-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathfunctions.c
1954 lines (1649 loc) · 45.3 KB
/
functions.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
/*
* Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
*
* This copyrighted material is made available to anyone wishing to use,
* modify, copy, or redistribute it subject to the terms and conditions
* of the GNU Lesser General Public License v.2.1.
*
* You should have received a copy of the GNU Lesser General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include "logging.h"
#include "functions.h"
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <signal.h>
#include <sys/stat.h>
#include <time.h>
#include <unistd.h>
#define BYTE_SHIFT 3
/*
* Magic for persistent mirrors: "MiRr"
* Following on-disk header information is stolen from
* drivers/md/dm-log.c
*/
#define MIRROR_MAGIC 0x4D695272
#define MIRROR_DISK_VERSION 2
#define LOG_OFFSET 2
#define RESYNC_HISTORY 50
//static char resync_history[RESYNC_HISTORY][128];
//static int idx = 0;
#define LOG_SPRINT(_lc, f, arg...) do { \
lc->idx++; \
lc->idx = lc->idx % RESYNC_HISTORY; \
sprintf(lc->resync_history[lc->idx], f, ## arg); \
} while (0)
struct log_header {
uint32_t magic;
uint32_t version;
uint64_t nr_regions;
};
struct log_c {
struct dm_list list;
char uuid[DM_UUID_LEN];
uint64_t luid;
time_t delay; /* limits how fast a resume can happen after suspend */
int touched;
int in_sync; /* An in-sync that stays set until suspend/resume */
uint32_t region_size;
uint32_t region_count;
uint64_t sync_count;
dm_bitset_t clean_bits;
dm_bitset_t sync_bits;
uint32_t recoverer;
uint64_t recovering_region; /* -1 means not recovering */
uint64_t skip_bit_warning; /* used to warn if region skipped */
int sync_search;
int resume_override;
uint32_t block_on_error;
enum sync {
DEFAULTSYNC, /* Synchronize if necessary */
NOSYNC, /* Devices known to be already in sync */
FORCESYNC, /* Force a sync to happen */
} sync;
uint32_t state; /* current operational state of the log */
struct dm_list mark_list;
uint32_t recovery_halted;
struct recovery_request *recovery_request_list;
int disk_fd; /* -1 means no disk log */
int log_dev_failed;
uint64_t disk_nr_regions;
size_t disk_size; /* size of disk_buffer in bytes */
void *disk_buffer; /* aligned memory for O_DIRECT */
int idx;
char resync_history[RESYNC_HISTORY][128];
};
struct mark_entry {
struct dm_list list;
uint32_t nodeid;
uint64_t region;
};
struct recovery_request {
uint64_t region;
struct recovery_request *next;
};
static DM_LIST_INIT(log_list);
static DM_LIST_INIT(log_pending_list);
static int log_test_bit(dm_bitset_t bs, int bit)
{
return dm_bit(bs, bit) ? 1 : 0;
}
static void log_set_bit(struct log_c *lc, dm_bitset_t bs, int bit)
{
dm_bit_set(bs, bit);
lc->touched = 1;
}
static void log_clear_bit(struct log_c *lc, dm_bitset_t bs, int bit)
{
dm_bit_clear(bs, bit);
lc->touched = 1;
}
static uint64_t find_next_zero_bit(dm_bitset_t bs, unsigned start)
{
for (; dm_bit(bs, start); start++)
if (start >= *bs)
return (uint64_t)-1;
return start;
}
static uint64_t count_bits32(dm_bitset_t bs)
{
unsigned i, size = bs[0]/(unsigned)DM_BITS_PER_INT + 1;
unsigned count = 0;
for (i = 1; i <= size; i++)
count += hweight32(bs[i]);
return (uint64_t)count;
}
/*
* get_log
*
* Returns: log if found, NULL otherwise
*/
static struct log_c *get_log(const char *uuid, uint64_t luid)
{
struct log_c *lc;
dm_list_iterate_items(lc, &log_list)
if (!strcmp(lc->uuid, uuid) &&
(!luid || (luid == lc->luid)))
return lc;
return NULL;
}
/*
* get_pending_log
*
* Pending logs are logs that have been 'clog_ctr'ed, but
* have not joined the CPG (via clog_resume).
*
* Returns: log if found, NULL otherwise
*/
static struct log_c *get_pending_log(const char *uuid, uint64_t luid)
{
struct log_c *lc;
dm_list_iterate_items(lc, &log_pending_list)
if (!strcmp(lc->uuid, uuid) &&
(!luid || (luid == lc->luid)))
return lc;
return NULL;
}
static void header_to_disk(struct log_header *mem, struct log_header *disk)
{
memcpy(disk, mem, sizeof(struct log_header));
}
static void header_from_disk(struct log_header *mem, struct log_header *disk)
{
memcpy(mem, disk, sizeof(struct log_header));
}
static int rw_log(struct log_c *lc, int do_write)
{
int r;
r = (int)lseek(lc->disk_fd, 0, SEEK_SET);
if (r < 0) {
LOG_ERROR("[%s] rw_log: lseek failure: %s",
SHORT_UUID(lc->uuid), strerror(errno));
return -errno;
}
if (do_write) {
/* FIXME Cope with full set of non-error conditions */
r = write(lc->disk_fd, lc->disk_buffer, lc->disk_size);
if (r < 0) {
LOG_ERROR("[%s] rw_log: write failure: %s",
SHORT_UUID(lc->uuid), strerror(errno));
return -EIO; /* Failed disk write */
}
return 0;
}
/* Read */
/* FIXME Cope with full set of non-error conditions */
r = read(lc->disk_fd, lc->disk_buffer, lc->disk_size);
if (r < 0)
LOG_ERROR("[%s] rw_log: read failure: %s",
SHORT_UUID(lc->uuid), strerror(errno));
if (r != lc->disk_size)
return -EIO; /* Failed disk read */
return 0;
}
/*
* read_log
* @lc
*
* Valid return codes:
* -EINVAL: Invalid header, bits not copied
* -EIO: Unable to read disk log
* 0: Valid header, disk bit -> lc->clean_bits
*
* Returns: 0 on success, -EXXX on failure
*/
static int read_log(struct log_c *lc)
{
struct log_header lh;
size_t bitset_size;
memset(&lh, 0, sizeof(struct log_header));
if (rw_log(lc, 0))
return -EIO; /* Failed disk read */
header_from_disk(&lh, lc->disk_buffer);
if (lh.magic != MIRROR_MAGIC)
return -EINVAL;
lc->disk_nr_regions = lh.nr_regions;
/* Read disk bits into sync_bits */
bitset_size = lc->region_count / 8;
bitset_size += (lc->region_count % 8) ? 1 : 0;
/* 'lc->clean_bits + 1' becasue dm_bitset_t leads with a uint32_t */
memcpy(lc->clean_bits + 1, (char *)lc->disk_buffer + 1024, bitset_size);
return 0;
}
/*
* write_log
* @lc
*
* Returns: 0 on success, -EIO on failure
*/
static int write_log(struct log_c *lc)
{
struct log_header lh;
size_t bitset_size;
lh.magic = MIRROR_MAGIC;
lh.version = MIRROR_DISK_VERSION;
lh.nr_regions = lc->region_count;
header_to_disk(&lh, lc->disk_buffer);
/* Write disk bits from clean_bits */
bitset_size = lc->region_count / 8;
bitset_size += (lc->region_count % 8) ? 1 : 0;
/* 'lc->clean_bits + 1' becasue dm_bitset_t leads with a uint32_t */
memcpy((char *)lc->disk_buffer + 1024, lc->clean_bits + 1, bitset_size);
if (rw_log(lc, 1)) {
lc->log_dev_failed = 1;
return -EIO; /* Failed disk write */
}
return 0;
}
/* FIXME Rewrite this function taking advantage of the udev changes (where in use) to improve its efficiency! */
static int find_disk_path(char *major_minor_str, char *path_rtn, int *unlink_path __attribute__((unused)))
{
int r;
DIR *dp;
struct dirent *dep;
struct stat statbuf;
int major, minor;
if (!strstr(major_minor_str, ":")) {
r = stat(major_minor_str, &statbuf);
if (r)
return -errno;
if (!S_ISBLK(statbuf.st_mode))
return -EINVAL;
sprintf(path_rtn, "%s", major_minor_str);
return 0;
}
r = sscanf(major_minor_str, "%d:%d", &major, &minor);
if (r != 2)
return -EINVAL;
/* FIXME dm_dir() */
LOG_DBG("Checking /dev/mapper for device %d:%d", major, minor);
/* Check /dev/mapper dir */
dp = opendir("/dev/mapper");
if (!dp)
return -ENOENT;
while ((dep = readdir(dp)) != NULL) {
/*
* FIXME: This is racy. By the time the path is used,
* it may point to something else. 'fstat' will be
* required upon opening to ensure we got what we
* wanted.
*/
sprintf(path_rtn, "/dev/mapper/%s", dep->d_name);
if (stat(path_rtn, &statbuf) < 0) {
LOG_DBG("Unable to stat %s", path_rtn);
continue;
}
if (S_ISBLK(statbuf.st_mode) &&
(major(statbuf.st_rdev) == major) &&
(minor(statbuf.st_rdev) == minor)) {
LOG_DBG(" %s: YES", dep->d_name);
closedir(dp);
return 0;
} else {
LOG_DBG(" %s: NO", dep->d_name);
}
}
closedir(dp);
/* FIXME Find out why this was here and deal with underlying problem. */
LOG_DBG("Path not found for %d/%d", major, minor);
return -ENOENT;
// LOG_DBG("Creating /dev/mapper/%d-%d", major, minor);
// sprintf(path_rtn, "/dev/mapper/%d-%d", major, minor);
// r = mknod(path_rtn, S_IFBLK | S_IRUSR | S_IWUSR, MKDEV(major, minor));
/*
* If we have to make the path, we unlink it after we open it
*/
// *unlink_path = 1;
// return r ? -errno : 0;
}
static int _clog_ctr(char *uuid, uint64_t luid,
int argc, char **argv, uint64_t device_size)
{
int i;
int r = 0;
char *p;
uint64_t region_size;
uint64_t region_count;
struct log_c *lc = NULL;
enum sync log_sync = DEFAULTSYNC;
uint32_t block_on_error = 0;
int disk_log;
char disk_path[128];
int unlink_path = 0;
long page_size;
int pages;
/* If core log request, then argv[0] will be region_size */
if (!strtoll(argv[0], &p, 0) || *p) {
disk_log = 1;
if ((argc < 2) || (argc > 4)) {
LOG_ERROR("Too %s arguments to clustered-disk log type",
(argc < 3) ? "few" : "many");
r = -EINVAL;
goto fail;
}
r = find_disk_path(argv[0], disk_path, &unlink_path);
if (r) {
LOG_ERROR("Unable to find path to device %s", argv[0]);
goto fail;
}
LOG_DBG("Clustered log disk is %s", disk_path);
} else {
disk_log = 0;
if ((argc < 1) || (argc > 3)) {
LOG_ERROR("Too %s arguments to clustered-core log type",
(argc < 2) ? "few" : "many");
r = -EINVAL;
goto fail;
}
}
if (!(region_size = strtoll(argv[disk_log], &p, 0)) || *p) {
LOG_ERROR("Invalid region_size argument to clustered-%s log type",
(disk_log) ? "disk" : "core");
r = -EINVAL;
goto fail;
}
region_count = device_size / region_size;
if (device_size % region_size) {
/*
* I can't remember if device_size must be a multiple
* of region_size, so check it anyway.
*/
region_count++;
}
for (i = 0; i < argc; i++) {
if (!strcmp(argv[i], "sync"))
log_sync = FORCESYNC;
else if (!strcmp(argv[i], "nosync"))
log_sync = NOSYNC;
else if (!strcmp(argv[i], "block_on_error"))
block_on_error = 1;
}
lc = dm_zalloc(sizeof(*lc));
if (!lc) {
LOG_ERROR("Unable to allocate cluster log context");
r = -ENOMEM;
goto fail;
}
lc->region_size = region_size;
lc->region_count = region_count;
lc->sync = log_sync;
lc->block_on_error = block_on_error;
lc->sync_search = 0;
lc->recovering_region = (uint64_t)-1;
lc->skip_bit_warning = region_count;
lc->disk_fd = -1;
lc->log_dev_failed = 0;
strncpy(lc->uuid, uuid, DM_UUID_LEN);
lc->luid = luid;
if (get_log(lc->uuid, lc->luid) ||
get_pending_log(lc->uuid, lc->luid)) {
LOG_ERROR("[%s/%" PRIu64 "u] Log already exists, unable to create.",
SHORT_UUID(lc->uuid), lc->luid);
dm_free(lc);
return -EINVAL;
}
dm_list_init(&lc->mark_list);
lc->clean_bits = dm_bitset_create(NULL, region_count);
if (!lc->clean_bits) {
LOG_ERROR("Unable to allocate clean bitset");
r = -ENOMEM;
goto fail;
}
lc->sync_bits = dm_bitset_create(NULL, region_count);
if (!lc->sync_bits) {
LOG_ERROR("Unable to allocate sync bitset");
r = -ENOMEM;
goto fail;
}
if (log_sync == NOSYNC)
dm_bit_set_all(lc->sync_bits);
lc->sync_count = (log_sync == NOSYNC) ? region_count : 0;
if (disk_log) {
if ((page_size = sysconf(_SC_PAGESIZE)) < 0) {
LOG_ERROR("Unable to read pagesize: %s",
strerror(errno));
r = errno;
goto fail;
}
pages = *(lc->clean_bits) / page_size;
pages += *(lc->clean_bits) % page_size ? 1 : 0;
pages += 1; /* for header */
r = open(disk_path, O_RDWR | O_DIRECT);
if (r < 0) {
LOG_ERROR("Unable to open log device, %s: %s",
disk_path, strerror(errno));
r = errno;
goto fail;
}
if (unlink_path)
if (unlink(disk_path) < 0) {
LOG_DBG("Warning: Unable to unlink log device, %s: %s",
disk_path, strerror(errno));
}
lc->disk_fd = r;
lc->disk_size = pages * page_size;
r = posix_memalign(&(lc->disk_buffer), page_size,
lc->disk_size);
if (r) {
LOG_ERROR("Unable to allocate memory for disk_buffer");
goto fail;
}
memset(lc->disk_buffer, 0, lc->disk_size);
LOG_DBG("Disk log ready");
}
dm_list_add(&log_pending_list, &lc->list);
return 0;
fail:
if (lc) {
if (lc->disk_fd >= 0 && close(lc->disk_fd))
LOG_ERROR("Close device error, %s: %s",
disk_path, strerror(errno));
free(lc->disk_buffer);
dm_free(lc->sync_bits);
dm_free(lc->clean_bits);
dm_free(lc);
}
return r;
}
/*
* clog_ctr
* @rq
*
* rq->data should contain constructor string as follows:
* <log_type> [disk] <region_size> [[no]sync] <device_len>
* The kernel is responsible for adding the <dev_len> argument
* to the end; otherwise, we cannot compute the region_count.
*
* FIXME: Currently relies on caller to fill in rq->error
*/
static int clog_dtr(struct dm_ulog_request *rq);
static int clog_ctr(struct dm_ulog_request *rq)
{
int argc, i, r = 0;
char *p, **argv = NULL;
char *dev_size_str;
uint64_t device_size;
/* Sanity checks */
if (!rq->data_size) {
LOG_ERROR("Received constructor request with no data");
return -EINVAL;
}
if (strlen(rq->data) > rq->data_size) {
LOG_ERROR("Received constructor request with bad data");
LOG_ERROR("strlen(rq->data)[%d] != rq->data_size[%llu]",
(int)strlen(rq->data),
(unsigned long long)rq->data_size);
LOG_ERROR("rq->data = '%s' [%d]",
rq->data, (int)strlen(rq->data));
return -EINVAL;
}
/* Split up args */
for (argc = 0, p = rq->data; (p = strstr(p, " ")); p++, argc++)
*p = '\0';
argv = malloc(argc * sizeof(char *));
if (!argv)
return -ENOMEM;
p = dev_size_str = rq->data;
p += strlen(p) + 1;
for (i = 0; i < argc; i++, p = p + strlen(p) + 1)
argv[i] = p;
if (strcmp(argv[0], "clustered-disk") &&
strcmp(argv[0], "clustered-core")) {
LOG_ERROR("Unsupported userspace log type, \"%s\"", argv[0]);
free(argv);
return -EINVAL;
}
if (!(device_size = strtoll(dev_size_str, &p, 0)) || *p) {
LOG_ERROR("Invalid device size argument: %s", dev_size_str);
free(argv);
return -EINVAL;
}
r = _clog_ctr(rq->uuid, rq->luid, argc - 1, argv + 1, device_size);
/* We join the CPG when we resume */
/* No returning data */
if ((rq->version > 1) && !strcmp(argv[0], "clustered-disk"))
rq->data_size = sprintf(rq->data, "%s", argv[1]) + 1;
else
rq->data_size = 0;
if (r) {
LOG_ERROR("Failed to create cluster log (%s)", rq->uuid);
for (i = 0; i < argc; i++)
LOG_ERROR("argv[%d] = %s", i, argv[i]);
}
else
LOG_DBG("[%s] Cluster log created",
SHORT_UUID(rq->uuid));
free(argv);
return r;
}
/*
* clog_dtr
* @rq
*
*/
static int clog_dtr(struct dm_ulog_request *rq)
{
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (lc) {
/*
* The log should not be on the official list. There
* should have been a suspend first.
*/
LOG_ERROR("[%s] DTR before SUS: leaving CPG",
SHORT_UUID(rq->uuid));
destroy_cluster_cpg(rq->uuid);
} else if (!(lc = get_pending_log(rq->uuid, rq->luid))) {
LOG_ERROR("clog_dtr called on log that is not official or pending");
return -EINVAL;
}
LOG_DBG("[%s] Cluster log removed", SHORT_UUID(lc->uuid));
dm_list_del(&lc->list);
if (lc->disk_fd != -1)
close(lc->disk_fd);
if (lc->disk_buffer)
free(lc->disk_buffer);
dm_free(lc->clean_bits);
dm_free(lc->sync_bits);
dm_free(lc);
return 0;
}
/*
* clog_presuspend
* @rq
*
*/
static int clog_presuspend(struct dm_ulog_request *rq)
{
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc)
return -EINVAL;
if (lc->touched)
LOG_DBG("WARNING: log still marked as 'touched' during suspend");
lc->recovery_halted = 1;
return 0;
}
/*
* clog_postsuspend
* @rq
*
*/
static int clog_postsuspend(struct dm_ulog_request *rq)
{
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc)
return -EINVAL;
LOG_DBG("[%s] clog_postsuspend: leaving CPG", SHORT_UUID(lc->uuid));
destroy_cluster_cpg(rq->uuid);
lc->state = LOG_SUSPENDED;
lc->recovering_region = (uint64_t)-1;
lc->recoverer = (uint32_t)-1;
lc->delay = time(NULL);
return 0;
}
/*
* cluster_postsuspend
* @rq
*
*/
int cluster_postsuspend(char *uuid, uint64_t luid)
{
struct log_c *lc = get_log(uuid, luid);
if (!lc)
return -EINVAL;
LOG_DBG("[%s] clog_postsuspend: finalizing", SHORT_UUID(lc->uuid));
lc->resume_override = 0;
/* move log to pending list */
dm_list_del(&lc->list);
dm_list_add(&log_pending_list, &lc->list);
return 0;
}
/*
* clog_resume
* @rq
*
* Does the main work of resuming.
*/
static int clog_resume(struct dm_ulog_request *rq)
{
uint32_t i;
int commit_log = 0;
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc)
return -EINVAL;
lc->in_sync = 0;
switch (lc->resume_override) {
case 1000:
LOG_ERROR("[%s] Additional resume issued before suspend",
SHORT_UUID(rq->uuid));
#ifdef DEBUG
kill(getpid(), SIGUSR1);
#endif
return 0;
case 0:
lc->resume_override = 1000;
if (lc->disk_fd == -1) {
LOG_DBG("[%s] Master resume.",
SHORT_UUID(lc->uuid));
goto no_disk;
}
LOG_DBG("[%s] Master resume: reading disk log",
SHORT_UUID(lc->uuid));
commit_log = 1;
break;
case 1:
LOG_ERROR("Error:: partial bit loading (just sync_bits)");
return -EINVAL;
case 2:
LOG_ERROR("Error:: partial bit loading (just clean_bits)");
return -EINVAL;
case 3:
LOG_DBG("[%s] Non-master resume: bits pre-loaded",
SHORT_UUID(lc->uuid));
lc->resume_override = 1000;
goto out;
default:
LOG_ERROR("Error:: multiple loading of bits (%d)",
lc->resume_override);
return -EINVAL;
}
if (lc->log_dev_failed) {
LOG_ERROR("Log device has failed, unable to read bits");
rq->error = 0; /* We can handle this so far */
lc->disk_nr_regions = 0;
} else
rq->error = read_log(lc);
switch (rq->error) {
case 0:
if (lc->disk_nr_regions < lc->region_count)
LOG_DBG("[%s] Mirror has grown, updating log bits",
SHORT_UUID(lc->uuid));
else if (lc->disk_nr_regions > lc->region_count)
LOG_DBG("[%s] Mirror has shrunk, updating log bits",
SHORT_UUID(lc->uuid));
break;
case -EINVAL:
LOG_DBG("[%s] (Re)initializing mirror log - resync issued.",
SHORT_UUID(lc->uuid));
lc->disk_nr_regions = 0;
break;
default:
LOG_ERROR("Failed to read disk log");
lc->disk_nr_regions = 0;
break;
}
no_disk:
/* If mirror has grown, set bits appropriately */
if (lc->sync == NOSYNC)
for (i = lc->disk_nr_regions; i < lc->region_count; i++)
log_set_bit(lc, lc->clean_bits, i);
else
for (i = lc->disk_nr_regions; i < lc->region_count; i++)
log_clear_bit(lc, lc->clean_bits, i);
/* Clear any old bits if device has shrunk */
for (i = lc->region_count; i % 32; i++)
log_clear_bit(lc, lc->clean_bits, i);
/* copy clean across to sync */
dm_bit_copy(lc->sync_bits, lc->clean_bits);
if (commit_log && (lc->disk_fd >= 0)) {
rq->error = write_log(lc);
if (rq->error)
LOG_ERROR("Failed initial disk log write");
else
LOG_DBG("Disk log initialized");
lc->touched = 0;
}
out:
/*
* Clear any old bits if device has shrunk - necessary
* for non-master resume
*/
for (i = lc->region_count; i % 32; i++) {
log_clear_bit(lc, lc->clean_bits, i);
log_clear_bit(lc, lc->sync_bits, i);
}
lc->sync_count = count_bits32(lc->sync_bits);
LOG_SPRINT(lc, "[%s] Initial sync_count = %llu",
SHORT_UUID(lc->uuid), (unsigned long long)lc->sync_count);
lc->sync_search = 0;
lc->state = LOG_RESUMED;
lc->recovery_halted = 0;
return rq->error;
}
/*
* local_resume
* @rq
*
* If the log is pending, we must first join the cpg and
* put the log in the official list.
*
*/
int local_resume(struct dm_ulog_request *rq)
{
int r;
time_t t;
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc) {
/* Is the log in the pending list? */
lc = get_pending_log(rq->uuid, rq->luid);
if (!lc) {
LOG_ERROR("clog_resume called on log that is not official or pending");
return -EINVAL;
}
t = time(NULL);
t -= lc->delay;
/*
* This should be considered a temporary fix. It addresses
* a problem that exists when nodes suspend/resume in rapid
* succession. While the problem is very rare, it has been
* seen to happen in real-world-like testing.
*
* The problem:
* - Node A joins cluster
* - Node B joins cluster
* - Node A prepares checkpoint
* - Node A gets ready to write checkpoint
* - Node B leaves
* - Node B joins
* - Node A finishes write of checkpoint
* - Node B receives checkpoint meant for previous session
* -- Node B can now be non-coherent
*
* This timer will solve the problem for now, but could be
* replaced by a generation number sent with the resume
* command from the kernel. The generation number would
* be included in the name of the checkpoint to prevent
* reading stale data.
*/
if ((t < 3) && (t >= 0))
sleep(3 - t);
/* Join the CPG */
r = create_cluster_cpg(rq->uuid, rq->luid);
if (r) {
LOG_ERROR("clog_resume: Failed to create cluster CPG");
return r;
}
/* move log to official list */
dm_list_del(&lc->list);
dm_list_add(&log_list, &lc->list);
}
return 0;
}
/*
* clog_get_region_size
* @rq
*
* Since this value doesn't change, the kernel
* should not need to talk to server to get this
* The function is here for completness
*
* Returns: 0 on success, -EXXX on failure
*/
static int clog_get_region_size(struct dm_ulog_request *rq)
{
uint64_t *rtn = (uint64_t *)rq->data;
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc && !(lc = get_pending_log(rq->uuid, rq->luid)))
return -EINVAL;
*rtn = lc->region_size;
rq->data_size = sizeof(*rtn);
return 0;
}
/*
* clog_is_clean
* @rq
*
* Returns: 1 if clean, 0 otherwise
*/
static int clog_is_clean(struct dm_ulog_request *rq)
{
int64_t *rtn = (int64_t *)rq->data;
uint64_t *region = (uint64_t *)rq->data;
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc)
return -EINVAL;
*rtn = log_test_bit(lc->clean_bits, *region);
rq->data_size = sizeof(*rtn);
return 0;
}
/*
* clog_in_sync
* @rq
*
* We ignore any request for non-block. That
* should be handled elsewhere. (If the request
* has come this far, it has already blocked.)
*
* Returns: 1 if in-sync, 0 otherwise
*/
static int clog_in_sync(struct dm_ulog_request *rq)
{
int64_t *rtn = (int64_t *)rq->data;
uint64_t *region_p = (uint64_t *)rq->data;
uint64_t region = *region_p;
struct log_c *lc = get_log(rq->uuid, rq->luid);
if (!lc)
return -EINVAL;
if (region > lc->region_count)
return -EINVAL;
*rtn = log_test_bit(lc->sync_bits, region);
/*
* If the mirror was successfully recovered, we want to always
* force every machine to write to all devices - otherwise,
* corruption will occur. Here's how:
* Node1 suffers a failure and marks a region out-of-sync
* Node2 attempts a write, gets by is_remote_recovering,
* and queries the sync status of the region - finding
* it out-of-sync.
* Node2 thinks the write should be a nosync write, but it
* hasn't suffered the drive failure that Node1 has yet.
* It then issues a generic_make_request directly to
* the primary image only - which is exactly the device
* that has suffered the failure.
* Node2 suffers a lost write - which completely bypasses the
* mirror layer because it had gone through generic_m_r.
* The file system will likely explode at this point due to
* I/O errors. If it wasn't the primary that failed, it is
* easily possible in this case to issue writes to just one
* of the remaining images - also leaving the mirror inconsistent.
*
* We let in_sync() return 1 in a cluster regardless of what is
* in the bitmap once recovery has successfully completed on a