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
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
|
// SPDX-License-Identifier: GPL-2.0
#include "bcachefs.h"
#include "alloc_foreground.h"
#include "buckets.h"
#include "checksum.h"
#include "error.h"
#include "journal.h"
#include "journal_io.h"
#include "journal_reclaim.h"
#include "replicas.h"
#include "trace.h"
struct journal_list {
struct closure cl;
struct mutex lock;
struct list_head *head;
int ret;
};
#define JOURNAL_ENTRY_ADD_OK 0
#define JOURNAL_ENTRY_ADD_OUT_OF_RANGE 5
/*
* Given a journal entry we just read, add it to the list of journal entries to
* be replayed:
*/
static int journal_entry_add(struct bch_fs *c, struct bch_dev *ca,
struct journal_list *jlist, struct jset *j)
{
struct journal_replay *i, *pos;
struct list_head *where;
size_t bytes = vstruct_bytes(j);
__le64 last_seq;
int ret;
last_seq = !list_empty(jlist->head)
? list_last_entry(jlist->head, struct journal_replay,
list)->j.last_seq
: 0;
/* Is this entry older than the range we need? */
if (le64_to_cpu(j->seq) < le64_to_cpu(last_seq)) {
ret = JOURNAL_ENTRY_ADD_OUT_OF_RANGE;
goto out;
}
/* Drop entries we don't need anymore */
list_for_each_entry_safe(i, pos, jlist->head, list) {
if (le64_to_cpu(i->j.seq) >= le64_to_cpu(j->last_seq))
break;
list_del(&i->list);
kvpfree(i, offsetof(struct journal_replay, j) +
vstruct_bytes(&i->j));
}
list_for_each_entry_reverse(i, jlist->head, list) {
/* Duplicate? */
if (le64_to_cpu(j->seq) == le64_to_cpu(i->j.seq)) {
fsck_err_on(bytes != vstruct_bytes(&i->j) ||
memcmp(j, &i->j, bytes), c,
"found duplicate but non identical journal entries (seq %llu)",
le64_to_cpu(j->seq));
goto found;
}
if (le64_to_cpu(j->seq) > le64_to_cpu(i->j.seq)) {
where = &i->list;
goto add;
}
}
where = jlist->head;
add:
i = kvpmalloc(offsetof(struct journal_replay, j) + bytes, GFP_KERNEL);
if (!i) {
ret = -ENOMEM;
goto out;
}
list_add(&i->list, where);
i->devs.nr = 0;
unsafe_memcpy(&i->j, j, bytes, "embedded variable length struct");
found:
if (!bch2_dev_list_has_dev(i->devs, ca->dev_idx))
bch2_dev_list_add_dev(&i->devs, ca->dev_idx);
else
fsck_err_on(1, c, "duplicate journal entries on same device");
ret = JOURNAL_ENTRY_ADD_OK;
out:
fsck_err:
return ret;
}
static struct nonce journal_nonce(const struct jset *jset)
{
return (struct nonce) {{
[0] = 0,
[1] = ((__le32 *) &jset->seq)[0],
[2] = ((__le32 *) &jset->seq)[1],
[3] = BCH_NONCE_JOURNAL,
}};
}
/* this fills in a range with empty jset_entries: */
static void journal_entry_null_range(void *start, void *end)
{
struct jset_entry *entry;
for (entry = start; entry != end; entry = vstruct_next(entry))
memset(entry, 0, sizeof(*entry));
}
#define JOURNAL_ENTRY_REREAD 5
#define JOURNAL_ENTRY_NONE 6
#define JOURNAL_ENTRY_BAD 7
#define journal_entry_err(c, msg, ...) \
({ \
switch (write) { \
case READ: \
mustfix_fsck_err(c, msg, ##__VA_ARGS__); \
break; \
case WRITE: \
bch_err(c, "corrupt metadata before write:\n" \
msg, ##__VA_ARGS__); \
if (bch2_fs_inconsistent(c)) { \
ret = BCH_FSCK_ERRORS_NOT_FIXED; \
goto fsck_err; \
} \
break; \
} \
true; \
})
#define journal_entry_err_on(cond, c, msg, ...) \
((cond) ? journal_entry_err(c, msg, ##__VA_ARGS__) : false)
static int journal_validate_key(struct bch_fs *c, struct jset *jset,
struct jset_entry *entry,
struct bkey_i *k, enum btree_node_type key_type,
const char *type, int write)
{
void *next = vstruct_next(entry);
const char *invalid;
unsigned version = le32_to_cpu(jset->version);
int ret = 0;
if (journal_entry_err_on(!k->k.u64s, c,
"invalid %s in journal: k->u64s 0", type)) {
entry->u64s = cpu_to_le16((u64 *) k - entry->_data);
journal_entry_null_range(vstruct_next(entry), next);
return 0;
}
if (journal_entry_err_on((void *) bkey_next(k) >
(void *) vstruct_next(entry), c,
"invalid %s in journal: extends past end of journal entry",
type)) {
entry->u64s = cpu_to_le16((u64 *) k - entry->_data);
journal_entry_null_range(vstruct_next(entry), next);
return 0;
}
if (journal_entry_err_on(k->k.format != KEY_FORMAT_CURRENT, c,
"invalid %s in journal: bad format %u",
type, k->k.format)) {
le16_add_cpu(&entry->u64s, -k->k.u64s);
memmove(k, bkey_next(k), next - (void *) bkey_next(k));
journal_entry_null_range(vstruct_next(entry), next);
return 0;
}
if (JSET_BIG_ENDIAN(jset) != CPU_BIG_ENDIAN)
bch2_bkey_swab(NULL, bkey_to_packed(k));
if (!write &&
version < bcachefs_metadata_version_bkey_renumber)
bch2_bkey_renumber(key_type, bkey_to_packed(k), write);
invalid = bch2_bkey_invalid(c, bkey_i_to_s_c(k), key_type);
if (invalid) {
char buf[160];
bch2_bkey_val_to_text(&PBUF(buf), c, bkey_i_to_s_c(k));
mustfix_fsck_err(c, "invalid %s in journal: %s\n%s",
type, invalid, buf);
le16_add_cpu(&entry->u64s, -k->k.u64s);
memmove(k, bkey_next(k), next - (void *) bkey_next(k));
journal_entry_null_range(vstruct_next(entry), next);
return 0;
}
if (write &&
version < bcachefs_metadata_version_bkey_renumber)
bch2_bkey_renumber(key_type, bkey_to_packed(k), write);
fsck_err:
return ret;
}
static int journal_entry_validate_btree_keys(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
struct bkey_i *k;
vstruct_for_each(entry, k) {
int ret = journal_validate_key(c, jset, entry, k,
__btree_node_type(entry->level,
entry->btree_id),
"key", write);
if (ret)
return ret;
}
return 0;
}
static int journal_entry_validate_btree_root(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
struct bkey_i *k = entry->start;
int ret = 0;
if (journal_entry_err_on(!entry->u64s ||
le16_to_cpu(entry->u64s) != k->k.u64s, c,
"invalid btree root journal entry: wrong number of keys")) {
void *next = vstruct_next(entry);
/*
* we don't want to null out this jset_entry,
* just the contents, so that later we can tell
* we were _supposed_ to have a btree root
*/
entry->u64s = 0;
journal_entry_null_range(vstruct_next(entry), next);
return 0;
}
return journal_validate_key(c, jset, entry, k, BKEY_TYPE_BTREE,
"btree root", write);
fsck_err:
return ret;
}
static int journal_entry_validate_prio_ptrs(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
/* obsolete, don't care: */
return 0;
}
static int journal_entry_validate_blacklist(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
int ret = 0;
if (journal_entry_err_on(le16_to_cpu(entry->u64s) != 1, c,
"invalid journal seq blacklist entry: bad size")) {
journal_entry_null_range(entry, vstruct_next(entry));
}
fsck_err:
return ret;
}
static int journal_entry_validate_blacklist_v2(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
struct jset_entry_blacklist_v2 *bl_entry;
int ret = 0;
if (journal_entry_err_on(le16_to_cpu(entry->u64s) != 2, c,
"invalid journal seq blacklist entry: bad size")) {
journal_entry_null_range(entry, vstruct_next(entry));
goto out;
}
bl_entry = container_of(entry, struct jset_entry_blacklist_v2, entry);
if (journal_entry_err_on(le64_to_cpu(bl_entry->start) >
le64_to_cpu(bl_entry->end), c,
"invalid journal seq blacklist entry: start > end")) {
journal_entry_null_range(entry, vstruct_next(entry));
}
out:
fsck_err:
return ret;
}
static int journal_entry_validate_usage(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
struct jset_entry_usage *u =
container_of(entry, struct jset_entry_usage, entry);
unsigned bytes = jset_u64s(le16_to_cpu(entry->u64s)) * sizeof(u64);
int ret = 0;
if (journal_entry_err_on(bytes < sizeof(*u),
c,
"invalid journal entry usage: bad size")) {
journal_entry_null_range(entry, vstruct_next(entry));
return ret;
}
fsck_err:
return ret;
}
static int journal_entry_validate_data_usage(struct bch_fs *c,
struct jset *jset,
struct jset_entry *entry,
int write)
{
struct jset_entry_data_usage *u =
container_of(entry, struct jset_entry_data_usage, entry);
unsigned bytes = jset_u64s(le16_to_cpu(entry->u64s)) * sizeof(u64);
int ret = 0;
if (journal_entry_err_on(bytes < sizeof(*u) ||
bytes < sizeof(*u) + u->r.nr_devs,
c,
"invalid journal entry usage: bad size")) {
journal_entry_null_range(entry, vstruct_next(entry));
return ret;
}
fsck_err:
return ret;
}
struct jset_entry_ops {
int (*validate)(struct bch_fs *, struct jset *,
struct jset_entry *, int);
};
static const struct jset_entry_ops bch2_jset_entry_ops[] = {
#define x(f, nr) \
[BCH_JSET_ENTRY_##f] = (struct jset_entry_ops) { \
.validate = journal_entry_validate_##f, \
},
BCH_JSET_ENTRY_TYPES()
#undef x
};
static int journal_entry_validate(struct bch_fs *c, struct jset *jset,
struct jset_entry *entry, int write)
{
return entry->type < BCH_JSET_ENTRY_NR
? bch2_jset_entry_ops[entry->type].validate(c, jset,
entry, write)
: 0;
}
static int jset_validate_entries(struct bch_fs *c, struct jset *jset,
int write)
{
struct jset_entry *entry;
int ret = 0;
vstruct_for_each(jset, entry) {
if (journal_entry_err_on(vstruct_next(entry) >
vstruct_last(jset), c,
"journal entry extends past end of jset")) {
jset->u64s = cpu_to_le32((u64 *) entry - jset->_data);
break;
}
ret = journal_entry_validate(c, jset, entry, write);
if (ret)
break;
}
fsck_err:
return ret;
}
static int jset_validate(struct bch_fs *c,
struct jset *jset, u64 sector,
unsigned bucket_sectors_left,
unsigned sectors_read,
int write)
{
size_t bytes = vstruct_bytes(jset);
struct bch_csum csum;
unsigned version;
int ret = 0;
if (le64_to_cpu(jset->magic) != jset_magic(c))
return JOURNAL_ENTRY_NONE;
version = le32_to_cpu(jset->version);
if ((version != BCH_JSET_VERSION_OLD &&
version < bcachefs_metadata_version_min) ||
version >= bcachefs_metadata_version_max) {
bch_err(c, "unknown journal entry version %u", jset->version);
return BCH_FSCK_UNKNOWN_VERSION;
}
if (journal_entry_err_on(bytes > bucket_sectors_left << 9, c,
"journal entry too big (%zu bytes), sector %lluu",
bytes, sector)) {
/* XXX: note we might have missing journal entries */
return JOURNAL_ENTRY_BAD;
}
if (bytes > sectors_read << 9)
return JOURNAL_ENTRY_REREAD;
if (fsck_err_on(!bch2_checksum_type_valid(c, JSET_CSUM_TYPE(jset)), c,
"journal entry with unknown csum type %llu sector %lluu",
JSET_CSUM_TYPE(jset), sector))
return JOURNAL_ENTRY_BAD;
csum = csum_vstruct(c, JSET_CSUM_TYPE(jset), journal_nonce(jset), jset);
if (journal_entry_err_on(bch2_crc_cmp(csum, jset->csum), c,
"journal checksum bad, sector %llu", sector)) {
/* XXX: retry IO, when we start retrying checksum errors */
/* XXX: note we might have missing journal entries */
return JOURNAL_ENTRY_BAD;
}
bch2_encrypt(c, JSET_CSUM_TYPE(jset), journal_nonce(jset),
jset->encrypted_start,
vstruct_end(jset) - (void *) jset->encrypted_start);
if (journal_entry_err_on(le64_to_cpu(jset->last_seq) > le64_to_cpu(jset->seq), c,
"invalid journal entry: last_seq > seq"))
jset->last_seq = jset->seq;
return 0;
fsck_err:
return ret;
}
struct journal_read_buf {
void *data;
size_t size;
};
static int journal_read_buf_realloc(struct journal_read_buf *b,
size_t new_size)
{
void *n;
/* the bios are sized for this many pages, max: */
if (new_size > JOURNAL_ENTRY_SIZE_MAX)
return -ENOMEM;
new_size = roundup_pow_of_two(new_size);
n = kvpmalloc(new_size, GFP_KERNEL);
if (!n)
return -ENOMEM;
kvpfree(b->data, b->size);
b->data = n;
b->size = new_size;
return 0;
}
static int journal_read_bucket(struct bch_dev *ca,
struct journal_read_buf *buf,
struct journal_list *jlist,
unsigned bucket)
{
struct bch_fs *c = ca->fs;
struct journal_device *ja = &ca->journal;
struct jset *j = NULL;
unsigned sectors, sectors_read = 0;
u64 offset = bucket_to_sector(ca, ja->buckets[bucket]),
end = offset + ca->mi.bucket_size;
bool saw_bad = false;
int ret = 0;
pr_debug("reading %u", bucket);
while (offset < end) {
if (!sectors_read) {
struct bio *bio;
unsigned nr_bvecs;
reread:
sectors_read = min_t(unsigned,
end - offset, buf->size >> 9);
nr_bvecs = buf_pages(buf->data, sectors_read << 9);
bio = bio_kmalloc(nr_bvecs, GFP_KERNEL);
bio_init(bio, ca->disk_sb.bdev, bio->bi_inline_vecs, nr_bvecs, REQ_OP_READ);
bio->bi_iter.bi_sector = offset;
bch2_bio_map(bio, buf->data, sectors_read << 9);
ret = submit_bio_wait(bio);
kfree(bio);
if (bch2_dev_io_err_on(ret, ca,
"journal read from sector %llu",
offset) ||
bch2_meta_read_fault("journal"))
return -EIO;
j = buf->data;
}
ret = jset_validate(c, j, offset,
end - offset, sectors_read,
READ);
switch (ret) {
case BCH_FSCK_OK:
break;
case JOURNAL_ENTRY_REREAD:
if (vstruct_bytes(j) > buf->size) {
ret = journal_read_buf_realloc(buf,
vstruct_bytes(j));
if (ret)
return ret;
}
goto reread;
case JOURNAL_ENTRY_NONE:
if (!saw_bad)
return 0;
sectors = c->opts.block_size;
goto next_block;
case JOURNAL_ENTRY_BAD:
saw_bad = true;
sectors = c->opts.block_size;
goto next_block;
default:
return ret;
}
/*
* This happens sometimes if we don't have discards on -
* when we've partially overwritten a bucket with new
* journal entries. We don't need the rest of the
* bucket:
*/
if (le64_to_cpu(j->seq) < ja->bucket_seq[bucket])
return 0;
ja->bucket_seq[bucket] = le64_to_cpu(j->seq);
mutex_lock(&jlist->lock);
ret = journal_entry_add(c, ca, jlist, j);
mutex_unlock(&jlist->lock);
switch (ret) {
case JOURNAL_ENTRY_ADD_OK:
break;
case JOURNAL_ENTRY_ADD_OUT_OF_RANGE:
break;
default:
return ret;
}
sectors = vstruct_sectors(j, c->block_bits);
next_block:
pr_debug("next");
offset += sectors;
sectors_read -= sectors;
j = ((void *) j) + (sectors << 9);
}
return 0;
}
static void bch2_journal_read_device(struct closure *cl)
{
struct journal_device *ja =
container_of(cl, struct journal_device, read);
struct bch_dev *ca = container_of(ja, struct bch_dev, journal);
struct journal_list *jlist =
container_of(cl->parent, struct journal_list, cl);
struct journal_read_buf buf = { NULL, 0 };
u64 min_seq = U64_MAX;
unsigned i;
int ret;
if (!ja->nr)
goto out;
ret = journal_read_buf_realloc(&buf, PAGE_SIZE);
if (ret)
goto err;
pr_debug("%u journal buckets", ja->nr);
for (i = 0; i < ja->nr; i++) {
ret = journal_read_bucket(ca, &buf, jlist, i);
if (ret)
goto err;
}
/* Find the journal bucket with the highest sequence number: */
for (i = 0; i < ja->nr; i++) {
if (ja->bucket_seq[i] > ja->bucket_seq[ja->cur_idx])
ja->cur_idx = i;
min_seq = min(ja->bucket_seq[i], min_seq);
}
/*
* If there's duplicate journal entries in multiple buckets (which
* definitely isn't supposed to happen, but...) - make sure to start
* cur_idx at the last of those buckets, so we don't deadlock trying to
* allocate
*/
while (ja->bucket_seq[ja->cur_idx] > min_seq &&
ja->bucket_seq[ja->cur_idx] >
ja->bucket_seq[(ja->cur_idx + 1) % ja->nr])
ja->cur_idx = (ja->cur_idx + 1) % ja->nr;
ja->sectors_free = 0;
/*
* Set dirty_idx to indicate the entire journal is full and needs to be
* reclaimed - journal reclaim will immediately reclaim whatever isn't
* pinned when it first runs:
*/
ja->discard_idx = ja->dirty_idx_ondisk =
ja->dirty_idx = (ja->cur_idx + 1) % ja->nr;
out:
kvpfree(buf.data, buf.size);
percpu_ref_put(&ca->io_ref);
closure_return(cl);
return;
err:
mutex_lock(&jlist->lock);
jlist->ret = ret;
mutex_unlock(&jlist->lock);
goto out;
}
int bch2_journal_read(struct bch_fs *c, struct list_head *list)
{
struct journal_list jlist;
struct journal_replay *i;
struct bch_dev *ca;
unsigned iter;
size_t keys = 0, entries = 0;
bool degraded = false;
int ret = 0;
closure_init_stack(&jlist.cl);
mutex_init(&jlist.lock);
jlist.head = list;
jlist.ret = 0;
for_each_member_device(ca, c, iter) {
if (!test_bit(BCH_FS_REBUILD_REPLICAS, &c->flags) &&
!(bch2_dev_has_data(c, ca) & (1 << BCH_DATA_JOURNAL)))
continue;
if ((ca->mi.state == BCH_MEMBER_STATE_RW ||
ca->mi.state == BCH_MEMBER_STATE_RO) &&
percpu_ref_tryget(&ca->io_ref))
closure_call(&ca->journal.read,
bch2_journal_read_device,
system_unbound_wq,
&jlist.cl);
else
degraded = true;
}
closure_sync(&jlist.cl);
if (jlist.ret)
return jlist.ret;
list_for_each_entry(i, list, list) {
struct jset_entry *entry;
struct bkey_i *k, *_n;
struct bch_replicas_padded replicas;
char buf[80];
ret = jset_validate_entries(c, &i->j, READ);
if (ret)
goto fsck_err;
/*
* If we're mounting in degraded mode - if we didn't read all
* the devices - this is wrong:
*/
bch2_devlist_to_replicas(&replicas.e, BCH_DATA_JOURNAL, i->devs);
if (!degraded &&
(test_bit(BCH_FS_REBUILD_REPLICAS, &c->flags) ||
fsck_err_on(!bch2_replicas_marked(c, &replicas.e, false), c,
"superblock not marked as containing replicas %s",
(bch2_replicas_entry_to_text(&PBUF(buf),
&replicas.e), buf)))) {
ret = bch2_mark_replicas(c, &replicas.e);
if (ret)
return ret;
}
for_each_jset_key(k, _n, entry, &i->j)
keys++;
entries++;
}
if (!list_empty(list)) {
i = list_last_entry(list, struct journal_replay, list);
bch_info(c, "journal read done, %zu keys in %zu entries, seq %llu",
keys, entries, le64_to_cpu(i->j.seq));
}
fsck_err:
return ret;
}
/* journal write: */
static void __journal_write_alloc(struct journal *j,
struct journal_buf *w,
struct dev_alloc_list *devs_sorted,
unsigned sectors,
unsigned *replicas,
unsigned replicas_want)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct journal_device *ja;
struct bch_dev *ca;
unsigned i;
if (*replicas >= replicas_want)
return;
for (i = 0; i < devs_sorted->nr; i++) {
ca = rcu_dereference(c->devs[devs_sorted->devs[i]]);
if (!ca)
continue;
ja = &ca->journal;
/*
* Check that we can use this device, and aren't already using
* it:
*/
if (!ca->mi.durability ||
ca->mi.state != BCH_MEMBER_STATE_RW ||
!ja->nr ||
bch2_bkey_has_device(bkey_i_to_s_c(&w->key),
ca->dev_idx) ||
sectors > ja->sectors_free)
continue;
bch2_dev_stripe_increment(c, ca, &j->wp.stripe);
bch2_bkey_append_ptr(&w->key,
(struct bch_extent_ptr) {
.offset = bucket_to_sector(ca,
ja->buckets[ja->cur_idx]) +
ca->mi.bucket_size -
ja->sectors_free,
.dev = ca->dev_idx,
});
ja->sectors_free -= sectors;
ja->bucket_seq[ja->cur_idx] = le64_to_cpu(w->data->seq);
*replicas += ca->mi.durability;
if (*replicas >= replicas_want)
break;
}
}
/**
* journal_next_bucket - move on to the next journal bucket if possible
*/
static int journal_write_alloc(struct journal *j, struct journal_buf *w,
unsigned sectors)
{
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct journal_device *ja;
struct bch_dev *ca;
struct dev_alloc_list devs_sorted;
unsigned i, replicas = 0, replicas_want =
READ_ONCE(c->opts.metadata_replicas);
rcu_read_lock();
devs_sorted = bch2_dev_alloc_list(c, &j->wp.stripe,
&c->rw_devs[BCH_DATA_JOURNAL]);
__journal_write_alloc(j, w, &devs_sorted,
sectors, &replicas, replicas_want);
if (replicas >= replicas_want)
goto done;
for (i = 0; i < devs_sorted.nr; i++) {
ca = rcu_dereference(c->devs[devs_sorted.devs[i]]);
if (!ca)
continue;
ja = &ca->journal;
if (sectors > ja->sectors_free &&
sectors <= ca->mi.bucket_size &&
bch2_journal_dev_buckets_available(j, ja,
journal_space_discarded)) {
ja->cur_idx = (ja->cur_idx + 1) % ja->nr;
ja->sectors_free = ca->mi.bucket_size;
/*
* ja->bucket_seq[ja->cur_idx] must always have
* something sensible:
*/
ja->bucket_seq[ja->cur_idx] = le64_to_cpu(w->data->seq);
}
}
__journal_write_alloc(j, w, &devs_sorted,
sectors, &replicas, replicas_want);
done:
rcu_read_unlock();
return replicas >= c->opts.metadata_replicas_required ? 0 : -EROFS;
}
static void journal_write_compact(struct jset *jset)
{
struct jset_entry *i, *next, *prev = NULL;
/*
* Simple compaction, dropping empty jset_entries (from journal
* reservations that weren't fully used) and merging jset_entries that
* can be.
*
* If we wanted to be really fancy here, we could sort all the keys in
* the jset and drop keys that were overwritten - probably not worth it:
*/
vstruct_for_each_safe(jset, i, next) {
unsigned u64s = le16_to_cpu(i->u64s);
/* Empty entry: */
if (!u64s)
continue;
/* Can we merge with previous entry? */
if (prev &&
i->btree_id == prev->btree_id &&
i->level == prev->level &&
i->type == prev->type &&
i->type == BCH_JSET_ENTRY_btree_keys &&
le16_to_cpu(prev->u64s) + u64s <= U16_MAX) {
memmove_u64s_down(vstruct_next(prev),
i->_data,
u64s);
le16_add_cpu(&prev->u64s, u64s);
continue;
}
/* Couldn't merge, move i into new position (after prev): */
prev = prev ? vstruct_next(prev) : jset->start;
if (i != prev)
memmove_u64s_down(prev, i, jset_u64s(u64s));
}
prev = prev ? vstruct_next(prev) : jset->start;
jset->u64s = cpu_to_le32((u64 *) prev - jset->_data);
}
static void journal_buf_realloc(struct journal *j, struct journal_buf *buf)
{
/* we aren't holding j->lock: */
unsigned new_size = READ_ONCE(j->buf_size_want);
void *new_buf;
if (buf->buf_size >= new_size)
return;
new_buf = kvpmalloc(new_size, GFP_NOIO|__GFP_NOWARN);
if (!new_buf)
return;
memcpy(new_buf, buf->data, buf->buf_size);
kvpfree(buf->data, buf->buf_size);
buf->data = new_buf;
buf->buf_size = new_size;
}
static void journal_write_done(struct closure *cl)
{
struct journal *j = container_of(cl, struct journal, io);
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct journal_buf *w = journal_prev_buf(j);
struct bch_devs_list devs =
bch2_bkey_devs(bkey_i_to_s_c(&w->key));
struct bch_replicas_padded replicas;
u64 seq = le64_to_cpu(w->data->seq);
u64 last_seq = le64_to_cpu(w->data->last_seq);
bch2_time_stats_update(j->write_time, j->write_start_time);
if (!devs.nr) {
bch_err(c, "unable to write journal to sufficient devices");
goto err;
}
bch2_devlist_to_replicas(&replicas.e, BCH_DATA_JOURNAL, devs);
if (bch2_mark_replicas(c, &replicas.e))
goto err;
spin_lock(&j->lock);
if (seq >= j->pin.front)
journal_seq_pin(j, seq)->devs = devs;
j->seq_ondisk = seq;
j->last_seq_ondisk = last_seq;
bch2_journal_space_available(j);
/*
* Updating last_seq_ondisk may let bch2_journal_reclaim_work() discard
* more buckets:
*
* Must come before signaling write completion, for
* bch2_fs_journal_stop():
*/
mod_delayed_work(c->journal_reclaim_wq, &j->reclaim_work, 0);
out:
/* also must come before signalling write completion: */
closure_debug_destroy(cl);
BUG_ON(!j->reservations.prev_buf_unwritten);
atomic64_sub(((union journal_res_state) { .prev_buf_unwritten = 1 }).v,
&j->reservations.counter);
closure_wake_up(&w->wait);
journal_wake(j);
if (test_bit(JOURNAL_NEED_WRITE, &j->flags))
mod_delayed_work(system_freezable_wq, &j->write_work, 0);
spin_unlock(&j->lock);
return;
err:
bch2_fatal_error(c);
spin_lock(&j->lock);
goto out;
}
static void journal_write_endio(struct bio *bio)
{
struct bch_dev *ca = bio->bi_private;
struct journal *j = &ca->fs->journal;
if (bch2_dev_io_err_on(bio->bi_status, ca, "journal write") ||
bch2_meta_write_fault("journal")) {
struct journal_buf *w = journal_prev_buf(j);
unsigned long flags;
spin_lock_irqsave(&j->err_lock, flags);
bch2_bkey_drop_device(bkey_i_to_s(&w->key), ca->dev_idx);
spin_unlock_irqrestore(&j->err_lock, flags);
}
closure_put(&j->io);
percpu_ref_put(&ca->io_ref);
}
void bch2_journal_write(struct closure *cl)
{
struct journal *j = container_of(cl, struct journal, io);
struct bch_fs *c = container_of(j, struct bch_fs, journal);
struct bch_dev *ca;
struct journal_buf *w = journal_prev_buf(j);
struct jset_entry *start, *end;
struct jset *jset;
struct bio *bio;
struct bch_extent_ptr *ptr;
bool validate_before_checksum = false;
unsigned i, sectors, bytes, u64s;
int ret;
bch2_journal_pin_put(j, le64_to_cpu(w->data->seq));
journal_buf_realloc(j, w);
jset = w->data;
j->write_start_time = local_clock();
start = vstruct_last(jset);
end = bch2_journal_super_entries_add_common(c, start,
le64_to_cpu(jset->seq));
u64s = (u64 *) end - (u64 *) start;
BUG_ON(u64s > j->entry_u64s_reserved);
le32_add_cpu(&jset->u64s, u64s);
BUG_ON(vstruct_sectors(jset, c->block_bits) > w->sectors);
journal_write_compact(jset);
jset->read_clock = cpu_to_le16(c->bucket_clock[READ].hand);
jset->write_clock = cpu_to_le16(c->bucket_clock[WRITE].hand);
jset->magic = cpu_to_le64(jset_magic(c));
jset->version = c->sb.version < bcachefs_metadata_version_new_versioning
? cpu_to_le32(BCH_JSET_VERSION_OLD)
: cpu_to_le32(c->sb.version);
SET_JSET_BIG_ENDIAN(jset, CPU_BIG_ENDIAN);
SET_JSET_CSUM_TYPE(jset, bch2_meta_checksum_type(c));
if (bch2_csum_type_is_encryption(JSET_CSUM_TYPE(jset)))
validate_before_checksum = true;
if (le32_to_cpu(jset->version) <
bcachefs_metadata_version_bkey_renumber)
validate_before_checksum = true;
if (validate_before_checksum &&
jset_validate_entries(c, jset, WRITE))
goto err;
bch2_encrypt(c, JSET_CSUM_TYPE(jset), journal_nonce(jset),
jset->encrypted_start,
vstruct_end(jset) - (void *) jset->encrypted_start);
jset->csum = csum_vstruct(c, JSET_CSUM_TYPE(jset),
journal_nonce(jset), jset);
if (!validate_before_checksum &&
jset_validate_entries(c, jset, WRITE))
goto err;
sectors = vstruct_sectors(jset, c->block_bits);
BUG_ON(sectors > w->sectors);
bytes = vstruct_bytes(jset);
memset((void *) jset + bytes, 0, (sectors << 9) - bytes);
spin_lock(&j->lock);
ret = journal_write_alloc(j, w, sectors);
/*
* write is allocated, no longer need to account for it in
* bch2_journal_space_available():
*/
w->sectors = 0;
/*
* journal entry has been compacted and allocated, recalculate space
* available:
*/
bch2_journal_space_available(j);
spin_unlock(&j->lock);
if (ret) {
bch_err(c, "Unable to allocate journal write");
bch2_fatal_error(c);
continue_at(cl, journal_write_done, system_highpri_wq);
return;
}
/*
* XXX: we really should just disable the entire journal in nochanges
* mode
*/
if (c->opts.nochanges)
goto no_io;
extent_for_each_ptr(bkey_i_to_s_extent(&w->key), ptr) {
ca = bch_dev_bkey_exists(c, ptr->dev);
if (!percpu_ref_tryget(&ca->io_ref)) {
/* XXX: fix this */
bch_err(c, "missing device for journal write\n");
continue;
}
this_cpu_add(ca->io_done->sectors[WRITE][BCH_DATA_JOURNAL],
sectors);
bio = ca->journal.bio;
bio_reset(bio, ca->disk_sb.bdev,
REQ_OP_WRITE|REQ_SYNC|REQ_META|REQ_PREFLUSH|REQ_FUA);
bio->bi_iter.bi_sector = ptr->offset;
bio->bi_end_io = journal_write_endio;
bio->bi_private = ca;
bch2_bio_map(bio, jset, sectors << 9);
trace_journal_write(bio);
closure_bio_submit(bio, cl);
ca->journal.bucket_seq[ca->journal.cur_idx] = le64_to_cpu(jset->seq);
}
for_each_rw_member(ca, c, i)
if (journal_flushes_device(ca) &&
!bch2_extent_has_device(bkey_i_to_s_c_extent(&w->key), i)) {
percpu_ref_get(&ca->io_ref);
bio = ca->journal.bio;
bio_reset(bio, ca->disk_sb.bdev, REQ_OP_FLUSH);
bio->bi_end_io = journal_write_endio;
bio->bi_private = ca;
closure_bio_submit(bio, cl);
}
no_io:
bch2_bucket_seq_cleanup(c);
continue_at(cl, journal_write_done, system_highpri_wq);
return;
err:
bch2_inconsistent_error(c);
continue_at(cl, journal_write_done, system_highpri_wq);
}
|