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
|
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012 by Delphix. All rights reserved.
*/
#include <sys/zfs_context.h>
#include <sys/dmu_objset.h>
#include <sys/dmu_traverse.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_dir.h>
#include <sys/dsl_pool.h>
#include <sys/dnode.h>
#include <sys/spa.h>
#include <sys/zio.h>
#include <sys/dmu_impl.h>
#include <sys/sa.h>
#include <sys/sa_impl.h>
#include <sys/callb.h>
int zfs_pd_blks_max = 100;
typedef struct prefetch_data {
kmutex_t pd_mtx;
kcondvar_t pd_cv;
int pd_blks_max;
int pd_blks_fetched;
int pd_flags;
boolean_t pd_cancel;
boolean_t pd_exited;
} prefetch_data_t;
typedef struct traverse_data {
spa_t *td_spa;
uint64_t td_objset;
blkptr_t *td_rootbp;
uint64_t td_min_txg;
zbookmark_t *td_resume;
int td_flags;
prefetch_data_t *td_pfd;
blkptr_cb_t *td_func;
void *td_arg;
} traverse_data_t;
static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object);
static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *,
uint64_t objset, uint64_t object);
static int
traverse_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg)
{
traverse_data_t *td = arg;
zbookmark_t zb;
if (bp->blk_birth == 0)
return (0);
if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(td->td_spa))
return (0);
SET_BOOKMARK(&zb, td->td_objset, ZB_ZIL_OBJECT, ZB_ZIL_LEVEL,
bp->blk_cksum.zc_word[ZIL_ZC_SEQ]);
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg);
return (0);
}
static int
traverse_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg)
{
traverse_data_t *td = arg;
if (lrc->lrc_txtype == TX_WRITE) {
lr_write_t *lr = (lr_write_t *)lrc;
blkptr_t *bp = &lr->lr_blkptr;
zbookmark_t zb;
if (bp->blk_birth == 0)
return (0);
if (claim_txg == 0 || bp->blk_birth < claim_txg)
return (0);
SET_BOOKMARK(&zb, td->td_objset, lr->lr_foid,
ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp));
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL,
td->td_arg);
}
return (0);
}
static void
traverse_zil(traverse_data_t *td, zil_header_t *zh)
{
uint64_t claim_txg = zh->zh_claim_txg;
zilog_t *zilog;
/*
* We only want to visit blocks that have been claimed but not yet
* replayed; plus, in read-only mode, blocks that are already stable.
*/
if (claim_txg == 0 && spa_writeable(td->td_spa))
return;
zilog = zil_alloc(spa_get_dsl(td->td_spa)->dp_meta_objset, zh);
(void) zil_parse(zilog, traverse_zil_block, traverse_zil_record, td,
claim_txg);
zil_free(zilog);
}
typedef enum resume_skip {
RESUME_SKIP_ALL,
RESUME_SKIP_NONE,
RESUME_SKIP_CHILDREN
} resume_skip_t;
/*
* Returns RESUME_SKIP_ALL if td indicates that we are resuming a traversal and
* the block indicated by zb does not need to be visited at all. Returns
* RESUME_SKIP_CHILDREN if we are resuming a post traversal and we reach the
* resume point. This indicates that this block should be visited but not its
* children (since they must have been visited in a previous traversal).
* Otherwise returns RESUME_SKIP_NONE.
*/
static resume_skip_t
resume_skip_check(traverse_data_t *td, const dnode_phys_t *dnp,
const zbookmark_t *zb)
{
if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume)) {
/*
* If we already visited this bp & everything below,
* don't bother doing it again.
*/
if (zbookmark_is_before(dnp, zb, td->td_resume))
return (RESUME_SKIP_ALL);
/*
* If we found the block we're trying to resume from, zero
* the bookmark out to indicate that we have resumed.
*/
ASSERT3U(zb->zb_object, <=, td->td_resume->zb_object);
if (bcmp(zb, td->td_resume, sizeof (*zb)) == 0) {
bzero(td->td_resume, sizeof (*zb));
if (td->td_flags & TRAVERSE_POST)
return (RESUME_SKIP_CHILDREN);
}
}
return (RESUME_SKIP_NONE);
}
static void
traverse_pause(traverse_data_t *td, const zbookmark_t *zb)
{
ASSERT(td->td_resume != NULL);
ASSERT0(zb->zb_level);
bcopy(zb, td->td_resume, sizeof (*td->td_resume));
}
static void
traverse_prefetch_metadata(traverse_data_t *td,
const blkptr_t *bp, const zbookmark_t *zb)
{
uint32_t flags = ARC_NOWAIT | ARC_PREFETCH;
if (!(td->td_flags & TRAVERSE_PREFETCH_METADATA))
return;
/*
* If we are in the process of resuming, don't prefetch, because
* some children will not be needed (and in fact may have already
* been freed).
*/
if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume))
return;
if (BP_IS_HOLE(bp) || bp->blk_birth <= td->td_min_txg)
return;
if (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE)
return;
(void) arc_read(NULL, td->td_spa, bp, NULL, NULL,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
}
static int
traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
const blkptr_t *bp, const zbookmark_t *zb)
{
zbookmark_t czb;
int err = 0, lasterr = 0;
arc_buf_t *buf = NULL;
prefetch_data_t *pd = td->td_pfd;
boolean_t hard = td->td_flags & TRAVERSE_HARD;
boolean_t pause = B_FALSE;
switch (resume_skip_check(td, dnp, zb)) {
case RESUME_SKIP_ALL:
return (0);
case RESUME_SKIP_CHILDREN:
goto post;
case RESUME_SKIP_NONE:
break;
default:
ASSERT(0);
}
if (BP_IS_HOLE(bp)) {
err = td->td_func(td->td_spa, NULL, NULL, zb, dnp, td->td_arg);
return (err);
}
if (bp->blk_birth <= td->td_min_txg)
return (0);
if (pd && !pd->pd_exited &&
((pd->pd_flags & TRAVERSE_PREFETCH_DATA) ||
BP_GET_TYPE(bp) == DMU_OT_DNODE || BP_GET_LEVEL(bp) > 0)) {
mutex_enter(&pd->pd_mtx);
ASSERT(pd->pd_blks_fetched >= 0);
while (pd->pd_blks_fetched == 0 && !pd->pd_exited)
cv_wait(&pd->pd_cv, &pd->pd_mtx);
pd->pd_blks_fetched--;
cv_broadcast(&pd->pd_cv);
mutex_exit(&pd->pd_mtx);
}
if (td->td_flags & TRAVERSE_PRE) {
err = td->td_func(td->td_spa, NULL, bp, zb, dnp,
td->td_arg);
if (err == TRAVERSE_VISIT_NO_CHILDREN)
return (0);
if (err == ERESTART)
pause = B_TRUE; /* handle pausing at a common point */
if (err != 0)
goto post;
}
if (BP_GET_LEVEL(bp) > 0) {
uint32_t flags = ARC_WAIT;
int i;
blkptr_t *cbp;
int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err)
return (err);
cbp = buf->b_data;
for (i = 0; i < epb; i++) {
SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object,
zb->zb_level - 1,
zb->zb_blkid * epb + i);
traverse_prefetch_metadata(td, &cbp[i], &czb);
}
/* recursively visitbp() blocks below this */
for (i = 0; i < epb; i++) {
SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object,
zb->zb_level - 1,
zb->zb_blkid * epb + i);
err = traverse_visitbp(td, dnp, &cbp[i], &czb);
if (err) {
if (!hard)
break;
lasterr = err;
}
}
} else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) {
uint32_t flags = ARC_WAIT;
int i;
int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err)
return (err);
dnp = buf->b_data;
for (i = 0; i < epb; i++) {
prefetch_dnode_metadata(td, &dnp[i], zb->zb_objset,
zb->zb_blkid * epb + i);
}
/* recursively visitbp() blocks below this */
for (i = 0; i < epb; i++) {
err = traverse_dnode(td, &dnp[i], zb->zb_objset,
zb->zb_blkid * epb + i);
if (err) {
if (!hard)
break;
lasterr = err;
}
}
} else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) {
uint32_t flags = ARC_WAIT;
objset_phys_t *osp;
dnode_phys_t *dnp;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err)
return (err);
osp = buf->b_data;
dnp = &osp->os_meta_dnode;
prefetch_dnode_metadata(td, dnp, zb->zb_objset,
DMU_META_DNODE_OBJECT);
if (arc_buf_size(buf) >= sizeof (objset_phys_t)) {
prefetch_dnode_metadata(td, &osp->os_userused_dnode,
zb->zb_objset, DMU_USERUSED_OBJECT);
prefetch_dnode_metadata(td, &osp->os_groupused_dnode,
zb->zb_objset, DMU_USERUSED_OBJECT);
}
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_META_DNODE_OBJECT);
if (err && hard) {
lasterr = err;
err = 0;
}
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
dnp = &osp->os_userused_dnode;
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_USERUSED_OBJECT);
}
if (err && hard) {
lasterr = err;
err = 0;
}
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
dnp = &osp->os_groupused_dnode;
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_GROUPUSED_OBJECT);
}
}
if (buf)
(void) arc_buf_remove_ref(buf, &buf);
post:
if (err == 0 && lasterr == 0 && (td->td_flags & TRAVERSE_POST)) {
err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg);
if (err == ERESTART)
pause = B_TRUE;
}
if (pause && td->td_resume != NULL) {
ASSERT3U(err, ==, ERESTART);
ASSERT(!hard);
traverse_pause(td, zb);
}
return (err != 0 ? err : lasterr);
}
static void
prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object)
{
int j;
zbookmark_t czb;
for (j = 0; j < dnp->dn_nblkptr; j++) {
SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j);
traverse_prefetch_metadata(td, &dnp->dn_blkptr[j], &czb);
}
if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
traverse_prefetch_metadata(td, &dnp->dn_spill, &czb);
}
}
static int
traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object)
{
int j, err = 0, lasterr = 0;
zbookmark_t czb;
boolean_t hard = (td->td_flags & TRAVERSE_HARD);
for (j = 0; j < dnp->dn_nblkptr; j++) {
SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j);
err = traverse_visitbp(td, dnp, &dnp->dn_blkptr[j], &czb);
if (err) {
if (!hard)
break;
lasterr = err;
}
}
if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
err = traverse_visitbp(td, dnp, &dnp->dn_spill, &czb);
if (err) {
if (!hard)
return (err);
lasterr = err;
}
}
return (err != 0 ? err : lasterr);
}
/* ARGSUSED */
static int
traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
const zbookmark_t *zb, const dnode_phys_t *dnp, void *arg)
{
prefetch_data_t *pfd = arg;
uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH;
ASSERT(pfd->pd_blks_fetched >= 0);
if (pfd->pd_cancel)
return (EINTR);
if (bp == NULL || !((pfd->pd_flags & TRAVERSE_PREFETCH_DATA) ||
BP_GET_TYPE(bp) == DMU_OT_DNODE || BP_GET_LEVEL(bp) > 0) ||
BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG)
return (0);
mutex_enter(&pfd->pd_mtx);
while (!pfd->pd_cancel && pfd->pd_blks_fetched >= pfd->pd_blks_max)
cv_wait(&pfd->pd_cv, &pfd->pd_mtx);
pfd->pd_blks_fetched++;
cv_broadcast(&pfd->pd_cv);
mutex_exit(&pfd->pd_mtx);
(void) arc_read(NULL, spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, zb);
return (0);
}
static void
traverse_prefetch_thread(void *arg)
{
traverse_data_t *td_main = arg;
traverse_data_t td = *td_main;
zbookmark_t czb;
td.td_func = traverse_prefetcher;
td.td_arg = td_main->td_pfd;
td.td_pfd = NULL;
SET_BOOKMARK(&czb, td.td_objset,
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
(void) traverse_visitbp(&td, NULL, td.td_rootbp, &czb);
mutex_enter(&td_main->td_pfd->pd_mtx);
td_main->td_pfd->pd_exited = B_TRUE;
cv_broadcast(&td_main->td_pfd->pd_cv);
mutex_exit(&td_main->td_pfd->pd_mtx);
}
/*
* NB: dataset must not be changing on-disk (eg, is a snapshot or we are
* in syncing context).
*/
static int
traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp,
uint64_t txg_start, zbookmark_t *resume, int flags,
blkptr_cb_t func, void *arg)
{
traverse_data_t td;
prefetch_data_t pd = { 0 };
zbookmark_t czb;
int err;
ASSERT(ds == NULL || objset == ds->ds_object);
ASSERT(!(flags & TRAVERSE_PRE) || !(flags & TRAVERSE_POST));
/*
* The data prefetching mechanism (the prefetch thread) is incompatible
* with resuming from a bookmark.
*/
ASSERT(resume == NULL || !(flags & TRAVERSE_PREFETCH_DATA));
td.td_spa = spa;
td.td_objset = objset;
td.td_rootbp = rootbp;
td.td_min_txg = txg_start;
td.td_resume = resume;
td.td_func = func;
td.td_arg = arg;
td.td_pfd = &pd;
td.td_flags = flags;
pd.pd_blks_max = zfs_pd_blks_max;
pd.pd_flags = flags;
mutex_init(&pd.pd_mtx, NULL, MUTEX_DEFAULT, NULL);
cv_init(&pd.pd_cv, NULL, CV_DEFAULT, NULL);
/* See comment on ZIL traversal in dsl_scan_visitds. */
if (ds != NULL && !dsl_dataset_is_snapshot(ds)) {
objset_t *os;
err = dmu_objset_from_ds(ds, &os);
if (err)
return (err);
traverse_zil(&td, &os->os_zil_header);
}
if (!(flags & TRAVERSE_PREFETCH_DATA) ||
0 == taskq_dispatch(system_taskq, traverse_prefetch_thread,
&td, TQ_NOQUEUE))
pd.pd_exited = B_TRUE;
SET_BOOKMARK(&czb, td.td_objset,
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
err = traverse_visitbp(&td, NULL, rootbp, &czb);
mutex_enter(&pd.pd_mtx);
pd.pd_cancel = B_TRUE;
cv_broadcast(&pd.pd_cv);
while (!pd.pd_exited)
cv_wait(&pd.pd_cv, &pd.pd_mtx);
mutex_exit(&pd.pd_mtx);
mutex_destroy(&pd.pd_mtx);
cv_destroy(&pd.pd_cv);
return (err);
}
/*
* NB: dataset must not be changing on-disk (eg, is a snapshot or we are
* in syncing context).
*/
int
traverse_dataset(dsl_dataset_t *ds, uint64_t txg_start, int flags,
blkptr_cb_t func, void *arg)
{
return (traverse_impl(ds->ds_dir->dd_pool->dp_spa, ds, ds->ds_object,
&ds->ds_phys->ds_bp, txg_start, NULL, flags, func, arg));
}
int
traverse_dataset_destroyed(spa_t *spa, blkptr_t *blkptr,
uint64_t txg_start, zbookmark_t *resume, int flags,
blkptr_cb_t func, void *arg)
{
return (traverse_impl(spa, NULL, ZB_DESTROYED_OBJSET,
blkptr, txg_start, resume, flags, func, arg));
}
/*
* NB: pool must not be changing on-disk (eg, from zdb or sync context).
*/
int
traverse_pool(spa_t *spa, uint64_t txg_start, int flags,
blkptr_cb_t func, void *arg)
{
int err, lasterr = 0;
uint64_t obj;
dsl_pool_t *dp = spa_get_dsl(spa);
objset_t *mos = dp->dp_meta_objset;
boolean_t hard = (flags & TRAVERSE_HARD);
/* visit the MOS */
err = traverse_impl(spa, NULL, 0, spa_get_rootblkptr(spa),
txg_start, NULL, flags, func, arg);
if (err)
return (err);
/* visit each dataset */
for (obj = 1; err == 0 || (err != ESRCH && hard);
err = dmu_object_next(mos, &obj, FALSE, txg_start)) {
dmu_object_info_t doi;
err = dmu_object_info(mos, obj, &doi);
if (err) {
if (!hard)
return (err);
lasterr = err;
continue;
}
if (doi.doi_type == DMU_OT_DSL_DATASET) {
dsl_dataset_t *ds;
uint64_t txg = txg_start;
rw_enter(&dp->dp_config_rwlock, RW_READER);
err = dsl_dataset_hold_obj(dp, obj, FTAG, &ds);
rw_exit(&dp->dp_config_rwlock);
if (err) {
if (!hard)
return (err);
lasterr = err;
continue;
}
if (ds->ds_phys->ds_prev_snap_txg > txg)
txg = ds->ds_phys->ds_prev_snap_txg;
err = traverse_dataset(ds, txg, flags, func, arg);
dsl_dataset_rele(ds, FTAG);
if (err) {
if (!hard)
return (err);
lasterr = err;
}
}
}
if (err == ESRCH)
err = 0;
return (err != 0 ? err : lasterr);
}
|