forked from torvalds/linux
-
Notifications
You must be signed in to change notification settings - Fork 0
/
xfs_reflink.c
1639 lines (1443 loc) · 45.3 KB
/
xfs_reflink.c
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
// SPDX-License-Identifier: GPL-2.0+
/*
* Copyright (C) 2016 Oracle. All Rights Reserved.
* Author: Darrick J. Wong <[email protected]>
*/
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_inode.h"
#include "xfs_trans.h"
#include "xfs_bmap.h"
#include "xfs_bmap_util.h"
#include "xfs_trace.h"
#include "xfs_icache.h"
#include "xfs_btree.h"
#include "xfs_refcount_btree.h"
#include "xfs_refcount.h"
#include "xfs_bmap_btree.h"
#include "xfs_trans_space.h"
#include "xfs_bit.h"
#include "xfs_alloc.h"
#include "xfs_quota.h"
#include "xfs_reflink.h"
#include "xfs_iomap.h"
#include "xfs_sb.h"
#include "xfs_ag_resv.h"
/*
* Copy on Write of Shared Blocks
*
* XFS must preserve "the usual" file semantics even when two files share
* the same physical blocks. This means that a write to one file must not
* alter the blocks in a different file; the way that we'll do that is
* through the use of a copy-on-write mechanism. At a high level, that
* means that when we want to write to a shared block, we allocate a new
* block, write the data to the new block, and if that succeeds we map the
* new block into the file.
*
* XFS provides a "delayed allocation" mechanism that defers the allocation
* of disk blocks to dirty-but-not-yet-mapped file blocks as long as
* possible. This reduces fragmentation by enabling the filesystem to ask
* for bigger chunks less often, which is exactly what we want for CoW.
*
* The delalloc mechanism begins when the kernel wants to make a block
* writable (write_begin or page_mkwrite). If the offset is not mapped, we
* create a delalloc mapping, which is a regular in-core extent, but without
* a real startblock. (For delalloc mappings, the startblock encodes both
* a flag that this is a delalloc mapping, and a worst-case estimate of how
* many blocks might be required to put the mapping into the BMBT.) delalloc
* mappings are a reservation against the free space in the filesystem;
* adjacent mappings can also be combined into fewer larger mappings.
*
* As an optimization, the CoW extent size hint (cowextsz) creates
* outsized aligned delalloc reservations in the hope of landing out of
* order nearby CoW writes in a single extent on disk, thereby reducing
* fragmentation and improving future performance.
*
* D: --RRRRRRSSSRRRRRRRR--- (data fork)
* C: ------DDDDDDD--------- (CoW fork)
*
* When dirty pages are being written out (typically in writepage), the
* delalloc reservations are converted into unwritten mappings by
* allocating blocks and replacing the delalloc mapping with real ones.
* A delalloc mapping can be replaced by several unwritten ones if the
* free space is fragmented.
*
* D: --RRRRRRSSSRRRRRRRR---
* C: ------UUUUUUU---------
*
* We want to adapt the delalloc mechanism for copy-on-write, since the
* write paths are similar. The first two steps (creating the reservation
* and allocating the blocks) are exactly the same as delalloc except that
* the mappings must be stored in a separate CoW fork because we do not want
* to disturb the mapping in the data fork until we're sure that the write
* succeeded. IO completion in this case is the process of removing the old
* mapping from the data fork and moving the new mapping from the CoW fork to
* the data fork. This will be discussed shortly.
*
* For now, unaligned directio writes will be bounced back to the page cache.
* Block-aligned directio writes will use the same mechanism as buffered
* writes.
*
* Just prior to submitting the actual disk write requests, we convert
* the extents representing the range of the file actually being written
* (as opposed to extra pieces created for the cowextsize hint) to real
* extents. This will become important in the next step:
*
* D: --RRRRRRSSSRRRRRRRR---
* C: ------UUrrUUU---------
*
* CoW remapping must be done after the data block write completes,
* because we don't want to destroy the old data fork map until we're sure
* the new block has been written. Since the new mappings are kept in a
* separate fork, we can simply iterate these mappings to find the ones
* that cover the file blocks that we just CoW'd. For each extent, simply
* unmap the corresponding range in the data fork, map the new range into
* the data fork, and remove the extent from the CoW fork. Because of
* the presence of the cowextsize hint, however, we must be careful
* only to remap the blocks that we've actually written out -- we must
* never remap delalloc reservations nor CoW staging blocks that have
* yet to be written. This corresponds exactly to the real extents in
* the CoW fork:
*
* D: --RRRRRRrrSRRRRRRRR---
* C: ------UU--UUU---------
*
* Since the remapping operation can be applied to an arbitrary file
* range, we record the need for the remap step as a flag in the ioend
* instead of declaring a new IO type. This is required for direct io
* because we only have ioend for the whole dio, and we have to be able to
* remember the presence of unwritten blocks and CoW blocks with a single
* ioend structure. Better yet, the more ground we can cover with one
* ioend, the better.
*/
/*
* Given an AG extent, find the lowest-numbered run of shared blocks
* within that range and return the range in fbno/flen. If
* find_end_of_shared is true, return the longest contiguous extent of
* shared blocks. If there are no shared extents, fbno and flen will
* be set to NULLAGBLOCK and 0, respectively.
*/
int
xfs_reflink_find_shared(
struct xfs_mount *mp,
struct xfs_trans *tp,
xfs_agnumber_t agno,
xfs_agblock_t agbno,
xfs_extlen_t aglen,
xfs_agblock_t *fbno,
xfs_extlen_t *flen,
bool find_end_of_shared)
{
struct xfs_buf *agbp;
struct xfs_btree_cur *cur;
int error;
error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
if (error)
return error;
if (!agbp)
return -ENOMEM;
cur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno);
error = xfs_refcount_find_shared(cur, agbno, aglen, fbno, flen,
find_end_of_shared);
xfs_btree_del_cursor(cur, error);
xfs_trans_brelse(tp, agbp);
return error;
}
/*
* Trim the mapping to the next block where there's a change in the
* shared/unshared status. More specifically, this means that we
* find the lowest-numbered extent of shared blocks that coincides with
* the given block mapping. If the shared extent overlaps the start of
* the mapping, trim the mapping to the end of the shared extent. If
* the shared region intersects the mapping, trim the mapping to the
* start of the shared extent. If there are no shared regions that
* overlap, just return the original extent.
*/
int
xfs_reflink_trim_around_shared(
struct xfs_inode *ip,
struct xfs_bmbt_irec *irec,
bool *shared)
{
xfs_agnumber_t agno;
xfs_agblock_t agbno;
xfs_extlen_t aglen;
xfs_agblock_t fbno;
xfs_extlen_t flen;
int error = 0;
/* Holes, unwritten, and delalloc extents cannot be shared */
if (!xfs_is_cow_inode(ip) || !xfs_bmap_is_real_extent(irec)) {
*shared = false;
return 0;
}
trace_xfs_reflink_trim_around_shared(ip, irec);
agno = XFS_FSB_TO_AGNO(ip->i_mount, irec->br_startblock);
agbno = XFS_FSB_TO_AGBNO(ip->i_mount, irec->br_startblock);
aglen = irec->br_blockcount;
error = xfs_reflink_find_shared(ip->i_mount, NULL, agno, agbno,
aglen, &fbno, &flen, true);
if (error)
return error;
*shared = false;
if (fbno == NULLAGBLOCK) {
/* No shared blocks at all. */
return 0;
} else if (fbno == agbno) {
/*
* The start of this extent is shared. Truncate the
* mapping at the end of the shared region so that a
* subsequent iteration starts at the start of the
* unshared region.
*/
irec->br_blockcount = flen;
*shared = true;
return 0;
} else {
/*
* There's a shared extent midway through this extent.
* Truncate the mapping at the start of the shared
* extent so that a subsequent iteration starts at the
* start of the shared region.
*/
irec->br_blockcount = fbno - agbno;
return 0;
}
}
bool
xfs_inode_need_cow(
struct xfs_inode *ip,
struct xfs_bmbt_irec *imap,
bool *shared)
{
/* We can't update any real extents in always COW mode. */
if (xfs_is_always_cow_inode(ip) &&
!isnullstartblock(imap->br_startblock)) {
*shared = true;
return 0;
}
/* Trim the mapping to the nearest shared extent boundary. */
return xfs_reflink_trim_around_shared(ip, imap, shared);
}
static int
xfs_reflink_convert_cow_locked(
struct xfs_inode *ip,
xfs_fileoff_t offset_fsb,
xfs_filblks_t count_fsb)
{
struct xfs_iext_cursor icur;
struct xfs_bmbt_irec got;
struct xfs_btree_cur *dummy_cur = NULL;
int dummy_logflags;
int error = 0;
if (!xfs_iext_lookup_extent(ip, ip->i_cowfp, offset_fsb, &icur, &got))
return 0;
do {
if (got.br_startoff >= offset_fsb + count_fsb)
break;
if (got.br_state == XFS_EXT_NORM)
continue;
if (WARN_ON_ONCE(isnullstartblock(got.br_startblock)))
return -EIO;
xfs_trim_extent(&got, offset_fsb, count_fsb);
if (!got.br_blockcount)
continue;
got.br_state = XFS_EXT_NORM;
error = xfs_bmap_add_extent_unwritten_real(NULL, ip,
XFS_COW_FORK, &icur, &dummy_cur, &got,
&dummy_logflags);
if (error)
return error;
} while (xfs_iext_next_extent(ip->i_cowfp, &icur, &got));
return error;
}
/* Convert all of the unwritten CoW extents in a file's range to real ones. */
int
xfs_reflink_convert_cow(
struct xfs_inode *ip,
xfs_off_t offset,
xfs_off_t count)
{
struct xfs_mount *mp = ip->i_mount;
xfs_fileoff_t offset_fsb = XFS_B_TO_FSBT(mp, offset);
xfs_fileoff_t end_fsb = XFS_B_TO_FSB(mp, offset + count);
xfs_filblks_t count_fsb = end_fsb - offset_fsb;
int error;
ASSERT(count != 0);
xfs_ilock(ip, XFS_ILOCK_EXCL);
error = xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return error;
}
/*
* Find the extent that maps the given range in the COW fork. Even if the extent
* is not shared we might have a preallocation for it in the COW fork. If so we
* use it that rather than trigger a new allocation.
*/
static int
xfs_find_trim_cow_extent(
struct xfs_inode *ip,
struct xfs_bmbt_irec *imap,
bool *shared,
bool *found)
{
xfs_fileoff_t offset_fsb = imap->br_startoff;
xfs_filblks_t count_fsb = imap->br_blockcount;
struct xfs_iext_cursor icur;
struct xfs_bmbt_irec got;
*found = false;
/*
* If we don't find an overlapping extent, trim the range we need to
* allocate to fit the hole we found.
*/
if (!xfs_iext_lookup_extent(ip, ip->i_cowfp, offset_fsb, &icur, &got))
got.br_startoff = offset_fsb + count_fsb;
if (got.br_startoff > offset_fsb) {
xfs_trim_extent(imap, imap->br_startoff,
got.br_startoff - imap->br_startoff);
return xfs_inode_need_cow(ip, imap, shared);
}
*shared = true;
if (isnullstartblock(got.br_startblock)) {
xfs_trim_extent(imap, got.br_startoff, got.br_blockcount);
return 0;
}
/* real extent found - no need to allocate */
xfs_trim_extent(&got, offset_fsb, count_fsb);
*imap = got;
*found = true;
return 0;
}
/* Allocate all CoW reservations covering a range of blocks in a file. */
int
xfs_reflink_allocate_cow(
struct xfs_inode *ip,
struct xfs_bmbt_irec *imap,
bool *shared,
uint *lockmode,
bool convert_now)
{
struct xfs_mount *mp = ip->i_mount;
xfs_fileoff_t offset_fsb = imap->br_startoff;
xfs_filblks_t count_fsb = imap->br_blockcount;
struct xfs_trans *tp;
int nimaps, error = 0;
bool found;
xfs_filblks_t resaligned;
xfs_extlen_t resblks = 0;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
if (!ip->i_cowfp) {
ASSERT(!xfs_is_reflink_inode(ip));
xfs_ifork_init_cow(ip);
}
error = xfs_find_trim_cow_extent(ip, imap, shared, &found);
if (error || !*shared)
return error;
if (found)
goto convert;
resaligned = xfs_aligned_fsb_count(imap->br_startoff,
imap->br_blockcount, xfs_get_cowextsz_hint(ip));
resblks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
xfs_iunlock(ip, *lockmode);
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, 0, 0, &tp);
*lockmode = XFS_ILOCK_EXCL;
xfs_ilock(ip, *lockmode);
if (error)
return error;
error = xfs_qm_dqattach_locked(ip, false);
if (error)
goto out_trans_cancel;
/*
* Check for an overlapping extent again now that we dropped the ilock.
*/
error = xfs_find_trim_cow_extent(ip, imap, shared, &found);
if (error || !*shared)
goto out_trans_cancel;
if (found) {
xfs_trans_cancel(tp);
goto convert;
}
error = xfs_trans_reserve_quota_nblks(tp, ip, resblks, 0,
XFS_QMOPT_RES_REGBLKS);
if (error)
goto out_trans_cancel;
xfs_trans_ijoin(tp, ip, 0);
/* Allocate the entire reservation as unwritten blocks. */
nimaps = 1;
error = xfs_bmapi_write(tp, ip, imap->br_startoff, imap->br_blockcount,
XFS_BMAPI_COWFORK | XFS_BMAPI_PREALLOC,
resblks, imap, &nimaps);
if (error)
goto out_unreserve;
xfs_inode_set_cowblocks_tag(ip);
error = xfs_trans_commit(tp);
if (error)
return error;
/*
* Allocation succeeded but the requested range was not even partially
* satisfied? Bail out!
*/
if (nimaps == 0)
return -ENOSPC;
convert:
xfs_trim_extent(imap, offset_fsb, count_fsb);
/*
* COW fork extents are supposed to remain unwritten until we're ready
* to initiate a disk write. For direct I/O we are going to write the
* data and need the conversion, but for buffered writes we're done.
*/
if (!convert_now || imap->br_state == XFS_EXT_NORM)
return 0;
trace_xfs_reflink_convert_cow(ip, imap);
return xfs_reflink_convert_cow_locked(ip, offset_fsb, count_fsb);
out_unreserve:
xfs_trans_unreserve_quota_nblks(tp, ip, (long)resblks, 0,
XFS_QMOPT_RES_REGBLKS);
out_trans_cancel:
xfs_trans_cancel(tp);
return error;
}
/*
* Cancel CoW reservations for some block range of an inode.
*
* If cancel_real is true this function cancels all COW fork extents for the
* inode; if cancel_real is false, real extents are not cleared.
*
* Caller must have already joined the inode to the current transaction. The
* inode will be joined to the transaction returned to the caller.
*/
int
xfs_reflink_cancel_cow_blocks(
struct xfs_inode *ip,
struct xfs_trans **tpp,
xfs_fileoff_t offset_fsb,
xfs_fileoff_t end_fsb,
bool cancel_real)
{
struct xfs_ifork *ifp = XFS_IFORK_PTR(ip, XFS_COW_FORK);
struct xfs_bmbt_irec got, del;
struct xfs_iext_cursor icur;
int error = 0;
if (!xfs_inode_has_cow_data(ip))
return 0;
if (!xfs_iext_lookup_extent_before(ip, ifp, &end_fsb, &icur, &got))
return 0;
/* Walk backwards until we're out of the I/O range... */
while (got.br_startoff + got.br_blockcount > offset_fsb) {
del = got;
xfs_trim_extent(&del, offset_fsb, end_fsb - offset_fsb);
/* Extent delete may have bumped ext forward */
if (!del.br_blockcount) {
xfs_iext_prev(ifp, &icur);
goto next_extent;
}
trace_xfs_reflink_cancel_cow(ip, &del);
if (isnullstartblock(del.br_startblock)) {
error = xfs_bmap_del_extent_delay(ip, XFS_COW_FORK,
&icur, &got, &del);
if (error)
break;
} else if (del.br_state == XFS_EXT_UNWRITTEN || cancel_real) {
ASSERT((*tpp)->t_firstblock == NULLFSBLOCK);
/* Free the CoW orphan record. */
xfs_refcount_free_cow_extent(*tpp, del.br_startblock,
del.br_blockcount);
xfs_bmap_add_free(*tpp, del.br_startblock,
del.br_blockcount, NULL);
/* Roll the transaction */
error = xfs_defer_finish(tpp);
if (error)
break;
/* Remove the mapping from the CoW fork. */
xfs_bmap_del_extent_cow(ip, &icur, &got, &del);
/* Remove the quota reservation */
error = xfs_trans_reserve_quota_nblks(NULL, ip,
-(long)del.br_blockcount, 0,
XFS_QMOPT_RES_REGBLKS);
if (error)
break;
} else {
/* Didn't do anything, push cursor back. */
xfs_iext_prev(ifp, &icur);
}
next_extent:
if (!xfs_iext_get_extent(ifp, &icur, &got))
break;
}
/* clear tag if cow fork is emptied */
if (!ifp->if_bytes)
xfs_inode_clear_cowblocks_tag(ip);
return error;
}
/*
* Cancel CoW reservations for some byte range of an inode.
*
* If cancel_real is true this function cancels all COW fork extents for the
* inode; if cancel_real is false, real extents are not cleared.
*/
int
xfs_reflink_cancel_cow_range(
struct xfs_inode *ip,
xfs_off_t offset,
xfs_off_t count,
bool cancel_real)
{
struct xfs_trans *tp;
xfs_fileoff_t offset_fsb;
xfs_fileoff_t end_fsb;
int error;
trace_xfs_reflink_cancel_cow_range(ip, offset, count);
ASSERT(ip->i_cowfp);
offset_fsb = XFS_B_TO_FSBT(ip->i_mount, offset);
if (count == NULLFILEOFF)
end_fsb = NULLFILEOFF;
else
end_fsb = XFS_B_TO_FSB(ip->i_mount, offset + count);
/* Start a rolling transaction to remove the mappings */
error = xfs_trans_alloc(ip->i_mount, &M_RES(ip->i_mount)->tr_write,
0, 0, 0, &tp);
if (error)
goto out;
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, 0);
/* Scrape out the old CoW reservations */
error = xfs_reflink_cancel_cow_blocks(ip, &tp, offset_fsb, end_fsb,
cancel_real);
if (error)
goto out_cancel;
error = xfs_trans_commit(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return error;
out_cancel:
xfs_trans_cancel(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
out:
trace_xfs_reflink_cancel_cow_range_error(ip, error, _RET_IP_);
return error;
}
/*
* Remap part of the CoW fork into the data fork.
*
* We aim to remap the range starting at @offset_fsb and ending at @end_fsb
* into the data fork; this function will remap what it can (at the end of the
* range) and update @end_fsb appropriately. Each remap gets its own
* transaction because we can end up merging and splitting bmbt blocks for
* every remap operation and we'd like to keep the block reservation
* requirements as low as possible.
*/
STATIC int
xfs_reflink_end_cow_extent(
struct xfs_inode *ip,
xfs_fileoff_t offset_fsb,
xfs_fileoff_t *end_fsb)
{
struct xfs_bmbt_irec got, del;
struct xfs_iext_cursor icur;
struct xfs_mount *mp = ip->i_mount;
struct xfs_trans *tp;
struct xfs_ifork *ifp = XFS_IFORK_PTR(ip, XFS_COW_FORK);
xfs_filblks_t rlen;
unsigned int resblks;
int error;
/* No COW extents? That's easy! */
if (ifp->if_bytes == 0) {
*end_fsb = offset_fsb;
return 0;
}
resblks = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, 0,
XFS_TRANS_RESERVE, &tp);
if (error)
return error;
/*
* Lock the inode. We have to ijoin without automatic unlock because
* the lead transaction is the refcountbt record deletion; the data
* fork update follows as a deferred log item.
*/
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, 0);
/*
* In case of racing, overlapping AIO writes no COW extents might be
* left by the time I/O completes for the loser of the race. In that
* case we are done.
*/
if (!xfs_iext_lookup_extent_before(ip, ifp, end_fsb, &icur, &got) ||
got.br_startoff + got.br_blockcount <= offset_fsb) {
*end_fsb = offset_fsb;
goto out_cancel;
}
/*
* Structure copy @got into @del, then trim @del to the range that we
* were asked to remap. We preserve @got for the eventual CoW fork
* deletion; from now on @del represents the mapping that we're
* actually remapping.
*/
del = got;
xfs_trim_extent(&del, offset_fsb, *end_fsb - offset_fsb);
ASSERT(del.br_blockcount > 0);
/*
* Only remap real extents that contain data. With AIO, speculative
* preallocations can leak into the range we are called upon, and we
* need to skip them.
*/
if (!xfs_bmap_is_real_extent(&got)) {
*end_fsb = del.br_startoff;
goto out_cancel;
}
/* Unmap the old blocks in the data fork. */
rlen = del.br_blockcount;
error = __xfs_bunmapi(tp, ip, del.br_startoff, &rlen, 0, 1);
if (error)
goto out_cancel;
/* Trim the extent to whatever got unmapped. */
xfs_trim_extent(&del, del.br_startoff + rlen, del.br_blockcount - rlen);
trace_xfs_reflink_cow_remap(ip, &del);
/* Free the CoW orphan record. */
xfs_refcount_free_cow_extent(tp, del.br_startblock, del.br_blockcount);
/* Map the new blocks into the data fork. */
xfs_bmap_map_extent(tp, ip, &del);
/* Charge this new data fork mapping to the on-disk quota. */
xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_DELBCOUNT,
(long)del.br_blockcount);
/* Remove the mapping from the CoW fork. */
xfs_bmap_del_extent_cow(ip, &icur, &got, &del);
error = xfs_trans_commit(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
if (error)
return error;
/* Update the caller about how much progress we made. */
*end_fsb = del.br_startoff;
return 0;
out_cancel:
xfs_trans_cancel(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return error;
}
/*
* Remap parts of a file's data fork after a successful CoW.
*/
int
xfs_reflink_end_cow(
struct xfs_inode *ip,
xfs_off_t offset,
xfs_off_t count)
{
xfs_fileoff_t offset_fsb;
xfs_fileoff_t end_fsb;
int error = 0;
trace_xfs_reflink_end_cow(ip, offset, count);
offset_fsb = XFS_B_TO_FSBT(ip->i_mount, offset);
end_fsb = XFS_B_TO_FSB(ip->i_mount, offset + count);
/*
* Walk backwards until we're out of the I/O range. The loop function
* repeatedly cycles the ILOCK to allocate one transaction per remapped
* extent.
*
* If we're being called by writeback then the the pages will still
* have PageWriteback set, which prevents races with reflink remapping
* and truncate. Reflink remapping prevents races with writeback by
* taking the iolock and mmaplock before flushing the pages and
* remapping, which means there won't be any further writeback or page
* cache dirtying until the reflink completes.
*
* We should never have two threads issuing writeback for the same file
* region. There are also have post-eof checks in the writeback
* preparation code so that we don't bother writing out pages that are
* about to be truncated.
*
* If we're being called as part of directio write completion, the dio
* count is still elevated, which reflink and truncate will wait for.
* Reflink remapping takes the iolock and mmaplock and waits for
* pending dio to finish, which should prevent any directio until the
* remap completes. Multiple concurrent directio writes to the same
* region are handled by end_cow processing only occurring for the
* threads which succeed; the outcome of multiple overlapping direct
* writes is not well defined anyway.
*
* It's possible that a buffered write and a direct write could collide
* here (the buffered write stumbles in after the dio flushes and
* invalidates the page cache and immediately queues writeback), but we
* have never supported this 100%. If either disk write succeeds the
* blocks will be remapped.
*/
while (end_fsb > offset_fsb && !error)
error = xfs_reflink_end_cow_extent(ip, offset_fsb, &end_fsb);
if (error)
trace_xfs_reflink_end_cow_error(ip, error, _RET_IP_);
return error;
}
/*
* Free leftover CoW reservations that didn't get cleaned out.
*/
int
xfs_reflink_recover_cow(
struct xfs_mount *mp)
{
xfs_agnumber_t agno;
int error = 0;
if (!xfs_sb_version_hasreflink(&mp->m_sb))
return 0;
for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
error = xfs_refcount_recover_cow_leftovers(mp, agno);
if (error)
break;
}
return error;
}
/*
* Reflinking (Block) Ranges of Two Files Together
*
* First, ensure that the reflink flag is set on both inodes. The flag is an
* optimization to avoid unnecessary refcount btree lookups in the write path.
*
* Now we can iteratively remap the range of extents (and holes) in src to the
* corresponding ranges in dest. Let drange and srange denote the ranges of
* logical blocks in dest and src touched by the reflink operation.
*
* While the length of drange is greater than zero,
* - Read src's bmbt at the start of srange ("imap")
* - If imap doesn't exist, make imap appear to start at the end of srange
* with zero length.
* - If imap starts before srange, advance imap to start at srange.
* - If imap goes beyond srange, truncate imap to end at the end of srange.
* - Punch (imap start - srange start + imap len) blocks from dest at
* offset (drange start).
* - If imap points to a real range of pblks,
* > Increase the refcount of the imap's pblks
* > Map imap's pblks into dest at the offset
* (drange start + imap start - srange start)
* - Advance drange and srange by (imap start - srange start + imap len)
*
* Finally, if the reflink made dest longer, update both the in-core and
* on-disk file sizes.
*
* ASCII Art Demonstration:
*
* Let's say we want to reflink this source file:
*
* ----SSSSSSS-SSSSS----SSSSSS (src file)
* <-------------------->
*
* into this destination file:
*
* --DDDDDDDDDDDDDDDDDDD--DDD (dest file)
* <-------------------->
* '-' means a hole, and 'S' and 'D' are written blocks in the src and dest.
* Observe that the range has different logical offsets in either file.
*
* Consider that the first extent in the source file doesn't line up with our
* reflink range. Unmapping and remapping are separate operations, so we can
* unmap more blocks from the destination file than we remap.
*
* ----SSSSSSS-SSSSS----SSSSSS
* <------->
* --DDDDD---------DDDDD--DDD
* <------->
*
* Now remap the source extent into the destination file:
*
* ----SSSSSSS-SSSSS----SSSSSS
* <------->
* --DDDDD--SSSSSSSDDDDD--DDD
* <------->
*
* Do likewise with the second hole and extent in our range. Holes in the
* unmap range don't affect our operation.
*
* ----SSSSSSS-SSSSS----SSSSSS
* <---->
* --DDDDD--SSSSSSS-SSSSS-DDD
* <---->
*
* Finally, unmap and remap part of the third extent. This will increase the
* size of the destination file.
*
* ----SSSSSSS-SSSSS----SSSSSS
* <----->
* --DDDDD--SSSSSSS-SSSSS----SSS
* <----->
*
* Once we update the destination file's i_size, we're done.
*/
/*
* Ensure the reflink bit is set in both inodes.
*/
STATIC int
xfs_reflink_set_inode_flag(
struct xfs_inode *src,
struct xfs_inode *dest)
{
struct xfs_mount *mp = src->i_mount;
int error;
struct xfs_trans *tp;
if (xfs_is_reflink_inode(src) && xfs_is_reflink_inode(dest))
return 0;
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ichange, 0, 0, 0, &tp);
if (error)
goto out_error;
/* Lock both files against IO */
if (src->i_ino == dest->i_ino)
xfs_ilock(src, XFS_ILOCK_EXCL);
else
xfs_lock_two_inodes(src, XFS_ILOCK_EXCL, dest, XFS_ILOCK_EXCL);
if (!xfs_is_reflink_inode(src)) {
trace_xfs_reflink_set_inode_flag(src);
xfs_trans_ijoin(tp, src, XFS_ILOCK_EXCL);
src->i_d.di_flags2 |= XFS_DIFLAG2_REFLINK;
xfs_trans_log_inode(tp, src, XFS_ILOG_CORE);
xfs_ifork_init_cow(src);
} else
xfs_iunlock(src, XFS_ILOCK_EXCL);
if (src->i_ino == dest->i_ino)
goto commit_flags;
if (!xfs_is_reflink_inode(dest)) {
trace_xfs_reflink_set_inode_flag(dest);
xfs_trans_ijoin(tp, dest, XFS_ILOCK_EXCL);
dest->i_d.di_flags2 |= XFS_DIFLAG2_REFLINK;
xfs_trans_log_inode(tp, dest, XFS_ILOG_CORE);
xfs_ifork_init_cow(dest);
} else
xfs_iunlock(dest, XFS_ILOCK_EXCL);
commit_flags:
error = xfs_trans_commit(tp);
if (error)
goto out_error;
return error;
out_error:
trace_xfs_reflink_set_inode_flag_error(dest, error, _RET_IP_);
return error;
}
/*
* Update destination inode size & cowextsize hint, if necessary.
*/
int
xfs_reflink_update_dest(
struct xfs_inode *dest,
xfs_off_t newlen,
xfs_extlen_t cowextsize,
unsigned int remap_flags)
{
struct xfs_mount *mp = dest->i_mount;
struct xfs_trans *tp;
int error;
if (newlen <= i_size_read(VFS_I(dest)) && cowextsize == 0)
return 0;
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ichange, 0, 0, 0, &tp);
if (error)
goto out_error;
xfs_ilock(dest, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, dest, XFS_ILOCK_EXCL);
if (newlen > i_size_read(VFS_I(dest))) {
trace_xfs_reflink_update_inode_size(dest, newlen);
i_size_write(VFS_I(dest), newlen);
dest->i_d.di_size = newlen;
}
if (cowextsize) {
dest->i_d.di_cowextsize = cowextsize;
dest->i_d.di_flags2 |= XFS_DIFLAG2_COWEXTSIZE;
}
xfs_trans_log_inode(tp, dest, XFS_ILOG_CORE);
error = xfs_trans_commit(tp);
if (error)
goto out_error;
return error;
out_error:
trace_xfs_reflink_update_inode_size_error(dest, error, _RET_IP_);
return error;
}
/*
* Do we have enough reserve in this AG to handle a reflink? The refcount
* btree already reserved all the space it needs, but the rmap btree can grow
* infinitely, so we won't allow more reflinks when the AG is down to the
* btree reserves.
*/
static int
xfs_reflink_ag_has_free_space(
struct xfs_mount *mp,
xfs_agnumber_t agno)
{
struct xfs_perag *pag;
int error = 0;
if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
return 0;
pag = xfs_perag_get(mp, agno);
if (xfs_ag_resv_critical(pag, XFS_AG_RESV_RMAPBT) ||
xfs_ag_resv_critical(pag, XFS_AG_RESV_METADATA))
error = -ENOSPC;
xfs_perag_put(pag);
return error;
}
/*
* Unmap a range of blocks from a file, then map other blocks into the hole.
* The range to unmap is (destoff : destoff + srcioff + irec->br_blockcount).
* The extent irec is mapped into dest at irec->br_startoff.
*/
STATIC int
xfs_reflink_remap_extent(
struct xfs_inode *ip,
struct xfs_bmbt_irec *irec,
xfs_fileoff_t destoff,
xfs_off_t new_isize)
{
struct xfs_mount *mp = ip->i_mount;