File: trx0roll.c

package info (click to toggle)
mysql-5.1 5.1.73-1%2Bdeb6u1
  • links: PTS, VCS
  • area: main
  • in suites: squeeze-lts
  • size: 197,132 kB
  • ctags: 93,377
  • sloc: cpp: 579,952; ansic: 429,462; perl: 49,053; sh: 21,692; pascal: 21,272; yacc: 12,801; makefile: 4,545; xml: 4,114; sql: 3,297; lex: 1,265; asm: 1,023
file content (1361 lines) | stat: -rw-r--r-- 32,955 bytes parent folder | download | duplicates (2)
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
/*****************************************************************************

Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.

This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; version 2 of the License.

This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 
51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA

*****************************************************************************/

/**************************************************//**
@file trx/trx0roll.c
Transaction rollback

Created 3/26/1996 Heikki Tuuri
*******************************************************/

#include "trx0roll.h"

#ifdef UNIV_NONINL
#include "trx0roll.ic"
#endif

#include "fsp0fsp.h"
#include "mach0data.h"
#include "trx0rseg.h"
#include "trx0trx.h"
#include "trx0undo.h"
#include "trx0rec.h"
#include "que0que.h"
#include "usr0sess.h"
#include "srv0que.h"
#include "srv0start.h"
#include "row0undo.h"
#include "row0mysql.h"
#include "lock0lock.h"
#include "pars0pars.h"

/** This many pages must be undone before a truncate is tried within
rollback */
#define TRX_ROLL_TRUNC_THRESHOLD	1

/** In crash recovery, the current trx to be rolled back; NULL otherwise */
static const trx_t*	trx_roll_crash_recv_trx	= NULL;

/** In crash recovery we set this to the undo n:o of the current trx to be
rolled back. Then we can print how many % the rollback has progressed. */
static ib_int64_t	trx_roll_max_undo_no;

/** Auxiliary variable which tells the previous progress % we printed */
static ulint		trx_roll_progress_printed_pct;

/*******************************************************************//**
Rollback a transaction used in MySQL.
@return	error code or DB_SUCCESS */
UNIV_INTERN
int
trx_general_rollback_for_mysql(
/*===========================*/
	trx_t*		trx,	/*!< in: transaction handle */
	trx_savept_t*	savept)	/*!< in: pointer to savepoint undo number, if
				partial rollback requested, or NULL for
				complete rollback */
{
	mem_heap_t*	heap;
	que_thr_t*	thr;
	roll_node_t*	roll_node;

	/* Tell Innobase server that there might be work for
	utility threads: */

	srv_active_wake_master_thread();

	trx_start_if_not_started(trx);

	heap = mem_heap_create(512);

	roll_node = roll_node_create(heap);

	if (savept) {
		roll_node->partial = TRUE;
		roll_node->savept = *savept;
	}

	trx->error_state = DB_SUCCESS;

	thr = pars_complete_graph_for_exec(roll_node, trx, heap);

	ut_a(thr == que_fork_start_command(que_node_get_parent(thr)));
	que_run_threads(thr);

	mutex_enter(&kernel_mutex);

	while (trx->que_state != TRX_QUE_RUNNING) {

		mutex_exit(&kernel_mutex);

		os_thread_sleep(100000);

		mutex_enter(&kernel_mutex);
	}

	mutex_exit(&kernel_mutex);

	mem_heap_free(heap);

	ut_a(trx->error_state == DB_SUCCESS);

	/* Tell Innobase server that there might be work for
	utility threads: */

	srv_active_wake_master_thread();

	return((int) trx->error_state);
}

/*******************************************************************//**
Rollback a transaction used in MySQL.
@return	error code or DB_SUCCESS */
UNIV_INTERN
int
trx_rollback_for_mysql(
/*===================*/
	trx_t*	trx)	/*!< in: transaction handle */
{
	int	err;

	if (trx->conc_state == TRX_NOT_STARTED) {

		return(DB_SUCCESS);
	}

	trx->op_info = "rollback";

	/* If we are doing the XA recovery of prepared transactions, then
	the transaction object does not have an InnoDB session object, and we
	set a dummy session that we use for all MySQL transactions. */

	err = trx_general_rollback_for_mysql(trx, NULL);

	trx->op_info = "";

	return(err);
}

/*******************************************************************//**
Rollback the latest SQL statement for MySQL.
@return	error code or DB_SUCCESS */
UNIV_INTERN
int
trx_rollback_last_sql_stat_for_mysql(
/*=================================*/
	trx_t*	trx)	/*!< in: transaction handle */
{
	int	err;

	if (trx->conc_state == TRX_NOT_STARTED) {

		return(DB_SUCCESS);
	}

	trx->op_info = "rollback of SQL statement";

	err = trx_general_rollback_for_mysql(trx, &trx->last_sql_stat_start);
	/* The following call should not be needed, but we play safe: */
	trx_mark_sql_stat_end(trx);

	trx->op_info = "";

	return(err);
}

/*******************************************************************//**
Frees a single savepoint struct. */
UNIV_INTERN
void
trx_roll_savepoint_free(
/*=====================*/
	trx_t*			trx,	/*!< in: transaction handle */
	trx_named_savept_t*	savep)	/*!< in: savepoint to free */
{
	ut_a(savep != NULL);
	ut_a(UT_LIST_GET_LEN(trx->trx_savepoints) > 0);

	UT_LIST_REMOVE(trx_savepoints, trx->trx_savepoints, savep);
	mem_free(savep->name);
	mem_free(savep);
}

/*******************************************************************//**
Frees savepoint structs starting from savep, if savep == NULL then
free all savepoints. */
UNIV_INTERN
void
trx_roll_savepoints_free(
/*=====================*/
	trx_t*			trx,	/*!< in: transaction handle */
	trx_named_savept_t*	savep)	/*!< in: free all savepoints > this one;
					if this is NULL, free all savepoints
					of trx */
{
	trx_named_savept_t*	next_savep;

	if (savep == NULL) {
		savep = UT_LIST_GET_FIRST(trx->trx_savepoints);
	} else {
		savep = UT_LIST_GET_NEXT(trx_savepoints, savep);
	}

	while (savep != NULL) {
		next_savep = UT_LIST_GET_NEXT(trx_savepoints, savep);

		trx_roll_savepoint_free(trx, savep);

		savep = next_savep;
	}
}

/*******************************************************************//**
Rolls back a transaction back to a named savepoint. Modifications after the
savepoint are undone but InnoDB does NOT release the corresponding locks
which are stored in memory. If a lock is 'implicit', that is, a new inserted
row holds a lock where the lock information is carried by the trx id stored in
the row, these locks are naturally released in the rollback. Savepoints which
were set after this savepoint are deleted.
@return if no savepoint of the name found then DB_NO_SAVEPOINT,
otherwise DB_SUCCESS */
UNIV_INTERN
ulint
trx_rollback_to_savepoint_for_mysql(
/*================================*/
	trx_t*		trx,			/*!< in: transaction handle */
	const char*	savepoint_name,		/*!< in: savepoint name */
	ib_int64_t*	mysql_binlog_cache_pos)	/*!< out: the MySQL binlog cache
						position corresponding to this
						savepoint; MySQL needs this
						information to remove the
						binlog entries of the queries
						executed after the savepoint */
{
	trx_named_savept_t*	savep;
	ulint			err;

	savep = UT_LIST_GET_FIRST(trx->trx_savepoints);

	while (savep != NULL) {
		if (0 == ut_strcmp(savep->name, savepoint_name)) {
			/* Found */
			break;
		}
		savep = UT_LIST_GET_NEXT(trx_savepoints, savep);
	}

	if (savep == NULL) {

		return(DB_NO_SAVEPOINT);
	}

	if (trx->conc_state == TRX_NOT_STARTED) {
		ut_print_timestamp(stderr);
		fputs("  InnoDB: Error: transaction has a savepoint ", stderr);
		ut_print_name(stderr, trx, FALSE, savep->name);
		fputs(" though it is not started\n", stderr);
		return(DB_ERROR);
	}

	/* We can now free all savepoints strictly later than this one */

	trx_roll_savepoints_free(trx, savep);

	*mysql_binlog_cache_pos = savep->mysql_binlog_cache_pos;

	trx->op_info = "rollback to a savepoint";

	err = trx_general_rollback_for_mysql(trx, &savep->savept);

	/* Store the current undo_no of the transaction so that we know where
	to roll back if we have to roll back the next SQL statement: */

	trx_mark_sql_stat_end(trx);

	trx->op_info = "";

	return(err);
}

/*******************************************************************//**
Creates a named savepoint. If the transaction is not yet started, starts it.
If there is already a savepoint of the same name, this call erases that old
savepoint and replaces it with a new. Savepoints are deleted in a transaction
commit or rollback.
@return	always DB_SUCCESS */
UNIV_INTERN
ulint
trx_savepoint_for_mysql(
/*====================*/
	trx_t*		trx,			/*!< in: transaction handle */
	const char*	savepoint_name,		/*!< in: savepoint name */
	ib_int64_t	binlog_cache_pos)	/*!< in: MySQL binlog cache
						position corresponding to this
						connection at the time of the
						savepoint */
{
	trx_named_savept_t*	savep;

	ut_a(trx);
	ut_a(savepoint_name);

	trx_start_if_not_started(trx);

	savep = UT_LIST_GET_FIRST(trx->trx_savepoints);

	while (savep != NULL) {
		if (0 == ut_strcmp(savep->name, savepoint_name)) {
			/* Found */
			break;
		}
		savep = UT_LIST_GET_NEXT(trx_savepoints, savep);
	}

	if (savep) {
		/* There is a savepoint with the same name: free that */

		UT_LIST_REMOVE(trx_savepoints, trx->trx_savepoints, savep);

		mem_free(savep->name);
		mem_free(savep);
	}

	/* Create a new savepoint and add it as the last in the list */

	savep = mem_alloc(sizeof(trx_named_savept_t));

	savep->name = mem_strdup(savepoint_name);

	savep->savept = trx_savept_take(trx);

	savep->mysql_binlog_cache_pos = binlog_cache_pos;

	UT_LIST_ADD_LAST(trx_savepoints, trx->trx_savepoints, savep);

	return(DB_SUCCESS);
}

/*******************************************************************//**
Releases only the named savepoint. Savepoints which were set after this
savepoint are left as is.
@return if no savepoint of the name found then DB_NO_SAVEPOINT,
otherwise DB_SUCCESS */
UNIV_INTERN
ulint
trx_release_savepoint_for_mysql(
/*============================*/
	trx_t*		trx,			/*!< in: transaction handle */
	const char*	savepoint_name)		/*!< in: savepoint name */
{
	trx_named_savept_t*	savep;

	savep = UT_LIST_GET_FIRST(trx->trx_savepoints);

	/* Search for the savepoint by name and free if found. */
	while (savep != NULL) {
		if (0 == ut_strcmp(savep->name, savepoint_name)) {
			trx_roll_savepoint_free(trx, savep);
			return(DB_SUCCESS);
		}
		savep = UT_LIST_GET_NEXT(trx_savepoints, savep);
	}

	return(DB_NO_SAVEPOINT);
}

/*******************************************************************//**
Determines if this transaction is rolling back an incomplete transaction
in crash recovery.
@return TRUE if trx is an incomplete transaction that is being rolled
back in crash recovery */
UNIV_INTERN
ibool
trx_is_recv(
/*========*/
	const trx_t*	trx)	/*!< in: transaction */
{
	return(trx == trx_roll_crash_recv_trx);
}

/*******************************************************************//**
Returns a transaction savepoint taken at this point in time.
@return	savepoint */
UNIV_INTERN
trx_savept_t
trx_savept_take(
/*============*/
	trx_t*	trx)	/*!< in: transaction */
{
	trx_savept_t	savept;

	savept.least_undo_no = trx->undo_no;

	return(savept);
}

/*******************************************************************//**
Roll back an active transaction. */
static
void
trx_rollback_active(
/*================*/
	trx_t*	trx)	/*!< in/out: transaction */
{
	mem_heap_t*	heap;
	que_fork_t*	fork;
	que_thr_t*	thr;
	roll_node_t*	roll_node;
	dict_table_t*	table;
	ib_int64_t	rows_to_undo;
	const char*	unit		= "";
	ibool		dictionary_locked = FALSE;

	heap = mem_heap_create(512);

	fork = que_fork_create(NULL, NULL, QUE_FORK_RECOVERY, heap);
	fork->trx = trx;

	thr = que_thr_create(fork, heap);

	roll_node = roll_node_create(heap);

	thr->child = roll_node;
	roll_node->common.parent = thr;

	mutex_enter(&kernel_mutex);

	trx->graph = fork;

	ut_a(thr == que_fork_start_command(fork));

	trx_roll_crash_recv_trx	= trx;
	trx_roll_max_undo_no = ut_conv_dulint_to_longlong(trx->undo_no);
	trx_roll_progress_printed_pct = 0;
	rows_to_undo = trx_roll_max_undo_no;

	if (rows_to_undo > 1000000000) {
		rows_to_undo = rows_to_undo / 1000000;
		unit = "M";
	}

	ut_print_timestamp(stderr);
	fprintf(stderr,
		"  InnoDB: Rolling back trx with id " TRX_ID_FMT ", %lu%s"
		" rows to undo\n",
		TRX_ID_PREP_PRINTF(trx->id),
		(ulong) rows_to_undo, unit);
	mutex_exit(&kernel_mutex);

	trx->mysql_thread_id = os_thread_get_curr_id();

	trx->mysql_process_no = os_proc_get_number();

	if (trx_get_dict_operation(trx) != TRX_DICT_OP_NONE) {
		row_mysql_lock_data_dictionary(trx);
		dictionary_locked = TRUE;
	}

	que_run_threads(thr);

	mutex_enter(&kernel_mutex);

	while (trx->que_state != TRX_QUE_RUNNING) {

		mutex_exit(&kernel_mutex);

		fprintf(stderr,
			"InnoDB: Waiting for rollback of trx id %lu to end\n",
			(ulong) ut_dulint_get_low(trx->id));
		os_thread_sleep(100000);

		mutex_enter(&kernel_mutex);
	}

	mutex_exit(&kernel_mutex);

	if (trx_get_dict_operation(trx) != TRX_DICT_OP_NONE
	    && !ut_dulint_is_zero(trx->table_id)) {

		/* If the transaction was for a dictionary operation, we
		drop the relevant table, if it still exists */

		fprintf(stderr,
			"InnoDB: Dropping table with id %lu %lu"
			" in recovery if it exists\n",
			(ulong) ut_dulint_get_high(trx->table_id),
			(ulong) ut_dulint_get_low(trx->table_id));

		table = dict_table_get_on_id_low(trx->table_id);

		if (table) {
			ulint	err;

			fputs("InnoDB: Table found: dropping table ", stderr);
			ut_print_name(stderr, trx, TRUE, table->name);
			fputs(" in recovery\n", stderr);

			err = row_drop_table_for_mysql(table->name, trx, TRUE);
			trx_commit_for_mysql(trx);

			ut_a(err == (int) DB_SUCCESS);
		}
	}

	if (dictionary_locked) {
		row_mysql_unlock_data_dictionary(trx);
	}

	fprintf(stderr, "\nInnoDB: Rolling back of trx id " TRX_ID_FMT
		" completed\n",
		TRX_ID_PREP_PRINTF(trx->id));
	mem_heap_free(heap);

	trx_roll_crash_recv_trx	= NULL;
}

/*******************************************************************//**
Rollback or clean up any incomplete transactions which were
encountered in crash recovery.  If the transaction already was
committed, then we clean up a possible insert undo log. If the
transaction was not yet committed, then we roll it back. */
UNIV_INTERN
void
trx_rollback_or_clean_recovered(
/*============================*/
	ibool	all)	/*!< in: FALSE=roll back dictionary transactions;
			TRUE=roll back all non-PREPARED transactions */
{
	trx_t*	trx;

	mutex_enter(&kernel_mutex);

	if (!UT_LIST_GET_FIRST(trx_sys->trx_list)) {
		goto leave_function;
	}

	if (all) {
		fprintf(stderr,
			"InnoDB: Starting in background the rollback"
			" of uncommitted transactions\n");
	}

	mutex_exit(&kernel_mutex);

loop:
	mutex_enter(&kernel_mutex);

	for (trx = UT_LIST_GET_FIRST(trx_sys->trx_list); trx;
	     trx = UT_LIST_GET_NEXT(trx_list, trx)) {
		if (!trx->is_recovered) {
			continue;
		}

		switch (trx->conc_state) {
		case TRX_NOT_STARTED:
		case TRX_PREPARED:
			continue;

		case TRX_COMMITTED_IN_MEMORY:
			mutex_exit(&kernel_mutex);
			fprintf(stderr,
				"InnoDB: Cleaning up trx with id "
				TRX_ID_FMT "\n",
				TRX_ID_PREP_PRINTF(trx->id));
			trx_cleanup_at_db_startup(trx);
			goto loop;

		case TRX_ACTIVE:
			if (all || trx_get_dict_operation(trx)
			    != TRX_DICT_OP_NONE) {
				mutex_exit(&kernel_mutex);
				trx_rollback_active(trx);
				goto loop;
			}
		}
	}

	if (all) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
			"  InnoDB: Rollback of non-prepared"
			" transactions completed\n");
	}

leave_function:
	mutex_exit(&kernel_mutex);
}

/*******************************************************************//**
Rollback or clean up any incomplete transactions which were
encountered in crash recovery.  If the transaction already was
committed, then we clean up a possible insert undo log. If the
transaction was not yet committed, then we roll it back.
Note: this is done in a background thread.
@return	a dummy parameter */
UNIV_INTERN
os_thread_ret_t
trx_rollback_or_clean_all_recovered(
/*================================*/
	void*	arg __attribute__((unused)))
			/*!< in: a dummy parameter required by
			os_thread_create */
{
	trx_rollback_or_clean_recovered(TRUE);

	/* We count the number of threads in os_thread_exit(). A created
	thread should always use that to exit and not use return() to exit. */

	os_thread_exit(NULL);

	OS_THREAD_DUMMY_RETURN;
}

/*******************************************************************//**
Creates an undo number array.
@return	own: undo number array */
UNIV_INTERN
trx_undo_arr_t*
trx_undo_arr_create(void)
/*=====================*/
{
	trx_undo_arr_t*	arr;
	mem_heap_t*	heap;
	ulint		i;

	heap = mem_heap_create(1024);

	arr = mem_heap_alloc(heap, sizeof(trx_undo_arr_t));

	arr->infos = mem_heap_alloc(heap, sizeof(trx_undo_inf_t)
				    * UNIV_MAX_PARALLELISM);
	arr->n_cells = UNIV_MAX_PARALLELISM;
	arr->n_used = 0;

	arr->heap = heap;

	for (i = 0; i < UNIV_MAX_PARALLELISM; i++) {

		(trx_undo_arr_get_nth_info(arr, i))->in_use = FALSE;
	}

	return(arr);
}

/*******************************************************************//**
Frees an undo number array. */
UNIV_INTERN
void
trx_undo_arr_free(
/*==============*/
	trx_undo_arr_t*	arr)	/*!< in: undo number array */
{
	ut_ad(arr->n_used == 0);

	mem_heap_free(arr->heap);
}

/*******************************************************************//**
Stores info of an undo log record to the array if it is not stored yet.
@return	FALSE if the record already existed in the array */
static
ibool
trx_undo_arr_store_info(
/*====================*/
	trx_t*		trx,	/*!< in: transaction */
	undo_no_t	undo_no)/*!< in: undo number */
{
	trx_undo_inf_t*	cell;
	trx_undo_inf_t*	stored_here;
	trx_undo_arr_t*	arr;
	ulint		n_used;
	ulint		n;
	ulint		i;

	n = 0;
	arr = trx->undo_no_arr;
	n_used = arr->n_used;
	stored_here = NULL;

	for (i = 0;; i++) {
		cell = trx_undo_arr_get_nth_info(arr, i);

		if (!cell->in_use) {
			if (!stored_here) {
				/* Not in use, we may store here */
				cell->undo_no = undo_no;
				cell->in_use = TRUE;

				arr->n_used++;

				stored_here = cell;
			}
		} else {
			n++;

			if (0 == ut_dulint_cmp(cell->undo_no, undo_no)) {

				if (stored_here) {
					stored_here->in_use = FALSE;
					ut_ad(arr->n_used > 0);
					arr->n_used--;
				}

				ut_ad(arr->n_used == n_used);

				return(FALSE);
			}
		}

		if (n == n_used && stored_here) {

			ut_ad(arr->n_used == 1 + n_used);

			return(TRUE);
		}
	}
}

/*******************************************************************//**
Removes an undo number from the array. */
static
void
trx_undo_arr_remove_info(
/*=====================*/
	trx_undo_arr_t*	arr,	/*!< in: undo number array */
	undo_no_t	undo_no)/*!< in: undo number */
{
	trx_undo_inf_t*	cell;
	ulint		i;

	for (i = 0;; i++) {
		cell = trx_undo_arr_get_nth_info(arr, i);

		if (cell->in_use
		    && 0 == ut_dulint_cmp(cell->undo_no, undo_no)) {

			cell->in_use = FALSE;

			ut_ad(arr->n_used > 0);

			arr->n_used--;

			return;
		}
	}
}

/*******************************************************************//**
Gets the biggest undo number in an array.
@return	biggest value, ut_dulint_zero if the array is empty */
static
undo_no_t
trx_undo_arr_get_biggest(
/*=====================*/
	trx_undo_arr_t*	arr)	/*!< in: undo number array */
{
	trx_undo_inf_t*	cell;
	ulint		n_used;
	undo_no_t	biggest;
	ulint		n;
	ulint		i;

	n = 0;
	n_used = arr->n_used;
	biggest = ut_dulint_zero;

	for (i = 0;; i++) {
		cell = trx_undo_arr_get_nth_info(arr, i);

		if (cell->in_use) {
			n++;
			if (ut_dulint_cmp(cell->undo_no, biggest) > 0) {

				biggest = cell->undo_no;
			}
		}

		if (n == n_used) {
			return(biggest);
		}
	}
}

/***********************************************************************//**
Tries truncate the undo logs. */
UNIV_INTERN
void
trx_roll_try_truncate(
/*==================*/
	trx_t*	trx)	/*!< in/out: transaction */
{
	trx_undo_arr_t*	arr;
	undo_no_t	limit;
	undo_no_t	biggest;

	ut_ad(mutex_own(&(trx->undo_mutex)));
	ut_ad(mutex_own(&((trx->rseg)->mutex)));

	trx->pages_undone = 0;

	arr = trx->undo_no_arr;

	limit = trx->undo_no;

	if (arr->n_used > 0) {
		biggest = trx_undo_arr_get_biggest(arr);

		if (ut_dulint_cmp(biggest, limit) >= 0) {

			limit = ut_dulint_add(biggest, 1);
		}
	}

	if (trx->insert_undo) {
		trx_undo_truncate_end(trx, trx->insert_undo, limit);
	}

	if (trx->update_undo) {
		trx_undo_truncate_end(trx, trx->update_undo, limit);
	}
}

/***********************************************************************//**
Pops the topmost undo log record in a single undo log and updates the info
about the topmost record in the undo log memory struct.
@return	undo log record, the page s-latched */
static
trx_undo_rec_t*
trx_roll_pop_top_rec(
/*=================*/
	trx_t*		trx,	/*!< in: transaction */
	trx_undo_t*	undo,	/*!< in: undo log */
	mtr_t*		mtr)	/*!< in: mtr */
{
	page_t*		undo_page;
	ulint		offset;
	trx_undo_rec_t*	prev_rec;
	page_t*		prev_rec_page;

	ut_ad(mutex_own(&(trx->undo_mutex)));

	undo_page = trx_undo_page_get_s_latched(undo->space, undo->zip_size,
						undo->top_page_no, mtr);
	offset = undo->top_offset;

	/*	fprintf(stderr, "Thread %lu undoing trx %lu undo record %lu\n",
	os_thread_get_curr_id(), ut_dulint_get_low(trx->id),
	ut_dulint_get_low(undo->top_undo_no)); */

	prev_rec = trx_undo_get_prev_rec(undo_page + offset,
					 undo->hdr_page_no, undo->hdr_offset,
					 mtr);
	if (prev_rec == NULL) {

		undo->empty = TRUE;
	} else {
		prev_rec_page = page_align(prev_rec);

		if (prev_rec_page != undo_page) {

			trx->pages_undone++;
		}

		undo->top_page_no = page_get_page_no(prev_rec_page);
		undo->top_offset  = prev_rec - prev_rec_page;
		undo->top_undo_no = trx_undo_rec_get_undo_no(prev_rec);
	}

	return(undo_page + offset);
}

/********************************************************************//**
Pops the topmost record when the two undo logs of a transaction are seen
as a single stack of records ordered by their undo numbers. Inserts the
undo number of the popped undo record to the array of currently processed
undo numbers in the transaction. When the query thread finishes processing
of this undo record, it must be released with trx_undo_rec_release.
@return undo log record copied to heap, NULL if none left, or if the
undo number of the top record would be less than the limit */
UNIV_INTERN
trx_undo_rec_t*
trx_roll_pop_top_rec_of_trx(
/*========================*/
	trx_t*		trx,	/*!< in: transaction */
	undo_no_t	limit,	/*!< in: least undo number we need */
	roll_ptr_t*	roll_ptr,/*!< out: roll pointer to undo record */
	mem_heap_t*	heap)	/*!< in: memory heap where copied */
{
	trx_undo_t*	undo;
	trx_undo_t*	ins_undo;
	trx_undo_t*	upd_undo;
	trx_undo_rec_t*	undo_rec;
	trx_undo_rec_t*	undo_rec_copy;
	undo_no_t	undo_no;
	ibool		is_insert;
	trx_rseg_t*	rseg;
	ulint		progress_pct;
	mtr_t		mtr;

	rseg = trx->rseg;
try_again:
	mutex_enter(&(trx->undo_mutex));

	if (trx->pages_undone >= TRX_ROLL_TRUNC_THRESHOLD) {
		mutex_enter(&(rseg->mutex));

		trx_roll_try_truncate(trx);

		mutex_exit(&(rseg->mutex));
	}

	ins_undo = trx->insert_undo;
	upd_undo = trx->update_undo;

	if (!ins_undo || ins_undo->empty) {
		undo = upd_undo;
	} else if (!upd_undo || upd_undo->empty) {
		undo = ins_undo;
	} else if (ut_dulint_cmp(upd_undo->top_undo_no,
				 ins_undo->top_undo_no) > 0) {
		undo = upd_undo;
	} else {
		undo = ins_undo;
	}

	if (!undo || undo->empty
	    || (ut_dulint_cmp(limit, undo->top_undo_no) > 0)) {

		if ((trx->undo_no_arr)->n_used == 0) {
			/* Rollback is ending */

			mutex_enter(&(rseg->mutex));

			trx_roll_try_truncate(trx);

			mutex_exit(&(rseg->mutex));
		}

		mutex_exit(&(trx->undo_mutex));

		return(NULL);
	}

	if (undo == ins_undo) {
		is_insert = TRUE;
	} else {
		is_insert = FALSE;
	}

	*roll_ptr = trx_undo_build_roll_ptr(is_insert, (undo->rseg)->id,
					    undo->top_page_no,
					    undo->top_offset);
	mtr_start(&mtr);

	undo_rec = trx_roll_pop_top_rec(trx, undo, &mtr);

	undo_no = trx_undo_rec_get_undo_no(undo_rec);

	ut_ad(ut_dulint_cmp(ut_dulint_add(undo_no, 1), trx->undo_no) == 0);

	/* We print rollback progress info if we are in a crash recovery
	and the transaction has at least 1000 row operations to undo. */

	if (trx == trx_roll_crash_recv_trx && trx_roll_max_undo_no > 1000) {

		progress_pct = 100 - (ulint)
			((ut_conv_dulint_to_longlong(undo_no) * 100)
			 / trx_roll_max_undo_no);
		if (progress_pct != trx_roll_progress_printed_pct) {
			if (trx_roll_progress_printed_pct == 0) {
				fprintf(stderr,
					"\nInnoDB: Progress in percents:"
					" %lu", (ulong) progress_pct);
			} else {
				fprintf(stderr,
					" %lu", (ulong) progress_pct);
			}
			fflush(stderr);
			trx_roll_progress_printed_pct = progress_pct;
		}
	}

	trx->undo_no = undo_no;

	if (!trx_undo_arr_store_info(trx, undo_no)) {
		/* A query thread is already processing this undo log record */

		mutex_exit(&(trx->undo_mutex));

		mtr_commit(&mtr);

		goto try_again;
	}

	undo_rec_copy = trx_undo_rec_copy(undo_rec, heap);

	mutex_exit(&(trx->undo_mutex));

	mtr_commit(&mtr);

	return(undo_rec_copy);
}

/********************************************************************//**
Reserves an undo log record for a query thread to undo. This should be
called if the query thread gets the undo log record not using the pop
function above.
@return	TRUE if succeeded */
UNIV_INTERN
ibool
trx_undo_rec_reserve(
/*=================*/
	trx_t*		trx,	/*!< in/out: transaction */
	undo_no_t	undo_no)/*!< in: undo number of the record */
{
	ibool	ret;

	mutex_enter(&(trx->undo_mutex));

	ret = trx_undo_arr_store_info(trx, undo_no);

	mutex_exit(&(trx->undo_mutex));

	return(ret);
}

/*******************************************************************//**
Releases a reserved undo record. */
UNIV_INTERN
void
trx_undo_rec_release(
/*=================*/
	trx_t*		trx,	/*!< in/out: transaction */
	undo_no_t	undo_no)/*!< in: undo number */
{
	trx_undo_arr_t*	arr;

	mutex_enter(&(trx->undo_mutex));

	arr = trx->undo_no_arr;

	trx_undo_arr_remove_info(arr, undo_no);

	mutex_exit(&(trx->undo_mutex));
}

/*********************************************************************//**
Starts a rollback operation. */
UNIV_INTERN
void
trx_rollback(
/*=========*/
	trx_t*		trx,	/*!< in: transaction */
	trx_sig_t*	sig,	/*!< in: signal starting the rollback */
	que_thr_t**	next_thr)/*!< in/out: next query thread to run;
				if the value which is passed in is
				a pointer to a NULL pointer, then the
				calling function can start running
				a new query thread; if the passed value is
				NULL, the parameter is ignored */
{
	que_t*		roll_graph;
	que_thr_t*	thr;
	/*	que_thr_t*	thr2; */

	ut_ad(mutex_own(&kernel_mutex));
	ut_ad((trx->undo_no_arr == NULL) || ((trx->undo_no_arr)->n_used == 0));

	/* Initialize the rollback field in the transaction */

	if (sig->type == TRX_SIG_TOTAL_ROLLBACK) {

		trx->roll_limit = ut_dulint_zero;

	} else if (sig->type == TRX_SIG_ROLLBACK_TO_SAVEPT) {

		trx->roll_limit = (sig->savept).least_undo_no;

	} else if (sig->type == TRX_SIG_ERROR_OCCURRED) {

		trx->roll_limit = trx->last_sql_stat_start.least_undo_no;
	} else {
		ut_error;
	}

	ut_a(ut_dulint_cmp(trx->roll_limit, trx->undo_no) <= 0);

	trx->pages_undone = 0;

	if (trx->undo_no_arr == NULL) {
		trx->undo_no_arr = trx_undo_arr_create();
	}

	/* Build a 'query' graph which will perform the undo operations */

	roll_graph = trx_roll_graph_build(trx);

	trx->graph = roll_graph;
	trx->que_state = TRX_QUE_ROLLING_BACK;

	thr = que_fork_start_command(roll_graph);

	ut_ad(thr);

	/*	thr2 = que_fork_start_command(roll_graph);

	ut_ad(thr2); */

	if (next_thr && (*next_thr == NULL)) {
		*next_thr = thr;
		/*		srv_que_task_enqueue_low(thr2); */
	} else {
		srv_que_task_enqueue_low(thr);
		/*		srv_que_task_enqueue_low(thr2); */
	}
}

/****************************************************************//**
Builds an undo 'query' graph for a transaction. The actual rollback is
performed by executing this query graph like a query subprocedure call.
The reply about the completion of the rollback will be sent by this
graph.
@return	own: the query graph */
UNIV_INTERN
que_t*
trx_roll_graph_build(
/*=================*/
	trx_t*	trx)	/*!< in: trx handle */
{
	mem_heap_t*	heap;
	que_fork_t*	fork;
	que_thr_t*	thr;
	/*	que_thr_t*	thr2; */

	ut_ad(mutex_own(&kernel_mutex));

	heap = mem_heap_create(512);
	fork = que_fork_create(NULL, NULL, QUE_FORK_ROLLBACK, heap);
	fork->trx = trx;

	thr = que_thr_create(fork, heap);
	/*	thr2 = que_thr_create(fork, heap); */

	thr->child = row_undo_node_create(trx, thr, heap);
	/*	thr2->child = row_undo_node_create(trx, thr2, heap); */

	return(fork);
}

/*********************************************************************//**
Finishes error processing after the necessary partial rollback has been
done. */
static
void
trx_finish_error_processing(
/*========================*/
	trx_t*	trx)	/*!< in: transaction */
{
	trx_sig_t*	sig;
	trx_sig_t*	next_sig;

	ut_ad(mutex_own(&kernel_mutex));

	sig = UT_LIST_GET_FIRST(trx->signals);

	while (sig != NULL) {
		next_sig = UT_LIST_GET_NEXT(signals, sig);

		if (sig->type == TRX_SIG_ERROR_OCCURRED) {

			trx_sig_remove(trx, sig);
		}

		sig = next_sig;
	}

	trx->que_state = TRX_QUE_RUNNING;
}

/*********************************************************************//**
Finishes a partial rollback operation. */
static
void
trx_finish_partial_rollback_off_kernel(
/*===================================*/
	trx_t*		trx,	/*!< in: transaction */
	que_thr_t**	next_thr)/*!< in/out: next query thread to run;
				if the value which is passed in is a pointer
				to a NULL pointer, then the calling function
				can start running a new query thread; if this
				parameter is NULL, it is ignored */
{
	trx_sig_t*	sig;

	ut_ad(mutex_own(&kernel_mutex));

	sig = UT_LIST_GET_FIRST(trx->signals);

	/* Remove the signal from the signal queue and send reply message
	to it */

	trx_sig_reply(sig, next_thr);
	trx_sig_remove(trx, sig);

	trx->que_state = TRX_QUE_RUNNING;
}

/****************************************************************//**
Finishes a transaction rollback. */
UNIV_INTERN
void
trx_finish_rollback_off_kernel(
/*===========================*/
	que_t*		graph,	/*!< in: undo graph which can now be freed */
	trx_t*		trx,	/*!< in: transaction */
	que_thr_t**	next_thr)/*!< in/out: next query thread to run;
				if the value which is passed in is
				a pointer to a NULL pointer, then the
				calling function can start running
				a new query thread; if this parameter is
				NULL, it is ignored */
{
	trx_sig_t*	sig;
	trx_sig_t*	next_sig;

	ut_ad(mutex_own(&kernel_mutex));

	ut_a(trx->undo_no_arr == NULL || trx->undo_no_arr->n_used == 0);

	/* Free the memory reserved by the undo graph */
	que_graph_free(graph);

	sig = UT_LIST_GET_FIRST(trx->signals);

	if (sig->type == TRX_SIG_ROLLBACK_TO_SAVEPT) {

		trx_finish_partial_rollback_off_kernel(trx, next_thr);

		return;

	} else if (sig->type == TRX_SIG_ERROR_OCCURRED) {

		trx_finish_error_processing(trx);

		return;
	}

#ifdef UNIV_DEBUG
	if (lock_print_waits) {
		fprintf(stderr, "Trx %lu rollback finished\n",
			(ulong) ut_dulint_get_low(trx->id));
	}
#endif /* UNIV_DEBUG */

	trx_commit_off_kernel(trx);

	/* Remove all TRX_SIG_TOTAL_ROLLBACK signals from the signal queue and
	send reply messages to them */

	trx->que_state = TRX_QUE_RUNNING;

	while (sig != NULL) {
		next_sig = UT_LIST_GET_NEXT(signals, sig);

		if (sig->type == TRX_SIG_TOTAL_ROLLBACK) {

			trx_sig_reply(sig, next_thr);

			trx_sig_remove(trx, sig);
		}

		sig = next_sig;
	}
}

/*********************************************************************//**
Creates a rollback command node struct.
@return	own: rollback node struct */
UNIV_INTERN
roll_node_t*
roll_node_create(
/*=============*/
	mem_heap_t*	heap)	/*!< in: mem heap where created */
{
	roll_node_t*	node;

	node = mem_heap_alloc(heap, sizeof(roll_node_t));
	node->common.type = QUE_NODE_ROLLBACK;
	node->state = ROLL_NODE_SEND;

	node->partial = FALSE;

	return(node);
}

/***********************************************************//**
Performs an execution step for a rollback command node in a query graph.
@return	query thread to run next, or NULL */
UNIV_INTERN
que_thr_t*
trx_rollback_step(
/*==============*/
	que_thr_t*	thr)	/*!< in: query thread */
{
	roll_node_t*	node;
	ulint		sig_no;
	trx_savept_t*	savept;

	node = thr->run_node;

	ut_ad(que_node_get_type(node) == QUE_NODE_ROLLBACK);

	if (thr->prev_node == que_node_get_parent(node)) {
		node->state = ROLL_NODE_SEND;
	}

	if (node->state == ROLL_NODE_SEND) {
		mutex_enter(&kernel_mutex);

		node->state = ROLL_NODE_WAIT;

		if (node->partial) {
			sig_no = TRX_SIG_ROLLBACK_TO_SAVEPT;
			savept = &(node->savept);
		} else {
			sig_no = TRX_SIG_TOTAL_ROLLBACK;
			savept = NULL;
		}

		/* Send a rollback signal to the transaction */

		trx_sig_send(thr_get_trx(thr), sig_no, TRX_SIG_SELF, thr,
			     savept, NULL);

		thr->state = QUE_THR_SIG_REPLY_WAIT;

		mutex_exit(&kernel_mutex);

		return(NULL);
	}

	ut_ad(node->state == ROLL_NODE_WAIT);

	thr->run_node = que_node_get_parent(node);

	return(thr);
}