forked from bloomberg/comdb2
-
Notifications
You must be signed in to change notification settings - Fork 0
/
bdb_osqltrn.c
1290 lines (1084 loc) · 39.4 KB
/
bdb_osqltrn.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 2015, 2017, Bloomberg Finance L.P.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
/**
* Snapshot/Serial sql transaction support;
*
*/
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <errno.h>
#include <pthread.h>
#include <limits.h>
#include <list.h>
#include <ctrace.h>
#include <alloca.h>
#include <build/db.h>
#include "bdb_int.h"
#include "bdb_cursor.h"
#include "bdb_api.h"
#include "bdb_osqltrn.h"
#include "bdb_osqllog.h"
#include "bdb_osqlbkfill.h"
#include "bdb_osqlcur.h"
#include "locks.h"
#include "locks_wrap.h"
#include <list.h>
#include <ctrace.h>
#include <logmsg.h>
#ifdef NEWSI_STAT
#include <time.h>
#include <sys/time.h>
#include <util.h>
extern struct timeval logical_undo_time;
#endif
unsigned int bdb_osql_trn_total_count = 0;
extern int request_durable_lsn_from_master(bdb_state_type *bdb_state,
uint32_t *durable_file,
uint32_t *durable_offset,
uint32_t *durable_gen);
/**
* Each snapshot/serializable transaction registers one bdb_osql_trn
*
*/
struct bdb_osql_trn {
tran_type *shadow_tran; /* shadow transaction */
bdb_osql_log_t *first; /* first log to process */
pthread_mutex_t log_mtx; /* mutex for first */
int trak; /* trak this transaction */
LINKC_T(struct bdb_osql_trn) lnk; /* link */
LISTC_T(bdb_osql_log_t) bkfill_list; /* backfill list */
int cancelled; /* set if resource limitation forced this to abort */
bdb_osql_log_t *last_reg; /* last registered log */
};
/**
* Snapshot/Serializable transaction repository
*
*/
typedef struct bdb_osql_trn_repo {
LISTC_T(bdb_osql_trn_t) trns; /* shadow_transactions */
int trak; /* global debug */
} bdb_osql_trn_repo_t;
/* snapshot/serializable session repository */
bdb_osql_trn_repo_t *trn_repo;
/* for creatn/deltn of trns */
pthread_mutex_t trn_repo_mtx = PTHREAD_MUTEX_INITIALIZER;
static int bdb_osql_trn_create_backfill(bdb_state_type *bdb_state,
bdb_osql_trn_t *trn, int *bdberr,
int epoch, int file, int offset,
struct bfillhndl *bkfill_hndl);
static int
bdb_osql_trn_create_backfill_active_trans(bdb_state_type *bdb_state,
bdb_osql_trn_t *trn, int *bdberr,
struct bfillhndl **ret_bkfill_hndl);
int bdb_prepare_newsi_bkfill(bdb_state_type *bdb_state, uint64_t **list,
int *count, DB_LSN *oldest);
/**
* Create the snapshot/serializable transaction repository
*
*/
int bdb_osql_trn_repo_init(int *bdberr)
{
bdb_osql_trn_repo_t *tmp = NULL;
tmp = (bdb_osql_trn_repo_t *)malloc(sizeof(*tmp));
if (!tmp) {
*bdberr = BDBERR_MALLOC;
return -1;
}
listc_init(&tmp->trns, offsetof(bdb_osql_trn_t, lnk));
tmp->trak = 0;
trn_repo = tmp;
return 0;
}
/**
* Enable/disable tracking for trn repo
*
*/
void bdb_osql_trn_trak(int flag)
{
if (!trn_repo)
return;
trn_repo->trak = flag;
}
#ifdef _LINUX_SOURCE
#include <unistd.h>
#include <sys/syscall.h>
void verify_pthread_mutex(pthread_mutex_t *lock)
{
return;
// WTF? The world is turning upside-down .. valgrind is hitting this ..
/*
pid_t tid = syscall(__NR_gettid);
assert(tid == lock->__data.__owner);
*/
}
#else
void verify_pthread_mutex(pthread_mutex_t *lock) { return; }
#endif
void bdb_verify_repo_lock() { verify_pthread_mutex(&trn_repo_mtx); }
/**
* lock the snapshot/serializable transaction repository
*
*/
inline void bdb_osql_trn_repo_lock()
{
Pthread_mutex_lock(&trn_repo_mtx);
}
/**
* unlock the snapshot/serializable transaction repository
*
*/
inline void bdb_osql_trn_repo_unlock()
{
Pthread_mutex_unlock(&trn_repo_mtx);
}
/**
* Destroy the snapshot/serializable transaction repository
*
*/
int bdb_osql_trn_repo_destroy(int *bdberr)
{
bdb_osql_trn_repo_t *tmp = trn_repo;
bdb_osql_trn_t *trn = NULL, *trn2 = NULL;
if (!tmp)
return 0;
Pthread_mutex_lock(&trn_repo_mtx);
trn_repo = NULL;
Pthread_mutex_unlock(&trn_repo_mtx);
/* this should be clean */
LISTC_FOR_EACH_SAFE(&tmp->trns, trn, trn2, lnk)
{
/* this is possible if there are pending cancelled
transactions */
ctrace("%s: pending transaction\n", __func__);
bdb_osql_trn_unregister(trn, bdberr);
}
free(tmp);
return 0;
}
extern int gbl_extended_sql_debug_trace;
extern int gbl_new_snapisol;
DB_LSN bdb_gbl_recoverable_lsn;
int32_t bdb_gbl_recoverable_timestamp = 0;
pthread_mutex_t bdb_gbl_recoverable_lsn_mutex;
DB_LSN bdb_asof_current_lsn = {0};
DB_LSN bdb_latest_commit_lsn = {0};
uint32_t bdb_latest_commit_gen = 0;
pthread_mutex_t bdb_asof_current_lsn_mutex;
pthread_cond_t bdb_asof_current_lsn_cond;
void bdb_set_gbl_recoverable_lsn(void *lsn, int32_t timestamp)
{
if (!gbl_new_snapisol_asof)
return;
Pthread_mutex_lock(&bdb_gbl_recoverable_lsn_mutex);
bdb_gbl_recoverable_timestamp = timestamp;
bdb_gbl_recoverable_lsn = *(DB_LSN *)lsn;
Pthread_mutex_unlock(&bdb_gbl_recoverable_lsn_mutex);
}
int bdb_is_timestamp_recoverable(bdb_state_type *bdb_state, int32_t timestamp)
{
Pthread_mutex_lock(&bdb_gbl_recoverable_lsn_mutex);
if (timestamp < bdb_gbl_recoverable_timestamp) {
Pthread_mutex_unlock(&bdb_gbl_recoverable_lsn_mutex);
return 0;
}
Pthread_mutex_unlock(&bdb_gbl_recoverable_lsn_mutex);
return 1;
}
static inline void cleanup_trn(bdb_osql_trn_t *trn)
{
Pthread_mutex_destroy(&trn->log_mtx);
free(trn);
}
/**
* Register a shadow transaction with the repository
* Called upon transaction begin/start
*
*/
bdb_osql_trn_t *bdb_osql_trn_register(bdb_state_type *bdb_state,
tran_type *shadow_tran, int trak,
int *bdberr, int epoch, uint32_t file,
uint32_t offset, int is_ha_retry)
{
bdb_osql_trn_t *trn = NULL;
int rc = 0, durable_lsns = bdb_state->attr->durable_lsns;
uint32_t durable_gen = 0;
int backfill_required = 0;
struct bfillhndl *bkfill_hndl = NULL;
if (gbl_extended_sql_debug_trace) {
logmsg(LOGMSG_USER,
"%s line %d called with epoch=%d lsn=[%d][%d] "
"is_retry=%d\n",
__func__, __LINE__, epoch, file, offset, is_ha_retry);
}
if (!trn_repo) // pre-check
return NULL;
if (!is_ha_retry)
file = 0;
if ((shadow_tran->tranclass == TRANCLASS_SNAPISOL ||
shadow_tran->tranclass == TRANCLASS_SERIALIZABLE)) {
int behind = 0;
/* Assert that we have an lsn for ha-retries */
if (is_ha_retry)
assert(file);
/* Don't request an LSN if we're given one */
if (epoch || file)
backfill_required = 1;
if (gbl_rowlocks)
backfill_required = 1;
/* Request our startpoint from the master */
if (durable_lsns || file) {
DB_LSN my_lsn, arg_lsn;
uint32_t my_gen;
/* Request my startpoint */
if (!file && (rc = request_durable_lsn_from_master(
bdb_state, &file, &offset, &durable_gen)) != 0) {
*bdberr = BDBERR_NOT_DURABLE;
return NULL;
}
/* Get our current lsn */
BDB_READLOCK("bdb_current_lsn");
bdb_state->dbenv->get_rep_gen(bdb_state->dbenv, &my_gen);
__log_txn_lsn(bdb_state->dbenv, &my_lsn, NULL, NULL);
BDB_RELLOCK();
arg_lsn.file = file;
arg_lsn.offset = offset;
/* Verify generation numbers for durable-lsn case */
if (durable_gen && my_gen != durable_gen)
behind = 1;
/* Verify lsns for both cases */
else if (log_compare(&my_lsn, &arg_lsn) < 0)
behind = 1;
/* Error out if we are too far behind */
if (behind) {
if (gbl_extended_sql_debug_trace) {
logmsg(LOGMSG_USER,
"%s line %d: returning not-durable, durable_gen=%u,"
" my_gen=%u req_lsn=[%d][%d] my_lsn=[%d][%d]\n",
__func__, __LINE__, durable_gen, my_gen, file,
offset, my_lsn.file, my_lsn.offset);
}
*bdberr = BDBERR_NOT_DURABLE;
return NULL;
}
/* Otherwise we'll do backfill */
backfill_required = 1;
}
}
trn = (bdb_osql_trn_t *)calloc(1, sizeof(bdb_osql_trn_t));
if (!trn) {
*bdberr = BDBERR_MALLOC;
return NULL;
}
Pthread_mutex_init(&trn->log_mtx, NULL);
trn->shadow_tran = shadow_tran;
listc_init(&trn->bkfill_list, offsetof(bdb_osql_log_t, lnk));
Pthread_mutex_lock(&trn_repo_mtx);
if (!trn_repo) {
cleanup_trn(trn);
trn = NULL;
goto done;
}
trn->trak = (trak & SQL_DBG_BDBTRN) || (trn_repo->trak);
/* Set while holding the trn_repo lock */
shadow_tran->startgenid = bdb_get_commit_genid_generation(
bdb_state, &shadow_tran->snapy_commit_lsn,
&shadow_tran->snapy_commit_generation);
if (shadow_tran->snapy_commit_lsn.file == 0) {
shadow_tran->startgenid =
bdb_get_current_lsn(bdb_state, &(shadow_tran->birth_lsn.file),
&(shadow_tran->birth_lsn.offset));
} else {
shadow_tran->birth_lsn = shadow_tran->snapy_commit_lsn;
}
/* lock the translist and create a backfill handler here
* if we need backfill to avoid active logical
* transactions exiting the system
* after we set our startgenid and birthlsn
*/
if (backfill_required) {
if (gbl_new_snapisol)
rc = bdb_prepare_newsi_bkfill(bdb_state,
&shadow_tran->bkfill_txn_list,
&shadow_tran->bkfill_txn_count,
&shadow_tran->oldest_txn_at_start);
else
rc = bdb_osql_trn_create_backfill_active_trans(
bdb_state, trn, bdberr, &bkfill_hndl);
if (rc) {
logmsg(LOGMSG_ERROR,
"%s:%d failed to create backfill active trans, rc %d\n",
__func__, __LINE__, rc);
cleanup_trn(trn);
trn = NULL;
goto done;
}
}
listc_abl(&trn_repo->trns, trn);
++bdb_osql_trn_total_count;
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_TRN: registered %p rc=%d for shadow=%p genid=%llx "
"birthlsn[%d][%d]\n",
trn, rc, shadow_tran, shadow_tran->startgenid,
shadow_tran->birth_lsn.file, shadow_tran->birth_lsn.offset);
done:
Pthread_mutex_unlock(&trn_repo_mtx);
if (gbl_new_snapisol) {
shadow_tran->pglogs_queue_hash =
hash_init_o(offsetof(struct pglogs_queue_cursor, fileid),
DB_FILE_ID_LEN * sizeof(unsigned char));
}
if (!rc && backfill_required) {
if (gbl_new_snapisol) {
if (epoch && gbl_new_snapisol_asof) {
rc = bdb_get_lsn_context_from_timestamp(
bdb_state, epoch, &shadow_tran->asof_lsn,
&shadow_tran->startgenid, bdberr);
if (gbl_extended_sql_debug_trace) {
logmsg(LOGMSG_USER, "%s line %d get_lsn_context from "
"epoch=%d lsn=[%d][%d] rc=%d\n", __func__,
__LINE__, epoch, file, offset, rc);
}
Pthread_mutex_lock(&bdb_gbl_recoverable_lsn_mutex);
if (!rc && ((bdb_gbl_recoverable_lsn.file == 0 &&
bdb_gbl_recoverable_lsn.offset == 1) ||
(log_compare(&shadow_tran->asof_lsn,
&bdb_gbl_recoverable_lsn) < 0))) {
shadow_tran->asof_lsn.file = 0;
shadow_tran->asof_lsn.offset = 1;
rc = -1;
} else if (rc) {
logmsg(LOGMSG_ERROR, "%s failed to get lsn from timestamp\n",
__func__);
if (epoch >= bdb_gbl_recoverable_timestamp) {
shadow_tran->asof_lsn = bdb_gbl_recoverable_lsn;
rc = 0;
}
}
Pthread_mutex_unlock(&bdb_gbl_recoverable_lsn_mutex);
} else if (file && gbl_new_snapisol_asof) {
if (gbl_extended_sql_debug_trace) {
logmsg(LOGMSG_USER, "%s line %d creating asof backfill lsn=[%d][%d]\n",
__func__, __LINE__, file, offset);
}
shadow_tran->asof_lsn.file = file;
shadow_tran->asof_lsn.offset = offset;
Pthread_mutex_lock(&bdb_gbl_recoverable_lsn_mutex);
if ((bdb_gbl_recoverable_lsn.file == 0 &&
bdb_gbl_recoverable_lsn.offset == 1) ||
(log_compare(&shadow_tran->asof_lsn,
&bdb_gbl_recoverable_lsn) < 0)) {
shadow_tran->asof_lsn.file = 0;
shadow_tran->asof_lsn.offset = 1;
rc = -1;
}
Pthread_mutex_unlock(&bdb_gbl_recoverable_lsn_mutex);
if (!rc) {
bdb_get_context_from_lsn(bdb_state, &shadow_tran->asof_lsn,
&shadow_tran->startgenid, bdberr);
}
}
if (shadow_tran->asof_lsn.file != 0 &&
shadow_tran->asof_lsn.offset != 1) {
int counter = 0;
DB_LSN latest_commit_lsn;
Pthread_mutex_lock(&bdb_asof_current_lsn_mutex);
assert(bdb_latest_commit_lsn.file);
latest_commit_lsn = bdb_latest_commit_lsn;
// Make sure we have everything copied into pglogs
while (log_compare(&bdb_asof_current_lsn, &latest_commit_lsn) <
0) {
struct timespec now;
if (counter++ > 3) {
logmsg(LOGMSG_ERROR,
"%s: waiting for asof_lsn %d:%d to become "
"current with req %d:%d, count=%d\n",
__func__, bdb_asof_current_lsn.file,
bdb_asof_current_lsn.offset, latest_commit_lsn.file,
latest_commit_lsn.offset, counter);
}
rc = clock_gettime(CLOCK_REALTIME, &now);
now.tv_sec += 1;
pthread_cond_timedwait(&bdb_asof_current_lsn_cond,
&bdb_asof_current_lsn_mutex, &now);
}
Pthread_mutex_unlock(&bdb_asof_current_lsn_mutex);
shadow_tran->asof_hashtbl =
hash_init_o(PGLOGS_KEY_OFFSET, PAGE_KEY_SIZE);
bdb_checkpoint_list_get_ckplsn_before_lsn(
shadow_tran->asof_lsn, &shadow_tran->asof_ref_lsn);
#ifdef ASOF_TRACE
printf("snapshot shadow_tran %p began as of lsn[%u][%u], "
"ref_lsn[%u][%u], startgenid %llx\n",
shadow_tran, shadow_tran->asof_lsn.file,
shadow_tran->asof_lsn.offset,
shadow_tran->asof_ref_lsn.file,
shadow_tran->asof_ref_lsn.offset,
shadow_tran->startgenid);
#endif
} else if (rc) {
// not able to recover back to that point
logmsg(LOGMSG_ERROR, "%s:%d unable to recover db to epoch %d, "
"file %d, offset %d\n",
__func__, __LINE__, epoch, file, offset);
*bdberr = BDBERR_NO_LOG;
}
}
if (!gbl_new_snapisol || !gbl_new_snapisol_asof) {
if (gbl_extended_sql_debug_trace) {
logmsg(LOGMSG_USER, "%s line %d creating old-style backfill epoch=%d lsn=[%d][%d]\n",
__func__, __LINE__, epoch, file, offset);
}
rc = bdb_osql_trn_create_backfill(bdb_state, trn, bdberr, epoch,
file, offset, bkfill_hndl);
}
if (rc) {
logmsg(LOGMSG_ERROR, "fail to backfill %d %d\n", rc, *bdberr);
Pthread_mutex_lock(&trn_repo_mtx);
if (trn_repo)
listc_rfl(&trn_repo->trns, trn);
Pthread_mutex_unlock(&trn_repo_mtx);
cleanup_trn(trn);
trn = NULL;
}
}
return trn;
}
/**
* Unregister a shadow transaction with the repository
* Called upon transaction commit/rollback
*
*/
int bdb_osql_trn_unregister(bdb_osql_trn_t *trn, int *bdberr)
{
int rc = 0;
Pthread_mutex_lock(&trn_repo_mtx);
if (!trn_repo) {
Pthread_mutex_unlock(&trn_repo_mtx);
return 0;
}
listc_rfl(&trn_repo->trns, trn);
Pthread_mutex_unlock(&trn_repo_mtx);
/*
fprintf( stderr, "%d %s:%d UNregistered %p\n",
pthread_self(), __FILE__, __LINE__, trn->shadow_tran);
*/
/* clean the logs if we point to anything */
if (trn->first) {
/* at this point trn is unlinked, no more updates for trn */
rc = bdb_osql_log_unregister(trn->shadow_tran, trn->first,
trn->last_reg, trn->trak);
if (rc < 0)
logmsg(LOGMSG_WARN, "%s %d bdb_osql_log_unregister LOG LEAK POSSIBLE\n",
__func__, rc);
} else {
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_TRN: %p has not logs shadow=%p\n", trn,
trn->shadow_tran);
}
/* now clean this structure */
Pthread_mutex_destroy(&trn->log_mtx);
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_TRN: unregistered %p rc=%d for shadow=%p\n", trn,
rc, trn->shadow_tran);
free(trn);
return rc;
}
/**
* Mark "aborted" all snapisol transactions that point to "log"
* as first argument
* (Upon updating the shadows, each transaction will actually abort)
*
*/
int bdb_osql_trn_cancel_clients(bdb_osql_log_t *log, int lock_repo, int *bdberr)
{
bdb_osql_trn_t *trn = NULL;
if (lock_repo) {
/* lock the repository so no transactions will move in/out */
Pthread_mutex_lock(&trn_repo_mtx);
}
if (trn_repo) {
LISTC_FOR_EACH(&trn_repo->trns, trn, lnk)
{
assert(trn->shadow_tran->tranclass == TRANCLASS_SNAPISOL ||
trn->shadow_tran->tranclass == TRANCLASS_SERIALIZABLE);
/* lock transaction mutex while registering the snapshot update */
Pthread_mutex_lock(&trn->log_mtx);
if (trn->first == log) {
trn->cancelled = 1;
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_LOG: cancel transaction %p start "
"genid=%llx due to vlog forced cleanup\n",
trn, trn->shadow_tran->startgenid);
} else {
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_LOG: not affected transaction %p "
"start genid=%llx\n",
trn, trn->shadow_tran->startgenid);
}
Pthread_mutex_unlock(&trn->log_mtx);
}
} else {
logmsg(LOGMSG_WARN, "%s: No trn repo???\n", __func__);
}
if (lock_repo) {
Pthread_mutex_unlock(&trn_repo_mtx);
}
return 0;
}
void bdb_osql_trn_count_clients(int *count, int lock_repo)
{
*count = 0;
if (lock_repo) {
Pthread_mutex_lock(&trn_repo_mtx);
}
bdb_verify_repo_lock();
if (trn_repo) {
*count = listc_size(&trn_repo->trns);
}
if (lock_repo) {
Pthread_mutex_unlock(&trn_repo_mtx);
}
}
void bdb_osql_trn_clients_status()
{
Pthread_mutex_lock(&trn_repo_mtx);
int count;
bdb_osql_trn_count_clients(&count, 0);
logmsg(LOGMSG_USER, "snapshot registered: %u\n", bdb_osql_trn_total_count);
logmsg(LOGMSG_USER, "active snapshot: %u\n", count);
Pthread_mutex_unlock(&trn_repo_mtx);
}
/**
* Check all the shadow transactions and register them
* with the log if any are viable (oldest_genid<birth_genid)
* If there are no clients for this transaction, set empty
*
*/
int bdb_osql_trn_check_clients(bdb_osql_log_t *log, int *empty, int lock_repo,
int *bdberr)
{
bdb_osql_trn_t *trn = NULL;
unsigned long long genid = bdb_osql_log_get_genid(log); /* debug only */
int rc = 0;
int clients = 0;
if (lock_repo) {
/* lock the repository so no transactions will move in/out */
Pthread_mutex_lock(&trn_repo_mtx);
}
if (trn_repo) {
/* this is also run during restart recovery;
there is no sql session/client here, and
the trn repo is not initialized yet
*/
LISTC_FOR_EACH(&trn_repo->trns, trn, lnk)
{
assert(trn->shadow_tran->tranclass == TRANCLASS_SNAPISOL ||
trn->shadow_tran->tranclass == TRANCLASS_SERIALIZABLE);
if (bdb_osql_log_undo_required(trn->shadow_tran, log)) {
/* printf("XXX %p LOG %p Found %llx vs %llx\n", pthread_self(),
log,
trn->shadow_tran->startgenid, genid); */
/* replication is changing genid-s committed before
the start of our transaction, need to update its snapshot
*/
clients++;
/*
fprintf(stderr, "%d %s:%d session %llx found, commit genid
=%llx, clients=%d\n",
pthread_self(), __FILE__, __LINE__,
trn->shadow_tran->startgenid, genid, clients);
*/
/* lock transaction mutex while registering the snapshot update
*/
Pthread_mutex_lock(&trn->log_mtx);
if (!trn->first) {
trn->first = log;
}
trn->last_reg = log;
if (trn->trak)
logmsg(LOGMSG_USER, "TRK_LOG: marking transaction %p first=%p "
"(%llx < %llx)\n",
trn, trn->first, trn->shadow_tran->startgenid,
genid);
Pthread_mutex_unlock(&trn->log_mtx);
} else {
/* printf ( "XXX %d LOG %p Skipping transaction %llx<= %llx\n",
pthread_self(), log,
trn->shadow_tran->startgenid, genid); */
if (trn->trak)
logmsg(LOGMSG_USER,
"TRK_LOG: skipping transaction %p (%llx<= %llx)\n",
trn, trn->shadow_tran->startgenid, genid);
}
}
} else {
logmsg(LOGMSG_WARN, "%s: No trn repo???\n", __func__);
}
if (clients) {
rc = bdb_osql_log_register_clients(log, clients, bdberr);
if (rc)
logmsg(LOGMSG_ERROR, "%s: register_clients %d %d\n", __func__, rc,
*bdberr);
*empty = 0;
} else
*empty = 1;
if (lock_repo) {
Pthread_mutex_unlock(&trn_repo_mtx);
}
return rc;
}
struct clients_update_req {
bdb_state_type *bdb_state;
tran_type *shadow_tran;
struct page_logical_lsn_key *keylist;
unsigned int nkeys;
};
struct clients_relink_req {
bdb_state_type *bdb_state;
tran_type *shadow_tran;
unsigned char *fileid;
db_pgno_t pgno;
db_pgno_t prev_pgno;
db_pgno_t next_pgno;
DB_LSN lsn;
};
int bdb_insert_pglogs_int(hash_t *pglogs_hashtbl, unsigned char *fileid,
db_pgno_t pgno, DB_LSN lsn);
static pthread_mutex_t clients_update_req_lk = PTHREAD_MUTEX_INITIALIZER;
static pool_t *clients_update_req_pool = NULL;
struct clients_update_req *allocate_clients_update_req(void)
{
struct clients_update_req *rq = NULL;
Pthread_mutex_lock(&clients_update_req_lk);
if (clients_update_req_pool == NULL)
clients_update_req_pool = pool_setalloc_init(
sizeof(struct clients_update_req), 0, malloc, free);
rq = pool_getablk(clients_update_req_pool);
Pthread_mutex_unlock(&clients_update_req_lk);
return rq;
}
void free_clients_update_req(struct clients_update_req *rq)
{
Pthread_mutex_lock(&clients_update_req_lk);
pool_relablk(clients_update_req_pool, rq);
Pthread_mutex_unlock(&clients_update_req_lk);
}
static pthread_mutex_t clients_relink_req_lk = PTHREAD_MUTEX_INITIALIZER;
static pool_t *clients_relink_req_pool = NULL;
struct clients_relink_req *allocate_clients_relink_req(void)
{
struct clients_relink_req *rq = NULL;
Pthread_mutex_lock(&clients_relink_req_lk);
if (clients_relink_req_pool == NULL)
clients_relink_req_pool = pool_setalloc_init(
sizeof(struct clients_relink_req), 0, malloc, free);
rq = pool_getablk(clients_relink_req_pool);
Pthread_mutex_unlock(&clients_relink_req_lk);
return rq;
}
void free_clients_relink_req(struct clients_relink_req *rq)
{
Pthread_mutex_lock(&clients_relink_req_lk);
pool_relablk(clients_relink_req_pool, rq);
Pthread_mutex_unlock(&clients_relink_req_lk);
}
/**
* Check all the shadow transactions and
* get the oldest asof_lsn among all begin-as-of transactions
*
*/
int bdb_osql_trn_get_oldest_asof_reflsn(DB_LSN *lsnout)
{
bdb_osql_trn_t *trn = NULL;
tran_type *shadow_tran = NULL;
lsnout->file = 0;
lsnout->offset = 1;
Pthread_mutex_lock(&trn_repo_mtx);
if (trn_repo) {
LISTC_FOR_EACH(&trn_repo->trns, trn, lnk)
{
assert(trn->shadow_tran->tranclass == TRANCLASS_SNAPISOL ||
trn->shadow_tran->tranclass == TRANCLASS_SERIALIZABLE);
shadow_tran = trn->shadow_tran;
assert(shadow_tran);
if (shadow_tran->asof_ref_lsn.file != 0 &&
shadow_tran->asof_ref_lsn.offset != 1) {
if (lsnout->file == 0)
*lsnout = shadow_tran->asof_ref_lsn;
else if (log_compare(&shadow_tran->asof_ref_lsn, lsnout) < 0)
*lsnout = shadow_tran->asof_ref_lsn;
}
}
} else {
logmsg(LOGMSG_WARN, "%s: No trn repo???\n", __func__);
}
Pthread_mutex_unlock(&trn_repo_mtx);
return 0;
}
/**
* Check all the shadow transactions and
* see if deleting log file will affect active begin-as-of transaction
*
*/
int bdb_osql_trn_asof_ok_to_delete_log(int filenum)
{
int rc = 0;
DB_LSN oldest_reference;
rc = bdb_osql_trn_get_oldest_asof_reflsn(&oldest_reference);
if (rc) {
logmsg(LOGMSG_FATAL, "%s: failed to get oldest asof lsn\n", __func__);
abort();
}
if (oldest_reference.file != 0 && oldest_reference.offset != 1) {
if (oldest_reference.file <= filenum)
return 0;
}
return 1;
}
/**
* NOTE: log retrieval for update
* - lock trn
* - if there is no log (first == NULL) return
* - get a copy of the tail of the log (get last)
* - unlock trn
* - process each log from first to last log
* - unlock trn
*/
/**
* Returns the first log
*
*/
bdb_osql_log_t *bdb_osql_trn_first_log(bdb_osql_trn_t *trn, int *bdberr)
{
bdb_osql_log_t *log = NULL;
*bdberr = 0;
/* lock transaction mutex while registering the snapshot update */
Pthread_mutex_lock(&trn->log_mtx);
log = trn->first;
Pthread_mutex_unlock(&trn->log_mtx);
return log;
}
/**
* Returns the next log
*
*/
bdb_osql_log_t *bdb_osql_trn_next_log(bdb_osql_trn_t *trn, bdb_osql_log_t *log,
int *bdberr)
{
bdb_osql_log_t *next = NULL;
unsigned long long genid;
*bdberr = 0;
/*TODO: no need for this locking! */
/* lock transaction mutex while registering the snapshot update */
Pthread_mutex_lock(&trn->log_mtx);
next = bdb_osql_log_next(log, &genid);
Pthread_mutex_unlock(&trn->log_mtx);
return next;
}
/**
* Returns the shadow_tran
*
*/
struct tran_tag *bdb_osql_trn_get_shadow_tran(bdb_osql_trn_t *trn)
{
return trn->shadow_tran;
}
extern int gbl_rowlocks;
/**
* Create a backfill handler for a snapshot/serializable transaction
* from the outstanding active logical transactions
*/
static int
bdb_osql_trn_create_backfill_active_trans(bdb_state_type *bdb_state,
bdb_osql_trn_t *trn, int *bdberr,
struct bfillhndl **ret_bkfill_hndl)
{
int rc;
/* get the list of lsn-s */
/* only care about active transactions for now */
rc = bdb_osqlbkfill_create(bdb_state, ret_bkfill_hndl, bdberr, 0, 0, 0,
trn->shadow_tran);
if (rc) {
logmsg(LOGMSG_ERROR, "%s:%d error retrieving backfill handler %d %d\n",
__func__, __LINE__, rc, *bdberr);
return rc;
}
return 0;
}
static int bdb_osql_trn_process_bfillhndl(bdb_state_type *bdb_state,
bdb_osql_trn_t *trn, int *bdberr,
struct bfillhndl *bkfill_hndl,
int bkfill_active_trans)
{
bdb_osql_log_t *log;
DB_LOGC *cur;
DB_LSN lsn;
int rc;
cur = NULL;
log = NULL;
rc = 0;
if (bkfill_hndl == NULL)
return 0;
/* get a log cursor */
rc = bdb_state->dbenv->log_cursor(bdb_state->dbenv, &cur, 0);
if (rc) {
logmsg(LOGMSG_ERROR, "%s:%d log_cursor rc %d\n", __FILE__, __LINE__, rc);
return rc;
}
rc = bdb_osqlbkfill_firstlsn(bkfill_hndl, &lsn, bdberr);