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
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
7134
7135
7136
7137
7138
7139
7140
7141
7142
7143
7144
7145
7146
7147
7148
7149
7150
7151
7152
7153
7154
7155
7156
7157
7158
7159
7160
7161
7162
7163
7164
7165
7166
7167
7168
7169
7170
7171
7172
7173
7174
7175
7176
7177
7178
7179
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
7195
7196
7197
7198
7199
7200
7201
7202
7203
7204
7205
7206
7207
7208
7209
7210
7211
7212
7213
7214
7215
7216
7217
7218
7219
7220
7221
7222
7223
7224
7225
7226
7227
7228
7229
7230
7231
7232
7233
7234
7235
7236
7237
7238
7239
7240
7241
7242
7243
7244
7245
7246
7247
7248
7249
7250
7251
7252
7253
7254
7255
7256
7257
7258
7259
7260
7261
7262
7263
7264
7265
7266
7267
7268
7269
7270
7271
7272
7273
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
7289
7290
7291
7292
7293
7294
7295
7296
7297
7298
7299
7300
7301
7302
7303
7304
7305
7306
7307
7308
7309
7310
7311
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
7332
7333
7334
7335
7336
7337
7338
7339
7340
7341
7342
7343
7344
7345
7346
7347
7348
7349
7350
7351
7352
7353
7354
7355
7356
7357
7358
7359
7360
7361
7362
7363
7364
7365
7366
7367
7368
7369
7370
7371
7372
7373
7374
7375
7376
7377
7378
7379
7380
7381
7382
7383
7384
7385
7386
7387
7388
7389
7390
7391
7392
7393
7394
7395
7396
7397
7398
7399
7400
7401
7402
7403
7404
7405
7406
7407
7408
7409
7410
7411
7412
7413
7414
7415
7416
7417
7418
7419
7420
7421
7422
7423
7424
7425
7426
7427
7428
7429
7430
7431
7432
7433
7434
7435
7436
7437
7438
7439
7440
7441
7442
7443
7444
7445
7446
7447
7448
7449
7450
7451
7452
7453
7454
7455
7456
7457
7458
7459
7460
7461
7462
7463
7464
7465
7466
7467
7468
7469
7470
7471
7472
7473
7474
7475
7476
7477
7478
7479
7480
7481
7482
7483
7484
7485
7486
7487
7488
7489
7490
7491
7492
7493
7494
7495
7496
7497
7498
7499
7500
7501
7502
7503
7504
7505
7506
7507
7508
7509
7510
7511
7512
7513
7514
7515
7516
7517
7518
7519
7520
7521
7522
7523
7524
7525
7526
7527
7528
7529
7530
7531
7532
7533
7534
7535
7536
7537
7538
7539
7540
7541
7542
7543
7544
7545
7546
7547
7548
7549
7550
7551
7552
7553
7554
7555
7556
7557
7558
7559
7560
7561
7562
7563
7564
7565
7566
7567
7568
7569
7570
7571
7572
7573
7574
7575
7576
7577
7578
7579
7580
7581
7582
7583
7584
7585
7586
7587
7588
7589
7590
7591
7592
7593
7594
7595
7596
7597
7598
7599
7600
7601
7602
7603
7604
7605
7606
7607
7608
7609
7610
7611
7612
7613
7614
7615
7616
7617
7618
7619
7620
7621
7622
7623
7624
7625
7626
7627
7628
7629
7630
7631
7632
7633
7634
7635
7636
7637
7638
7639
7640
7641
7642
7643
7644
7645
7646
7647
7648
7649
7650
7651
7652
7653
7654
7655
7656
7657
7658
7659
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
7675
7676
7677
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
7693
7694
7695
7696
7697
7698
7699
7700
7701
7702
7703
7704
7705
7706
7707
7708
7709
7710
7711
7712
7713
7714
7715
7716
7717
7718
7719
7720
7721
7722
7723
7724
7725
7726
7727
7728
7729
7730
7731
7732
7733
7734
7735
7736
7737
7738
7739
7740
7741
7742
7743
7744
7745
7746
7747
7748
7749
7750
7751
7752
7753
7754
7755
7756
7757
7758
7759
7760
7761
7762
7763
7764
7765
7766
7767
7768
7769
7770
7771
7772
7773
7774
7775
7776
7777
7778
7779
7780
7781
7782
7783
7784
7785
7786
7787
7788
7789
7790
7791
7792
7793
7794
7795
7796
7797
7798
7799
7800
7801
7802
7803
7804
7805
7806
7807
7808
7809
7810
7811
7812
7813
7814
7815
7816
7817
7818
7819
7820
7821
7822
7823
7824
7825
7826
7827
7828
7829
7830
7831
7832
7833
7834
7835
7836
7837
7838
7839
7840
7841
7842
7843
7844
7845
7846
7847
7848
7849
7850
7851
7852
7853
7854
7855
7856
7857
7858
7859
7860
7861
7862
7863
7864
7865
7866
7867
7868
7869
7870
7871
7872
7873
7874
7875
7876
7877
7878
7879
7880
7881
7882
7883
7884
7885
7886
7887
7888
7889
7890
7891
7892
7893
7894
7895
7896
7897
7898
7899
7900
7901
7902
7903
7904
7905
7906
7907
7908
7909
7910
7911
7912
7913
7914
7915
7916
7917
7918
7919
7920
7921
7922
7923
7924
7925
7926
7927
7928
7929
7930
7931
7932
7933
7934
7935
7936
7937
7938
7939
7940
7941
7942
7943
7944
7945
7946
7947
7948
7949
7950
7951
7952
7953
7954
7955
7956
7957
7958
7959
7960
7961
7962
7963
7964
7965
7966
7967
7968
7969
7970
7971
7972
7973
7974
7975
7976
7977
7978
7979
7980
7981
7982
7983
7984
7985
7986
7987
7988
7989
7990
7991
7992
7993
7994
7995
7996
7997
7998
7999
8000
8001
8002
8003
8004
8005
8006
8007
8008
8009
8010
8011
8012
8013
8014
8015
8016
8017
8018
8019
8020
8021
8022
8023
8024
8025
8026
8027
8028
8029
8030
8031
8032
8033
8034
8035
8036
8037
8038
8039
8040
8041
8042
8043
8044
8045
8046
8047
8048
8049
8050
8051
8052
8053
8054
8055
8056
8057
8058
8059
8060
8061
8062
8063
8064
8065
8066
8067
8068
8069
8070
8071
8072
8073
8074
8075
8076
8077
8078
8079
8080
8081
8082
8083
8084
8085
8086
8087
8088
8089
8090
8091
8092
8093
8094
8095
8096
8097
8098
8099
8100
8101
8102
8103
8104
8105
8106
8107
8108
8109
8110
8111
8112
8113
8114
8115
8116
8117
8118
8119
8120
8121
8122
8123
8124
8125
8126
8127
8128
8129
8130
8131
8132
8133
8134
8135
8136
8137
8138
8139
8140
8141
8142
8143
8144
8145
8146
8147
8148
8149
8150
8151
8152
8153
8154
8155
8156
8157
8158
8159
8160
8161
8162
8163
8164
8165
8166
8167
8168
8169
8170
8171
8172
8173
8174
8175
8176
8177
8178
8179
8180
8181
8182
8183
8184
8185
8186
8187
8188
8189
8190
8191
8192
8193
8194
8195
8196
8197
8198
8199
8200
8201
8202
8203
8204
8205
8206
8207
8208
8209
8210
8211
8212
8213
8214
8215
8216
8217
8218
8219
8220
8221
8222
8223
8224
8225
8226
8227
8228
8229
8230
8231
8232
8233
8234
8235
8236
8237
8238
8239
8240
8241
8242
8243
8244
8245
8246
8247
8248
8249
8250
8251
8252
8253
8254
8255
8256
8257
8258
8259
8260
8261
8262
8263
8264
8265
8266
8267
8268
8269
8270
8271
8272
8273
8274
8275
8276
8277
8278
8279
8280
8281
8282
8283
8284
8285
8286
8287
8288
8289
8290
8291
8292
8293
8294
8295
8296
8297
8298
8299
8300
8301
8302
8303
8304
8305
8306
8307
8308
8309
8310
8311
8312
8313
8314
8315
8316
8317
8318
8319
8320
8321
8322
8323
8324
8325
8326
8327
8328
8329
8330
8331
8332
8333
8334
|
/*****************************************************************************
Copyright (c) 1996, 2016, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2014, 2017, MariaDB Corporation
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 lock/lock0lock.cc
The transaction lock system
Created 5/7/1996 Heikki Tuuri
*******************************************************/
#define LOCK_MODULE_IMPLEMENTATION
#include "lock0lock.h"
#include "lock0priv.h"
#ifdef UNIV_NONINL
#include "lock0lock.ic"
#include "lock0priv.ic"
#endif
#include "ha_prototypes.h"
#include "usr0sess.h"
#include "trx0purge.h"
#include "dict0mem.h"
#include "dict0boot.h"
#include "trx0sys.h"
#include "pars0pars.h" /* pars_complete_graph_for_exec() */
#include "que0que.h" /* que_node_get_parent() */
#include "row0mysql.h" /* row_mysql_handle_errors() */
#include "row0sel.h" /* sel_node_create(), sel_node_t */
#include "row0types.h" /* sel_node_t */
#include "srv0mon.h"
#include "ut0vec.h"
#include "btr0btr.h"
#include "dict0boot.h"
#include <set>
#include "mysql/plugin.h"
#include <mysql/service_wsrep.h>
#include <string>
#include <sstream>
/* Restricts the length of search we will do in the waits-for
graph of transactions */
#define LOCK_MAX_N_STEPS_IN_DEADLOCK_CHECK 1000000
/* Restricts the search depth we will do in the waits-for graph of
transactions */
#define LOCK_MAX_DEPTH_IN_DEADLOCK_CHECK 200
/* When releasing transaction locks, this specifies how often we release
the lock mutex for a moment to give also others access to it */
#define LOCK_RELEASE_INTERVAL 1000
/* Safety margin when creating a new record lock: this many extra records
can be inserted to the page without need to create a lock with a bigger
bitmap */
#define LOCK_PAGE_BITMAP_MARGIN 64
/** Lock scheduling algorithm */
ulong innodb_lock_schedule_algorithm = INNODB_LOCK_SCHEDULE_ALGORITHM_FCFS;
/* An explicit record lock affects both the record and the gap before it.
An implicit x-lock does not affect the gap, it only locks the index
record from read or update.
If a transaction has modified or inserted an index record, then
it owns an implicit x-lock on the record. On a secondary index record,
a transaction has an implicit x-lock also if it has modified the
clustered index record, the max trx id of the page where the secondary
index record resides is >= trx id of the transaction (or database recovery
is running), and there are no explicit non-gap lock requests on the
secondary index record.
This complicated definition for a secondary index comes from the
implementation: we want to be able to determine if a secondary index
record has an implicit x-lock, just by looking at the present clustered
index record, not at the historical versions of the record. The
complicated definition can be explained to the user so that there is
nondeterminism in the access path when a query is answered: we may,
or may not, access the clustered index record and thus may, or may not,
bump into an x-lock set there.
Different transaction can have conflicting locks set on the gap at the
same time. The locks on the gap are purely inhibitive: an insert cannot
be made, or a select cursor may have to wait if a different transaction
has a conflicting lock on the gap. An x-lock on the gap does not give
the right to insert into the gap.
An explicit lock can be placed on a user record or the supremum record of
a page. The locks on the supremum record are always thought to be of the gap
type, though the gap bit is not set. When we perform an update of a record
where the size of the record changes, we may temporarily store its explicit
locks on the infimum record of the page, though the infimum otherwise never
carries locks.
A waiting record lock can also be of the gap type. A waiting lock request
can be granted when there is no conflicting mode lock request by another
transaction ahead of it in the explicit lock queue.
In version 4.0.5 we added yet another explicit lock type: LOCK_REC_NOT_GAP.
It only locks the record it is placed on, not the gap before the record.
This lock type is necessary to emulate an Oracle-like READ COMMITTED isolation
level.
-------------------------------------------------------------------------
RULE 1: If there is an implicit x-lock on a record, and there are non-gap
-------
lock requests waiting in the queue, then the transaction holding the implicit
x-lock also has an explicit non-gap record x-lock. Therefore, as locks are
released, we can grant locks to waiting lock requests purely by looking at
the explicit lock requests in the queue.
RULE 3: Different transactions cannot have conflicting granted non-gap locks
-------
on a record at the same time. However, they can have conflicting granted gap
locks.
RULE 4: If a there is a waiting lock request in a queue, no lock request,
-------
gap or not, can be inserted ahead of it in the queue. In record deletes
and page splits new gap type locks can be created by the database manager
for a transaction, and without rule 4, the waits-for graph of transactions
might become cyclic without the database noticing it, as the deadlock check
is only performed when a transaction itself requests a lock!
-------------------------------------------------------------------------
An insert is allowed to a gap if there are no explicit lock requests by
other transactions on the next record. It does not matter if these lock
requests are granted or waiting, gap bit set or not, with the exception
that a gap type request set by another transaction to wait for
its turn to do an insert is ignored. On the other hand, an
implicit x-lock by another transaction does not prevent an insert, which
allows for more concurrency when using an Oracle-style sequence number
generator for the primary key with many transactions doing inserts
concurrently.
A modify of a record is allowed if the transaction has an x-lock on the
record, or if other transactions do not have any non-gap lock requests on the
record.
A read of a single user record with a cursor is allowed if the transaction
has a non-gap explicit, or an implicit lock on the record, or if the other
transactions have no x-lock requests on the record. At a page supremum a
read is always allowed.
In summary, an implicit lock is seen as a granted x-lock only on the
record, not on the gap. An explicit lock with no gap bit set is a lock
both on the record and the gap. If the gap bit is set, the lock is only
on the gap. Different transaction cannot own conflicting locks on the
record at the same time, but they may own conflicting locks on the gap.
Granted locks on a record give an access right to the record, but gap type
locks just inhibit operations.
NOTE: Finding out if some transaction has an implicit x-lock on a secondary
index record can be cumbersome. We may have to look at previous versions of
the corresponding clustered index record to find out if a delete marked
secondary index record was delete marked by an active transaction, not by
a committed one.
FACT A: If a transaction has inserted a row, it can delete it any time
without need to wait for locks.
PROOF: The transaction has an implicit x-lock on every index record inserted
for the row, and can thus modify each record without the need to wait. Q.E.D.
FACT B: If a transaction has read some result set with a cursor, it can read
it again, and retrieves the same result set, if it has not modified the
result set in the meantime. Hence, there is no phantom problem. If the
biggest record, in the alphabetical order, touched by the cursor is removed,
a lock wait may occur, otherwise not.
PROOF: When a read cursor proceeds, it sets an s-lock on each user record
it passes, and a gap type s-lock on each page supremum. The cursor must
wait until it has these locks granted. Then no other transaction can
have a granted x-lock on any of the user records, and therefore cannot
modify the user records. Neither can any other transaction insert into
the gaps which were passed over by the cursor. Page splits and merges,
and removal of obsolete versions of records do not affect this, because
when a user record or a page supremum is removed, the next record inherits
its locks as gap type locks, and therefore blocks inserts to the same gap.
Also, if a page supremum is inserted, it inherits its locks from the successor
record. When the cursor is positioned again at the start of the result set,
the records it will touch on its course are either records it touched
during the last pass or new inserted page supremums. It can immediately
access all these records, and when it arrives at the biggest record, it
notices that the result set is complete. If the biggest record was removed,
lock wait can occur because the next record only inherits a gap type lock,
and a wait may be needed. Q.E.D. */
/* If an index record should be changed or a new inserted, we must check
the lock on the record or the next. When a read cursor starts reading,
we will set a record level s-lock on each record it passes, except on the
initial record on which the cursor is positioned before we start to fetch
records. Our index tree search has the convention that the B-tree
cursor is positioned BEFORE the first possibly matching record in
the search. Optimizations are possible here: if the record is searched
on an equality condition to a unique key, we could actually set a special
lock on the record, a lock which would not prevent any insert before
this record. In the next key locking an x-lock set on a record also
prevents inserts just before that record.
There are special infimum and supremum records on each page.
A supremum record can be locked by a read cursor. This records cannot be
updated but the lock prevents insert of a user record to the end of
the page.
Next key locks will prevent the phantom problem where new rows
could appear to SELECT result sets after the select operation has been
performed. Prevention of phantoms ensures the serilizability of
transactions.
What should we check if an insert of a new record is wanted?
Only the lock on the next record on the same page, because also the
supremum record can carry a lock. An s-lock prevents insertion, but
what about an x-lock? If it was set by a searched update, then there
is implicitly an s-lock, too, and the insert should be prevented.
What if our transaction owns an x-lock to the next record, but there is
a waiting s-lock request on the next record? If this s-lock was placed
by a read cursor moving in the ascending order in the index, we cannot
do the insert immediately, because when we finally commit our transaction,
the read cursor should see also the new inserted record. So we should
move the read cursor backward from the next record for it to pass over
the new inserted record. This move backward may be too cumbersome to
implement. If we in this situation just enqueue a second x-lock request
for our transaction on the next record, then the deadlock mechanism
notices a deadlock between our transaction and the s-lock request
transaction. This seems to be an ok solution.
We could have the convention that granted explicit record locks,
lock the corresponding records from changing, and also lock the gaps
before them from inserting. A waiting explicit lock request locks the gap
before from inserting. Implicit record x-locks, which we derive from the
transaction id in the clustered index record, only lock the record itself
from modification, not the gap before it from inserting.
How should we store update locks? If the search is done by a unique
key, we could just modify the record trx id. Otherwise, we could put a record
x-lock on the record. If the update changes ordering fields of the
clustered index record, the inserted new record needs no record lock in
lock table, the trx id is enough. The same holds for a secondary index
record. Searched delete is similar to update.
PROBLEM:
What about waiting lock requests? If a transaction is waiting to make an
update to a record which another modified, how does the other transaction
know to send the end-lock-wait signal to the waiting transaction? If we have
the convention that a transaction may wait for just one lock at a time, how
do we preserve it if lock wait ends?
PROBLEM:
Checking the trx id label of a secondary index record. In the case of a
modification, not an insert, is this necessary? A secondary index record
is modified only by setting or resetting its deleted flag. A secondary index
record contains fields to uniquely determine the corresponding clustered
index record. A secondary index record is therefore only modified if we
also modify the clustered index record, and the trx id checking is done
on the clustered index record, before we come to modify the secondary index
record. So, in the case of delete marking or unmarking a secondary index
record, we do not have to care about trx ids, only the locks in the lock
table must be checked. In the case of a select from a secondary index, the
trx id is relevant, and in this case we may have to search the clustered
index record.
PROBLEM: How to update record locks when page is split or merged, or
--------------------------------------------------------------------
a record is deleted or updated?
If the size of fields in a record changes, we perform the update by
a delete followed by an insert. How can we retain the locks set or
waiting on the record? Because a record lock is indexed in the bitmap
by the heap number of the record, when we remove the record from the
record list, it is possible still to keep the lock bits. If the page
is reorganized, we could make a table of old and new heap numbers,
and permute the bitmaps in the locks accordingly. We can add to the
table a row telling where the updated record ended. If the update does
not require a reorganization of the page, we can simply move the lock
bits for the updated record to the position determined by its new heap
number (we may have to allocate a new lock, if we run out of the bitmap
in the old one).
A more complicated case is the one where the reinsertion of the
updated record is done pessimistically, because the structure of the
tree may change.
PROBLEM: If a supremum record is removed in a page merge, or a record
---------------------------------------------------------------------
removed in a purge, what to do to the waiting lock requests? In a split to
the right, we just move the lock requests to the new supremum. If a record
is removed, we could move the waiting lock request to its inheritor, the
next record in the index. But, the next record may already have lock
requests on its own queue. A new deadlock check should be made then. Maybe
it is easier just to release the waiting transactions. They can then enqueue
new lock requests on appropriate records.
PROBLEM: When a record is inserted, what locks should it inherit from the
-------------------------------------------------------------------------
upper neighbor? An insert of a new supremum record in a page split is
always possible, but an insert of a new user record requires that the upper
neighbor does not have any lock requests by other transactions, granted or
waiting, in its lock queue. Solution: We can copy the locks as gap type
locks, so that also the waiting locks are transformed to granted gap type
locks on the inserted record. */
#define LOCK_STACK_SIZE OS_THREAD_MAX_N
/* LOCK COMPATIBILITY MATRIX
* IS IX S X AI
* IS + + + - +
* IX + + - - +
* S + - + - -
* X - - - - -
* AI + + - - -
*
* Note that for rows, InnoDB only acquires S or X locks.
* For tables, InnoDB normally acquires IS or IX locks.
* S or X table locks are only acquired for LOCK TABLES.
* Auto-increment (AI) locks are needed because of
* statement-level MySQL binlog.
* See also lock_mode_compatible().
*/
static const byte lock_compatibility_matrix[5][5] = {
/** IS IX S X AI */
/* IS */ { TRUE, TRUE, TRUE, FALSE, TRUE},
/* IX */ { TRUE, TRUE, FALSE, FALSE, TRUE},
/* S */ { TRUE, FALSE, TRUE, FALSE, FALSE},
/* X */ { FALSE, FALSE, FALSE, FALSE, FALSE},
/* AI */ { TRUE, TRUE, FALSE, FALSE, FALSE}
};
/* STRONGER-OR-EQUAL RELATION (mode1=row, mode2=column)
* IS IX S X AI
* IS + - - - -
* IX + + - - -
* S + - + - -
* X + + + + +
* AI - - - - +
* See lock_mode_stronger_or_eq().
*/
static const byte lock_strength_matrix[5][5] = {
/** IS IX S X AI */
/* IS */ { TRUE, FALSE, FALSE, FALSE, FALSE},
/* IX */ { TRUE, TRUE, FALSE, FALSE, FALSE},
/* S */ { TRUE, FALSE, TRUE, FALSE, FALSE},
/* X */ { TRUE, TRUE, TRUE, TRUE, TRUE},
/* AI */ { FALSE, FALSE, FALSE, FALSE, TRUE}
};
/** Deadlock check context. */
struct lock_deadlock_ctx_t {
const trx_t* start; /*!< Joining transaction that is
requesting a lock in an incompatible
mode */
const lock_t* wait_lock; /*!< Lock that trx wants */
ib_uint64_t mark_start; /*!< Value of lock_mark_count at
the start of the deadlock check. */
ulint depth; /*!< Stack depth */
ulint cost; /*!< Calculation steps thus far */
ibool too_deep; /*!< TRUE if search was too deep and
was aborted */
};
/** DFS visited node information used during deadlock checking. */
struct lock_stack_t {
const lock_t* lock; /*!< Current lock */
const lock_t* wait_lock; /*!< Waiting for lock */
ulint heap_no; /*!< heap number if rec lock */
};
/*********************************************************************//**
Checks if a waiting record lock request still has to wait in a queue.
@return lock that is causing the wait */
static
const lock_t*
lock_rec_has_to_wait_in_queue(
/*==========================*/
const lock_t* wait_lock); /*!< in: waiting record lock */
/*************************************************************//**
Grants a lock to a waiting lock request and releases the waiting transaction.
The caller must hold lock_sys->mutex. */
static
void
lock_grant(
/*=======*/
lock_t* lock, /*!< in/out: waiting lock request */
bool owns_trx_mutex); /*!< in: whether lock->trx->mutex is owned */
extern "C" void thd_report_wait_for(MYSQL_THD thd, MYSQL_THD other_thd);
extern "C" int thd_need_wait_for(const MYSQL_THD thd);
extern "C"
int thd_need_ordering_with(const MYSQL_THD thd, const MYSQL_THD other_thd);
extern "C"
int thd_deadlock_victim_preference(const MYSQL_THD thd1, const MYSQL_THD thd2);
/** Stack to use during DFS search. Currently only a single stack is required
because there is no parallel deadlock check. This stack is protected by
the lock_sys_t::mutex. */
static lock_stack_t* lock_stack;
#ifdef UNIV_DEBUG
/** The count of the types of locks. */
static const ulint lock_types = UT_ARR_SIZE(lock_compatibility_matrix);
#endif /* UNIV_DEBUG */
#ifdef UNIV_PFS_MUTEX
/* Key to register mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t lock_sys_mutex_key;
/* Key to register mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t lock_sys_wait_mutex_key;
#endif /* UNIV_PFS_MUTEX */
/* Buffer to collect THDs to report waits for. */
struct thd_wait_reports {
struct thd_wait_reports *next; /*!< List link */
ulint used; /*!< How many elements in waitees[] */
trx_t *waitees[64]; /*!< Trxs for thd_report_wait_for() */
};
#ifdef UNIV_DEBUG
UNIV_INTERN ibool lock_print_waits = FALSE;
/*********************************************************************//**
Validates the lock system.
@return TRUE if ok */
static
bool
lock_validate();
/*============*/
/*********************************************************************//**
Validates the record lock queues on a page.
@return TRUE if ok */
static
ibool
lock_rec_validate_page(
/*===================*/
const buf_block_t* block) /*!< in: buffer block */
MY_ATTRIBUTE((nonnull, warn_unused_result));
#endif /* UNIV_DEBUG */
/* The lock system */
UNIV_INTERN lock_sys_t* lock_sys = NULL;
/** We store info on the latest deadlock error to this buffer. InnoDB
Monitor will then fetch it and print */
UNIV_INTERN ibool lock_deadlock_found = FALSE;
/** Only created if !srv_read_only_mode */
static FILE* lock_latest_err_file;
/********************************************************************//**
Checks if a joining lock request results in a deadlock. If a deadlock is
found this function will resolve the dadlock by choosing a victim transaction
and rolling it back. It will attempt to resolve all deadlocks. The returned
transaction id will be the joining transaction id or 0 if some other
transaction was chosen as a victim and rolled back or no deadlock found.
@return id of transaction chosen as victim or 0 */
static
trx_id_t
lock_deadlock_check_and_resolve(
/*===========================*/
const lock_t* lock, /*!< in: lock the transaction is requesting */
const trx_t* trx); /*!< in: transaction */
/*********************************************************************//**
Gets the nth bit of a record lock.
@return TRUE if bit set also if i == ULINT_UNDEFINED return FALSE*/
UNIV_INLINE
ibool
lock_rec_get_nth_bit(
/*=================*/
const lock_t* lock, /*!< in: record lock */
ulint i) /*!< in: index of the bit */
{
const byte* b;
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
if (i >= lock->un_member.rec_lock.n_bits) {
return(FALSE);
}
b = ((const byte*) &lock[1]) + (i / 8);
return(1 & *b >> (i % 8));
}
/*********************************************************************//**
Reports that a transaction id is insensible, i.e., in the future. */
UNIV_INTERN
void
lock_report_trx_id_insanity(
/*========================*/
trx_id_t trx_id, /*!< in: trx id */
const rec_t* rec, /*!< in: user record */
dict_index_t* index, /*!< in: index */
const ulint* offsets, /*!< in: rec_get_offsets(rec, index) */
trx_id_t max_trx_id) /*!< in: trx_sys_get_max_trx_id() */
{
ut_print_timestamp(stderr);
fputs(" InnoDB: Error: transaction id associated with record\n",
stderr);
rec_print_new(stderr, rec, offsets);
fputs("InnoDB: in ", stderr);
dict_index_name_print(stderr, NULL, index);
fprintf(stderr, "\n"
"InnoDB: is " TRX_ID_FMT " which is higher than the"
" global trx id counter " TRX_ID_FMT "!\n"
"InnoDB: The table is corrupt. You have to do"
" dump + drop + reimport.\n",
trx_id, max_trx_id);
}
/*********************************************************************//**
Checks that a transaction id is sensible, i.e., not in the future.
@return true if ok */
#ifdef UNIV_DEBUG
UNIV_INTERN
#else
static MY_ATTRIBUTE((nonnull, warn_unused_result))
#endif
bool
lock_check_trx_id_sanity(
/*=====================*/
trx_id_t trx_id, /*!< in: trx id */
const rec_t* rec, /*!< in: user record */
dict_index_t* index, /*!< in: index */
const ulint* offsets) /*!< in: rec_get_offsets(rec, index) */
{
bool is_ok;
trx_id_t max_trx_id;
ut_ad(rec_offs_validate(rec, index, offsets));
max_trx_id = trx_sys_get_max_trx_id();
is_ok = trx_id < max_trx_id;
if (UNIV_UNLIKELY(!is_ok)) {
lock_report_trx_id_insanity(trx_id,
rec, index, offsets, max_trx_id);
}
return(is_ok);
}
/*********************************************************************//**
Checks that a record is seen in a consistent read.
@return true if sees, or false if an earlier version of the record
should be retrieved */
UNIV_INTERN
bool
lock_clust_rec_cons_read_sees(
/*==========================*/
const rec_t* rec, /*!< in: user record which should be read or
passed over by a read cursor */
dict_index_t* index, /*!< in: clustered index */
const ulint* offsets,/*!< in: rec_get_offsets(rec, index) */
read_view_t* view) /*!< in: consistent read view */
{
trx_id_t trx_id;
ut_ad(dict_index_is_clust(index));
ut_ad(page_rec_is_user_rec(rec));
ut_ad(rec_offs_validate(rec, index, offsets));
/* NOTE that we call this function while holding the search
system latch. */
trx_id = row_get_rec_trx_id(rec, index, offsets);
return(read_view_sees_trx_id(view, trx_id));
}
/*********************************************************************//**
Checks that a non-clustered index record is seen in a consistent read.
NOTE that a non-clustered index page contains so little information on
its modifications that also in the case false, the present version of
rec may be the right, but we must check this from the clustered index
record.
@return true if certainly sees, or false if an earlier version of the
clustered index record might be needed */
UNIV_INTERN
bool
lock_sec_rec_cons_read_sees(
/*========================*/
const rec_t* rec, /*!< in: user record which
should be read or passed over
by a read cursor */
const read_view_t* view) /*!< in: consistent read view */
{
trx_id_t max_trx_id;
ut_ad(page_rec_is_user_rec(rec));
/* NOTE that we might call this function while holding the search
system latch. */
if (recv_recovery_is_on()) {
return(false);
}
max_trx_id = page_get_max_trx_id(page_align(rec));
ut_ad(max_trx_id);
return(max_trx_id < view->up_limit_id);
}
/*********************************************************************//**
Creates the lock system at database start. */
UNIV_INTERN
void
lock_sys_create(
/*============*/
ulint n_cells) /*!< in: number of slots in lock hash table */
{
ulint lock_sys_sz;
lock_sys_sz = sizeof(*lock_sys)
+ OS_THREAD_MAX_N * sizeof(srv_slot_t);
lock_sys = static_cast<lock_sys_t*>(mem_zalloc(lock_sys_sz));
lock_stack = static_cast<lock_stack_t*>(
mem_zalloc(sizeof(*lock_stack) * LOCK_STACK_SIZE));
void* ptr = &lock_sys[1];
lock_sys->waiting_threads = static_cast<srv_slot_t*>(ptr);
lock_sys->last_slot = lock_sys->waiting_threads;
mutex_create(lock_sys_mutex_key, &lock_sys->mutex, SYNC_LOCK_SYS);
mutex_create(lock_sys_wait_mutex_key,
&lock_sys->wait_mutex, SYNC_LOCK_WAIT_SYS);
lock_sys->timeout_event = os_event_create();
lock_sys->rec_hash = hash_create(n_cells);
lock_sys->rec_num = 0;
if (!srv_read_only_mode) {
lock_latest_err_file = os_file_create_tmpfile(NULL);
ut_a(lock_latest_err_file);
}
}
/*********************************************************************//**
Closes the lock system at database shutdown. */
UNIV_INTERN
void
lock_sys_close(void)
/*================*/
{
if (lock_latest_err_file != NULL) {
fclose(lock_latest_err_file);
lock_latest_err_file = NULL;
}
hash_table_free(lock_sys->rec_hash);
mutex_free(&lock_sys->mutex);
mutex_free(&lock_sys->wait_mutex);
os_event_free(lock_sys->timeout_event);
for (srv_slot_t* slot = lock_sys->waiting_threads;
slot < lock_sys->waiting_threads + OS_THREAD_MAX_N; slot++) {
ut_ad(!slot->in_use);
ut_ad(!slot->thr);
if (slot->event != NULL)
os_event_free(slot->event);
}
mem_free(lock_stack);
mem_free(lock_sys);
lock_sys = NULL;
lock_stack = NULL;
}
/*********************************************************************//**
Gets the size of a lock struct.
@return size in bytes */
UNIV_INTERN
ulint
lock_get_size(void)
/*===============*/
{
return((ulint) sizeof(lock_t));
}
/*********************************************************************//**
Gets the mode of a lock.
@return mode */
UNIV_INLINE
enum lock_mode
lock_get_mode(
/*==========*/
const lock_t* lock) /*!< in: lock */
{
ut_ad(lock);
return(static_cast<enum lock_mode>(lock->type_mode & LOCK_MODE_MASK));
}
/*********************************************************************//**
Gets the wait flag of a lock.
@return LOCK_WAIT if waiting, 0 if not */
UNIV_INLINE
ulint
lock_get_wait(
/*==========*/
const lock_t* lock) /*!< in: lock */
{
ut_ad(lock);
return(lock->type_mode & LOCK_WAIT);
}
/*********************************************************************//**
Gets the source table of an ALTER TABLE transaction. The table must be
covered by an IX or IS table lock.
@return the source table of transaction, if it is covered by an IX or
IS table lock; dest if there is no source table, and NULL if the
transaction is locking more than two tables or an inconsistency is
found */
UNIV_INTERN
dict_table_t*
lock_get_src_table(
/*===============*/
trx_t* trx, /*!< in: transaction */
dict_table_t* dest, /*!< in: destination of ALTER TABLE */
enum lock_mode* mode) /*!< out: lock mode of the source table */
{
dict_table_t* src;
lock_t* lock;
ut_ad(!lock_mutex_own());
src = NULL;
*mode = LOCK_NONE;
/* The trx mutex protects the trx_locks for our purposes.
Other transactions could want to convert one of our implicit
record locks to an explicit one. For that, they would need our
trx mutex. Waiting locks can be removed while only holding
lock_sys->mutex, but this is a running transaction and cannot
thus be holding any waiting locks. */
trx_mutex_enter(trx);
for (lock = UT_LIST_GET_FIRST(trx->lock.trx_locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(trx_locks, lock)) {
lock_table_t* tab_lock;
enum lock_mode lock_mode;
if (!(lock_get_type_low(lock) & LOCK_TABLE)) {
/* We are only interested in table locks. */
continue;
}
tab_lock = &lock->un_member.tab_lock;
if (dest == tab_lock->table) {
/* We are not interested in the destination table. */
continue;
} else if (!src) {
/* This presumably is the source table. */
src = tab_lock->table;
if (UT_LIST_GET_LEN(src->locks) != 1
|| UT_LIST_GET_FIRST(src->locks) != lock) {
/* We only support the case when
there is only one lock on this table. */
src = NULL;
goto func_exit;
}
} else if (src != tab_lock->table) {
/* The transaction is locking more than
two tables (src and dest): abort */
src = NULL;
goto func_exit;
}
/* Check that the source table is locked by
LOCK_IX or LOCK_IS. */
lock_mode = lock_get_mode(lock);
if (lock_mode == LOCK_IX || lock_mode == LOCK_IS) {
if (*mode != LOCK_NONE && *mode != lock_mode) {
/* There are multiple locks on src. */
src = NULL;
goto func_exit;
}
*mode = lock_mode;
}
}
if (!src) {
/* No source table lock found: flag the situation to caller */
src = dest;
}
func_exit:
trx_mutex_exit(trx);
return(src);
}
/*********************************************************************//**
Determine if the given table is exclusively "owned" by the given
transaction, i.e., transaction holds LOCK_IX and possibly LOCK_AUTO_INC
on the table.
@return TRUE if table is only locked by trx, with LOCK_IX, and
possibly LOCK_AUTO_INC */
UNIV_INTERN
ibool
lock_is_table_exclusive(
/*====================*/
const dict_table_t* table, /*!< in: table */
const trx_t* trx) /*!< in: transaction */
{
const lock_t* lock;
ibool ok = FALSE;
ut_ad(table);
ut_ad(trx);
lock_mutex_enter();
for (lock = UT_LIST_GET_FIRST(table->locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(locks, &lock->un_member.tab_lock)) {
if (lock->trx != trx) {
/* A lock on the table is held
by some other transaction. */
goto not_ok;
}
if (!(lock_get_type_low(lock) & LOCK_TABLE)) {
/* We are interested in table locks only. */
continue;
}
switch (lock_get_mode(lock)) {
case LOCK_IX:
ok = TRUE;
break;
case LOCK_AUTO_INC:
/* It is allowed for trx to hold an
auto_increment lock. */
break;
default:
not_ok:
/* Other table locks than LOCK_IX are not allowed. */
ok = FALSE;
goto func_exit;
}
}
func_exit:
lock_mutex_exit();
return(ok);
}
/*********************************************************************//**
Sets the wait flag of a lock and the back pointer in trx to lock. */
UNIV_INLINE
void
lock_set_lock_and_trx_wait(
/*=======================*/
lock_t* lock, /*!< in: lock */
trx_t* trx) /*!< in/out: trx */
{
ut_ad(lock);
ut_ad(lock->trx == trx);
ut_ad(trx->lock.wait_lock == NULL);
ut_ad(lock_mutex_own());
ut_ad(trx_mutex_own(trx));
trx->lock.wait_lock = lock;
lock->type_mode |= LOCK_WAIT;
}
/**********************************************************************//**
The back pointer to a waiting lock request in the transaction is set to NULL
and the wait bit in lock type_mode is reset. */
UNIV_INLINE
void
lock_reset_lock_and_trx_wait(
/*=========================*/
lock_t* lock) /*!< in/out: record lock */
{
ut_ad(lock_get_wait(lock));
ut_ad(lock_mutex_own());
if (lock->trx->lock.wait_lock &&
lock->trx->lock.wait_lock != lock) {
const char* stmt=NULL;
const char* stmt2=NULL;
size_t stmt_len;
trx_id_t trx_id = 0;
stmt = innobase_get_stmt(lock->trx->mysql_thd, &stmt_len);
if (lock->trx->lock.wait_lock &&
lock->trx->lock.wait_lock->trx) {
trx_id = lock->trx->lock.wait_lock->trx->id;
stmt2 = innobase_get_stmt(lock->trx->lock.wait_lock->trx->mysql_thd, &stmt_len);
}
ib_logf(IB_LOG_LEVEL_INFO,
"Trx id " TRX_ID_FMT
" is waiting a lock in statement %s"
" for this trx id " TRX_ID_FMT
" and statement %s wait_lock %p",
lock->trx->id,
stmt ? stmt : "NULL",
trx_id,
stmt2 ? stmt2 : "NULL",
lock->trx->lock.wait_lock);
ut_ad(lock->trx->lock.wait_lock == lock);
}
lock->trx->lock.wait_lock = NULL;
lock->type_mode &= ~LOCK_WAIT;
}
/*********************************************************************//**
Gets the gap flag of a record lock.
@return LOCK_GAP or 0 */
UNIV_INLINE
ulint
lock_rec_get_gap(
/*=============*/
const lock_t* lock) /*!< in: record lock */
{
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
return(lock->type_mode & LOCK_GAP);
}
/*********************************************************************//**
Gets the LOCK_REC_NOT_GAP flag of a record lock.
@return LOCK_REC_NOT_GAP or 0 */
UNIV_INLINE
ulint
lock_rec_get_rec_not_gap(
/*=====================*/
const lock_t* lock) /*!< in: record lock */
{
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
return(lock->type_mode & LOCK_REC_NOT_GAP);
}
/*********************************************************************//**
Gets the waiting insert flag of a record lock.
@return LOCK_INSERT_INTENTION or 0 */
UNIV_INLINE
ulint
lock_rec_get_insert_intention(
/*==========================*/
const lock_t* lock) /*!< in: record lock */
{
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
return(lock->type_mode & LOCK_INSERT_INTENTION);
}
/*********************************************************************//**
Calculates if lock mode 1 is stronger or equal to lock mode 2.
@return nonzero if mode1 stronger or equal to mode2 */
UNIV_INLINE
ulint
lock_mode_stronger_or_eq(
/*=====================*/
enum lock_mode mode1, /*!< in: lock mode */
enum lock_mode mode2) /*!< in: lock mode */
{
ut_ad((ulint) mode1 < lock_types);
ut_ad((ulint) mode2 < lock_types);
return(lock_strength_matrix[mode1][mode2]);
}
/*********************************************************************//**
Calculates if lock mode 1 is compatible with lock mode 2.
@return nonzero if mode1 compatible with mode2 */
UNIV_INLINE
ulint
lock_mode_compatible(
/*=================*/
enum lock_mode mode1, /*!< in: lock mode */
enum lock_mode mode2) /*!< in: lock mode */
{
ut_ad((ulint) mode1 < lock_types);
ut_ad((ulint) mode2 < lock_types);
return(lock_compatibility_matrix[mode1][mode2]);
}
/*********************************************************************//**
Checks if a lock request for a new lock has to wait for request lock2.
@return TRUE if new lock has to wait for lock2 to be removed */
UNIV_INLINE
ibool
lock_rec_has_to_wait(
/*=================*/
#ifdef WITH_WSREP
ibool for_locking, /*!< is caller locking or releasing */
#endif /* WITH_WSREP */
const trx_t* trx, /*!< in: trx of new lock */
ulint type_mode,/*!< in: precise mode of the new lock
to set: LOCK_S or LOCK_X, possibly
ORed to LOCK_GAP or LOCK_REC_NOT_GAP,
LOCK_INSERT_INTENTION */
const lock_t* lock2, /*!< in: another record lock; NOTE that
it is assumed that this has a lock bit
set on the same record as in the new
lock we are setting */
ibool lock_is_on_supremum) /*!< in: TRUE if we are setting the
lock on the 'supremum' record of an
index page: we know then that the lock
request is really for a 'gap' type lock */
{
ut_ad(trx && lock2);
ut_ad(lock_get_type_low(lock2) == LOCK_REC);
if (trx != lock2->trx
&& !lock_mode_compatible(static_cast<enum lock_mode>(
LOCK_MODE_MASK & type_mode),
lock_get_mode(lock2))) {
/* We have somewhat complex rules when gap type record locks
cause waits */
if ((lock_is_on_supremum || (type_mode & LOCK_GAP))
&& !(type_mode & LOCK_INSERT_INTENTION)) {
/* Gap type locks without LOCK_INSERT_INTENTION flag
do not need to wait for anything. This is because
different users can have conflicting lock types
on gaps. */
return(FALSE);
}
if (!(type_mode & LOCK_INSERT_INTENTION)
&& lock_rec_get_gap(lock2)) {
/* Record lock (LOCK_ORDINARY or LOCK_REC_NOT_GAP
does not need to wait for a gap type lock */
return(FALSE);
}
if ((type_mode & LOCK_GAP)
&& lock_rec_get_rec_not_gap(lock2)) {
/* Lock on gap does not need to wait for
a LOCK_REC_NOT_GAP type lock */
return(FALSE);
}
if (lock_rec_get_insert_intention(lock2)) {
/* No lock request needs to wait for an insert
intention lock to be removed. This is ok since our
rules allow conflicting locks on gaps. This eliminates
a spurious deadlock caused by a next-key lock waiting
for an insert intention lock; when the insert
intention lock was granted, the insert deadlocked on
the waiting next-key lock.
Also, insert intention locks do not disturb each
other. */
return(FALSE);
}
if ((type_mode & LOCK_GAP || lock_rec_get_gap(lock2)) &&
!thd_need_ordering_with(trx->mysql_thd,
lock2->trx->mysql_thd)) {
/* If the upper server layer has already decided on the
commit order between the transaction requesting the
lock and the transaction owning the lock, we do not
need to wait for gap locks. Such ordeering by the upper
server layer happens in parallel replication, where the
commit order is fixed to match the original order on the
master.
Such gap locks are mainly needed to get serialisability
between transactions so that they will be binlogged in
the correct order so that statement-based replication
will give the correct results. Since the right order
was already determined on the master, we do not need
to enforce it again here.
Skipping the locks is not essential for correctness,
since in case of deadlock we will just kill the later
transaction and retry it. But it can save some
unnecessary rollbacks and retries. */
return (FALSE);
}
#ifdef WITH_WSREP
/* if BF thread is locking and has conflict with another BF
thread, we need to look at trx ordering and lock types */
if (wsrep_thd_is_BF(trx->mysql_thd, FALSE) &&
wsrep_thd_is_BF(lock2->trx->mysql_thd, TRUE)) {
if (wsrep_debug) {
fprintf(stderr,
"BF-BF lock conflict, locking: %lu\n",
for_locking);
lock_rec_print(stderr, lock2);
}
if (wsrep_trx_order_before(trx->mysql_thd,
lock2->trx->mysql_thd) &&
(type_mode & LOCK_MODE_MASK) == LOCK_X &&
(lock2->type_mode & LOCK_MODE_MASK) == LOCK_X)
{
if (for_locking || wsrep_debug) {
/* exclusive lock conflicts are not
accepted */
fprintf(stderr,
"BF-BF X lock conflict,"
"mode: %lu supremum: %lu\n",
type_mode, lock_is_on_supremum);
fprintf(stderr,
"conflicts states: my %d locked %d\n",
wsrep_thd_conflict_state(trx->mysql_thd, FALSE),
wsrep_thd_conflict_state(lock2->trx->mysql_thd, FALSE) );
lock_rec_print(stderr, lock2);
if (for_locking) return FALSE;
//abort();
}
} else {
/* if lock2->index->n_uniq <=
lock2->index->n_user_defined_cols
operation is on uniq index
*/
if (wsrep_debug) fprintf(stderr,
"BF conflict, modes: %lu %lu, "
"idx: %s-%s n_uniq %u n_user %u\n",
type_mode, lock2->type_mode,
lock2->index->name,
lock2->index->table_name,
lock2->index->n_uniq,
lock2->index->n_user_defined_cols);
return FALSE;
}
}
#endif /* WITH_WSREP */
return(TRUE);
}
return(FALSE);
}
/*********************************************************************//**
Checks if a lock request lock1 has to wait for request lock2.
@return TRUE if lock1 has to wait for lock2 to be removed */
UNIV_INTERN
ibool
lock_has_to_wait(
/*=============*/
const lock_t* lock1, /*!< in: waiting lock */
const lock_t* lock2) /*!< in: another lock; NOTE that it is
assumed that this has a lock bit set
on the same record as in lock1 if the
locks are record locks */
{
ut_ad(lock1 && lock2);
if (lock1->trx != lock2->trx
&& !lock_mode_compatible(lock_get_mode(lock1),
lock_get_mode(lock2))) {
if (lock_get_type_low(lock1) == LOCK_REC) {
ut_ad(lock_get_type_low(lock2) == LOCK_REC);
/* If this lock request is for a supremum record
then the second bit on the lock bitmap is set */
#ifdef WITH_WSREP
return(lock_rec_has_to_wait(FALSE, lock1->trx,
#else
return(lock_rec_has_to_wait(lock1->trx,
#endif /* WITH_WSREP */
lock1->type_mode, lock2,
lock_rec_get_nth_bit(
lock1, 1)));
}
return(TRUE);
}
return(FALSE);
}
/*============== RECORD LOCK BASIC FUNCTIONS ============================*/
/*********************************************************************//**
Gets the number of bits in a record lock bitmap.
@return number of bits */
UNIV_INLINE
ulint
lock_rec_get_n_bits(
/*================*/
const lock_t* lock) /*!< in: record lock */
{
return(lock->un_member.rec_lock.n_bits);
}
/**********************************************************************//**
Sets the nth bit of a record lock to TRUE. */
UNIV_INLINE
void
lock_rec_set_nth_bit(
/*=================*/
lock_t* lock, /*!< in: record lock */
ulint i) /*!< in: index of the bit */
{
ulint byte_index;
ulint bit_index;
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
ut_ad(i < lock->un_member.rec_lock.n_bits);
byte_index = i / 8;
bit_index = i % 8;
((byte*) &lock[1])[byte_index] |= 1 << bit_index;
}
/**********************************************************************//**
Looks for a set bit in a record lock bitmap. Returns ULINT_UNDEFINED,
if none found.
@return bit index == heap number of the record, or ULINT_UNDEFINED if
none found */
UNIV_INTERN
ulint
lock_rec_find_set_bit(
/*==================*/
const lock_t* lock) /*!< in: record lock with at least one bit set */
{
ulint i;
for (i = 0; i < lock_rec_get_n_bits(lock); i++) {
if (lock_rec_get_nth_bit(lock, i)) {
return(i);
}
}
return(ULINT_UNDEFINED);
}
/**********************************************************************//**
Resets the nth bit of a record lock. */
UNIV_INLINE
void
lock_rec_reset_nth_bit(
/*===================*/
lock_t* lock, /*!< in: record lock */
ulint i) /*!< in: index of the bit which must be set to TRUE
when this function is called */
{
ulint byte_index;
ulint bit_index;
ut_ad(lock);
ut_ad(lock_get_type_low(lock) == LOCK_REC);
ut_ad(i < lock->un_member.rec_lock.n_bits);
byte_index = i / 8;
bit_index = i % 8;
((byte*) &lock[1])[byte_index] &= ~(1 << bit_index);
}
/*********************************************************************//**
Gets the first or next record lock on a page.
@return next lock, NULL if none exists */
UNIV_INLINE
const lock_t*
lock_rec_get_next_on_page_const(
/*============================*/
const lock_t* lock) /*!< in: a record lock */
{
ulint space;
ulint page_no;
ut_ad(lock_mutex_own());
ut_ad(lock_get_type_low(lock) == LOCK_REC);
space = lock->un_member.rec_lock.space;
page_no = lock->un_member.rec_lock.page_no;
for (;;) {
lock = static_cast<const lock_t*>(HASH_GET_NEXT(hash, lock));
if (!lock) {
break;
}
if ((lock->un_member.rec_lock.space == space)
&& (lock->un_member.rec_lock.page_no == page_no)) {
break;
}
}
return(lock);
}
/*********************************************************************//**
Gets the first or next record lock on a page.
@return next lock, NULL if none exists */
UNIV_INLINE
lock_t*
lock_rec_get_next_on_page(
/*======================*/
lock_t* lock) /*!< in: a record lock */
{
return((lock_t*) lock_rec_get_next_on_page_const(lock));
}
/*********************************************************************//**
Gets the first record lock on a page, where the page is identified by its
file address.
@return first lock, NULL if none exists */
UNIV_INLINE
lock_t*
lock_rec_get_first_on_page_addr(
/*============================*/
ulint space, /*!< in: space */
ulint page_no)/*!< in: page number */
{
lock_t* lock;
ut_ad(lock_mutex_own());
for (lock = static_cast<lock_t*>(
HASH_GET_FIRST(lock_sys->rec_hash,
lock_rec_hash(space, page_no)));
lock != NULL;
lock = static_cast<lock_t*>(HASH_GET_NEXT(hash, lock))) {
if (lock->un_member.rec_lock.space == space
&& lock->un_member.rec_lock.page_no == page_no) {
break;
}
}
return(lock);
}
/*********************************************************************//**
Determines if there are explicit record locks on a page.
@return an explicit record lock on the page, or NULL if there are none */
UNIV_INTERN
lock_t*
lock_rec_expl_exist_on_page(
/*========================*/
ulint space, /*!< in: space id */
ulint page_no)/*!< in: page number */
{
lock_t* lock;
lock_mutex_enter();
lock = lock_rec_get_first_on_page_addr(space, page_no);
lock_mutex_exit();
return(lock);
}
/*********************************************************************//**
Gets the first record lock on a page, where the page is identified by a
pointer to it.
@return first lock, NULL if none exists */
UNIV_INLINE
lock_t*
lock_rec_get_first_on_page(
/*=======================*/
const buf_block_t* block) /*!< in: buffer block */
{
ulint hash;
lock_t* lock;
ulint space = buf_block_get_space(block);
ulint page_no = buf_block_get_page_no(block);
ut_ad(lock_mutex_own());
hash = buf_block_get_lock_hash_val(block);
for (lock = static_cast<lock_t*>(
HASH_GET_FIRST( lock_sys->rec_hash, hash));
lock != NULL;
lock = static_cast<lock_t*>(HASH_GET_NEXT(hash, lock))) {
if ((lock->un_member.rec_lock.space == space)
&& (lock->un_member.rec_lock.page_no == page_no)) {
break;
}
}
return(lock);
}
/*********************************************************************//**
Gets the next explicit lock request on a record.
@return next lock, NULL if none exists or if heap_no == ULINT_UNDEFINED */
UNIV_INLINE
lock_t*
lock_rec_get_next(
/*==============*/
ulint heap_no,/*!< in: heap number of the record */
lock_t* lock) /*!< in: lock */
{
ut_ad(lock_mutex_own());
do {
ut_ad(lock_get_type_low(lock) == LOCK_REC);
lock = lock_rec_get_next_on_page(lock);
} while (lock && !lock_rec_get_nth_bit(lock, heap_no));
return(lock);
}
/*********************************************************************//**
Gets the next explicit lock request on a record.
@return next lock, NULL if none exists or if heap_no == ULINT_UNDEFINED */
UNIV_INLINE
const lock_t*
lock_rec_get_next_const(
/*====================*/
ulint heap_no,/*!< in: heap number of the record */
const lock_t* lock) /*!< in: lock */
{
return(lock_rec_get_next(heap_no, (lock_t*) lock));
}
/*********************************************************************//**
Gets the first explicit lock request on a record.
@return first lock, NULL if none exists */
UNIV_INLINE
lock_t*
lock_rec_get_first(
/*===============*/
const buf_block_t* block, /*!< in: block containing the record */
ulint heap_no)/*!< in: heap number of the record */
{
lock_t* lock;
ut_ad(lock_mutex_own());
for (lock = lock_rec_get_first_on_page(block); lock;
lock = lock_rec_get_next_on_page(lock)) {
if (lock_rec_get_nth_bit(lock, heap_no)) {
break;
}
}
return(lock);
}
/*********************************************************************//**
Resets the record lock bitmap to zero. NOTE: does not touch the wait_lock
pointer in the transaction! This function is used in lock object creation
and resetting. */
static
void
lock_rec_bitmap_reset(
/*==================*/
lock_t* lock) /*!< in: record lock */
{
ulint n_bytes;
ut_ad(lock_get_type_low(lock) == LOCK_REC);
/* Reset to zero the bitmap which resides immediately after the lock
struct */
n_bytes = lock_rec_get_n_bits(lock) / 8;
ut_ad((lock_rec_get_n_bits(lock) % 8) == 0);
memset(&lock[1], 0, n_bytes);
}
/*********************************************************************//**
Copies a record lock to heap.
@return copy of lock */
static
lock_t*
lock_rec_copy(
/*==========*/
const lock_t* lock, /*!< in: record lock */
mem_heap_t* heap) /*!< in: memory heap */
{
ulint size;
ut_ad(lock_get_type_low(lock) == LOCK_REC);
size = sizeof(lock_t) + lock_rec_get_n_bits(lock) / 8;
return(static_cast<lock_t*>(mem_heap_dup(heap, lock, size)));
}
/*********************************************************************//**
Gets the previous record lock set on a record.
@return previous lock on the same record, NULL if none exists */
UNIV_INTERN
const lock_t*
lock_rec_get_prev(
/*==============*/
const lock_t* in_lock,/*!< in: record lock */
ulint heap_no)/*!< in: heap number of the record */
{
lock_t* lock;
ulint space;
ulint page_no;
lock_t* found_lock = NULL;
ut_ad(lock_mutex_own());
ut_ad(lock_get_type_low(in_lock) == LOCK_REC);
space = in_lock->un_member.rec_lock.space;
page_no = in_lock->un_member.rec_lock.page_no;
for (lock = lock_rec_get_first_on_page_addr(space, page_no);
/* No op */;
lock = lock_rec_get_next_on_page(lock)) {
ut_ad(lock);
if (lock == in_lock) {
return(found_lock);
}
if (lock_rec_get_nth_bit(lock, heap_no)) {
found_lock = lock;
}
}
}
/*============= FUNCTIONS FOR ANALYZING TABLE LOCK QUEUE ================*/
/*********************************************************************//**
Checks if a transaction has the specified table lock, or stronger. This
function should only be called by the thread that owns the transaction.
@return lock or NULL */
UNIV_INLINE
const lock_t*
lock_table_has(
/*===========*/
const trx_t* trx, /*!< in: transaction */
const dict_table_t* table, /*!< in: table */
enum lock_mode mode) /*!< in: lock mode */
{
lint i;
if (ib_vector_is_empty(trx->lock.table_locks)) {
return(NULL);
}
/* Look for stronger locks the same trx already has on the table */
for (i = ib_vector_size(trx->lock.table_locks) - 1; i >= 0; --i) {
const lock_t* lock;
enum lock_mode lock_mode;
lock = *static_cast<const lock_t**>(
ib_vector_get(trx->lock.table_locks, i));
if (lock == NULL) {
continue;
}
lock_mode = lock_get_mode(lock);
ut_ad(trx == lock->trx);
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
ut_ad(lock->un_member.tab_lock.table != NULL);
if (table == lock->un_member.tab_lock.table
&& lock_mode_stronger_or_eq(lock_mode, mode)) {
ut_ad(!lock_get_wait(lock));
return(lock);
}
}
return(NULL);
}
/*============= FUNCTIONS FOR ANALYZING RECORD LOCK QUEUE ================*/
/*********************************************************************//**
Checks if a transaction has a GRANTED explicit lock on rec stronger or equal
to precise_mode.
@return lock or NULL */
UNIV_INLINE
lock_t*
lock_rec_has_expl(
/*==============*/
ulint precise_mode,/*!< in: LOCK_S or LOCK_X
possibly ORed to LOCK_GAP or
LOCK_REC_NOT_GAP, for a
supremum record we regard this
always a gap type request */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
trx_id_t trx_id) /*!< in: transaction id */
{
lock_t* lock;
ut_ad(lock_mutex_own());
ut_ad((precise_mode & LOCK_MODE_MASK) == LOCK_S
|| (precise_mode & LOCK_MODE_MASK) == LOCK_X);
ut_ad(!(precise_mode & LOCK_INSERT_INTENTION));
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (lock->trx->id == trx_id
&& !lock_rec_get_insert_intention(lock)
&& lock_mode_stronger_or_eq(
lock_get_mode(lock),
static_cast<enum lock_mode>(
precise_mode & LOCK_MODE_MASK))
&& !lock_get_wait(lock)
&& (!lock_rec_get_rec_not_gap(lock)
|| (precise_mode & LOCK_REC_NOT_GAP)
|| heap_no == PAGE_HEAP_NO_SUPREMUM)
&& (!lock_rec_get_gap(lock)
|| (precise_mode & LOCK_GAP)
|| heap_no == PAGE_HEAP_NO_SUPREMUM)) {
return(lock);
}
}
return(NULL);
}
#ifdef WITH_WSREP
static
void
lock_rec_discard(lock_t* in_lock);
#endif
#ifdef UNIV_DEBUG
/*********************************************************************//**
Checks if some other transaction has a lock request in the queue.
@return lock or NULL */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
const lock_t*
lock_rec_other_has_expl_req(
/*========================*/
enum lock_mode mode, /*!< in: LOCK_S or LOCK_X */
ulint gap, /*!< in: LOCK_GAP if also gap
locks are taken into account,
or 0 if not */
ulint wait, /*!< in: LOCK_WAIT if also
waiting locks are taken into
account, or 0 if not */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
trx_id_t trx_id) /*!< in: transaction */
{
const lock_t* lock;
ut_ad(lock_mutex_own());
ut_ad(mode == LOCK_X || mode == LOCK_S);
ut_ad(gap == 0 || gap == LOCK_GAP);
ut_ad(wait == 0 || wait == LOCK_WAIT);
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next_const(heap_no, lock)) {
if (lock->trx->id != trx_id
&& (gap
|| !(lock_rec_get_gap(lock)
|| heap_no == PAGE_HEAP_NO_SUPREMUM))
&& (wait || !lock_get_wait(lock))
&& lock_mode_stronger_or_eq(lock_get_mode(lock), mode)) {
return(lock);
}
}
return(NULL);
}
#endif /* UNIV_DEBUG */
#ifdef WITH_WSREP
static
void
wsrep_kill_victim(
const trx_t * const trx,
const lock_t *lock)
{
ut_ad(lock_mutex_own());
ut_ad(trx_mutex_own(lock->trx));
/* quit for native mysql */
if (!wsrep_on(trx->mysql_thd)) return;
my_bool bf_this = wsrep_thd_is_BF(trx->mysql_thd, FALSE);
my_bool bf_other = wsrep_thd_is_BF(lock->trx->mysql_thd, TRUE);
if ((bf_this && !bf_other) ||
(bf_this && bf_other && wsrep_trx_order_before(
trx->mysql_thd, lock->trx->mysql_thd))) {
if (lock->trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
if (wsrep_debug) {
fprintf(stderr, "WSREP: BF victim waiting\n");
}
/* cannot release lock, until our lock
is in the queue*/
} else if (lock->trx != trx) {
if (wsrep_log_conflicts) {
if (bf_this) {
fputs("\n*** Priority TRANSACTION:\n",
stderr);
} else {
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
wsrep_trx_print_locking(stderr, trx, 3000);
if (bf_other) {
fputs("\n*** Priority TRANSACTION:\n",
stderr);
} else {
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
wsrep_trx_print_locking(stderr, lock->trx, 3000);
fputs("*** WAITING FOR THIS LOCK TO BE GRANTED:\n",
stderr);
if (lock_get_type(lock) == LOCK_REC) {
lock_rec_print(stderr, lock);
} else {
lock_table_print(stderr, lock);
}
}
lock->trx->abort_type = TRX_WSREP_ABORT;
wsrep_innobase_kill_one_trx(trx->mysql_thd,
(const trx_t*) trx, lock->trx, TRUE);
lock->trx->abort_type = TRX_SERVER_ABORT;
}
}
}
#endif
/*********************************************************************//**
Checks if some other transaction has a conflicting explicit lock request
in the queue, so that we have to wait.
@return lock or NULL */
static
const lock_t*
lock_rec_other_has_conflicting(
/*===========================*/
enum lock_mode mode, /*!< in: LOCK_S or LOCK_X,
possibly ORed to LOCK_GAP or
LOC_REC_NOT_GAP,
LOCK_INSERT_INTENTION */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
const trx_t* trx) /*!< in: our transaction */
{
const lock_t* lock;
ibool is_supremum;
ut_ad(lock_mutex_own());
is_supremum = (heap_no == PAGE_HEAP_NO_SUPREMUM);
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next_const(heap_no, lock)) {
#ifdef WITH_WSREP
if (lock_rec_has_to_wait(TRUE, trx, mode, lock, is_supremum)) {
if (wsrep_on(trx->mysql_thd)) {
trx_mutex_enter(lock->trx);
wsrep_kill_victim(trx, lock);
trx_mutex_exit(lock->trx);
}
#else
if (lock_rec_has_to_wait(trx, mode, lock, is_supremum)) {
#endif /* WITH_WSREP */
return(lock);
}
}
return(NULL);
}
/*********************************************************************//**
Looks for a suitable type record lock struct by the same trx on the same page.
This can be used to save space when a new record lock should be set on a page:
no new struct is needed, if a suitable old is found.
@return lock or NULL */
UNIV_INLINE
lock_t*
lock_rec_find_similar_on_page(
/*==========================*/
ulint type_mode, /*!< in: lock type_mode field */
ulint heap_no, /*!< in: heap number of the record */
lock_t* lock, /*!< in: lock_rec_get_first_on_page() */
const trx_t* trx) /*!< in: transaction */
{
ut_ad(lock_mutex_own());
for (/* No op */;
lock != NULL;
lock = lock_rec_get_next_on_page(lock)) {
if (lock->trx == trx
&& lock->type_mode == type_mode
&& lock_rec_get_n_bits(lock) > heap_no) {
return(lock);
}
}
return(NULL);
}
/*********************************************************************//**
Checks if some transaction has an implicit x-lock on a record in a secondary
index.
@return transaction id of the transaction which has the x-lock, or 0;
NOTE that this function can return false positives but never false
negatives. The caller must confirm all positive results by calling
trx_is_active(). */
static
trx_id_t
lock_sec_rec_some_has_impl(
/*=======================*/
const rec_t* rec, /*!< in: user record */
dict_index_t* index, /*!< in: secondary index */
const ulint* offsets)/*!< in: rec_get_offsets(rec, index) */
{
trx_id_t trx_id;
trx_id_t max_trx_id;
const page_t* page = page_align(rec);
ut_ad(!lock_mutex_own());
ut_ad(!mutex_own(&trx_sys->mutex));
ut_ad(!dict_index_is_clust(index));
ut_ad(page_rec_is_user_rec(rec));
ut_ad(rec_offs_validate(rec, index, offsets));
max_trx_id = page_get_max_trx_id(page);
/* Some transaction may have an implicit x-lock on the record only
if the max trx id for the page >= min trx id for the trx list, or
database recovery is running. We do not write the changes of a page
max trx id to the log, and therefore during recovery, this value
for a page may be incorrect. */
if (max_trx_id < trx_rw_min_trx_id() && !recv_recovery_is_on()) {
trx_id = 0;
} else if (!lock_check_trx_id_sanity(max_trx_id, rec, index, offsets)) {
buf_page_print(page, 0, 0);
/* The page is corrupt: try to avoid a crash by returning 0 */
trx_id = 0;
/* In this case it is possible that some transaction has an implicit
x-lock. We have to look in the clustered index. */
} else {
trx_id = row_vers_impl_x_locked(rec, index, offsets);
}
return(trx_id);
}
#ifdef UNIV_DEBUG
/*********************************************************************//**
Checks if some transaction, other than given trx_id, has an explicit
lock on the given rec, in the given precise_mode.
@return the transaction, whose id is not equal to trx_id, that has an
explicit lock on the given rec, in the given precise_mode or NULL.*/
static
trx_t*
lock_rec_other_trx_holds_expl(
/*==========================*/
ulint precise_mode, /*!< in: LOCK_S or LOCK_X
possibly ORed to LOCK_GAP or
LOCK_REC_NOT_GAP. */
trx_id_t trx_id, /*!< in: trx holding implicit
lock on rec */
const rec_t* rec, /*!< in: user record */
const buf_block_t* block) /*!< in: buffer block
containing the record */
{
trx_t* holds = NULL;
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
trx_id_t* impl_trx_desc = trx_find_descriptor(trx_sys->descriptors,
trx_sys->descr_n_used,
trx_id);
if (impl_trx_desc) {
ut_ad(trx_id == *impl_trx_desc);
ulint heap_no = page_rec_get_heap_no(rec);
ulint rw_trx_count = trx_sys->descr_n_used;
trx_id_t* rw_trx_snapshot = static_cast<trx_id_t *>
(ut_malloc(sizeof(trx_id_t) * rw_trx_count));
memcpy(rw_trx_snapshot, trx_sys->descriptors,
sizeof(trx_id_t) * rw_trx_count);
mutex_exit(&trx_sys->mutex);
for (ulint i = 0; i < rw_trx_count; i++) {
lock_t* expl_lock = lock_rec_has_expl(precise_mode,
block, heap_no,
rw_trx_snapshot[i]);
if (expl_lock && expl_lock->trx->id != trx_id) {
/* An explicit lock is held by trx other than
the trx holding the implicit lock. */
holds = expl_lock->trx;
break;
}
}
ut_free(rw_trx_snapshot);
} else {
mutex_exit(&trx_sys->mutex);
}
lock_mutex_exit();
return(holds);
}
#endif /* UNIV_DEBUG */
/*********************************************************************//**
Return approximate number or record locks (bits set in the bitmap) for
this transaction. Since delete-marked records may be removed, the
record count will not be precise.
The caller must be holding lock_sys->mutex. */
UNIV_INTERN
ulint
lock_number_of_rows_locked(
/*=======================*/
const trx_lock_t* trx_lock) /*!< in: transaction locks */
{
const lock_t* lock;
ulint n_records = 0;
ut_ad(lock_mutex_own());
for (lock = UT_LIST_GET_FIRST(trx_lock->trx_locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(trx_locks, lock)) {
if (lock_get_type_low(lock) == LOCK_REC) {
ulint n_bit;
ulint n_bits = lock_rec_get_n_bits(lock);
for (n_bit = 0; n_bit < n_bits; n_bit++) {
if (lock_rec_get_nth_bit(lock, n_bit)) {
n_records++;
}
}
}
}
return(n_records);
}
/*============== RECORD LOCK CREATION AND QUEUE MANAGEMENT =============*/
#ifdef WITH_WSREP
static
void
wsrep_print_wait_locks(
/*============*/
lock_t* c_lock) /* conflicting lock to print */
{
if (wsrep_debug && c_lock->trx->lock.wait_lock != c_lock) {
fprintf(stderr, "WSREP: c_lock != wait lock\n");
if (lock_get_type_low(c_lock) & LOCK_TABLE)
lock_table_print(stderr, c_lock);
else
lock_rec_print(stderr, c_lock);
if (lock_get_type_low(c_lock->trx->lock.wait_lock) & LOCK_TABLE)
lock_table_print(stderr, c_lock->trx->lock.wait_lock);
else
lock_rec_print(stderr, c_lock->trx->lock.wait_lock);
}
}
#endif /* WITH_WSREP */
/*********************************************************************//**
Check if lock1 has higher priority than lock2.
NULL has lowest priority.
If neither of them is wait lock, the first one has higher priority.
If only one of them is a wait lock, it has lower priority.
Otherwise, the one with an older transaction has higher priority.
@returns true if lock1 has higher priority, false otherwise. */
bool
has_higher_priority(
lock_t *lock1,
lock_t *lock2)
{
if (lock1 == NULL) {
return false;
} else if (lock2 == NULL) {
return true;
}
// No preference. Compre them by wait mode and trx age.
if (!lock_get_wait(lock1)) {
return true;
} else if (!lock_get_wait(lock2)) {
return false;
}
return lock1->trx->start_time_micro <= lock2->trx->start_time_micro;
}
/*********************************************************************//**
Insert a lock to the hash list according to the mode (whether it is a wait
lock) and the age of the transaction the it is associated with.
If the lock is not a wait lock, insert it to the head of the hash list.
Otherwise, insert it to the middle of the wait locks according to the age of
the transaciton. */
static
dberr_t
lock_rec_insert_by_trx_age(
lock_t *in_lock) /*!< in: lock to be insert */{
ulint space;
ulint page_no;
ulint rec_fold;
lock_t* node;
lock_t* next;
hash_cell_t* cell;
space = in_lock->un_member.rec_lock.space;
page_no = in_lock->un_member.rec_lock.page_no;
rec_fold = lock_rec_fold(space, page_no);
cell = hash_get_nth_cell(lock_sys->rec_hash,
hash_calc_hash(rec_fold, lock_sys->rec_hash));
node = (lock_t *) cell->node;
// If in_lock is not a wait lock, we insert it to the head of the list.
if (node == NULL || !lock_get_wait(in_lock) || has_higher_priority(in_lock, node)) {
cell->node = in_lock;
in_lock->hash = node;
if (lock_get_wait(in_lock)) {
lock_grant(in_lock, true);
return DB_SUCCESS_LOCKED_REC;
}
return DB_SUCCESS;
}
while (node != NULL && has_higher_priority((lock_t *) node->hash,
in_lock)) {
node = (lock_t *) node->hash;
}
next = (lock_t *) node->hash;
node->hash = in_lock;
in_lock->hash = next;
if (lock_get_wait(in_lock) && !lock_rec_has_to_wait_in_queue(in_lock)) {
lock_grant(in_lock, true);
if (cell->node != in_lock) {
// Move it to the front of the queue
node->hash = in_lock->hash;
next = (lock_t *) cell->node;
cell->node = in_lock;
in_lock->hash = next;
}
return DB_SUCCESS_LOCKED_REC;
}
return DB_SUCCESS;
}
static
bool
lock_queue_validate(
const lock_t *in_lock) /*!< in: lock whose hash list is to be validated */
{
ulint space;
ulint page_no;
ulint rec_fold;
hash_cell_t* cell;
lock_t* next;
bool wait_lock = false;
if (in_lock == NULL) {
return true;
}
space = in_lock->un_member.rec_lock.space;
page_no = in_lock->un_member.rec_lock.page_no;
rec_fold = lock_rec_fold(space, page_no);
cell = hash_get_nth_cell(lock_sys->rec_hash,
hash_calc_hash(rec_fold, lock_sys->rec_hash));
next = (lock_t *) cell->node;
while (next != NULL) {
// If this is a granted lock, check that there's no wait lock before it.
if (!lock_get_wait(next)) {
ut_ad(!wait_lock);
} else {
wait_lock = true;
}
next = (lock_t *) next->hash;
}
return true;
}
static
void
lock_rec_insert_to_head(
lock_t *in_lock, /*!< in: lock to be insert */
ulint rec_fold) /*!< in: rec_fold of the page */
{
hash_cell_t* cell;
lock_t* node;
if (in_lock == NULL) {
return;
}
cell = hash_get_nth_cell(lock_sys->rec_hash,
hash_calc_hash(rec_fold, lock_sys->rec_hash));
node = (lock_t *) cell->node;
if (node != in_lock) {
cell->node = in_lock;
in_lock->hash = node;
}
}
/*********************************************************************//**
Creates a new record lock and inserts it to the lock queue. Does NOT check
for deadlocks or lock compatibility!
@return created lock */
static
lock_t*
lock_rec_create(
/*============*/
#ifdef WITH_WSREP
lock_t* const c_lock, /* conflicting lock */
que_thr_t* thr,
#endif
ulint type_mode,/*!< in: lock mode and wait
flag, type is ignored and
replaced by LOCK_REC */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
dict_index_t* index, /*!< in: index of record */
trx_t* trx, /*!< in/out: transaction */
ibool caller_owns_trx_mutex)
/*!< in: TRUE if caller owns
trx mutex */
{
lock_t* lock;
ulint page_no;
ulint space;
ulint rec_fold;
ulint n_bits;
ulint n_bytes;
bool wait_lock;
const page_t* page;
ut_ad(lock_mutex_own());
ut_ad(caller_owns_trx_mutex == trx_mutex_own(trx));
ut_ad(dict_index_is_clust(index) || !dict_index_is_online_ddl(index));
/* Non-locking autocommit read-only transactions should not set
any locks. */
assert_trx_in_list(trx);
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
page = block->frame;
btr_assert_not_corrupted(block, index);
/* If rec is the supremum record, then we reset the gap and
LOCK_REC_NOT_GAP bits, as all locks on the supremum are
automatically of the gap type */
if (UNIV_UNLIKELY(heap_no == PAGE_HEAP_NO_SUPREMUM)) {
ut_ad(!(type_mode & LOCK_REC_NOT_GAP));
type_mode = type_mode & ~(LOCK_GAP | LOCK_REC_NOT_GAP);
}
wait_lock = type_mode & LOCK_WAIT;
/* Make lock bitmap bigger by a safety margin */
n_bits = page_dir_get_n_heap(page) + LOCK_PAGE_BITMAP_MARGIN;
n_bytes = 1 + n_bits / 8;
lock = static_cast<lock_t*>(
mem_heap_alloc(trx->lock.lock_heap, sizeof(lock_t) + n_bytes));
lock->trx = trx;
lock->type_mode = (type_mode & ~LOCK_TYPE_MASK) | LOCK_REC;
lock->index = index;
lock->un_member.rec_lock.space = space;
lock->un_member.rec_lock.page_no = page_no;
lock->un_member.rec_lock.n_bits = n_bytes * 8;
rec_fold = lock_rec_fold(space, page_no);
/* Reset to zero the bitmap which resides immediately after the
lock struct */
lock_rec_bitmap_reset(lock);
/* Set the bit corresponding to rec */
lock_rec_set_nth_bit(lock, heap_no);
lock->requested_time = ut_time();
lock->wait_time = 0;
index->table->n_rec_locks++;
ut_ad(index->table->n_ref_count > 0 || !index->table->can_be_evicted);
#ifdef WITH_WSREP
if (c_lock &&
wsrep_on(trx->mysql_thd) &&
wsrep_thd_is_BF(trx->mysql_thd, FALSE)) {
lock_t *hash = (lock_t *)c_lock->hash;
lock_t *prev = NULL;
while (hash &&
wsrep_thd_is_BF(((lock_t *)hash)->trx->mysql_thd, TRUE) &&
wsrep_trx_order_before(
((lock_t *)hash)->trx->mysql_thd,
trx->mysql_thd)) {
prev = hash;
hash = (lock_t *)hash->hash;
}
lock->hash = hash;
if (prev) {
prev->hash = lock;
} else {
c_lock->hash = lock;
}
/*
* delayed conflict resolution '...kill_one_trx' was not called,
* if victim was waiting for some other lock
*/
trx_mutex_enter(c_lock->trx);
if (c_lock->trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
c_lock->trx->lock.was_chosen_as_deadlock_victim = TRUE;
if (wsrep_debug) {
wsrep_print_wait_locks(c_lock);
}
trx->lock.que_state = TRX_QUE_LOCK_WAIT;
lock_set_lock_and_trx_wait(lock, trx);
UT_LIST_ADD_LAST(trx_locks, trx->lock.trx_locks, lock);
ut_ad(thr != NULL);
trx->lock.wait_thr = thr;
thr->state = QUE_THR_LOCK_WAIT;
/* have to release trx mutex for the duration of
victim lock release. This will eventually call
lock_grant, which wants to grant trx mutex again
*/
if (caller_owns_trx_mutex) {
trx_mutex_exit(trx);
}
lock_cancel_waiting_and_release(
c_lock->trx->lock.wait_lock);
if (caller_owns_trx_mutex) {
trx_mutex_enter(trx);
}
/* trx might not wait for c_lock, but some other lock
does not matter if wait_lock was released above
*/
if (c_lock->trx->lock.wait_lock == c_lock) {
lock_reset_lock_and_trx_wait(lock);
}
trx_mutex_exit(c_lock->trx);
if (wsrep_debug) {
fprintf(
stderr,
"WSREP: c_lock canceled %llu\n",
(ulonglong) c_lock->trx->id);
}
/* have to bail out here to avoid lock_set_lock... */
return(lock);
}
trx_mutex_exit(c_lock->trx);
} else if (innodb_lock_schedule_algorithm == INNODB_LOCK_SCHEDULE_ALGORITHM_VATS
&& !thd_is_replication_slave_thread(lock->trx->mysql_thd)) {
if (wait_lock) {
HASH_INSERT(lock_t, hash, lock_sys->rec_hash, rec_fold, lock);
} else {
lock_rec_insert_to_head(lock, rec_fold);
}
} else {
HASH_INSERT(lock_t, hash, lock_sys->rec_hash, rec_fold, lock);
}
#else
if (innodb_lock_schedule_algorithm == INNODB_LOCK_SCHEDULE_ALGORITHM_VATS
&& !thd_is_replication_slave_thread(lock->trx->mysql_thd)) {
if (wait_lock) {
HASH_INSERT(lock_t, hash, lock_sys->rec_hash, rec_fold, lock);
} else {
lock_rec_insert_to_head(lock, rec_fold);
}
} else {
HASH_INSERT(lock_t, hash, lock_sys->rec_hash, rec_fold, lock);
}
#endif /* WITH_WSREP */
lock_sys->rec_num++;
if (!caller_owns_trx_mutex) {
trx_mutex_enter(trx);
}
ut_ad(trx_mutex_own(trx));
if (type_mode & LOCK_WAIT) {
lock_set_lock_and_trx_wait(lock, trx);
}
UT_LIST_ADD_LAST(trx_locks, trx->lock.trx_locks, lock);
if (!caller_owns_trx_mutex) {
trx_mutex_exit(trx);
}
MONITOR_INC(MONITOR_RECLOCK_CREATED);
MONITOR_INC(MONITOR_NUM_RECLOCK);
return(lock);
}
/*********************************************************************//**
Enqueues a waiting request for a lock which cannot be granted immediately.
Checks for deadlocks.
@return DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED, or
DB_SUCCESS_LOCKED_REC; DB_SUCCESS_LOCKED_REC means that
there was a deadlock, but another transaction was chosen as a victim,
and we got the lock immediately: no need to wait then */
static
dberr_t
lock_rec_enqueue_waiting(
/*=====================*/
#ifdef WITH_WSREP
lock_t* c_lock, /* conflicting lock */
#endif
ulint type_mode,/*!< in: lock mode this
transaction is requesting:
LOCK_S or LOCK_X, possibly
ORed with LOCK_GAP or
LOCK_REC_NOT_GAP, ORed with
LOCK_INSERT_INTENTION if this
waiting lock request is set
when performing an insert of
an index record */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
dict_index_t* index, /*!< in: index of record */
que_thr_t* thr) /*!< in: query thread */
{
trx_t* trx;
lock_t* lock;
trx_id_t victim_trx_id;
ulint sec;
ulint ms;
ulint space;
ulint page_no;
dberr_t err;
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
ut_ad(dict_index_is_clust(index) || !dict_index_is_online_ddl(index));
trx = thr_get_trx(thr);
ut_ad(trx_mutex_own(trx));
/* Test if there already is some other reason to suspend thread:
we do not enqueue a lock request if the query thread should be
stopped anyway */
if (que_thr_stop(thr)) {
ut_error;
return(DB_QUE_THR_SUSPENDED);
}
switch (trx_get_dict_operation(trx)) {
case TRX_DICT_OP_NONE:
break;
case TRX_DICT_OP_TABLE:
case TRX_DICT_OP_INDEX:
ut_print_timestamp(stderr);
fputs(" InnoDB: Error: a record lock wait happens"
" in a dictionary operation!\n"
"InnoDB: ", stderr);
dict_index_name_print(stderr, trx, index);
fputs(".\n"
"InnoDB: Submit a detailed bug report"
" to http://bugs.mysql.com\n",
stderr);
ut_ad(0);
}
/* Enqueue the lock request that will wait to be granted, note that
we already own the trx mutex. */
lock = lock_rec_create(
#ifdef WITH_WSREP
c_lock, thr,
#endif /* WITH_WSREP */
type_mode | LOCK_WAIT, block, heap_no, index, trx, TRUE);
/* Release the mutex to obey the latching order.
This is safe, because lock_deadlock_check_and_resolve()
is invoked when a lock wait is enqueued for the currently
running transaction. Because trx is a running transaction
(it is not currently suspended because of a lock wait),
its state can only be changed by this thread, which is
currently associated with the transaction. */
trx_mutex_exit(trx);
victim_trx_id = lock_deadlock_check_and_resolve(lock, trx);
trx_mutex_enter(trx);
if (victim_trx_id != 0) {
ut_ad(victim_trx_id == trx->id);
lock_reset_lock_and_trx_wait(lock);
lock_rec_reset_nth_bit(lock, heap_no);
return(DB_DEADLOCK);
} else if (trx->lock.wait_lock == NULL) {
/* If there was a deadlock but we chose another
transaction as a victim, it is possible that we
already have the lock now granted! */
err = DB_SUCCESS_LOCKED_REC;
} else {
trx->lock.que_state = TRX_QUE_LOCK_WAIT;
trx->lock.was_chosen_as_deadlock_victim = FALSE;
trx->lock.wait_started = ut_time();
if (UNIV_UNLIKELY(trx->take_stats)) {
ut_usectime(&sec, &ms);
trx->lock_que_wait_ustarted = (ib_uint64_t)sec * 1000000 + ms;
}
ut_a(que_thr_stop(thr));
#ifdef UNIV_DEBUG
if (lock_print_waits) {
fprintf(stderr, "Lock wait for trx " TRX_ID_FMT " in index ",
trx->id);
ut_print_name(stderr, trx, FALSE, index->name);
}
#endif /* UNIV_DEBUG */
MONITOR_INC(MONITOR_LOCKREC_WAIT);
trx->n_rec_lock_waits++;
err = DB_LOCK_WAIT;
}
// Move it only when it does not cause a deadlock.
if (err != DB_DEADLOCK
&& innodb_lock_schedule_algorithm
== INNODB_LOCK_SCHEDULE_ALGORITHM_VATS
&& !thd_is_replication_slave_thread(lock->trx->mysql_thd)) {
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
HASH_DELETE(lock_t, hash, lock_sys->rec_hash,
lock_rec_fold(space, page_no), lock);
dberr_t res = lock_rec_insert_by_trx_age(lock);
if (res != DB_SUCCESS) {
return res;
}
}
return err;
}
/*********************************************************************//**
Adds a record lock request in the record queue. The request is normally
added as the last in the queue, but if there are no waiting lock requests
on the record, and the request to be added is not a waiting request, we
can reuse a suitable record lock object already existing on the same page,
just setting the appropriate bit in its bitmap. This is a low-level function
which does NOT check for deadlocks or lock compatibility!
@return lock where the bit was set */
static
lock_t*
lock_rec_add_to_queue(
/*==================*/
ulint type_mode,/*!< in: lock mode, wait, gap
etc. flags; type is ignored
and replaced by LOCK_REC */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of the record */
dict_index_t* index, /*!< in: index of record */
trx_t* trx, /*!< in/out: transaction */
ibool caller_owns_trx_mutex)
/*!< in: TRUE if caller owns the
transaction mutex */
{
lock_t* lock;
lock_t* first_lock;
ut_ad(lock_mutex_own());
ut_ad(caller_owns_trx_mutex == trx_mutex_own(trx));
ut_ad(dict_index_is_clust(index)
|| dict_index_get_online_status(index) != ONLINE_INDEX_CREATION);
#ifdef UNIV_DEBUG
switch (type_mode & LOCK_MODE_MASK) {
case LOCK_X:
case LOCK_S:
break;
default:
ut_error;
}
if (!(type_mode & (LOCK_WAIT | LOCK_GAP))) {
enum lock_mode mode = (type_mode & LOCK_MODE_MASK) == LOCK_S
? LOCK_X
: LOCK_S;
const lock_t* other_lock
= lock_rec_other_has_expl_req(mode, 0, LOCK_WAIT,
block, heap_no, trx->id);
#ifdef WITH_WSREP
/* this can potentionally assert with wsrep */
if (wsrep_thd_is_wsrep(trx->mysql_thd)) {
if (wsrep_debug && other_lock) {
fprintf(stderr,
"WSREP: InnoDB assert ignored\n");
}
} else {
ut_a(!other_lock);
}
#else
ut_a(!other_lock);
#endif /* WITH_WSREP */
}
#endif /* UNIV_DEBUG */
type_mode |= LOCK_REC;
/* If rec is the supremum record, then we can reset the gap bit, as
all locks on the supremum are automatically of the gap type, and we
try to avoid unnecessary memory consumption of a new record lock
struct for a gap type lock */
if (UNIV_UNLIKELY(heap_no == PAGE_HEAP_NO_SUPREMUM)) {
ut_ad(!(type_mode & LOCK_REC_NOT_GAP));
/* There should never be LOCK_REC_NOT_GAP on a supremum
record, but let us play safe */
type_mode = type_mode & ~(LOCK_GAP | LOCK_REC_NOT_GAP);
}
/* Look for a waiting lock request on the same record or on a gap */
for (first_lock = lock = lock_rec_get_first_on_page(block);
lock != NULL;
lock = lock_rec_get_next_on_page(lock)) {
if (lock_get_wait(lock)
&& lock_rec_get_nth_bit(lock, heap_no)) {
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(trx->mysql_thd, FALSE)) {
if (wsrep_debug) {
fprintf(stderr,
"BF skipping wait: "
TRX_ID_FMT "\n",
trx->id);
lock_rec_print(stderr, lock);
}
} else
#endif
goto somebody_waits;
}
}
if (UNIV_LIKELY(!(type_mode & LOCK_WAIT))) {
/* Look for a similar record lock on the same page:
if one is found and there are no waiting lock requests,
we can just set the bit */
lock = lock_rec_find_similar_on_page(
type_mode, heap_no, first_lock, trx);
if (lock) {
lock_rec_set_nth_bit(lock, heap_no);
return(lock);
}
}
somebody_waits:
#ifdef WITH_WSREP
return(lock_rec_create(NULL, NULL,
type_mode, block, heap_no, index, trx,
caller_owns_trx_mutex));
#else
return(lock_rec_create(
type_mode, block, heap_no, index, trx,
caller_owns_trx_mutex));
#endif /* WITH_WSREP */
}
/** Record locking request status */
enum lock_rec_req_status {
/** Failed to acquire a lock */
LOCK_REC_FAIL,
/** Succeeded in acquiring a lock (implicit or already acquired) */
LOCK_REC_SUCCESS,
/** Explicitly created a new lock */
LOCK_REC_SUCCESS_CREATED
};
/*********************************************************************//**
This is a fast routine for locking a record in the most common cases:
there are no explicit locks on the page, or there is just one lock, owned
by this transaction, and of the right type_mode. This is a low-level function
which does NOT look at implicit locks! Checks lock compatibility within
explicit locks. This function sets a normal next-key lock, or in the case of
a page supremum record, a gap type lock.
@return whether the locking succeeded */
UNIV_INLINE
enum lock_rec_req_status
lock_rec_lock_fast(
/*===============*/
ibool impl, /*!< in: if TRUE, no lock is set
if no wait is necessary: we
assume that the caller will
set an implicit lock */
ulint mode, /*!< in: lock mode: LOCK_X or
LOCK_S possibly ORed to either
LOCK_GAP or LOCK_REC_NOT_GAP */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of record */
dict_index_t* index, /*!< in: index of record */
que_thr_t* thr) /*!< in: query thread */
{
lock_t* lock;
trx_t* trx;
enum lock_rec_req_status status = LOCK_REC_SUCCESS;
ut_ad(lock_mutex_own());
ut_ad((LOCK_MODE_MASK & mode) != LOCK_S
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IS));
ut_ad((LOCK_MODE_MASK & mode) != LOCK_X
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IX));
ut_ad((LOCK_MODE_MASK & mode) == LOCK_S
|| (LOCK_MODE_MASK & mode) == LOCK_X);
ut_ad(mode - (LOCK_MODE_MASK & mode) == LOCK_GAP
|| mode - (LOCK_MODE_MASK & mode) == 0
|| mode - (LOCK_MODE_MASK & mode) == LOCK_REC_NOT_GAP);
ut_ad(dict_index_is_clust(index) || !dict_index_is_online_ddl(index));
DBUG_EXECUTE_IF("innodb_report_deadlock", return(LOCK_REC_FAIL););
lock = lock_rec_get_first_on_page(block);
trx = thr_get_trx(thr);
if (lock == NULL) {
if (!impl) {
/* Note that we don't own the trx mutex. */
#ifdef WITH_WSREP
lock = lock_rec_create(NULL, thr,
mode, block, heap_no, index, trx, FALSE);
#else
lock = lock_rec_create(
mode, block, heap_no, index, trx, FALSE);
#endif
}
status = LOCK_REC_SUCCESS_CREATED;
} else {
trx_mutex_enter(trx);
if (lock_rec_get_next_on_page(lock)
|| lock->trx != trx
|| lock->type_mode != (mode | LOCK_REC)
|| lock_rec_get_n_bits(lock) <= heap_no) {
status = LOCK_REC_FAIL;
} else if (!impl) {
/* If the nth bit of the record lock is already set
then we do not set a new lock bit, otherwise we do
set */
if (!lock_rec_get_nth_bit(lock, heap_no)) {
lock_rec_set_nth_bit(lock, heap_no);
status = LOCK_REC_SUCCESS_CREATED;
}
}
trx_mutex_exit(trx);
}
return(status);
}
/*********************************************************************//**
This is the general, and slower, routine for locking a record. This is a
low-level function which does NOT look at implicit locks! Checks lock
compatibility within explicit locks. This function sets a normal next-key
lock, or in the case of a page supremum record, a gap type lock.
@return DB_SUCCESS, DB_SUCCESS_LOCKED_REC, DB_LOCK_WAIT, DB_DEADLOCK,
or DB_QUE_THR_SUSPENDED */
static
dberr_t
lock_rec_lock_slow(
/*===============*/
ibool impl, /*!< in: if TRUE, no lock is set
if no wait is necessary: we
assume that the caller will
set an implicit lock */
ulint mode, /*!< in: lock mode: LOCK_X or
LOCK_S possibly ORed to either
LOCK_GAP or LOCK_REC_NOT_GAP */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of record */
dict_index_t* index, /*!< in: index of record */
que_thr_t* thr) /*!< in: query thread */
{
trx_t* trx;
#ifdef WITH_WSREP
lock_t* c_lock(NULL);
#endif
dberr_t err = DB_SUCCESS;
ut_ad(lock_mutex_own());
ut_ad((LOCK_MODE_MASK & mode) != LOCK_S
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IS));
ut_ad((LOCK_MODE_MASK & mode) != LOCK_X
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IX));
ut_ad((LOCK_MODE_MASK & mode) == LOCK_S
|| (LOCK_MODE_MASK & mode) == LOCK_X);
ut_ad(mode - (LOCK_MODE_MASK & mode) == LOCK_GAP
|| mode - (LOCK_MODE_MASK & mode) == 0
|| mode - (LOCK_MODE_MASK & mode) == LOCK_REC_NOT_GAP);
ut_ad(dict_index_is_clust(index) || !dict_index_is_online_ddl(index));
DBUG_EXECUTE_IF("innodb_report_deadlock", return(DB_DEADLOCK););
trx = thr_get_trx(thr);
trx_mutex_enter(trx);
if (lock_rec_has_expl(mode, block, heap_no, trx->id)) {
/* The trx already has a strong enough lock on rec: do
nothing */
#ifdef WITH_WSREP
} else if ((c_lock = (lock_t *)lock_rec_other_has_conflicting(
static_cast<enum lock_mode>(mode),
block, heap_no, trx))) {
#else
} else if (lock_rec_other_has_conflicting(
static_cast<enum lock_mode>(mode),
block, heap_no, trx)) {
#endif /* WITH_WSREP */
/* If another transaction has a non-gap conflicting
request in the queue, as this transaction does not
have a lock strong enough already granted on the
record, we have to wait. */
#ifdef WITH_WSREP
/* c_lock is NULL here if jump to enqueue_waiting happened
but it's ok because lock is not NULL in that case and c_lock
is not used. */
err = lock_rec_enqueue_waiting(c_lock,
mode, block, heap_no, index, thr);
#else
err = lock_rec_enqueue_waiting(
mode, block, heap_no, index, thr);
#endif /* WITH_WSREP */
} else if (!impl) {
/* Set the requested lock on the record, note that
we already own the transaction mutex. */
lock_rec_add_to_queue(
LOCK_REC | mode, block, heap_no, index, trx, TRUE);
err = DB_SUCCESS_LOCKED_REC;
}
trx_mutex_exit(trx);
return(err);
}
/*********************************************************************//**
Tries to lock the specified record in the mode requested. If not immediately
possible, enqueues a waiting lock request. This is a low-level function
which does NOT look at implicit locks! Checks lock compatibility within
explicit locks. This function sets a normal next-key lock, or in the case
of a page supremum record, a gap type lock.
@return DB_SUCCESS, DB_SUCCESS_LOCKED_REC, DB_LOCK_WAIT, DB_DEADLOCK,
or DB_QUE_THR_SUSPENDED */
static
dberr_t
lock_rec_lock(
/*==========*/
ibool impl, /*!< in: if TRUE, no lock is set
if no wait is necessary: we
assume that the caller will
set an implicit lock */
ulint mode, /*!< in: lock mode: LOCK_X or
LOCK_S possibly ORed to either
LOCK_GAP or LOCK_REC_NOT_GAP */
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no,/*!< in: heap number of record */
dict_index_t* index, /*!< in: index of record */
que_thr_t* thr) /*!< in: query thread */
{
ut_ad(lock_mutex_own());
ut_ad((LOCK_MODE_MASK & mode) != LOCK_S
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IS));
ut_ad((LOCK_MODE_MASK & mode) != LOCK_X
|| lock_table_has(thr_get_trx(thr), index->table, LOCK_IX));
ut_ad((LOCK_MODE_MASK & mode) == LOCK_S
|| (LOCK_MODE_MASK & mode) == LOCK_X);
ut_ad(mode - (LOCK_MODE_MASK & mode) == LOCK_GAP
|| mode - (LOCK_MODE_MASK & mode) == LOCK_REC_NOT_GAP
|| mode - (LOCK_MODE_MASK & mode) == 0);
ut_ad(dict_index_is_clust(index) || !dict_index_is_online_ddl(index));
/* We try a simplified and faster subroutine for the most
common cases */
switch (lock_rec_lock_fast(impl, mode, block, heap_no, index, thr)) {
case LOCK_REC_SUCCESS:
return(DB_SUCCESS);
case LOCK_REC_SUCCESS_CREATED:
return(DB_SUCCESS_LOCKED_REC);
case LOCK_REC_FAIL:
return(lock_rec_lock_slow(impl, mode, block,
heap_no, index, thr));
}
ut_error;
return(DB_ERROR);
}
/*********************************************************************//**
Checks if a waiting record lock request still has to wait in a queue.
@return lock that is causing the wait */
static
const lock_t*
lock_rec_has_to_wait_in_queue(
/*==========================*/
const lock_t* wait_lock) /*!< in: waiting record lock */
{
const lock_t* lock;
ulint space;
ulint page_no;
ulint heap_no;
ulint bit_mask;
ulint bit_offset;
ut_ad(lock_mutex_own());
ut_ad(lock_get_wait(wait_lock));
ut_ad(lock_get_type_low(wait_lock) == LOCK_REC);
space = wait_lock->un_member.rec_lock.space;
page_no = wait_lock->un_member.rec_lock.page_no;
heap_no = lock_rec_find_set_bit(wait_lock);
bit_offset = heap_no / 8;
bit_mask = static_cast<ulint>(1 << (heap_no % 8));
for (lock = lock_rec_get_first_on_page_addr(space, page_no);
lock != wait_lock;
lock = lock_rec_get_next_on_page_const(lock)) {
const byte* p = (const byte*) &lock[1];
if (heap_no < lock_rec_get_n_bits(lock)
&& (p[bit_offset] & bit_mask)
&& lock_has_to_wait(wait_lock, lock)) {
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(wait_lock->trx->mysql_thd, FALSE) &&
wsrep_thd_is_BF(lock->trx->mysql_thd, TRUE)) {
/* don't wait for another BF lock */
continue;
}
#endif
return(lock);
}
}
return(NULL);
}
/*************************************************************//**
Grants a lock to a waiting lock request and releases the waiting transaction.
The caller must hold lock_sys->mutex but not lock->trx->mutex. */
static
void
lock_grant(
/*=======*/
lock_t* lock, /*!< in/out: waiting lock request */
bool owns_trx_mutex) /*!< in: whether lock->trx->mutex is owned */
{
ut_ad(lock_mutex_own());
lock_reset_lock_and_trx_wait(lock);
if (!owns_trx_mutex) {
trx_mutex_enter(lock->trx);
}
if (lock_get_mode(lock) == LOCK_AUTO_INC) {
dict_table_t* table = lock->un_member.tab_lock.table;
if (UNIV_UNLIKELY(table->autoinc_trx == lock->trx)) {
fprintf(stderr,
"InnoDB: Error: trx already had"
" an AUTO-INC lock!\n");
} else {
table->autoinc_trx = lock->trx;
ib_vector_push(lock->trx->autoinc_locks, &lock);
}
}
#ifdef UNIV_DEBUG
if (lock_print_waits) {
fprintf(stderr, "Lock wait for trx " TRX_ID_FMT " ends\n",
lock->trx->id);
}
#endif /* UNIV_DEBUG */
/* If we are resolving a deadlock by choosing another transaction
as a victim, then our original transaction may not be in the
TRX_QUE_LOCK_WAIT state, and there is no need to end the lock wait
for it */
if (lock->trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
que_thr_t* thr;
thr = que_thr_end_lock_wait(lock->trx);
if (thr != NULL) {
lock_wait_release_thread_if_suspended(thr);
}
}
/* Cumulate total lock wait time for statistics */
if (lock_get_type_low(lock) & LOCK_TABLE) {
lock->trx->total_table_lock_wait_time +=
(ulint)difftime(ut_time(), lock->trx->lock.wait_started);
} else {
lock->trx->total_rec_lock_wait_time +=
(ulint)difftime(ut_time(), lock->trx->lock.wait_started);
}
lock->wait_time = (ulint)difftime(ut_time(), lock->requested_time);
if (!owns_trx_mutex) {
trx_mutex_exit(lock->trx);
}
}
/*************************************************************//**
Cancels a waiting record lock request and releases the waiting transaction
that requested it. NOTE: does NOT check if waiting lock requests behind this
one can now be granted! */
static
void
lock_rec_cancel(
/*============*/
lock_t* lock) /*!< in: waiting record lock request */
{
que_thr_t* thr;
ut_ad(lock_mutex_own());
ut_ad(lock_get_type_low(lock) == LOCK_REC);
/* Reset the bit (there can be only one set bit) in the lock bitmap */
lock_rec_reset_nth_bit(lock, lock_rec_find_set_bit(lock));
/* Reset the wait flag and the back pointer to lock in trx */
lock_reset_lock_and_trx_wait(lock);
/* The following function releases the trx from lock wait */
trx_mutex_enter(lock->trx);
thr = que_thr_end_lock_wait(lock->trx);
if (thr != NULL) {
lock_wait_release_thread_if_suspended(thr);
}
trx_mutex_exit(lock->trx);
}
static
void
lock_grant_and_move_on_page(
ulint space,
ulint page_no)
{
lock_t* lock;
lock_t* next;
lock_t* previous;
ulint rec_fold = lock_rec_fold(space, page_no);
previous = (lock_t *) hash_get_nth_cell(lock_sys->rec_hash,
hash_calc_hash(rec_fold, lock_sys->rec_hash))->node;
if (previous == NULL) {
return;
}
if (previous->un_member.rec_lock.space == space &&
previous->un_member.rec_lock.page_no == page_no) {
lock = previous;
}
else {
next = (lock_t *) previous->hash;
while (next &&
(next->un_member.rec_lock.space != space ||
next->un_member.rec_lock.page_no != page_no)) {
previous = next;
next = (lock_t *) previous->hash;
}
lock = (lock_t *) previous->hash;
}
ut_ad(previous->hash == lock || previous == lock);
/* Grant locks if there are no conflicting locks ahead.
Move granted locks to the head of the list. */
for (;lock != NULL;) {
/* If the lock is a wait lock on this page, and it does not need to wait. */
if ((lock->un_member.rec_lock.space == space)
&& (lock->un_member.rec_lock.page_no == page_no)
&& lock_get_wait(lock)
&& !lock_rec_has_to_wait_in_queue(lock)) {
lock_grant(lock, false);
if (previous != NULL) {
/* Move the lock to the head of the list. */
HASH_GET_NEXT(hash, previous) = HASH_GET_NEXT(hash, lock);
lock_rec_insert_to_head(lock, rec_fold);
} else {
/* Already at the head of the list. */
previous = lock;
}
/* Move on to the next lock. */
lock = static_cast<lock_t *>(HASH_GET_NEXT(hash, previous));
} else {
previous = lock;
lock = static_cast<lock_t *>(HASH_GET_NEXT(hash, lock));
}
}
}
/*************************************************************//**
Removes a record lock request, waiting or granted, from the queue and
grants locks to other transactions in the queue if they now are entitled
to a lock. NOTE: all record locks contained in in_lock are removed. */
static
void
lock_rec_dequeue_from_page(
/*=======================*/
lock_t* in_lock) /*!< in: record lock object: all
record locks which are contained in
this lock object are removed;
transactions waiting behind will
get their lock requests granted,
if they are now qualified to it */
{
ulint space;
ulint page_no;
lock_t* lock;
trx_lock_t* trx_lock;
ut_ad(lock_mutex_own());
ut_ad(lock_get_type_low(in_lock) == LOCK_REC);
/* We may or may not be holding in_lock->trx->mutex here. */
trx_lock = &in_lock->trx->lock;
space = in_lock->un_member.rec_lock.space;
page_no = in_lock->un_member.rec_lock.page_no;
in_lock->index->table->n_rec_locks--;
HASH_DELETE(lock_t, hash, lock_sys->rec_hash,
lock_rec_fold(space, page_no), in_lock);
lock_sys->rec_num--;
UT_LIST_REMOVE(trx_locks, trx_lock->trx_locks, in_lock);
MONITOR_INC(MONITOR_RECLOCK_REMOVED);
MONITOR_DEC(MONITOR_NUM_RECLOCK);
if (innodb_lock_schedule_algorithm
== INNODB_LOCK_SCHEDULE_ALGORITHM_FCFS ||
thd_is_replication_slave_thread(in_lock->trx->mysql_thd)) {
/* Check if waiting locks in the queue can now be granted: grant
locks if there are no conflicting locks ahead. Stop at the first
X lock that is waiting or has been granted. */
for (lock = lock_rec_get_first_on_page_addr(space, page_no);
lock != NULL;
lock = lock_rec_get_next_on_page(lock)) {
if (lock_get_wait(lock)
&& !lock_rec_has_to_wait_in_queue(lock)) {
/* Grant the lock */
ut_ad(lock->trx != in_lock->trx);
lock_grant(lock, false);
}
}
} else {
lock_grant_and_move_on_page(space, page_no);
}
}
/*************************************************************//**
Removes a record lock request, waiting or granted, from the queue. */
static
void
lock_rec_discard(
/*=============*/
lock_t* in_lock) /*!< in: record lock object: all
record locks which are contained
in this lock object are removed */
{
ulint space;
ulint page_no;
trx_lock_t* trx_lock;
ut_ad(lock_mutex_own());
ut_ad(lock_get_type_low(in_lock) == LOCK_REC);
trx_lock = &in_lock->trx->lock;
space = in_lock->un_member.rec_lock.space;
page_no = in_lock->un_member.rec_lock.page_no;
in_lock->index->table->n_rec_locks--;
HASH_DELETE(lock_t, hash, lock_sys->rec_hash,
lock_rec_fold(space, page_no), in_lock);
lock_sys->rec_num--;
UT_LIST_REMOVE(trx_locks, trx_lock->trx_locks, in_lock);
MONITOR_INC(MONITOR_RECLOCK_REMOVED);
MONITOR_DEC(MONITOR_NUM_RECLOCK);
}
/*************************************************************//**
Removes record lock objects set on an index page which is discarded. This
function does not move locks, or check for waiting locks, therefore the
lock bitmaps must already be reset when this function is called. */
static
void
lock_rec_free_all_from_discard_page(
/*================================*/
const buf_block_t* block) /*!< in: page to be discarded */
{
ulint space;
ulint page_no;
lock_t* lock;
lock_t* next_lock;
ut_ad(lock_mutex_own());
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
lock = lock_rec_get_first_on_page_addr(space, page_no);
while (lock != NULL) {
ut_ad(lock_rec_find_set_bit(lock) == ULINT_UNDEFINED);
ut_ad(!lock_get_wait(lock));
next_lock = lock_rec_get_next_on_page(lock);
lock_rec_discard(lock);
lock = next_lock;
}
}
/*============= RECORD LOCK MOVING AND INHERITING ===================*/
/*************************************************************//**
Resets the lock bits for a single record. Releases transactions waiting for
lock requests here. */
static
void
lock_rec_reset_and_release_wait(
/*============================*/
const buf_block_t* block, /*!< in: buffer block containing
the record */
ulint heap_no)/*!< in: heap number of record */
{
lock_t* lock;
ut_ad(lock_mutex_own());
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (lock_get_wait(lock)) {
lock_rec_cancel(lock);
} else {
lock_rec_reset_nth_bit(lock, heap_no);
}
}
}
/*************************************************************//**
Makes a record to inherit the locks (except LOCK_INSERT_INTENTION type)
of another record as gap type locks, but does not reset the lock bits of
the other record. Also waiting lock requests on rec are inherited as
GRANTED gap locks. */
static
void
lock_rec_inherit_to_gap(
/*====================*/
const buf_block_t* heir_block, /*!< in: block containing the
record which inherits */
const buf_block_t* block, /*!< in: block containing the
record from which inherited;
does NOT reset the locks on
this record */
ulint heir_heap_no, /*!< in: heap_no of the
inheriting record */
ulint heap_no) /*!< in: heap_no of the
donating record */
{
lock_t* lock;
ut_ad(lock_mutex_own());
/* If srv_locks_unsafe_for_binlog is TRUE or session is using
READ COMMITTED isolation level, we do not want locks set
by an UPDATE or a DELETE to be inherited as gap type locks. But we
DO want S-locks/X-locks(taken for replace) set by a consistency
constraint to be inherited also then */
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (!lock_rec_get_insert_intention(lock)
&& !((srv_locks_unsafe_for_binlog
|| lock->trx->isolation_level
<= TRX_ISO_READ_COMMITTED)
&& lock_get_mode(lock) ==
(lock->trx->duplicates ? LOCK_S : LOCK_X))) {
lock_rec_add_to_queue(
LOCK_REC | LOCK_GAP | lock_get_mode(lock),
heir_block, heir_heap_no, lock->index,
lock->trx, FALSE);
}
}
}
/*************************************************************//**
Makes a record to inherit the gap locks (except LOCK_INSERT_INTENTION type)
of another record as gap type locks, but does not reset the lock bits of the
other record. Also waiting lock requests are inherited as GRANTED gap locks. */
static
void
lock_rec_inherit_to_gap_if_gap_lock(
/*================================*/
const buf_block_t* block, /*!< in: buffer block */
ulint heir_heap_no, /*!< in: heap_no of
record which inherits */
ulint heap_no) /*!< in: heap_no of record
from which inherited;
does NOT reset the locks
on this record */
{
lock_t* lock;
lock_mutex_enter();
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (!lock_rec_get_insert_intention(lock)
&& (heap_no == PAGE_HEAP_NO_SUPREMUM
|| !lock_rec_get_rec_not_gap(lock))) {
lock_rec_add_to_queue(
LOCK_REC | LOCK_GAP | lock_get_mode(lock),
block, heir_heap_no, lock->index,
lock->trx, FALSE);
}
}
lock_mutex_exit();
}
/*************************************************************//**
Moves the locks of a record to another record and resets the lock bits of
the donating record. */
static
void
lock_rec_move(
/*==========*/
const buf_block_t* receiver, /*!< in: buffer block containing
the receiving record */
const buf_block_t* donator, /*!< in: buffer block containing
the donating record */
ulint receiver_heap_no,/*!< in: heap_no of the record
which gets the locks; there
must be no lock requests
on it! */
ulint donator_heap_no)/*!< in: heap_no of the record
which gives the locks */
{
lock_t* lock;
ut_ad(lock_mutex_own());
ut_ad(lock_rec_get_first(receiver, receiver_heap_no) == NULL);
for (lock = lock_rec_get_first(donator, donator_heap_no);
lock != NULL;
lock = lock_rec_get_next(donator_heap_no, lock)) {
const ulint type_mode = lock->type_mode;
lock_rec_reset_nth_bit(lock, donator_heap_no);
if (UNIV_UNLIKELY(type_mode & LOCK_WAIT)) {
lock_reset_lock_and_trx_wait(lock);
}
/* Note that we FIRST reset the bit, and then set the lock:
the function works also if donator == receiver */
lock_rec_add_to_queue(
type_mode, receiver, receiver_heap_no,
lock->index, lock->trx, FALSE);
}
ut_ad(lock_rec_get_first(donator, donator_heap_no) == NULL);
}
/*************************************************************//**
Updates the lock table when we have reorganized a page. NOTE: we copy
also the locks set on the infimum of the page; the infimum may carry
locks if an update of a record is occurring on the page, and its locks
were temporarily stored on the infimum. */
UNIV_INTERN
void
lock_move_reorganize_page(
/*======================*/
const buf_block_t* block, /*!< in: old index page, now
reorganized */
const buf_block_t* oblock) /*!< in: copy of the old, not
reorganized page */
{
lock_t* lock;
UT_LIST_BASE_NODE_T(lock_t) old_locks;
mem_heap_t* heap = NULL;
ulint comp;
lock_mutex_enter();
lock = lock_rec_get_first_on_page(block);
if (lock == NULL) {
lock_mutex_exit();
return;
}
heap = mem_heap_create(256);
/* Copy first all the locks on the page to heap and reset the
bitmaps in the original locks; chain the copies of the locks
using the trx_locks field in them. */
UT_LIST_INIT(old_locks);
do {
/* Make a copy of the lock */
lock_t* old_lock = lock_rec_copy(lock, heap);
UT_LIST_ADD_LAST(trx_locks, old_locks, old_lock);
/* Reset bitmap of lock */
lock_rec_bitmap_reset(lock);
if (lock_get_wait(lock)) {
lock_reset_lock_and_trx_wait(lock);
}
lock = lock_rec_get_next_on_page(lock);
} while (lock != NULL);
comp = page_is_comp(block->frame);
ut_ad(comp == page_is_comp(oblock->frame));
for (lock = UT_LIST_GET_FIRST(old_locks); lock;
lock = UT_LIST_GET_NEXT(trx_locks, lock)) {
/* NOTE: we copy also the locks set on the infimum and
supremum of the page; the infimum may carry locks if an
update of a record is occurring on the page, and its locks
were temporarily stored on the infimum */
page_cur_t cur1;
page_cur_t cur2;
page_cur_set_before_first(block, &cur1);
page_cur_set_before_first(oblock, &cur2);
/* Set locks according to old locks */
for (;;) {
ulint old_heap_no;
ulint new_heap_no;
ut_ad(comp || !memcmp(page_cur_get_rec(&cur1),
page_cur_get_rec(&cur2),
rec_get_data_size_old(
page_cur_get_rec(
&cur2))));
if (UNIV_LIKELY(comp)) {
old_heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur2));
new_heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur1));
} else {
old_heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur2));
new_heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur1));
}
if (lock_rec_get_nth_bit(lock, old_heap_no)) {
/* Clear the bit in old_lock. */
ut_d(lock_rec_reset_nth_bit(lock,
old_heap_no));
/* NOTE that the old lock bitmap could be too
small for the new heap number! */
lock_rec_add_to_queue(
lock->type_mode, block, new_heap_no,
lock->index, lock->trx, FALSE);
/* if (new_heap_no == PAGE_HEAP_NO_SUPREMUM
&& lock_get_wait(lock)) {
fprintf(stderr,
"---\n--\n!!!Lock reorg: supr type %lu\n",
lock->type_mode);
} */
}
if (UNIV_UNLIKELY
(new_heap_no == PAGE_HEAP_NO_SUPREMUM)) {
ut_ad(old_heap_no == PAGE_HEAP_NO_SUPREMUM);
break;
}
page_cur_move_to_next(&cur1);
page_cur_move_to_next(&cur2);
}
#ifdef UNIV_DEBUG
{
ulint i = lock_rec_find_set_bit(lock);
/* Check that all locks were moved. */
if (UNIV_UNLIKELY(i != ULINT_UNDEFINED)) {
fprintf(stderr,
"lock_move_reorganize_page():"
" %lu not moved in %p\n",
(ulong) i, (void*) lock);
ut_error;
}
}
#endif /* UNIV_DEBUG */
}
lock_mutex_exit();
mem_heap_free(heap);
#ifdef UNIV_DEBUG_LOCK_VALIDATE
ut_ad(lock_rec_validate_page(block));
#endif
}
/*************************************************************//**
Moves the explicit locks on user records to another page if a record
list end is moved to another page. */
UNIV_INTERN
void
lock_move_rec_list_end(
/*===================*/
const buf_block_t* new_block, /*!< in: index page to move to */
const buf_block_t* block, /*!< in: index page */
const rec_t* rec) /*!< in: record on page: this
is the first record moved */
{
lock_t* lock;
const ulint comp = page_rec_is_comp(rec);
lock_mutex_enter();
/* Note: when we move locks from record to record, waiting locks
and possible granted gap type locks behind them are enqueued in
the original order, because new elements are inserted to a hash
table to the end of the hash chain, and lock_rec_add_to_queue
does not reuse locks if there are waiters in the queue. */
for (lock = lock_rec_get_first_on_page(block); lock;
lock = lock_rec_get_next_on_page(lock)) {
page_cur_t cur1;
page_cur_t cur2;
const ulint type_mode = lock->type_mode;
page_cur_position(rec, block, &cur1);
if (page_cur_is_before_first(&cur1)) {
page_cur_move_to_next(&cur1);
}
page_cur_set_before_first(new_block, &cur2);
page_cur_move_to_next(&cur2);
/* Copy lock requests on user records to new page and
reset the lock bits on the old */
while (!page_cur_is_after_last(&cur1)) {
ulint heap_no;
if (comp) {
heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur1));
} else {
heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur1));
ut_ad(!memcmp(page_cur_get_rec(&cur1),
page_cur_get_rec(&cur2),
rec_get_data_size_old(
page_cur_get_rec(&cur2))));
}
if (lock_rec_get_nth_bit(lock, heap_no)) {
lock_rec_reset_nth_bit(lock, heap_no);
if (UNIV_UNLIKELY(type_mode & LOCK_WAIT)) {
lock_reset_lock_and_trx_wait(lock);
}
if (comp) {
heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur2));
} else {
heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur2));
}
lock_rec_add_to_queue(
type_mode, new_block, heap_no,
lock->index, lock->trx, FALSE);
}
page_cur_move_to_next(&cur1);
page_cur_move_to_next(&cur2);
}
}
lock_mutex_exit();
#ifdef UNIV_DEBUG_LOCK_VALIDATE
ut_ad(lock_rec_validate_page(block));
ut_ad(lock_rec_validate_page(new_block));
#endif
}
/*************************************************************//**
Moves the explicit locks on user records to another page if a record
list start is moved to another page. */
UNIV_INTERN
void
lock_move_rec_list_start(
/*=====================*/
const buf_block_t* new_block, /*!< in: index page to
move to */
const buf_block_t* block, /*!< in: index page */
const rec_t* rec, /*!< in: record on page:
this is the first
record NOT copied */
const rec_t* old_end) /*!< in: old
previous-to-last
record on new_page
before the records
were copied */
{
lock_t* lock;
const ulint comp = page_rec_is_comp(rec);
ut_ad(block->frame == page_align(rec));
ut_ad(new_block->frame == page_align(old_end));
lock_mutex_enter();
for (lock = lock_rec_get_first_on_page(block); lock;
lock = lock_rec_get_next_on_page(lock)) {
page_cur_t cur1;
page_cur_t cur2;
const ulint type_mode = lock->type_mode;
page_cur_set_before_first(block, &cur1);
page_cur_move_to_next(&cur1);
page_cur_position(old_end, new_block, &cur2);
page_cur_move_to_next(&cur2);
/* Copy lock requests on user records to new page and
reset the lock bits on the old */
while (page_cur_get_rec(&cur1) != rec) {
ulint heap_no;
if (comp) {
heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur1));
} else {
heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur1));
ut_ad(!memcmp(page_cur_get_rec(&cur1),
page_cur_get_rec(&cur2),
rec_get_data_size_old(
page_cur_get_rec(
&cur2))));
}
if (lock_rec_get_nth_bit(lock, heap_no)) {
lock_rec_reset_nth_bit(lock, heap_no);
if (UNIV_UNLIKELY(type_mode & LOCK_WAIT)) {
lock_reset_lock_and_trx_wait(lock);
}
if (comp) {
heap_no = rec_get_heap_no_new(
page_cur_get_rec(&cur2));
} else {
heap_no = rec_get_heap_no_old(
page_cur_get_rec(&cur2));
}
lock_rec_add_to_queue(
type_mode, new_block, heap_no,
lock->index, lock->trx, FALSE);
}
page_cur_move_to_next(&cur1);
page_cur_move_to_next(&cur2);
}
#ifdef UNIV_DEBUG
if (page_rec_is_supremum(rec)) {
ulint i;
for (i = PAGE_HEAP_NO_USER_LOW;
i < lock_rec_get_n_bits(lock); i++) {
if (UNIV_UNLIKELY
(lock_rec_get_nth_bit(lock, i))) {
fprintf(stderr,
"lock_move_rec_list_start():"
" %lu not moved in %p\n",
(ulong) i, (void*) lock);
ut_error;
}
}
}
#endif /* UNIV_DEBUG */
}
lock_mutex_exit();
#ifdef UNIV_DEBUG_LOCK_VALIDATE
ut_ad(lock_rec_validate_page(block));
#endif
}
/*************************************************************//**
Updates the lock table when a page is split to the right. */
UNIV_INTERN
void
lock_update_split_right(
/*====================*/
const buf_block_t* right_block, /*!< in: right page */
const buf_block_t* left_block) /*!< in: left page */
{
ulint heap_no = lock_get_min_heap_no(right_block);
lock_mutex_enter();
/* Move the locks on the supremum of the left page to the supremum
of the right page */
lock_rec_move(right_block, left_block,
PAGE_HEAP_NO_SUPREMUM, PAGE_HEAP_NO_SUPREMUM);
/* Inherit the locks to the supremum of left page from the successor
of the infimum on right page */
lock_rec_inherit_to_gap(left_block, right_block,
PAGE_HEAP_NO_SUPREMUM, heap_no);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is merged to the right. */
UNIV_INTERN
void
lock_update_merge_right(
/*====================*/
const buf_block_t* right_block, /*!< in: right page to
which merged */
const rec_t* orig_succ, /*!< in: original
successor of infimum
on the right page
before merge */
const buf_block_t* left_block) /*!< in: merged index
page which will be
discarded */
{
lock_mutex_enter();
/* Inherit the locks from the supremum of the left page to the
original successor of infimum on the right page, to which the left
page was merged */
lock_rec_inherit_to_gap(right_block, left_block,
page_rec_get_heap_no(orig_succ),
PAGE_HEAP_NO_SUPREMUM);
/* Reset the locks on the supremum of the left page, releasing
waiting transactions */
lock_rec_reset_and_release_wait(left_block,
PAGE_HEAP_NO_SUPREMUM);
lock_rec_free_all_from_discard_page(left_block);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when the root page is copied to another in
btr_root_raise_and_insert. Note that we leave lock structs on the
root page, even though they do not make sense on other than leaf
pages: the reason is that in a pessimistic update the infimum record
of the root page will act as a dummy carrier of the locks of the record
to be updated. */
UNIV_INTERN
void
lock_update_root_raise(
/*===================*/
const buf_block_t* block, /*!< in: index page to which copied */
const buf_block_t* root) /*!< in: root page */
{
lock_mutex_enter();
/* Move the locks on the supremum of the root to the supremum
of block */
lock_rec_move(block, root,
PAGE_HEAP_NO_SUPREMUM, PAGE_HEAP_NO_SUPREMUM);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is copied to another and the original page
is removed from the chain of leaf pages, except if page is the root! */
UNIV_INTERN
void
lock_update_copy_and_discard(
/*=========================*/
const buf_block_t* new_block, /*!< in: index page to
which copied */
const buf_block_t* block) /*!< in: index page;
NOT the root! */
{
lock_mutex_enter();
/* Move the locks on the supremum of the old page to the supremum
of new_page */
lock_rec_move(new_block, block,
PAGE_HEAP_NO_SUPREMUM, PAGE_HEAP_NO_SUPREMUM);
lock_rec_free_all_from_discard_page(block);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is split to the left. */
UNIV_INTERN
void
lock_update_split_left(
/*===================*/
const buf_block_t* right_block, /*!< in: right page */
const buf_block_t* left_block) /*!< in: left page */
{
ulint heap_no = lock_get_min_heap_no(right_block);
lock_mutex_enter();
/* Inherit the locks to the supremum of the left page from the
successor of the infimum on the right page */
lock_rec_inherit_to_gap(left_block, right_block,
PAGE_HEAP_NO_SUPREMUM, heap_no);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is merged to the left. */
UNIV_INTERN
void
lock_update_merge_left(
/*===================*/
const buf_block_t* left_block, /*!< in: left page to
which merged */
const rec_t* orig_pred, /*!< in: original predecessor
of supremum on the left page
before merge */
const buf_block_t* right_block) /*!< in: merged index page
which will be discarded */
{
const rec_t* left_next_rec;
ut_ad(left_block->frame == page_align(orig_pred));
lock_mutex_enter();
left_next_rec = page_rec_get_next_const(orig_pred);
if (!page_rec_is_supremum(left_next_rec)) {
/* Inherit the locks on the supremum of the left page to the
first record which was moved from the right page */
lock_rec_inherit_to_gap(left_block, left_block,
page_rec_get_heap_no(left_next_rec),
PAGE_HEAP_NO_SUPREMUM);
/* Reset the locks on the supremum of the left page,
releasing waiting transactions */
lock_rec_reset_and_release_wait(left_block,
PAGE_HEAP_NO_SUPREMUM);
}
/* Move the locks from the supremum of right page to the supremum
of the left page */
lock_rec_move(left_block, right_block,
PAGE_HEAP_NO_SUPREMUM, PAGE_HEAP_NO_SUPREMUM);
lock_rec_free_all_from_discard_page(right_block);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is split and merged to
two pages. */
UNIV_INTERN
void
lock_update_split_and_merge(
const buf_block_t* left_block, /*!< in: left page to which merged */
const rec_t* orig_pred, /*!< in: original predecessor of
supremum on the left page before merge*/
const buf_block_t* right_block) /*!< in: right page from which merged */
{
const rec_t* left_next_rec;
ut_a(left_block && right_block);
ut_a(orig_pred);
lock_mutex_enter();
left_next_rec = page_rec_get_next_const(orig_pred);
/* Inherit the locks on the supremum of the left page to the
first record which was moved from the right page */
lock_rec_inherit_to_gap(
left_block, left_block,
page_rec_get_heap_no(left_next_rec),
PAGE_HEAP_NO_SUPREMUM);
/* Reset the locks on the supremum of the left page,
releasing waiting transactions */
lock_rec_reset_and_release_wait(left_block,
PAGE_HEAP_NO_SUPREMUM);
/* Inherit the locks to the supremum of the left page from the
successor of the infimum on the right page */
lock_rec_inherit_to_gap(left_block, right_block,
PAGE_HEAP_NO_SUPREMUM,
lock_get_min_heap_no(right_block));
lock_mutex_exit();
}
/*************************************************************//**
Resets the original locks on heir and replaces them with gap type locks
inherited from rec. */
UNIV_INTERN
void
lock_rec_reset_and_inherit_gap_locks(
/*=================================*/
const buf_block_t* heir_block, /*!< in: block containing the
record which inherits */
const buf_block_t* block, /*!< in: block containing the
record from which inherited;
does NOT reset the locks on
this record */
ulint heir_heap_no, /*!< in: heap_no of the
inheriting record */
ulint heap_no) /*!< in: heap_no of the
donating record */
{
lock_mutex_enter();
lock_rec_reset_and_release_wait(heir_block, heir_heap_no);
lock_rec_inherit_to_gap(heir_block, block, heir_heap_no, heap_no);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a page is discarded. */
UNIV_INTERN
void
lock_update_discard(
/*================*/
const buf_block_t* heir_block, /*!< in: index page
which will inherit the locks */
ulint heir_heap_no, /*!< in: heap_no of the record
which will inherit the locks */
const buf_block_t* block) /*!< in: index page
which will be discarded */
{
const page_t* page = block->frame;
const rec_t* rec;
ulint heap_no;
lock_mutex_enter();
if (!lock_rec_get_first_on_page(block)) {
/* No locks exist on page, nothing to do */
lock_mutex_exit();
return;
}
/* Inherit all the locks on the page to the record and reset all
the locks on the page */
if (page_is_comp(page)) {
rec = page + PAGE_NEW_INFIMUM;
do {
heap_no = rec_get_heap_no_new(rec);
lock_rec_inherit_to_gap(heir_block, block,
heir_heap_no, heap_no);
lock_rec_reset_and_release_wait(block, heap_no);
rec = page + rec_get_next_offs(rec, TRUE);
} while (heap_no != PAGE_HEAP_NO_SUPREMUM);
} else {
rec = page + PAGE_OLD_INFIMUM;
do {
heap_no = rec_get_heap_no_old(rec);
lock_rec_inherit_to_gap(heir_block, block,
heir_heap_no, heap_no);
lock_rec_reset_and_release_wait(block, heap_no);
rec = page + rec_get_next_offs(rec, FALSE);
} while (heap_no != PAGE_HEAP_NO_SUPREMUM);
}
lock_rec_free_all_from_discard_page(block);
lock_mutex_exit();
}
/*************************************************************//**
Updates the lock table when a new user record is inserted. */
UNIV_INTERN
void
lock_update_insert(
/*===============*/
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec) /*!< in: the inserted record */
{
ulint receiver_heap_no;
ulint donator_heap_no;
ut_ad(block->frame == page_align(rec));
/* Inherit the gap-locking locks for rec, in gap mode, from the next
record */
if (page_rec_is_comp(rec)) {
receiver_heap_no = rec_get_heap_no_new(rec);
donator_heap_no = rec_get_heap_no_new(
page_rec_get_next_low(rec, TRUE));
} else {
receiver_heap_no = rec_get_heap_no_old(rec);
donator_heap_no = rec_get_heap_no_old(
page_rec_get_next_low(rec, FALSE));
}
lock_rec_inherit_to_gap_if_gap_lock(
block, receiver_heap_no, donator_heap_no);
}
/*************************************************************//**
Updates the lock table when a record is removed. */
UNIV_INTERN
void
lock_update_delete(
/*===============*/
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec) /*!< in: the record to be removed */
{
const page_t* page = block->frame;
ulint heap_no;
ulint next_heap_no;
ut_ad(page == page_align(rec));
if (page_is_comp(page)) {
heap_no = rec_get_heap_no_new(rec);
next_heap_no = rec_get_heap_no_new(page
+ rec_get_next_offs(rec,
TRUE));
} else {
heap_no = rec_get_heap_no_old(rec);
next_heap_no = rec_get_heap_no_old(page
+ rec_get_next_offs(rec,
FALSE));
}
lock_mutex_enter();
/* Let the next record inherit the locks from rec, in gap mode */
lock_rec_inherit_to_gap(block, block, next_heap_no, heap_no);
/* Reset the lock bits on rec and release waiting transactions */
lock_rec_reset_and_release_wait(block, heap_no);
lock_mutex_exit();
}
/*********************************************************************//**
Stores on the page infimum record the explicit locks of another record.
This function is used to store the lock state of a record when it is
updated and the size of the record changes in the update. The record
is moved in such an update, perhaps to another page. The infimum record
acts as a dummy carrier record, taking care of lock releases while the
actual record is being moved. */
UNIV_INTERN
void
lock_rec_store_on_page_infimum(
/*===========================*/
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec) /*!< in: record whose lock state
is stored on the infimum
record of the same page; lock
bits are reset on the
record */
{
ulint heap_no = page_rec_get_heap_no(rec);
ut_ad(block->frame == page_align(rec));
lock_mutex_enter();
lock_rec_move(block, block, PAGE_HEAP_NO_INFIMUM, heap_no);
lock_mutex_exit();
}
/*********************************************************************//**
Restores the state of explicit lock requests on a single record, where the
state was stored on the infimum of the page. */
UNIV_INTERN
void
lock_rec_restore_from_page_infimum(
/*===============================*/
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec, /*!< in: record whose lock state
is restored */
const buf_block_t* donator)/*!< in: page (rec is not
necessarily on this page)
whose infimum stored the lock
state; lock bits are reset on
the infimum */
{
ulint heap_no = page_rec_get_heap_no(rec);
lock_mutex_enter();
lock_rec_move(block, donator, heap_no, PAGE_HEAP_NO_INFIMUM);
lock_mutex_exit();
}
/*=========== DEADLOCK CHECKING ======================================*/
/*********************************************************************//**
rewind(3) the file used for storing the latest detected deadlock and
print a heading message to stderr if printing of all deadlocks to stderr
is enabled. */
UNIV_INLINE
void
lock_deadlock_start_print()
/*=======================*/
{
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
rewind(lock_latest_err_file);
ut_print_timestamp(lock_latest_err_file);
if (srv_print_all_deadlocks) {
ut_print_timestamp(stderr);
fprintf(stderr, "InnoDB: transactions deadlock detected, "
"dumping detailed information.\n");
ut_print_timestamp(stderr);
}
}
/*********************************************************************//**
Print a message to the deadlock file and possibly to stderr. */
UNIV_INLINE
void
lock_deadlock_fputs(
/*================*/
const char* msg) /*!< in: message to print */
{
if (!srv_read_only_mode) {
fputs(msg, lock_latest_err_file);
if (srv_print_all_deadlocks) {
fputs(msg, stderr);
}
}
}
/*********************************************************************//**
Print transaction data to the deadlock file and possibly to stderr. */
UNIV_INLINE
void
lock_deadlock_trx_print(
/*====================*/
const trx_t* trx, /*!< in: transaction */
ulint max_query_len) /*!< in: max query length to print,
or 0 to use the default max length */
{
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
ulint n_rec_locks = lock_number_of_rows_locked(&trx->lock);
ulint n_trx_locks = UT_LIST_GET_LEN(trx->lock.trx_locks);
ulint heap_size = mem_heap_get_size(trx->lock.lock_heap);
mutex_enter(&trx_sys->mutex);
trx_print_low(lock_latest_err_file, trx, max_query_len,
n_rec_locks, n_trx_locks, heap_size);
if (srv_print_all_deadlocks) {
trx_print_low(stderr, trx, max_query_len,
n_rec_locks, n_trx_locks, heap_size);
}
mutex_exit(&trx_sys->mutex);
}
/*********************************************************************//**
Print lock data to the deadlock file and possibly to stderr. */
UNIV_INLINE
void
lock_deadlock_lock_print(
/*=====================*/
const lock_t* lock) /*!< in: record or table type lock */
{
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
if (lock_get_type_low(lock) == LOCK_REC) {
lock_rec_print(lock_latest_err_file, lock);
if (srv_print_all_deadlocks) {
lock_rec_print(stderr, lock);
}
} else {
lock_table_print(lock_latest_err_file, lock);
if (srv_print_all_deadlocks) {
lock_table_print(stderr, lock);
}
}
}
/** Used in deadlock tracking. Protected by lock_sys->mutex. */
static ib_uint64_t lock_mark_counter = 0;
/** Check if the search is too deep. */
#define lock_deadlock_too_deep(c) \
(c->depth > LOCK_MAX_DEPTH_IN_DEADLOCK_CHECK \
|| c->cost > LOCK_MAX_N_STEPS_IN_DEADLOCK_CHECK)
/********************************************************************//**
Get the next lock in the queue that is owned by a transaction whose
sub-tree has not already been searched.
@return next lock or NULL if at end of queue */
static
const lock_t*
lock_get_next_lock(
/*===============*/
const lock_deadlock_ctx_t*
ctx, /*!< in: deadlock context */
const lock_t* lock, /*!< in: lock in the queue */
ulint heap_no)/*!< in: heap no if rec lock else
ULINT_UNDEFINED */
{
ut_ad(lock_mutex_own());
do {
if (lock_get_type_low(lock) == LOCK_REC) {
ut_ad(heap_no != ULINT_UNDEFINED);
lock = lock_rec_get_next_const(heap_no, lock);
} else {
ut_ad(heap_no == ULINT_UNDEFINED);
ut_ad(lock_get_type_low(lock) == LOCK_TABLE);
lock = UT_LIST_GET_NEXT(un_member.tab_lock.locks, lock);
}
} while (lock != NULL
&& lock->trx->lock.deadlock_mark > ctx->mark_start);
ut_ad(lock == NULL
|| lock_get_type_low(lock) == lock_get_type_low(ctx->wait_lock));
return(lock);
}
/********************************************************************//**
Get the first lock to search. The search starts from the current
wait_lock. What we are really interested in is an edge from the
current wait_lock's owning transaction to another transaction that has
a lock ahead in the queue. We skip locks where the owning transaction's
sub-tree has already been searched.
@return first lock or NULL */
static
const lock_t*
lock_get_first_lock(
/*================*/
const lock_deadlock_ctx_t*
ctx, /*!< in: deadlock context */
ulint* heap_no)/*!< out: heap no if rec lock,
else ULINT_UNDEFINED */
{
const lock_t* lock;
ut_ad(lock_mutex_own());
lock = ctx->wait_lock;
if (lock_get_type_low(lock) == LOCK_REC) {
*heap_no = lock_rec_find_set_bit(lock);
ut_ad(*heap_no != ULINT_UNDEFINED);
lock = lock_rec_get_first_on_page_addr(
lock->un_member.rec_lock.space,
lock->un_member.rec_lock.page_no);
/* Position on the first lock on the physical record. */
if (!lock_rec_get_nth_bit(lock, *heap_no)) {
lock = lock_rec_get_next_const(*heap_no, lock);
}
} else {
*heap_no = ULINT_UNDEFINED;
ut_ad(lock_get_type_low(lock) == LOCK_TABLE);
dict_table_t* table = lock->un_member.tab_lock.table;
lock = UT_LIST_GET_FIRST(table->locks);
}
ut_a(lock != NULL);
ut_a(lock != ctx->wait_lock ||
innodb_lock_schedule_algorithm == INNODB_LOCK_SCHEDULE_ALGORITHM_VATS);
ut_ad(lock_get_type_low(lock) == lock_get_type_low(ctx->wait_lock));
return(lock);
}
/********************************************************************//**
Notify that a deadlock has been detected and print the conflicting
transaction info. */
static
void
lock_deadlock_notify(
/*=================*/
const lock_deadlock_ctx_t* ctx, /*!< in: deadlock context */
const lock_t* lock) /*!< in: lock causing
deadlock */
{
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
lock_deadlock_start_print();
lock_deadlock_fputs("\n*** (1) TRANSACTION:\n");
lock_deadlock_trx_print(ctx->wait_lock->trx, 3000);
lock_deadlock_fputs("*** (1) WAITING FOR THIS LOCK TO BE GRANTED:\n");
lock_deadlock_lock_print(ctx->wait_lock);
lock_deadlock_fputs("*** (2) TRANSACTION:\n");
lock_deadlock_trx_print(lock->trx, 3000);
lock_deadlock_fputs("*** (2) HOLDS THE LOCK(S):\n");
lock_deadlock_lock_print(lock);
/* It is possible that the joining transaction was granted its
lock when we rolled back some other waiting transaction. */
if (ctx->start->lock.wait_lock != 0) {
lock_deadlock_fputs(
"*** (2) WAITING FOR THIS LOCK TO BE GRANTED:\n");
lock_deadlock_lock_print(ctx->start->lock.wait_lock);
}
#ifdef UNIV_DEBUG
if (lock_print_waits) {
fputs("Deadlock detected\n", stderr);
}
#endif /* UNIV_DEBUG */
}
/********************************************************************//**
Select the victim transaction that should be rolledback.
@return victim transaction */
static
const trx_t*
lock_deadlock_select_victim(
/*========================*/
const lock_deadlock_ctx_t* ctx) /*!< in: deadlock context */
{
ut_ad(lock_mutex_own());
ut_ad(ctx->start->lock.wait_lock != 0);
ut_ad(ctx->wait_lock->trx != ctx->start);
if (trx_weight_ge(ctx->wait_lock->trx, ctx->start)) {
/* The joining transaction is 'smaller',
choose it as the victim and roll it back. */
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(ctx->start->mysql_thd, TRUE)) {
return(ctx->wait_lock->trx);
}
else
#endif /* WITH_WSREP */
return(ctx->start);
}
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(ctx->wait_lock->trx->mysql_thd, TRUE)) {
return(ctx->start);
}
else
#endif /* WITH_WSREP */
return(ctx->wait_lock->trx);
}
/********************************************************************//**
Pop the deadlock search state from the stack.
@return stack slot instance that was on top of the stack. */
static
const lock_stack_t*
lock_deadlock_pop(
/*==============*/
lock_deadlock_ctx_t* ctx) /*!< in/out: context */
{
ut_ad(lock_mutex_own());
ut_ad(ctx->depth > 0);
return(&lock_stack[--ctx->depth]);
}
/********************************************************************//**
Push the deadlock search state onto the stack.
@return slot that was used in the stack */
static
lock_stack_t*
lock_deadlock_push(
/*===============*/
lock_deadlock_ctx_t* ctx, /*!< in/out: context */
const lock_t* lock, /*!< in: current lock */
ulint heap_no) /*!< in: heap number */
{
ut_ad(lock_mutex_own());
/* Save current search state. */
if (LOCK_STACK_SIZE > ctx->depth) {
lock_stack_t* stack;
stack = &lock_stack[ctx->depth++];
stack->lock = lock;
stack->heap_no = heap_no;
stack->wait_lock = ctx->wait_lock;
return(stack);
}
return(NULL);
}
/********************************************************************//**
Looks iteratively for a deadlock. Note: the joining transaction may
have been granted its lock by the deadlock checks.
@return 0 if no deadlock else the victim transaction id.*/
static
trx_id_t
lock_deadlock_search(
/*=================*/
lock_deadlock_ctx_t* ctx, /*!< in/out: deadlock context */
struct thd_wait_reports*waitee_ptr) /*!< in/out: list of waitees */
{
const lock_t* lock;
ulint heap_no;
ut_ad(lock_mutex_own());
ut_ad(!trx_mutex_own(ctx->start));
ut_ad(ctx->start != NULL);
ut_ad(ctx->wait_lock != NULL);
assert_trx_in_list(ctx->wait_lock->trx);
ut_ad(ctx->mark_start <= lock_mark_counter);
/* Look at the locks ahead of wait_lock in the lock queue. */
lock = lock_get_first_lock(ctx, &heap_no);
for (;;) {
/* We should never visit the same sub-tree more than once. */
ut_ad(lock == NULL
|| lock->trx->lock.deadlock_mark <= ctx->mark_start);
while (ctx->depth > 0 && lock == NULL) {
const lock_stack_t* stack;
/* Restore previous search state. */
stack = lock_deadlock_pop(ctx);
lock = stack->lock;
heap_no = stack->heap_no;
ctx->wait_lock = stack->wait_lock;
lock = lock_get_next_lock(ctx, lock, heap_no);
}
if (lock == NULL) {
break;
} else if (lock == ctx->wait_lock) {
/* We can mark this subtree as searched */
ut_ad(lock->trx->lock.deadlock_mark <= ctx->mark_start);
lock->trx->lock.deadlock_mark = ++lock_mark_counter;
/* We are not prepared for an overflow. This 64-bit
counter should never wrap around. At 10^9 increments
per second, it would take 10^3 years of uptime. */
ut_ad(lock_mark_counter > 0);
lock = NULL;
} else if (!lock_has_to_wait(ctx->wait_lock, lock)) {
/* No conflict, next lock */
lock = lock_get_next_lock(ctx, lock, heap_no);
} else if (lock->trx == ctx->start) {
/* Found a cycle. */
lock_deadlock_notify(ctx, lock);
return(lock_deadlock_select_victim(ctx)->id);
} else if (lock_deadlock_too_deep(ctx)) {
/* Search too deep to continue. */
ctx->too_deep = TRUE;
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(ctx->start->mysql_thd, TRUE)) {
return(ctx->wait_lock->trx->id);
}
else
#endif /* WITH_WSREP */
/* Select the joining transaction as the victim. */
return(ctx->start->id);
} else {
/* We do not need to report autoinc locks to the upper
layer. These locks are released before commit, so they
can not cause deadlocks with binlog-fixed commit
order. */
if (waitee_ptr &&
(lock_get_type_low(lock) != LOCK_TABLE ||
lock_get_mode(lock) != LOCK_AUTO_INC)) {
if (waitee_ptr->used ==
sizeof(waitee_ptr->waitees) /
sizeof(waitee_ptr->waitees[0])) {
waitee_ptr->next =
(struct thd_wait_reports *)
mem_alloc(sizeof(*waitee_ptr));
waitee_ptr = waitee_ptr->next;
if (!waitee_ptr) {
ctx->too_deep = TRUE;
return(ctx->start->id);
}
waitee_ptr->next = NULL;
waitee_ptr->used = 0;
}
waitee_ptr->waitees[waitee_ptr->used++] = lock->trx;
}
if (lock->trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
/* Another trx ahead has requested a lock in an
incompatible mode, and is itself waiting for a lock. */
++ctx->cost;
/* Save current search state. */
if (!lock_deadlock_push(ctx, lock, heap_no)) {
/* Unable to save current search state, stack
size not big enough. */
ctx->too_deep = TRUE;
#ifdef WITH_WSREP
if (wsrep_thd_is_BF(ctx->start->mysql_thd, TRUE))
return(lock->trx->id);
else
#endif /* WITH_WSREP */
return(ctx->start->id);
}
ctx->wait_lock = lock->trx->lock.wait_lock;
lock = lock_get_first_lock(ctx, &heap_no);
if (lock->trx->lock.deadlock_mark > ctx->mark_start) {
lock = lock_get_next_lock(ctx, lock, heap_no);
}
} else {
lock = lock_get_next_lock(ctx, lock, heap_no);
}
}
}
ut_a(lock == NULL && ctx->depth == 0);
/* No deadlock found. */
return(0);
}
/********************************************************************//**
Print info about transaction that was rolled back. */
static
void
lock_deadlock_joining_trx_print(
/*============================*/
const trx_t* trx, /*!< in: transaction rolled back */
const lock_t* lock) /*!< in: lock trx wants */
{
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
/* If the lock search exceeds the max step
or the max depth, the current trx will be
the victim. Print its information. */
lock_deadlock_start_print();
lock_deadlock_fputs(
"TOO DEEP OR LONG SEARCH IN THE LOCK TABLE"
" WAITS-FOR GRAPH, WE WILL ROLL BACK"
" FOLLOWING TRANSACTION \n\n"
"*** TRANSACTION:\n");
lock_deadlock_trx_print(trx, 3000);
lock_deadlock_fputs("*** WAITING FOR THIS LOCK TO BE GRANTED:\n");
lock_deadlock_lock_print(lock);
}
/********************************************************************//**
Rollback transaction selected as the victim. */
static
void
lock_deadlock_trx_rollback(
/*=======================*/
lock_deadlock_ctx_t* ctx) /*!< in: deadlock context */
{
trx_t* trx;
ut_ad(lock_mutex_own());
trx = ctx->wait_lock->trx;
lock_deadlock_fputs("*** WE ROLL BACK TRANSACTION (1)\n");
trx_mutex_enter(trx);
trx->lock.was_chosen_as_deadlock_victim = TRUE;
lock_cancel_waiting_and_release(trx->lock.wait_lock);
trx_mutex_exit(trx);
}
static
void
lock_report_waiters_to_mysql(
/*=======================*/
struct thd_wait_reports* waitee_buf_ptr, /*!< in: set of trxs */
THD* mysql_thd, /*!< in: THD */
trx_id_t victim_trx_id) /*!< in: Trx selected
as deadlock victim, if
any */
{
struct thd_wait_reports* p;
struct thd_wait_reports* q;
ulint i;
p = waitee_buf_ptr;
while (p) {
i = 0;
while (i < p->used) {
trx_t *w_trx = p->waitees[i];
/* There is no need to report waits to a trx already
selected as a victim. */
if (w_trx->id != victim_trx_id) {
/* If thd_report_wait_for() decides to kill the
transaction, then we will get a call back into
innobase_kill_query. We mark this by setting
current_lock_mutex_owner, so we can avoid trying
to recursively take lock_sys->mutex. */
w_trx->abort_type = TRX_REPLICATION_ABORT;
thd_report_wait_for(mysql_thd, w_trx->mysql_thd);
w_trx->abort_type = TRX_SERVER_ABORT;
}
++i;
}
q = p->next;
if (p != waitee_buf_ptr) {
mem_free(p);
}
p = q;
}
}
/********************************************************************//**
Checks if a joining lock request results in a deadlock. If a deadlock is
found this function will resolve the dadlock by choosing a victim transaction
and rolling it back. It will attempt to resolve all deadlocks. The returned
transaction id will be the joining transaction id or 0 if some other
transaction was chosen as a victim and rolled back or no deadlock found.
@return id of transaction chosen as victim or 0 */
static
trx_id_t
lock_deadlock_check_and_resolve(
/*============================*/
const lock_t* lock, /*!< in: lock the transaction is requesting */
const trx_t* trx) /*!< in: transaction */
{
trx_id_t victim_trx_id;
struct thd_wait_reports waitee_buf;
struct thd_wait_reports*waitee_buf_ptr;
THD* start_mysql_thd;
ut_ad(trx != NULL);
ut_ad(lock != NULL);
ut_ad(lock_mutex_own());
assert_trx_in_list(trx);
start_mysql_thd = trx->mysql_thd;
if (start_mysql_thd && thd_need_wait_for(start_mysql_thd)) {
waitee_buf_ptr = &waitee_buf;
} else {
waitee_buf_ptr = NULL;
}
/* Try and resolve as many deadlocks as possible. */
do {
lock_deadlock_ctx_t ctx;
/* Reset the context. */
ctx.cost = 0;
ctx.depth = 0;
ctx.start = trx;
ctx.too_deep = FALSE;
ctx.wait_lock = lock;
ctx.mark_start = lock_mark_counter;
if (waitee_buf_ptr) {
waitee_buf_ptr->next = NULL;
waitee_buf_ptr->used = 0;
}
victim_trx_id = lock_deadlock_search(&ctx, waitee_buf_ptr);
/* Report waits to upper layer, as needed. */
if (waitee_buf_ptr) {
lock_report_waiters_to_mysql(waitee_buf_ptr,
start_mysql_thd,
victim_trx_id);
}
/* Search too deep, we rollback the joining transaction. */
if (ctx.too_deep) {
ut_a(trx == ctx.start);
ut_a(victim_trx_id == trx->id);
#ifdef WITH_WSREP
if (!wsrep_thd_is_BF(ctx.start->mysql_thd, TRUE))
{
#endif /* WITH_WSREP */
if (!srv_read_only_mode) {
lock_deadlock_joining_trx_print(trx, lock);
}
#ifdef WITH_WSREP
} else {
/* BF processor */;
}
#endif /* WITH_WSREP */
} else if (victim_trx_id != 0 && victim_trx_id != trx->id) {
ut_ad(victim_trx_id == ctx.wait_lock->trx->id);
lock_deadlock_trx_rollback(&ctx);
lock_deadlock_found = TRUE;
MONITOR_INC(MONITOR_DEADLOCK);
srv_stats.lock_deadlock_count.inc();
}
} while (victim_trx_id != 0 && victim_trx_id != trx->id);
/* If the joining transaction was selected as the victim. */
if (victim_trx_id != 0) {
ut_a(victim_trx_id == trx->id);
MONITOR_INC(MONITOR_DEADLOCK);
srv_stats.lock_deadlock_count.inc();
lock_deadlock_fputs("*** WE ROLL BACK TRANSACTION (2)\n");
lock_deadlock_found = TRUE;
}
return(victim_trx_id);
}
/*========================= TABLE LOCKS ==============================*/
/*********************************************************************//**
Creates a table lock object and adds it as the last in the lock queue
of the table. Does NOT check for deadlocks or lock compatibility.
@return own: new lock object */
UNIV_INLINE
lock_t*
lock_table_create(
/*==============*/
#ifdef WITH_WSREP
lock_t* c_lock, /*!< in: conflicting lock */
#endif
dict_table_t* table, /*!< in/out: database table
in dictionary cache */
ulint type_mode,/*!< in: lock mode possibly ORed with
LOCK_WAIT */
trx_t* trx) /*!< in: trx */
{
lock_t* lock;
ut_ad(table && trx);
ut_ad(lock_mutex_own());
ut_ad(trx_mutex_own(trx));
/* Non-locking autocommit read-only transactions should not set
any locks. */
assert_trx_in_list(trx);
if ((type_mode & LOCK_MODE_MASK) == LOCK_AUTO_INC) {
++table->n_waiting_or_granted_auto_inc_locks;
}
/* For AUTOINC locking we reuse the lock instance only if
there is no wait involved else we allocate the waiting lock
from the transaction lock heap. */
if (type_mode == LOCK_AUTO_INC) {
lock = table->autoinc_lock;
table->autoinc_trx = trx;
ib_vector_push(trx->autoinc_locks, &lock);
} else {
lock = static_cast<lock_t*>(
mem_heap_alloc(trx->lock.lock_heap, sizeof(*lock)));
}
lock->type_mode = type_mode | LOCK_TABLE;
lock->trx = trx;
lock->requested_time = ut_time();
lock->wait_time = 0;
lock->un_member.tab_lock.table = table;
ut_ad(table->n_ref_count > 0 || !table->can_be_evicted);
UT_LIST_ADD_LAST(trx_locks, trx->lock.trx_locks, lock);
#ifdef WITH_WSREP
if (wsrep_thd_is_wsrep(trx->mysql_thd)) {
if (c_lock && wsrep_thd_is_BF(trx->mysql_thd, FALSE)) {
UT_LIST_INSERT_AFTER(
un_member.tab_lock.locks, table->locks, c_lock, lock);
} else {
UT_LIST_ADD_LAST(un_member.tab_lock.locks, table->locks, lock);
}
if (c_lock) {
trx_mutex_enter(c_lock->trx);
}
if (c_lock && c_lock->trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
c_lock->trx->lock.was_chosen_as_deadlock_victim = TRUE;
if (wsrep_debug) {
wsrep_print_wait_locks(c_lock);
wsrep_print_wait_locks(c_lock->trx->lock.wait_lock);
}
/* have to release trx mutex for the duration of
victim lock release. This will eventually call
lock_grant, which wants to grant trx mutex again
*/
/* caller has trx_mutex, have to release for lock cancel */
trx_mutex_exit(trx);
lock_cancel_waiting_and_release(c_lock->trx->lock.wait_lock);
trx_mutex_enter(trx);
/* trx might not wait for c_lock, but some other lock
does not matter if wait_lock was released above
*/
if (c_lock->trx->lock.wait_lock == c_lock) {
lock_reset_lock_and_trx_wait(lock);
}
if (wsrep_debug) {
fprintf(stderr, "WSREP: c_lock canceled %llu\n",
(ulonglong) c_lock->trx->id);
}
}
if (c_lock) {
trx_mutex_exit(c_lock->trx);
}
} else {
#endif /* WITH_WSREP */
UT_LIST_ADD_LAST(un_member.tab_lock.locks, table->locks, lock);
#ifdef WITH_WSREP
}
#endif /* WITH_WSREP */
if (UNIV_UNLIKELY(type_mode & LOCK_WAIT)) {
lock_set_lock_and_trx_wait(lock, trx);
}
ib_vector_push(lock->trx->lock.table_locks, &lock);
MONITOR_INC(MONITOR_TABLELOCK_CREATED);
MONITOR_INC(MONITOR_NUM_TABLELOCK);
return(lock);
}
/*************************************************************//**
Pops autoinc lock requests from the transaction's autoinc_locks. We
handle the case where there are gaps in the array and they need to
be popped off the stack. */
UNIV_INLINE
void
lock_table_pop_autoinc_locks(
/*=========================*/
trx_t* trx) /*!< in/out: transaction that owns the AUTOINC locks */
{
ut_ad(lock_mutex_own());
ut_ad(!ib_vector_is_empty(trx->autoinc_locks));
/* Skip any gaps, gaps are NULL lock entries in the
trx->autoinc_locks vector. */
do {
ib_vector_pop(trx->autoinc_locks);
if (ib_vector_is_empty(trx->autoinc_locks)) {
return;
}
} while (*(lock_t**) ib_vector_get_last(trx->autoinc_locks) == NULL);
}
/*************************************************************//**
Removes an autoinc lock request from the transaction's autoinc_locks. */
UNIV_INLINE
void
lock_table_remove_autoinc_lock(
/*===========================*/
lock_t* lock, /*!< in: table lock */
trx_t* trx) /*!< in/out: transaction that owns the lock */
{
lock_t* autoinc_lock;
lint i = ib_vector_size(trx->autoinc_locks) - 1;
ut_ad(lock_mutex_own());
ut_ad(lock_get_mode(lock) == LOCK_AUTO_INC);
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
ut_ad(!ib_vector_is_empty(trx->autoinc_locks));
/* With stored functions and procedures the user may drop
a table within the same "statement". This special case has
to be handled by deleting only those AUTOINC locks that were
held by the table being dropped. */
autoinc_lock = *static_cast<lock_t**>(
ib_vector_get(trx->autoinc_locks, i));
/* This is the default fast case. */
if (autoinc_lock == lock) {
lock_table_pop_autoinc_locks(trx);
} else {
/* The last element should never be NULL */
ut_a(autoinc_lock != NULL);
/* Handle freeing the locks from within the stack. */
while (--i >= 0) {
autoinc_lock = *static_cast<lock_t**>(
ib_vector_get(trx->autoinc_locks, i));
if (UNIV_LIKELY(autoinc_lock == lock)) {
void* null_var = NULL;
ib_vector_set(trx->autoinc_locks, i, &null_var);
return;
}
}
/* Must find the autoinc lock. */
ut_error;
}
}
/*************************************************************//**
Removes a table lock request from the queue and the trx list of locks;
this is a low-level function which does NOT check if waiting requests
can now be granted. */
UNIV_INLINE
void
lock_table_remove_low(
/*==================*/
lock_t* lock) /*!< in/out: table lock */
{
trx_t* trx;
dict_table_t* table;
ut_ad(lock_mutex_own());
trx = lock->trx;
table = lock->un_member.tab_lock.table;
/* Remove the table from the transaction's AUTOINC vector, if
the lock that is being released is an AUTOINC lock. */
if (lock_get_mode(lock) == LOCK_AUTO_INC) {
/* The table's AUTOINC lock can get transferred to
another transaction before we get here. */
if (table->autoinc_trx == trx) {
table->autoinc_trx = NULL;
}
/* The locks must be freed in the reverse order from
the one in which they were acquired. This is to avoid
traversing the AUTOINC lock vector unnecessarily.
We only store locks that were granted in the
trx->autoinc_locks vector (see lock_table_create()
and lock_grant()). Therefore it can be empty and we
need to check for that. */
if (!lock_get_wait(lock)
&& !ib_vector_is_empty(trx->autoinc_locks)) {
lock_table_remove_autoinc_lock(lock, trx);
}
ut_a(table->n_waiting_or_granted_auto_inc_locks > 0);
table->n_waiting_or_granted_auto_inc_locks--;
}
UT_LIST_REMOVE(trx_locks, trx->lock.trx_locks, lock);
UT_LIST_REMOVE(un_member.tab_lock.locks, table->locks, lock);
MONITOR_INC(MONITOR_TABLELOCK_REMOVED);
MONITOR_DEC(MONITOR_NUM_TABLELOCK);
}
/*********************************************************************//**
Enqueues a waiting request for a table lock which cannot be granted
immediately. Checks for deadlocks.
@return DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED, or
DB_SUCCESS; DB_SUCCESS means that there was a deadlock, but another
transaction was chosen as a victim, and we got the lock immediately:
no need to wait then */
static
dberr_t
lock_table_enqueue_waiting(
/*=======================*/
#ifdef WITH_WSREP
lock_t* c_lock, /*!< in: conflicting lock */
#endif
ulint mode, /*!< in: lock mode this transaction is
requesting */
dict_table_t* table, /*!< in/out: table */
que_thr_t* thr) /*!< in: query thread */
{
trx_t* trx;
lock_t* lock;
trx_id_t victim_trx_id;
ulint sec;
ulint ms;
ut_ad(lock_mutex_own());
ut_ad(!srv_read_only_mode);
trx = thr_get_trx(thr);
ut_ad(trx_mutex_own(trx));
/* Test if there already is some other reason to suspend thread:
we do not enqueue a lock request if the query thread should be
stopped anyway */
if (que_thr_stop(thr)) {
ut_error;
return(DB_QUE_THR_SUSPENDED);
}
switch (trx_get_dict_operation(trx)) {
case TRX_DICT_OP_NONE:
break;
case TRX_DICT_OP_TABLE:
case TRX_DICT_OP_INDEX:
ut_print_timestamp(stderr);
fputs(" InnoDB: Error: a table lock wait happens"
" in a dictionary operation!\n"
"InnoDB: Table name ", stderr);
ut_print_name(stderr, trx, TRUE, table->name);
fputs(".\n"
"InnoDB: Submit a detailed bug report"
" to http://bugs.mysql.com\n",
stderr);
ut_ad(0);
}
/* Enqueue the lock request that will wait to be granted */
#ifdef WITH_WSREP
if (trx->lock.was_chosen_as_deadlock_victim) {
return(DB_DEADLOCK);
}
lock = lock_table_create(c_lock, table, mode | LOCK_WAIT, trx);
#else
lock = lock_table_create(table, mode | LOCK_WAIT, trx);
#endif /* WITH_WSREP */
/* Release the mutex to obey the latching order.
This is safe, because lock_deadlock_check_and_resolve()
is invoked when a lock wait is enqueued for the currently
running transaction. Because trx is a running transaction
(it is not currently suspended because of a lock wait),
its state can only be changed by this thread, which is
currently associated with the transaction. */
trx_mutex_exit(trx);
victim_trx_id = lock_deadlock_check_and_resolve(lock, trx);
trx_mutex_enter(trx);
if (victim_trx_id != 0) {
ut_ad(victim_trx_id == trx->id);
/* The order here is important, we don't want to
lose the state of the lock before calling remove. */
lock_table_remove_low(lock);
lock_reset_lock_and_trx_wait(lock);
return(DB_DEADLOCK);
} else if (trx->lock.wait_lock == NULL) {
/* Deadlock resolution chose another transaction as a victim,
and we accidentally got our lock granted! */
return(DB_SUCCESS);
}
trx->lock.que_state = TRX_QUE_LOCK_WAIT;
trx->lock.wait_started = ut_time();
trx->lock.was_chosen_as_deadlock_victim = FALSE;
trx->n_table_lock_waits++;
if (UNIV_UNLIKELY(trx->take_stats)) {
ut_usectime(&sec, &ms);
trx->lock_que_wait_ustarted = (ib_uint64_t)sec * 1000000 + ms;
}
ut_a(que_thr_stop(thr));
MONITOR_INC(MONITOR_TABLELOCK_WAIT);
return(DB_LOCK_WAIT);
}
/*********************************************************************//**
Checks if other transactions have an incompatible mode lock request in
the lock queue.
@return lock or NULL */
UNIV_INLINE
const lock_t*
lock_table_other_has_incompatible(
/*==============================*/
const trx_t* trx, /*!< in: transaction, or NULL if all
transactions should be included */
ulint wait, /*!< in: LOCK_WAIT if also
waiting locks are taken into
account, or 0 if not */
const dict_table_t* table, /*!< in: table */
enum lock_mode mode) /*!< in: lock mode */
{
const lock_t* lock;
ut_ad(lock_mutex_own());
for (lock = UT_LIST_GET_LAST(table->locks);
lock != NULL;
lock = UT_LIST_GET_PREV(un_member.tab_lock.locks, lock)) {
if (lock->trx != trx
&& !lock_mode_compatible(lock_get_mode(lock), mode)
&& (wait || !lock_get_wait(lock))) {
#ifdef WITH_WSREP
if(wsrep_thd_is_wsrep(trx->mysql_thd)) {
if (wsrep_debug) {
fprintf(stderr, "WSREP: trx "
TRX_ID_FMT
" table lock abort\n",
trx->id);
}
trx_mutex_enter(lock->trx);
wsrep_kill_victim((trx_t *)trx, (lock_t *)lock);
trx_mutex_exit(lock->trx);
}
#endif
return(lock);
}
}
return(NULL);
}
/*********************************************************************//**
Locks the specified database table in the mode given. If the lock cannot
be granted immediately, the query thread is put to wait.
@return DB_SUCCESS, DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_table(
/*=======*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG bit is set,
does nothing */
dict_table_t* table, /*!< in/out: database table
in dictionary cache */
enum lock_mode mode, /*!< in: lock mode */
que_thr_t* thr) /*!< in: query thread */
{
#ifdef WITH_WSREP
lock_t *c_lock = NULL;
#endif
trx_t* trx;
dberr_t err;
const lock_t* wait_for;
ut_ad(table != NULL);
ut_ad(thr != NULL);
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
ut_a(flags == 0);
trx = thr_get_trx(thr);
if (UNIV_UNLIKELY(trx->fake_changes && mode == LOCK_IX)) {
mode = LOCK_IS;
}
/* Look for equal or stronger locks the same trx already
has on the table. No need to acquire the lock mutex here
because only this transacton can add/access table locks
to/from trx_t::table_locks. */
if (lock_table_has(trx, table, mode)) {
return(DB_SUCCESS);
}
lock_mutex_enter();
DBUG_EXECUTE_IF("fatal-semaphore-timeout",
{ os_thread_sleep(3600000000); });
/* We have to check if the new lock is compatible with any locks
other transactions have in the table lock queue. */
#ifdef WITH_WSREP
wait_for = lock_table_other_has_incompatible(
trx, LOCK_WAIT, table, mode);
#else
wait_for = lock_table_other_has_incompatible(
trx, LOCK_WAIT, table, mode);
#endif
trx_mutex_enter(trx);
/* Another trx has a request on the table in an incompatible
mode: this trx may have to wait */
if (wait_for != NULL) {
#ifdef WITH_WSREP
err = lock_table_enqueue_waiting((ib_lock_t*)wait_for, mode | flags, table, thr);
#else
err = lock_table_enqueue_waiting(mode | flags, table, thr);
#endif
} else {
#ifdef WITH_WSREP
lock_table_create(c_lock, table, mode | flags, trx);
#else
lock_table_create(table, mode | flags, trx);
#endif
ut_a(!flags || mode == LOCK_S || mode == LOCK_X);
err = DB_SUCCESS;
}
lock_mutex_exit();
trx_mutex_exit(trx);
return(err);
}
/*********************************************************************//**
Creates a table IX lock object for a resurrected transaction. */
UNIV_INTERN
void
lock_table_ix_resurrect(
/*====================*/
dict_table_t* table, /*!< in/out: table */
trx_t* trx) /*!< in/out: transaction */
{
ut_ad(trx->is_recovered);
if (lock_table_has(trx, table, LOCK_IX)) {
return;
}
lock_mutex_enter();
/* We have to check if the new lock is compatible with any locks
other transactions have in the table lock queue. */
ut_ad(!lock_table_other_has_incompatible(
trx, LOCK_WAIT, table, LOCK_IX));
trx_mutex_enter(trx);
#ifdef WITH_WSREP
lock_table_create(NULL, table, LOCK_IX, trx);
#else
lock_table_create(table, LOCK_IX, trx);
#endif
lock_mutex_exit();
trx_mutex_exit(trx);
}
/*********************************************************************//**
Checks if a waiting table lock request still has to wait in a queue.
@return TRUE if still has to wait */
static
ibool
lock_table_has_to_wait_in_queue(
/*============================*/
const lock_t* wait_lock) /*!< in: waiting table lock */
{
const dict_table_t* table;
const lock_t* lock;
ut_ad(lock_mutex_own());
ut_ad(lock_get_wait(wait_lock));
table = wait_lock->un_member.tab_lock.table;
for (lock = UT_LIST_GET_FIRST(table->locks);
lock != wait_lock;
lock = UT_LIST_GET_NEXT(un_member.tab_lock.locks, lock)) {
if (lock_has_to_wait(wait_lock, lock)) {
return(TRUE);
}
}
return(FALSE);
}
/*************************************************************//**
Removes a table lock request, waiting or granted, from the queue and grants
locks to other transactions in the queue, if they now are entitled to a
lock. */
static
void
lock_table_dequeue(
/*===============*/
lock_t* in_lock)/*!< in/out: table lock object; transactions waiting
behind will get their lock requests granted, if
they are now qualified to it */
{
lock_t* lock;
ut_ad(lock_mutex_own());
ut_a(lock_get_type_low(in_lock) == LOCK_TABLE);
lock = UT_LIST_GET_NEXT(un_member.tab_lock.locks, in_lock);
lock_table_remove_low(in_lock);
/* Check if waiting locks in the queue can now be granted: grant
locks if there are no conflicting locks ahead. */
for (/* No op */;
lock != NULL;
lock = UT_LIST_GET_NEXT(un_member.tab_lock.locks, lock)) {
if (lock_get_wait(lock)
&& !lock_table_has_to_wait_in_queue(lock)) {
/* Grant the lock */
ut_ad(in_lock->trx != lock->trx);
lock_grant(lock, false);
}
}
}
/*=========================== LOCK RELEASE ==============================*/
static
void
lock_grant_and_move_on_rec(
lock_t* first_lock,
ulint heap_no)
{
lock_t* lock;
lock_t* previous;
ulint space;
ulint page_no;
ulint rec_fold;
space = first_lock->un_member.rec_lock.space;
page_no = first_lock->un_member.rec_lock.page_no;
rec_fold = lock_rec_fold(space, page_no);
previous = (lock_t *) hash_get_nth_cell(lock_sys->rec_hash,
hash_calc_hash(rec_fold, lock_sys->rec_hash))->node;
if (previous == NULL) {
return;
}
if (previous == first_lock) {
lock = previous;
} else {
while (previous->hash &&
previous->hash != first_lock) {
previous = (lock_t *) previous->hash;
}
lock = (lock_t *) previous->hash;
}
/* Grant locks if there are no conflicting locks ahead.
Move granted locks to the head of the list. */
for (;lock != NULL;) {
/* If the lock is a wait lock on this page, and it does not need to wait. */
if (lock->un_member.rec_lock.space == space
&& lock->un_member.rec_lock.page_no == page_no
&& lock_rec_get_nth_bit(lock, heap_no)
&& lock_get_wait(lock)
&& !lock_rec_has_to_wait_in_queue(lock)) {
lock_grant(lock, false);
if (previous != NULL) {
/* Move the lock to the head of the list. */
HASH_GET_NEXT(hash, previous) = HASH_GET_NEXT(hash, lock);
lock_rec_insert_to_head(lock, rec_fold);
} else {
/* Already at the head of the list. */
previous = lock;
}
/* Move on to the next lock. */
lock = static_cast<lock_t *>(HASH_GET_NEXT(hash, previous));
} else {
previous = lock;
lock = static_cast<lock_t *>(HASH_GET_NEXT(hash, lock));
}
}
}
/*************************************************************//**
Removes a granted record lock of a transaction from the queue and grants
locks to other transactions waiting in the queue if they now are entitled
to a lock. */
UNIV_INTERN
void
lock_rec_unlock(
/*============*/
trx_t* trx, /*!< in/out: transaction that has
set a record lock */
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec, /*!< in: record */
enum lock_mode lock_mode)/*!< in: LOCK_S or LOCK_X */
{
lock_t* first_lock;
lock_t* lock;
ulint heap_no;
const char* stmt;
size_t stmt_len;
ut_ad(trx);
ut_ad(rec);
ut_ad(block->frame == page_align(rec));
ut_ad(!trx->lock.wait_lock);
ut_ad(trx_state_eq(trx, TRX_STATE_ACTIVE));
heap_no = page_rec_get_heap_no(rec);
lock_mutex_enter();
trx_mutex_enter(trx);
first_lock = lock_rec_get_first(block, heap_no);
/* Find the last lock with the same lock_mode and transaction
on the record. */
for (lock = first_lock; lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (lock->trx == trx && lock_get_mode(lock) == lock_mode) {
goto released;
}
}
lock_mutex_exit();
trx_mutex_exit(trx);
stmt = innobase_get_stmt(trx->mysql_thd, &stmt_len);
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Error: unlock row could not"
" find a %lu mode lock on the record\n",
(ulong) lock_mode);
ut_print_timestamp(stderr);
fprintf(stderr, " InnoDB: current statement: %.*s\n",
(int) stmt_len, stmt);
return;
released:
ut_a(!lock_get_wait(lock));
lock_rec_reset_nth_bit(lock, heap_no);
if (innodb_lock_schedule_algorithm
== INNODB_LOCK_SCHEDULE_ALGORITHM_FCFS ||
thd_is_replication_slave_thread(lock->trx->mysql_thd)) {
/* Check if we can now grant waiting lock requests */
for (lock = first_lock; lock != NULL;
lock = lock_rec_get_next(heap_no, lock)) {
if (lock_get_wait(lock)
&& !lock_rec_has_to_wait_in_queue(lock)) {
/* Grant the lock */
ut_ad(trx != lock->trx);
lock_grant(lock, false);
}
}
} else {
lock_grant_and_move_on_rec(first_lock, heap_no);
}
lock_mutex_exit();
trx_mutex_exit(trx);
}
/*********************************************************************//**
Releases transaction locks, and releases possible other transactions waiting
because of these locks. */
static
void
lock_release(
/*=========*/
trx_t* trx) /*!< in/out: transaction */
{
lock_t* lock;
ulint count = 0;
trx_id_t max_trx_id;
ut_ad(lock_mutex_own());
ut_ad(!trx_mutex_own(trx));
max_trx_id = trx_sys_get_max_trx_id();
for (lock = UT_LIST_GET_LAST(trx->lock.trx_locks);
lock != NULL;
lock = UT_LIST_GET_LAST(trx->lock.trx_locks)) {
if (lock_get_type_low(lock) == LOCK_REC) {
#ifdef UNIV_DEBUG
/* Check if the transcation locked a record
in a system table in X mode. It should have set
the dict_op code correctly if it did. */
if (lock->index->table->id < DICT_HDR_FIRST_ID
&& lock_get_mode(lock) == LOCK_X) {
ut_ad(lock_get_mode(lock) != LOCK_IX);
ut_ad(trx->dict_operation != TRX_DICT_OP_NONE);
}
#endif /* UNIV_DEBUG */
lock_rec_dequeue_from_page(lock);
} else {
dict_table_t* table;
table = lock->un_member.tab_lock.table;
#ifdef UNIV_DEBUG
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
/* Check if the transcation locked a system table
in IX mode. It should have set the dict_op code
correctly if it did. */
if (table->id < DICT_HDR_FIRST_ID
&& (lock_get_mode(lock) == LOCK_X
|| lock_get_mode(lock) == LOCK_IX)) {
ut_ad(trx->dict_operation != TRX_DICT_OP_NONE);
}
#endif /* UNIV_DEBUG */
if (lock_get_mode(lock) != LOCK_IS
&& trx->undo_no != 0) {
/* The trx may have modified the table. We
block the use of the MySQL query cache for
all currently active transactions. */
table->query_cache_inv_trx_id = max_trx_id;
}
lock_table_dequeue(lock);
}
if (count == LOCK_RELEASE_INTERVAL) {
/* Release the mutex for a while, so that we
do not monopolize it */
lock_mutex_exit();
lock_mutex_enter();
count = 0;
}
++count;
}
/* We don't remove the locks one by one from the vector for
efficiency reasons. We simply reset it because we would have
released all the locks anyway. */
ib_vector_reset(trx->lock.table_locks);
ut_a(UT_LIST_GET_LEN(trx->lock.trx_locks) == 0);
ut_a(ib_vector_is_empty(trx->autoinc_locks));
ut_a(ib_vector_is_empty(trx->lock.table_locks));
mem_heap_empty(trx->lock.lock_heap);
}
/* True if a lock mode is S or X */
#define IS_LOCK_S_OR_X(lock) \
(lock_get_mode(lock) == LOCK_S \
|| lock_get_mode(lock) == LOCK_X)
/*********************************************************************//**
Removes table locks of the transaction on a table to be dropped. */
static
void
lock_trx_table_locks_remove(
/*========================*/
const lock_t* lock_to_remove) /*!< in: lock to remove */
{
lint i;
trx_t* trx = lock_to_remove->trx;
ut_ad(lock_mutex_own());
/* It is safe to read this because we are holding the lock mutex */
if (!trx->lock.cancel) {
trx_mutex_enter(trx);
} else {
ut_ad(trx_mutex_own(trx));
}
for (i = ib_vector_size(trx->lock.table_locks) - 1; i >= 0; --i) {
const lock_t* lock;
lock = *static_cast<lock_t**>(
ib_vector_get(trx->lock.table_locks, i));
if (lock == NULL) {
continue;
}
ut_a(trx == lock->trx);
ut_a(lock_get_type_low(lock) & LOCK_TABLE);
ut_a(lock->un_member.tab_lock.table != NULL);
if (lock == lock_to_remove) {
void* null_var = NULL;
ib_vector_set(trx->lock.table_locks, i, &null_var);
if (!trx->lock.cancel) {
trx_mutex_exit(trx);
}
return;
}
}
if (!trx->lock.cancel) {
trx_mutex_exit(trx);
}
/* Lock must exist in the vector. */
ut_error;
}
/*********************************************************************//**
Removes locks of a transaction on a table to be dropped.
If remove_also_table_sx_locks is TRUE then table-level S and X locks are
also removed in addition to other table-level and record-level locks.
No lock that is going to be removed is allowed to be a wait lock. */
static
void
lock_remove_all_on_table_for_trx(
/*=============================*/
dict_table_t* table, /*!< in: table to be dropped */
trx_t* trx, /*!< in: a transaction */
ibool remove_also_table_sx_locks)/*!< in: also removes
table S and X locks */
{
lock_t* lock;
lock_t* prev_lock;
ut_ad(lock_mutex_own());
for (lock = UT_LIST_GET_LAST(trx->lock.trx_locks);
lock != NULL;
lock = prev_lock) {
prev_lock = UT_LIST_GET_PREV(trx_locks, lock);
if (lock_get_type_low(lock) == LOCK_REC
&& lock->index->table == table) {
ut_a(!lock_get_wait(lock));
lock_rec_discard(lock);
} else if (lock_get_type_low(lock) & LOCK_TABLE
&& lock->un_member.tab_lock.table == table
&& (remove_also_table_sx_locks
|| !IS_LOCK_S_OR_X(lock))) {
ut_a(!lock_get_wait(lock));
lock_trx_table_locks_remove(lock);
lock_table_remove_low(lock);
}
}
}
/*******************************************************************//**
Remove any explicit record locks held by recovering transactions on
the table.
@return number of recovered transactions examined */
static
ulint
lock_remove_recovered_trx_record_locks(
/*===================================*/
dict_table_t* table) /*!< in: check if there are any locks
held on records in this table or on the
table itself */
{
trx_t* trx;
ulint n_recovered_trx = 0;
ut_a(table != NULL);
ut_ad(lock_mutex_own());
mutex_enter(&trx_sys->mutex);
for (trx = UT_LIST_GET_FIRST(trx_sys->rw_trx_list);
trx != NULL;
trx = UT_LIST_GET_NEXT(trx_list, trx)) {
lock_t* lock;
lock_t* next_lock;
assert_trx_in_rw_list(trx);
if (!trx->is_recovered) {
continue;
}
/* Because we are holding the lock_sys->mutex,
implicit locks cannot be converted to explicit ones
while we are scanning the explicit locks. */
for (lock = UT_LIST_GET_FIRST(trx->lock.trx_locks);
lock != NULL;
lock = next_lock) {
ut_a(lock->trx == trx);
/* Recovered transactions can't wait on a lock. */
ut_a(!lock_get_wait(lock));
next_lock = UT_LIST_GET_NEXT(trx_locks, lock);
switch (lock_get_type_low(lock)) {
default:
ut_error;
case LOCK_TABLE:
if (lock->un_member.tab_lock.table == table) {
lock_trx_table_locks_remove(lock);
lock_table_remove_low(lock);
}
break;
case LOCK_REC:
if (lock->index->table == table) {
lock_rec_discard(lock);
}
}
}
++n_recovered_trx;
}
mutex_exit(&trx_sys->mutex);
return(n_recovered_trx);
}
/*********************************************************************//**
Removes locks on a table to be dropped or truncated.
If remove_also_table_sx_locks is TRUE then table-level S and X locks are
also removed in addition to other table-level and record-level locks.
No lock, that is going to be removed, is allowed to be a wait lock. */
UNIV_INTERN
void
lock_remove_all_on_table(
/*=====================*/
dict_table_t* table, /*!< in: table to be dropped
or truncated */
ibool remove_also_table_sx_locks)/*!< in: also removes
table S and X locks */
{
lock_t* lock;
lock_mutex_enter();
for (lock = UT_LIST_GET_FIRST(table->locks);
lock != NULL;
/* No op */) {
lock_t* prev_lock;
prev_lock = UT_LIST_GET_PREV(un_member.tab_lock.locks, lock);
/* If we should remove all locks (remove_also_table_sx_locks
is TRUE), or if the lock is not table-level S or X lock,
then check we are not going to remove a wait lock. */
if (remove_also_table_sx_locks
|| !(lock_get_type(lock) == LOCK_TABLE
&& IS_LOCK_S_OR_X(lock))) {
ut_a(!lock_get_wait(lock));
}
lock_remove_all_on_table_for_trx(
table, lock->trx, remove_also_table_sx_locks);
if (prev_lock == NULL) {
if (lock == UT_LIST_GET_FIRST(table->locks)) {
/* lock was not removed, pick its successor */
lock = UT_LIST_GET_NEXT(
un_member.tab_lock.locks, lock);
} else {
/* lock was removed, pick the first one */
lock = UT_LIST_GET_FIRST(table->locks);
}
} else if (UT_LIST_GET_NEXT(un_member.tab_lock.locks,
prev_lock) != lock) {
/* If lock was removed by
lock_remove_all_on_table_for_trx() then pick the
successor of prev_lock ... */
lock = UT_LIST_GET_NEXT(
un_member.tab_lock.locks, prev_lock);
} else {
/* ... otherwise pick the successor of lock. */
lock = UT_LIST_GET_NEXT(
un_member.tab_lock.locks, lock);
}
}
/* Note: Recovered transactions don't have table level IX or IS locks
but can have implicit record locks that have been converted to explicit
record locks. Such record locks cannot be freed by traversing the
transaction lock list in dict_table_t (as above). */
if (!lock_sys->rollback_complete
&& lock_remove_recovered_trx_record_locks(table) == 0) {
lock_sys->rollback_complete = TRUE;
}
lock_mutex_exit();
}
/*===================== VALIDATION AND DEBUGGING ====================*/
/*********************************************************************//**
Prints info of a table lock. */
UNIV_INTERN
void
lock_table_print(
/*=============*/
FILE* file, /*!< in: file where to print */
const lock_t* lock) /*!< in: table type lock */
{
ut_ad(lock_mutex_own());
ut_a(lock_get_type_low(lock) == LOCK_TABLE);
fputs("TABLE LOCK table ", file);
ut_print_name(file, lock->trx, TRUE,
lock->un_member.tab_lock.table->name);
fprintf(file, " trx id " TRX_ID_FMT, lock->trx->id);
if (lock_get_mode(lock) == LOCK_S) {
fputs(" lock mode S", file);
} else if (lock_get_mode(lock) == LOCK_X) {
fputs(" lock mode X", file);
} else if (lock_get_mode(lock) == LOCK_IS) {
fputs(" lock mode IS", file);
} else if (lock_get_mode(lock) == LOCK_IX) {
fputs(" lock mode IX", file);
} else if (lock_get_mode(lock) == LOCK_AUTO_INC) {
fputs(" lock mode AUTO-INC", file);
} else {
fprintf(file, " unknown lock mode %lu",
(ulong) lock_get_mode(lock));
}
if (lock_get_wait(lock)) {
fputs(" waiting", file);
}
fprintf(file, " lock hold time %lu wait time before grant %lu ",
(ulint)difftime(ut_time(), lock->requested_time),
lock->wait_time);
putc('\n', file);
}
/*********************************************************************//**
Prints info of a record lock. */
UNIV_INTERN
void
lock_rec_print(
/*===========*/
FILE* file, /*!< in: file where to print */
const lock_t* lock) /*!< in: record type lock */
{
const buf_block_t* block;
ulint space;
ulint page_no;
ulint i;
mtr_t mtr;
mem_heap_t* heap = NULL;
ulint offsets_[REC_OFFS_NORMAL_SIZE];
ulint* offsets = offsets_;
rec_offs_init(offsets_);
ut_ad(lock_mutex_own());
ut_a(lock_get_type_low(lock) == LOCK_REC);
space = lock->un_member.rec_lock.space;
page_no = lock->un_member.rec_lock.page_no;
fprintf(file, "RECORD LOCKS space id %lu page no %lu n bits %lu ",
(ulong) space, (ulong) page_no,
(ulong) lock_rec_get_n_bits(lock));
dict_index_name_print(file, lock->trx, lock->index);
/* Print number of table locks */
fprintf(file, " trx table locks %lu total table locks %lu ",
ib_vector_size(lock->trx->lock.table_locks),
UT_LIST_GET_LEN(lock->index->table->locks));
fprintf(file, " trx id " TRX_ID_FMT, lock->trx->id);
if (lock_get_mode(lock) == LOCK_S) {
fputs(" lock mode S", file);
} else if (lock_get_mode(lock) == LOCK_X) {
fputs(" lock_mode X", file);
} else {
ut_error;
}
if (lock_rec_get_gap(lock)) {
fputs(" locks gap before rec", file);
}
if (lock_rec_get_rec_not_gap(lock)) {
fputs(" locks rec but not gap", file);
}
if (lock_rec_get_insert_intention(lock)) {
fputs(" insert intention", file);
}
if (lock_get_wait(lock)) {
fputs(" waiting", file);
}
mtr_start(&mtr);
fprintf(file, " lock hold time %lu wait time before grant %lu ",
(ulint)difftime(ut_time(), lock->requested_time),
lock->wait_time);
putc('\n', file);
if ( srv_show_verbose_locks ) {
block = buf_page_try_get(space, page_no, &mtr);
for (i = 0; i < lock_rec_get_n_bits(lock); ++i) {
if (!lock_rec_get_nth_bit(lock, i)) {
continue;
}
fprintf(file, "Record lock, heap no %lu", (ulong) i);
if (block) {
const rec_t* rec;
rec = page_find_rec_with_heap_no(
buf_block_get_frame(block), i);
offsets = rec_get_offsets(
rec, lock->index, offsets,
ULINT_UNDEFINED, &heap);
putc(' ', file);
rec_print_new(file, rec, offsets);
}
putc('\n', file);
}
}
mtr_commit(&mtr);
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
}
#ifdef UNIV_DEBUG
/* Print the number of lock structs from lock_print_info_summary() only
in non-production builds for performance reasons, see
http://bugs.mysql.com/36942 */
#define PRINT_NUM_OF_LOCK_STRUCTS
#endif /* UNIV_DEBUG */
#ifdef PRINT_NUM_OF_LOCK_STRUCTS
/*********************************************************************//**
Calculates the number of record lock structs in the record lock hash table.
@return number of record locks */
static
ulint
lock_get_n_rec_locks(void)
/*======================*/
{
ulint n_locks = 0;
ulint i;
ut_ad(lock_mutex_own());
for (i = 0; i < hash_get_n_cells(lock_sys->rec_hash); i++) {
const lock_t* lock;
for (lock = static_cast<const lock_t*>(
HASH_GET_FIRST(lock_sys->rec_hash, i));
lock != 0;
lock = static_cast<const lock_t*>(
HASH_GET_NEXT(hash, lock))) {
n_locks++;
}
}
return(n_locks);
}
#endif /* PRINT_NUM_OF_LOCK_STRUCTS */
/*********************************************************************//**
Prints info of locks for all transactions.
@return FALSE if not able to obtain lock mutex
and exits without printing info */
UNIV_INTERN
ibool
lock_print_info_summary(
/*====================*/
FILE* file, /*!< in: file where to print */
ibool nowait) /*!< in: whether to wait for the lock mutex */
{
/* if nowait is FALSE, wait on the lock mutex,
otherwise return immediately if fail to obtain the
mutex. */
if (!nowait) {
lock_mutex_enter();
} else if (lock_mutex_enter_nowait()) {
fputs("FAIL TO OBTAIN LOCK MUTEX, "
"SKIP LOCK INFO PRINTING\n", file);
return(FALSE);
}
if (lock_deadlock_found) {
fputs("------------------------\n"
"LATEST DETECTED DEADLOCK\n"
"------------------------\n", file);
if (!srv_read_only_mode) {
ut_copy_file(file, lock_latest_err_file);
}
}
fputs("------------\n"
"TRANSACTIONS\n"
"------------\n", file);
fprintf(file, "Trx id counter " TRX_ID_FMT "\n",
trx_sys_get_max_trx_id());
fprintf(file,
"Purge done for trx's n:o < " TRX_ID_FMT
" undo n:o < " TRX_ID_FMT " state: ",
purge_sys->iter.trx_no,
purge_sys->iter.undo_no);
/* Note: We are reading the state without the latch. One because it
will violate the latching order and two because we are merely querying
the state of the variable for display. */
switch (purge_sys->state){
case PURGE_STATE_INIT:
/* Should never be in this state while the system is running. */
ut_error;
case PURGE_STATE_EXIT:
fprintf(file, "exited");
break;
case PURGE_STATE_DISABLED:
fprintf(file, "disabled");
break;
case PURGE_STATE_RUN:
fprintf(file, "running");
/* Check if it is waiting for more data to arrive. */
if (!purge_sys->running) {
fprintf(file, " but idle");
}
break;
case PURGE_STATE_STOP:
fprintf(file, "stopped");
break;
}
fprintf(file, "\n");
fprintf(file,
"History list length %lu\n",
(ulong) trx_sys->rseg_history_len);
#ifdef PRINT_NUM_OF_LOCK_STRUCTS
fprintf(file,
"Total number of lock structs in row lock hash table %lu\n",
(ulong) lock_get_n_rec_locks());
#endif /* PRINT_NUM_OF_LOCK_STRUCTS */
return(TRUE);
}
/*********************************************************************//**
Prints info of locks for each transaction. This function assumes that the
caller holds the lock mutex and more importantly it will release the lock
mutex on behalf of the caller. (This should be fixed in the future). */
UNIV_INTERN
void
lock_print_info_all_transactions(
/*=============================*/
FILE* file) /*!< in: file where to print */
{
const lock_t* lock;
ibool load_page_first = TRUE;
ulint nth_trx = 0;
ulint nth_lock = 0;
ulint i;
mtr_t mtr;
const trx_t* trx;
trx_list_t* trx_list = &trx_sys->rw_trx_list;
fprintf(file, "LIST OF TRANSACTIONS FOR EACH SESSION:\n");
ut_ad(lock_mutex_own());
mutex_enter(&trx_sys->mutex);
/* First print info on non-active transactions */
/* NOTE: information of auto-commit non-locking read-only
transactions will be omitted here. The information will be
available from INFORMATION_SCHEMA.INNODB_TRX. */
for (trx = UT_LIST_GET_FIRST(trx_sys->mysql_trx_list);
trx != NULL;
trx = UT_LIST_GET_NEXT(mysql_trx_list, trx)) {
ut_ad(trx->in_mysql_trx_list);
/* See state transitions and locking rules in trx0trx.h */
if (trx_state_eq(trx, TRX_STATE_NOT_STARTED)) {
fputs("---", file);
trx_print_latched(file, trx, 600);
}
}
loop:
/* Since we temporarily release lock_sys->mutex and
trx_sys->mutex when reading a database page in below,
variable trx may be obsolete now and we must loop
through the trx list to get probably the same trx,
or some other trx. */
for (trx = UT_LIST_GET_FIRST(*trx_list), i = 0;
trx && (i < nth_trx);
trx = UT_LIST_GET_NEXT(trx_list, trx), i++) {
assert_trx_in_list(trx);
ut_ad(trx->read_only == (trx_list == &trx_sys->ro_trx_list));
}
ut_ad(trx == NULL
|| trx->read_only == (trx_list == &trx_sys->ro_trx_list));
if (trx == NULL) {
/* Check the read-only transaction list next. */
if (trx_list == &trx_sys->rw_trx_list) {
trx_list = &trx_sys->ro_trx_list;
nth_trx = 0;
nth_lock = 0;
goto loop;
}
lock_mutex_exit();
mutex_exit(&trx_sys->mutex);
ut_ad(lock_validate());
return;
}
assert_trx_in_list(trx);
if (nth_lock == 0) {
fputs("---", file);
trx_print_latched(file, trx, 600);
if (trx->read_view) {
fprintf(file,
"Trx read view will not see trx with"
" id >= " TRX_ID_FMT
", sees < " TRX_ID_FMT "\n",
trx->read_view->low_limit_id,
trx->read_view->up_limit_id);
}
/* Total trx lock waits and times */
fprintf(file, "Trx #rec lock waits %lu #table lock waits %lu\n",
trx->n_rec_lock_waits, trx->n_table_lock_waits);
fprintf(file, "Trx total rec lock wait time %lu SEC\n",
trx->total_rec_lock_wait_time);
fprintf(file, "Trx total table lock wait time %lu SEC\n",
trx->total_table_lock_wait_time);
if (trx->lock.que_state == TRX_QUE_LOCK_WAIT) {
fprintf(file,
"------- TRX HAS BEEN WAITING %lu SEC"
" FOR THIS LOCK TO BE GRANTED:\n",
(ulong) difftime(ut_time(),
trx->lock.wait_started));
if (lock_get_type_low(trx->lock.wait_lock) == LOCK_REC) {
lock_rec_print(file, trx->lock.wait_lock);
} else {
lock_table_print(file, trx->lock.wait_lock);
}
fputs("------------------\n", file);
}
}
if (!srv_print_innodb_lock_monitor || !srv_show_locks_held) {
nth_trx++;
goto loop;
}
i = 0;
/* Look at the note about the trx loop above why we loop here:
lock may be an obsolete pointer now. */
lock = UT_LIST_GET_FIRST(trx->lock.trx_locks);
while (lock && (i < nth_lock)) {
lock = UT_LIST_GET_NEXT(trx_locks, lock);
i++;
}
if (lock == NULL) {
nth_trx++;
nth_lock = 0;
goto loop;
}
if (lock_get_type_low(lock) == LOCK_REC) {
if (load_page_first) {
ulint space_id = lock->un_member.rec_lock.space;
/* Check if the space is exists or not. only
when the space is valid, try to get the page. */
fil_space_t* space = fil_space_acquire(space_id);
ulint page_no = lock->un_member.rec_lock.page_no;
if (!space) {
/* It is a single table tablespace and
the .ibd file is missing (TRUNCATE
TABLE probably stole the locks): just
print the lock without attempting to
load the page in the buffer pool. */
fprintf(file, "RECORD LOCKS on"
" non-existing space: " ULINTPF "\n",
space_id);
goto print_rec;
}
const ulint zip_size = fsp_flags_get_zip_size(space->flags);
lock_mutex_exit();
mutex_exit(&trx_sys->mutex);
if (srv_show_verbose_locks) {
DEBUG_SYNC_C("innodb_monitor_before_lock_page_read");
if (space) {
mtr_start(&mtr);
buf_page_get_gen(space_id, zip_size,
page_no, RW_NO_LATCH,
NULL,
BUF_GET_POSSIBLY_FREED,
__FILE__, __LINE__,
&mtr);
mtr_commit(&mtr);
}
}
fil_space_release(space);
load_page_first = FALSE;
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
goto loop;
}
print_rec:
lock_rec_print(file, lock);
} else {
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
lock_table_print(file, lock);
}
load_page_first = TRUE;
nth_lock++;
if (nth_lock >= srv_show_locks_held) {
fputs("TOO MANY LOCKS PRINTED FOR THIS TRX:"
" SUPPRESSING FURTHER PRINTS\n",
file);
nth_trx++;
nth_lock = 0;
}
goto loop;
}
#ifdef UNIV_DEBUG
/*********************************************************************//**
Find the the lock in the trx_t::trx_lock_t::table_locks vector.
@return TRUE if found */
static
ibool
lock_trx_table_locks_find(
/*======================*/
trx_t* trx, /*!< in: trx to validate */
const lock_t* find_lock) /*!< in: lock to find */
{
lint i;
ibool found = FALSE;
trx_mutex_enter(trx);
for (i = ib_vector_size(trx->lock.table_locks) - 1; i >= 0; --i) {
const lock_t* lock;
lock = *static_cast<const lock_t**>(
ib_vector_get(trx->lock.table_locks, i));
if (lock == NULL) {
continue;
} else if (lock == find_lock) {
/* Can't be duplicates. */
ut_a(!found);
found = TRUE;
}
ut_a(trx == lock->trx);
ut_a(lock_get_type_low(lock) & LOCK_TABLE);
ut_a(lock->un_member.tab_lock.table != NULL);
}
trx_mutex_exit(trx);
return(found);
}
/*********************************************************************//**
Validates the lock queue on a table.
@return TRUE if ok */
static
ibool
lock_table_queue_validate(
/*======================*/
const dict_table_t* table) /*!< in: table */
{
const lock_t* lock;
ut_ad(lock_mutex_own());
ut_ad(mutex_own(&trx_sys->mutex));
for (lock = UT_LIST_GET_FIRST(table->locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(un_member.tab_lock.locks, lock)) {
/* lock->trx->state cannot change from or to NOT_STARTED
while we are holding the trx_sys->mutex. It may change
from ACTIVE to PREPARED, but it may not change to
COMMITTED, because we are holding the lock_sys->mutex. */
ut_ad(trx_assert_started(lock->trx));
if (!lock_get_wait(lock)) {
ut_a(!lock_table_other_has_incompatible(
lock->trx, 0, table,
lock_get_mode(lock)));
} else {
ut_a(lock_table_has_to_wait_in_queue(lock));
}
ut_a(lock_trx_table_locks_find(lock->trx, lock));
}
return(TRUE);
}
/*********************************************************************//**
Validates the lock queue on a single record.
@return TRUE if ok */
static
ibool
lock_rec_queue_validate(
/*====================*/
ibool locked_lock_trx_sys,
/*!< in: if the caller holds
both the lock mutex and
trx_sys_t->lock. */
const buf_block_t* block, /*!< in: buffer block containing rec */
const rec_t* rec, /*!< in: record to look at */
const dict_index_t* index, /*!< in: index, or NULL if not known */
const ulint* offsets)/*!< in: rec_get_offsets(rec, index) */
{
const lock_t* lock;
ulint heap_no;
ut_a(rec);
ut_a(block->frame == page_align(rec));
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(!page_rec_is_comp(rec) == !rec_offs_comp(offsets));
ut_ad(lock_mutex_own() == locked_lock_trx_sys);
ut_ad(!index || dict_index_is_clust(index)
|| !dict_index_is_online_ddl(index));
heap_no = page_rec_get_heap_no(rec);
if (!locked_lock_trx_sys) {
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
}
if (!page_rec_is_user_rec(rec)) {
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next_const(heap_no, lock)) {
ut_a(trx_in_trx_list(lock->trx));
if (lock_get_wait(lock)) {
ut_a(lock_rec_has_to_wait_in_queue(lock));
}
if (index) {
ut_a(lock->index == index);
}
}
goto func_exit;
}
if (!index);
else if (dict_index_is_clust(index)) {
trx_id_t trx_id;
trx_id_t* trx_desc;
/* Unlike the non-debug code, this invariant can only succeed
if the check and assertion are covered by the lock mutex. */
trx_id = lock_clust_rec_some_has_impl(rec, index, offsets);
trx_desc = trx_find_descriptor(trx_sys->descriptors,
trx_sys->descr_n_used,
trx_id);
ut_ad(lock_mutex_own());
/* trx_id cannot be committed until lock_mutex_exit()
because lock_trx_release_locks() acquires lock_sys->mutex */
if (trx_desc != NULL
&& lock_rec_other_has_expl_req(LOCK_S, 0, LOCK_WAIT,
block, heap_no, trx_id)) {
ut_ad(trx_id == *trx_desc);
ut_a(lock_rec_has_expl(LOCK_X | LOCK_REC_NOT_GAP,
block, heap_no, trx_id));
}
}
for (lock = lock_rec_get_first(block, heap_no);
lock != NULL;
lock = lock_rec_get_next_const(heap_no, lock)) {
ut_a(trx_in_trx_list(lock->trx));
if (index) {
ut_a(lock->index == index);
}
if (!lock_rec_get_gap(lock) && !lock_get_wait(lock)) {
enum lock_mode mode;
if (lock_get_mode(lock) == LOCK_S) {
mode = LOCK_X;
} else {
mode = LOCK_S;
}
const lock_t* other_lock
= lock_rec_other_has_expl_req(
mode, 0, 0, block, heap_no,
lock->trx->id);
#ifdef WITH_WSREP
ut_a(!other_lock
|| wsrep_thd_is_BF(lock->trx->mysql_thd, FALSE)
|| wsrep_thd_is_BF(other_lock->trx->mysql_thd, FALSE));
#else
ut_a(!other_lock);
#endif /* WITH_WSREP */
} else if (lock_get_wait(lock) && !lock_rec_get_gap(lock)
&& innodb_lock_schedule_algorithm == INNODB_LOCK_SCHEDULE_ALGORITHM_FCFS) {
// If using VATS, it's possible that a wait lock is inserted to a place in the list
// such that it does not need to wait.
ut_a(lock_rec_has_to_wait_in_queue(lock));
}
}
ut_ad(innodb_lock_schedule_algorithm == INNODB_LOCK_SCHEDULE_ALGORITHM_FCFS ||
lock_queue_validate(lock));
func_exit:
if (!locked_lock_trx_sys) {
lock_mutex_exit();
mutex_exit(&trx_sys->mutex);
}
return(TRUE);
}
/*********************************************************************//**
Validates the record lock queues on a page.
@return TRUE if ok */
static
ibool
lock_rec_validate_page(
/*===================*/
const buf_block_t* block) /*!< in: buffer block */
{
const lock_t* lock;
const rec_t* rec;
ulint nth_lock = 0;
ulint nth_bit = 0;
ulint i;
mem_heap_t* heap = NULL;
ulint offsets_[REC_OFFS_NORMAL_SIZE];
ulint* offsets = offsets_;
rec_offs_init(offsets_);
ut_ad(!lock_mutex_own());
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
loop:
lock = lock_rec_get_first_on_page_addr(buf_block_get_space(block),
buf_block_get_page_no(block));
if (!lock) {
goto function_exit;
}
#if defined UNIV_DEBUG_FILE_ACCESSES || defined UNIV_DEBUG
ut_a(!block->page.file_page_was_freed);
#endif
for (i = 0; i < nth_lock; i++) {
lock = lock_rec_get_next_on_page_const(lock);
if (!lock) {
goto function_exit;
}
}
ut_a(trx_in_trx_list(lock->trx));
# ifdef UNIV_SYNC_DEBUG
/* Only validate the record queues when this thread is not
holding a space->latch. Deadlocks are possible due to
latching order violation when UNIV_DEBUG is defined while
UNIV_SYNC_DEBUG is not. */
if (!sync_thread_levels_contains(SYNC_FSP))
# endif /* UNIV_SYNC_DEBUG */
for (i = nth_bit; i < lock_rec_get_n_bits(lock); i++) {
if (i == 1 || lock_rec_get_nth_bit(lock, i)) {
rec = page_find_rec_with_heap_no(block->frame, i);
ut_a(rec);
offsets = rec_get_offsets(rec, lock->index, offsets,
ULINT_UNDEFINED, &heap);
#if 0
fprintf(stderr,
"Validating %u %u\n",
block->page.space, block->page.offset);
#endif
/* If this thread is holding the file space
latch (fil_space_t::latch), the following
check WILL break the latching order and may
cause a deadlock of threads. */
lock_rec_queue_validate(
TRUE, block, rec, lock->index, offsets);
nth_bit = i + 1;
goto loop;
}
}
nth_bit = 0;
nth_lock++;
goto loop;
function_exit:
lock_mutex_exit();
mutex_exit(&trx_sys->mutex);
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
return(TRUE);
}
/*********************************************************************//**
Validates the table locks.
@return TRUE if ok */
static
ibool
lock_validate_table_locks(
/*======================*/
const trx_list_t* trx_list) /*!< in: trx list */
{
const trx_t* trx;
ut_ad(lock_mutex_own());
ut_ad(mutex_own(&trx_sys->mutex));
ut_ad(trx_list == &trx_sys->rw_trx_list
|| trx_list == &trx_sys->ro_trx_list);
for (trx = UT_LIST_GET_FIRST(*trx_list);
trx != NULL;
trx = UT_LIST_GET_NEXT(trx_list, trx)) {
const lock_t* lock;
assert_trx_in_list(trx);
ut_ad(trx->read_only == (trx_list == &trx_sys->ro_trx_list));
for (lock = UT_LIST_GET_FIRST(trx->lock.trx_locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(trx_locks, lock)) {
if (lock_get_type_low(lock) & LOCK_TABLE) {
lock_table_queue_validate(
lock->un_member.tab_lock.table);
}
}
}
return(TRUE);
}
/*********************************************************************//**
Validate record locks up to a limit.
@return lock at limit or NULL if no more locks in the hash bucket */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
const lock_t*
lock_rec_validate(
/*==============*/
ulint start, /*!< in: lock_sys->rec_hash
bucket */
ib_uint64_t* limit) /*!< in/out: upper limit of
(space, page_no) */
{
ut_ad(lock_mutex_own());
ut_ad(mutex_own(&trx_sys->mutex));
for (const lock_t* lock = static_cast<const lock_t*>(
HASH_GET_FIRST(lock_sys->rec_hash, start));
lock != NULL;
lock = static_cast<const lock_t*>(HASH_GET_NEXT(hash, lock))) {
ib_uint64_t current;
ut_a(trx_in_trx_list(lock->trx));
ut_a(lock_get_type(lock) == LOCK_REC);
current = ut_ull_create(
lock->un_member.rec_lock.space,
lock->un_member.rec_lock.page_no);
if (current > *limit) {
*limit = current + 1;
return(lock);
}
}
return(0);
}
/*********************************************************************//**
Validate a record lock's block */
static
void
lock_rec_block_validate(
/*====================*/
ulint space_id,
ulint page_no)
{
/* The lock and the block that it is referring to may be freed at
this point. We pass BUF_GET_POSSIBLY_FREED to skip a debug check.
If the lock exists in lock_rec_validate_page() we assert
!block->page.file_page_was_freed. */
buf_block_t* block;
mtr_t mtr;
/* Make sure that the tablespace is not deleted while we are
trying to access the page. */
if (fil_space_t* space = fil_space_acquire(space_id)) {
mtr_start(&mtr);
block = buf_page_get_gen(
space_id, fsp_flags_get_zip_size(space->flags),
page_no, RW_X_LATCH, NULL,
BUF_GET_POSSIBLY_FREED,
__FILE__, __LINE__, &mtr);
buf_block_dbg_add_level(block, SYNC_NO_ORDER_CHECK);
ut_ad(lock_rec_validate_page(block));
mtr_commit(&mtr);
fil_space_release(space);
}
}
/*********************************************************************//**
Validates the lock system.
@return TRUE if ok */
static
bool
lock_validate()
/*===========*/
{
typedef std::pair<ulint, ulint> page_addr_t;
typedef std::set<page_addr_t> page_addr_set;
page_addr_set pages;
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
ut_a(lock_validate_table_locks(&trx_sys->rw_trx_list));
ut_a(lock_validate_table_locks(&trx_sys->ro_trx_list));
/* Iterate over all the record locks and validate the locks. We
don't want to hog the lock_sys_t::mutex and the trx_sys_t::mutex.
Release both mutexes during the validation check. */
for (ulint i = 0; i < hash_get_n_cells(lock_sys->rec_hash); i++) {
const lock_t* lock;
ib_uint64_t limit = 0;
while ((lock = lock_rec_validate(i, &limit)) != 0) {
ulint space = lock->un_member.rec_lock.space;
ulint page_no = lock->un_member.rec_lock.page_no;
pages.insert(std::make_pair(space, page_no));
}
}
mutex_exit(&trx_sys->mutex);
lock_mutex_exit();
for (page_addr_set::const_iterator it = pages.begin();
it != pages.end();
++it) {
lock_rec_block_validate((*it).first, (*it).second);
}
return(true);
}
#endif /* UNIV_DEBUG */
/*============ RECORD LOCK CHECKS FOR ROW OPERATIONS ====================*/
/*********************************************************************//**
Checks if locks of other transactions prevent an immediate insert of
a record. If they do, first tests if the query thread should anyway
be suspended for some reason; if not, then puts the transaction and
the query thread to the lock wait state and inserts a waiting request
for a gap x-lock to the lock queue.
@return DB_SUCCESS, DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_rec_insert_check_and_lock(
/*===========================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG bit is
set, does nothing */
const rec_t* rec, /*!< in: record after which to insert */
buf_block_t* block, /*!< in/out: buffer block of rec */
dict_index_t* index, /*!< in: index */
que_thr_t* thr, /*!< in: query thread */
mtr_t* mtr, /*!< in/out: mini-transaction */
ibool* inherit)/*!< out: set to TRUE if the new
inserted record maybe should inherit
LOCK_GAP type locks from the successor
record */
{
const rec_t* next_rec;
trx_t* trx;
lock_t* lock;
dberr_t err;
ulint next_rec_heap_no;
ibool inherit_in = *inherit;
#ifdef WITH_WSREP
lock_t* c_lock=NULL;
#endif
ut_ad(block->frame == page_align(rec));
ut_ad(!dict_index_is_online_ddl(index)
|| dict_index_is_clust(index)
|| (flags & BTR_CREATE_FLAG));
ut_ad((flags & BTR_NO_LOCKING_FLAG) || thr);
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
trx = thr_get_trx(thr);
if (UNIV_UNLIKELY(trx->fake_changes)) {
return(DB_SUCCESS);
}
next_rec = page_rec_get_next_const(rec);
next_rec_heap_no = page_rec_get_heap_no(next_rec);
lock_mutex_enter();
/* Because this code is invoked for a running transaction by
the thread that is serving the transaction, it is not necessary
to hold trx->mutex here. */
/* When inserting a record into an index, the table must be at
least IX-locked. When we are building an index, we would pass
BTR_NO_LOCKING_FLAG and skip the locking altogether. */
ut_ad(lock_table_has(trx, index->table, LOCK_IX));
lock = lock_rec_get_first(block, next_rec_heap_no);
if (UNIV_LIKELY(lock == NULL)) {
/* We optimize CPU time usage in the simplest case */
lock_mutex_exit();
if (inherit_in && !dict_index_is_clust(index)) {
/* Update the page max trx id field */
page_update_max_trx_id(block,
buf_block_get_page_zip(block),
trx->id, mtr);
}
*inherit = FALSE;
return(DB_SUCCESS);
}
*inherit = TRUE;
/* If another transaction has an explicit lock request which locks
the gap, waiting or granted, on the successor, the insert has to wait.
An exception is the case where the lock by the another transaction
is a gap type lock which it placed to wait for its turn to insert. We
do not consider that kind of a lock conflicting with our insert. This
eliminates an unnecessary deadlock which resulted when 2 transactions
had to wait for their insert. Both had waiting gap type lock requests
on the successor, which produced an unnecessary deadlock. */
#ifdef WITH_WSREP
if ((c_lock = (ib_lock_t*)lock_rec_other_has_conflicting(
static_cast<enum lock_mode>(
LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION),
block, next_rec_heap_no, trx))) {
#else
if (lock_rec_other_has_conflicting(
static_cast<enum lock_mode>(
LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION),
block, next_rec_heap_no, trx)) {
#endif /* WITH_WSREP */
/* Note that we may get DB_SUCCESS also here! */
trx_mutex_enter(trx);
#ifdef WITH_WSREP
err = lock_rec_enqueue_waiting(c_lock,
LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION,
block, next_rec_heap_no, index, thr);
#else
err = lock_rec_enqueue_waiting(
LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION,
block, next_rec_heap_no, index, thr);
#endif /* WITH_WSREP */
trx_mutex_exit(trx);
} else {
err = DB_SUCCESS;
}
lock_mutex_exit();
switch (err) {
case DB_SUCCESS_LOCKED_REC:
err = DB_SUCCESS;
/* fall through */
case DB_SUCCESS:
if (!inherit_in || dict_index_is_clust(index)) {
break;
}
/* Update the page max trx id field */
page_update_max_trx_id(block,
buf_block_get_page_zip(block),
trx->id, mtr);
default:
/* We only care about the two return values. */
break;
}
#ifdef UNIV_DEBUG
{
mem_heap_t* heap = NULL;
ulint offsets_[REC_OFFS_NORMAL_SIZE];
const ulint* offsets;
rec_offs_init(offsets_);
offsets = rec_get_offsets(next_rec, index, offsets_,
ULINT_UNDEFINED, &heap);
ut_ad(lock_rec_queue_validate(
FALSE, block, next_rec, index, offsets));
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
}
#endif /* UNIV_DEBUG */
return(err);
}
/*********************************************************************//**
If a transaction has an implicit x-lock on a record, but no explicit x-lock
set on the record, sets one for it. */
static
void
lock_rec_convert_impl_to_expl(
/*==========================*/
const buf_block_t* block, /*!< in: buffer block of rec */
const rec_t* rec, /*!< in: user record on page */
dict_index_t* index, /*!< in: index of record */
const ulint* offsets)/*!< in: rec_get_offsets(rec, index) */
{
trx_id_t trx_id;
ut_ad(!lock_mutex_own());
ut_ad(page_rec_is_user_rec(rec));
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(!page_rec_is_comp(rec) == !rec_offs_comp(offsets));
if (dict_index_is_clust(index)) {
trx_id = lock_clust_rec_some_has_impl(rec, index, offsets);
/* The clustered index record was last modified by
this transaction. The transaction may have been
committed a long time ago. */
} else {
ut_ad(!dict_index_is_online_ddl(index));
trx_id = lock_sec_rec_some_has_impl(rec, index, offsets);
/* The transaction can be committed before the
trx_is_active(trx_id, NULL) check below, because we are not
holding lock_mutex. */
ut_ad(!lock_rec_other_trx_holds_expl(LOCK_S | LOCK_REC_NOT_GAP,
trx_id, rec, block));
}
if (trx_id != 0) {
trx_id_t* impl_trx_desc;
ulint heap_no = page_rec_get_heap_no(rec);
lock_mutex_enter();
/* If the transaction is still active and has no
explicit x-lock set on the record, set one for it */
mutex_enter(&trx_sys->mutex);
impl_trx_desc = trx_find_descriptor(trx_sys->descriptors,
trx_sys->descr_n_used,
trx_id);
mutex_exit(&trx_sys->mutex);
/* trx_id cannot be committed until lock_mutex_exit()
because lock_trx_release_locks() acquires lock_sys->mutex */
if (impl_trx_desc != NULL
&& !lock_rec_has_expl(LOCK_X | LOCK_REC_NOT_GAP, block,
heap_no, trx_id)) {
ulint type_mode = (LOCK_REC | LOCK_X
| LOCK_REC_NOT_GAP);
mutex_enter(&trx_sys->mutex);
trx_t* impl_trx = trx_rw_get_active_trx_by_id(trx_id,
NULL);
mutex_exit(&trx_sys->mutex);
ut_ad(impl_trx != NULL);
lock_rec_add_to_queue(
type_mode, block, heap_no, index,
impl_trx, FALSE);
}
lock_mutex_exit();
}
}
/*********************************************************************//**
Checks if locks of other transactions prevent an immediate modify (update,
delete mark, or delete unmark) of a clustered index record. If they do,
first tests if the query thread should anyway be suspended for some
reason; if not, then puts the transaction and the query thread to the
lock wait state and inserts a waiting request for a record x-lock to the
lock queue.
@return DB_SUCCESS, DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_clust_rec_modify_check_and_lock(
/*=================================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG
bit is set, does nothing */
const buf_block_t* block, /*!< in: buffer block of rec */
const rec_t* rec, /*!< in: record which should be
modified */
dict_index_t* index, /*!< in: clustered index */
const ulint* offsets,/*!< in: rec_get_offsets(rec, index) */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err;
ulint heap_no;
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(dict_index_is_clust(index));
ut_ad(block->frame == page_align(rec));
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
if (UNIV_UNLIKELY(thr_get_trx(thr)->fake_changes)) {
return(DB_SUCCESS);
}
heap_no = rec_offs_comp(offsets)
? rec_get_heap_no_new(rec)
: rec_get_heap_no_old(rec);
/* If a transaction has no explicit x-lock set on the record, set one
for it */
lock_rec_convert_impl_to_expl(block, rec, index, offsets);
lock_mutex_enter();
trx_t* trx = thr_get_trx(thr);
ut_ad(lock_table_has(trx, index->table, LOCK_IX));
err = lock_rec_lock(TRUE, LOCK_X | LOCK_REC_NOT_GAP,
block, heap_no, index, thr);
MONITOR_INC(MONITOR_NUM_RECLOCK_REQ);
lock_mutex_exit();
ut_ad(lock_rec_queue_validate(FALSE, block, rec, index, offsets));
if (UNIV_UNLIKELY(err == DB_SUCCESS_LOCKED_REC)) {
err = DB_SUCCESS;
}
return(err);
}
/*********************************************************************//**
Checks if locks of other transactions prevent an immediate modify (delete
mark or delete unmark) of a secondary index record.
@return DB_SUCCESS, DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_sec_rec_modify_check_and_lock(
/*===============================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG
bit is set, does nothing */
buf_block_t* block, /*!< in/out: buffer block of rec */
const rec_t* rec, /*!< in: record which should be
modified; NOTE: as this is a secondary
index, we always have to modify the
clustered index record first: see the
comment below */
dict_index_t* index, /*!< in: secondary index */
que_thr_t* thr, /*!< in: query thread
(can be NULL if BTR_NO_LOCKING_FLAG) */
mtr_t* mtr) /*!< in/out: mini-transaction */
{
dberr_t err;
ulint heap_no;
ut_ad(!dict_index_is_clust(index));
ut_ad(!dict_index_is_online_ddl(index) || (flags & BTR_CREATE_FLAG));
ut_ad(block->frame == page_align(rec));
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
if (UNIV_UNLIKELY(thr_get_trx(thr)->fake_changes)) {
return(DB_SUCCESS);
}
heap_no = page_rec_get_heap_no(rec);
/* Another transaction cannot have an implicit lock on the record,
because when we come here, we already have modified the clustered
index record, and this would not have been possible if another active
transaction had modified this secondary index record. */
trx_t* trx = thr_get_trx(thr);
lock_mutex_enter();
ut_ad(lock_table_has(trx, index->table, LOCK_IX));
err = lock_rec_lock(TRUE, LOCK_X | LOCK_REC_NOT_GAP,
block, heap_no, index, thr);
MONITOR_INC(MONITOR_NUM_RECLOCK_REQ);
lock_mutex_exit();
#ifdef UNIV_DEBUG
{
mem_heap_t* heap = NULL;
ulint offsets_[REC_OFFS_NORMAL_SIZE];
const ulint* offsets;
rec_offs_init(offsets_);
offsets = rec_get_offsets(rec, index, offsets_,
ULINT_UNDEFINED, &heap);
ut_ad(lock_rec_queue_validate(
FALSE, block, rec, index, offsets));
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
}
#endif /* UNIV_DEBUG */
if (err == DB_SUCCESS || err == DB_SUCCESS_LOCKED_REC) {
/* Update the page max trx id field */
/* It might not be necessary to do this if
err == DB_SUCCESS (no new lock created),
but it should not cost too much performance. */
page_update_max_trx_id(block,
buf_block_get_page_zip(block),
thr_get_trx(thr)->id, mtr);
err = DB_SUCCESS;
}
return(err);
}
/*********************************************************************//**
Like lock_clust_rec_read_check_and_lock(), but reads a
secondary index record.
@return DB_SUCCESS, DB_SUCCESS_LOCKED_REC, DB_LOCK_WAIT, DB_DEADLOCK,
or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_sec_rec_read_check_and_lock(
/*=============================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG
bit is set, does nothing */
const buf_block_t* block, /*!< in: buffer block of rec */
const rec_t* rec, /*!< in: user record or page
supremum record which should
be read or passed over by a
read cursor */
dict_index_t* index, /*!< in: secondary index */
const ulint* offsets,/*!< in: rec_get_offsets(rec, index) */
enum lock_mode mode, /*!< in: mode of the lock which
the read cursor should set on
records: LOCK_S or LOCK_X; the
latter is possible in
SELECT FOR UPDATE */
ulint gap_mode,/*!< in: LOCK_ORDINARY, LOCK_GAP, or
LOCK_REC_NOT_GAP */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err;
ulint heap_no;
ut_ad(!dict_index_is_clust(index));
ut_ad(!dict_index_is_online_ddl(index));
ut_ad(block->frame == page_align(rec));
ut_ad(page_rec_is_user_rec(rec) || page_rec_is_supremum(rec));
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(mode == LOCK_X || mode == LOCK_S);
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
if (UNIV_UNLIKELY((thr && thr_get_trx(thr)->fake_changes))) {
if (!srv_fake_changes_locks) {
return(DB_SUCCESS);
}
if (mode == LOCK_X) {
mode = LOCK_S;
}
}
heap_no = page_rec_get_heap_no(rec);
/* Some transaction may have an implicit x-lock on the record only
if the max trx id for the page >= min trx id for the trx list or a
database recovery is running. */
if ((page_get_max_trx_id(block->frame) >= trx_rw_min_trx_id()
|| recv_recovery_is_on())
&& !page_rec_is_supremum(rec)) {
lock_rec_convert_impl_to_expl(block, rec, index, offsets);
}
trx_t* trx = thr_get_trx(thr);
lock_mutex_enter();
ut_ad(mode != LOCK_X
|| lock_table_has(trx, index->table, LOCK_IX));
ut_ad(mode != LOCK_S
|| lock_table_has(trx, index->table, LOCK_IS));
err = lock_rec_lock(FALSE, mode | gap_mode,
block, heap_no, index, thr);
MONITOR_INC(MONITOR_NUM_RECLOCK_REQ);
lock_mutex_exit();
ut_ad(lock_rec_queue_validate(FALSE, block, rec, index, offsets));
return(err);
}
/*********************************************************************//**
Checks if locks of other transactions prevent an immediate read, or passing
over by a read cursor, of a clustered index record. If they do, first tests
if the query thread should anyway be suspended for some reason; if not, then
puts the transaction and the query thread to the lock wait state and inserts a
waiting request for a record lock to the lock queue. Sets the requested mode
lock on the record.
@return DB_SUCCESS, DB_SUCCESS_LOCKED_REC, DB_LOCK_WAIT, DB_DEADLOCK,
or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_clust_rec_read_check_and_lock(
/*===============================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG
bit is set, does nothing */
const buf_block_t* block, /*!< in: buffer block of rec */
const rec_t* rec, /*!< in: user record or page
supremum record which should
be read or passed over by a
read cursor */
dict_index_t* index, /*!< in: clustered index */
const ulint* offsets,/*!< in: rec_get_offsets(rec, index) */
enum lock_mode mode, /*!< in: mode of the lock which
the read cursor should set on
records: LOCK_S or LOCK_X; the
latter is possible in
SELECT FOR UPDATE */
ulint gap_mode,/*!< in: LOCK_ORDINARY, LOCK_GAP, or
LOCK_REC_NOT_GAP */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err;
ulint heap_no;
ut_ad(dict_index_is_clust(index));
ut_ad(block->frame == page_align(rec));
ut_ad(page_rec_is_user_rec(rec) || page_rec_is_supremum(rec));
ut_ad(gap_mode == LOCK_ORDINARY || gap_mode == LOCK_GAP
|| gap_mode == LOCK_REC_NOT_GAP);
ut_ad(rec_offs_validate(rec, index, offsets));
if (flags & BTR_NO_LOCKING_FLAG) {
return(DB_SUCCESS);
}
if (UNIV_UNLIKELY((thr && thr_get_trx(thr)->fake_changes))) {
if (!srv_fake_changes_locks) {
return(DB_SUCCESS);
}
if (mode == LOCK_X) {
mode = LOCK_S;
}
}
heap_no = page_rec_get_heap_no(rec);
if (UNIV_LIKELY(heap_no != PAGE_HEAP_NO_SUPREMUM)) {
lock_rec_convert_impl_to_expl(block, rec, index, offsets);
}
lock_mutex_enter();
trx_t* trx = thr_get_trx(thr);
ut_ad(mode != LOCK_X
|| lock_table_has(trx, index->table, LOCK_IX));
ut_ad(mode != LOCK_S
|| lock_table_has(trx, index->table, LOCK_IS));
err = lock_rec_lock(FALSE, mode | gap_mode,
block, heap_no, index, thr);
MONITOR_INC(MONITOR_NUM_RECLOCK_REQ);
lock_mutex_exit();
ut_ad(lock_rec_queue_validate(FALSE, block, rec, index, offsets));
return(err);
}
/*********************************************************************//**
Checks if locks of other transactions prevent an immediate read, or passing
over by a read cursor, of a clustered index record. If they do, first tests
if the query thread should anyway be suspended for some reason; if not, then
puts the transaction and the query thread to the lock wait state and inserts a
waiting request for a record lock to the lock queue. Sets the requested mode
lock on the record. This is an alternative version of
lock_clust_rec_read_check_and_lock() that does not require the parameter
"offsets".
@return DB_SUCCESS, DB_LOCK_WAIT, DB_DEADLOCK, or DB_QUE_THR_SUSPENDED */
UNIV_INTERN
dberr_t
lock_clust_rec_read_check_and_lock_alt(
/*===================================*/
ulint flags, /*!< in: if BTR_NO_LOCKING_FLAG
bit is set, does nothing */
const buf_block_t* block, /*!< in: buffer block of rec */
const rec_t* rec, /*!< in: user record or page
supremum record which should
be read or passed over by a
read cursor */
dict_index_t* index, /*!< in: clustered index */
enum lock_mode mode, /*!< in: mode of the lock which
the read cursor should set on
records: LOCK_S or LOCK_X; the
latter is possible in
SELECT FOR UPDATE */
ulint gap_mode,/*!< in: LOCK_ORDINARY, LOCK_GAP, or
LOCK_REC_NOT_GAP */
que_thr_t* thr) /*!< in: query thread */
{
mem_heap_t* tmp_heap = NULL;
ulint offsets_[REC_OFFS_NORMAL_SIZE];
ulint* offsets = offsets_;
dberr_t err;
rec_offs_init(offsets_);
offsets = rec_get_offsets(rec, index, offsets,
ULINT_UNDEFINED, &tmp_heap);
err = lock_clust_rec_read_check_and_lock(flags, block, rec, index,
offsets, mode, gap_mode, thr);
if (tmp_heap) {
mem_heap_free(tmp_heap);
}
if (UNIV_UNLIKELY(err == DB_SUCCESS_LOCKED_REC)) {
err = DB_SUCCESS;
}
return(err);
}
/*******************************************************************//**
Release the last lock from the transaction's autoinc locks. */
UNIV_INLINE
void
lock_release_autoinc_last_lock(
/*===========================*/
ib_vector_t* autoinc_locks) /*!< in/out: vector of AUTOINC locks */
{
ulint last;
lock_t* lock;
ut_ad(lock_mutex_own());
ut_a(!ib_vector_is_empty(autoinc_locks));
/* The lock to be release must be the last lock acquired. */
last = ib_vector_size(autoinc_locks) - 1;
lock = *static_cast<lock_t**>(ib_vector_get(autoinc_locks, last));
/* Should have only AUTOINC locks in the vector. */
ut_a(lock_get_mode(lock) == LOCK_AUTO_INC);
ut_a(lock_get_type(lock) == LOCK_TABLE);
ut_a(lock->un_member.tab_lock.table != NULL);
/* This will remove the lock from the trx autoinc_locks too. */
lock_table_dequeue(lock);
/* Remove from the table vector too. */
lock_trx_table_locks_remove(lock);
}
/*******************************************************************//**
Check if a transaction holds any autoinc locks.
@return TRUE if the transaction holds any AUTOINC locks. */
static
ibool
lock_trx_holds_autoinc_locks(
/*=========================*/
const trx_t* trx) /*!< in: transaction */
{
ut_a(trx->autoinc_locks != NULL);
return(!ib_vector_is_empty(trx->autoinc_locks));
}
/*******************************************************************//**
Release all the transaction's autoinc locks. */
static
void
lock_release_autoinc_locks(
/*=======================*/
trx_t* trx) /*!< in/out: transaction */
{
ut_ad(lock_mutex_own());
/* If this is invoked for a running transaction by the thread
that is serving the transaction, then it is not necessary to
hold trx->mutex here. */
ut_a(trx->autoinc_locks != NULL);
/* We release the locks in the reverse order. This is to
avoid searching the vector for the element to delete at
the lower level. See (lock_table_remove_low()) for details. */
while (!ib_vector_is_empty(trx->autoinc_locks)) {
/* lock_table_remove_low() will also remove the lock from
the transaction's autoinc_locks vector. */
lock_release_autoinc_last_lock(trx->autoinc_locks);
}
/* Should release all locks. */
ut_a(ib_vector_is_empty(trx->autoinc_locks));
}
/*******************************************************************//**
Gets the type of a lock. Non-inline version for using outside of the
lock module.
@return LOCK_TABLE or LOCK_REC */
UNIV_INTERN
ulint
lock_get_type(
/*==========*/
const lock_t* lock) /*!< in: lock */
{
return(lock_get_type_low(lock));
}
/*******************************************************************//**
Gets the trx of the lock. Non-inline version for using outside of the
lock module.
@return trx_t* */
UNIV_INTERN
trx_t*
lock_get_trx(
/*=========*/
const lock_t* lock) /*!< in: lock */
{
return (lock->trx);
}
/*******************************************************************//**
Gets the id of the transaction owning a lock.
@return transaction id */
UNIV_INTERN
trx_id_t
lock_get_trx_id(
/*============*/
const lock_t* lock) /*!< in: lock */
{
return(lock->trx->id);
}
/*******************************************************************//**
Gets the mode of a lock in a human readable string.
The string should not be free()'d or modified.
@return lock mode */
UNIV_INTERN
const char*
lock_get_mode_str(
/*==============*/
const lock_t* lock) /*!< in: lock */
{
ibool is_gap_lock;
is_gap_lock = lock_get_type_low(lock) == LOCK_REC
&& lock_rec_get_gap(lock);
switch (lock_get_mode(lock)) {
case LOCK_S:
if (is_gap_lock) {
return("S,GAP");
} else {
return("S");
}
case LOCK_X:
if (is_gap_lock) {
return("X,GAP");
} else {
return("X");
}
case LOCK_IS:
if (is_gap_lock) {
return("IS,GAP");
} else {
return("IS");
}
case LOCK_IX:
if (is_gap_lock) {
return("IX,GAP");
} else {
return("IX");
}
case LOCK_AUTO_INC:
return("AUTO_INC");
default:
return("UNKNOWN");
}
}
/*******************************************************************//**
Gets the type of a lock in a human readable string.
The string should not be free()'d or modified.
@return lock type */
UNIV_INTERN
const char*
lock_get_type_str(
/*==============*/
const lock_t* lock) /*!< in: lock */
{
switch (lock_get_type_low(lock)) {
case LOCK_REC:
return("RECORD");
case LOCK_TABLE:
return("TABLE");
default:
return("UNKNOWN");
}
}
/*******************************************************************//**
Gets the table on which the lock is.
@return table */
UNIV_INLINE
dict_table_t*
lock_get_table(
/*===========*/
const lock_t* lock) /*!< in: lock */
{
switch (lock_get_type_low(lock)) {
case LOCK_REC:
ut_ad(dict_index_is_clust(lock->index)
|| !dict_index_is_online_ddl(lock->index));
return(lock->index->table);
case LOCK_TABLE:
return(lock->un_member.tab_lock.table);
default:
ut_error;
return(NULL);
}
}
/*******************************************************************//**
Gets the id of the table on which the lock is.
@return id of the table */
UNIV_INTERN
table_id_t
lock_get_table_id(
/*==============*/
const lock_t* lock) /*!< in: lock */
{
dict_table_t* table;
table = lock_get_table(lock);
return(table->id);
}
/*******************************************************************//**
Gets the name of the table on which the lock is.
The string should not be free()'d or modified.
@return name of the table */
UNIV_INTERN
const char*
lock_get_table_name(
/*================*/
const lock_t* lock) /*!< in: lock */
{
dict_table_t* table;
table = lock_get_table(lock);
return(table->name);
}
/*******************************************************************//**
For a record lock, gets the index on which the lock is.
@return index */
UNIV_INTERN
const dict_index_t*
lock_rec_get_index(
/*===============*/
const lock_t* lock) /*!< in: lock */
{
ut_a(lock_get_type_low(lock) == LOCK_REC);
ut_ad(dict_index_is_clust(lock->index)
|| !dict_index_is_online_ddl(lock->index));
return(lock->index);
}
/*******************************************************************//**
For a record lock, gets the name of the index on which the lock is.
The string should not be free()'d or modified.
@return name of the index */
UNIV_INTERN
const char*
lock_rec_get_index_name(
/*====================*/
const lock_t* lock) /*!< in: lock */
{
ut_a(lock_get_type_low(lock) == LOCK_REC);
ut_ad(dict_index_is_clust(lock->index)
|| !dict_index_is_online_ddl(lock->index));
return(lock->index->name);
}
/*******************************************************************//**
For a record lock, gets the tablespace number on which the lock is.
@return tablespace number */
UNIV_INTERN
ulint
lock_rec_get_space_id(
/*==================*/
const lock_t* lock) /*!< in: lock */
{
ut_a(lock_get_type_low(lock) == LOCK_REC);
return(lock->un_member.rec_lock.space);
}
/*******************************************************************//**
For a record lock, gets the page number on which the lock is.
@return page number */
UNIV_INTERN
ulint
lock_rec_get_page_no(
/*=================*/
const lock_t* lock) /*!< in: lock */
{
ut_a(lock_get_type_low(lock) == LOCK_REC);
return(lock->un_member.rec_lock.page_no);
}
/*********************************************************************//**
Cancels a waiting lock request and releases possible other transactions
waiting behind it. */
UNIV_INTERN
void
lock_cancel_waiting_and_release(
/*============================*/
lock_t* lock) /*!< in/out: waiting lock request */
{
que_thr_t* thr;
ut_ad(lock_mutex_own());
ut_ad(trx_mutex_own(lock->trx));
lock->trx->lock.cancel = TRUE;
if (lock_get_type_low(lock) == LOCK_REC) {
lock_rec_dequeue_from_page(lock);
} else {
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
if (lock->trx->autoinc_locks != NULL) {
/* Release the transaction's AUTOINC locks. */
lock_release_autoinc_locks(lock->trx);
}
lock_table_dequeue(lock);
}
/* Reset the wait flag and the back pointer to lock in trx. */
lock_reset_lock_and_trx_wait(lock);
/* The following function releases the trx from lock wait. */
thr = que_thr_end_lock_wait(lock->trx);
if (thr != NULL) {
lock_wait_release_thread_if_suspended(thr);
}
lock->trx->lock.cancel = FALSE;
}
/*********************************************************************//**
Unlocks AUTO_INC type locks that were possibly reserved by a trx. This
function should be called at the the end of an SQL statement, by the
connection thread that owns the transaction (trx->mysql_thd). */
UNIV_INTERN
void
lock_unlock_table_autoinc(
/*======================*/
trx_t* trx) /*!< in/out: transaction */
{
ut_ad(!lock_mutex_own());
ut_ad(!trx_mutex_own(trx));
ut_ad(!trx->lock.wait_lock);
/* This can be invoked on NOT_STARTED, ACTIVE, PREPARED,
but not COMMITTED transactions. */
ut_ad(trx_state_eq(trx, TRX_STATE_NOT_STARTED)
|| !trx_state_eq(trx, TRX_STATE_COMMITTED_IN_MEMORY));
/* This function is invoked for a running transaction by the
thread that is serving the transaction. Therefore it is not
necessary to hold trx->mutex here. */
if (lock_trx_holds_autoinc_locks(trx)) {
lock_mutex_enter();
lock_release_autoinc_locks(trx);
lock_mutex_exit();
}
}
/*********************************************************************//**
Releases a transaction's locks, and releases possible other transactions
waiting because of these locks. Change the state of the transaction to
TRX_STATE_COMMITTED_IN_MEMORY. */
UNIV_INTERN
void
lock_trx_release_locks(
/*===================*/
trx_t* trx) /*!< in/out: transaction */
{
assert_trx_in_list(trx);
if (trx_state_eq(trx, TRX_STATE_PREPARED)) {
mutex_enter(&trx_sys->mutex);
ut_a(trx_sys->n_prepared_trx > 0);
trx_sys->n_prepared_trx--;
if (trx->is_recovered) {
ut_a(trx_sys->n_prepared_recovered_trx > 0);
trx_sys->n_prepared_recovered_trx--;
}
mutex_exit(&trx_sys->mutex);
} else {
ut_ad(trx_state_eq(trx, TRX_STATE_ACTIVE));
}
/* The transition of trx->state to TRX_STATE_COMMITTED_IN_MEMORY
is protected by both the lock_sys->mutex and the trx->mutex.
We also lock trx_sys->mutex, because state transition to
TRX_STATE_COMMITTED_IN_MEMORY must be atomic with removing trx
from the descriptors array. */
lock_mutex_enter();
mutex_enter(&trx_sys->mutex);
trx_mutex_enter(trx);
/* The following assignment makes the transaction committed in memory
and makes its changes to data visible to other transactions.
NOTE that there is a small discrepancy from the strict formal
visibility rules here: a human user of the database can see
modifications made by another transaction T even before the necessary
log segment has been flushed to the disk. If the database happens to
crash before the flush, the user has seen modifications from T which
will never be a committed transaction. However, any transaction T2
which sees the modifications of the committing transaction T, and
which also itself makes modifications to the database, will get an lsn
larger than the committing transaction T. In the case where the log
flush fails, and T never gets committed, also T2 will never get
committed. */
/*--------------------------------------*/
trx->state = TRX_STATE_COMMITTED_IN_MEMORY;
/* The following also removes trx from trx_serial_list */
trx_release_descriptor(trx);
/*--------------------------------------*/
/* If the background thread trx_rollback_or_clean_recovered()
is still active then there is a chance that the rollback
thread may see this trx as COMMITTED_IN_MEMORY and goes ahead
to clean it up calling trx_cleanup_at_db_startup(). This can
happen in the case we are committing a trx here that is left
in PREPARED state during the crash. Note that commit of the
rollback of a PREPARED trx happens in the recovery thread
while the rollback of other transactions happen in the
background thread. To avoid this race we unconditionally unset
the is_recovered flag. */
trx->is_recovered = FALSE;
trx_mutex_exit(trx);
mutex_exit(&trx_sys->mutex);
lock_release(trx);
lock_mutex_exit();
}
/*********************************************************************//**
Check whether the transaction has already been rolled back because it
was selected as a deadlock victim, or if it has to wait then cancel
the wait lock.
@return DB_DEADLOCK, DB_LOCK_WAIT or DB_SUCCESS */
UNIV_INTERN
dberr_t
lock_trx_handle_wait(
/*=================*/
trx_t* trx) /*!< in/out: trx lock state */
{
dberr_t err;
lock_mutex_enter();
trx_mutex_enter(trx);
if (trx->lock.was_chosen_as_deadlock_victim) {
err = DB_DEADLOCK;
} else if (trx->lock.wait_lock != NULL) {
lock_cancel_waiting_and_release(trx->lock.wait_lock);
err = DB_LOCK_WAIT;
} else {
/* The lock was probably granted before we got here. */
err = DB_SUCCESS;
}
lock_mutex_exit();
trx_mutex_exit(trx);
return(err);
}
/*********************************************************************//**
Get the number of locks on a table.
@return number of locks */
UNIV_INTERN
ulint
lock_table_get_n_locks(
/*===================*/
const dict_table_t* table) /*!< in: table */
{
ulint n_table_locks;
lock_mutex_enter();
n_table_locks = UT_LIST_GET_LEN(table->locks);
lock_mutex_exit();
return(n_table_locks);
}
#ifdef UNIV_DEBUG
/*******************************************************************//**
Do an exhaustive check for any locks (table or rec) against the table.
@return lock if found */
static
const lock_t*
lock_table_locks_lookup(
/*====================*/
const dict_table_t* table, /*!< in: check if there are
any locks held on records in
this table or on the table
itself */
const trx_list_t* trx_list) /*!< in: trx list to check */
{
trx_t* trx;
ut_a(table != NULL);
ut_ad(lock_mutex_own());
ut_ad(mutex_own(&trx_sys->mutex));
ut_ad(trx_list == &trx_sys->rw_trx_list
|| trx_list == &trx_sys->ro_trx_list);
for (trx = UT_LIST_GET_FIRST(*trx_list);
trx != NULL;
trx = UT_LIST_GET_NEXT(trx_list, trx)) {
const lock_t* lock;
assert_trx_in_list(trx);
ut_ad(trx->read_only == (trx_list == &trx_sys->ro_trx_list));
for (lock = UT_LIST_GET_FIRST(trx->lock.trx_locks);
lock != NULL;
lock = UT_LIST_GET_NEXT(trx_locks, lock)) {
ut_a(lock->trx == trx);
if (lock_get_type_low(lock) == LOCK_REC) {
ut_ad(!dict_index_is_online_ddl(lock->index)
|| dict_index_is_clust(lock->index));
if (lock->index->table == table) {
return(lock);
}
} else if (lock->un_member.tab_lock.table == table) {
return(lock);
}
}
}
return(NULL);
}
#endif /* UNIV_DEBUG */
/*******************************************************************//**
Check if there are any locks (table or rec) against table.
@return TRUE if table has either table or record locks. */
UNIV_INTERN
ibool
lock_table_has_locks(
/*=================*/
const dict_table_t* table) /*!< in: check if there are any locks
held on records in this table or on the
table itself */
{
ibool has_locks;
lock_mutex_enter();
has_locks = UT_LIST_GET_LEN(table->locks) > 0 || table->n_rec_locks > 0;
#ifdef UNIV_DEBUG
if (!has_locks) {
mutex_enter(&trx_sys->mutex);
ut_ad(!lock_table_locks_lookup(table, &trx_sys->rw_trx_list));
ut_ad(!lock_table_locks_lookup(table, &trx_sys->ro_trx_list));
mutex_exit(&trx_sys->mutex);
}
#endif /* UNIV_DEBUG */
lock_mutex_exit();
return(has_locks);
}
#ifdef UNIV_DEBUG
/*******************************************************************//**
Check if the transaction holds any locks on the sys tables
or its records.
@return the strongest lock found on any sys table or 0 for none */
UNIV_INTERN
const lock_t*
lock_trx_has_sys_table_locks(
/*=========================*/
const trx_t* trx) /*!< in: transaction to check */
{
lint i;
const lock_t* strongest_lock = 0;
lock_mode strongest = LOCK_NONE;
lock_mutex_enter();
/* Find a valid mode. Note: ib_vector_size() can be 0. */
for (i = ib_vector_size(trx->lock.table_locks) - 1; i >= 0; --i) {
const lock_t* lock;
lock = *static_cast<const lock_t**>(
ib_vector_get(trx->lock.table_locks, i));
if (lock != NULL
&& dict_is_sys_table(lock->un_member.tab_lock.table->id)) {
strongest = lock_get_mode(lock);
ut_ad(strongest != LOCK_NONE);
strongest_lock = lock;
break;
}
}
if (strongest == LOCK_NONE) {
lock_mutex_exit();
return(NULL);
}
for (/* No op */; i >= 0; --i) {
const lock_t* lock;
lock = *static_cast<const lock_t**>(
ib_vector_get(trx->lock.table_locks, i));
if (lock == NULL) {
continue;
}
ut_ad(trx == lock->trx);
ut_ad(lock_get_type_low(lock) & LOCK_TABLE);
ut_ad(lock->un_member.tab_lock.table != NULL);
lock_mode mode = lock_get_mode(lock);
if (dict_is_sys_table(lock->un_member.tab_lock.table->id)
&& lock_mode_stronger_or_eq(mode, strongest)) {
strongest = mode;
strongest_lock = lock;
}
}
lock_mutex_exit();
return(strongest_lock);
}
/*******************************************************************//**
Check if the transaction holds an exclusive lock on a record.
@return whether the locks are held */
UNIV_INTERN
bool
lock_trx_has_rec_x_lock(
/*====================*/
const trx_t* trx, /*!< in: transaction to check */
const dict_table_t* table, /*!< in: table to check */
const buf_block_t* block, /*!< in: buffer block of the record */
ulint heap_no)/*!< in: record heap number */
{
enum lock_mode intention_lock;
enum lock_mode rec_lock;
ut_ad(heap_no > PAGE_HEAP_NO_SUPREMUM);
if (UNIV_UNLIKELY(trx->fake_changes)) {
intention_lock = LOCK_IS;
rec_lock = LOCK_S;
} else {
intention_lock = LOCK_IX;
rec_lock = LOCK_X;
}
lock_mutex_enter();
ut_a(lock_table_has(trx, table, intention_lock));
if (UNIV_LIKELY(srv_fake_changes_locks)) {
ut_a(lock_rec_has_expl(rec_lock | LOCK_REC_NOT_GAP,
block, heap_no, trx->id));
}
lock_mutex_exit();
return(true);
}
#endif /* UNIV_DEBUG */
/*******************************************************************//**
Get lock mode and table/index name
@return string containing lock info */
std::string
lock_get_info(
const lock_t* lock)
{
std::string info;
std::string mode("mode ");
std::string index("index ");
std::string table("table ");
std::string n_uniq(" n_uniq");
std::string n_user(" n_user");
std::string lock_mode((lock_get_mode_str(lock)));
std::string iname(lock->index->name);
std::string tname(lock->index->table_name);
#define SSTR( x ) reinterpret_cast< std::ostringstream & >( \
( std::ostringstream() << std::dec << x ) ).str()
info = mode + lock_mode
+ index + iname
+ table + tname
+ n_uniq + SSTR(lock->index->n_uniq)
+ n_user + SSTR(lock->index->n_user_defined_cols);
return info;
}
|