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
|
// SPDX-License-Identifier: GPL-2.0
#include "bcachefs.h"
#include "alloc_foreground.h"
#include "bkey_buf.h"
#include "btree_update.h"
#include "buckets.h"
#include "data_update.h"
#include "ec.h"
#include "extents.h"
#include "io.h"
#include "keylist.h"
#include "move.h"
#include "nocow_locking.h"
#include "subvolume.h"
#include "trace.h"
static int insert_snapshot_whiteouts(struct btree_trans *trans,
enum btree_id id,
struct bpos old_pos,
struct bpos new_pos)
{
struct bch_fs *c = trans->c;
struct btree_iter iter, iter2;
struct bkey_s_c k, k2;
snapshot_id_list s;
struct bkey_i *update;
int ret;
if (!btree_type_has_snapshots(id))
return 0;
darray_init(&s);
if (!bch2_snapshot_has_children(c, old_pos.snapshot))
return 0;
bch2_trans_iter_init(trans, &iter, id, old_pos,
BTREE_ITER_NOT_EXTENTS|
BTREE_ITER_ALL_SNAPSHOTS);
while (1) {
k = bch2_btree_iter_prev(&iter);
ret = bkey_err(k);
if (ret)
break;
if (!k.k)
break;
if (!bkey_eq(old_pos, k.k->p))
break;
if (bch2_snapshot_is_ancestor(c, k.k->p.snapshot, old_pos.snapshot) &&
!snapshot_list_has_ancestor(c, &s, k.k->p.snapshot)) {
struct bpos whiteout_pos = new_pos;
whiteout_pos.snapshot = k.k->p.snapshot;
bch2_trans_iter_init(trans, &iter2, id, whiteout_pos,
BTREE_ITER_NOT_EXTENTS|
BTREE_ITER_INTENT);
k2 = bch2_btree_iter_peek_slot(&iter2);
ret = bkey_err(k2);
if (!ret && k2.k->type == KEY_TYPE_deleted) {
update = bch2_trans_kmalloc(trans, sizeof(struct bkey_i));
ret = PTR_ERR_OR_ZERO(update);
if (ret)
break;
bkey_init(&update->k);
update->k.p = whiteout_pos;
update->k.type = KEY_TYPE_whiteout;
ret = bch2_trans_update(trans, &iter2, update,
BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
}
bch2_trans_iter_exit(trans, &iter2);
if (ret)
break;
ret = snapshot_list_add(c, &s, k.k->p.snapshot);
if (ret)
break;
}
}
bch2_trans_iter_exit(trans, &iter);
darray_exit(&s);
return ret;
}
static void bch2_bkey_mark_dev_cached(struct bkey_s k, unsigned dev)
{
struct bkey_ptrs ptrs = bch2_bkey_ptrs(k);
struct bch_extent_ptr *ptr;
bkey_for_each_ptr(ptrs, ptr)
if (ptr->dev == dev) {
bch2_extent_ptr_set_cached(k, ptr);
return;
}
}
static int __bch2_data_update_index_update(struct btree_trans *trans,
struct bch_write_op *op)
{
struct bch_fs *c = op->c;
struct btree_iter iter;
struct data_update *m =
container_of(op, struct data_update, op);
struct keylist *keys = &op->insert_keys;
struct bkey_buf _new, _insert;
int ret = 0;
bch2_bkey_buf_init(&_new);
bch2_bkey_buf_init(&_insert);
bch2_bkey_buf_realloc(&_insert, c, U8_MAX);
bch2_trans_iter_init(trans, &iter, m->btree_id,
bkey_start_pos(&bch2_keylist_front(keys)->k),
BTREE_ITER_SLOTS|BTREE_ITER_INTENT);
while (1) {
struct bkey_s_c k;
struct bkey_s_c old = bkey_i_to_s_c(m->k.k);
struct bkey_i *insert;
struct bkey_i_extent *new;
const union bch_extent_entry *entry;
struct extent_ptr_decoded p;
struct bpos next_pos;
bool did_work = false;
bool should_check_enospc;
s64 i_sectors_delta = 0, disk_sectors_delta = 0;
unsigned i;
bch2_trans_begin(trans);
k = bch2_btree_iter_peek_slot(&iter);
ret = bkey_err(k);
if (ret)
goto err;
new = bkey_i_to_extent(bch2_keylist_front(keys));
if (!bch2_extents_match(k, old))
goto nomatch;
bkey_reassemble(_insert.k, k);
insert = _insert.k;
bch2_bkey_buf_copy(&_new, c, bch2_keylist_front(keys));
new = bkey_i_to_extent(_new.k);
bch2_cut_front(iter.pos, &new->k_i);
bch2_cut_front(iter.pos, insert);
bch2_cut_back(new->k.p, insert);
bch2_cut_back(insert->k.p, &new->k_i);
/*
* @old: extent that we read from
* @insert: key that we're going to update, initialized from
* extent currently in btree - same as @old unless we raced with
* other updates
* @new: extent with new pointers that we'll be adding to @insert
*
* Fist, drop rewrite_ptrs from @new:
*/
i = 0;
bkey_for_each_ptr_decode(old.k, bch2_bkey_ptrs_c(old), p, entry) {
if (((1U << i) & m->data_opts.rewrite_ptrs) &&
bch2_extent_has_ptr(old, p, bkey_i_to_s_c(insert))) {
/*
* If we're going to be adding a pointer to the
* same device, we have to drop the old one -
* otherwise, we can just mark it cached:
*/
if (bch2_bkey_has_device(bkey_i_to_s_c(&new->k_i), p.ptr.dev))
bch2_bkey_drop_device_noerror(bkey_i_to_s(insert), p.ptr.dev);
else
bch2_bkey_mark_dev_cached(bkey_i_to_s(insert), p.ptr.dev);
}
i++;
}
/* Add new ptrs: */
extent_for_each_ptr_decode(extent_i_to_s(new), p, entry) {
const struct bch_extent_ptr *existing_ptr =
bch2_bkey_has_device(bkey_i_to_s_c(insert), p.ptr.dev);
if (existing_ptr && existing_ptr->cached) {
/*
* We're replacing a cached pointer with a non
* cached pointer:
*/
bch2_bkey_drop_device_noerror(bkey_i_to_s(insert),
existing_ptr->dev);
} else if (existing_ptr) {
/*
* raced with another move op? extent already
* has a pointer to the device we just wrote
* data to
*/
continue;
}
bch2_extent_ptr_decoded_append(insert, &p);
did_work = true;
}
if (!did_work)
goto nomatch;
bch2_bkey_narrow_crcs(insert, (struct bch_extent_crc_unpacked) { 0 });
bch2_extent_normalize(c, bkey_i_to_s(insert));
ret = bch2_sum_sector_overwrites(trans, &iter, insert,
&should_check_enospc,
&i_sectors_delta,
&disk_sectors_delta);
if (ret)
goto err;
if (disk_sectors_delta > (s64) op->res.sectors) {
ret = bch2_disk_reservation_add(c, &op->res,
disk_sectors_delta - op->res.sectors,
!should_check_enospc
? BCH_DISK_RESERVATION_NOFAIL : 0);
if (ret)
goto out;
}
next_pos = insert->k.p;
if (!bkey_eq(bkey_start_pos(&insert->k), bkey_start_pos(k.k))) {
ret = insert_snapshot_whiteouts(trans, m->btree_id, k.k->p,
bkey_start_pos(&insert->k));
if (ret)
goto err;
}
if (!bkey_eq(insert->k.p, k.k->p)) {
ret = insert_snapshot_whiteouts(trans, m->btree_id,
k.k->p, insert->k.p);
if (ret)
goto err;
}
ret = bch2_trans_update(trans, &iter, insert,
BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE) ?:
bch2_trans_commit(trans, &op->res,
NULL,
BTREE_INSERT_NOFAIL|
m->data_opts.btree_insert_flags);
if (!ret) {
bch2_btree_iter_set_pos(&iter, next_pos);
this_cpu_add(c->counters[BCH_COUNTER_move_extent_finish], new->k.size);
trace_move_extent_finish(&new->k);
}
err:
if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
ret = 0;
if (ret)
break;
next:
while (bkey_ge(iter.pos, bch2_keylist_front(keys)->k.p)) {
bch2_keylist_pop_front(keys);
if (bch2_keylist_empty(keys))
goto out;
}
continue;
nomatch:
if (m->ctxt && m->ctxt->stats) {
BUG_ON(k.k->p.offset <= iter.pos.offset);
atomic64_inc(&m->ctxt->stats->keys_raced);
atomic64_add(k.k->p.offset - iter.pos.offset,
&m->ctxt->stats->sectors_raced);
}
this_cpu_add(c->counters[BCH_COUNTER_move_extent_fail], new->k.size);
trace_move_extent_fail(&new->k);
bch2_btree_iter_advance(&iter);
goto next;
}
out:
bch2_trans_iter_exit(trans, &iter);
bch2_bkey_buf_exit(&_insert, c);
bch2_bkey_buf_exit(&_new, c);
BUG_ON(bch2_err_matches(ret, BCH_ERR_transaction_restart));
return ret;
}
int bch2_data_update_index_update(struct bch_write_op *op)
{
return bch2_trans_run(op->c, __bch2_data_update_index_update(&trans, op));
}
void bch2_data_update_read_done(struct data_update *m,
struct bch_extent_crc_unpacked crc)
{
/* write bio must own pages: */
BUG_ON(!m->op.wbio.bio.bi_vcnt);
m->op.crc = crc;
m->op.wbio.bio.bi_iter.bi_size = crc.compressed_size << 9;
closure_call(&m->op.cl, bch2_write, NULL, NULL);
}
void bch2_data_update_exit(struct data_update *update)
{
struct bch_fs *c = update->op.c;
struct bkey_ptrs_c ptrs =
bch2_bkey_ptrs_c(bkey_i_to_s_c(update->k.k));
const struct bch_extent_ptr *ptr;
bkey_for_each_ptr(ptrs, ptr) {
if (c->opts.nocow_enabled)
bch2_bucket_nocow_unlock(&c->nocow_locks,
PTR_BUCKET_POS(c, ptr), 0);
percpu_ref_put(&bch_dev_bkey_exists(c, ptr->dev)->ref);
}
bch2_bkey_buf_exit(&update->k, c);
bch2_disk_reservation_put(c, &update->op.res);
bch2_bio_free_pages_pool(c, &update->op.wbio.bio);
}
void bch2_update_unwritten_extent(struct btree_trans *trans,
struct data_update *update)
{
struct bch_fs *c = update->op.c;
struct bio *bio = &update->op.wbio.bio;
struct bkey_i_extent *e;
struct write_point *wp;
struct bch_extent_ptr *ptr;
struct closure cl;
struct btree_iter iter;
struct bkey_s_c k;
int ret;
closure_init_stack(&cl);
bch2_keylist_init(&update->op.insert_keys, update->op.inline_keys);
while (bio_sectors(bio)) {
unsigned sectors = bio_sectors(bio);
bch2_trans_iter_init(trans, &iter, update->btree_id, update->op.pos,
BTREE_ITER_SLOTS);
ret = lockrestart_do(trans, ({
k = bch2_btree_iter_peek_slot(&iter);
bkey_err(k);
}));
bch2_trans_iter_exit(trans, &iter);
if (ret || !bch2_extents_match(k, bkey_i_to_s_c(update->k.k)))
break;
e = bkey_extent_init(update->op.insert_keys.top);
e->k.p = update->op.pos;
ret = bch2_alloc_sectors_start_trans(trans,
update->op.target,
false,
update->op.write_point,
&update->op.devs_have,
update->op.nr_replicas,
update->op.nr_replicas,
update->op.alloc_reserve,
0, &cl, &wp);
if (bch2_err_matches(ret, BCH_ERR_operation_blocked)) {
bch2_trans_unlock(trans);
closure_sync(&cl);
continue;
}
if (ret)
return;
sectors = min(sectors, wp->sectors_free);
bch2_key_resize(&e->k, sectors);
bch2_open_bucket_get(c, wp, &update->op.open_buckets);
bch2_alloc_sectors_append_ptrs(c, wp, &e->k_i, sectors, false);
bch2_alloc_sectors_done(c, wp);
bio_advance(bio, sectors << 9);
update->op.pos.offset += sectors;
extent_for_each_ptr(extent_i_to_s(e), ptr)
ptr->unwritten = true;
bch2_keylist_push(&update->op.insert_keys);
ret = __bch2_data_update_index_update(trans, &update->op);
bch2_open_buckets_put(c, &update->op.open_buckets);
if (ret)
break;
}
if ((atomic_read(&cl.remaining) & CLOSURE_REMAINING_MASK) != 1) {
bch2_trans_unlock(trans);
closure_sync(&cl);
}
}
int bch2_data_update_init(struct btree_trans *trans,
struct moving_context *ctxt,
struct data_update *m,
struct write_point_specifier wp,
struct bch_io_opts io_opts,
struct data_update_opts data_opts,
enum btree_id btree_id,
struct bkey_s_c k)
{
struct bch_fs *c = trans->c;
struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k);
const union bch_extent_entry *entry;
struct extent_ptr_decoded p;
const struct bch_extent_ptr *ptr;
unsigned i, reserve_sectors = k.k->size * data_opts.extra_replicas;
unsigned ptrs_locked = 0;
int ret;
bch2_bkey_buf_init(&m->k);
bch2_bkey_buf_reassemble(&m->k, c, k);
m->btree_id = btree_id;
m->data_opts = data_opts;
bch2_write_op_init(&m->op, c, io_opts);
m->op.pos = bkey_start_pos(k.k);
m->op.version = k.k->version;
m->op.target = data_opts.target;
m->op.write_point = wp;
m->op.nr_replicas = 0;
m->op.flags |= BCH_WRITE_PAGES_STABLE|
BCH_WRITE_PAGES_OWNED|
BCH_WRITE_DATA_ENCODED|
BCH_WRITE_MOVE|
m->data_opts.write_flags;
m->op.compression_type =
bch2_compression_opt_to_type[io_opts.background_compression ?:
io_opts.compression];
if (m->data_opts.btree_insert_flags & BTREE_INSERT_USE_RESERVE)
m->op.alloc_reserve = RESERVE_movinggc;
bkey_for_each_ptr(ptrs, ptr)
percpu_ref_get(&bch_dev_bkey_exists(c, ptr->dev)->ref);
i = 0;
bkey_for_each_ptr_decode(k.k, ptrs, p, entry) {
bool locked;
if (((1U << i) & m->data_opts.rewrite_ptrs)) {
BUG_ON(p.ptr.cached);
if (crc_is_compressed(p.crc))
reserve_sectors += k.k->size;
m->op.nr_replicas += bch2_extent_ptr_durability(c, &p);
} else if (!p.ptr.cached) {
bch2_dev_list_add_dev(&m->op.devs_have, p.ptr.dev);
}
/*
* op->csum_type is normally initialized from the fs/file's
* current options - but if an extent is encrypted, we require
* that it stays encrypted:
*/
if (bch2_csum_type_is_encryption(p.crc.csum_type)) {
m->op.nonce = p.crc.nonce + p.crc.offset;
m->op.csum_type = p.crc.csum_type;
}
if (p.crc.compression_type == BCH_COMPRESSION_TYPE_incompressible)
m->op.incompressible = true;
if (c->opts.nocow_enabled) {
if (ctxt) {
move_ctxt_wait_event(ctxt, trans,
(locked = bch2_bucket_nocow_trylock(&c->nocow_locks,
PTR_BUCKET_POS(c, &p.ptr), 0)) ||
!atomic_read(&ctxt->read_sectors));
if (!locked)
bch2_bucket_nocow_lock(&c->nocow_locks,
PTR_BUCKET_POS(c, &p.ptr), 0);
} else {
if (!bch2_bucket_nocow_trylock(&c->nocow_locks,
PTR_BUCKET_POS(c, &p.ptr), 0)) {
ret = -BCH_ERR_nocow_lock_blocked;
goto err;
}
}
ptrs_locked |= (1U << i);
}
i++;
}
if (reserve_sectors) {
ret = bch2_disk_reservation_add(c, &m->op.res, reserve_sectors,
m->data_opts.extra_replicas
? 0
: BCH_DISK_RESERVATION_NOFAIL);
if (ret)
goto err;
}
m->op.nr_replicas += m->data_opts.extra_replicas;
m->op.nr_replicas_required = m->op.nr_replicas;
BUG_ON(!m->op.nr_replicas);
/* Special handling required: */
if (bkey_extent_is_unwritten(k))
return -BCH_ERR_unwritten_extent_update;
return 0;
err:
i = 0;
bkey_for_each_ptr_decode(k.k, ptrs, p, entry) {
if ((1U << i) & ptrs_locked)
bch2_bucket_nocow_unlock(&c->nocow_locks,
PTR_BUCKET_POS(c, &p.ptr), 0);
percpu_ref_put(&bch_dev_bkey_exists(c, p.ptr.dev)->ref);
i++;
}
bch2_bkey_buf_exit(&m->k, c);
bch2_bio_free_pages_pool(c, &m->op.wbio.bio);
return ret;
}
void bch2_data_update_opts_normalize(struct bkey_s_c k, struct data_update_opts *opts)
{
struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k);
const struct bch_extent_ptr *ptr;
unsigned i = 0;
bkey_for_each_ptr(ptrs, ptr) {
if ((opts->rewrite_ptrs & (1U << i)) && ptr->cached) {
opts->kill_ptrs |= 1U << i;
opts->rewrite_ptrs ^= 1U << i;
}
i++;
}
}
|