projects / linux-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
bcachefs: Fix restart handling in for_each_btree_key()
[linux-block.git] / fs / bcachefs / fsck.c
CommitLineData
1c6fdbd8
KO		 1// SPDX-License-Identifier: GPL-2.0
2
3#include "bcachefs.h"
07a1006a 4#include "bkey_buf.h"
1c6fdbd8
KO		 5#include "btree_update.h"
6#include "dirent.h"
7#include "error.h"
96385742 8#include "fs-common.h"
1c6fdbd8
KO		 9#include "fsck.h"
10#include "inode.h"
11#include "keylist.h"
14b393ee 12#include "subvolume.h"
1c6fdbd8
KO		 13#include "super.h"
14#include "xattr.h"
15
fc51b041 16#include <linux/bsearch.h>
1c6fdbd8 17#include <linux/dcache.h> /* struct qstr */
1c6fdbd8
KO		 18
19#define QSTR(n) { { { .len = strlen(n) } }, .name = n }
20
ef1669ff
KO		 21static s64 bch2_count_inode_sectors(struct btree_trans *trans, u64 inum,
22 u32 snapshot)
424eb881 23{
67e0dd8f 24 struct btree_iter iter;
424eb881
KO		 25 struct bkey_s_c k;
26 u64 sectors = 0;
94f651e2 27 int ret;
424eb881 28
41f8b09e 29 for_each_btree_key(trans, iter, BTREE_ID_extents,
ef1669ff 30 SPOS(inum, 0, snapshot), 0, k, ret) {
424eb881
KO		 31 if (k.k->p.inode != inum)
32 break;
33
34 if (bkey_extent_is_allocation(k.k))
35 sectors += k.k->size;
36 }
37
67e0dd8f 38 bch2_trans_iter_exit(trans, &iter);
94f651e2
KO		 39
40 return ret ?: sectors;
424eb881
KO		 41}
42
ef1669ff
KO		 43static s64 bch2_count_subdirs(struct btree_trans *trans, u64 inum,
44 u32 snapshot)
45{
46 struct btree_iter iter;
47 struct bkey_s_c k;
48 struct bkey_s_c_dirent d;
49 u64 subdirs = 0;
50 int ret;
51
52 for_each_btree_key(trans, iter, BTREE_ID_dirents,
53 SPOS(inum, 0, snapshot), 0, k, ret) {
54 if (k.k->p.inode != inum)
55 break;
56
57 if (k.k->type != KEY_TYPE_dirent)
58 continue;
59
60 d = bkey_s_c_to_dirent(k);
61 if (d.v->d_type == DT_DIR)
62 subdirs++;
63 }
64
65 bch2_trans_iter_exit(trans, &iter);
66
67 return ret ?: subdirs;
68}
69
81ed9ce3
KO		 70static int __snapshot_lookup_subvol(struct btree_trans *trans, u32 snapshot,
71 u32 *subvol)
72{
73 struct btree_iter iter;
74 struct bkey_s_c k;
75 int ret;
76
77 bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots,
78 POS(0, snapshot), 0);
79 k = bch2_btree_iter_peek_slot(&iter);
80 ret = bkey_err(k);
81 if (ret)
82 goto err;
83
84 if (k.k->type != KEY_TYPE_snapshot) {
85 bch_err(trans->c, "snapshot %u not fonud", snapshot);
86 ret = -ENOENT;
87 goto err;
88 }
89
90 *subvol = le32_to_cpu(bkey_s_c_to_snapshot(k).v->subvol);
91err:
92 bch2_trans_iter_exit(trans, &iter);
93 return ret;
94
95}
96
97static int snapshot_lookup_subvol(struct btree_trans *trans, u32 snapshot,
98 u32 *subvol)
99{
100 return lockrestart_do(trans, __snapshot_lookup_subvol(trans, snapshot, subvol));
101}
102
ef1669ff
KO		 103static int __subvol_lookup(struct btree_trans *trans, u32 subvol,
104 u32 *snapshot, u64 *inum)
81ed9ce3 105{
97996ddf 106 struct bch_subvolume s;
81ed9ce3
KO		 107 int ret;
108
97996ddf 109 ret = bch2_subvolume_get(trans, subvol, false, 0, &s);
81ed9ce3 110
97996ddf
KO		 111 *snapshot = le32_to_cpu(s.snapshot);
112 *inum = le64_to_cpu(s.inode);
81ed9ce3 113 return ret;
81ed9ce3
KO		 114}
115
ef1669ff
KO		 116static int subvol_lookup(struct btree_trans *trans, u32 subvol,
117 u32 *snapshot, u64 *inum)
81ed9ce3 118{
ef1669ff 119 return lockrestart_do(trans, __subvol_lookup(trans, subvol, snapshot, inum));
81ed9ce3
KO		 120}
121
58686a25
KO		 122static int __lookup_inode(struct btree_trans *trans, u64 inode_nr,
123 struct bch_inode_unpacked *inode,
124 u32 *snapshot)
8a85b20c 125{
67e0dd8f 126 struct btree_iter iter;
8a85b20c
KO		 127 struct bkey_s_c k;
128 int ret;
129
67e0dd8f 130 bch2_trans_iter_init(trans, &iter, BTREE_ID_inodes,
ef1669ff 131 SPOS(0, inode_nr, *snapshot), 0);
67e0dd8f 132 k = bch2_btree_iter_peek_slot(&iter);
8a85b20c
KO		 133 ret = bkey_err(k);
134 if (ret)
135 goto err;
136
ef1669ff 137 *snapshot = iter.pos.snapshot;
8a85b20c
KO		 138 ret = k.k->type == KEY_TYPE_inode
139 ? bch2_inode_unpack(bkey_s_c_to_inode(k), inode)
140 : -ENOENT;
141err:
67e0dd8f 142 bch2_trans_iter_exit(trans, &iter);
8a85b20c
KO		 143 return ret;
144}
145
58686a25
KO		 146static int lookup_inode(struct btree_trans *trans, u64 inode_nr,
147 struct bch_inode_unpacked *inode,
148 u32 *snapshot)
149{
150 return lockrestart_do(trans, __lookup_inode(trans, inode_nr, inode, snapshot));
151}
152
ef1669ff
KO		 153static int __lookup_dirent(struct btree_trans *trans,
154 struct bch_hash_info hash_info,
155 subvol_inum dir, struct qstr *name,
156 u64 *target, unsigned *type)
157{
158 struct btree_iter iter;
159 struct bkey_s_c_dirent d;
160 int ret;
161
162 ret = bch2_hash_lookup(trans, &iter, bch2_dirent_hash_desc,
163 &hash_info, dir, name, 0);
164 if (ret)
165 return ret;
166
167 d = bkey_s_c_to_dirent(bch2_btree_iter_peek_slot(&iter));
168 *target = le64_to_cpu(d.v->d_inum);
169 *type = d.v->d_type;
170 bch2_trans_iter_exit(trans, &iter);
171 return 0;
172}
173
174static int lookup_dirent(struct btree_trans *trans,
175 struct bch_hash_info hash_info,
176 subvol_inum dir, struct qstr *name,
177 u64 *target, unsigned *type)
178{
179 return lockrestart_do(trans,
180 __lookup_dirent(trans, hash_info, dir, name, target, type));
181}
182
58686a25
KO		 183static int __write_inode(struct btree_trans *trans,
184 struct bch_inode_unpacked *inode,
185 u32 snapshot)
8a85b20c 186{
67e0dd8f
KO		 187 struct btree_iter iter;
188 int ret;
189
190 bch2_trans_iter_init(trans, &iter, BTREE_ID_inodes,
191 SPOS(0, inode->bi_inum, snapshot),
192 BTREE_ITER_INTENT);
193
194 ret = bch2_btree_iter_traverse(&iter) ?:
195 bch2_inode_write(trans, &iter, inode);
196 bch2_trans_iter_exit(trans, &iter);
58686a25
KO		 197 return ret;
198}
199
200static int write_inode(struct btree_trans *trans,
201 struct bch_inode_unpacked *inode,
202 u32 snapshot)
203{
8a85b20c
KO		 204 int ret = __bch2_trans_do(trans, NULL, NULL,
205 BTREE_INSERT_NOFAIL|
206 BTREE_INSERT_LAZY_RW,
58686a25 207 __write_inode(trans, inode, snapshot));
8a85b20c
KO		 208 if (ret)
209 bch_err(trans->c, "error in fsck: error %i updating inode", ret);
210 return ret;
211}
212
ef1669ff
KO		 213static int fsck_inode_rm(struct btree_trans *trans, u64 inum, u32 snapshot)
214{
215 struct btree_iter iter = { NULL };
216 struct bkey_i_inode_generation delete;
217 struct bch_inode_unpacked inode_u;
218 struct bkey_s_c k;
219 int ret;
220
221 ret = bch2_btree_delete_range_trans(trans, BTREE_ID_extents,
222 SPOS(inum, 0, snapshot),
223 SPOS(inum, U64_MAX, snapshot),
224 0, NULL) ?:
225 bch2_btree_delete_range_trans(trans, BTREE_ID_dirents,
226 SPOS(inum, 0, snapshot),
227 SPOS(inum, U64_MAX, snapshot),
228 0, NULL) ?:
229 bch2_btree_delete_range_trans(trans, BTREE_ID_xattrs,
230 SPOS(inum, 0, snapshot),
231 SPOS(inum, U64_MAX, snapshot),
232 0, NULL);
233 if (ret)
234 goto err;
235retry:
236 bch2_trans_begin(trans);
237
238 bch2_trans_iter_init(trans, &iter, BTREE_ID_inodes,
239 SPOS(0, inum, snapshot), BTREE_ITER_INTENT);
240 k = bch2_btree_iter_peek_slot(&iter);
241
242 ret = bkey_err(k);
243 if (ret)
244 goto err;
245
246 if (k.k->type != KEY_TYPE_inode) {
247 bch2_fs_inconsistent(trans->c,
248 "inode %llu:%u not found when deleting",
249 inum, snapshot);
250 ret = -EIO;
251 goto err;
252 }
253
254 bch2_inode_unpack(bkey_s_c_to_inode(k), &inode_u);
255
256 /* Subvolume root? */
257 if (inode_u.bi_subvol) {
258 ret = bch2_subvolume_delete(trans, inode_u.bi_subvol, -1);
259 if (ret)
260 goto err;
261 }
262
263 bkey_inode_generation_init(&delete.k_i);
264 delete.k.p = iter.pos;
265 delete.v.bi_generation = cpu_to_le32(inode_u.bi_generation + 1);
266
267 ret = bch2_trans_update(trans, &iter, &delete.k_i, 0) ?:
268 bch2_trans_commit(trans, NULL, NULL,
269 BTREE_INSERT_NOFAIL);
270err:
271 bch2_trans_iter_exit(trans, &iter);
272 if (ret == -EINTR)
273 goto retry;
274
275 return ret;
276}
277
ae8bbb9f 278static int __remove_dirent(struct btree_trans *trans, struct bpos pos)
1c6fdbd8 279{
0f238367 280 struct bch_fs *c = trans->c;
67e0dd8f 281 struct btree_iter iter;
1c6fdbd8
KO		 282 struct bch_inode_unpacked dir_inode;
283 struct bch_hash_info dir_hash_info;
1c6fdbd8 284 int ret;
1c6fdbd8 285
ae8bbb9f 286 ret = lookup_inode(trans, pos.inode, &dir_inode, NULL);
b1fd23df
KO		 287 if (ret)
288 return ret;
1c6fdbd8
KO		 289
290 dir_hash_info = bch2_hash_info_init(c, &dir_inode);
291
67e0dd8f 292 bch2_trans_iter_init(trans, &iter, BTREE_ID_dirents, pos, BTREE_ITER_INTENT);
b1fd23df 293
ae8bbb9f 294 ret = bch2_hash_delete_at(trans, bch2_dirent_hash_desc,
42d23732 295 &dir_hash_info, &iter, 0);
67e0dd8f 296 bch2_trans_iter_exit(trans, &iter);
ae8bbb9f 297 return ret;
b1fd23df
KO		 298}
299
ae8bbb9f 300static int remove_dirent(struct btree_trans *trans, struct bpos pos)
b1fd23df 301{
ae8bbb9f
KO		 302 int ret = __bch2_trans_do(trans, NULL, NULL,
303 BTREE_INSERT_NOFAIL|
304 BTREE_INSERT_LAZY_RW,
305 __remove_dirent(trans, pos));
306 if (ret)
307 bch_err(trans->c, "remove_dirent: err %i deleting dirent", ret);
308 return ret;
1c6fdbd8
KO		 309}
310
58686a25 311/* Get lost+found, create if it doesn't exist: */
ef1669ff 312static int lookup_lostfound(struct btree_trans *trans, u32 subvol,
58686a25 313 struct bch_inode_unpacked *lostfound)
1c6fdbd8 314{
58686a25
KO		 315 struct bch_fs *c = trans->c;
316 struct bch_inode_unpacked root;
317 struct bch_hash_info root_hash_info;
318 struct qstr lostfound_str = QSTR("lost+found");
ef1669ff
KO		 319 subvol_inum root_inum = { .subvol = subvol };
320 u64 inum = 0;
321 unsigned d_type = 0;
58686a25
KO		 322 u32 snapshot;
323 int ret;
324
ef1669ff
KO		 325 ret = subvol_lookup(trans, subvol, &snapshot, &root_inum.inum);
326 if (ret)
327 return ret;
81ed9ce3 328
ef1669ff
KO		 329 ret = lookup_inode(trans, root_inum.inum, &root, &snapshot);
330 if (ret) {
331 bch_err(c, "error fetching subvol root: %i", ret);
58686a25 332 return ret;
ef1669ff 333 }
58686a25
KO		 334
335 root_hash_info = bch2_hash_info_init(c, &root);
ef1669ff
KO		 336
337 ret = lookup_dirent(trans, root_hash_info, root_inum,
338 &lostfound_str, &inum, &d_type);
339 if (ret == -ENOENT) {
58686a25
KO		 340 bch_notice(c, "creating lost+found");
341 goto create_lostfound;
342 }
343
ef1669ff
KO		 344 if (ret) {
345 bch_err(c, "error looking up lost+found: %i", ret);
346 return ret;
347 }
348
349 if (d_type != DT_DIR) {
350 bch_err(c, "error looking up lost+found: not a directory");
351 return ret;
352
353 }
354
58686a25
KO		 355 ret = lookup_inode(trans, inum, lostfound, &snapshot);
356 if (ret && ret != -ENOENT) {
357 /*
358 * The check_dirents pass has already run, dangling dirents
359 * shouldn't exist here:
360 */
361 bch_err(c, "error looking up lost+found: %i", ret);
362 return ret;
363 }
364
365 if (ret == -ENOENT) {
366create_lostfound:
367 bch2_inode_init_early(c, lostfound);
368
369 ret = __bch2_trans_do(trans, NULL, NULL,
370 BTREE_INSERT_NOFAIL|
371 BTREE_INSERT_LAZY_RW,
ef1669ff
KO		 372 bch2_create_trans(trans, root_inum, &root,
373 lostfound, &lostfound_str,
42d23732
KO		 374 0, 0, S_IFDIR|0700, 0, NULL, NULL,
375 (subvol_inum) { }, 0));
58686a25
KO		 376 if (ret)
377 bch_err(c, "error creating lost+found: %i", ret);
378 }
379
380 return 0;
381}
382
383static int reattach_inode(struct btree_trans *trans,
81ed9ce3 384 struct bch_inode_unpacked *inode,
ef1669ff 385 u32 inode_snapshot)
58686a25
KO		 386{
387 struct bch_hash_info dir_hash;
388 struct bch_inode_unpacked lostfound;
1c6fdbd8
KO		 389 char name_buf[20];
390 struct qstr name;
176cf4bf 391 u64 dir_offset = 0;
81ed9ce3 392 u32 subvol;
1c6fdbd8
KO		 393 int ret;
394
ef1669ff 395 ret = snapshot_lookup_subvol(trans, inode_snapshot, &subvol);
81ed9ce3
KO		 396 if (ret)
397 return ret;
398
399 ret = lookup_lostfound(trans, subvol, &lostfound);
176cf4bf 400 if (ret)
d3ff7fec 401 return ret;
176cf4bf 402
58686a25
KO		 403 if (S_ISDIR(inode->bi_mode)) {
404 lostfound.bi_nlink++;
176cf4bf 405
58686a25 406 ret = write_inode(trans, &lostfound, U32_MAX);
176cf4bf 407 if (ret)
d3ff7fec 408 return ret;
176cf4bf
KO		 409 }
410
58686a25 411 dir_hash = bch2_hash_info_init(trans->c, &lostfound);
176cf4bf 412
58686a25
KO		 413 snprintf(name_buf, sizeof(name_buf), "%llu", inode->bi_inum);
414 name = (struct qstr) QSTR(name_buf);
176cf4bf 415
58686a25 416 ret = __bch2_trans_do(trans, NULL, NULL, BTREE_INSERT_LAZY_RW,
ef1669ff
KO		 417 bch2_dirent_create(trans,
418 (subvol_inum) {
419 .subvol = subvol,
420 .inum = lostfound.bi_inum,
421 },
422 &dir_hash,
423 mode_to_type(inode->bi_mode),
424 &name, inode->bi_inum, &dir_offset,
425 BCH_HASH_SET_MUST_CREATE));
58686a25
KO		 426 if (ret) {
427 bch_err(trans->c, "error %i reattaching inode %llu",
428 ret, inode->bi_inum);
429 return ret;
430 }
176cf4bf 431
58686a25
KO		 432 inode->bi_dir = lostfound.bi_inum;
433 inode->bi_dir_offset = dir_offset;
1c6fdbd8 434
ef1669ff 435 return write_inode(trans, inode, inode_snapshot);
1c6fdbd8
KO		 436}
437
d3ff7fec
KO		 438static int remove_backpointer(struct btree_trans *trans,
439 struct bch_inode_unpacked *inode)
440{
67e0dd8f 441 struct btree_iter iter;
d3ff7fec
KO		 442 struct bkey_s_c k;
443 int ret;
444
67e0dd8f
KO		 445 bch2_trans_iter_init(trans, &iter, BTREE_ID_dirents,
446 POS(inode->bi_dir, inode->bi_dir_offset), 0);
447 k = bch2_btree_iter_peek_slot(&iter);
d3ff7fec
KO		 448 ret = bkey_err(k);
449 if (ret)
450 goto out;
451 if (k.k->type != KEY_TYPE_dirent) {
452 ret = -ENOENT;
453 goto out;
454 }
455
ae8bbb9f 456 ret = remove_dirent(trans, k.k->p);
d3ff7fec 457out:
67e0dd8f 458 bch2_trans_iter_exit(trans, &iter);
d3ff7fec
KO		 459 return ret;
460}
461
ef1669ff
KO		 462static int snapshots_seen_update(struct bch_fs *c, struct snapshots_seen *s, struct bpos pos)
463{
464 pos.snapshot = snapshot_t(c, pos.snapshot)->equiv;
465
466 if (bkey_cmp(s->pos, pos))
467 s->nr = 0;
468 s->pos = pos;
469
ef1669ff
KO		 470 /* Might get called multiple times due to lock restarts */
471 if (s->nr && s->d[s->nr - 1] == pos.snapshot)
472 return 0;
473
18443cb9 474 return snapshots_seen_add(c, s, pos.snapshot);
ef1669ff
KO		 475}
476
477/**
478 * key_visible_in_snapshot - returns true if @id is a descendent of @ancestor,
479 * and @ancestor hasn't been overwritten in @seen
480 *
481 * That is, returns whether key in @ancestor snapshot is visible in @id snapshot
482 */
483static bool key_visible_in_snapshot(struct bch_fs *c, struct snapshots_seen *seen,
484 u32 id, u32 ancestor)
485{
486 ssize_t i;
487
488 BUG_ON(id > ancestor);
489
490 id = snapshot_t(c, id)->equiv;
491 ancestor = snapshot_t(c, ancestor)->equiv;
492
493 /* @ancestor should be the snapshot most recently added to @seen */
494 BUG_ON(!seen->nr || seen->d[seen->nr - 1] != ancestor);
495 BUG_ON(seen->pos.snapshot != ancestor);
496
497 if (id == ancestor)
498 return true;
499
500 if (!bch2_snapshot_is_ancestor(c, id, ancestor))
501 return false;
502
503 for (i = seen->nr - 2;
504 i >= 0 && seen->d[i] >= id;
505 --i)
506 if (bch2_snapshot_is_ancestor(c, id, seen->d[i]) &&
507 bch2_snapshot_is_ancestor(c, seen->d[i], ancestor))
508 return false;
509
510 return true;
511}
512
513/**
514 * ref_visible - given a key with snapshot id @src that points to a key with
515 * snapshot id @dst, test whether there is some snapshot in which @dst is
516 * visible.
517 *
518 * This assumes we're visiting @src keys in natural key order.
519 *
520 * @s - list of snapshot IDs already seen at @src
521 * @src - snapshot ID of src key
522 * @dst - snapshot ID of dst key
523 */
524static int ref_visible(struct bch_fs *c, struct snapshots_seen *s,
525 u32 src, u32 dst)
526{
527 return dst <= src
528 ? key_visible_in_snapshot(c, s, dst, src)
529 : bch2_snapshot_is_ancestor(c, src, dst);
530}
531
532#define for_each_visible_inode(_c, _s, _w, _snapshot, _i) \
533 for (_i = (_w)->d; _i < (_w)->d + (_w)->nr && (_i)->snapshot <= (_snapshot); _i++)\
534 if (key_visible_in_snapshot(_c, _s, _i->snapshot, _snapshot))
535
1c6fdbd8 536struct inode_walker {
ef1669ff
KO		 537 bool first_this_inode;
538 u64 cur_inum;
539
540 size_t nr;
541 size_t size;
542 struct inode_walker_entry {
543 struct bch_inode_unpacked inode;
544 u32 snapshot;
545 u64 count;
546 } *d;
1c6fdbd8
KO		 547};
548
ef1669ff
KO		 549static void inode_walker_exit(struct inode_walker *w)
550{
551 kfree(w->d);
552 w->d = NULL;
553}
554
1c6fdbd8
KO		 555static struct inode_walker inode_walker_init(void)
556{
ef1669ff
KO		 557 return (struct inode_walker) { 0, };
558}
559
560static int inode_walker_realloc(struct inode_walker *w)
561{
562 if (w->nr == w->size) {
563 size_t new_size = max_t(size_t, 8UL, w->size * 2);
564 void *d = krealloc(w->d, new_size * sizeof(w->d[0]),
565 GFP_KERNEL);
566 if (!d)
567 return -ENOMEM;
568
569 w->d = d;
570 w->size = new_size;
571 }
572
573 return 0;
574}
575
576static int add_inode(struct bch_fs *c, struct inode_walker *w,
577 struct bkey_s_c_inode inode)
578{
579 struct bch_inode_unpacked u;
580 int ret;
581
582 ret = inode_walker_realloc(w);
583 if (ret)
584 return ret;
585
586 BUG_ON(bch2_inode_unpack(inode, &u));
587
588 w->d[w->nr++] = (struct inode_walker_entry) {
589 .inode = u,
590 .snapshot = snapshot_t(c, inode.k->p.snapshot)->equiv,
1c6fdbd8 591 };
ef1669ff
KO		 592
593 return 0;
1c6fdbd8
KO		 594}
595
914f2786 596static int __walk_inode(struct btree_trans *trans,
ef1669ff 597 struct inode_walker *w, struct bpos pos)
1c6fdbd8 598{
ef1669ff
KO		 599 struct bch_fs *c = trans->c;
600 struct btree_iter iter;
601 struct bkey_s_c k;
602 unsigned i, ancestor_pos;
603 int ret;
1c6fdbd8 604
ef1669ff 605 pos.snapshot = snapshot_t(c, pos.snapshot)->equiv;
1c6fdbd8 606
ef1669ff 607 if (pos.inode == w->cur_inum) {
6bd13057 608 w->first_this_inode = false;
ef1669ff 609 goto lookup_snapshot;
1c6fdbd8
KO		 610 }
611
ef1669ff
KO		 612 w->nr = 0;
613
614 for_each_btree_key(trans, iter, BTREE_ID_inodes, POS(0, pos.inode),
615 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
616 if (k.k->p.offset != pos.inode)
617 break;
618
619 if (k.k->type == KEY_TYPE_inode)
620 add_inode(c, w, bkey_s_c_to_inode(k));
621 }
622 bch2_trans_iter_exit(trans, &iter);
623
624 if (ret)
625 return ret;
626
627 w->cur_inum = pos.inode;
628 w->first_this_inode = true;
629lookup_snapshot:
630 for (i = 0; i < w->nr; i++)
631 if (bch2_snapshot_is_ancestor(c, pos.snapshot, w->d[i].snapshot))
632 goto found;
633 return INT_MAX;
634found:
635 BUG_ON(pos.snapshot > w->d[i].snapshot);
636
637 if (pos.snapshot != w->d[i].snapshot) {
638 ancestor_pos = i;
639
640 while (i && w->d[i - 1].snapshot > pos.snapshot)
641 --i;
642
643 ret = inode_walker_realloc(w);
644 if (ret)
645 return ret;
646
647 array_insert_item(w->d, w->nr, i, w->d[ancestor_pos]);
648 w->d[i].snapshot = pos.snapshot;
649 w->d[i].count = 0;
650 }
651
652 return i;
1c6fdbd8
KO		 653}
654
914f2786 655static int walk_inode(struct btree_trans *trans,
ef1669ff 656 struct inode_walker *w, struct bpos pos)
914f2786 657{
ef1669ff
KO		 658 return lockrestart_do(trans, __walk_inode(trans, w, pos));
659}
660
661static int __get_visible_inodes(struct btree_trans *trans,
662 struct inode_walker *w,
663 struct snapshots_seen *s,
664 u64 inum)
665{
666 struct bch_fs *c = trans->c;
667 struct btree_iter iter;
668 struct bkey_s_c k;
669 int ret;
670
671 w->nr = 0;
672
673 for_each_btree_key(trans, iter, BTREE_ID_inodes, POS(0, inum),
674 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
675 if (k.k->p.offset != inum)
676 break;
677
678 if (k.k->type != KEY_TYPE_inode)
679 continue;
680
681 if (ref_visible(c, s, s->pos.snapshot, k.k->p.snapshot)) {
682 add_inode(c, w, bkey_s_c_to_inode(k));
683 if (k.k->p.snapshot >= s->pos.snapshot)
684 break;
685 }
686 }
687 bch2_trans_iter_exit(trans, &iter);
688
689 return ret;
690}
691
692static int check_key_has_snapshot(struct btree_trans *trans,
693 struct btree_iter *iter,
694 struct bkey_s_c k)
695{
696 struct bch_fs *c = trans->c;
697 char buf[200];
698 int ret = 0;
699
700 if (fsck_err_on(!snapshot_t(c, k.k->p.snapshot)->equiv, c,
701 "key in missing snapshot: %s",
702 (bch2_bkey_val_to_text(&PBUF(buf), c, k), buf))) {
703 ret = __bch2_trans_do(trans, NULL, NULL, BTREE_INSERT_LAZY_RW,
704 bch2_btree_delete_at(trans, iter,
705 BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE));
706 return ret ?: -EINTR;
707 }
708fsck_err:
709 return ret;
914f2786
KO		 710}
711
7ac2c55e
KO		 712static int hash_redo_key(struct btree_trans *trans,
713 const struct bch_hash_desc desc,
714 struct bch_hash_info *hash_info,
715 struct btree_iter *k_iter, struct bkey_s_c k)
1c6fdbd8 716{
ef1669ff
KO		 717 bch_err(trans->c, "hash_redo_key() not implemented yet");
718 return -EINVAL;
719#if 0
e3b4b48c 720 struct bkey_i *delete;
1c6fdbd8 721 struct bkey_i *tmp;
1c6fdbd8 722
e3b4b48c
KO		 723 delete = bch2_trans_kmalloc(trans, sizeof(*delete));
724 if (IS_ERR(delete))
725 return PTR_ERR(delete);
726
b1fd23df
KO		 727 tmp = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
728 if (IS_ERR(tmp))
729 return PTR_ERR(tmp);
1c6fdbd8
KO		 730
731 bkey_reassemble(tmp, k);
732
e3b4b48c
KO		 733 bkey_init(&delete->k);
734 delete->k.p = k_iter->pos;
8c3f6da9
KO		 735 return bch2_btree_iter_traverse(k_iter) ?:
736 bch2_trans_update(trans, k_iter, delete, 0) ?:
bc3f8b25 737 bch2_hash_set(trans, desc, hash_info, k_iter->pos.inode, tmp, 0);
ef1669ff 738#endif
1c6fdbd8
KO		 739}
740
424eb881
KO		 741static int fsck_hash_delete_at(struct btree_trans *trans,
742 const struct bch_hash_desc desc,
1c6fdbd8 743 struct bch_hash_info *info,
424eb881 744 struct btree_iter *iter)
1c6fdbd8 745{
1c6fdbd8 746 int ret;
1c6fdbd8 747retry:
42d23732 748 ret = bch2_hash_delete_at(trans, desc, info, iter, 0) ?:
424eb881 749 bch2_trans_commit(trans, NULL, NULL,
134915f3
KO		 750 BTREE_INSERT_NOFAIL|
751 BTREE_INSERT_LAZY_RW);
424eb881
KO		 752 if (ret == -EINTR) {
753 ret = bch2_btree_iter_traverse(iter);
754 if (!ret)
755 goto retry;
756 }
1c6fdbd8 757
1c6fdbd8
KO		 758 return ret;
759}
760
7ac2c55e
KO		 761static int hash_check_key(struct btree_trans *trans,
762 const struct bch_hash_desc desc,
763 struct bch_hash_info *hash_info,
764 struct btree_iter *k_iter, struct bkey_s_c hash_k)
d69f41d6 765{
424eb881 766 struct bch_fs *c = trans->c;
67e0dd8f 767 struct btree_iter iter = { NULL };
d69f41d6 768 char buf[200];
7ac2c55e
KO		 769 struct bkey_s_c k;
770 u64 hash;
d69f41d6
KO		 771 int ret = 0;
772
7ac2c55e
KO		 773 if (hash_k.k->type != desc.key_type)
774 return 0;
775
776 hash = desc.hash_bkey(hash_info, hash_k);
777
778 if (likely(hash == hash_k.k->p.offset))
d69f41d6
KO		 779 return 0;
780
7ac2c55e
KO		 781 if (hash_k.k->p.offset < hash)
782 goto bad_hash;
d69f41d6 783
7ac2c55e
KO		 784 for_each_btree_key(trans, iter, desc.btree_id, POS(hash_k.k->p.inode, hash),
785 BTREE_ITER_SLOTS, k, ret) {
786 if (!bkey_cmp(k.k->p, hash_k.k->p))
d69f41d6
KO		 787 break;
788
7ac2c55e
KO		 789 if (fsck_err_on(k.k->type == desc.key_type &&
790 !desc.cmp_bkey(k, hash_k), c,
d69f41d6 791 "duplicate hash table keys:\n%s",
319f9ac3 792 (bch2_bkey_val_to_text(&PBUF(buf), c,
7ac2c55e
KO		 793 hash_k), buf))) {
794 ret = fsck_hash_delete_at(trans, desc, hash_info, k_iter);
d69f41d6
KO		 795 if (ret)
796 return ret;
797 ret = 1;
798 break;
799 }
d69f41d6 800
7ac2c55e 801 if (bkey_deleted(k.k)) {
67e0dd8f 802 bch2_trans_iter_exit(trans, &iter);
7ac2c55e 803 goto bad_hash;
1c6fdbd8 804 }
1c6fdbd8 805
7ac2c55e 806 }
67e0dd8f 807 bch2_trans_iter_exit(trans, &iter);
1c6fdbd8 808 return ret;
7ac2c55e
KO		 809bad_hash:
810 if (fsck_err(c, "hash table key at wrong offset: btree %u inode %llu offset %llu, "
e3b4b48c
KO		 811 "hashed to %llu\n%s",
812 desc.btree_id, hash_k.k->p.inode, hash_k.k->p.offset, hash,
7ac2c55e 813 (bch2_bkey_val_to_text(&PBUF(buf), c, hash_k), buf)) == FSCK_ERR_IGNORE)
741daa5b
KO		 814 return 0;
815
7ac2c55e
KO		 816 ret = __bch2_trans_do(trans, NULL, NULL,
817 BTREE_INSERT_NOFAIL|BTREE_INSERT_LAZY_RW,
818 hash_redo_key(trans, desc, hash_info, k_iter, hash_k));
819 if (ret) {
820 bch_err(c, "hash_redo_key err %i", ret);
821 return ret;
741daa5b 822 }
7ac2c55e 823 return -EINTR;
741daa5b 824fsck_err:
741daa5b 825 return ret;
741daa5b
KO		 826}
827
5c16add5
KO		 828static int check_inode(struct btree_trans *trans,
829 struct btree_iter *iter,
ef1669ff
KO		 830 struct bch_inode_unpacked *prev,
831 struct bch_inode_unpacked u)
5c16add5
KO		 832{
833 struct bch_fs *c = trans->c;
5c16add5
KO		 834 bool do_update = false;
835 int ret = 0;
836
ef1669ff
KO		 837 if (fsck_err_on(prev &&
838 (prev->bi_hash_seed != u.bi_hash_seed ||
839 mode_to_type(prev->bi_mode) != mode_to_type(u.bi_mode)), c,
840 "inodes in different snapshots don't match")) {
841 bch_err(c, "repair not implemented yet");
842 return -EINVAL;
843 }
5c16add5
KO		 844
845 if (u.bi_flags & BCH_INODE_UNLINKED &&
846 (!c->sb.clean ||
847 fsck_err(c, "filesystem marked clean, but inode %llu unlinked",
848 u.bi_inum))) {
5c16add5
KO		 849 bch2_trans_unlock(trans);
850 bch2_fs_lazy_rw(c);
851
ef1669ff 852 ret = fsck_inode_rm(trans, u.bi_inum, iter->pos.snapshot);
5c16add5
KO		 853 if (ret)
854 bch_err(c, "error in fsck: error %i while deleting inode", ret);
855 return ret;
856 }
857
858 if (u.bi_flags & BCH_INODE_I_SIZE_DIRTY &&
859 (!c->sb.clean ||
860 fsck_err(c, "filesystem marked clean, but inode %llu has i_size dirty",
861 u.bi_inum))) {
862 bch_verbose(c, "truncating inode %llu", u.bi_inum);
863
864 bch2_trans_unlock(trans);
865 bch2_fs_lazy_rw(c);
866
867 /*
868 * XXX: need to truncate partial blocks too here - or ideally
869 * just switch units to bytes and that issue goes away
870 */
871 ret = bch2_btree_delete_range_trans(trans, BTREE_ID_extents,
ef1669ff
KO		 872 SPOS(u.bi_inum, round_up(u.bi_size, block_bytes(c)) >> 9,
873 iter->pos.snapshot),
5c16add5 874 POS(u.bi_inum, U64_MAX),
ef1669ff 875 0, NULL);
5c16add5
KO		 876 if (ret) {
877 bch_err(c, "error in fsck: error %i truncating inode", ret);
878 return ret;
879 }
880
881 /*
882 * We truncated without our normal sector accounting hook, just
883 * make sure we recalculate it:
884 */
885 u.bi_flags |= BCH_INODE_I_SECTORS_DIRTY;
886
887 u.bi_flags &= ~BCH_INODE_I_SIZE_DIRTY;
888 do_update = true;
889 }
890
891 if (u.bi_flags & BCH_INODE_I_SECTORS_DIRTY &&
892 (!c->sb.clean ||
893 fsck_err(c, "filesystem marked clean, but inode %llu has i_sectors dirty",
894 u.bi_inum))) {
895 s64 sectors;
896
897 bch_verbose(c, "recounting sectors for inode %llu",
898 u.bi_inum);
899
ef1669ff 900 sectors = bch2_count_inode_sectors(trans, u.bi_inum, iter->pos.snapshot);
5c16add5
KO		 901 if (sectors < 0) {
902 bch_err(c, "error in fsck: error %i recounting inode sectors",
903 (int) sectors);
904 return sectors;
905 }
906
907 u.bi_sectors = sectors;
908 u.bi_flags &= ~BCH_INODE_I_SECTORS_DIRTY;
909 do_update = true;
910 }
911
d3ff7fec
KO		 912 if (u.bi_flags & BCH_INODE_BACKPTR_UNTRUSTED) {
913 u.bi_dir = 0;
914 u.bi_dir_offset = 0;
915 u.bi_flags &= ~BCH_INODE_BACKPTR_UNTRUSTED;
5c16add5
KO		 916 do_update = true;
917 }
918
919 if (do_update) {
ef1669ff 920 ret = write_inode(trans, &u, iter->pos.snapshot);
5c16add5
KO		 921 if (ret)
922 bch_err(c, "error in fsck: error %i "
923 "updating inode", ret);
924 }
925fsck_err:
926 return ret;
927}
928
929noinline_for_stack
930static int check_inodes(struct bch_fs *c, bool full)
931{
932 struct btree_trans trans;
67e0dd8f 933 struct btree_iter iter;
5c16add5
KO		 934 struct bkey_s_c k;
935 struct bkey_s_c_inode inode;
ef1669ff 936 struct bch_inode_unpacked prev, u;
5c16add5
KO		 937 int ret;
938
ef1669ff
KO		 939 memset(&prev, 0, sizeof(prev));
940
5c16add5
KO		 941 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
942
909004d2
KO		 943 for_each_btree_key(&trans, iter, BTREE_ID_inodes, POS_MIN,
944 BTREE_ITER_INTENT|
ef1669ff
KO		 945 BTREE_ITER_PREFETCH|
946 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
947 ret = check_key_has_snapshot(&trans, &iter, k);
948 if (ret)
949 break;
950
951 /*
952 * if snapshot id isn't a leaf node, skip it - deletion in
953 * particular is not atomic, so on the internal snapshot nodes
954 * we can see inodes marked for deletion after a clean shutdown
955 */
956 if (bch2_snapshot_internal_node(c, k.k->p.snapshot))
957 continue;
958
5c16add5
KO		 959 if (k.k->type != KEY_TYPE_inode)
960 continue;
961
962 inode = bkey_s_c_to_inode(k);
963
ef1669ff
KO		 964 if (!full &&
965 !(inode.v->bi_flags & (BCH_INODE_I_SIZE_DIRTY|
966 BCH_INODE_I_SECTORS_DIRTY|
967 BCH_INODE_UNLINKED)))
968 continue;
969
970 BUG_ON(bch2_inode_unpack(inode, &u));
971
972 ret = check_inode(&trans, &iter,
973 full && prev.bi_inum == u.bi_inum
974 ? &prev : NULL, u);
975 if (ret)
976 break;
977
978 prev = u;
5c16add5 979 }
67e0dd8f 980 bch2_trans_iter_exit(&trans, &iter);
5c16add5
KO		 981
982 BUG_ON(ret == -EINTR);
983
9a796fdb
KO		 984 bch2_trans_exit(&trans);
985 return ret;
5c16add5
KO		 986}
987
ef1669ff
KO		 988noinline_for_stack
989static int check_subvols(struct bch_fs *c)
990{
991 struct btree_trans trans;
992 struct btree_iter iter;
993 struct bkey_s_c k;
994 int ret;
995
996 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
997
998 for_each_btree_key(&trans, iter, BTREE_ID_subvolumes, POS_MIN,
999 0, k, ret) {
1000 }
1001 bch2_trans_iter_exit(&trans, &iter);
1002
1003 bch2_trans_exit(&trans);
1004 return ret;
1005}
1006
1007/*
1008 * Checking for overlapping extents needs to be reimplemented
1009 */
1010#if 0
abcecb49 1011static int fix_overlapping_extent(struct btree_trans *trans,
e3e464ac
KO		 1012 struct bkey_s_c k, struct bpos cut_at)
1013{
67e0dd8f 1014 struct btree_iter iter;
e3e464ac
KO		 1015 struct bkey_i *u;
1016 int ret;
1017
1018 u = bch2_trans_kmalloc(trans, bkey_bytes(k.k));
1019 ret = PTR_ERR_OR_ZERO(u);
1020 if (ret)
1021 return ret;
1022
1023 bkey_reassemble(u, k);
1024 bch2_cut_front(cut_at, u);
1025
e3e464ac 1026
ef1669ff
KO		 1027 /*
1028 * We don't want to go through the extent_handle_overwrites path:
1029 *
1030 * XXX: this is going to screw up disk accounting, extent triggers
1031 * assume things about extent overwrites - we should be running the
1032 * triggers manually here
1033 */
1034 bch2_trans_iter_init(trans, &iter, BTREE_ID_extents, u->k.p,
1035 BTREE_ITER_INTENT|BTREE_ITER_NOT_EXTENTS);
1036
1037 BUG_ON(iter.flags & BTREE_ITER_IS_EXTENTS);
1038 ret = bch2_btree_iter_traverse(&iter) ?:
1039 bch2_trans_update(trans, &iter, u, BTREE_TRIGGER_NORUN) ?:
1040 bch2_trans_commit(trans, NULL, NULL,
1041 BTREE_INSERT_NOFAIL|
1042 BTREE_INSERT_LAZY_RW);
1043 bch2_trans_iter_exit(trans, &iter);
1044 return ret;
1045}
1046#endif
1047
1048static int inode_backpointer_exists(struct btree_trans *trans,
1049 struct bch_inode_unpacked *inode,
1050 u32 snapshot)
1051{
1052 struct btree_iter iter;
1053 struct bkey_s_c k;
488f9776
KO		 1054 u32 target_subvol, target_snapshot;
1055 u64 target_inum;
ef1669ff
KO		 1056 int ret;
1057
1058 bch2_trans_iter_init(trans, &iter, BTREE_ID_dirents,
1059 SPOS(inode->bi_dir, inode->bi_dir_offset, snapshot), 0);
1060 k = bch2_btree_iter_peek_slot(&iter);
1061 ret = bkey_err(k);
1062 if (ret)
1063 goto out;
1064 if (k.k->type != KEY_TYPE_dirent)
1065 goto out;
1066
488f9776
KO		 1067 ret = __bch2_dirent_read_target(trans, bkey_s_c_to_dirent(k),
1068 &target_subvol,
1069 &target_snapshot,
1070 &target_inum,
1071 true);
1072 if (ret)
1073 goto out;
1074
1075 ret = target_inum == inode->bi_inum;
ef1669ff
KO		 1076out:
1077 bch2_trans_iter_exit(trans, &iter);
1078 return ret;
1079}
1080
1081static bool inode_backpointer_matches(struct bkey_s_c_dirent d,
1082 struct bch_inode_unpacked *inode)
1083{
1084 return d.k->p.inode == inode->bi_dir &&
1085 d.k->p.offset == inode->bi_dir_offset;
1086}
1087
1088static int check_i_sectors(struct btree_trans *trans, struct inode_walker *w)
1089{
1090 struct bch_fs *c = trans->c;
1091 struct inode_walker_entry *i;
1092 int ret = 0, ret2 = 0;
1093 s64 count2;
1094
1095 for (i = w->d; i < w->d + w->nr; i++) {
1096 if (i->inode.bi_sectors == i->count)
1097 continue;
1098
1099 count2 = lockrestart_do(trans,
1100 bch2_count_inode_sectors(trans, w->cur_inum, i->snapshot));
1101
1102 if (i->count != count2) {
1103 bch_err(c, "fsck counted i_sectors wrong: got %llu should be %llu",
1104 i->count, count2);
1105 i->count = count2;
1106 if (i->inode.bi_sectors == i->count)
1107 continue;
1108 }
1109
1110 if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_I_SECTORS_DIRTY), c,
1111 "inode %llu:%u has incorrect i_sectors: got %llu, should be %llu",
1112 w->cur_inum, i->snapshot,
1113 i->inode.bi_sectors, i->count) == FSCK_ERR_IGNORE)
1114 continue;
1115
1116 i->inode.bi_sectors = i->count;
1117 ret = write_inode(trans, &i->inode, i->snapshot);
1118 if (ret)
1119 break;
1120 ret2 = -EINTR;
1121 }
1122fsck_err:
1123 return ret ?: ret2;
1124}
1125
1126static int check_extent(struct btree_trans *trans, struct btree_iter *iter,
1127 struct inode_walker *inode,
1128 struct snapshots_seen *s)
1129{
1130 struct bch_fs *c = trans->c;
1131 struct bkey_s_c k;
1132 struct inode_walker_entry *i;
1133 char buf[200];
1134 int ret = 0;
1135
1136 k = bch2_btree_iter_peek(iter);
1137 if (!k.k)
1138 return 0;
1139
1140 ret = bkey_err(k);
1141 if (ret)
1142 return ret;
1143
1144 ret = check_key_has_snapshot(trans, iter, k);
1145 if (ret)
1146 return ret;
1147
1148 ret = snapshots_seen_update(c, s, k.k->p);
1149 if (ret)
1150 return ret;
1151
1152 if (k.k->type == KEY_TYPE_whiteout)
1153 return 0;
1154
1155 if (inode->cur_inum != k.k->p.inode) {
1156 ret = check_i_sectors(trans, inode);
1157 if (ret)
1158 return ret;
1159 }
1160#if 0
1161 if (bkey_cmp(prev.k->k.p, bkey_start_pos(k.k)) > 0) {
1162 char buf1[200];
1163 char buf2[200];
1164
1165 bch2_bkey_val_to_text(&PBUF(buf1), c, bkey_i_to_s_c(prev.k));
1166 bch2_bkey_val_to_text(&PBUF(buf2), c, k);
1167
1168 if (fsck_err(c, "overlapping extents:\n%s\n%s", buf1, buf2))
1169 return fix_overlapping_extent(trans, k, prev.k->k.p) ?: -EINTR;
1170 }
1171#endif
1172 ret = __walk_inode(trans, inode, k.k->p);
1173 if (ret < 0)
1174 return ret;
1175
1176 if (fsck_err_on(ret == INT_MAX, c,
1177 "extent in missing inode:\n %s",
1178 (bch2_bkey_val_to_text(&PBUF(buf), c, k), buf)))
1179 return __bch2_trans_do(trans, NULL, NULL, BTREE_INSERT_LAZY_RW,
1180 bch2_btree_delete_at(trans, iter,
1181 BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE));
1182
1183 if (ret == INT_MAX)
1184 return 0;
e3e464ac 1185
ef1669ff
KO		 1186 i = inode->d + ret;
1187 ret = 0;
e3e464ac 1188
ef1669ff
KO		 1189 if (fsck_err_on(!S_ISREG(i->inode.bi_mode) &&
1190 !S_ISLNK(i->inode.bi_mode), c,
1191 "extent in non regular inode mode %o:\n %s",
1192 i->inode.bi_mode,
1193 (bch2_bkey_val_to_text(&PBUF(buf), c, k), buf)))
1194 return __bch2_trans_do(trans, NULL, NULL, BTREE_INSERT_LAZY_RW,
1195 bch2_btree_delete_at(trans, iter,
1196 BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE));
1197
1198 if (!bch2_snapshot_internal_node(c, k.k->p.snapshot)) {
1199 for_each_visible_inode(c, s, inode, k.k->p.snapshot, i) {
1200 if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_I_SIZE_DIRTY) &&
1201 k.k->type != KEY_TYPE_reservation &&
1202 k.k->p.offset > round_up(i->inode.bi_size, block_bytes(c)) >> 9, c,
1203 "extent type %u offset %llu past end of inode %llu, i_size %llu",
1204 k.k->type, k.k->p.offset, k.k->p.inode, i->inode.bi_size)) {
1205 bch2_fs_lazy_rw(c);
1206 return bch2_btree_delete_range_trans(trans, BTREE_ID_extents,
1207 SPOS(k.k->p.inode, round_up(i->inode.bi_size, block_bytes(c)) >> 9,
1208 k.k->p.snapshot),
1209 POS(k.k->p.inode, U64_MAX),
1210 0, NULL) ?: -EINTR;
1211 }
1212 }
1213 }
8a85b20c 1214
ef1669ff
KO		 1215 if (bkey_extent_is_allocation(k.k))
1216 for_each_visible_inode(c, s, inode, k.k->p.snapshot, i)
1217 i->count += k.k->size;
1218#if 0
1219 bch2_bkey_buf_reassemble(&prev, c, k);
1220#endif
8a85b20c 1221
ef1669ff 1222fsck_err:
8a85b20c
KO		 1223 return ret;
1224}
1225
1c6fdbd8
KO		 1226/*
1227 * Walk extents: verify that extents have a corresponding S_ISREG inode, and
1228 * that i_size an i_sectors are consistent
1229 */
1230noinline_for_stack
1231static int check_extents(struct bch_fs *c)
1232{
1233 struct inode_walker w = inode_walker_init();
ef1669ff 1234 struct snapshots_seen s;
424eb881 1235 struct btree_trans trans;
67e0dd8f 1236 struct btree_iter iter;
1c6fdbd8
KO		 1237 int ret = 0;
1238
ef1669ff
KO		 1239#if 0
1240 struct bkey_buf prev;
07a1006a 1241 bch2_bkey_buf_init(&prev);
f7005e01 1242 prev.k->k = KEY(0, 0, 0);
ef1669ff
KO		 1243#endif
1244 snapshots_seen_init(&s);
20bceecb 1245 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
424eb881 1246
1c6fdbd8
KO		 1247 bch_verbose(c, "checking extents");
1248
67e0dd8f
KO		 1249 bch2_trans_iter_init(&trans, &iter, BTREE_ID_extents,
1250 POS(BCACHEFS_ROOT_INO, 0),
1251 BTREE_ITER_INTENT|
ef1669ff
KO		 1252 BTREE_ITER_PREFETCH|
1253 BTREE_ITER_ALL_SNAPSHOTS);
1254
1255 do {
1256 ret = lockrestart_do(&trans,
1257 check_extent(&trans, &iter, &w, &s));
1258 if (ret)
1259 break;
1260 } while (bch2_btree_iter_advance(&iter));
1261 bch2_trans_iter_exit(&trans, &iter);
1262#if 0
1263 bch2_bkey_buf_exit(&prev, c);
1264#endif
1265 inode_walker_exit(&w);
1266 bch2_trans_exit(&trans);
1267 snapshots_seen_exit(&s);
1268
1269 return ret;
1270}
1271
1272static int check_subdir_count(struct btree_trans *trans, struct inode_walker *w)
1273{
1274 struct bch_fs *c = trans->c;
1275 struct inode_walker_entry *i;
1276 int ret = 0, ret2 = 0;
1277 s64 count2;
1278
1279 for (i = w->d; i < w->d + w->nr; i++) {
1280 if (i->inode.bi_nlink == i->count)
1281 continue;
1282
1283 count2 = lockrestart_do(trans,
1284 bch2_count_subdirs(trans, w->cur_inum, i->snapshot));
1285
1286 if (i->count != count2) {
1287 bch_err(c, "fsck counted subdirectories wrong: got %llu should be %llu",
1288 i->count, count2);
1289 i->count = count2;
1290 if (i->inode.bi_nlink == i->count)
1291 continue;
1292 }
1293
1294 if (fsck_err_on(i->inode.bi_nlink != i->count, c,
1295 "directory %llu:%u with wrong i_nlink: got %u, should be %llu",
1296 w->cur_inum, i->snapshot, i->inode.bi_nlink, i->count)) {
1297 i->inode.bi_nlink = i->count;
1298 ret = write_inode(trans, &i->inode, i->snapshot);
abcecb49
KO		 1299 if (ret)
1300 break;
ef1669ff 1301 ret2 = -EINTR;
abcecb49 1302 }
ef1669ff
KO		 1303 }
1304fsck_err:
1305 return ret ?: ret2;
1306}
abcecb49 1307
ef1669ff
KO		 1308static int check_dirent_target(struct btree_trans *trans,
1309 struct btree_iter *iter,
1310 struct bkey_s_c_dirent d,
1311 struct bch_inode_unpacked *target,
1312 u32 target_snapshot)
1313{
1314 struct bch_fs *c = trans->c;
1315 bool backpointer_exists = true;
1316 char buf[200];
1317 int ret = 0;
1318
1319 if (!target->bi_dir &&
1320 !target->bi_dir_offset) {
1321 target->bi_dir = d.k->p.inode;
1322 target->bi_dir_offset = d.k->p.offset;
1323
1324 ret = write_inode(trans, target, target_snapshot);
1325 if (ret)
1326 goto err;
1327 }
1328
1329 if (!inode_backpointer_matches(d, target)) {
1330 ret = inode_backpointer_exists(trans, target, d.k->p.snapshot);
1331 if (ret < 0)
1332 goto err;
e3e464ac 1333
ef1669ff
KO		 1334 backpointer_exists = ret;
1335 ret = 0;
e3e464ac 1336
ef1669ff
KO		 1337 if (fsck_err_on(S_ISDIR(target->bi_mode) &&
1338 backpointer_exists, c,
1339 "directory %llu with multiple links",
1340 target->bi_inum)) {
1341 ret = remove_dirent(trans, d.k->p);
1342 if (ret)
1343 goto err;
1344 return 0;
e3e464ac 1345 }
e3e464ac 1346
ef1669ff
KO		 1347 if (fsck_err_on(backpointer_exists &&
1348 !target->bi_nlink, c,
1349 "inode %llu has multiple links but i_nlink 0",
1350 target->bi_inum)) {
1351 target->bi_nlink++;
1352 target->bi_flags &= ~BCH_INODE_UNLINKED;
1c6fdbd8 1353
ef1669ff
KO		 1354 ret = write_inode(trans, target, target_snapshot);
1355 if (ret)
1356 goto err;
1c6fdbd8
KO		 1357 }
1358
ef1669ff 1359 if (fsck_err_on(!backpointer_exists, c,
6e0c886d 1360 "inode %llu:%u has wrong backpointer:\n"
ef1669ff
KO		 1361 "got %llu:%llu\n"
1362 "should be %llu:%llu",
6e0c886d 1363 target->bi_inum, target_snapshot,
ef1669ff
KO		 1364 target->bi_dir,
1365 target->bi_dir_offset,
1366 d.k->p.inode,
1367 d.k->p.offset)) {
1368 target->bi_dir = d.k->p.inode;
1369 target->bi_dir_offset = d.k->p.offset;
1370
1371 ret = write_inode(trans, target, target_snapshot);
1372 if (ret)
1373 goto err;
1c6fdbd8 1374 }
ef1669ff 1375 }
1c6fdbd8 1376
ef1669ff
KO		 1377 if (fsck_err_on(vfs_d_type(d.v->d_type) != mode_to_type(target->bi_mode), c,
1378 "incorrect d_type: should be %u:\n%s",
1379 mode_to_type(target->bi_mode),
1380 (bch2_bkey_val_to_text(&PBUF(buf), c, d.s_c), buf))) {
1381 struct bkey_i_dirent *n;
1c6fdbd8 1382
ef1669ff
KO		 1383 n = kmalloc(bkey_bytes(d.k), GFP_KERNEL);
1384 if (!n) {
1385 ret = -ENOMEM;
1386 goto err;
1387 }
1388
1389 bkey_reassemble(&n->k_i, d.s_c);
1390 n->v.d_type = mode_to_type(target->bi_mode);
1391
1392 ret = __bch2_trans_do(trans, NULL, NULL,
1393 BTREE_INSERT_NOFAIL|
1394 BTREE_INSERT_LAZY_RW,
1395 bch2_trans_update(trans, iter, &n->k_i, 0));
1396 kfree(n);
1397 if (ret)
1398 goto err;
1c6fdbd8 1399 }
ef1669ff 1400err:
1c6fdbd8 1401fsck_err:
ef1669ff 1402 return ret;
1c6fdbd8
KO		 1403}
1404
914f2786
KO		 1405static int check_dirent(struct btree_trans *trans, struct btree_iter *iter,
1406 struct bch_hash_info *hash_info,
ef1669ff
KO		 1407 struct inode_walker *dir,
1408 struct inode_walker *target,
1409 struct snapshots_seen *s)
1c6fdbd8 1410{
914f2786 1411 struct bch_fs *c = trans->c;
1c6fdbd8 1412 struct bkey_s_c k;
914f2786 1413 struct bkey_s_c_dirent d;
ef1669ff 1414 struct inode_walker_entry *i;
914f2786 1415 u32 target_snapshot;
b9e1adf5 1416 u32 target_subvol;
ef1669ff 1417 u64 target_inum;
1c6fdbd8 1418 char buf[200];
914f2786 1419 int ret;
1c6fdbd8 1420
914f2786
KO		 1421 k = bch2_btree_iter_peek(iter);
1422 if (!k.k)
38200544 1423 return 0;
1c6fdbd8 1424
914f2786
KO		 1425 ret = bkey_err(k);
1426 if (ret)
1427 return ret;
d69f41d6 1428
ef1669ff
KO		 1429 ret = check_key_has_snapshot(trans, iter, k);
1430 if (ret)
1431 return ret;
1c6fdbd8 1432
ef1669ff 1433 ret = snapshots_seen_update(c, s, k.k->p);
914f2786
KO		 1434 if (ret)
1435 return ret;
8a85b20c 1436
ef1669ff
KO		 1437 if (k.k->type == KEY_TYPE_whiteout)
1438 return 0;
1439
1440 if (dir->cur_inum != k.k->p.inode) {
1441 ret = check_subdir_count(trans, dir);
1442 if (ret)
1443 return ret;
1444 }
1445
1446 ret = __walk_inode(trans, dir, k.k->p);
1447 if (ret < 0)
1448 return ret;
1c6fdbd8 1449
ef1669ff 1450 if (fsck_err_on(ret == INT_MAX, c,
914f2786 1451 "dirent in nonexisting directory:\n%s",
ef1669ff
KO		 1452 (bch2_bkey_val_to_text(&PBUF(buf), c, k), buf)))
1453 return __bch2_trans_do(trans, NULL, NULL, BTREE_INSERT_LAZY_RW,
1454 bch2_btree_delete_at(trans, iter,
1455 BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE));
1456
1457 if (ret == INT_MAX)
1458 return 0;
1459
1460 i = dir->d + ret;
1461 ret = 0;
1462
1463 if (fsck_err_on(!S_ISDIR(i->inode.bi_mode), c,
914f2786 1464 "dirent in non directory inode type %u:\n%s",
ef1669ff 1465 mode_to_type(i->inode.bi_mode),
914f2786
KO		 1466 (bch2_bkey_val_to_text(&PBUF(buf), c, k), buf)))
1467 return __bch2_trans_do(trans, NULL, NULL, 0,
1468 bch2_btree_delete_at(trans, iter, 0));
8a85b20c 1469
ef1669ff
KO		 1470 if (dir->first_this_inode)
1471 *hash_info = bch2_hash_info_init(c, &dir->d[0].inode);
1c6fdbd8 1472
914f2786
KO		 1473 ret = hash_check_key(trans, bch2_dirent_hash_desc,
1474 hash_info, iter, k);
1475 if (ret < 0)
1476 return ret;
1477 if (ret) /* dirent has been deleted */
1478 return 0;
7ac2c55e 1479
914f2786
KO		 1480 if (k.k->type != KEY_TYPE_dirent)
1481 return 0;
1c6fdbd8 1482
914f2786 1483 d = bkey_s_c_to_dirent(k);
1c6fdbd8 1484
ef1669ff 1485 ret = __bch2_dirent_read_target(trans, d,
6fed42bb
KO		 1486 &target_subvol,
1487 &target_snapshot,
ef1669ff
KO		 1488 &target_inum,
1489 true);
b9e1adf5
KO		 1490 if (ret && ret != -ENOENT)
1491 return ret;
1492
ef1669ff
KO		 1493 if (fsck_err_on(ret, c,
1494 "dirent points to missing subvolume %llu",
1495 le64_to_cpu(d.v->d_inum)))
1496 return remove_dirent(trans, d.k->p);
1c6fdbd8 1497
ef1669ff
KO		 1498 if (target_subvol) {
1499 struct bch_inode_unpacked subvol_root;
914f2786 1500
ef1669ff
KO		 1501 ret = __lookup_inode(trans, target_inum,
1502 &subvol_root, &target_snapshot);
1503 if (ret && ret != -ENOENT)
1504 return ret;
914f2786 1505
ef1669ff
KO		 1506 if (fsck_err_on(ret, c,
1507 "subvolume %u points to missing subvolume root %llu",
1508 target_subvol,
1509 target_inum)) {
1510 bch_err(c, "repair not implemented yet");
1511 return -EINVAL;
1512 }
914f2786 1513
ef1669ff
KO		 1514 if (fsck_err_on(subvol_root.bi_subvol != target_subvol, c,
1515 "subvol root %llu has wrong bi_subvol field: got %u, should be %u",
1516 target_inum,
1517 subvol_root.bi_subvol, target_subvol)) {
1518 subvol_root.bi_subvol = target_subvol;
1519 ret = write_inode(trans, &subvol_root, target_snapshot);
1520 if (ret)
1521 return ret;
1522 }
914f2786 1523
ef1669ff
KO		 1524 ret = check_dirent_target(trans, iter, d, &subvol_root,
1525 target_snapshot);
914f2786
KO		 1526 if (ret)
1527 return ret;
ef1669ff
KO		 1528 } else {
1529 ret = __get_visible_inodes(trans, target, s, target_inum);
1530 if (ret)
914f2786 1531 return ret;
1c6fdbd8 1532
ef1669ff
KO		 1533 if (fsck_err_on(!target->nr, c,
1534 "dirent points to missing inode:\n%s",
1535 (bch2_bkey_val_to_text(&PBUF(buf), c,
1536 k), buf))) {
1537 ret = remove_dirent(trans, d.k->p);
1538 if (ret)
1539 return ret;
1c6fdbd8
KO		 1540 }
1541
ef1669ff
KO		 1542 for (i = target->d; i < target->d + target->nr; i++) {
1543 ret = check_dirent_target(trans, iter, d,
1544 &i->inode, i->snapshot);
1545 if (ret)
1546 return ret;
d3ff7fec 1547 }
914f2786 1548 }
d3ff7fec 1549
ef1669ff
KO		 1550 if (d.v->d_type == DT_DIR)
1551 for_each_visible_inode(c, s, dir, d.k->p.snapshot, i)
1552 i->count++;
1c6fdbd8 1553
914f2786
KO		 1554fsck_err:
1555 return ret;
1556}
1c6fdbd8 1557
914f2786
KO		 1558/*
1559 * Walk dirents: verify that they all have a corresponding S_ISDIR inode,
1560 * validate d_type
1561 */
1562noinline_for_stack
1563static int check_dirents(struct bch_fs *c)
1564{
ef1669ff
KO		 1565 struct inode_walker dir = inode_walker_init();
1566 struct inode_walker target = inode_walker_init();
1567 struct snapshots_seen s;
914f2786
KO		 1568 struct bch_hash_info hash_info;
1569 struct btree_trans trans;
67e0dd8f 1570 struct btree_iter iter;
914f2786 1571 int ret = 0;
1c6fdbd8 1572
914f2786
KO		 1573 bch_verbose(c, "checking dirents");
1574
ef1669ff 1575 snapshots_seen_init(&s);
914f2786 1576 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
1c6fdbd8 1577
67e0dd8f
KO		 1578 bch2_trans_iter_init(&trans, &iter, BTREE_ID_dirents,
1579 POS(BCACHEFS_ROOT_INO, 0),
1580 BTREE_ITER_INTENT|
ef1669ff
KO		 1581 BTREE_ITER_PREFETCH|
1582 BTREE_ITER_ALL_SNAPSHOTS);
914f2786 1583
38200544 1584 do {
914f2786 1585 ret = lockrestart_do(&trans,
ef1669ff
KO		 1586 check_dirent(&trans, &iter, &hash_info,
1587 &dir, &target, &s));
914f2786
KO		 1588 if (ret)
1589 break;
67e0dd8f
KO		 1590 } while (bch2_btree_iter_advance(&iter));
1591 bch2_trans_iter_exit(&trans, &iter);
914f2786 1592
ef1669ff
KO		 1593 bch2_trans_exit(&trans);
1594 snapshots_seen_exit(&s);
1595 inode_walker_exit(&dir);
1596 inode_walker_exit(&target);
1597 return ret;
1c6fdbd8
KO		 1598}
1599
1600/*
1601 * Walk xattrs: verify that they all have a corresponding inode
1602 */
1603noinline_for_stack
1604static int check_xattrs(struct bch_fs *c)
1605{
1606 struct inode_walker w = inode_walker_init();
7ac2c55e 1607 struct bch_hash_info hash_info;
d69f41d6 1608 struct btree_trans trans;
67e0dd8f 1609 struct btree_iter iter;
1c6fdbd8
KO		 1610 struct bkey_s_c k;
1611 int ret = 0;
1612
1613 bch_verbose(c, "checking xattrs");
1614
20bceecb 1615 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
d69f41d6 1616
67e0dd8f
KO		 1617 bch2_trans_iter_init(&trans, &iter, BTREE_ID_xattrs,
1618 POS(BCACHEFS_ROOT_INO, 0),
1619 BTREE_ITER_INTENT|
ef1669ff
KO		 1620 BTREE_ITER_PREFETCH|
1621 BTREE_ITER_ALL_SNAPSHOTS);
6bd13057 1622retry:
ef1669ff
KO		 1623 bch2_trans_begin(&trans);
1624
67e0dd8f 1625 while ((k = bch2_btree_iter_peek(&iter)).k &&
abcecb49 1626 !(ret = bkey_err(k))) {
ef1669ff 1627 ret = check_key_has_snapshot(&trans, &iter, k);
1c6fdbd8
KO		 1628 if (ret)
1629 break;
1630
ef1669ff
KO		 1631 ret = walk_inode(&trans, &w, k.k->p);
1632 if (ret < 0)
1633 break;
1634
1635 if (fsck_err_on(ret == INT_MAX, c,
1c6fdbd8
KO		 1636 "xattr for missing inode %llu",
1637 k.k->p.inode)) {
67e0dd8f 1638 ret = bch2_btree_delete_at(&trans, &iter, 0);
1c6fdbd8 1639 if (ret)
abcecb49 1640 break;
1c6fdbd8
KO		 1641 continue;
1642 }
1643
ef1669ff
KO		 1644 if (ret == INT_MAX)
1645 goto next;
1646 ret = 0;
1647
1648 if (w.first_this_inode)
1649 hash_info = bch2_hash_info_init(c, &w.d[0].inode);
1c6fdbd8 1650
424eb881 1651 ret = hash_check_key(&trans, bch2_xattr_hash_desc,
67e0dd8f 1652 &hash_info, &iter, k);
1c6fdbd8 1653 if (ret)
abcecb49 1654 break;
ef1669ff 1655next:
67e0dd8f 1656 bch2_btree_iter_advance(&iter);
1c6fdbd8 1657 }
1c6fdbd8 1658fsck_err:
6bd13057
KO		 1659 if (ret == -EINTR)
1660 goto retry;
abcecb49 1661
67e0dd8f 1662 bch2_trans_iter_exit(&trans, &iter);
9a796fdb
KO		 1663 bch2_trans_exit(&trans);
1664 return ret;
1c6fdbd8
KO		 1665}
1666
1667/* Get root directory, create if it doesn't exist: */
ef1669ff 1668static int check_root(struct bch_fs *c)
1c6fdbd8 1669{
ef1669ff
KO		 1670 struct btree_trans trans;
1671 struct bch_inode_unpacked root_inode;
d3ff7fec 1672 u32 snapshot;
ef1669ff 1673 u64 inum;
1c6fdbd8
KO		 1674 int ret;
1675
ef1669ff
KO		 1676 bch2_trans_init(&trans, c, 0, 0);
1677
1c6fdbd8
KO		 1678 bch_verbose(c, "checking root directory");
1679
ef1669ff 1680 ret = subvol_lookup(&trans, BCACHEFS_ROOT_SUBVOL, &snapshot, &inum);
1c6fdbd8
KO		 1681 if (ret && ret != -ENOENT)
1682 return ret;
1683
ef1669ff
KO		 1684 if (mustfix_fsck_err_on(ret, c, "root subvol missing")) {
1685 struct bkey_i_subvolume root_subvol;
1c6fdbd8 1686
ef1669ff
KO		 1687 snapshot = U32_MAX;
1688 inum = BCACHEFS_ROOT_INO;
1c6fdbd8 1689
ef1669ff
KO		 1690 bkey_subvolume_init(&root_subvol.k_i);
1691 root_subvol.k.p.offset = BCACHEFS_ROOT_SUBVOL;
1692 root_subvol.v.flags = 0;
1693 root_subvol.v.snapshot = cpu_to_le32(snapshot);
1694 root_subvol.v.inode = cpu_to_le64(inum);
1695 ret = __bch2_trans_do(&trans, NULL, NULL,
1696 BTREE_INSERT_NOFAIL|
1697 BTREE_INSERT_LAZY_RW,
1698 __bch2_btree_insert(&trans, BTREE_ID_subvolumes, &root_subvol.k_i));
1699 if (ret) {
1700 bch_err(c, "error writing root subvol: %i", ret);
1701 goto err;
1702 }
1703
1704 }
1705
1706 ret = lookup_inode(&trans, BCACHEFS_ROOT_INO, &root_inode, &snapshot);
1707 if (ret && ret != -ENOENT)
1708 return ret;
1c6fdbd8 1709
ef1669ff
KO		 1710 if (mustfix_fsck_err_on(ret, c, "root directory missing") ||
1711 mustfix_fsck_err_on(!S_ISDIR(root_inode.bi_mode), c,
1712 "root inode not a directory")) {
1713 bch2_inode_init(c, &root_inode, 0, 0, S_IFDIR|0755,
1714 0, NULL);
1715 root_inode.bi_inum = inum;
1c6fdbd8 1716
ef1669ff
KO		 1717 ret = write_inode(&trans, &root_inode, snapshot);
1718 if (ret)
1719 bch_err(c, "error writing root inode: %i", ret);
1720 }
1721err:
1722fsck_err:
1723 bch2_trans_exit(&trans);
1724 return ret;
1c6fdbd8
KO		 1725}
1726
1c6fdbd8
KO		 1727struct pathbuf {
1728 size_t nr;
1729 size_t size;
1730
1731 struct pathbuf_entry {
1732 u64 inum;
6e0c886d 1733 u32 snapshot;
1c6fdbd8
KO		 1734 } *entries;
1735};
1736
6e0c886d
KO		 1737static bool path_is_dup(struct pathbuf *p, u64 inum, u32 snapshot)
1738{
1739 struct pathbuf_entry *i;
1740
1741 for (i = p->entries; i < p->entries + p->nr; i++)
1742 if (i->inum == inum &&
1743 i->snapshot == snapshot)
1744 return true;
1745
1746 return false;
1747}
1748
1749static int path_down(struct pathbuf *p, u64 inum, u32 snapshot)
1c6fdbd8
KO		 1750{
1751 if (p->nr == p->size) {
1752 size_t new_size = max_t(size_t, 256UL, p->size * 2);
1753 void *n = krealloc(p->entries,
1754 new_size * sizeof(p->entries[0]),
1755 GFP_KERNEL);
d3ff7fec 1756 if (!n) {
1c6fdbd8 1757 return -ENOMEM;
d3ff7fec 1758 }
1c6fdbd8
KO		 1759
1760 p->entries = n;
1761 p->size = new_size;
1762 };
1763
1764 p->entries[p->nr++] = (struct pathbuf_entry) {
6e0c886d
KO		 1765 .inum = inum,
1766 .snapshot = snapshot,
1c6fdbd8
KO		 1767 };
1768 return 0;
1769}
1770
6e0c886d
KO		 1771/*
1772 * Check that a given inode is reachable from the root:
1773 *
1774 * XXX: we should also be verifying that inodes are in the right subvolumes
1775 */
d3ff7fec 1776static int check_path(struct btree_trans *trans,
d3ff7fec 1777 struct pathbuf *p,
81ed9ce3
KO		 1778 struct bch_inode_unpacked *inode,
1779 u32 snapshot)
1c6fdbd8 1780{
d3ff7fec 1781 struct bch_fs *c = trans->c;
1c6fdbd8
KO		 1782 int ret = 0;
1783
ef1669ff 1784 snapshot = snapshot_t(c, snapshot)->equiv;
d3ff7fec 1785 p->nr = 0;
424eb881 1786
488f9776
KO		 1787 while (!(inode->bi_inum == BCACHEFS_ROOT_INO &&
1788 inode->bi_subvol == BCACHEFS_ROOT_SUBVOL)) {
6e0c886d
KO		 1789 u32 parent_snapshot = snapshot;
1790
488f9776
KO		 1791 if (inode->bi_parent_subvol) {
1792 u64 inum;
1793
1794 ret = subvol_lookup(trans, inode->bi_parent_subvol,
6e0c886d 1795 &parent_snapshot, &inum);
488f9776
KO		 1796 if (ret)
1797 break;
1798 }
1799
d3ff7fec 1800 ret = lockrestart_do(trans,
6e0c886d 1801 inode_backpointer_exists(trans, inode, parent_snapshot));
d3ff7fec
KO		 1802 if (ret < 0)
1803 break;
1c6fdbd8 1804
d3ff7fec 1805 if (!ret) {
ef1669ff
KO		 1806 if (fsck_err(c, "unreachable inode %llu:%u, type %u nlink %u backptr %llu:%llu",
1807 inode->bi_inum, snapshot,
d3ff7fec
KO		 1808 mode_to_type(inode->bi_mode),
1809 inode->bi_nlink,
1810 inode->bi_dir,
1811 inode->bi_dir_offset))
81ed9ce3 1812 ret = reattach_inode(trans, inode, snapshot);
d3ff7fec
KO		 1813 break;
1814 }
1815 ret = 0;
1c6fdbd8 1816
d3ff7fec
KO		 1817 if (!S_ISDIR(inode->bi_mode))
1818 break;
1c6fdbd8 1819
6e0c886d 1820 ret = path_down(p, inode->bi_inum, snapshot);
d3ff7fec
KO		 1821 if (ret) {
1822 bch_err(c, "memory allocation failure");
1823 return ret;
1824 }
1c6fdbd8 1825
6e0c886d
KO		 1826 snapshot = parent_snapshot;
1827
1828 ret = lookup_inode(trans, inode->bi_dir, inode, &snapshot);
1829 if (ret) {
1830 /* Should have been caught in dirents pass */
1831 bch_err(c, "error looking up parent directory: %i", ret);
1832 break;
1833 }
1834
1835 if (path_is_dup(p, inode->bi_inum, snapshot)) {
1836 struct pathbuf_entry *i;
1c6fdbd8 1837
d3ff7fec 1838 /* XXX print path */
6e0c886d
KO		 1839 bch_err(c, "directory structure loop");
1840
1841 for (i = p->entries; i < p->entries + p->nr; i++)
1842 pr_err("%llu:%u", i->inum, i->snapshot);
1843 pr_err("%llu:%u", inode->bi_inum, snapshot);
1844
d3ff7fec
KO		 1845 if (!fsck_err(c, "directory structure loop"))
1846 return 0;
1c6fdbd8 1847
d3ff7fec 1848 ret = lockrestart_do(trans,
81ed9ce3 1849 remove_backpointer(trans, inode));
1c6fdbd8 1850 if (ret) {
d3ff7fec
KO		 1851 bch_err(c, "error removing dirent: %i", ret);
1852 break;
1c6fdbd8
KO		 1853 }
1854
81ed9ce3 1855 ret = reattach_inode(trans, inode, snapshot);
1c6fdbd8 1856 }
1c6fdbd8 1857 }
d3ff7fec
KO		 1858fsck_err:
1859 if (ret)
1860 bch_err(c, "%s: err %i", __func__, ret);
1861 return ret;
1862}
1c6fdbd8 1863
d3ff7fec
KO		 1864/*
1865 * Check for unreachable inodes, as well as loops in the directory structure:
1866 * After check_dirents(), if an inode backpointer doesn't exist that means it's
1867 * unreachable:
1868 */
58686a25 1869static int check_directory_structure(struct bch_fs *c)
d3ff7fec
KO		 1870{
1871 struct btree_trans trans;
67e0dd8f 1872 struct btree_iter iter;
d3ff7fec
KO		 1873 struct bkey_s_c k;
1874 struct bch_inode_unpacked u;
1875 struct pathbuf path = { 0, 0, NULL };
1876 int ret;
1c6fdbd8 1877
d3ff7fec 1878 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
1c6fdbd8 1879
909004d2
KO		 1880 for_each_btree_key(&trans, iter, BTREE_ID_inodes, POS_MIN,
1881 BTREE_ITER_INTENT|
ef1669ff
KO		 1882 BTREE_ITER_PREFETCH|
1883 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
d3ff7fec 1884 if (k.k->type != KEY_TYPE_inode)
1c6fdbd8
KO		 1885 continue;
1886
d3ff7fec
KO		 1887 ret = bch2_inode_unpack(bkey_s_c_to_inode(k), &u);
1888 if (ret) {
1889 /* Should have been caught earlier in fsck: */
1890 bch_err(c, "error unpacking inode %llu: %i", k.k->p.offset, ret);
1891 break;
1c6fdbd8 1892 }
1c6fdbd8 1893
ef1669ff
KO		 1894 if (u.bi_flags & BCH_INODE_UNLINKED)
1895 continue;
1896
81ed9ce3 1897 ret = check_path(&trans, &path, &u, iter.pos.snapshot);
d3ff7fec
KO		 1898 if (ret)
1899 break;
1c6fdbd8 1900 }
67e0dd8f 1901 bch2_trans_iter_exit(&trans, &iter);
d3ff7fec
KO		 1902
1903 BUG_ON(ret == -EINTR);
1904
f24fab9c
KO		 1905 kfree(path.entries);
1906
9a796fdb
KO		 1907 bch2_trans_exit(&trans);
1908 return ret;
1c6fdbd8
KO		 1909}
1910
fc51b041
KO		 1911struct nlink_table {
1912 size_t nr;
1913 size_t size;
1c6fdbd8 1914
fc51b041
KO		 1915 struct nlink {
1916 u64 inum;
1917 u32 snapshot;
1918 u32 count;
1919 } *d;
1920};
1c6fdbd8 1921
fc51b041 1922static int add_nlink(struct nlink_table *t, u64 inum, u32 snapshot)
1c6fdbd8 1923{
fc51b041
KO		 1924 if (t->nr == t->size) {
1925 size_t new_size = max_t(size_t, 128UL, t->size * 2);
1926 void *d = kvmalloc(new_size * sizeof(t->d[0]), GFP_KERNEL);
1927 if (!d) {
1928 return -ENOMEM;
1929 }
1c6fdbd8 1930
82355e28
KO		 1931 if (t->d)
1932 memcpy(d, t->d, t->size * sizeof(t->d[0]));
fc51b041 1933 kvfree(t->d);
1c6fdbd8 1934
fc51b041
KO		 1935 t->d = d;
1936 t->size = new_size;
2bb748a6
KO		 1937 }
1938
fc51b041
KO		 1939
1940 t->d[t->nr++] = (struct nlink) {
1941 .inum = inum,
1942 .snapshot = snapshot,
1943 };
1944
1945 return 0;
1946}
1947
1948static int nlink_cmp(const void *_l, const void *_r)
1949{
1950 const struct nlink *l = _l;
1951 const struct nlink *r = _r;
1952
1953 return cmp_int(l->inum, r->inum) ?: cmp_int(l->snapshot, r->snapshot);
1954}
1955
ef1669ff
KO		 1956static void inc_link(struct bch_fs *c, struct snapshots_seen *s,
1957 struct nlink_table *links,
1958 u64 range_start, u64 range_end, u64 inum, u32 snapshot)
fc51b041
KO		 1959{
1960 struct nlink *link, key = {
1961 .inum = inum, .snapshot = U32_MAX,
1962 };
1963
1964 if (inum < range_start || inum >= range_end)
1c6fdbd8 1965 return;
fc51b041
KO		 1966
1967 link = __inline_bsearch(&key, links->d, links->nr,
1968 sizeof(links->d[0]), nlink_cmp);
ef1669ff
KO		 1969 if (!link)
1970 return;
1971
1972 while (link > links->d && link[0].inum == link[-1].inum)
1973 --link;
1974
1975 for (; link < links->d + links->nr && link->inum == inum; link++)
1976 if (ref_visible(c, s, snapshot, link->snapshot)) {
1977 link->count++;
1978 if (link->snapshot >= snapshot)
1979 break;
1980 }
fc51b041
KO		 1981}
1982
1983noinline_for_stack
1984static int check_nlinks_find_hardlinks(struct bch_fs *c,
1985 struct nlink_table *t,
1986 u64 start, u64 *end)
1987{
1988 struct btree_trans trans;
67e0dd8f 1989 struct btree_iter iter;
fc51b041
KO		 1990 struct bkey_s_c k;
1991 struct bkey_s_c_inode inode;
1992 struct bch_inode_unpacked u;
1993 int ret = 0;
1994
1995 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
1996
1997 for_each_btree_key(&trans, iter, BTREE_ID_inodes,
909004d2
KO		 1998 POS(0, start),
1999 BTREE_ITER_INTENT|
ef1669ff
KO		 2000 BTREE_ITER_PREFETCH|
2001 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
fc51b041
KO		 2002 if (k.k->type != KEY_TYPE_inode)
2003 continue;
2004
2005 inode = bkey_s_c_to_inode(k);
2006
2007 /*
2008 * Backpointer and directory structure checks are sufficient for
2009 * directories, since they can't have hardlinks:
2010 */
2011 if (S_ISDIR(le16_to_cpu(inode.v->bi_mode)))
2012 continue;
2013
2014 /* Should never fail, checked by bch2_inode_invalid: */
2015 BUG_ON(bch2_inode_unpack(inode, &u));
2016
2017 if (!u.bi_nlink)
2018 continue;
2019
2020 ret = add_nlink(t, k.k->p.offset, k.k->p.snapshot);
2021 if (ret) {
2022 *end = k.k->p.offset;
2023 ret = 0;
2024 break;
2025 }
2026
1c6fdbd8 2027 }
67e0dd8f 2028 bch2_trans_iter_exit(&trans, &iter);
fc51b041
KO		 2029 bch2_trans_exit(&trans);
2030
2031 if (ret)
2032 bch_err(c, "error in fsck: btree error %i while walking inodes", ret);
1c6fdbd8 2033
fc51b041 2034 return ret;
1c6fdbd8
KO		 2035}
2036
2037noinline_for_stack
fc51b041
KO		 2038static int check_nlinks_walk_dirents(struct bch_fs *c, struct nlink_table *links,
2039 u64 range_start, u64 range_end)
1c6fdbd8 2040{
424eb881 2041 struct btree_trans trans;
ef1669ff 2042 struct snapshots_seen s;
67e0dd8f 2043 struct btree_iter iter;
1c6fdbd8
KO		 2044 struct bkey_s_c k;
2045 struct bkey_s_c_dirent d;
1c6fdbd8
KO		 2046 int ret;
2047
ef1669ff
KO		 2048 snapshots_seen_init(&s);
2049
20bceecb 2050 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
424eb881 2051
909004d2
KO		 2052 for_each_btree_key(&trans, iter, BTREE_ID_dirents, POS_MIN,
2053 BTREE_ITER_INTENT|
ef1669ff
KO		 2054 BTREE_ITER_PREFETCH|
2055 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
2056 ret = snapshots_seen_update(c, &s, k.k->p);
2057 if (ret)
2058 break;
2059
1c6fdbd8 2060 switch (k.k->type) {
26609b61 2061 case KEY_TYPE_dirent:
1c6fdbd8 2062 d = bkey_s_c_to_dirent(k);
1c6fdbd8 2063
ef1669ff
KO		 2064 if (d.v->d_type != DT_DIR &&
2065 d.v->d_type != DT_SUBVOL)
2066 inc_link(c, &s, links, range_start, range_end,
2067 le64_to_cpu(d.v->d_inum),
2068 d.k->p.snapshot);
1c6fdbd8
KO		 2069 break;
2070 }
2071
424eb881 2072 bch2_trans_cond_resched(&trans);
1c6fdbd8 2073 }
67e0dd8f 2074 bch2_trans_iter_exit(&trans, &iter);
abcecb49 2075
1c6fdbd8 2076 if (ret)
619f5bee 2077 bch_err(c, "error in fsck: btree error %i while walking dirents", ret);
1c6fdbd8 2078
ef1669ff
KO		 2079 bch2_trans_exit(&trans);
2080 snapshots_seen_exit(&s);
1c6fdbd8
KO		 2081 return ret;
2082}
2083
1c6fdbd8 2084noinline_for_stack
fc51b041
KO		 2085static int check_nlinks_update_hardlinks(struct bch_fs *c,
2086 struct nlink_table *links,
1c6fdbd8
KO		 2087 u64 range_start, u64 range_end)
2088{
0564b167 2089 struct btree_trans trans;
67e0dd8f 2090 struct btree_iter iter;
1c6fdbd8 2091 struct bkey_s_c k;
fc51b041
KO		 2092 struct bkey_s_c_inode inode;
2093 struct bch_inode_unpacked u;
2094 struct nlink *link = links->d;
b906aadd 2095 int ret = 0;
1c6fdbd8 2096
20bceecb 2097 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
0564b167 2098
b906aadd 2099 for_each_btree_key(&trans, iter, BTREE_ID_inodes,
909004d2
KO		 2100 POS(0, range_start),
2101 BTREE_ITER_INTENT|
ef1669ff
KO		 2102 BTREE_ITER_PREFETCH|
2103 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
fc51b041 2104 if (k.k->p.offset >= range_end)
1c6fdbd8
KO		 2105 break;
2106
b906aadd
KO		 2107 if (k.k->type != KEY_TYPE_inode)
2108 continue;
1c6fdbd8 2109
fc51b041
KO		 2110 inode = bkey_s_c_to_inode(k);
2111 if (S_ISDIR(le16_to_cpu(inode.v->bi_mode)))
2112 continue;
2113
2114 BUG_ON(bch2_inode_unpack(inode, &u));
1c6fdbd8 2115
fc51b041
KO		 2116 if (!u.bi_nlink)
2117 continue;
2118
ef1669ff
KO		 2119 while ((cmp_int(link->inum, k.k->p.offset) ?:
2120 cmp_int(link->snapshot, k.k->p.snapshot)) < 0) {
fc51b041
KO		 2121 link++;
2122 BUG_ON(link >= links->d + links->nr);
2123 }
2124
2125 if (fsck_err_on(bch2_inode_nlink_get(&u) != link->count, c,
2126 "inode %llu has wrong i_nlink (type %u i_nlink %u, should be %u)",
2127 u.bi_inum, mode_to_type(u.bi_mode),
2128 bch2_inode_nlink_get(&u), link->count)) {
2129 bch2_inode_nlink_set(&u, link->count);
2130
ea0531f8 2131 ret = write_inode(&trans, &u, k.k->p.snapshot);
fc51b041
KO		 2132 if (ret)
2133 bch_err(c, "error in fsck: error %i updating inode", ret);
2134 }
1c6fdbd8 2135 }
fc51b041 2136fsck_err:
67e0dd8f 2137 bch2_trans_iter_exit(&trans, &iter);
0564b167
KO		 2138 bch2_trans_exit(&trans);
2139
b906aadd
KO		 2140 if (ret)
2141 bch_err(c, "error in fsck: btree error %i while walking inodes", ret);
1c6fdbd8 2142
b906aadd 2143 return ret;
1c6fdbd8
KO		 2144}
2145
2146noinline_for_stack
58686a25 2147static int check_nlinks(struct bch_fs *c)
1c6fdbd8 2148{
fc51b041 2149 struct nlink_table links = { 0 };
1c6fdbd8
KO		 2150 u64 this_iter_range_start, next_iter_range_start = 0;
2151 int ret = 0;
2152
2153 bch_verbose(c, "checking inode nlinks");
2154
1c6fdbd8
KO		 2155 do {
2156 this_iter_range_start = next_iter_range_start;
2157 next_iter_range_start = U64_MAX;
2158
fc51b041
KO		 2159 ret = check_nlinks_find_hardlinks(c, &links,
2160 this_iter_range_start,
2161 &next_iter_range_start);
2162
2163 ret = check_nlinks_walk_dirents(c, &links,
1c6fdbd8 2164 this_iter_range_start,
fc51b041 2165 next_iter_range_start);
1c6fdbd8
KO		 2166 if (ret)
2167 break;
2168
fc51b041 2169 ret = check_nlinks_update_hardlinks(c, &links,
1c6fdbd8
KO		 2170 this_iter_range_start,
2171 next_iter_range_start);
2172 if (ret)
2173 break;
2174
fc51b041 2175 links.nr = 0;
1c6fdbd8
KO		 2176 } while (next_iter_range_start != U64_MAX);
2177
fc51b041 2178 kvfree(links.d);
1c6fdbd8
KO		 2179
2180 return ret;
2181}
2182
bfe88863
KO		 2183static int fix_reflink_p_key(struct btree_trans *trans, struct btree_iter *iter)
2184{
2185 struct bkey_s_c k;
2186 struct bkey_s_c_reflink_p p;
2187 struct bkey_i_reflink_p *u;
2188 int ret;
2189
2190 k = bch2_btree_iter_peek(iter);
2191 if (!k.k)
2192 return 0;
2193
2194 ret = bkey_err(k);
2195 if (ret)
2196 return ret;
2197
2198 if (k.k->type != KEY_TYPE_reflink_p)
2199 return 0;
2200
2201 p = bkey_s_c_to_reflink_p(k);
2202
6d76aefe 2203 if (!p.v->front_pad && !p.v->back_pad)
bfe88863
KO		 2204 return 0;
2205
2206 u = bch2_trans_kmalloc(trans, sizeof(*u));
2207 ret = PTR_ERR_OR_ZERO(u);
2208 if (ret)
2209 return ret;
2210
2211 bkey_reassemble(&u->k_i, k);
6d76aefe
KO		 2212 u->v.front_pad = 0;
2213 u->v.back_pad = 0;
bfe88863
KO		 2214
2215 return bch2_trans_update(trans, iter, &u->k_i, 0);
2216}
2217
2218static int fix_reflink_p(struct bch_fs *c)
2219{
2220 struct btree_trans trans;
2221 struct btree_iter iter;
2222 struct bkey_s_c k;
2223 int ret;
2224
2225 if (c->sb.version >= bcachefs_metadata_version_reflink_p_fix)
2226 return 0;
2227
2228 bch2_trans_init(&trans, c, BTREE_ITER_MAX, 0);
2229
2230 for_each_btree_key(&trans, iter, BTREE_ID_extents, POS_MIN,
2231 BTREE_ITER_INTENT|
2232 BTREE_ITER_PREFETCH|
2233 BTREE_ITER_ALL_SNAPSHOTS, k, ret) {
2234 if (k.k->type == KEY_TYPE_reflink_p) {
2235 ret = __bch2_trans_do(&trans, NULL, NULL,
2236 BTREE_INSERT_NOFAIL|
2237 BTREE_INSERT_LAZY_RW,
2238 fix_reflink_p_key(&trans, &iter));
2239 if (ret)
2240 break;
2241 }
2242 }
2243 bch2_trans_iter_exit(&trans, &iter);
2244
2245 bch2_trans_exit(&trans);
2246 return ret;
2247}
2248
1c6fdbd8
KO		 2249/*
2250 * Checks for inconsistencies that shouldn't happen, unless we have a bug.
2251 * Doesn't fix them yet, mainly because they haven't yet been observed:
2252 */
619f5bee 2253int bch2_fsck_full(struct bch_fs *c)
1c6fdbd8 2254{
14b393ee
KO		 2255 return bch2_fs_snapshots_check(c) ?:
2256 check_inodes(c, true) ?:
ef1669ff 2257 check_subvols(c) ?:
5c16add5 2258 check_extents(c) ?:
1c6fdbd8
KO		 2259 check_dirents(c) ?:
2260 check_xattrs(c) ?:
ef1669ff 2261 check_root(c) ?:
58686a25 2262 check_directory_structure(c) ?:
bfe88863
KO		 2263 check_nlinks(c) ?:
2264 fix_reflink_p(c);
1c6fdbd8
KO		 2265}
2266
619f5bee 2267int bch2_fsck_walk_inodes_only(struct bch_fs *c)
1c6fdbd8 2268{
5c16add5 2269 return check_inodes(c, false);
1c6fdbd8 2270}
Linux 6.x block layer and io_uring tree(s)