summaryrefslogtreecommitdiff
path: root/fs/bcachefs/subvolume.c
blob: 6407d19edc0e5cc87be979376a906858d0137ae3 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
// SPDX-License-Identifier: GPL-2.0

#include "bcachefs.h"
#include "btree_key_cache.h"
#include "btree_update.h"
#include "errcode.h"
#include "error.h"
#include "fs.h"
#include "subvolume.h"

/* Snapshot tree: */

void bch2_snapshot_to_text(struct printbuf *out, struct bch_fs *c,
			   struct bkey_s_c k)
{
	struct bkey_s_c_snapshot s = bkey_s_c_to_snapshot(k);

	prt_printf(out, "is_subvol %llu deleted %llu parent %10u children %10u %10u subvol %u",
	       BCH_SNAPSHOT_SUBVOL(s.v),
	       BCH_SNAPSHOT_DELETED(s.v),
	       le32_to_cpu(s.v->parent),
	       le32_to_cpu(s.v->children[0]),
	       le32_to_cpu(s.v->children[1]),
	       le32_to_cpu(s.v->subvol));
}

int bch2_snapshot_invalid(const struct bch_fs *c, struct bkey_s_c k,
			  unsigned flags, struct printbuf *err)
{
	struct bkey_s_c_snapshot s;
	u32 i, id;

	if (bkey_gt(k.k->p, POS(0, U32_MAX)) ||
	    bkey_lt(k.k->p, POS(0, 1))) {
		prt_printf(err, "bad pos");
		return -BCH_ERR_invalid_bkey;
	}

	if (bkey_val_bytes(k.k) != sizeof(struct bch_snapshot)) {
		prt_printf(err, "bad val size (%zu != %zu)",
		       bkey_val_bytes(k.k), sizeof(struct bch_snapshot));
		return -BCH_ERR_invalid_bkey;
	}

	s = bkey_s_c_to_snapshot(k);

	id = le32_to_cpu(s.v->parent);
	if (id && id <= k.k->p.offset) {
		prt_printf(err, "bad parent node (%u <= %llu)",
		       id, k.k->p.offset);
		return -BCH_ERR_invalid_bkey;
	}

	if (le32_to_cpu(s.v->children[0]) < le32_to_cpu(s.v->children[1])) {
		prt_printf(err, "children not normalized");
		return -BCH_ERR_invalid_bkey;
	}

	if (s.v->children[0] &&
	    s.v->children[0] == s.v->children[1]) {
		prt_printf(err, "duplicate child nodes");
		return -BCH_ERR_invalid_bkey;
	}

	for (i = 0; i < 2; i++) {
		id = le32_to_cpu(s.v->children[i]);

		if (id >= k.k->p.offset) {
			prt_printf(err, "bad child node (%u >= %llu)",
			       id, k.k->p.offset);
			return -BCH_ERR_invalid_bkey;
		}
	}

	return 0;
}

int bch2_mark_snapshot(struct btree_trans *trans,
		       enum btree_id btree, unsigned level,
		       struct bkey_s_c old, struct bkey_s_c new,
		       unsigned flags)
{
	struct bch_fs *c = trans->c;
	struct snapshot_t *t;

	t = genradix_ptr_alloc(&c->snapshots,
			       U32_MAX - new.k->p.offset,
			       GFP_KERNEL);
	if (!t)
		return -BCH_ERR_ENOMEM_mark_snapshot;

	if (new.k->type == KEY_TYPE_snapshot) {
		struct bkey_s_c_snapshot s = bkey_s_c_to_snapshot(new);

		t->parent	= le32_to_cpu(s.v->parent);
		t->children[0]	= le32_to_cpu(s.v->children[0]);
		t->children[1]	= le32_to_cpu(s.v->children[1]);
		t->subvol	= BCH_SNAPSHOT_SUBVOL(s.v) ? le32_to_cpu(s.v->subvol) : 0;
	} else {
		t->parent	= 0;
		t->children[0]	= 0;
		t->children[1]	= 0;
		t->subvol	= 0;
	}

	return 0;
}

static int snapshot_lookup(struct btree_trans *trans, u32 id,
			   struct bch_snapshot *s)
{
	struct btree_iter iter;
	struct bkey_s_c k;
	int ret;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots, POS(0, id),
			     BTREE_ITER_WITH_UPDATES);
	k = bch2_btree_iter_peek_slot(&iter);
	ret = bkey_err(k) ?: k.k->type == KEY_TYPE_snapshot ? 0 : -ENOENT;

	if (!ret)
		*s = *bkey_s_c_to_snapshot(k).v;

	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

static int snapshot_live(struct btree_trans *trans, u32 id)
{
	struct bch_snapshot v;
	int ret;

	if (!id)
		return 0;

	ret = snapshot_lookup(trans, id, &v);
	if (ret == -ENOENT)
		bch_err(trans->c, "snapshot node %u not found", id);
	if (ret)
		return ret;

	return !BCH_SNAPSHOT_DELETED(&v);
}

static int bch2_snapshot_set_equiv(struct btree_trans *trans, struct bkey_s_c k)
{
	struct bch_fs *c = trans->c;
	unsigned i, nr_live = 0, live_idx = 0;
	struct bkey_s_c_snapshot snap;
	u32 id = k.k->p.offset, child[2];

	if (k.k->type != KEY_TYPE_snapshot)
		return 0;

	snap = bkey_s_c_to_snapshot(k);

	child[0] = le32_to_cpu(snap.v->children[0]);
	child[1] = le32_to_cpu(snap.v->children[1]);

	for (i = 0; i < 2; i++) {
		int ret = snapshot_live(trans, child[i]);

		if (ret < 0)
			return ret;

		if (ret)
			live_idx = i;
		nr_live += ret;
	}

	snapshot_t(c, id)->equiv = nr_live == 1
		? snapshot_t(c, child[live_idx])->equiv
		: id;
	return 0;
}

/* fsck: */
static int check_snapshot(struct btree_trans *trans,
			  struct btree_iter *iter,
			  struct bkey_s_c k)
{
	struct bch_fs *c = trans->c;
	struct bkey_s_c_snapshot s;
	struct bch_subvolume subvol;
	struct bch_snapshot v;
	struct printbuf buf = PRINTBUF;
	bool should_have_subvol;
	u32 i, id;
	int ret = 0;

	if (k.k->type != KEY_TYPE_snapshot)
		return 0;

	s = bkey_s_c_to_snapshot(k);
	id = le32_to_cpu(s.v->parent);
	if (id) {
		ret = snapshot_lookup(trans, id, &v);
		if (ret == -ENOENT)
			bch_err(c, "snapshot with nonexistent parent:\n  %s",
				(bch2_bkey_val_to_text(&buf, c, s.s_c), buf.buf));
		if (ret)
			goto err;

		if (le32_to_cpu(v.children[0]) != s.k->p.offset &&
		    le32_to_cpu(v.children[1]) != s.k->p.offset) {
			bch_err(c, "snapshot parent %u missing pointer to child %llu",
				id, s.k->p.offset);
			ret = -EINVAL;
			goto err;
		}
	}

	for (i = 0; i < 2 && s.v->children[i]; i++) {
		id = le32_to_cpu(s.v->children[i]);

		ret = snapshot_lookup(trans, id, &v);
		if (ret == -ENOENT)
			bch_err(c, "snapshot node %llu has nonexistent child %u",
				s.k->p.offset, id);
		if (ret)
			goto err;

		if (le32_to_cpu(v.parent) != s.k->p.offset) {
			bch_err(c, "snapshot child %u has wrong parent (got %u should be %llu)",
				id, le32_to_cpu(v.parent), s.k->p.offset);
			ret = -EINVAL;
			goto err;
		}
	}

	should_have_subvol = BCH_SNAPSHOT_SUBVOL(s.v) &&
		!BCH_SNAPSHOT_DELETED(s.v);

	if (should_have_subvol) {
		id = le32_to_cpu(s.v->subvol);
		ret = bch2_subvolume_get(trans, id, 0, false, &subvol);
		if (ret == -ENOENT)
			bch_err(c, "snapshot points to nonexistent subvolume:\n  %s",
				(bch2_bkey_val_to_text(&buf, c, s.s_c), buf.buf));
		if (ret)
			goto err;

		if (BCH_SNAPSHOT_SUBVOL(s.v) != (le32_to_cpu(subvol.snapshot) == s.k->p.offset)) {
			bch_err(c, "snapshot node %llu has wrong BCH_SNAPSHOT_SUBVOL",
				s.k->p.offset);
			ret = -EINVAL;
			goto err;
		}
	} else {
		if (fsck_err_on(s.v->subvol, c, "snapshot should not point to subvol:\n  %s",
				(bch2_bkey_val_to_text(&buf, c, s.s_c), buf.buf))) {
			struct bkey_i_snapshot *u = bch2_trans_kmalloc(trans, sizeof(*u));

			ret = PTR_ERR_OR_ZERO(u);
			if (ret)
				goto err;

			bkey_reassemble(&u->k_i, s.s_c);
			u->v.subvol = 0;
			ret = bch2_trans_update(trans, iter, &u->k_i, 0);
			if (ret)
				goto err;
		}
	}

	if (BCH_SNAPSHOT_DELETED(s.v))
		set_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags);
err:
fsck_err:
	printbuf_exit(&buf);
	return ret;
}

int bch2_fs_check_snapshots(struct bch_fs *c)
{
	struct btree_trans trans;
	struct btree_iter iter;
	struct bkey_s_c k;
	int ret;

	bch2_trans_init(&trans, c, 0, 0);

	ret = for_each_btree_key_commit(&trans, iter,
			BTREE_ID_snapshots, POS_MIN,
			BTREE_ITER_PREFETCH, k,
			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
		check_snapshot(&trans, &iter, k));

	if (ret)
		bch_err(c, "error %i checking snapshots", ret);

	bch2_trans_exit(&trans);
	return ret;
}

static int check_subvol(struct btree_trans *trans,
			struct btree_iter *iter,
			struct bkey_s_c k)
{
	struct bkey_s_c_subvolume subvol;
	struct bch_snapshot snapshot;
	unsigned snapid;
	int ret;

	if (k.k->type != KEY_TYPE_subvolume)
		return 0;

	subvol = bkey_s_c_to_subvolume(k);
	snapid = le32_to_cpu(subvol.v->snapshot);
	ret = snapshot_lookup(trans, snapid, &snapshot);

	if (ret == -ENOENT)
		bch_err(trans->c, "subvolume %llu points to nonexistent snapshot %u",
			k.k->p.offset, snapid);
	if (ret)
		return ret;

	if (BCH_SUBVOLUME_UNLINKED(subvol.v)) {
		ret = bch2_subvolume_delete(trans, iter->pos.offset);
		if (ret && !bch2_err_matches(ret, BCH_ERR_transaction_restart))
			bch_err(trans->c, "error deleting subvolume %llu: %s",
				iter->pos.offset, bch2_err_str(ret));
		if (ret)
			return ret;
	}

	return 0;
}

int bch2_fs_check_subvols(struct bch_fs *c)
{
	struct btree_trans trans;
	struct btree_iter iter;
	struct bkey_s_c k;
	int ret;

	bch2_trans_init(&trans, c, 0, 0);

	ret = for_each_btree_key_commit(&trans, iter,
			BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k,
			NULL, NULL, BTREE_INSERT_LAZY_RW|BTREE_INSERT_NOFAIL,
		check_subvol(&trans, &iter, k));

	bch2_trans_exit(&trans);

	return ret;
}

void bch2_fs_snapshots_exit(struct bch_fs *c)
{
	genradix_free(&c->snapshots);
}

int bch2_fs_snapshots_start(struct bch_fs *c)
{
	struct btree_trans trans;
	struct btree_iter iter;
	struct bkey_s_c k;
	int ret = 0;

	bch2_trans_init(&trans, c, 0, 0);

	for_each_btree_key2(&trans, iter, BTREE_ID_snapshots,
			   POS_MIN, 0, k,
		bch2_mark_snapshot(&trans, BTREE_ID_snapshots, 0, bkey_s_c_null, k, 0) ?:
		bch2_snapshot_set_equiv(&trans, k));

	bch2_trans_exit(&trans);

	if (ret)
		bch_err(c, "error starting snapshots: %s", bch2_err_str(ret));
	return ret;
}

/*
 * Mark a snapshot as deleted, for future cleanup:
 */
static int bch2_snapshot_node_set_deleted(struct btree_trans *trans, u32 id)
{
	struct btree_iter iter;
	struct bkey_i_snapshot *s;
	int ret = 0;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots, POS(0, id),
			     BTREE_ITER_INTENT);
	s = bch2_bkey_get_mut_typed(trans, &iter, snapshot);
	ret = PTR_ERR_OR_ZERO(s);
	if (unlikely(ret)) {
		bch2_fs_inconsistent_on(ret == -ENOENT, trans->c, "missing snapshot %u", id);
		goto err;
	}

	/* already deleted? */
	if (BCH_SNAPSHOT_DELETED(&s->v))
		goto err;

	SET_BCH_SNAPSHOT_DELETED(&s->v, true);
	SET_BCH_SNAPSHOT_SUBVOL(&s->v, false);
	s->v.subvol = 0;

	ret = bch2_trans_update(trans, &iter, &s->k_i, 0);
	if (ret)
		goto err;
err:
	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

static int bch2_snapshot_node_delete(struct btree_trans *trans, u32 id)
{
	struct btree_iter iter, p_iter = (struct btree_iter) { NULL };
	struct bkey_s_c k;
	struct bkey_s_c_snapshot s;
	u32 parent_id;
	unsigned i;
	int ret = 0;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots, POS(0, id),
			     BTREE_ITER_INTENT);
	k = bch2_btree_iter_peek_slot(&iter);
	ret = bkey_err(k);
	if (ret)
		goto err;

	if (k.k->type != KEY_TYPE_snapshot) {
		bch2_fs_inconsistent(trans->c, "missing snapshot %u", id);
		ret = -ENOENT;
		goto err;
	}

	s = bkey_s_c_to_snapshot(k);

	BUG_ON(!BCH_SNAPSHOT_DELETED(s.v));
	parent_id = le32_to_cpu(s.v->parent);

	if (parent_id) {
		struct bkey_i_snapshot *parent;

		bch2_trans_iter_init(trans, &p_iter, BTREE_ID_snapshots,
				     POS(0, parent_id),
				     BTREE_ITER_INTENT);
		parent = bch2_bkey_get_mut_typed(trans, &p_iter, snapshot);
		ret = PTR_ERR_OR_ZERO(parent);
		if (unlikely(ret)) {
			bch2_fs_inconsistent_on(ret == -ENOENT, trans->c, "missing snapshot %u", parent_id);
			goto err;
		}

		for (i = 0; i < 2; i++)
			if (le32_to_cpu(parent->v.children[i]) == id)
				break;

		if (i == 2)
			bch_err(trans->c, "snapshot %u missing child pointer to %u",
				parent_id, id);
		else
			parent->v.children[i] = 0;

		if (le32_to_cpu(parent->v.children[0]) <
		    le32_to_cpu(parent->v.children[1]))
			swap(parent->v.children[0],
			     parent->v.children[1]);

		ret = bch2_trans_update(trans, &p_iter, &parent->k_i, 0);
		if (ret)
			goto err;
	}

	ret = bch2_btree_delete_at(trans, &iter, 0);
err:
	bch2_trans_iter_exit(trans, &p_iter);
	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

int bch2_snapshot_node_create(struct btree_trans *trans, u32 parent,
			      u32 *new_snapids,
			      u32 *snapshot_subvols,
			      unsigned nr_snapids)
{
	struct btree_iter iter;
	struct bkey_i_snapshot *n;
	struct bkey_s_c k;
	unsigned i;
	int ret = 0;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots,
			     POS_MIN, BTREE_ITER_INTENT);
	k = bch2_btree_iter_peek(&iter);
	ret = bkey_err(k);
	if (ret)
		goto err;

	for (i = 0; i < nr_snapids; i++) {
		k = bch2_btree_iter_prev_slot(&iter);
		ret = bkey_err(k);
		if (ret)
			goto err;

		if (!k.k || !k.k->p.offset) {
			ret = -BCH_ERR_ENOSPC_snapshot_create;
			goto err;
		}

		n = bch2_bkey_alloc(trans, &iter, snapshot);
		ret = PTR_ERR_OR_ZERO(n);
		if (ret)
			goto err;

		n->v.flags	= 0;
		n->v.parent	= cpu_to_le32(parent);
		n->v.subvol	= cpu_to_le32(snapshot_subvols[i]);
		n->v.pad	= 0;
		SET_BCH_SNAPSHOT_SUBVOL(&n->v, true);

		ret   = bch2_trans_update(trans, &iter, &n->k_i, 0) ?:
			bch2_mark_snapshot(trans, BTREE_ID_snapshots, 0,
					   bkey_s_c_null, bkey_i_to_s_c(&n->k_i), 0);
		if (ret)
			goto err;

		new_snapids[i]	= iter.pos.offset;
	}

	if (parent) {
		bch2_btree_iter_set_pos(&iter, POS(0, parent));
		n = bch2_bkey_get_mut_typed(trans, &iter, snapshot);
		ret = PTR_ERR_OR_ZERO(n);
		if (unlikely(ret)) {
			if (ret == -ENOENT)
				bch_err(trans->c, "snapshot %u not found", parent);
			goto err;
		}

		if (n->v.children[0] || n->v.children[1]) {
			bch_err(trans->c, "Trying to add child snapshot nodes to parent that already has children");
			ret = -EINVAL;
			goto err;
		}

		n->v.children[0] = cpu_to_le32(new_snapids[0]);
		n->v.children[1] = cpu_to_le32(new_snapids[1]);
		n->v.subvol = 0;
		SET_BCH_SNAPSHOT_SUBVOL(&n->v, false);
		ret   = bch2_trans_update(trans, &iter, &n->k_i, 0);
		if (ret)
			goto err;
	}
err:
	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

static int snapshot_delete_key(struct btree_trans *trans,
			       struct btree_iter *iter,
			       struct bkey_s_c k,
			       snapshot_id_list *deleted,
			       snapshot_id_list *equiv_seen,
			       struct bpos *last_pos)
{
	struct bch_fs *c = trans->c;
	u32 equiv = snapshot_t(c, k.k->p.snapshot)->equiv;

	if (!bkey_eq(k.k->p, *last_pos))
		equiv_seen->nr = 0;
	*last_pos = k.k->p;

	if (snapshot_list_has_id(deleted, k.k->p.snapshot) ||
	    snapshot_list_has_id(equiv_seen, equiv)) {
		return bch2_btree_delete_at(trans, iter,
					    BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
	} else {
		return snapshot_list_add(c, equiv_seen, equiv);
	}
}

static int bch2_delete_redundant_snapshot(struct btree_trans *trans, struct btree_iter *iter,
					  struct bkey_s_c k)
{
	struct bkey_s_c_snapshot snap;
	u32 children[2];
	int ret;

	if (k.k->type != KEY_TYPE_snapshot)
		return 0;

	snap = bkey_s_c_to_snapshot(k);
	if (BCH_SNAPSHOT_DELETED(snap.v) ||
	    BCH_SNAPSHOT_SUBVOL(snap.v))
		return 0;

	children[0] = le32_to_cpu(snap.v->children[0]);
	children[1] = le32_to_cpu(snap.v->children[1]);

	ret   = snapshot_live(trans, children[0]) ?:
		snapshot_live(trans, children[1]);
	if (ret < 0)
		return ret;

	if (!ret)
		return bch2_snapshot_node_set_deleted(trans, k.k->p.offset);
	return 0;
}

int bch2_delete_dead_snapshots(struct bch_fs *c)
{
	struct btree_trans trans;
	struct btree_iter iter;
	struct bkey_s_c k;
	struct bkey_s_c_snapshot snap;
	snapshot_id_list deleted = { 0 };
	u32 i, id;
	int ret = 0;

	if (!test_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags))
		return 0;

	if (!test_bit(BCH_FS_STARTED, &c->flags)) {
		ret = bch2_fs_read_write_early(c);
		if (ret) {
			bch_err(c, "error deleleting dead snapshots: error going rw: %s", bch2_err_str(ret));
			return ret;
		}
	}

	bch2_trans_init(&trans, c, 0, 0);

	/*
	 * For every snapshot node: If we have no live children and it's not
	 * pointed to by a subvolume, delete it:
	 */
	ret = for_each_btree_key_commit(&trans, iter, BTREE_ID_snapshots,
			POS_MIN, 0, k,
			NULL, NULL, 0,
		bch2_delete_redundant_snapshot(&trans, &iter, k));
	if (ret) {
		bch_err(c, "error deleting redundant snapshots: %s", bch2_err_str(ret));
		goto err;
	}

	for_each_btree_key2(&trans, iter, BTREE_ID_snapshots,
			   POS_MIN, 0, k,
		bch2_snapshot_set_equiv(&trans, k));
	if (ret) {
		bch_err(c, "error in bch2_snapshots_set_equiv: %s", bch2_err_str(ret));
		goto err;
	}

	for_each_btree_key(&trans, iter, BTREE_ID_snapshots,
			   POS_MIN, 0, k, ret) {
		if (k.k->type != KEY_TYPE_snapshot)
			continue;

		snap = bkey_s_c_to_snapshot(k);
		if (BCH_SNAPSHOT_DELETED(snap.v)) {
			ret = snapshot_list_add(c, &deleted, k.k->p.offset);
			if (ret)
				break;
		}
	}
	bch2_trans_iter_exit(&trans, &iter);

	if (ret) {
		bch_err(c, "error walking snapshots: %s", bch2_err_str(ret));
		goto err;
	}

	for (id = 0; id < BTREE_ID_NR; id++) {
		struct bpos last_pos = POS_MIN;
		snapshot_id_list equiv_seen = { 0 };

		if (!btree_type_has_snapshots(id))
			continue;

		ret = for_each_btree_key_commit(&trans, iter,
				id, POS_MIN,
				BTREE_ITER_PREFETCH|BTREE_ITER_ALL_SNAPSHOTS, k,
				NULL, NULL, BTREE_INSERT_NOFAIL,
			snapshot_delete_key(&trans, &iter, k, &deleted, &equiv_seen, &last_pos));

		darray_exit(&equiv_seen);

		if (ret) {
			bch_err(c, "error deleting snapshot keys: %s", bch2_err_str(ret));
			goto err;
		}
	}

	for (i = 0; i < deleted.nr; i++) {
		ret = commit_do(&trans, NULL, NULL, 0,
			bch2_snapshot_node_delete(&trans, deleted.data[i]));
		if (ret) {
			bch_err(c, "error deleting snapshot %u: %s",
				deleted.data[i], bch2_err_str(ret));
			goto err;
		}
	}

	clear_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags);
err:
	darray_exit(&deleted);
	bch2_trans_exit(&trans);
	return ret;
}

static void bch2_delete_dead_snapshots_work(struct work_struct *work)
{
	struct bch_fs *c = container_of(work, struct bch_fs, snapshot_delete_work);

	bch2_delete_dead_snapshots(c);
	bch2_write_ref_put(c, BCH_WRITE_REF_delete_dead_snapshots);
}

void bch2_delete_dead_snapshots_async(struct bch_fs *c)
{
	if (bch2_write_ref_tryget(c, BCH_WRITE_REF_delete_dead_snapshots) &&
	    !queue_work(c->write_ref_wq, &c->snapshot_delete_work))
		bch2_write_ref_put(c, BCH_WRITE_REF_delete_dead_snapshots);
}

static int bch2_delete_dead_snapshots_hook(struct btree_trans *trans,
					   struct btree_trans_commit_hook *h)
{
	struct bch_fs *c = trans->c;

	set_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags);

	if (!test_bit(BCH_FS_FSCK_DONE, &c->flags))
		return 0;

	bch2_delete_dead_snapshots_async(c);
	return 0;
}

/* Subvolumes: */

int bch2_subvolume_invalid(const struct bch_fs *c, struct bkey_s_c k,
			   unsigned flags, struct printbuf *err)
{
	if (bkey_lt(k.k->p, SUBVOL_POS_MIN) ||
	    bkey_gt(k.k->p, SUBVOL_POS_MAX)) {
		prt_printf(err, "invalid pos");
		return -BCH_ERR_invalid_bkey;
	}

	if (bkey_val_bytes(k.k) != sizeof(struct bch_subvolume)) {
		prt_printf(err, "incorrect value size (%zu != %zu)",
		       bkey_val_bytes(k.k), sizeof(struct bch_subvolume));
		return -BCH_ERR_invalid_bkey;
	}

	return 0;
}

void bch2_subvolume_to_text(struct printbuf *out, struct bch_fs *c,
			    struct bkey_s_c k)
{
	struct bkey_s_c_subvolume s = bkey_s_c_to_subvolume(k);

	prt_printf(out, "root %llu snapshot id %u",
	       le64_to_cpu(s.v->inode),
	       le32_to_cpu(s.v->snapshot));
}

static __always_inline int
bch2_subvolume_get_inlined(struct btree_trans *trans, unsigned subvol,
			   bool inconsistent_if_not_found,
			   int iter_flags,
			   struct bch_subvolume *s)
{
	struct btree_iter iter;
	struct bkey_s_c k;
	int ret;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_subvolumes, POS(0, subvol),
			     iter_flags);
	k = bch2_btree_iter_peek_slot(&iter);
	ret = bkey_err(k) ?: k.k->type == KEY_TYPE_subvolume ? 0 : -ENOENT;

	if (ret == -ENOENT && inconsistent_if_not_found)
		bch2_fs_inconsistent(trans->c, "missing subvolume %u", subvol);
	if (!ret)
		*s = *bkey_s_c_to_subvolume(k).v;

	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

int bch2_subvolume_get(struct btree_trans *trans, unsigned subvol,
		       bool inconsistent_if_not_found,
		       int iter_flags,
		       struct bch_subvolume *s)
{
	return bch2_subvolume_get_inlined(trans, subvol, inconsistent_if_not_found, iter_flags, s);
}

int bch2_snapshot_get_subvol(struct btree_trans *trans, u32 snapshot,
			     struct bch_subvolume *subvol)
{
	struct bch_snapshot snap;

	return  snapshot_lookup(trans, snapshot, &snap) ?:
		bch2_subvolume_get(trans, le32_to_cpu(snap.subvol), true, 0, subvol);
}

int bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvol,
				u32 *snapid)
{
	struct bch_subvolume s;
	int ret;

	ret = bch2_subvolume_get_inlined(trans, subvol, true,
					 BTREE_ITER_CACHED|
					 BTREE_ITER_WITH_UPDATES,
					 &s);
	if (!ret)
		*snapid = le32_to_cpu(s.snapshot);
	return ret;
}

/*
 * Delete subvolume, mark snapshot ID as deleted, queue up snapshot
 * deletion/cleanup:
 */
int bch2_subvolume_delete(struct btree_trans *trans, u32 subvolid)
{
	struct btree_iter iter;
	struct bkey_s_c k;
	struct bkey_s_c_subvolume subvol;
	struct btree_trans_commit_hook *h;
	u32 snapid;
	int ret = 0;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_subvolumes,
			     POS(0, subvolid),
			     BTREE_ITER_CACHED|
			     BTREE_ITER_INTENT);
	k = bch2_btree_iter_peek_slot(&iter);
	ret = bkey_err(k);
	if (ret)
		goto err;

	if (k.k->type != KEY_TYPE_subvolume) {
		bch2_fs_inconsistent(trans->c, "missing subvolume %u", subvolid);
		ret = -EIO;
		goto err;
	}

	subvol = bkey_s_c_to_subvolume(k);
	snapid = le32_to_cpu(subvol.v->snapshot);

	ret = bch2_btree_delete_at(trans, &iter, 0);
	if (ret)
		goto err;

	ret = bch2_snapshot_node_set_deleted(trans, snapid);
	if (ret)
		goto err;

	h = bch2_trans_kmalloc(trans, sizeof(*h));
	ret = PTR_ERR_OR_ZERO(h);
	if (ret)
		goto err;

	h->fn = bch2_delete_dead_snapshots_hook;
	bch2_trans_commit_hook(trans, h);
err:
	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

void bch2_subvolume_wait_for_pagecache_and_delete(struct work_struct *work)
{
	struct bch_fs *c = container_of(work, struct bch_fs,
				snapshot_wait_for_pagecache_and_delete_work);
	snapshot_id_list s;
	u32 *id;
	int ret = 0;

	while (!ret) {
		mutex_lock(&c->snapshots_unlinked_lock);
		s = c->snapshots_unlinked;
		darray_init(&c->snapshots_unlinked);
		mutex_unlock(&c->snapshots_unlinked_lock);

		if (!s.nr)
			break;

		bch2_evict_subvolume_inodes(c, &s);

		for (id = s.data; id < s.data + s.nr; id++) {
			ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_NOFAIL,
				      bch2_subvolume_delete(&trans, *id));
			if (ret) {
				bch_err(c, "error deleting subvolume %u: %s", *id, bch2_err_str(ret));
				break;
			}
		}

		darray_exit(&s);
	}

	bch2_write_ref_put(c, BCH_WRITE_REF_snapshot_delete_pagecache);
}

struct subvolume_unlink_hook {
	struct btree_trans_commit_hook	h;
	u32				subvol;
};

int bch2_subvolume_wait_for_pagecache_and_delete_hook(struct btree_trans *trans,
						      struct btree_trans_commit_hook *_h)
{
	struct subvolume_unlink_hook *h = container_of(_h, struct subvolume_unlink_hook, h);
	struct bch_fs *c = trans->c;
	int ret = 0;

	mutex_lock(&c->snapshots_unlinked_lock);
	if (!snapshot_list_has_id(&c->snapshots_unlinked, h->subvol))
		ret = snapshot_list_add(c, &c->snapshots_unlinked, h->subvol);
	mutex_unlock(&c->snapshots_unlinked_lock);

	if (ret)
		return ret;

	if (!bch2_write_ref_tryget(c, BCH_WRITE_REF_snapshot_delete_pagecache))
		return -EROFS;

	if (!queue_work(c->write_ref_wq, &c->snapshot_wait_for_pagecache_and_delete_work))
		bch2_write_ref_put(c, BCH_WRITE_REF_snapshot_delete_pagecache);
	return 0;
}

int bch2_subvolume_unlink(struct btree_trans *trans, u32 subvolid)
{
	struct btree_iter iter;
	struct bkey_i_subvolume *n;
	struct subvolume_unlink_hook *h;
	int ret = 0;

	bch2_trans_iter_init(trans, &iter, BTREE_ID_subvolumes,
			     POS(0, subvolid),
			     BTREE_ITER_CACHED|
			     BTREE_ITER_INTENT);
	n = bch2_bkey_get_mut_typed(trans, &iter, subvolume);
	ret = PTR_ERR_OR_ZERO(n);
	if (unlikely(ret)) {
		bch2_fs_inconsistent_on(ret == -ENOENT, trans->c, "missing subvolume %u", subvolid);
		goto err;
	}

	SET_BCH_SUBVOLUME_UNLINKED(&n->v, true);

	ret = bch2_trans_update(trans, &iter, &n->k_i, 0);
	if (ret)
		goto err;

	h = bch2_trans_kmalloc(trans, sizeof(*h));
	ret = PTR_ERR_OR_ZERO(h);
	if (ret)
		goto err;

	h->h.fn		= bch2_subvolume_wait_for_pagecache_and_delete_hook;
	h->subvol	= subvolid;
	bch2_trans_commit_hook(trans, &h->h);
err:
	bch2_trans_iter_exit(trans, &iter);
	return ret;
}

int bch2_subvolume_create(struct btree_trans *trans, u64 inode,
			  u32 src_subvolid,
			  u32 *new_subvolid,
			  u32 *new_snapshotid,
			  bool ro)
{
	struct bch_fs *c = trans->c;
	struct btree_iter dst_iter, src_iter = (struct btree_iter) { NULL };
	struct bkey_i_subvolume *new_subvol = NULL;
	struct bkey_i_subvolume *src_subvol = NULL;
	struct bkey_s_c k;
	u32 parent = 0, new_nodes[2], snapshot_subvols[2];
	int ret = 0;

	for_each_btree_key(trans, dst_iter, BTREE_ID_subvolumes, SUBVOL_POS_MIN,
			   BTREE_ITER_SLOTS|BTREE_ITER_INTENT, k, ret) {
		if (bkey_gt(k.k->p, SUBVOL_POS_MAX))
			break;

		/*
		 * bch2_subvolume_delete() doesn't flush the btree key cache -
		 * ideally it would but that's tricky
		 */
		if (bkey_deleted(k.k) &&
		    !bch2_btree_key_cache_find(c, BTREE_ID_subvolumes, dst_iter.pos))
			goto found_slot;
	}

	if (!ret)
		ret = -BCH_ERR_ENOSPC_subvolume_create;
	goto err;
found_slot:
	snapshot_subvols[0] = dst_iter.pos.offset;
	snapshot_subvols[1] = src_subvolid;

	if (src_subvolid) {
		/* Creating a snapshot: */

		bch2_trans_iter_init(trans, &src_iter, BTREE_ID_subvolumes,
				     POS(0, src_subvolid),
				     BTREE_ITER_CACHED|
				     BTREE_ITER_INTENT);
		src_subvol = bch2_bkey_get_mut_typed(trans, &src_iter, subvolume);
		ret = PTR_ERR_OR_ZERO(src_subvol);
		if (unlikely(ret)) {
			bch2_fs_inconsistent_on(ret == -ENOENT, trans->c,
						"subvolume %u not found", src_subvolid);
			goto err;
		}

		parent = le32_to_cpu(src_subvol->v.snapshot);
	}

	ret = bch2_snapshot_node_create(trans, parent, new_nodes,
					snapshot_subvols,
					src_subvolid ? 2 : 1);
	if (ret)
		goto err;

	if (src_subvolid) {
		src_subvol->v.snapshot = cpu_to_le32(new_nodes[1]);
		ret = bch2_trans_update(trans, &src_iter, &src_subvol->k_i, 0);
		if (ret)
			goto err;
	}

	new_subvol = bch2_bkey_alloc(trans, &dst_iter, subvolume);
	ret = PTR_ERR_OR_ZERO(new_subvol);
	if (ret)
		goto err;

	new_subvol->v.flags	= 0;
	new_subvol->v.snapshot	= cpu_to_le32(new_nodes[0]);
	new_subvol->v.inode	= cpu_to_le64(inode);
	SET_BCH_SUBVOLUME_RO(&new_subvol->v, ro);
	SET_BCH_SUBVOLUME_SNAP(&new_subvol->v, src_subvolid != 0);
	ret = bch2_trans_update(trans, &dst_iter, &new_subvol->k_i, 0);
	if (ret)
		goto err;

	*new_subvolid	= new_subvol->k.p.offset;
	*new_snapshotid	= new_nodes[0];
err:
	bch2_trans_iter_exit(trans, &src_iter);
	bch2_trans_iter_exit(trans, &dst_iter);
	return ret;
}

int bch2_fs_subvolumes_init(struct bch_fs *c)
{
	INIT_WORK(&c->snapshot_delete_work, bch2_delete_dead_snapshots_work);
	INIT_WORK(&c->snapshot_wait_for_pagecache_and_delete_work,
		  bch2_subvolume_wait_for_pagecache_and_delete);
	mutex_init(&c->snapshots_unlinked_lock);
	return 0;
}