read(&acct->nr_running))
		return true;

	rcu_read_lock();
	free_worker = !hlist_nulls_empty(&wqe->free_list);
	rcu_read_unlock();
	if (free_worker)
		return true;

	if (atomic_read(&wqe->wq->user->processes) >= acct->max_workers &&
	    !(capable(CAP_SYS_RESOURCE) || capable(CAP_SYS_ADMIN)))
		return false;

	return true;
}

static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
{
	struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
	unsigned long flags;

	/*
	 * Do early check to see if we need a new unbound worker, and if we do,
	 * if we're allowed to do so. This isn't 100% accurate as there's a
	 * gap between this check and incrementing the value, but that's OK.
	 * It's close enough to not be an issue, fork() has the same delay.
	 */
	if (unlikely(!io_wq_can_queue(wqe, acct, work))) {
		work->flags |= IO_WQ_WORK_CANCEL;
		work->func(&work);
		return;
	}

	spin_lock_irqsave(&wqe->lock, flags);
	list_add_tail(&work->list, &wqe->work_list);
	wqe->flags &= ~IO_WQE_FLAG_STALLED;
	spin_unlock_irqrestore(&wqe->lock, flags);

	if (!atomic_read(&acct->nr_running))
		io_wqe_wake_worker(wqe, acct);
}

void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work)
{
	struct io_wqe *wqe = wq->wqes[numa_node_id()];

	io_wqe_enqueue(wqe, work);
}

/*
 * Enqueue work, hashed by some key. Work items that hash to the same value
 * will not be done in parallel. Used to limit concurrent writes, generally
 * hashed by inode.
 */
void io_wq_enqueue_hashed(struct io_wq *wq, struct io_wq_work *work, void *val)
{
	struct io_wqe *wqe = wq->wqes[numa_node_id()];
	unsigned bit;


	bit = hash_ptr(val, IO_WQ_HASH_ORDER);
	work->flags |= (IO_WQ_WORK_HASHED | (bit << IO_WQ_HASH_SHIFT));
	io_wqe_enqueue(wqe, work);
}

static bool io_wqe_worker_send_sig(struct io_worker *worker, void *data)
{
	send_sig(SIGINT, worker->task, 1);
	return false;
}

/*
 * Iterate the passed in list and call the specific function for each
 * worker that isn't exiting
 */
static bool io_wq_for_each_worker(struct io_wqe *wqe,
				  bool (*func)(struct io_worker *, void *),
				  void *data)
{
	struct io_worker *worker;
	bool ret = false;

	list_for_each_entry_rcu(worker, &wqe->all_list, all_list) {
		if (io_worker_get(worker)) {
			ret = func(worker, data);
			io_worker_release(worker);
			if (ret)
				break;
		}
	}

	return ret;
}

void io_wq_cancel_all(struct io_wq *wq)
{
	int i;

	set_bit(IO_WQ_BIT_CANCEL, &wq->state);

	/*
	 * Browse both lists, as there's a gap between handing work off
	 * to a worker and the worker putting itself on the busy_list
	 */
	rcu_read_lock();
	for (i = 0; i < wq->nr_wqes; i++) {
		struct io_wqe *wqe = wq->wqes[i];

		io_wq_for_each_worker(wqe, io_wqe_worker_send_sig, NULL);
	}
	rcu_read_unlock();
}

struct io_cb_cancel_data {
	struct io_wqe *wqe;
	work_cancel_fn *cancel;
	void *caller_data;
};

static bool io_work_cancel(struct io_worker *worker, void *cancel_data)
{
	struct io_cb_cancel_data *data = cancel_data;
	unsigned long flags;
	bool ret = false;

	/*
	 * Hold the lock to avoid ->cur_work going out of scope, caller
	 * may dereference the passed in work.
	 */
	spin_lock_irqsave(&worker->lock, flags);
	if (worker->cur_work &&
	    data->cancel(worker->cur_work, data->caller_data)) {
		send_sig(SIGINT, worker->task, 1);
		ret = true;
	}
	spin_unlock_irqrestore(&worker->lock, flags);

	return ret;
}

static enum io_wq_cancel io_wqe_cancel_cb_work(struct io_wqe *wqe,
					       work_cancel_fn *cancel,
					       void *cancel_data)
{
	struct io_cb_cancel_data data = {
		.wqe = wqe,
		.cancel = cancel,
		.caller_data = cancel_data,
	};
	struct io_wq_work *work;
	unsigned long flags;
	bool found = false;

	spin_lock_irqsave(&wqe->lock, flags);
	list_for_each_entry(work, &wqe->work_list, list) {
		if (cancel(work, cancel_data)) {
			list_del(&work->list);
			found = true;
			break;
		}
	}
	spin_unlock_irqrestore(&wqe->lock, flags);

	if (found) {
		work->flags |= IO_WQ_WORK_CANCEL;
		work->func(&work);
		return IO_WQ_CANCEL_OK;
	}

	rcu_read_lock();
	found = io_wq_for_each_worker(wqe, io_work_cancel, &data);
	rcu_read_unlock();
	return found ? IO_WQ_CANCEL_RUNNING : IO_WQ_CANCEL_NOTFOUND;
}

enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
				  void *data)
{
	enum io_wq_cancel ret = IO_WQ_CANCEL_NOTFOUND;
	int i;

	for (i = 0; i < wq->nr_wqes; i++) {
		struct io_wqe *wqe = wq->wqes[i];

		ret = io_wqe_cancel_cb_work(wqe, cancel, data);
		if (ret != IO_WQ_CANCEL_NOTFOUND)
			break;
	}

	return ret;
}

static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
{
	struct io_wq_work *work = data;
	unsigned long flags;
	bool ret = false;

	if (worker->cur_work != work)
		return false;

	spin_lock_irqsave(&worker->lock, flags);
	if (worker->cur_work == work) {
		send_sig(SIGINT, worker->task, 1);
		ret = true;
	}
	spin_unlock_irqrestore(&worker->lock, flags);

	return ret;
}

static enum io_wq_cancel io_wqe_cancel_work(struct io_wqe *wqe,
					    struct io_wq_work *cwork)
{
	struct io_wq_work *work;
	unsigned long flags;
	bool found = false;

	cwork->flags |= IO_WQ_WORK_CANCEL;

	/*
	 * First check pending list, if we're lucky we can just remove it
	 * from there. CANCEL_OK means that the work is returned as-new,
	 * no completion will be posted for it.
	 */
	spin_lock_irqsave(&wqe->lock, flags);
	list_for_each_entry(work, &wqe->work_list, list) {
		if (work == cwork) {
			list_del(&work->list);
			found = true;
			break;
		}
	}
	spin_unlock_irqrestore(&wqe->lock, flags);

	if (found) {
		work->flags |= IO_WQ_WORK_CANCEL;
		work->func(&work);
		return IO_WQ_CANCEL_OK;
	}

	/*
	 * Now check if a free (going busy) or busy worker has the work
	 * currently running. If we find it there, we'll return CANCEL_RUNNING
	 * as an indication that we attempte to signal cancellation. The
	 * completion will run normally in this case.
	 */
	rcu_read_lock();
	found = io_wq_for_each_worker(wqe, io_wq_worker_cancel, cwork);
	rcu_read_unlock();
	return found ? IO_WQ_CANCEL_RUNNING : IO_WQ_CANCEL_NOTFOUND;
}

enum io_wq_cancel io_wq_cancel_work(struct io_wq *wq, struct io_wq_work *cwork)
{
	enum io_wq_cancel ret = IO_WQ_CANCEL_NOTFOUND;
	int i;

	for (i = 0; i < wq->nr_wqes; i++) {
		struct io_wqe *wqe = wq->wqes[i];

		ret = io_wqe_cancel_work(wqe, cwork);
		if (ret != IO_WQ_CANCEL_NOTFOUND)
			break;
	}

	return ret;
}

struct io_wq_flush_data {
	struct io_wq_work work;
	struct completion done;
};

static void io_wq_flush_func(struct io_wq_work **workptr)
{
	struct io_wq_work *work = *workptr;
	struct io_wq_flush_data *data;

	data = container_of(work, struct io_wq_flush_data, work);
	complete(&data->done);
}

/*
 * Doesn't wait for previously queued work to finish. When this completes,
 * it just means that previously queued work was started.
 */
void io_wq_flush(struct io_wq *wq)
{
	struct io_wq_flush_data data;
	int i;

	for (i = 0; i < wq->nr_wqes; i++) {
		struct io_wqe *wqe = wq->wqes[i];

		init_completion(&data.done);
		INIT_IO_WORK(&data.work, io_wq_flush_func);
		data.work.flags |= IO_WQ_WORK_INTERNAL;
		io_wqe_enqueue(wqe, &data.work);
		wait_for_completion(&data.done);
	}
}

struct io_wq *io_wq_create(unsigned bounded, struct mm_struct *mm,
			   struct user_struct *user, get_work_fn *get_work,
			   put_work_fn *put_work)
{
	int ret = -ENOMEM, i, node;
	struct io_wq *wq;

	wq = kcalloc(1, sizeof(*wq), GFP_KERNEL);
	if (!wq)
		return ERR_PTR(-ENOMEM);

	wq->nr_wqes = num_online_nodes();
	wq->wqes = kcalloc(wq->nr_wqes, sizeof(struct io_wqe *), GFP_KERNEL);
	if (!wq->wqes) {
		kfree(wq);
		return ERR_PTR(-ENOMEM);
	}

	wq->get_work = get_work;
	wq->put_work = put_work;

	/* caller must already hold a reference to this */
	wq->user = user;

	i = 0;
	for_each_online_node(node) {
		struct io_wqe *wqe;

		wqe = kcalloc_node(1, sizeof(struct io_wqe), GFP_KERNEL, node);
		if (!wqe)
			break;
		wq->wqes[i] = wqe;
		wqe->node = node;
		wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
		atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
		if (user) {
			wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
					task_rlimit(current, RLIMIT_NPROC);
		}
		atomic_set(&wqe->acct[IO_WQ_ACCT_UNBOUND].nr_running, 0);
		wqe->node = node;
		wqe->wq = wq;
		spin_lock_init(&wqe->lock);
		INIT_LIST_HEAD(&wqe->work_list);
		INIT_HLIST_NULLS_HEAD(&wqe->free_list, 0);
		INIT_HLIST_NULLS_HEAD(&wqe->busy_list, 1);
		INIT_LIST_HEAD(&wqe->all_list);

		i++;
	}

	init_completion(&wq->done);

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

	/* caller must have already done mmgrab() on this mm */
	wq->mm = mm;

	wq->manager = kthread_create(io_wq_manager, wq, "io_wq_manager");
	if (!IS_ERR(wq->manager)) {
		wake_up_process(wq->manager);
		wait_for_completion(&wq->done);
		if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
			ret = -ENOMEM;
			goto err;
		}
		reinit_completion(&wq->done);
		return wq;
	}

	ret = PTR_ERR(wq->manager);
	complete(&wq->done);
err:
	for (i = 0; i < wq->nr_wqes; i++)
		kfree(wq->wqes[i]);
	kfree(wq->wqes);
	kfree(wq);
	return ERR_PTR(ret);
}

static bool io_wq_worker_wake(struct io_worker *worker, void *data)
{
	wake_up_process(worker->task);
	return false;
}

void io_wq_destroy(struct io_wq *wq)
{
	int i;

	set_bit(IO_WQ_BIT_EXIT, &wq->state);
	if (wq->manager)
		kthread_stop(wq->manager);

	rcu_read_lock();
	for (i = 0; i < wq->nr_wqes; i++) {
		struct io_wqe *wqe = wq->wqes[i];

		if (!wqe)
			continue;
		io_wq_for_each_worker(wqe, io_wq_worker_wake, NULL);
	}
	rcu_read_unlock();

	wait_for_completion(&wq->done);

	for (i = 0; i < wq->nr_wqes; i++)
		kfree(wq->wqes[i]);
	kfree(wq->wqes);
	kfree(wq);
}