-
Notifications
You must be signed in to change notification settings - Fork 4k
/
Copy pathsql_base.cc
10715 lines (9215 loc) · 399 KB
/
sql_base.cc
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) 2000, 2023, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License, version 2.0,
as published by the Free Software Foundation.
This program is also distributed with certain software (including
but not limited to OpenSSL) that is licensed under separate terms,
as designated in a particular file or component or in included license
documentation. The authors of MySQL hereby grant you an additional
permission to link the program and your derivative works with the
separately licensed software that they have included with MySQL.
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, version 2.0, 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 */
/* Basic functions needed by many modules */
#include "sql/sql_base.h"
#include <fcntl.h>
#include <limits.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include <atomic>
#include <functional>
#include <memory>
#include <unordered_map>
#include <utility>
#include "ft_global.h"
#include "m_string.h"
#include "map_helpers.h"
#include "mf_wcomp.h" // wild_one, wild_many
#include "mutex_lock.h"
#include "my_alloc.h"
#include "my_bitmap.h"
#include "my_byteorder.h"
#include "my_compiler.h"
#include "my_dbug.h"
#include "my_dir.h"
#include "my_io.h"
#include "my_macros.h"
#include "my_psi_config.h"
#include "my_sqlcommand.h"
#include "my_sys.h"
#include "my_systime.h"
#include "my_table_map.h"
#include "my_thread_local.h"
#include "mysql/binlog/event/table_id.h"
#include "mysql/components/services/bits/mysql_cond_bits.h"
#include "mysql/components/services/bits/psi_bits.h"
#include "mysql/components/services/bits/psi_cond_bits.h"
#include "mysql/components/services/bits/psi_mutex_bits.h"
#include "mysql/components/services/log_builtins.h"
#include "mysql/my_loglevel.h"
#include "mysql/plugin.h"
#include "mysql/psi/mysql_cond.h"
#include "mysql/psi/mysql_file.h"
#include "mysql/psi/mysql_mutex.h"
#include "mysql/psi/mysql_statement.h"
#include "mysql/psi/mysql_table.h"
#include "mysql/psi/mysql_thread.h"
#include "mysql/psi/psi_table.h"
#include "mysql/service_mysql_alloc.h"
#include "mysql/strings/m_ctype.h"
#include "mysql/thread_type.h"
#include "mysql_com.h"
#include "mysqld_error.h"
#include "nulls.h"
#include "scope_guard.h"
#include "sql/auth/auth_acls.h"
#include "sql/auth/auth_common.h" // check_table_access
#include "sql/auth/sql_security_ctx.h"
#include "sql/binlog.h" // mysql_bin_log
#include "sql/check_stack.h"
#include "sql/dd/cache/dictionary_client.h"
#include "sql/dd/dd_schema.h"
#include "sql/dd/dd_table.h" // dd::table_exists
#include "sql/dd/dd_tablespace.h" // dd::fill_table_and_parts_tablespace_name
#include "sql/dd/string_type.h"
#include "sql/dd/types/abstract_table.h"
#include "sql/dd/types/column.h"
#include "sql/dd/types/column_statistics.h"
#include "sql/dd/types/foreign_key.h" // dd::Foreign_key
#include "sql/dd/types/function.h"
#include "sql/dd/types/procedure.h"
#include "sql/dd/types/schema.h"
#include "sql/dd/types/table.h" // dd::Table
#include "sql/dd/types/view.h"
#include "sql/dd_table_share.h" // open_table_def
#include "sql/debug_sync.h" // DEBUG_SYNC
#include "sql/derror.h" // ER_THD
#include "sql/error_handler.h" // Internal_error_handler
#include "sql/field.h"
#include "sql/handler.h"
#include "sql/histograms/histogram.h"
#include "sql/histograms/table_histograms.h"
#include "sql/item.h"
#include "sql/item_cmpfunc.h" // Item_func_eq
#include "sql/item_func.h"
#include "sql/item_subselect.h"
#include "sql/lock.h" // mysql_lock_remove
#include "sql/log.h"
#include "sql/log_event.h" // Query_log_event
#include "sql/mysqld.h" // replica_open_temp_tables
#include "sql/mysqld_thd_manager.h" // Global_THD_manage
#include "sql/nested_join.h"
#include "sql/partition_info.h" // partition_info
#include "sql/psi_memory_key.h" // key_memory_TABLE
#include "sql/query_options.h"
#include "sql/rpl_gtid.h"
#include "sql/rpl_handler.h" // RUN_HOOK
#include "sql/rpl_replica_commit_order_manager.h" // has_commit_order_manager
#include "sql/rpl_rli.h" //Relay_log_information
#include "sql/session_tracker.h"
#include "sql/sp.h" // Sroutine_hash_entry
#include "sql/sp_cache.h" // sp_cache_version
#include "sql/sp_head.h" // sp_head
#include "sql/sql_audit.h" // mysql_event_tracking_table_access_notify
#include "sql/sql_backup_lock.h" // acquire_shared_backup_lock
#include "sql/sql_class.h" // THD
#include "sql/sql_const.h"
#include "sql/sql_data_change.h"
#include "sql/sql_db.h" // check_schema_readonly
#include "sql/sql_error.h" // Sql_condition
#include "sql/sql_executor.h" // unwrap_rollup_group
#include "sql/sql_handler.h" // mysql_ha_flush_tables
#include "sql/sql_lex.h"
#include "sql/sql_list.h"
#include "sql/sql_parse.h" // is_update_query
#include "sql/sql_prepare.h" // Reprepare_observer
#include "sql/sql_select.h" // reset_statement_timer
#include "sql/sql_show.h" // append_identifier
#include "sql/sql_sort.h"
#include "sql/sql_table.h" // build_table_filename
#include "sql/sql_update.h" // records_are_comparable
#include "sql/sql_view.h" // mysql_make_view
#include "sql/strfunc.h"
#include "sql/system_variables.h"
#include "sql/table.h" // Table_ref
#include "sql/table_cache.h" // table_cache_manager
#include "sql/table_trigger_dispatcher.h" // Table_trigger_dispatcher
#include "sql/thd_raii.h"
#include "sql/transaction.h" // trans_rollback_stmt
#include "sql/transaction_info.h"
#include "sql/trigger_chain.h" // Trigger_chain
#include "sql/xa.h"
#include "sql_string.h"
#include "strmake.h"
#include "strxnmov.h"
#include "template_utils.h"
#include "thr_mutex.h"
using std::equal_to;
using std::hash;
using std::pair;
using std::string;
using std::unique_ptr;
using std::unordered_map;
/**
The maximum length of a key in the table definition cache.
The key consists of the schema name, a '\0' character, the table
name and a '\0' character. Hence NAME_LEN * 2 + 1 + 1.
Additionally, the key can be suffixed with either 4 + 4 extra bytes
for slave tmp tables, or with a single extra byte for tables in a
secondary storage engine. Add 4 + 4 to account for either of these
suffixes.
*/
static constexpr const size_t MAX_DBKEY_LENGTH{NAME_LEN * 2 + 1 + 1 + 4 + 4};
static constexpr long STACK_MIN_SIZE_FOR_OPEN{1024 * 80};
/**
This internal handler is used to trap ER_NO_SUCH_TABLE and
ER_WRONG_MRG_TABLE errors during CHECK/REPAIR TABLE for MERGE
tables.
*/
class Repair_mrg_table_error_handler : public Internal_error_handler {
public:
Repair_mrg_table_error_handler()
: m_handled_errors(false), m_unhandled_errors(false) {}
bool handle_condition(THD *, uint sql_errno, const char *,
Sql_condition::enum_severity_level *,
const char *) override {
if (sql_errno == ER_NO_SUCH_TABLE || sql_errno == ER_WRONG_MRG_TABLE) {
m_handled_errors = true;
return true;
}
m_unhandled_errors = true;
return false;
}
/**
Returns true if there were ER_NO_SUCH_/WRONG_MRG_TABLE and there
were no unhandled errors. false otherwise.
*/
bool safely_trapped_errors() {
/*
Check for m_handled_errors is here for extra safety.
It can be useful in situation when call to open_table()
fails because some error which was suppressed by another
error handler (e.g. in case of MDL deadlock which we
decided to solve by back-off and retry).
*/
return (m_handled_errors && (!m_unhandled_errors));
}
private:
bool m_handled_errors;
bool m_unhandled_errors;
};
/**
@defgroup Data_Dictionary Data Dictionary
@{
*/
/**
LOCK_open protects the following variables/objects:
1) The table_def_cache
This is the hash table mapping table name to a table
share object. The hash table can only be manipulated
while holding LOCK_open.
2) last_table_id
Generation of a new unique table_map_id for a table
share is done through incrementing last_table_id, a
global variable used for this purpose.
3) LOCK_open protects the initialisation of the table share
object and all its members, however, it does not protect
reading the .frm file from where the table share is
initialised. In get_table_share, the lock is temporarily
released while opening the table definition in order to
allow a higher degree of concurrency. Concurrent access
to the same share is controlled by introducing a condition
variable for signaling when opening the share is completed.
4) In particular the share->ref_count is updated each time
a new table object is created that refers to a table share.
This update is protected by LOCK_open.
5) oldest_unused_share, end_of_unused_share and share->next
and share->prev are variables to handle the lists of table
share objects, these can only be read and manipulated while
holding the LOCK_open mutex.
6) table_def_shutdown_in_progress can be updated only while
holding LOCK_open and ALL table cache mutexes.
7) refresh_version
This variable can only be updated while holding LOCK_open AND
all table cache mutexes.
8) share->version
This variable is initialised while holding LOCK_open. It can only
be updated while holding LOCK_open AND all table cache mutexes.
So if a table share is found through a reference its version won't
change if any of those mutexes are held.
9) share->m_flush_tickets
10) share->m_histograms
Inserting, acquiring, and releasing histograms from the collection
of histograms on the share is protected by LOCK_open.
See the comments in table_histograms.h for further details.
*/
mysql_mutex_t LOCK_open;
/**
COND_open synchronizes concurrent opening of the same share:
If a thread calls get_table_share, it releases the LOCK_open
mutex while reading the definition from file. If a different
thread calls get_table_share for the same share at this point
in time, it will find the share in the TDC, but with the
m_open_in_progress flag set to true. This will make the
(second) thread wait for the COND_open condition, while the
first thread completes opening the table definition.
When the first thread is done reading the table definition,
it will set m_open_in_progress to false and broadcast the
COND_open condition. Then, all threads waiting for COND_open
will wake up and, re-search the TDC for the share, and:
1) If the share is gone, the thread will continue to allocate
and open the table definition. This happens, e.g., if the
first thread failed when opening the table definition and
had to destroy the share.
2) If the share is still in the cache, and m_open_in_progress
is still true, the thread will wait for the condition again.
This happens if a different thread finished opening a
different share.
3) If the share is still in the cache, and m_open_in_progress
has become false, the thread will check if the share is ok
(no error), increment the ref counter, and return the share.
*/
mysql_cond_t COND_open;
#ifdef HAVE_PSI_INTERFACE
static PSI_mutex_key key_LOCK_open;
static PSI_cond_key key_COND_open;
static PSI_mutex_info all_tdc_mutexes[] = {
{&key_LOCK_open, "LOCK_open", PSI_FLAG_SINGLETON, 0, PSI_DOCUMENT_ME}};
static PSI_cond_info all_tdc_conds[] = {
{&key_COND_open, "COND_open", 0, 0, PSI_DOCUMENT_ME}};
/**
Initialize performance schema instrumentation points
used by the table cache.
*/
static void init_tdc_psi_keys(void) {
const char *category = "sql";
int count;
count = static_cast<int>(array_elements(all_tdc_mutexes));
mysql_mutex_register(category, all_tdc_mutexes, count);
count = static_cast<int>(array_elements(all_tdc_conds));
mysql_cond_register(category, all_tdc_conds, count);
}
#endif /* HAVE_PSI_INTERFACE */
using Table_definition_cache =
malloc_unordered_map<std::string,
std::unique_ptr<TABLE_SHARE, Table_share_deleter>>;
Table_definition_cache *table_def_cache;
static TABLE_SHARE *oldest_unused_share, end_of_unused_share;
static bool table_def_shutdown_in_progress = false;
static bool check_and_update_table_version(THD *thd, Table_ref *tables,
TABLE_SHARE *table_share);
static bool open_table_entry_fini(THD *thd, TABLE_SHARE *share,
const dd::Table *table, TABLE *entry);
static bool auto_repair_table(THD *thd, Table_ref *table_list);
static TABLE *find_temporary_table(THD *thd, const char *table_key,
size_t table_key_length);
static bool tdc_open_view(THD *thd, Table_ref *table_list,
const char *cache_key, size_t cache_key_length);
static bool add_view_place_holder(THD *thd, Table_ref *table_list);
/**
Create a table cache/table definition cache key for a table. The
table is neither a temporary table nor a table in a secondary
storage engine.
@note
The table cache_key is created from:
db_name + \0
table_name + \0
@param[in] db_name the database name
@param[in] table_name the table name
@param[out] key buffer for the key to be created (must be of
size MAX_DBKEY_LENGTH)
@return the length of the key
*/
static size_t create_table_def_key(const char *db_name, const char *table_name,
char *key) {
/*
In theory caller should ensure that both db and table_name are
not longer than NAME_LEN bytes. In practice we play safe to avoid
buffer overruns.
*/
assert(strlen(db_name) <= NAME_LEN && strlen(table_name) <= NAME_LEN);
return strmake(strmake(key, db_name, NAME_LEN) + 1, table_name, NAME_LEN) -
key + 1;
}
/**
Create a table cache/table definition cache key for a temporary table.
The key is constructed by appending the following to the key
generated by #create_table_def_key():
- 4 bytes for master thread id
- 4 bytes pseudo thread id
@param[in] thd thread context
@param[in] db_name the database name
@param[in] table_name the table name
@param[out] key buffer for the key to be created (must be of
size MAX_DBKEY_LENGTH)
@return the length of the key
*/
static size_t create_table_def_key_tmp(const THD *thd, const char *db_name,
const char *table_name, char *key) {
const size_t key_length = create_table_def_key(db_name, table_name, key);
int4store(key + key_length, thd->server_id);
int4store(key + key_length + 4, thd->variables.pseudo_thread_id);
return key_length + TMP_TABLE_KEY_EXTRA;
}
/**
Create a table cache/table definition cache key for a table in a
secondary storage engine.
The key is constructed by appending a single byte with the value 1
to the key generated by #create_table_def_key().
@param db_name the database name
@param table_name the table name
@return the key
*/
std::string create_table_def_key_secondary(const char *db_name,
const char *table_name) {
char key[MAX_DBKEY_LENGTH];
size_t key_length = create_table_def_key(db_name, table_name, key);
// Add a single byte to distinguish the secondary table from the
// primary table. Their db name and table name are identical.
key[key_length++] = 1;
return {key, key_length};
}
/**
Get table cache key for a table list element.
@param [in] table_list Table list element.
@param [out] key On return points to table cache key for the table.
@note Unlike create_table_def_key() call this function doesn't construct
key in a buffer provider by caller. Instead it relies on the fact
that table list element for which key is requested has properly
initialized MDL_request object and the fact that table definition
cache key is suffix of key used in MDL subsystem. So to get table
definition key it simply needs to return pointer to appropriate
part of MDL_key object nested in this table list element.
Indeed, this means that lifetime of key produced by this call is
limited by the lifetime of table list element which it got as
parameter.
@return Length of key.
*/
size_t get_table_def_key(const Table_ref *table_list, const char **key) {
/*
This call relies on the fact that Table_ref::mdl_request::key object
is properly initialized, so table definition cache can be produced
from key used by MDL subsystem.
strcase is converted to strcasecmp because information_schema tables
can be accessed with lower case and upper case table names.
*/
assert(!my_strcasecmp(system_charset_info, table_list->get_db_name(),
table_list->mdl_request.key.db_name()) &&
!my_strcasecmp(system_charset_info, table_list->get_table_name(),
table_list->mdl_request.key.name()));
*key = (const char *)table_list->mdl_request.key.ptr() + 1;
return table_list->mdl_request.key.length() - 1;
}
/*****************************************************************************
Functions to handle table definition cache (TABLE_SHARE)
*****************************************************************************/
void Table_share_deleter::operator()(TABLE_SHARE *share) const {
DBUG_TRACE;
mysql_mutex_assert_owner(&LOCK_open);
if (share->prev) {
/* remove from old_unused_share list */
*share->prev = share->next;
share->next->prev = share->prev;
}
free_table_share(share);
}
bool table_def_init(void) {
#ifdef HAVE_PSI_INTERFACE
init_tdc_psi_keys();
#endif
mysql_mutex_init(key_LOCK_open, &LOCK_open, MY_MUTEX_INIT_FAST);
mysql_cond_init(key_COND_open, &COND_open);
oldest_unused_share = &end_of_unused_share;
end_of_unused_share.prev = &oldest_unused_share;
if (table_cache_manager.init()) {
mysql_cond_destroy(&COND_open);
mysql_mutex_destroy(&LOCK_open);
return true;
}
table_def_cache = new Table_definition_cache(key_memory_table_share);
return false;
}
/**
Notify table definition cache that process of shutting down server
has started so it has to keep number of TABLE and TABLE_SHARE objects
minimal in order to reduce number of references to pluggable engines.
*/
void table_def_start_shutdown(void) {
if (table_def_cache != nullptr) {
table_cache_manager.lock_all_and_tdc();
/*
Ensure that TABLE and TABLE_SHARE objects which are created for
tables that are open during process of plugins' shutdown are
immediately released. This keeps number of references to engine
plugins minimal and allows shutdown to proceed smoothly.
*/
table_def_shutdown_in_progress = true;
table_cache_manager.unlock_all_and_tdc();
/* Free all cached but unused TABLEs and TABLE_SHAREs. */
close_cached_tables(nullptr, nullptr, false, LONG_TIMEOUT);
}
}
void table_def_free(void) {
DBUG_TRACE;
if (table_def_cache != nullptr) {
/* Free table definitions. */
delete table_def_cache;
table_def_cache = nullptr;
table_cache_manager.destroy();
mysql_cond_destroy(&COND_open);
mysql_mutex_destroy(&LOCK_open);
}
}
uint cached_table_definitions(void) { return table_def_cache->size(); }
static TABLE_SHARE *process_found_table_share(THD *thd [[maybe_unused]],
TABLE_SHARE *share,
bool open_view) {
DBUG_TRACE;
mysql_mutex_assert_owner(&LOCK_open);
#if defined(ENABLED_DEBUG_SYNC)
if (!thd->is_attachable_ro_transaction_active())
DEBUG_SYNC(thd, "get_share_found_share");
#endif
/*
We found an existing table definition. Return it if we didn't get
an error when reading the table definition from file.
*/
if (share->error) {
/*
Table definition contained an error.
Note that we report ER_NO_SUCH_TABLE regardless of which error occurred
when the other thread tried to open the table definition (e.g. OOM).
*/
my_error(ER_NO_SUCH_TABLE, MYF(0), share->db.str, share->table_name.str);
return nullptr;
}
if (share->is_view && !open_view) {
my_error(ER_NO_SUCH_TABLE, MYF(0), share->db.str, share->table_name.str);
return nullptr;
}
share->increment_ref_count();
if (share->ref_count() == 1 && share->prev) {
/*
Share was not used before and it was in the old_unused_share list
Unlink share from this list
*/
DBUG_PRINT("info", ("Unlinking from not used list"));
*share->prev = share->next;
share->next->prev = share->prev;
share->next = nullptr;
share->prev = nullptr;
}
/* Free cache if too big */
while (table_def_cache->size() > table_def_size && oldest_unused_share->next)
table_def_cache->erase(to_string(oldest_unused_share->table_cache_key));
DBUG_PRINT("exit", ("share: %p ref_count: %u", share, share->ref_count()));
return share;
}
// MDL_release_locks_visitor subclass to release MDL for COLUMN_STATISTICS.
class Release_histogram_locks : public MDL_release_locks_visitor {
public:
bool release(MDL_ticket *ticket) override {
return ticket->get_key()->mdl_namespace() == MDL_key::COLUMN_STATISTICS;
}
};
/**
Read any existing histogram statistics from the data dictionary and store a
copy of them in the TABLE_SHARE.
This function is called while TABLE_SHARE is being set up and it should
therefore be safe to modify the collection of histograms on the share without
explicity locking LOCK_open.
@note We use short-lived MDL locks with explicit duration to protect the
histograms while reading them. We want to avoid using statement duration locks
on the histograms in order to prevent deadlocks of the following type:
Threads and commands:
Thread A: ALTER TABLE (TABLE_SHARE is not yet loaded in memory, so
read_histograms() will be invoked).
Thread B: ANALYZE TABLE ... UPDATE HISTOGRAM.
Problematic sequence of lock acquisitions:
A: Acquires SHARED_UPGRADABLE on table for ALTER TABLE.
A: Acquires SHARED_READ on histograms (with statement duration) when creating
TABLE_SHARE.
B: Acquires SHARED_READ on table.
B: Attempts to acquire EXCLUSIVE on histograms (in order to update them), but
gets stuck waiting for A to release SHARED_READ on histograms.
A: ((( A could release MDL LOCK on histograms here, if using explicit
duration, allowing B to progress )))
A: Attempts to upgrade the SHARED_UPGRADABLE on the table to EXCLUSIVE during
execution of the ALTER TABLE statement, but gets stuck waiting for thread B
to release SHARED_READ on table.
This deadlock scenario is prevented by releasing Thread A's lock on the
histograms early, before releasing its lock on the table, i.e. by using
explicit locks on the histograms rather than statement duration locks. When
Thread A releases the histogram locks, then Thread B can update the histograms
and eventually release its table lock, and finally Thread A can upgrade its
MDL lock and continue with its ALTER TABLE statement.
@param thd Thread handler
@param share The table share where to store the histograms
@param schema Schema definition
@param table_def Table definition
@retval true on error
@retval false on success
*/
static bool read_histograms(THD *thd, TABLE_SHARE *share,
const dd::Schema *schema,
const dd::Abstract_table *table_def) {
assert(share->m_histograms != nullptr);
Table_histograms *table_histograms =
Table_histograms::create(key_memory_table_share);
if (table_histograms == nullptr) return true;
auto table_histograms_guard =
create_scope_guard([table_histograms]() { table_histograms->destroy(); });
const dd::cache::Dictionary_client::Auto_releaser releaser(thd->dd_client());
MDL_request_list mdl_requests;
for (const auto column : table_def->columns()) {
if (column->is_se_hidden()) continue;
MDL_key mdl_key;
dd::Column_statistics::create_mdl_key(schema->name(), table_def->name(),
column->name(), &mdl_key);
MDL_request *request = new (thd->mem_root) MDL_request;
MDL_REQUEST_INIT_BY_KEY(request, &mdl_key, MDL_SHARED_READ, MDL_EXPLICIT);
mdl_requests.push_front(request);
}
if (thd->mdl_context.acquire_locks(&mdl_requests,
thd->variables.lock_wait_timeout))
return true; /* purecov: deadcode */
auto mdl_guard = create_scope_guard([&]() {
Release_histogram_locks histogram_mdl_releaser;
thd->mdl_context.release_locks(&histogram_mdl_releaser);
});
for (const auto column : table_def->columns()) {
if (column->is_se_hidden()) continue;
const histograms::Histogram *histogram = nullptr;
if (histograms::find_histogram(thd, schema->name().c_str(),
table_def->name().c_str(),
column->name().c_str(), &histogram)) {
// Any error is reported by the dictionary subsystem.
return true; /* purecov: deadcode */
}
if (histogram != nullptr) {
unsigned int field_index = column->ordinal_position() - 1;
if (table_histograms->insert_histogram(field_index, histogram))
return true;
}
}
if (share->m_histograms->insert(table_histograms)) return true;
table_histograms_guard.commit(); // Ownership transferred.
return false;
}
/** Update TABLE_SHARE with options from dd::Schema object */
static void update_schema_options(const dd::Schema *sch_obj,
TABLE_SHARE *share) {
assert(sch_obj != nullptr);
if (sch_obj != nullptr) {
if (sch_obj->read_only())
share->schema_read_only = TABLE_SHARE::Schema_read_only::RO_ON;
else
share->schema_read_only = TABLE_SHARE::Schema_read_only::RO_OFF;
}
}
/**
Get the TABLE_SHARE for a table.
Get a table definition from the table definition cache. If the share
does not exist, create a new one from the persistently stored table
definition, and temporarily release LOCK_open while retrieving it.
Re-lock LOCK_open when the table definition has been retrieved, and
broadcast this to other threads waiting for the share to become opened.
If the share exists, and is in the process of being opened, wait for
opening to complete before continuing.
@pre It is a precondition that the caller must own LOCK_open before
calling this function.
@note Callers of this function cannot rely on LOCK_open being
held for the duration of the call. It may be temporarily
released while the table definition is opened, and it may be
temporarily released while the thread is waiting for a different
thread to finish opening it.
@note After share->m_open_in_progress is set, there should be no wait
for resources like row- or metadata locks, table flushes, etc.
Otherwise, we may end up in deadlocks that will not be detected.
@param thd thread handle
@param db schema name
@param table_name table name
@param key table cache key
@param key_length length of key
@param open_view allow open of view
@param open_secondary get the share for a table in a secondary
storage engine
@return Pointer to the new TABLE_SHARE, or NULL if there was an error
*/
TABLE_SHARE *get_table_share(THD *thd, const char *db, const char *table_name,
const char *key, size_t key_length, bool open_view,
bool open_secondary) {
TABLE_SHARE *share;
bool open_table_err = false;
DBUG_TRACE;
/* Make sure we own LOCK_open */
mysql_mutex_assert_owner(&LOCK_open);
/*
To be able perform any operation on table we should own
some kind of metadata lock on it.
*/
assert(thd->mdl_context.owns_equal_or_stronger_lock(MDL_key::TABLE, db,
table_name, MDL_SHARED));
/*
Read table definition from the cache. If the share is being opened,
wait for the appropriate condition. The share may be destroyed if
open fails, so after cond_wait, we must repeat searching the
hash table.
*/
for (;;) {
auto it = table_def_cache->find(string(key, key_length));
if (it == table_def_cache->end()) {
if (thd->mdl_context.owns_equal_or_stronger_lock(
MDL_key::SCHEMA, db, "", MDL_INTENTION_EXCLUSIVE)) {
break;
}
mysql_mutex_unlock(&LOCK_open);
if (dd::mdl_lock_schema(thd, db, MDL_TRANSACTION)) {
// Lock LOCK_open again to preserve function contract
mysql_mutex_lock(&LOCK_open);
return nullptr;
}
mysql_mutex_lock(&LOCK_open);
// Need to re-try the find after getting the mutex again
continue;
}
share = it->second.get();
if (!share->m_open_in_progress)
return process_found_table_share(thd, share, open_view);
DEBUG_SYNC(thd, "get_share_before_COND_open_wait");
mysql_cond_wait(&COND_open, &LOCK_open);
}
/*
If alloc fails, the share object will not be present in the TDC, so no
thread will be waiting for m_open_in_progress. Hence, a broadcast is
not necessary.
*/
if (!(share = alloc_table_share(db, table_name, key, key_length,
open_secondary))) {
return nullptr;
}
/*
We assign a new table id under the protection of LOCK_open.
We do this instead of creating a new mutex
and using it for the sole purpose of serializing accesses to a
static variable, we assign the table id here. We assign it to the
share before inserting it into the table_def_cache to be really
sure that it cannot be read from the cache without having a table
id assigned.
CAVEAT. This means that the table cannot be used for
binlogging/replication purposes, unless get_table_share() has been
called directly or indirectly.
*/
assign_new_table_id(share);
table_def_cache->emplace(to_string(share->table_cache_key),
unique_ptr<TABLE_SHARE, Table_share_deleter>(share));
/*
We must increase ref_count prior to releasing LOCK_open
to keep the share from being deleted in tdc_remove_table()
and TABLE_SHARE::wait_for_old_version. We must also set
m_open_in_progress to indicate allocated but incomplete share.
*/
share->increment_ref_count(); // Mark in use
share->m_open_in_progress = true; // Mark being opened
/*
Temporarily release LOCK_open before opening the table definition,
which can be done without mutex protection.
*/
mysql_mutex_unlock(&LOCK_open);
#if defined(ENABLED_DEBUG_SYNC)
if (!thd->is_attachable_ro_transaction_active())
DEBUG_SYNC(thd, "get_share_before_open");
#endif
{
// We must make sure the schema is released and unlocked in the right order.
const dd::cache::Dictionary_client::Auto_releaser releaser(
thd->dd_client());
const dd::Schema *sch = nullptr;
const dd::Abstract_table *abstract_table = nullptr;
open_table_err = true; // Assume error to simplify code below.
if (thd->dd_client()->acquire(share->db.str, &sch) ||
thd->dd_client()->acquire(share->db.str, share->table_name.str,
&abstract_table)) {
} else if (sch == nullptr)
my_error(ER_BAD_DB_ERROR, MYF(0), share->db.str);
else if (abstract_table == nullptr)
my_error(ER_NO_SUCH_TABLE, MYF(0), share->db.str, share->table_name.str);
else if (abstract_table->type() == dd::enum_table_type::USER_VIEW ||
abstract_table->type() == dd::enum_table_type::SYSTEM_VIEW) {
if (!open_view) // We found a view but were trying to open table only.
my_error(ER_NO_SUCH_TABLE, MYF(0), share->db.str,
share->table_name.str);
else {
/*
Clone the view reference object and hold it in
TABLE_SHARE member view_object.
*/
share->is_view = true;
const dd::View *tmp_view =
dynamic_cast<const dd::View *>(abstract_table);
share->view_object = tmp_view->clone();
share->table_category =
get_table_category(share->db, share->table_name);
thd->status_var.opened_shares++;
global_aggregated_stats.get_shard(thd->thread_id()).opened_shares++;
open_table_err = false;
}
} else {
assert(abstract_table->type() == dd::enum_table_type::BASE_TABLE);
open_table_err = open_table_def(
thd, share, *dynamic_cast<const dd::Table *>(abstract_table));
/*
Update the table share with meta data from the schema object to
have it readily available to avoid performance degradation.
*/
if (!open_table_err) update_schema_options(sch, share);
/*
Read any existing histogram statistics from the data dictionary and
store a copy of them in the TABLE_SHARE. We only perform this step for
non-temporary tables, since temporary tables have share->m_histograms
set to nullptr.
We need to do this outside the protection of LOCK_open, since the data
dictionary might have to open tables in order to read histogram data
(such recursion will not work).
*/
if (!open_table_err && share->m_histograms != nullptr &&
read_histograms(thd, share, sch, abstract_table)) {
open_table_err = true;
}
}
}
/*
Get back LOCK_open before continuing. Notify all waiters that the
opening is finished, even if there was a failure while opening.
*/
mysql_mutex_lock(&LOCK_open);
share->m_open_in_progress = false;
mysql_cond_broadcast(&COND_open);
/*
Fake an open_table_def error in debug build, resulting in
ER_NO_SUCH_TABLE.
*/
DBUG_EXECUTE_IF("set_open_table_err", {
open_table_err = true;
my_error(ER_NO_SUCH_TABLE, MYF(0), share->db.str, share->table_name.str);
});
/*
If there was an error while opening the definition, delete the
share from the TDC, and (implicitly) destroy the share. Waiters
will detect that the share is gone, and repeat the attempt at
opening the table definition. The ref counter must be stepped
down to allow the share to be destroyed.
*/
if (open_table_err) {
share->error = true; // Allow waiters to detect the error
share->decrement_ref_count();
table_def_cache->erase(to_string(share->table_cache_key));
#if defined(ENABLED_DEBUG_SYNC)
if (!thd->is_attachable_ro_transaction_active())
DEBUG_SYNC(thd, "get_share_after_destroy");
#endif
return nullptr;
}
#ifdef HAVE_PSI_TABLE_INTERFACE
share->m_psi = PSI_TABLE_CALL(get_table_share)(
(share->tmp_table != NO_TMP_TABLE), share);
#else
share->m_psi = NULL;
#endif
DBUG_PRINT("exit", ("share: %p ref_count: %u", share, share->ref_count()));
/* If debug, assert that the share is actually present in the cache */
#ifndef NDEBUG
assert(table_def_cache->count(string(key, key_length)) != 0);
#endif
return share;
}
/**
Get a table share. If it didn't exist, try creating it from engine
For arguments and return values, see get_table_share()
*/
static TABLE_SHARE *get_table_share_with_discover(
THD *thd, Table_ref *table_list, const char *key, size_t key_length,
bool open_secondary, int *error)
{
TABLE_SHARE *share;
bool exists;
DBUG_TRACE;
share = get_table_share(thd, table_list->db, table_list->table_name, key,
key_length, true, open_secondary);
/*
If share is not NULL, we found an existing share.
If share is NULL, and there is no error, we're inside
pre-locking, which silences 'ER_NO_SUCH_TABLE' errors
with the intention to silently drop non-existing tables
from the pre-locking list. In this case we still need to try
auto-discover before returning a NULL share.
Or, we're inside SHOW CREATE VIEW, which
also installs a silencer for ER_NO_SUCH_TABLE error.
If share is NULL and the error is ER_NO_SUCH_TABLE, this is
the same as above, only that the error was not silenced by
pre-locking or SHOW CREATE VIEW.
In both these cases it won't harm to try to discover the
table.
Finally, if share is still NULL, it's a real error and we need
to abort.
@todo Rework alternative ways to deal with ER_NO_SUCH TABLE.
*/
if (share || (thd->is_error() &&