File: row0ins.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 (2616 lines) | stat: -rw-r--r-- 69,227 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
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
/*****************************************************************************

Copyright (c) 1996, 2012, Oracle and/or its affiliates. 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 Street, Suite 500, Boston, MA 02110-1335 USA

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

/**************************************************//**
@file row/row0ins.c
Insert into a table

Created 4/20/1996 Heikki Tuuri
*******************************************************/

#ifdef __WIN__
/* error LNK2001: unresolved external symbol _debug_sync_C_callback_ptr */
# define DEBUG_SYNC_C(dummy) ((void) 0)
#else
# include "my_global.h" /* HAVE_* */
# include "m_string.h" /* for my_sys.h */
# include "my_sys.h" /* DEBUG_SYNC_C */
#endif

#include "row0ins.h"

#ifdef UNIV_NONINL
#include "row0ins.ic"
#endif

#include "ha_prototypes.h"
#include "dict0dict.h"
#include "dict0boot.h"
#include "trx0undo.h"
#include "btr0btr.h"
#include "btr0cur.h"
#include "mach0data.h"
#include "que0que.h"
#include "row0upd.h"
#include "row0sel.h"
#include "row0row.h"
#include "rem0cmp.h"
#include "lock0lock.h"
#include "log0log.h"
#include "eval0eval.h"
#include "data0data.h"
#include "usr0sess.h"
#include "buf0lru.h"

#define	ROW_INS_PREV	1
#define	ROW_INS_NEXT	2

/*************************************************************************
IMPORTANT NOTE: Any operation that generates redo MUST check that there
is enough space in the redo log before for that operation. This is
done by calling log_free_check(). The reason for checking the
availability of the redo log space before the start of the operation is
that we MUST not hold any synchonization objects when performing the
check.
If you make a change in this module make sure that no codepath is
introduced where a call to log_free_check() is bypassed. */

/*********************************************************************//**
Creates an insert node struct.
@return	own: insert node struct */
UNIV_INTERN
ins_node_t*
ins_node_create(
/*============*/
	ulint		ins_type,	/*!< in: INS_VALUES, ... */
	dict_table_t*	table,		/*!< in: table where to insert */
	mem_heap_t*	heap)		/*!< in: mem heap where created */
{
	ins_node_t*	node;

	node = mem_heap_alloc(heap, sizeof(ins_node_t));

	node->common.type = QUE_NODE_INSERT;

	node->ins_type = ins_type;

	node->state = INS_NODE_SET_IX_LOCK;
	node->table = table;
	node->index = NULL;
	node->entry = NULL;

	node->select = NULL;

	node->trx_id = ut_dulint_zero;

	node->entry_sys_heap = mem_heap_create(128);

	node->magic_n = INS_NODE_MAGIC_N;

	return(node);
}

/***********************************************************//**
Creates an entry template for each index of a table. */
UNIV_INTERN
void
ins_node_create_entry_list(
/*=======================*/
	ins_node_t*	node)	/*!< in: row insert node */
{
	dict_index_t*	index;
	dtuple_t*	entry;

	ut_ad(node->entry_sys_heap);

	UT_LIST_INIT(node->entry_list);

	index = dict_table_get_first_index(node->table);

	while (index != NULL) {
		entry = row_build_index_entry(node->row, NULL, index,
					      node->entry_sys_heap);
		UT_LIST_ADD_LAST(tuple_list, node->entry_list, entry);

		index = dict_table_get_next_index(index);
	}
}

/*****************************************************************//**
Adds system field buffers to a row. */
static
void
row_ins_alloc_sys_fields(
/*=====================*/
	ins_node_t*	node)	/*!< in: insert node */
{
	dtuple_t*		row;
	dict_table_t*		table;
	mem_heap_t*		heap;
	const dict_col_t*	col;
	dfield_t*		dfield;
	byte*			ptr;

	row = node->row;
	table = node->table;
	heap = node->entry_sys_heap;

	ut_ad(row && table && heap);
	ut_ad(dtuple_get_n_fields(row) == dict_table_get_n_cols(table));

	/* 1. Allocate buffer for row id */

	col = dict_table_get_sys_col(table, DATA_ROW_ID);

	dfield = dtuple_get_nth_field(row, dict_col_get_no(col));

	ptr = mem_heap_zalloc(heap, DATA_ROW_ID_LEN);

	dfield_set_data(dfield, ptr, DATA_ROW_ID_LEN);

	node->row_id_buf = ptr;

	/* 3. Allocate buffer for trx id */

	col = dict_table_get_sys_col(table, DATA_TRX_ID);

	dfield = dtuple_get_nth_field(row, dict_col_get_no(col));
	ptr = mem_heap_zalloc(heap, DATA_TRX_ID_LEN);

	dfield_set_data(dfield, ptr, DATA_TRX_ID_LEN);

	node->trx_id_buf = ptr;

	/* 4. Allocate buffer for roll ptr */

	col = dict_table_get_sys_col(table, DATA_ROLL_PTR);

	dfield = dtuple_get_nth_field(row, dict_col_get_no(col));
	ptr = mem_heap_zalloc(heap, DATA_ROLL_PTR_LEN);

	dfield_set_data(dfield, ptr, DATA_ROLL_PTR_LEN);
}

/*********************************************************************//**
Sets a new row to insert for an INS_DIRECT node. This function is only used
if we have constructed the row separately, which is a rare case; this
function is quite slow. */
UNIV_INTERN
void
ins_node_set_new_row(
/*=================*/
	ins_node_t*	node,	/*!< in: insert node */
	dtuple_t*	row)	/*!< in: new row (or first row) for the node */
{
	node->state = INS_NODE_SET_IX_LOCK;
	node->index = NULL;
	node->entry = NULL;

	node->row = row;

	mem_heap_empty(node->entry_sys_heap);

	/* Create templates for index entries */

	ins_node_create_entry_list(node);

	/* Allocate from entry_sys_heap buffers for sys fields */

	row_ins_alloc_sys_fields(node);

	/* As we allocated a new trx id buf, the trx id should be written
	there again: */

	node->trx_id = ut_dulint_zero;
}

/*******************************************************************//**
Does an insert operation by updating a delete-marked existing record
in the index. This situation can occur if the delete-marked record is
kept in the index for consistent reads.
@return	DB_SUCCESS or error code */
static
ulint
row_ins_sec_index_entry_by_modify(
/*==============================*/
	ulint		mode,	/*!< in: BTR_MODIFY_LEAF or BTR_MODIFY_TREE,
				depending on whether mtr holds just a leaf
				latch or also a tree latch */
	btr_cur_t*	cursor,	/*!< in: B-tree cursor */
	const dtuple_t*	entry,	/*!< in: index entry to insert */
	que_thr_t*	thr,	/*!< in: query thread */
	mtr_t*		mtr)	/*!< in: mtr; must be committed before
				latching any further pages */
{
	big_rec_t*	dummy_big_rec;
	mem_heap_t*	heap;
	upd_t*		update;
	rec_t*		rec;
	ulint		err;

	rec = btr_cur_get_rec(cursor);

	ut_ad(!dict_index_is_clust(cursor->index));
	ut_ad(rec_get_deleted_flag(rec,
				   dict_table_is_comp(cursor->index->table)));

	/* We know that in the alphabetical ordering, entry and rec are
	identified. But in their binary form there may be differences if
	there are char fields in them. Therefore we have to calculate the
	difference. */

	heap = mem_heap_create(1024);

	update = row_upd_build_sec_rec_difference_binary(
		cursor->index, entry, rec, thr_get_trx(thr), heap);
	if (mode == BTR_MODIFY_LEAF) {
		/* Try an optimistic updating of the record, keeping changes
		within the page */

		err = btr_cur_optimistic_update(BTR_KEEP_SYS_FLAG, cursor,
						update, 0, thr, mtr);
		switch (err) {
		case DB_OVERFLOW:
		case DB_UNDERFLOW:
		case DB_ZIP_OVERFLOW:
			err = DB_FAIL;
		}
	} else {
		ut_a(mode == BTR_MODIFY_TREE);
		if (buf_LRU_buf_pool_running_out()) {

			err = DB_LOCK_TABLE_FULL;

			goto func_exit;
		}

		err = btr_cur_pessimistic_update(BTR_KEEP_SYS_FLAG, cursor,
						 &heap, &dummy_big_rec, update,
						 0, thr, mtr);
		ut_ad(!dummy_big_rec);
	}
func_exit:
	mem_heap_free(heap);

	return(err);
}

/*******************************************************************//**
Does an insert operation by delete unmarking and updating a delete marked
existing record in the index. This situation can occur if the delete marked
record is kept in the index for consistent reads.
@return	DB_SUCCESS, DB_FAIL, or error code */
static
ulint
row_ins_clust_index_entry_by_modify(
/*================================*/
	ulint		mode,	/*!< in: BTR_MODIFY_LEAF or BTR_MODIFY_TREE,
				depending on whether mtr holds just a leaf
				latch or also a tree latch */
	btr_cur_t*	cursor,	/*!< in: B-tree cursor */
	mem_heap_t**	heap,	/*!< in/out: pointer to memory heap, or NULL */
	big_rec_t**	big_rec,/*!< out: possible big rec vector of fields
				which have to be stored externally by the
				caller */
	const dtuple_t*	entry,	/*!< in: index entry to insert */
	que_thr_t*	thr,	/*!< in: query thread */
	mtr_t*		mtr)	/*!< in: mtr; must be committed before
				latching any further pages */
{
	rec_t*		rec;
	upd_t*		update;
	ulint		err;

	ut_ad(dict_index_is_clust(cursor->index));

	*big_rec = NULL;

	rec = btr_cur_get_rec(cursor);

	ut_ad(rec_get_deleted_flag(rec,
				   dict_table_is_comp(cursor->index->table)));

	if (!*heap) {
		*heap = mem_heap_create(1024);
	}

	/* Build an update vector containing all the fields to be modified;
	NOTE that this vector may NOT contain system columns trx_id or
	roll_ptr */

	update = row_upd_build_difference_binary(cursor->index, entry, rec,
						 thr_get_trx(thr), *heap);
	if (mode == BTR_MODIFY_LEAF) {
		/* Try optimistic updating of the record, keeping changes
		within the page */

		err = btr_cur_optimistic_update(0, cursor, update, 0, thr,
						mtr);
		switch (err) {
		case DB_OVERFLOW:
		case DB_UNDERFLOW:
		case DB_ZIP_OVERFLOW:
			err = DB_FAIL;
		}
	} else {
		ut_a(mode == BTR_MODIFY_TREE);
		if (buf_LRU_buf_pool_running_out()) {

			return(DB_LOCK_TABLE_FULL);

		}
		err = btr_cur_pessimistic_update(
			BTR_KEEP_POS_FLAG, cursor, heap, big_rec, update,
			0, thr, mtr);
	}

	return(err);
}

/*********************************************************************//**
Returns TRUE if in a cascaded update/delete an ancestor node of node
updates (not DELETE, but UPDATE) table.
@return	TRUE if an ancestor updates table */
static
ibool
row_ins_cascade_ancestor_updates_table(
/*===================================*/
	que_node_t*	node,	/*!< in: node in a query graph */
	dict_table_t*	table)	/*!< in: table */
{
	que_node_t*	parent;
	upd_node_t*	upd_node;

	parent = que_node_get_parent(node);

	while (que_node_get_type(parent) == QUE_NODE_UPDATE) {

		upd_node = parent;

		if (upd_node->table == table && upd_node->is_delete == FALSE) {

			return(TRUE);
		}

		parent = que_node_get_parent(parent);

		ut_a(parent);
	}

	return(FALSE);
}

/*********************************************************************//**
Returns the number of ancestor UPDATE or DELETE nodes of a
cascaded update/delete node.
@return	number of ancestors */
static
ulint
row_ins_cascade_n_ancestors(
/*========================*/
	que_node_t*	node)	/*!< in: node in a query graph */
{
	que_node_t*	parent;
	ulint		n_ancestors = 0;

	parent = que_node_get_parent(node);

	while (que_node_get_type(parent) == QUE_NODE_UPDATE) {
		n_ancestors++;

		parent = que_node_get_parent(parent);

		ut_a(parent);
	}

	return(n_ancestors);
}

/******************************************************************//**
Calculates the update vector node->cascade->update for a child table in
a cascaded update.
@return number of fields in the calculated update vector; the value
can also be 0 if no foreign key fields changed; the returned value is
ULINT_UNDEFINED if the column type in the child table is too short to
fit the new value in the parent table: that means the update fails */
static
ulint
row_ins_cascade_calc_update_vec(
/*============================*/
	upd_node_t*	node,		/*!< in: update node of the parent
					table */
	dict_foreign_t*	foreign,	/*!< in: foreign key constraint whose
					type is != 0 */
	mem_heap_t*	heap)		/*!< in: memory heap to use as
					temporary storage */
{
	upd_node_t*	cascade		= node->cascade_node;
	dict_table_t*	table		= foreign->foreign_table;
	dict_index_t*	index		= foreign->foreign_index;
	upd_t*		update;
	dict_table_t*	parent_table;
	dict_index_t*	parent_index;
	upd_t*		parent_update;
	ulint		n_fields_updated;
	ulint		parent_field_no;
	ulint		i;
	ulint		j;

	ut_a(node);
	ut_a(foreign);
	ut_a(cascade);
	ut_a(table);
	ut_a(index);

	/* Calculate the appropriate update vector which will set the fields
	in the child index record to the same value (possibly padded with
	spaces if the column is a fixed length CHAR or FIXBINARY column) as
	the referenced index record will get in the update. */

	parent_table = node->table;
	ut_a(parent_table == foreign->referenced_table);
	parent_index = foreign->referenced_index;
	parent_update = node->update;

	update = cascade->update;

	update->info_bits = 0;
	update->n_fields = foreign->n_fields;

	n_fields_updated = 0;

	for (i = 0; i < foreign->n_fields; i++) {

		parent_field_no = dict_table_get_nth_col_pos(
			parent_table,
			dict_index_get_nth_col_no(parent_index, i));

		for (j = 0; j < parent_update->n_fields; j++) {
			const upd_field_t*	parent_ufield
				= &parent_update->fields[j];

			if (parent_ufield->field_no == parent_field_no) {

				ulint			min_size;
				const dict_col_t*	col;
				ulint			ufield_len;
				upd_field_t*		ufield;

				col = dict_index_get_nth_col(index, i);

				/* A field in the parent index record is
				updated. Let us make the update vector
				field for the child table. */

				ufield = update->fields + n_fields_updated;

				ufield->field_no
					= dict_table_get_nth_col_pos(
					table, dict_col_get_no(col));

				ufield->orig_len = 0;
				ufield->exp = NULL;

				ufield->new_val = parent_ufield->new_val;
				ufield_len = dfield_get_len(&ufield->new_val);

				/* Clear the "external storage" flag */
				dfield_set_len(&ufield->new_val, ufield_len);

				/* Do not allow a NOT NULL column to be
				updated as NULL */

				if (dfield_is_null(&ufield->new_val)
				    && (col->prtype & DATA_NOT_NULL)) {

					return(ULINT_UNDEFINED);
				}

				/* If the new value would not fit in the
				column, do not allow the update */

				if (!dfield_is_null(&ufield->new_val)
				    && dtype_get_at_most_n_mbchars(
					col->prtype,
					col->mbminlen, col->mbmaxlen,
					col->len,
					ufield_len,
					dfield_get_data(&ufield->new_val))
				    < ufield_len) {

					return(ULINT_UNDEFINED);
				}

				/* If the parent column type has a different
				length than the child column type, we may
				need to pad with spaces the new value of the
				child column */

				min_size = dict_col_get_min_size(col);

				/* Because UNIV_SQL_NULL (the marker
				of SQL NULL values) exceeds all possible
				values of min_size, the test below will
				not hold for SQL NULL columns. */

				if (min_size > ufield_len) {

					char*		pad_start;
					const char*	pad_end;
					char*		padded_data
						= mem_heap_alloc(
							heap, min_size);
					pad_start = padded_data + ufield_len;
					pad_end = padded_data + min_size;

					memcpy(padded_data,
					       dfield_get_data(&ufield
							       ->new_val),
					       dfield_get_len(&ufield
							      ->new_val));

					switch (UNIV_EXPECT(col->mbminlen,1)) {
					default:
						ut_error;
						return(ULINT_UNDEFINED);
					case 1:
						if (UNIV_UNLIKELY
						    (dtype_get_charset_coll(
							    col->prtype)
						     == DATA_MYSQL_BINARY_CHARSET_COLL)) {
							/* Do not pad BINARY
							columns. */
							return(ULINT_UNDEFINED);
						}

						/* space=0x20 */
						memset(pad_start, 0x20,
						       pad_end - pad_start);
						break;
					case 2:
						/* space=0x0020 */
						ut_a(!(ufield_len % 2));
						ut_a(!(min_size % 2));
						do {
							*pad_start++ = 0x00;
							*pad_start++ = 0x20;
						} while (pad_start < pad_end);
						break;
					}

					dfield_set_data(&ufield->new_val,
							padded_data, min_size);
				}

				n_fields_updated++;
			}
		}
	}

	update->n_fields = n_fields_updated;

	return(n_fields_updated);
}

/*********************************************************************//**
Set detailed error message associated with foreign key errors for
the given transaction. */
static
void
row_ins_set_detailed(
/*=================*/
	trx_t*		trx,		/*!< in: transaction */
	dict_foreign_t*	foreign)	/*!< in: foreign key constraint */
{
	mutex_enter(&srv_misc_tmpfile_mutex);
	rewind(srv_misc_tmpfile);

	if (os_file_set_eof(srv_misc_tmpfile)) {
		ut_print_name(srv_misc_tmpfile, trx, TRUE,
			      foreign->foreign_table_name);
		dict_print_info_on_foreign_key_in_create_format(
			srv_misc_tmpfile, trx, foreign, FALSE);
		trx_set_detailed_error_from_file(trx, srv_misc_tmpfile);
	} else {
		trx_set_detailed_error(trx, "temp file operation failed");
	}

	mutex_exit(&srv_misc_tmpfile_mutex);
}

/*********************************************************************//**
Reports a foreign key error associated with an update or a delete of a
parent table index entry. */
static
void
row_ins_foreign_report_err(
/*=======================*/
	const char*	errstr,		/*!< in: error string from the viewpoint
					of the parent table */
	que_thr_t*	thr,		/*!< in: query thread whose run_node
					is an update node */
	dict_foreign_t*	foreign,	/*!< in: foreign key constraint */
	const rec_t*	rec,		/*!< in: a matching index record in the
					child table */
	const dtuple_t*	entry)		/*!< in: index entry in the parent
					table */
{
	FILE*	ef	= dict_foreign_err_file;
	trx_t*	trx	= thr_get_trx(thr);

	row_ins_set_detailed(trx, foreign);

	mutex_enter(&dict_foreign_err_mutex);
	rewind(ef);
	ut_print_timestamp(ef);
	fputs(" Transaction:\n", ef);
	trx_print(ef, trx, 600);

	fputs("Foreign key constraint fails for table ", ef);
	ut_print_name(ef, trx, TRUE, foreign->foreign_table_name);
	fputs(":\n", ef);
	dict_print_info_on_foreign_key_in_create_format(ef, trx, foreign,
							TRUE);
	putc('\n', ef);
	fputs(errstr, ef);
	fputs(" in parent table, in index ", ef);
	ut_print_name(ef, trx, FALSE, foreign->referenced_index->name);
	if (entry) {
		fputs(" tuple:\n", ef);
		dtuple_print(ef, entry);
	}
	fputs("\nBut in child table ", ef);
	ut_print_name(ef, trx, TRUE, foreign->foreign_table_name);
	fputs(", in index ", ef);
	ut_print_name(ef, trx, FALSE, foreign->foreign_index->name);
	if (rec) {
		fputs(", there is a record:\n", ef);
		rec_print(ef, rec, foreign->foreign_index);
	} else {
		fputs(", the record is not available\n", ef);
	}
	putc('\n', ef);

	mutex_exit(&dict_foreign_err_mutex);
}

/*********************************************************************//**
Reports a foreign key error to dict_foreign_err_file when we are trying
to add an index entry to a child table. Note that the adding may be the result
of an update, too. */
static
void
row_ins_foreign_report_add_err(
/*===========================*/
	trx_t*		trx,		/*!< in: transaction */
	dict_foreign_t*	foreign,	/*!< in: foreign key constraint */
	const rec_t*	rec,		/*!< in: a record in the parent table:
					it does not match entry because we
					have an error! */
	const dtuple_t*	entry)		/*!< in: index entry to insert in the
					child table */
{
	FILE*	ef	= dict_foreign_err_file;

	row_ins_set_detailed(trx, foreign);

	mutex_enter(&dict_foreign_err_mutex);
	rewind(ef);
	ut_print_timestamp(ef);
	fputs(" Transaction:\n", ef);
	trx_print(ef, trx, 600);
	fputs("Foreign key constraint fails for table ", ef);
	ut_print_name(ef, trx, TRUE, foreign->foreign_table_name);
	fputs(":\n", ef);
	dict_print_info_on_foreign_key_in_create_format(ef, trx, foreign,
							TRUE);
	fputs("\nTrying to add in child table, in index ", ef);
	ut_print_name(ef, trx, FALSE, foreign->foreign_index->name);
	if (entry) {
		fputs(" tuple:\n", ef);
		/* TODO: DB_TRX_ID and DB_ROLL_PTR may be uninitialized.
		It would be better to only display the user columns. */
		dtuple_print(ef, entry);
	}
	fputs("\nBut in parent table ", ef);
	ut_print_name(ef, trx, TRUE, foreign->referenced_table_name);
	fputs(", in index ", ef);
	ut_print_name(ef, trx, FALSE, foreign->referenced_index->name);
	fputs(",\nthe closest match we can find is record:\n", ef);
	if (rec && page_rec_is_supremum(rec)) {
		/* If the cursor ended on a supremum record, it is better
		to report the previous record in the error message, so that
		the user gets a more descriptive error message. */
		rec = page_rec_get_prev_const(rec);
	}

	if (rec) {
		rec_print(ef, rec, foreign->referenced_index);
	}
	putc('\n', ef);

	mutex_exit(&dict_foreign_err_mutex);
}

/*********************************************************************//**
Invalidate the query cache for the given table. */
static
void
row_ins_invalidate_query_cache(
/*===========================*/
	que_thr_t*	thr,		/*!< in: query thread whose run_node
					is an update node */
	const char*	name)		/*!< in: table name prefixed with
					database name and a '/' character */
{
	char*	buf;
	char*	ptr;
	ulint	len = strlen(name) + 1;

	buf = mem_strdupl(name, len);

	ptr = strchr(buf, '/');
	ut_a(ptr);
	*ptr = '\0';

	innobase_invalidate_query_cache(thr_get_trx(thr), buf, len);
	mem_free(buf);
}

/*********************************************************************//**
Perform referential actions or checks when a parent row is deleted or updated
and the constraint had an ON DELETE or ON UPDATE condition which was not
RESTRICT.
@return	DB_SUCCESS, DB_LOCK_WAIT, or error code */
static
ulint
row_ins_foreign_check_on_constraint(
/*================================*/
	que_thr_t*	thr,		/*!< in: query thread whose run_node
					is an update node */
	dict_foreign_t*	foreign,	/*!< in: foreign key constraint whose
					type is != 0 */
	btr_pcur_t*	pcur,		/*!< in: cursor placed on a matching
					index record in the child table */
	dtuple_t*	entry,		/*!< in: index entry in the parent
					table */
	mtr_t*		mtr)		/*!< in: mtr holding the latch of pcur
					page */
{
	upd_node_t*	node;
	upd_node_t*	cascade;
	dict_table_t*	table		= foreign->foreign_table;
	dict_index_t*	index;
	dict_index_t*	clust_index;
	dtuple_t*	ref;
	mem_heap_t*	upd_vec_heap	= NULL;
	const rec_t*	rec;
	const rec_t*	clust_rec;
	const buf_block_t* clust_block;
	upd_t*		update;
	ulint		n_to_update;
	ulint		err;
	ulint		i;
	trx_t*		trx;
	mem_heap_t*	tmp_heap	= NULL;

	ut_a(thr);
	ut_a(foreign);
	ut_a(pcur);
	ut_a(mtr);

	trx = thr_get_trx(thr);

	/* Since we are going to delete or update a row, we have to invalidate
	the MySQL query cache for table. A deadlock of threads is not possible
	here because the caller of this function does not hold any latches with
	the sync0sync.h rank above the kernel mutex. The query cache mutex has
	a rank just above the kernel mutex. */

	row_ins_invalidate_query_cache(thr, table->name);

	node = thr->run_node;

	if (node->is_delete && 0 == (foreign->type
				     & (DICT_FOREIGN_ON_DELETE_CASCADE
					| DICT_FOREIGN_ON_DELETE_SET_NULL))) {

		row_ins_foreign_report_err("Trying to delete",
					   thr, foreign,
					   btr_pcur_get_rec(pcur), entry);

		return(DB_ROW_IS_REFERENCED);
	}

	if (!node->is_delete && 0 == (foreign->type
				      & (DICT_FOREIGN_ON_UPDATE_CASCADE
					 | DICT_FOREIGN_ON_UPDATE_SET_NULL))) {

		/* This is an UPDATE */

		row_ins_foreign_report_err("Trying to update",
					   thr, foreign,
					   btr_pcur_get_rec(pcur), entry);

		return(DB_ROW_IS_REFERENCED);
	}

	if (node->cascade_node == NULL) {
		/* Extend our query graph by creating a child to current
		update node. The child is used in the cascade or set null
		operation. */

		node->cascade_heap = mem_heap_create(128);
		node->cascade_node = row_create_update_node_for_mysql(
			table, node->cascade_heap);
		que_node_set_parent(node->cascade_node, node);
	}

	/* Initialize cascade_node to do the operation we want. Note that we
	use the SAME cascade node to do all foreign key operations of the
	SQL DELETE: the table of the cascade node may change if there are
	several child tables to the table where the delete is done! */

	cascade = node->cascade_node;

	cascade->table = table;

	cascade->foreign = foreign;

	if (node->is_delete
	    && (foreign->type & DICT_FOREIGN_ON_DELETE_CASCADE)) {
		cascade->is_delete = TRUE;
	} else {
		cascade->is_delete = FALSE;

		if (foreign->n_fields > cascade->update_n_fields) {
			/* We have to make the update vector longer */

			cascade->update = upd_create(foreign->n_fields,
						     node->cascade_heap);
			cascade->update_n_fields = foreign->n_fields;
		}
	}

	/* We do not allow cyclic cascaded updating (DELETE is allowed,
	but not UPDATE) of the same table, as this can lead to an infinite
	cycle. Check that we are not updating the same table which is
	already being modified in this cascade chain. We have to check
	this also because the modification of the indexes of a 'parent'
	table may still be incomplete, and we must avoid seeing the indexes
	of the parent table in an inconsistent state! */

	if (!cascade->is_delete
	    && row_ins_cascade_ancestor_updates_table(cascade, table)) {

		/* We do not know if this would break foreign key
		constraints, but play safe and return an error */

		err = DB_ROW_IS_REFERENCED;

		row_ins_foreign_report_err(
			"Trying an update, possibly causing a cyclic"
			" cascaded update\n"
			"in the child table,", thr, foreign,
			btr_pcur_get_rec(pcur), entry);

		goto nonstandard_exit_func;
	}

	if (row_ins_cascade_n_ancestors(cascade) >= 15) {
		err = DB_ROW_IS_REFERENCED;

		row_ins_foreign_report_err(
			"Trying a too deep cascaded delete or update\n",
			thr, foreign, btr_pcur_get_rec(pcur), entry);

		goto nonstandard_exit_func;
	}

	index = btr_pcur_get_btr_cur(pcur)->index;

	ut_a(index == foreign->foreign_index);

	rec = btr_pcur_get_rec(pcur);

	if (dict_index_is_clust(index)) {
		/* pcur is already positioned in the clustered index of
		the child table */

		clust_index = index;
		clust_rec = rec;
		clust_block = btr_pcur_get_block(pcur);
	} else {
		/* We have to look for the record in the clustered index
		in the child table */

		clust_index = dict_table_get_first_index(table);

		tmp_heap = mem_heap_create(256);

		ref = row_build_row_ref(ROW_COPY_POINTERS, index, rec,
					tmp_heap);
		btr_pcur_open_with_no_init(clust_index, ref,
					   PAGE_CUR_LE, BTR_SEARCH_LEAF,
					   cascade->pcur, 0, mtr);

		clust_rec = btr_pcur_get_rec(cascade->pcur);
		clust_block = btr_pcur_get_block(cascade->pcur);

		if (!page_rec_is_user_rec(clust_rec)
		    || btr_pcur_get_low_match(cascade->pcur)
		    < dict_index_get_n_unique(clust_index)) {

			fputs("InnoDB: error in cascade of a foreign key op\n"
			      "InnoDB: ", stderr);
			dict_index_name_print(stderr, trx, index);

			fputs("\n"
			      "InnoDB: record ", stderr);
			rec_print(stderr, rec, index);
			fputs("\n"
			      "InnoDB: clustered record ", stderr);
			rec_print(stderr, clust_rec, clust_index);
			fputs("\n"
			      "InnoDB: Submit a detailed bug report to"
			      " http://bugs.mysql.com\n", stderr);

			err = DB_SUCCESS;

			goto nonstandard_exit_func;
		}
	}

	/* Set an X-lock on the row to delete or update in the child table */

	err = lock_table(0, table, LOCK_IX, thr);

	if (err == DB_SUCCESS) {
		/* Here it suffices to use a LOCK_REC_NOT_GAP type lock;
		we already have a normal shared lock on the appropriate
		gap if the search criterion was not unique */

		err = lock_clust_rec_read_check_and_lock_alt(
			0, clust_block, clust_rec, clust_index,
			LOCK_X, LOCK_REC_NOT_GAP, thr);
	}

	if (err != DB_SUCCESS) {

		goto nonstandard_exit_func;
	}

	if (rec_get_deleted_flag(clust_rec, dict_table_is_comp(table))) {
		/* This can happen if there is a circular reference of
		rows such that cascading delete comes to delete a row
		already in the process of being delete marked */
		err = DB_SUCCESS;

		goto nonstandard_exit_func;
	}

	if (node->is_delete
	    ? (foreign->type & DICT_FOREIGN_ON_DELETE_SET_NULL)
	    : (foreign->type & DICT_FOREIGN_ON_UPDATE_SET_NULL)) {

		/* Build the appropriate update vector which sets
		foreign->n_fields first fields in rec to SQL NULL */

		update = cascade->update;

		update->info_bits = 0;
		update->n_fields = foreign->n_fields;
		UNIV_MEM_INVALID(update->fields,
				 update->n_fields * sizeof *update->fields);

		for (i = 0; i < foreign->n_fields; i++) {
			upd_field_t*	ufield = &update->fields[i];

			ufield->field_no = dict_table_get_nth_col_pos(
				table,
				dict_index_get_nth_col_no(index, i));
			ufield->orig_len = 0;
			ufield->exp = NULL;
			dfield_set_null(&ufield->new_val);
		}
	}

	if (!node->is_delete
	    && (foreign->type & DICT_FOREIGN_ON_UPDATE_CASCADE)) {

		/* Build the appropriate update vector which sets changing
		foreign->n_fields first fields in rec to new values */

		upd_vec_heap = mem_heap_create(256);

		n_to_update = row_ins_cascade_calc_update_vec(node, foreign,
							      upd_vec_heap);
		if (n_to_update == ULINT_UNDEFINED) {
			err = DB_ROW_IS_REFERENCED;

			row_ins_foreign_report_err(
				"Trying a cascaded update where the"
				" updated value in the child\n"
				"table would not fit in the length"
				" of the column, or the value would\n"
				"be NULL and the column is"
				" declared as not NULL in the child table,",
				thr, foreign, btr_pcur_get_rec(pcur), entry);

			goto nonstandard_exit_func;
		}

		if (cascade->update->n_fields == 0) {

			/* The update does not change any columns referred
			to in this foreign key constraint: no need to do
			anything */

			err = DB_SUCCESS;

			goto nonstandard_exit_func;
		}
	}

	/* Store pcur position and initialize or store the cascade node
	pcur stored position */

	btr_pcur_store_position(pcur, mtr);

	if (index == clust_index) {
		btr_pcur_copy_stored_position(cascade->pcur, pcur);
	} else {
		btr_pcur_store_position(cascade->pcur, mtr);
	}

	mtr_commit(mtr);

	ut_a(cascade->pcur->rel_pos == BTR_PCUR_ON);

	cascade->state = UPD_NODE_UPDATE_CLUSTERED;

	err = row_update_cascade_for_mysql(thr, cascade,
					   foreign->foreign_table);

	if (foreign->foreign_table->n_foreign_key_checks_running == 0) {
		fprintf(stderr,
			"InnoDB: error: table %s has the counter 0"
			" though there is\n"
			"InnoDB: a FOREIGN KEY check running on it.\n",
			foreign->foreign_table->name);
	}

	/* Release the data dictionary latch for a while, so that we do not
	starve other threads from doing CREATE TABLE etc. if we have a huge
	cascaded operation running. The counter n_foreign_key_checks_running
	will prevent other users from dropping or ALTERing the table when we
	release the latch. */

	row_mysql_unfreeze_data_dictionary(thr_get_trx(thr));

	DEBUG_SYNC_C("innodb_dml_cascade_dict_unfreeze");

	row_mysql_freeze_data_dictionary(thr_get_trx(thr));

	mtr_start(mtr);

	/* Restore pcur position */

	btr_pcur_restore_position(BTR_SEARCH_LEAF, pcur, mtr);

	if (tmp_heap) {
		mem_heap_free(tmp_heap);
	}

	if (upd_vec_heap) {
		mem_heap_free(upd_vec_heap);
	}

	return(err);

nonstandard_exit_func:
	if (tmp_heap) {
		mem_heap_free(tmp_heap);
	}

	if (upd_vec_heap) {
		mem_heap_free(upd_vec_heap);
	}

	btr_pcur_store_position(pcur, mtr);

	mtr_commit(mtr);
	mtr_start(mtr);

	btr_pcur_restore_position(BTR_SEARCH_LEAF, pcur, mtr);

	return(err);
}

/*********************************************************************//**
Sets a shared lock on a record. Used in locking possible duplicate key
records and also in checking foreign key constraints.
@return	DB_SUCCESS, DB_SUCCESS_LOCKED_REC, or error code */
static
enum db_err
row_ins_set_shared_rec_lock(
/*========================*/
	ulint			type,	/*!< in: LOCK_ORDINARY, LOCK_GAP, or
					LOCK_REC_NOT_GAP type lock */
	const buf_block_t*	block,	/*!< in: buffer block of rec */
	const rec_t*		rec,	/*!< in: record */
	dict_index_t*		index,	/*!< in: index */
	const ulint*		offsets,/*!< in: rec_get_offsets(rec, index) */
	que_thr_t*		thr)	/*!< in: query thread */
{
	enum db_err	err;

	ut_ad(rec_offs_validate(rec, index, offsets));

	if (dict_index_is_clust(index)) {
		err = lock_clust_rec_read_check_and_lock(
			0, block, rec, index, offsets, LOCK_S, type, thr);
	} else {
		err = lock_sec_rec_read_check_and_lock(
			0, block, rec, index, offsets, LOCK_S, type, thr);
	}

	return(err);
}

/*********************************************************************//**
Sets a exclusive lock on a record. Used in locking possible duplicate key
records
@return	DB_SUCCESS, DB_SUCCESS_LOCKED_REC, or error code */
static
enum db_err
row_ins_set_exclusive_rec_lock(
/*===========================*/
	ulint			type,	/*!< in: LOCK_ORDINARY, LOCK_GAP, or
					LOCK_REC_NOT_GAP type lock */
	const buf_block_t*	block,	/*!< in: buffer block of rec */
	const rec_t*		rec,	/*!< in: record */
	dict_index_t*		index,	/*!< in: index */
	const ulint*		offsets,/*!< in: rec_get_offsets(rec, index) */
	que_thr_t*		thr)	/*!< in: query thread */
{
	enum db_err	err;

	ut_ad(rec_offs_validate(rec, index, offsets));

	if (dict_index_is_clust(index)) {
		err = lock_clust_rec_read_check_and_lock(
			0, block, rec, index, offsets, LOCK_X, type, thr);
	} else {
		err = lock_sec_rec_read_check_and_lock(
			0, block, rec, index, offsets, LOCK_X, type, thr);
	}

	return(err);
}

/***************************************************************//**
Checks if foreign key constraint fails for an index entry. Sets shared locks
which lock either the success or the failure of the constraint. NOTE that
the caller must have a shared latch on dict_operation_lock.
@return	DB_SUCCESS, DB_NO_REFERENCED_ROW, or DB_ROW_IS_REFERENCED */
UNIV_INTERN
ulint
row_ins_check_foreign_constraint(
/*=============================*/
	ibool		check_ref,/*!< in: TRUE if we want to check that
				the referenced table is ok, FALSE if we
				want to check the foreign key table */
	dict_foreign_t*	foreign,/*!< in: foreign constraint; NOTE that the
				tables mentioned in it must be in the
				dictionary cache if they exist at all */
	dict_table_t*	table,	/*!< in: if check_ref is TRUE, then the foreign
				table, else the referenced table */
	dtuple_t*	entry,	/*!< in: index entry for index */
	que_thr_t*	thr)	/*!< in: query thread */
{
	upd_node_t*	upd_node;
	dict_table_t*	check_table;
	dict_index_t*	check_index;
	ulint		n_fields_cmp;
	btr_pcur_t	pcur;
	int		cmp;
	ulint		err;
	ulint		i;
	mtr_t		mtr;
	trx_t*		trx		= thr_get_trx(thr);
	mem_heap_t*	heap		= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets		= offsets_;
	rec_offs_init(offsets_);

run_again:
#ifdef UNIV_SYNC_DEBUG
	ut_ad(rw_lock_own(&dict_operation_lock, RW_LOCK_SHARED));
#endif /* UNIV_SYNC_DEBUG */

	err = DB_SUCCESS;

	if (trx->check_foreigns == FALSE) {
		/* The user has suppressed foreign key checks currently for
		this session */
		goto exit_func;
	}

	/* If any of the foreign key fields in entry is SQL NULL, we
	suppress the foreign key check: this is compatible with Oracle,
	for example */

	for (i = 0; i < foreign->n_fields; i++) {
		if (UNIV_SQL_NULL == dfield_get_len(
			    dtuple_get_nth_field(entry, i))) {

			goto exit_func;
		}
	}

	if (que_node_get_type(thr->run_node) == QUE_NODE_UPDATE) {
		upd_node = thr->run_node;

		if (!(upd_node->is_delete) && upd_node->foreign == foreign) {
			/* If a cascaded update is done as defined by a
			foreign key constraint, do not check that
			constraint for the child row. In ON UPDATE CASCADE
			the update of the parent row is only half done when
			we come here: if we would check the constraint here
			for the child row it would fail.

			A QUESTION remains: if in the child table there are
			several constraints which refer to the same parent
			table, we should merge all updates to the child as
			one update? And the updates can be contradictory!
			Currently we just perform the update associated
			with each foreign key constraint, one after
			another, and the user has problems predicting in
			which order they are performed. */

			goto exit_func;
		}
	}

	if (check_ref) {
		check_table = foreign->referenced_table;
		check_index = foreign->referenced_index;
	} else {
		check_table = foreign->foreign_table;
		check_index = foreign->foreign_index;
	}

	if (check_table == NULL || check_table->ibd_file_missing
	    || check_index == NULL) {
		if (check_ref) {
			FILE*	ef = dict_foreign_err_file;

			row_ins_set_detailed(trx, foreign);

			mutex_enter(&dict_foreign_err_mutex);
			rewind(ef);
			ut_print_timestamp(ef);
			fputs(" Transaction:\n", ef);
			trx_print(ef, trx, 600);
			fputs("Foreign key constraint fails for table ", ef);
			ut_print_name(ef, trx, TRUE,
				      foreign->foreign_table_name);
			fputs(":\n", ef);
			dict_print_info_on_foreign_key_in_create_format(
				ef, trx, foreign, TRUE);
			fputs("\nTrying to add to index ", ef);
			ut_print_name(ef, trx, FALSE,
				      foreign->foreign_index->name);
			fputs(" tuple:\n", ef);
			dtuple_print(ef, entry);
			fputs("\nBut the parent table ", ef);
			ut_print_name(ef, trx, TRUE,
				      foreign->referenced_table_name);
			fputs("\nor its .ibd file does"
			      " not currently exist!\n", ef);
			mutex_exit(&dict_foreign_err_mutex);

			err = DB_NO_REFERENCED_ROW;
		}

		goto exit_func;
	}

	if (check_table != table) {
		/* We already have a LOCK_IX on table, but not necessarily
		on check_table */

		err = lock_table(0, check_table, LOCK_IS, thr);

		if (err != DB_SUCCESS) {

			goto do_possible_lock_wait;
		}
	}

	mtr_start(&mtr);

	/* Store old value on n_fields_cmp */

	n_fields_cmp = dtuple_get_n_fields_cmp(entry);

	dtuple_set_n_fields_cmp(entry, foreign->n_fields);

	btr_pcur_open(check_index, entry, PAGE_CUR_GE,
		      BTR_SEARCH_LEAF, &pcur, &mtr);

	/* Scan index records and check if there is a matching record */

	do {
		const rec_t*		rec = btr_pcur_get_rec(&pcur);
		const buf_block_t*	block = btr_pcur_get_block(&pcur);

		if (page_rec_is_infimum(rec)) {

			continue;
		}

		offsets = rec_get_offsets(rec, check_index,
					  offsets, ULINT_UNDEFINED, &heap);

		if (page_rec_is_supremum(rec)) {

			err = row_ins_set_shared_rec_lock(LOCK_ORDINARY, block,
							  rec, check_index,
							  offsets, thr);
			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
			case DB_SUCCESS:
				continue;
			default:
				goto end_scan;
			}
		}

		cmp = cmp_dtuple_rec(entry, rec, offsets);

		if (cmp == 0) {
			if (rec_get_deleted_flag(rec,
						 rec_offs_comp(offsets))) {
				err = row_ins_set_shared_rec_lock(
					LOCK_ORDINARY, block,
					rec, check_index, offsets, thr);
				switch (err) {
				case DB_SUCCESS_LOCKED_REC:
				case DB_SUCCESS:
					break;
				default:
					goto end_scan;
				}
			} else {
				/* Found a matching record. Lock only
				a record because we can allow inserts
				into gaps */

				err = row_ins_set_shared_rec_lock(
					LOCK_REC_NOT_GAP, block,
					rec, check_index, offsets, thr);

				switch (err) {
				case DB_SUCCESS_LOCKED_REC:
				case DB_SUCCESS:
					break;
				default:
					goto end_scan;
				}

				if (check_ref) {
					err = DB_SUCCESS;

					goto end_scan;
				} else if (foreign->type != 0) {
					/* There is an ON UPDATE or ON DELETE
					condition: check them in a separate
					function */

					err = row_ins_foreign_check_on_constraint(
						thr, foreign, &pcur, entry,
						&mtr);
					if (err != DB_SUCCESS) {
						/* Since reporting a plain
						"duplicate key" error
						message to the user in
						cases where a long CASCADE
						operation would lead to a
						duplicate key in some
						other table is very
						confusing, map duplicate
						key errors resulting from
						FK constraints to a
						separate error code. */

						if (err == DB_DUPLICATE_KEY) {
							err = DB_FOREIGN_DUPLICATE_KEY;
						}

						goto end_scan;
					}

					/* row_ins_foreign_check_on_constraint
					may have repositioned pcur on a
					different block */
					block = btr_pcur_get_block(&pcur);
				} else {
					row_ins_foreign_report_err(
						"Trying to delete or update",
						thr, foreign, rec, entry);

					err = DB_ROW_IS_REFERENCED;
					goto end_scan;
				}
			}
		} else {
			ut_a(cmp < 0);

			err = row_ins_set_shared_rec_lock(
				LOCK_GAP, block,
				rec, check_index, offsets, thr);

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
			case DB_SUCCESS:
				if (check_ref) {
					err = DB_NO_REFERENCED_ROW;
					row_ins_foreign_report_add_err(
						trx, foreign, rec, entry);
				} else {
					err = DB_SUCCESS;
				}
			}

			goto end_scan;
		}
	} while (btr_pcur_move_to_next(&pcur, &mtr));

	if (check_ref) {
		row_ins_foreign_report_add_err(
			trx, foreign, btr_pcur_get_rec(&pcur), entry);
		err = DB_NO_REFERENCED_ROW;
	} else {
		err = DB_SUCCESS;
	}

end_scan:
	btr_pcur_close(&pcur);

	mtr_commit(&mtr);

	/* Restore old value */
	dtuple_set_n_fields_cmp(entry, n_fields_cmp);

do_possible_lock_wait:
	if (err == DB_LOCK_WAIT) {
		trx->error_state = err;

		que_thr_stop_for_mysql(thr);

		srv_suspend_mysql_thread(thr);

		if (trx->error_state == DB_SUCCESS) {

			goto run_again;
		}

		err = trx->error_state;
	}

exit_func:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(err);
}

/***************************************************************//**
Checks if foreign key constraints fail for an index entry. If index
is not mentioned in any constraint, this function does nothing,
Otherwise does searches to the indexes of referenced tables and
sets shared locks which lock either the success or the failure of
a constraint.
@return	DB_SUCCESS or error code */
static
ulint
row_ins_check_foreign_constraints(
/*==============================*/
	dict_table_t*	table,	/*!< in: table */
	dict_index_t*	index,	/*!< in: index */
	dtuple_t*	entry,	/*!< in: index entry for index */
	que_thr_t*	thr)	/*!< in: query thread */
{
	dict_foreign_t*	foreign;
	ulint		err;
	trx_t*		trx;
	ibool		got_s_lock	= FALSE;

	trx = thr_get_trx(thr);

	foreign = UT_LIST_GET_FIRST(table->foreign_list);

	while (foreign) {
		if (foreign->foreign_index == index) {

			if (foreign->referenced_table == NULL) {
				dict_table_get(foreign->referenced_table_name,
					       FALSE);
			}

			if (0 == trx->dict_operation_lock_mode) {
				got_s_lock = TRUE;

				row_mysql_freeze_data_dictionary(trx);
			}

			if (foreign->referenced_table) {
				mutex_enter(&(dict_sys->mutex));

				(foreign->referenced_table
				 ->n_foreign_key_checks_running)++;

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

			/* NOTE that if the thread ends up waiting for a lock
			we will release dict_operation_lock temporarily!
			But the counter on the table protects the referenced
			table from being dropped while the check is running. */

			err = row_ins_check_foreign_constraint(
				TRUE, foreign, table, entry, thr);

			if (foreign->referenced_table) {
				mutex_enter(&(dict_sys->mutex));

				ut_a(foreign->referenced_table
				     ->n_foreign_key_checks_running > 0);
				(foreign->referenced_table
				 ->n_foreign_key_checks_running)--;

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

			if (got_s_lock) {
				row_mysql_unfreeze_data_dictionary(trx);
			}

			if (err != DB_SUCCESS) {
				return(err);
			}
		}

		foreign = UT_LIST_GET_NEXT(foreign_list, foreign);
	}

	return(DB_SUCCESS);
}

/***************************************************************//**
Checks if a unique key violation to rec would occur at the index entry
insert.
@return	TRUE if error */
static
ibool
row_ins_dupl_error_with_rec(
/*========================*/
	const rec_t*	rec,	/*!< in: user record; NOTE that we assume
				that the caller already has a record lock on
				the record! */
	const dtuple_t*	entry,	/*!< in: entry to insert */
	dict_index_t*	index,	/*!< in: index */
	const ulint*	offsets)/*!< in: rec_get_offsets(rec, index) */
{
	ulint	matched_fields;
	ulint	matched_bytes;
	ulint	n_unique;
	ulint	i;

	ut_ad(rec_offs_validate(rec, index, offsets));

	n_unique = dict_index_get_n_unique(index);

	matched_fields = 0;
	matched_bytes = 0;

	cmp_dtuple_rec_with_match(entry, rec, offsets,
				  &matched_fields, &matched_bytes);

	if (matched_fields < n_unique) {

		return(FALSE);
	}

	/* In a unique secondary index we allow equal key values if they
	contain SQL NULLs */

	if (!dict_index_is_clust(index)) {

		for (i = 0; i < n_unique; i++) {
			if (UNIV_SQL_NULL == dfield_get_len(
				    dtuple_get_nth_field(entry, i))) {

				return(FALSE);
			}
		}
	}

	return(!rec_get_deleted_flag(rec, rec_offs_comp(offsets)));
}

/***************************************************************//**
Scans a unique non-clustered index at a given index entry to determine
whether a uniqueness violation has occurred for the key value of the entry.
Set shared locks on possible duplicate records.
@return	DB_SUCCESS, DB_DUPLICATE_KEY, or DB_LOCK_WAIT */
static
ulint
row_ins_scan_sec_index_for_duplicate(
/*=================================*/
	dict_index_t*	index,	/*!< in: non-clustered unique index */
	dtuple_t*	entry,	/*!< in: index entry */
	que_thr_t*	thr)	/*!< in: query thread */
{
	ulint		n_unique;
	ulint		i;
	int		cmp;
	ulint		n_fields_cmp;
	btr_pcur_t	pcur;
	ulint		err		= DB_SUCCESS;
	ulint		allow_duplicates;
	mtr_t		mtr;
	mem_heap_t*	heap		= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets		= offsets_;
	rec_offs_init(offsets_);

	n_unique = dict_index_get_n_unique(index);

	/* If the secondary index is unique, but one of the fields in the
	n_unique first fields is NULL, a unique key violation cannot occur,
	since we define NULL != NULL in this case */

	for (i = 0; i < n_unique; i++) {
		if (UNIV_SQL_NULL == dfield_get_len(
			    dtuple_get_nth_field(entry, i))) {

			return(DB_SUCCESS);
		}
	}

	mtr_start(&mtr);

	/* Store old value on n_fields_cmp */

	n_fields_cmp = dtuple_get_n_fields_cmp(entry);

	dtuple_set_n_fields_cmp(entry, dict_index_get_n_unique(index));

	btr_pcur_open(index, entry, PAGE_CUR_GE, BTR_SEARCH_LEAF, &pcur, &mtr);

	allow_duplicates = thr_get_trx(thr)->duplicates;

	/* Scan index records and check if there is a duplicate */

	do {
		const rec_t*		rec	= btr_pcur_get_rec(&pcur);
		const buf_block_t*	block	= btr_pcur_get_block(&pcur);

		if (page_rec_is_infimum(rec)) {

			continue;
		}

		offsets = rec_get_offsets(rec, index, offsets,
					  ULINT_UNDEFINED, &heap);

		if (allow_duplicates) {

			/* If the SQL-query will update or replace
			duplicate key we will take X-lock for
			duplicates ( REPLACE, LOAD DATAFILE REPLACE,
			INSERT ON DUPLICATE KEY UPDATE). */

			err = row_ins_set_exclusive_rec_lock(
				LOCK_ORDINARY, block,
				rec, index, offsets, thr);
		} else {

			err = row_ins_set_shared_rec_lock(
				LOCK_ORDINARY, block,
				rec, index, offsets, thr);
		}

		switch (err) {
		case DB_SUCCESS_LOCKED_REC:
			err = DB_SUCCESS;
		case DB_SUCCESS:
			break;
		default:
			goto end_scan;
		}

		if (page_rec_is_supremum(rec)) {

			continue;
		}

		cmp = cmp_dtuple_rec(entry, rec, offsets);

		if (cmp == 0) {
			if (row_ins_dupl_error_with_rec(rec, entry,
							index, offsets)) {
				err = DB_DUPLICATE_KEY;

				thr_get_trx(thr)->error_info = index;

				goto end_scan;
			}
		} else {
			ut_a(cmp < 0);
			goto end_scan;
		}
	} while (btr_pcur_move_to_next(&pcur, &mtr));

end_scan:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	mtr_commit(&mtr);

	/* Restore old value */
	dtuple_set_n_fields_cmp(entry, n_fields_cmp);

	return(err);
}

/***************************************************************//**
Checks if a unique key violation error would occur at an index entry
insert. Sets shared locks on possible duplicate records. Works only
for a clustered index!
@return DB_SUCCESS if no error, DB_DUPLICATE_KEY if error,
DB_LOCK_WAIT if we have to wait for a lock on a possible duplicate
record */
static
ulint
row_ins_duplicate_error_in_clust(
/*=============================*/
	btr_cur_t*	cursor,	/*!< in: B-tree cursor */
	const dtuple_t*	entry,	/*!< in: entry to insert */
	que_thr_t*	thr,	/*!< in: query thread */
	mtr_t*		mtr)	/*!< in: mtr */
{
	ulint	err;
	rec_t*	rec;
	ulint	n_unique;
	trx_t*	trx		= thr_get_trx(thr);
	mem_heap_t*heap		= NULL;
	ulint	offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*	offsets		= offsets_;
	rec_offs_init(offsets_);

	UT_NOT_USED(mtr);

	ut_a(dict_index_is_clust(cursor->index));
	ut_ad(dict_index_is_unique(cursor->index));

	/* NOTE: For unique non-clustered indexes there may be any number
	of delete marked records with the same value for the non-clustered
	index key (remember multiversioning), and which differ only in
	the row refererence part of the index record, containing the
	clustered index key fields. For such a secondary index record,
	to avoid race condition, we must FIRST do the insertion and after
	that check that the uniqueness condition is not breached! */

	/* NOTE: A problem is that in the B-tree node pointers on an
	upper level may match more to the entry than the actual existing
	user records on the leaf level. So, even if low_match would suggest
	that a duplicate key violation may occur, this may not be the case. */

	n_unique = dict_index_get_n_unique(cursor->index);

	if (cursor->low_match >= n_unique) {

		rec = btr_cur_get_rec(cursor);

		if (!page_rec_is_infimum(rec)) {
			offsets = rec_get_offsets(rec, cursor->index, offsets,
						  ULINT_UNDEFINED, &heap);

			/* We set a lock on the possible duplicate: this
			is needed in logical logging of MySQL to make
			sure that in roll-forward we get the same duplicate
			errors as in original execution */

			if (trx->duplicates) {

				/* If the SQL-query will update or replace
				duplicate key we will take X-lock for
				duplicates ( REPLACE, LOAD DATAFILE REPLACE,
				INSERT ON DUPLICATE KEY UPDATE). */

				err = row_ins_set_exclusive_rec_lock(
					LOCK_REC_NOT_GAP,
					btr_cur_get_block(cursor),
					rec, cursor->index, offsets, thr);
			} else {

				err = row_ins_set_shared_rec_lock(
					LOCK_REC_NOT_GAP,
					btr_cur_get_block(cursor), rec,
					cursor->index, offsets, thr);
			}

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
			case DB_SUCCESS:
				break;
			default:
				goto func_exit;
			}

			if (row_ins_dupl_error_with_rec(
				    rec, entry, cursor->index, offsets)) {
				trx->error_info = cursor->index;
				err = DB_DUPLICATE_KEY;
				goto func_exit;
			}
		}
	}

	if (cursor->up_match >= n_unique) {

		rec = page_rec_get_next(btr_cur_get_rec(cursor));

		if (!page_rec_is_supremum(rec)) {
			offsets = rec_get_offsets(rec, cursor->index, offsets,
						  ULINT_UNDEFINED, &heap);

			if (trx->duplicates) {

				/* If the SQL-query will update or replace
				duplicate key we will take X-lock for
				duplicates ( REPLACE, LOAD DATAFILE REPLACE,
				INSERT ON DUPLICATE KEY UPDATE). */

				err = row_ins_set_exclusive_rec_lock(
					LOCK_REC_NOT_GAP,
					btr_cur_get_block(cursor),
					rec, cursor->index, offsets, thr);
			} else {

				err = row_ins_set_shared_rec_lock(
					LOCK_REC_NOT_GAP,
					btr_cur_get_block(cursor),
					rec, cursor->index, offsets, thr);
			}

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
			case DB_SUCCESS:
				break;
			default:
				goto func_exit;
			}

			if (row_ins_dupl_error_with_rec(
				    rec, entry, cursor->index, offsets)) {
				trx->error_info = cursor->index;
				err = DB_DUPLICATE_KEY;
				goto func_exit;
			}
		}

		ut_a(!dict_index_is_clust(cursor->index));
		/* This should never happen */
	}

	err = DB_SUCCESS;
func_exit:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(err);
}

/***************************************************************//**
Checks if an index entry has long enough common prefix with an existing
record so that the intended insert of the entry must be changed to a modify of
the existing record. In the case of a clustered index, the prefix must be
n_unique fields long, and in the case of a secondary index, all fields must be
equal.
@return 0 if no update, ROW_INS_PREV if previous should be updated;
currently we do the search so that only the low_match record can match
enough to the search tuple, not the next record */
UNIV_INLINE
ulint
row_ins_must_modify(
/*================*/
	btr_cur_t*	cursor)	/*!< in: B-tree cursor */
{
	ulint	enough_match;
	rec_t*	rec;

	/* NOTE: (compare to the note in row_ins_duplicate_error) Because node
	pointers on upper levels of the B-tree may match more to entry than
	to actual user records on the leaf level, we have to check if the
	candidate record is actually a user record. In a clustered index
	node pointers contain index->n_unique first fields, and in the case
	of a secondary index, all fields of the index. */

	enough_match = dict_index_get_n_unique_in_tree(cursor->index);

	if (cursor->low_match >= enough_match) {

		rec = btr_cur_get_rec(cursor);

		if (!page_rec_is_infimum(rec)) {

			return(ROW_INS_PREV);
		}
	}

	return(0);
}

/***************************************************************//**
Tries to insert an index entry to an index. If the index is clustered
and a record with the same unique key is found, the other record is
necessarily marked deleted by a committed transaction, or a unique key
violation error occurs. The delete marked record is then updated to an
existing record, and we must write an undo log record on the delete
marked record. If the index is secondary, and a record with exactly the
same fields is found, the other record is necessarily marked deleted.
It is then unmarked. Otherwise, the entry is just inserted to the index.
@return DB_SUCCESS, DB_LOCK_WAIT, DB_FAIL if pessimistic retry needed,
or error code */
static
ulint
row_ins_index_entry_low(
/*====================*/
	ulint		mode,	/*!< in: BTR_MODIFY_LEAF or BTR_MODIFY_TREE,
				depending on whether we wish optimistic or
				pessimistic descent down the index tree */
	dict_index_t*	index,	/*!< in: index */
	dtuple_t*	entry,	/*!< in/out: index entry to insert */
	ulint		n_ext,	/*!< in: number of externally stored columns */
	que_thr_t*	thr)	/*!< in: query thread */
{
	btr_cur_t	cursor;
	ulint		ignore_sec_unique	= 0;
	ulint		modify = 0; /* remove warning */
	rec_t*		insert_rec;
	rec_t*		rec;
	ulint*		offsets;
	ulint		err;
	ulint		n_unique;
	big_rec_t*	big_rec			= NULL;
	mtr_t		mtr;
	mem_heap_t*	heap			= NULL;

	log_free_check();

	mtr_start(&mtr);

	cursor.thr = thr;

	/* Note that we use PAGE_CUR_LE as the search mode, because then
	the function will return in both low_match and up_match of the
	cursor sensible values */

	if (!(thr_get_trx(thr)->check_unique_secondary)) {
		ignore_sec_unique = BTR_IGNORE_SEC_UNIQUE;
	}

	btr_cur_search_to_nth_level(index, 0, entry, PAGE_CUR_LE,
				    mode | BTR_INSERT | ignore_sec_unique,
				    &cursor, 0, __FILE__, __LINE__, &mtr);

	if (cursor.flag == BTR_CUR_INSERT_TO_IBUF) {
		/* The insertion was made to the insert buffer already during
		the search: we are done */

		err = DB_SUCCESS;

		goto function_exit;
	}

#ifdef UNIV_DEBUG
	{
		page_t*	page = btr_cur_get_page(&cursor);
		rec_t*	first_rec = page_rec_get_next(
			page_get_infimum_rec(page));

		ut_ad(page_rec_is_supremum(first_rec)
		      || rec_get_n_fields(first_rec, index)
		      == dtuple_get_n_fields(entry));
	}
#endif

	n_unique = dict_index_get_n_unique(index);

	if (dict_index_is_unique(index) && (cursor.up_match >= n_unique
					    || cursor.low_match >= n_unique)) {

		if (dict_index_is_clust(index)) {
			/* Note that the following may return also
			DB_LOCK_WAIT */

			err = row_ins_duplicate_error_in_clust(
				&cursor, entry, thr, &mtr);
			if (err != DB_SUCCESS) {

				goto function_exit;
			}
		} else {
			mtr_commit(&mtr);
			err = row_ins_scan_sec_index_for_duplicate(
				index, entry, thr);
			mtr_start(&mtr);

			if (err != DB_SUCCESS) {

				goto function_exit;
			}

			/* We did not find a duplicate and we have now
			locked with s-locks the necessary records to
			prevent any insertion of a duplicate by another
			transaction. Let us now reposition the cursor and
			continue the insertion. */

			btr_cur_search_to_nth_level(index, 0, entry,
						    PAGE_CUR_LE,
						    mode | BTR_INSERT,
						    &cursor, 0,
						    __FILE__, __LINE__, &mtr);
		}
	}

	modify = row_ins_must_modify(&cursor);

	if (modify != 0) {
		/* There is already an index entry with a long enough common
		prefix, we must convert the insert into a modify of an
		existing record */

		if (modify == ROW_INS_NEXT) {
			rec = page_rec_get_next(btr_cur_get_rec(&cursor));

			btr_cur_position(index, rec,
					 btr_cur_get_block(&cursor),&cursor);
		}

		if (dict_index_is_clust(index)) {
			err = row_ins_clust_index_entry_by_modify(
				mode, &cursor, &heap, &big_rec, entry,
				thr, &mtr);

			if (big_rec) {
				ut_a(err == DB_SUCCESS);
				/* Write out the externally stored
				columns while still x-latching
				index->lock and block->lock. Allocate
				pages for big_rec in the mtr that
				modified the B-tree, but be sure to skip
				any pages that were freed in mtr. We will
				write out the big_rec pages before
				committing the B-tree mini-transaction. If
				the system crashes so that crash recovery
				will not replay the mtr_commit(&mtr), the
				big_rec pages will be left orphaned until
				the pages are allocated for something else.

				TODO: If the allocation extends the
				tablespace, it will not be redo
				logged, in either mini-transaction.
				Tablespace extension should be
				redo-logged in the big_rec
				mini-transaction, so that recovery
				will not fail when the big_rec was
				written to the extended portion of the
				file, in case the file was somehow
				truncated in the crash. */

				rec = btr_cur_get_rec(&cursor);
				offsets = rec_get_offsets(
					rec, index, NULL,
					ULINT_UNDEFINED, &heap);

				DEBUG_SYNC_C("before_row_ins_upd_extern");
				err = btr_store_big_rec_extern_fields(
					index, btr_cur_get_block(&cursor),
					rec, offsets, big_rec, &mtr,
					BTR_STORE_INSERT_UPDATE);
				DEBUG_SYNC_C("after_row_ins_upd_extern");
				/* If writing big_rec fails (for
				example, because of DB_OUT_OF_FILE_SPACE),
				the record will be corrupted. Even if
				we did not update any externally
				stored columns, our update could cause
				the record to grow so that a
				non-updated column was selected for
				external storage. This non-update
				would not have been written to the
				undo log, and thus the record cannot
				be rolled back.

				However, because we have not executed
				mtr_commit(mtr) yet, the update will
				not be replayed in crash recovery, and
				the following assertion failure will
				effectively "roll back" the operation. */
				ut_a(err == DB_SUCCESS);
				goto stored_big_rec;
			}
		} else {
			ut_ad(!n_ext);
			err = row_ins_sec_index_entry_by_modify(
				mode, &cursor, entry, thr, &mtr);
		}
	} else {
		if (mode == BTR_MODIFY_LEAF) {
			err = btr_cur_optimistic_insert(
				0, &cursor, entry, &insert_rec, &big_rec,
				n_ext, thr, &mtr);
		} else {
			ut_a(mode == BTR_MODIFY_TREE);
			if (buf_LRU_buf_pool_running_out()) {

				err = DB_LOCK_TABLE_FULL;

				goto function_exit;
			}

			err = btr_cur_optimistic_insert(
				0, &cursor, entry, &insert_rec, &big_rec,
				n_ext, thr, &mtr);

			if (err == DB_FAIL) {
				err = btr_cur_pessimistic_insert(
					0, &cursor, entry, &insert_rec,
					&big_rec, n_ext, thr, &mtr);
			}
		}
	}

function_exit:
	mtr_commit(&mtr);

	if (UNIV_LIKELY_NULL(big_rec)) {
		DBUG_EXECUTE_IF(
			"row_ins_extern_checkpoint",
			log_make_checkpoint_at(IB_ULONGLONG_MAX, TRUE););

		mtr_start(&mtr);

		DEBUG_SYNC_C("before_row_ins_extern_latch");
		btr_cur_search_to_nth_level(index, 0, entry, PAGE_CUR_LE,
					    BTR_MODIFY_TREE, &cursor, 0,
					    __FILE__, __LINE__, &mtr);
		rec = btr_cur_get_rec(&cursor);
		offsets = rec_get_offsets(rec, index, NULL,
					  ULINT_UNDEFINED, &heap);

		DEBUG_SYNC_C("before_row_ins_extern");
		err = btr_store_big_rec_extern_fields(
			index, btr_cur_get_block(&cursor),
			rec, offsets, big_rec, &mtr, BTR_STORE_INSERT);
		DEBUG_SYNC_C("after_row_ins_extern");

stored_big_rec:
		if (modify) {
			dtuple_big_rec_free(big_rec);
		} else {
			dtuple_convert_back_big_rec(index, entry, big_rec);
		}

		mtr_commit(&mtr);
	}

	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(err);
}

/***************************************************************//**
Inserts an index entry to index. Tries first optimistic, then pessimistic
descent down the tree. If the entry matches enough to a delete marked record,
performs the insert by updating or delete unmarking the delete marked
record.
@return	DB_SUCCESS, DB_LOCK_WAIT, DB_DUPLICATE_KEY, or some other error code */
UNIV_INTERN
ulint
row_ins_index_entry(
/*================*/
	dict_index_t*	index,	/*!< in: index */
	dtuple_t*	entry,	/*!< in/out: index entry to insert */
	ulint		n_ext,	/*!< in: number of externally stored columns */
	ibool		foreign,/*!< in: TRUE=check foreign key constraints
				(foreign=FALSE only during CREATE INDEX) */
	que_thr_t*	thr)	/*!< in: query thread */
{
	ulint	err;

	if (foreign && UT_LIST_GET_FIRST(index->table->foreign_list)) {
		err = row_ins_check_foreign_constraints(index->table, index,
							entry, thr);
		if (err != DB_SUCCESS) {

			return(err);
		}
	}

	/* Try first optimistic descent to the B-tree */

	err = row_ins_index_entry_low(BTR_MODIFY_LEAF, index, entry,
				      n_ext, thr);
	if (err != DB_FAIL) {
		if (index == dict_table_get_first_index(index->table)
		    && thr_get_trx(thr)->mysql_thd != 0) {
			DEBUG_SYNC_C("row_ins_clust_index_entry_leaf_after");
		}
		return(err);
	}

	/* Try then pessimistic descent to the B-tree */

	err = row_ins_index_entry_low(BTR_MODIFY_TREE, index, entry,
				      n_ext, thr);
	return(err);
}

/***********************************************************//**
Sets the values of the dtuple fields in entry from the values of appropriate
columns in row. */
static
void
row_ins_index_entry_set_vals(
/*=========================*/
	dict_index_t*	index,	/*!< in: index */
	dtuple_t*	entry,	/*!< in: index entry to make */
	const dtuple_t*	row)	/*!< in: row */
{
	ulint	n_fields;
	ulint	i;

	ut_ad(entry && row);

	n_fields = dtuple_get_n_fields(entry);

	for (i = 0; i < n_fields; i++) {
		dict_field_t*	ind_field;
		dfield_t*	field;
		const dfield_t*	row_field;
		ulint		len;

		field = dtuple_get_nth_field(entry, i);
		ind_field = dict_index_get_nth_field(index, i);
		row_field = dtuple_get_nth_field(row, ind_field->col->ind);
		len = dfield_get_len(row_field);

		/* Check column prefix indexes */
		if (ind_field->prefix_len > 0
		    && dfield_get_len(row_field) != UNIV_SQL_NULL) {

			const	dict_col_t*	col
				= dict_field_get_col(ind_field);

			len = dtype_get_at_most_n_mbchars(
				col->prtype, col->mbminlen, col->mbmaxlen,
				ind_field->prefix_len,
				len, dfield_get_data(row_field));

			ut_ad(!dfield_is_ext(row_field));
		}

		dfield_set_data(field, dfield_get_data(row_field), len);
		if (dfield_is_ext(row_field)) {
			ut_ad(dict_index_is_clust(index));
			dfield_set_ext(field);
		}
	}
}

/***********************************************************//**
Inserts a single index entry to the table.
@return DB_SUCCESS if operation successfully completed, else error
code or DB_LOCK_WAIT */
static
ulint
row_ins_index_entry_step(
/*=====================*/
	ins_node_t*	node,	/*!< in: row insert node */
	que_thr_t*	thr)	/*!< in: query thread */
{
	ulint	err;

	ut_ad(dtuple_check_typed(node->row));

	row_ins_index_entry_set_vals(node->index, node->entry, node->row);

	ut_ad(dtuple_check_typed(node->entry));

	err = row_ins_index_entry(node->index, node->entry, 0, TRUE, thr);

	return(err);
}

/***********************************************************//**
Allocates a row id for row and inits the node->index field. */
UNIV_INLINE
void
row_ins_alloc_row_id_step(
/*======================*/
	ins_node_t*	node)	/*!< in: row insert node */
{
	dulint	row_id;

	ut_ad(node->state == INS_NODE_ALLOC_ROW_ID);

	if (dict_index_is_unique(dict_table_get_first_index(node->table))) {

		/* No row id is stored if the clustered index is unique */

		return;
	}

	/* Fill in row id value to row */

	row_id = dict_sys_get_new_row_id();

	dict_sys_write_row_id(node->row_id_buf, row_id);
}

/***********************************************************//**
Gets a row to insert from the values list. */
UNIV_INLINE
void
row_ins_get_row_from_values(
/*========================*/
	ins_node_t*	node)	/*!< in: row insert node */
{
	que_node_t*	list_node;
	dfield_t*	dfield;
	dtuple_t*	row;
	ulint		i;

	/* The field values are copied in the buffers of the select node and
	it is safe to use them until we fetch from select again: therefore
	we can just copy the pointers */

	row = node->row;

	i = 0;
	list_node = node->values_list;

	while (list_node) {
		eval_exp(list_node);

		dfield = dtuple_get_nth_field(row, i);
		dfield_copy_data(dfield, que_node_get_val(list_node));

		i++;
		list_node = que_node_get_next(list_node);
	}
}

/***********************************************************//**
Gets a row to insert from the select list. */
UNIV_INLINE
void
row_ins_get_row_from_select(
/*========================*/
	ins_node_t*	node)	/*!< in: row insert node */
{
	que_node_t*	list_node;
	dfield_t*	dfield;
	dtuple_t*	row;
	ulint		i;

	/* The field values are copied in the buffers of the select node and
	it is safe to use them until we fetch from select again: therefore
	we can just copy the pointers */

	row = node->row;

	i = 0;
	list_node = node->select->select_list;

	while (list_node) {
		dfield = dtuple_get_nth_field(row, i);
		dfield_copy_data(dfield, que_node_get_val(list_node));

		i++;
		list_node = que_node_get_next(list_node);
	}
}

/***********************************************************//**
Inserts a row to a table.
@return DB_SUCCESS if operation successfully completed, else error
code or DB_LOCK_WAIT */
static
ulint
row_ins(
/*====*/
	ins_node_t*	node,	/*!< in: row insert node */
	que_thr_t*	thr)	/*!< in: query thread */
{
	ulint	err;

	ut_ad(node && thr);

	if (node->state == INS_NODE_ALLOC_ROW_ID) {

		row_ins_alloc_row_id_step(node);

		node->index = dict_table_get_first_index(node->table);
		node->entry = UT_LIST_GET_FIRST(node->entry_list);

		if (node->ins_type == INS_SEARCHED) {

			row_ins_get_row_from_select(node);

		} else if (node->ins_type == INS_VALUES) {

			row_ins_get_row_from_values(node);
		}

		node->state = INS_NODE_INSERT_ENTRIES;
	}

	ut_ad(node->state == INS_NODE_INSERT_ENTRIES);

	while (node->index != NULL) {
		err = row_ins_index_entry_step(node, thr);

		if (err != DB_SUCCESS) {

			return(err);
		}

		node->index = dict_table_get_next_index(node->index);
		node->entry = UT_LIST_GET_NEXT(tuple_list, node->entry);
	}

	ut_ad(node->entry == NULL);

	node->state = INS_NODE_ALLOC_ROW_ID;

	return(DB_SUCCESS);
}

/***********************************************************//**
Inserts a row to a table. This is a high-level function used in SQL execution
graphs.
@return	query thread to run next or NULL */
UNIV_INTERN
que_thr_t*
row_ins_step(
/*=========*/
	que_thr_t*	thr)	/*!< in: query thread */
{
	ins_node_t*	node;
	que_node_t*	parent;
	sel_node_t*	sel_node;
	trx_t*		trx;
	ulint		err;

	ut_ad(thr);

	trx = thr_get_trx(thr);

	trx_start_if_not_started(trx);

	node = thr->run_node;

	ut_ad(que_node_get_type(node) == QUE_NODE_INSERT);

	parent = que_node_get_parent(node);
	sel_node = node->select;

	if (thr->prev_node == parent) {
		node->state = INS_NODE_SET_IX_LOCK;
	}

	/* If this is the first time this node is executed (or when
	execution resumes after wait for the table IX lock), set an
	IX lock on the table and reset the possible select node. MySQL's
	partitioned table code may also call an insert within the same
	SQL statement AFTER it has used this table handle to do a search.
	This happens, for example, when a row update moves it to another
	partition. In that case, we have already set the IX lock on the
	table during the search operation, and there is no need to set
	it again here. But we must write trx->id to node->trx_id_buf. */

	trx_write_trx_id(node->trx_id_buf, trx->id);

	if (node->state == INS_NODE_SET_IX_LOCK) {

		/* It may be that the current session has not yet started
		its transaction, or it has been committed: */

		if (UT_DULINT_EQ(trx->id, node->trx_id)) {
			/* No need to do IX-locking */

			goto same_trx;
		}

		err = lock_table(0, node->table, LOCK_IX, thr);

		if (err != DB_SUCCESS) {

			goto error_handling;
		}

		node->trx_id = trx->id;
same_trx:
		node->state = INS_NODE_ALLOC_ROW_ID;

		if (node->ins_type == INS_SEARCHED) {
			/* Reset the cursor */
			sel_node->state = SEL_NODE_OPEN;

			/* Fetch a row to insert */

			thr->run_node = sel_node;

			return(thr);
		}
	}

	if ((node->ins_type == INS_SEARCHED)
	    && (sel_node->state != SEL_NODE_FETCH)) {

		ut_ad(sel_node->state == SEL_NODE_NO_MORE_ROWS);

		/* No more rows to insert */
		thr->run_node = parent;

		return(thr);
	}

	/* DO THE CHECKS OF THE CONSISTENCY CONSTRAINTS HERE */

	err = row_ins(node, thr);

error_handling:
	trx->error_state = err;

	if (err != DB_SUCCESS) {
		/* err == DB_LOCK_WAIT or SQL error detected */
		return(NULL);
	}

	/* DO THE TRIGGER ACTIONS HERE */

	if (node->ins_type == INS_SEARCHED) {
		/* Fetch a row to insert */

		thr->run_node = sel_node;
	} else {
		thr->run_node = que_node_get_parent(node);
	}

	return(thr);
}