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
|
// SPDX-License-Identifier: GPL-2.0
#include "bcachefs.h"
#include "btree_key_cache.h"
#include "error.h"
#include "journal.h"
#include "journal_io.h"
#include "journal_reclaim.h"
#include "replicas.h"
#include "super.h"
#include "trace.h"
#include <linux/kthread.h>
#include <linux/sched/mm.h>
/* Free space calculations: */
static unsigned journal_space_from(struct journal_device *ja,
enum journal_space_from from)
{
switch (from) {
case journal_space_discarded:
return ja->discard_idx;
case journal_space_clean_ondisk:
return ja->dirty_idx_ondisk;
case journal_space_clean:
return ja->dirty_idx;
default:
BUG();
}
}
unsigned bch2_journal_dev_buckets_available(struct journal *j,
struct journal_device *ja,
enum journal_space_from from)
{
unsigned available = (journal_space_from(ja, from) -
ja->cur_idx - 1 + ja->nr) % ja->nr;
/*
* Don't use the last bucket unless writing the new last_seq
* will make another bucket available:
*/
if (available && ja->dirty_idx_ondisk == ja->dirty_idx)
--available;
return available;
}
static void journal_set_remaining(struct journal *j, unsigned u64s_remaining)
{
union journal_preres_state old, new;
u64 v = atomic64_read(&j->prereserved.counter);
do {
old.v = new.v = v;
new.remaining = u64s_remaining;
} while ((v = atomic64_cmpxchg(&j->prereserved.counter,
old.v, new.v)) != old.v);
}
static inline unsigned get_unwritten_sectors(struct journal *j, unsigned *idx)
{
unsigned sectors = 0;
while (!sectors && *idx != j->reservations.idx) {
sectors = j->buf[*idx].sectors;
*idx = (*idx + 1) & JOURNAL_BUF_MASK;
}
return sectors;
}
static struct journal_space
journal_dev_space_available(struct journal *j, struct bch_dev *ca,
enum journal_space_from from)
{
struct journal_device *ja = &ca->journal;
unsigned sectors, buckets, unwritten, idx = j->reservations.unwritten_idx;
if (from == journal_space_total)
return (struct journal_space) {
.next_entry = ca->mi.bucket_size,
.total = ca->mi.bucket_size * ja->nr,
};
buckets = bch2_journal_dev_buckets_available(j, ja, from);
sectors = ja->sectors_free;
/*
* We that we don't allocate the space for a journal entry
* until we write it out - thus, account for it here:
*/
while ((unwritten = get_unwritten_sectors(j, &idx))) {
/* entry won't fit on this device, skip: */
if (unwritten > ca->mi.bucket_size)
continue;
if (unwritten >= sectors) {
if (!buckets) {
sectors = 0;
break;
}
buckets--;
sectors = ca->mi.bucket_size;
}
sectors -= unwritten;
}
if (sectors < ca->mi.bucket_size && buckets) {
buckets--;
sectors = ca->mi.bucket_size;
}
return (struct journal_space) {
.next_entry = sectors,
.total = sectors + buckets * ca->mi.bucket_size,
};
}
static struct journal_space __journal_space_available(struct journal *j, unsigned nr_devs_want,
enum journal_space_from from)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct bch_dev *ca;
unsigned i, pos, nr_devs = 0;
struct journal_space space, dev_space[BCH_SB_MEMBERS_MAX];
BUG_ON(nr_devs_want > ARRAY_SIZE(dev_space));
rcu_read_lock();
for_each_member_device_rcu(ca, c, i,
&c->rw_devs[BCH_DATA_journal]) {
if (!ca->journal.nr)
continue;
space = journal_dev_space_available(j, ca, from);
if (!space.next_entry)
continue;
for (pos = 0; pos < nr_devs; pos++)
if (space.total > dev_space[pos].total)
break;
array_insert_item(dev_space, nr_devs, pos, space);
}
rcu_read_unlock();
if (nr_devs < nr_devs_want)
return (struct journal_space) { 0, 0 };
/*
* We sorted largest to smallest, and we want the smallest out of the
* @nr_devs_want largest devices:
*/
return dev_space[nr_devs_want - 1];
}
void bch2_journal_space_available(struct journal *j)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct bch_dev *ca;
unsigned clean, clean_ondisk, total;
s64 u64s_remaining = 0;
unsigned max_entry_size = min(j->buf[0].buf_size >> 9,
j->buf[1].buf_size >> 9);
unsigned i, nr_online = 0, nr_devs_want;
bool can_discard = false;
int ret = 0;
lockdep_assert_held(&j->lock);
rcu_read_lock();
for_each_member_device_rcu(ca, c, i,
&c->rw_devs[BCH_DATA_journal]) {
struct journal_device *ja = &ca->journal;
if (!ja->nr)
continue;
while (ja->dirty_idx != ja->cur_idx &&
ja->bucket_seq[ja->dirty_idx] < journal_last_seq(j))
ja->dirty_idx = (ja->dirty_idx + 1) % ja->nr;
while (ja->dirty_idx_ondisk != ja->dirty_idx &&
ja->bucket_seq[ja->dirty_idx_ondisk] < j->last_seq_ondisk)
ja->dirty_idx_ondisk = (ja->dirty_idx_ondisk + 1) % ja->nr;
if (ja->discard_idx != ja->dirty_idx_ondisk)
can_discard = true;
max_entry_size = min_t(unsigned, max_entry_size, ca->mi.bucket_size);
nr_online++;
}
rcu_read_unlock();
j->can_discard = can_discard;
if (nr_online < c->opts.metadata_replicas_required) {
ret = cur_entry_insufficient_devices;
goto out;
}
nr_devs_want = min_t(unsigned, nr_online, c->opts.metadata_replicas);
for (i = 0; i < journal_space_nr; i++)
j->space[i] = __journal_space_available(j, nr_devs_want, i);
clean_ondisk = j->space[journal_space_clean_ondisk].total;
clean = j->space[journal_space_clean].total;
total = j->space[journal_space_total].total;
if (!clean_ondisk &&
j->reservations.idx ==
j->reservations.unwritten_idx) {
struct printbuf buf = PRINTBUF;
__bch2_journal_debug_to_text(&buf, j);
bch_err(c, "journal stuck\n%s", buf.buf);
printbuf_exit(&buf);
bch2_fatal_error(c);
ret = cur_entry_journal_stuck;
} else if (!j->space[journal_space_discarded].next_entry)
ret = cur_entry_journal_full;
else if (!fifo_free(&j->pin))
ret = cur_entry_journal_pin_full;
if ((j->space[journal_space_clean_ondisk].next_entry <
j->space[journal_space_clean_ondisk].total) &&
(clean - clean_ondisk <= total / 8) &&
(clean_ondisk * 2 > clean ))
set_bit(JOURNAL_MAY_SKIP_FLUSH, &j->flags);
else
clear_bit(JOURNAL_MAY_SKIP_FLUSH, &j->flags);
u64s_remaining = (u64) clean << 6;
u64s_remaining -= (u64) total << 3;
u64s_remaining = max(0LL, u64s_remaining);
u64s_remaining /= 4;
u64s_remaining = min_t(u64, u64s_remaining, U32_MAX);
out:
j->cur_entry_sectors = !ret ? j->space[journal_space_discarded].next_entry : 0;
j->cur_entry_error = ret;
journal_set_remaining(j, u64s_remaining);
journal_check_may_get_unreserved(j);
if (!ret)
journal_wake(j);
}
/* Discards - last part of journal reclaim: */
static bool should_discard_bucket(struct journal *j, struct journal_device *ja)
{
bool ret;
spin_lock(&j->lock);
ret = ja->discard_idx != ja->dirty_idx_ondisk;
spin_unlock(&j->lock);
return ret;
}
/*
* Advance ja->discard_idx as long as it points to buckets that are no longer
* dirty, issuing discards if necessary:
*/
void bch2_journal_do_discards(struct journal *j)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct bch_dev *ca;
unsigned iter;
mutex_lock(&j->discard_lock);
for_each_rw_member(ca, c, iter) {
struct journal_device *ja = &ca->journal;
while (should_discard_bucket(j, ja)) {
if (!c->opts.nochanges &&
ca->mi.discard &&
bdev_max_discard_sectors(ca->disk_sb.bdev))
blkdev_issue_discard(ca->disk_sb.bdev,
bucket_to_sector(ca,
ja->buckets[ja->discard_idx]),
ca->mi.bucket_size, GFP_NOIO);
spin_lock(&j->lock);
ja->discard_idx = (ja->discard_idx + 1) % ja->nr;
bch2_journal_space_available(j);
spin_unlock(&j->lock);
}
}
mutex_unlock(&j->discard_lock);
}
/*
* Journal entry pinning - machinery for holding a reference on a given journal
* entry, holding it open to ensure it gets replayed during recovery:
*/
static void bch2_journal_reclaim_fast(struct journal *j)
{
struct journal_entry_pin_list temp;
bool popped = false;
lockdep_assert_held(&j->lock);
/*
* Unpin journal entries whose reference counts reached zero, meaning
* all btree nodes got written out
*/
while (!fifo_empty(&j->pin) &&
!atomic_read(&fifo_peek_front(&j->pin).count)) {
BUG_ON(!list_empty(&fifo_peek_front(&j->pin).list));
BUG_ON(!list_empty(&fifo_peek_front(&j->pin).flushed));
BUG_ON(!fifo_pop(&j->pin, temp));
popped = true;
}
if (popped)
bch2_journal_space_available(j);
}
void __bch2_journal_pin_put(struct journal *j, u64 seq)
{
struct journal_entry_pin_list *pin_list = journal_seq_pin(j, seq);
if (atomic_dec_and_test(&pin_list->count))
bch2_journal_reclaim_fast(j);
}
void bch2_journal_pin_put(struct journal *j, u64 seq)
{
struct journal_entry_pin_list *pin_list = journal_seq_pin(j, seq);
if (atomic_dec_and_test(&pin_list->count)) {
spin_lock(&j->lock);
bch2_journal_reclaim_fast(j);
spin_unlock(&j->lock);
}
}
static inline void __journal_pin_drop(struct journal *j,
struct journal_entry_pin *pin)
{
struct journal_entry_pin_list *pin_list;
if (!journal_pin_active(pin))
return;
if (j->flush_in_progress == pin)
j->flush_in_progress_dropped = true;
pin_list = journal_seq_pin(j, pin->seq);
pin->seq = 0;
list_del_init(&pin->list);
/*
* Unpinning a journal entry make make journal_next_bucket() succeed, if
* writing a new last_seq will now make another bucket available:
*/
if (atomic_dec_and_test(&pin_list->count) &&
pin_list == &fifo_peek_front(&j->pin))
bch2_journal_reclaim_fast(j);
else if (fifo_used(&j->pin) == 1 &&
atomic_read(&pin_list->count) == 1)
journal_wake(j);
}
void bch2_journal_pin_drop(struct journal *j,
struct journal_entry_pin *pin)
{
spin_lock(&j->lock);
__journal_pin_drop(j, pin);
spin_unlock(&j->lock);
}
void bch2_journal_pin_set(struct journal *j, u64 seq,
struct journal_entry_pin *pin,
journal_pin_flush_fn flush_fn)
{
struct journal_entry_pin_list *pin_list;
spin_lock(&j->lock);
if (seq < journal_last_seq(j)) {
/*
* bch2_journal_pin_copy() raced with bch2_journal_pin_drop() on
* the src pin - with the pin dropped, the entry to pin might no
* longer to exist, but that means there's no longer anything to
* copy and we can bail out here:
*/
spin_unlock(&j->lock);
return;
}
pin_list = journal_seq_pin(j, seq);
__journal_pin_drop(j, pin);
atomic_inc(&pin_list->count);
pin->seq = seq;
pin->flush = flush_fn;
if (flush_fn == bch2_btree_key_cache_journal_flush)
list_add(&pin->list, &pin_list->key_cache_list);
else if (flush_fn)
list_add(&pin->list, &pin_list->list);
else
list_add(&pin->list, &pin_list->flushed);
spin_unlock(&j->lock);
/*
* If the journal is currently full, we might want to call flush_fn
* immediately:
*/
journal_wake(j);
}
/**
* bch2_journal_pin_flush: ensure journal pin callback is no longer running
*/
void bch2_journal_pin_flush(struct journal *j, struct journal_entry_pin *pin)
{
BUG_ON(journal_pin_active(pin));
wait_event(j->pin_flush_wait, j->flush_in_progress != pin);
}
/*
* Journal reclaim: flush references to open journal entries to reclaim space in
* the journal
*
* May be done by the journal code in the background as needed to free up space
* for more journal entries, or as part of doing a clean shutdown, or to migrate
* data off of a specific device:
*/
static struct journal_entry_pin *
journal_get_next_pin(struct journal *j,
bool get_any,
bool get_key_cache,
u64 max_seq, u64 *seq)
{
struct journal_entry_pin_list *pin_list;
struct journal_entry_pin *ret = NULL;
fifo_for_each_entry_ptr(pin_list, &j->pin, *seq) {
if (*seq > max_seq && !get_any && !get_key_cache)
break;
if (*seq <= max_seq || get_any) {
ret = list_first_entry_or_null(&pin_list->list,
struct journal_entry_pin, list);
if (ret)
return ret;
}
if (*seq <= max_seq || get_any || get_key_cache) {
ret = list_first_entry_or_null(&pin_list->key_cache_list,
struct journal_entry_pin, list);
if (ret)
return ret;
}
}
return NULL;
}
/* returns true if we did work */
static size_t journal_flush_pins(struct journal *j, u64 seq_to_flush,
unsigned min_any,
unsigned min_key_cache)
{
struct journal_entry_pin *pin;
size_t nr_flushed = 0;
journal_pin_flush_fn flush_fn;
u64 seq;
int err;
lockdep_assert_held(&j->reclaim_lock);
while (1) {
cond_resched();
j->last_flushed = jiffies;
spin_lock(&j->lock);
pin = journal_get_next_pin(j,
min_any != 0,
min_key_cache != 0,
seq_to_flush, &seq);
if (pin) {
BUG_ON(j->flush_in_progress);
j->flush_in_progress = pin;
j->flush_in_progress_dropped = false;
flush_fn = pin->flush;
}
spin_unlock(&j->lock);
if (!pin)
break;
if (min_key_cache && pin->flush == bch2_btree_key_cache_journal_flush)
min_key_cache--;
if (min_any)
min_any--;
err = flush_fn(j, pin, seq);
spin_lock(&j->lock);
/* Pin might have been dropped or rearmed: */
if (likely(!err && !j->flush_in_progress_dropped))
list_move(&pin->list, &journal_seq_pin(j, seq)->flushed);
j->flush_in_progress = NULL;
j->flush_in_progress_dropped = false;
spin_unlock(&j->lock);
wake_up(&j->pin_flush_wait);
if (err)
break;
nr_flushed++;
}
return nr_flushed;
}
static u64 journal_seq_to_flush(struct journal *j)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct bch_dev *ca;
u64 seq_to_flush = 0;
unsigned iter;
spin_lock(&j->lock);
for_each_rw_member(ca, c, iter) {
struct journal_device *ja = &ca->journal;
unsigned nr_buckets, bucket_to_flush;
if (!ja->nr)
continue;
/* Try to keep the journal at most half full: */
nr_buckets = ja->nr / 2;
/* And include pre-reservations: */
nr_buckets += DIV_ROUND_UP(j->prereserved.reserved,
(ca->mi.bucket_size << 6) -
journal_entry_overhead(j));
nr_buckets = min(nr_buckets, ja->nr);
bucket_to_flush = (ja->cur_idx + nr_buckets) % ja->nr;
seq_to_flush = max(seq_to_flush,
ja->bucket_seq[bucket_to_flush]);
}
/* Also flush if the pin fifo is more than half full */
seq_to_flush = max_t(s64, seq_to_flush,
(s64) journal_cur_seq(j) -
(j->pin.size >> 1));
spin_unlock(&j->lock);
return seq_to_flush;
}
/**
* bch2_journal_reclaim - free up journal buckets
*
* Background journal reclaim writes out btree nodes. It should be run
* early enough so that we never completely run out of journal buckets.
*
* High watermarks for triggering background reclaim:
* - FIFO has fewer than 512 entries left
* - fewer than 25% journal buckets free
*
* Background reclaim runs until low watermarks are reached:
* - FIFO has more than 1024 entries left
* - more than 50% journal buckets free
*
* As long as a reclaim can complete in the time it takes to fill up
* 512 journal entries or 25% of all journal buckets, then
* journal_next_bucket() should not stall.
*/
static int __bch2_journal_reclaim(struct journal *j, bool direct)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
bool kthread = (current->flags & PF_KTHREAD) != 0;
u64 seq_to_flush;
size_t min_nr, min_key_cache, nr_flushed;
unsigned flags;
int ret = 0;
/*
* We can't invoke memory reclaim while holding the reclaim_lock -
* journal reclaim is required to make progress for memory reclaim
* (cleaning the caches), so we can't get stuck in memory reclaim while
* we're holding the reclaim lock:
*/
lockdep_assert_held(&j->reclaim_lock);
flags = memalloc_noreclaim_save();
do {
if (kthread && kthread_should_stop())
break;
if (bch2_journal_error(j)) {
ret = -EIO;
break;
}
bch2_journal_do_discards(j);
seq_to_flush = journal_seq_to_flush(j);
min_nr = 0;
/*
* If it's been longer than j->reclaim_delay_ms since we last flushed,
* make sure to flush at least one journal pin:
*/
if (time_after(jiffies, j->last_flushed +
msecs_to_jiffies(c->opts.journal_reclaim_delay)))
min_nr = 1;
if (j->prereserved.reserved * 4 > j->prereserved.remaining)
min_nr = 1;
if (fifo_free(&j->pin) <= 32)
min_nr = 1;
trace_journal_reclaim_start(c,
min_nr,
j->prereserved.reserved,
j->prereserved.remaining,
atomic_read(&c->btree_cache.dirty),
c->btree_cache.used,
atomic_long_read(&c->btree_key_cache.nr_dirty),
atomic_long_read(&c->btree_key_cache.nr_keys));
min_key_cache = min(bch2_nr_btree_keys_need_flush(c), (size_t) 128);
nr_flushed = journal_flush_pins(j, seq_to_flush,
min_nr, min_key_cache);
if (direct)
j->nr_direct_reclaim += nr_flushed;
else
j->nr_background_reclaim += nr_flushed;
trace_journal_reclaim_finish(c, nr_flushed);
if (nr_flushed)
wake_up(&j->reclaim_wait);
} while ((min_nr || min_key_cache) && nr_flushed && !direct);
memalloc_noreclaim_restore(flags);
return ret;
}
int bch2_journal_reclaim(struct journal *j)
{
return __bch2_journal_reclaim(j, true);
}
static int bch2_journal_reclaim_thread(void *arg)
{
struct journal *j = arg;
struct bch_fs *c = container_of(j, struct bch_fs, journal);
unsigned long delay, now;
int ret = 0;
set_freezable();
j->last_flushed = jiffies;
while (!ret && !kthread_should_stop()) {
j->reclaim_kicked = false;
mutex_lock(&j->reclaim_lock);
ret = __bch2_journal_reclaim(j, false);
mutex_unlock(&j->reclaim_lock);
now = jiffies;
delay = msecs_to_jiffies(c->opts.journal_reclaim_delay);
j->next_reclaim = j->last_flushed + delay;
if (!time_in_range(j->next_reclaim, now, now + delay))
j->next_reclaim = now + delay;
while (1) {
set_current_state(TASK_INTERRUPTIBLE|TASK_FREEZABLE);
if (kthread_should_stop())
break;
if (j->reclaim_kicked)
break;
if (time_after_eq(jiffies, j->next_reclaim))
break;
schedule_timeout(j->next_reclaim - jiffies);
}
__set_current_state(TASK_RUNNING);
}
return 0;
}
void bch2_journal_reclaim_stop(struct journal *j)
{
struct task_struct *p = j->reclaim_thread;
j->reclaim_thread = NULL;
if (p) {
kthread_stop(p);
put_task_struct(p);
}
}
int bch2_journal_reclaim_start(struct journal *j)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct task_struct *p;
if (j->reclaim_thread)
return 0;
p = kthread_create(bch2_journal_reclaim_thread, j,
"bch-reclaim/%s", c->name);
if (IS_ERR(p)) {
bch_err(c, "error creating journal reclaim thread: %li", PTR_ERR(p));
return PTR_ERR(p);
}
get_task_struct(p);
j->reclaim_thread = p;
wake_up_process(p);
return 0;
}
static int journal_flush_done(struct journal *j, u64 seq_to_flush,
bool *did_work)
{
int ret;
ret = bch2_journal_error(j);
if (ret)
return ret;
mutex_lock(&j->reclaim_lock);
if (journal_flush_pins(j, seq_to_flush, 0, 0))
*did_work = true;
spin_lock(&j->lock);
/*
* If journal replay hasn't completed, the unreplayed journal entries
* hold refs on their corresponding sequence numbers
*/
ret = !test_bit(JOURNAL_REPLAY_DONE, &j->flags) ||
journal_last_seq(j) > seq_to_flush ||
(fifo_used(&j->pin) == 1 &&
atomic_read(&fifo_peek_front(&j->pin).count) == 1);
spin_unlock(&j->lock);
mutex_unlock(&j->reclaim_lock);
return ret;
}
bool bch2_journal_flush_pins(struct journal *j, u64 seq_to_flush)
{
bool did_work = false;
if (!test_bit(JOURNAL_STARTED, &j->flags))
return false;
closure_wait_event(&j->async_wait,
journal_flush_done(j, seq_to_flush, &did_work));
return did_work;
}
int bch2_journal_flush_device_pins(struct journal *j, int dev_idx)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct journal_entry_pin_list *p;
u64 iter, seq = 0;
int ret = 0;
spin_lock(&j->lock);
fifo_for_each_entry_ptr(p, &j->pin, iter)
if (dev_idx >= 0
? bch2_dev_list_has_dev(p->devs, dev_idx)
: p->devs.nr < c->opts.metadata_replicas)
seq = iter;
spin_unlock(&j->lock);
bch2_journal_flush_pins(j, seq);
ret = bch2_journal_error(j);
if (ret)
return ret;
mutex_lock(&c->replicas_gc_lock);
bch2_replicas_gc_start(c, 1 << BCH_DATA_journal);
seq = 0;
spin_lock(&j->lock);
while (!ret && seq < j->pin.back) {
struct bch_replicas_padded replicas;
seq = max(seq, journal_last_seq(j));
bch2_devlist_to_replicas(&replicas.e, BCH_DATA_journal,
journal_seq_pin(j, seq)->devs);
seq++;
spin_unlock(&j->lock);
ret = bch2_mark_replicas(c, &replicas.e);
spin_lock(&j->lock);
}
spin_unlock(&j->lock);
ret = bch2_replicas_gc_end(c, ret);
mutex_unlock(&c->replicas_gc_lock);
return ret;
}
|