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
/*****************************************************************************
Copyright (c) 1997, 2010, Innobase Oy. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 59 Temple
Place, Suite 330, Boston, MA 02111-1307 USA
*****************************************************************************/
/**************************************************//**
@file ibuf/ibuf0ibuf.c
Insert buffer
Created 7/19/1997 Heikki Tuuri
*******************************************************/
#include "ibuf0ibuf.h"
/** Number of bits describing a single page */
#define IBUF_BITS_PER_PAGE 4
#if IBUF_BITS_PER_PAGE % 2
# error "IBUF_BITS_PER_PAGE must be an even number!"
#endif
/** The start address for an insert buffer bitmap page bitmap */
#define IBUF_BITMAP PAGE_DATA
#ifdef UNIV_NONINL
#include "ibuf0ibuf.ic"
#endif
#ifndef UNIV_HOTBACKUP
#include "buf0buf.h"
#include "buf0rea.h"
#include "fsp0fsp.h"
#include "trx0sys.h"
#include "fil0fil.h"
#include "thr0loc.h"
#include "rem0rec.h"
#include "btr0cur.h"
#include "btr0pcur.h"
#include "btr0btr.h"
#include "sync0sync.h"
#include "dict0boot.h"
#include "fut0lst.h"
#include "lock0lock.h"
#include "log0recv.h"
#include "que0que.h"
/* STRUCTURE OF AN INSERT BUFFER RECORD
In versions < 4.1.x:
1. The first field is the page number.
2. The second field is an array which stores type info for each subsequent
field. We store the information which affects the ordering of records, and
also the physical storage size of an SQL NULL value. E.g., for CHAR(10) it
is 10 bytes.
3. Next we have the fields of the actual index record.
In versions >= 4.1.x:
Note that contary to what we planned in the 1990's, there will only be one
insert buffer tree, and that is in the system tablespace of InnoDB.
1. The first field is the space id.
2. The second field is a one-byte marker (0) which differentiates records from
the < 4.1.x storage format.
3. The third field is the page number.
4. The fourth field contains the type info, where we have also added 2 bytes to
store the charset. In the compressed table format of 5.0.x we must add more
information here so that we can build a dummy 'index' struct which 5.0.x
can use in the binary search on the index page in the ibuf merge phase.
5. The rest of the fields contain the fields of the actual index record.
In versions >= 5.0.3:
The first byte of the fourth field is an additional marker (0) if the record
is in the compact format. The presence of this marker can be detected by
looking at the length of the field modulo DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE.
The high-order bit of the character set field in the type info is the
"nullable" flag for the field.
In versions >= 5.5:
The optional marker byte at the start of the fourth field is replaced by
mandatory 3 fields, totaling 4 bytes:
1. 2 bytes: Counter field, used to sort records within a (space id, page
no) in the order they were added. This is needed so that for example the
sequence of operations "INSERT x, DEL MARK x, INSERT x" is handled
correctly.
2. 1 byte: Operation type (see ibuf_op_t).
3. 1 byte: Flags. Currently only one flag exists, IBUF_REC_COMPACT.
To ensure older records, which do not have counters to enforce correct
sorting, are merged before any new records, ibuf_insert checks if we're
trying to insert to a position that contains old-style records, and if so,
refuses the insert. Thus, ibuf pages are gradually converted to the new
format as their corresponding buffer pool pages are read into memory.
*/
/* PREVENTING DEADLOCKS IN THE INSERT BUFFER SYSTEM
If an OS thread performs any operation that brings in disk pages from
non-system tablespaces into the buffer pool, or creates such a page there,
then the operation may have as a side effect an insert buffer index tree
compression. Thus, the tree latch of the insert buffer tree may be acquired
in the x-mode, and also the file space latch of the system tablespace may
be acquired in the x-mode.
Also, an insert to an index in a non-system tablespace can have the same
effect. How do we know this cannot lead to a deadlock of OS threads? There
is a problem with the i\o-handler threads: they break the latching order
because they own x-latches to pages which are on a lower level than the
insert buffer tree latch, its page latches, and the tablespace latch an
insert buffer operation can reserve.
The solution is the following: Let all the tree and page latches connected
with the insert buffer be later in the latching order than the fsp latch and
fsp page latches.
Insert buffer pages must be such that the insert buffer is never invoked
when these pages are accessed as this would result in a recursion violating
the latching order. We let a special i/o-handler thread take care of i/o to
the insert buffer pages and the ibuf bitmap pages, as well as the fsp bitmap
pages and the first inode page, which contains the inode of the ibuf tree: let
us call all these ibuf pages. To prevent deadlocks, we do not let a read-ahead
access both non-ibuf and ibuf pages.
Then an i/o-handler for the insert buffer never needs to access recursively the
insert buffer tree and thus obeys the latching order. On the other hand, other
i/o-handlers for other tablespaces may require access to the insert buffer,
but because all kinds of latches they need to access there are later in the
latching order, no violation of the latching order occurs in this case,
either.
A problem is how to grow and contract an insert buffer tree. As it is later
in the latching order than the fsp management, we have to reserve the fsp
latch first, before adding or removing pages from the insert buffer tree.
We let the insert buffer tree have its own file space management: a free
list of pages linked to the tree root. To prevent recursive using of the
insert buffer when adding pages to the tree, we must first load these pages
to memory, obtaining a latch on them, and only after that add them to the
free list of the insert buffer tree. More difficult is removing of pages
from the free list. If there is an excess of pages in the free list of the
ibuf tree, they might be needed if some thread reserves the fsp latch,
intending to allocate more file space. So we do the following: if a thread
reserves the fsp latch, we check the writer count field of the latch. If
this field has value 1, it means that the thread did not own the latch
before entering the fsp system, and the mtr of the thread contains no
modifications to the fsp pages. Now we are free to reserve the ibuf latch,
and check if there is an excess of pages in the free list. We can then, in a
separate mini-transaction, take them out of the free list and free them to
the fsp system.
To avoid deadlocks in the ibuf system, we divide file pages into three levels:
(1) non-ibuf pages,
(2) ibuf tree pages and the pages in the ibuf tree free list, and
(3) ibuf bitmap pages.
No OS thread is allowed to access higher level pages if it has latches to
lower level pages; even if the thread owns a B-tree latch it must not access
the B-tree non-leaf pages if it has latches on lower level pages. Read-ahead
is only allowed for level 1 and 2 pages. Dedicated i/o-handler threads handle
exclusively level 1 i/o. A dedicated i/o handler thread handles exclusively
level 2 i/o. However, if an OS thread does the i/o handling for itself, i.e.,
it uses synchronous aio, it can access any pages, as long as it obeys the
access order rules. */
/** Buffer pool size per the maximum insert buffer size */
#define IBUF_POOL_SIZE_PER_MAX_SIZE 2
/** Table name for the insert buffer. */
#define IBUF_TABLE_NAME "SYS_IBUF_TABLE"
/** Operations that can currently be buffered. */
UNIV_INTERN ibuf_use_t ibuf_use = IBUF_USE_ALL;
/** The insert buffer control structure */
UNIV_INTERN ibuf_t* ibuf = NULL;
/** Counter for ibuf_should_try() */
UNIV_INTERN ulint ibuf_flush_count = 0;
#ifdef UNIV_PFS_MUTEX
UNIV_INTERN mysql_pfs_key_t ibuf_pessimistic_insert_mutex_key;
UNIV_INTERN mysql_pfs_key_t ibuf_mutex_key;
UNIV_INTERN mysql_pfs_key_t ibuf_bitmap_mutex_key;
#endif /* UNIV_PFS_MUTEX */
#ifdef UNIV_IBUF_COUNT_DEBUG
/** Number of tablespaces in the ibuf_counts array */
#define IBUF_COUNT_N_SPACES 4
/** Number of pages within each tablespace in the ibuf_counts array */
#define IBUF_COUNT_N_PAGES 130000
/** Buffered entry counts for file pages, used in debugging */
static ulint ibuf_counts[IBUF_COUNT_N_SPACES][IBUF_COUNT_N_PAGES];
/******************************************************************//**
Checks that the indexes to ibuf_counts[][] are within limits. */
UNIV_INLINE
void
ibuf_count_check(
/*=============*/
ulint space_id, /*!< in: space identifier */
ulint page_no) /*!< in: page number */
{
if (space_id < IBUF_COUNT_N_SPACES && page_no < IBUF_COUNT_N_PAGES) {
return;
}
fprintf(stderr,
"InnoDB: UNIV_IBUF_COUNT_DEBUG limits space_id and page_no\n"
"InnoDB: and breaks crash recovery.\n"
"InnoDB: space_id=%lu, should be 0<=space_id<%lu\n"
"InnoDB: page_no=%lu, should be 0<=page_no<%lu\n",
(ulint) space_id, (ulint) IBUF_COUNT_N_SPACES,
(ulint) page_no, (ulint) IBUF_COUNT_N_PAGES);
ut_error;
}
#endif
/** @name Offsets to the per-page bits in the insert buffer bitmap */
/* @{ */
#define IBUF_BITMAP_FREE 0 /*!< Bits indicating the
amount of free space */
#define IBUF_BITMAP_BUFFERED 2 /*!< TRUE if there are buffered
changes for the page */
#define IBUF_BITMAP_IBUF 3 /*!< TRUE if page is a part of
the ibuf tree, excluding the
root page, or is in the free
list of the ibuf */
/* @} */
/* Various constants for checking the type of an ibuf record and extracting
data from it. For details, see the description of the record format at the
top of this file. */
/** @name Format of the fourth column of an insert buffer record
The fourth column in the MySQL 5.5 format contains an operation
type, counter, and some flags. */
/* @{ */
#define IBUF_REC_INFO_SIZE 4 /*!< Combined size of info fields at
the beginning of the fourth field */
#if IBUF_REC_INFO_SIZE >= DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE
# error "IBUF_REC_INFO_SIZE >= DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE"
#endif
/* Offsets for the fields at the beginning of the fourth field */
#define IBUF_REC_OFFSET_COUNTER 0 /*!< Operation counter */
#define IBUF_REC_OFFSET_TYPE 2 /*!< Type of operation */
#define IBUF_REC_OFFSET_FLAGS 3 /*!< Additional flags */
/* Record flag masks */
#define IBUF_REC_COMPACT 0x1 /*!< Set in
IBUF_REC_OFFSET_FLAGS if the
user index is in COMPACT
format or later */
/** The mutex used to block pessimistic inserts to ibuf trees */
static mutex_t ibuf_pessimistic_insert_mutex;
/** The mutex protecting the insert buffer structs */
static mutex_t ibuf_mutex;
/** The mutex protecting the insert buffer bitmaps */
static mutex_t ibuf_bitmap_mutex;
/** The area in pages from which contract looks for page numbers for merge */
#define IBUF_MERGE_AREA 8
/** Inside the merge area, pages which have at most 1 per this number less
buffered entries compared to maximum volume that can buffered for a single
page are merged along with the page whose buffer became full */
#define IBUF_MERGE_THRESHOLD 4
/** In ibuf_contract at most this number of pages is read to memory in one
batch, in order to merge the entries for them in the insert buffer */
#define IBUF_MAX_N_PAGES_MERGED IBUF_MERGE_AREA
/** If the combined size of the ibuf trees exceeds ibuf->max_size by this
many pages, we start to contract it in connection to inserts there, using
non-synchronous contract */
#define IBUF_CONTRACT_ON_INSERT_NON_SYNC 0
/** If the combined size of the ibuf trees exceeds ibuf->max_size by this
many pages, we start to contract it in connection to inserts there, using
synchronous contract */
#define IBUF_CONTRACT_ON_INSERT_SYNC 5
/** If the combined size of the ibuf trees exceeds ibuf->max_size by
this many pages, we start to contract it synchronous contract, but do
not insert */
#define IBUF_CONTRACT_DO_NOT_INSERT 10
/* TODO: how to cope with drop table if there are records in the insert
buffer for the indexes of the table? Is there actually any problem,
because ibuf merge is done to a page when it is read in, and it is
still physically like the index page even if the index would have been
dropped! So, there seems to be no problem. */
/******************************************************************//**
Sets the flag in the current OS thread local storage denoting that it is
inside an insert buffer routine. */
UNIV_INLINE
void
ibuf_enter(void)
/*============*/
{
ibool* ptr;
ptr = thr_local_get_in_ibuf_field();
ut_ad(*ptr == FALSE);
*ptr = TRUE;
}
/******************************************************************//**
Sets the flag in the current OS thread local storage denoting that it is
exiting an insert buffer routine. */
UNIV_INLINE
void
ibuf_exit(void)
/*===========*/
{
ibool* ptr;
ptr = thr_local_get_in_ibuf_field();
ut_ad(*ptr == TRUE);
*ptr = FALSE;
}
/******************************************************************//**
Returns TRUE if the current OS thread is performing an insert buffer
routine.
For instance, a read-ahead of non-ibuf pages is forbidden by threads
that are executing an insert buffer routine.
@return TRUE if inside an insert buffer routine */
UNIV_INTERN
ibool
ibuf_inside(void)
/*=============*/
{
return(*thr_local_get_in_ibuf_field());
}
/******************************************************************//**
Gets the ibuf header page and x-latches it.
@return insert buffer header page */
static
page_t*
ibuf_header_page_get(
/*=================*/
mtr_t* mtr) /*!< in: mtr */
{
buf_block_t* block;
ut_ad(!ibuf_inside());
block = buf_page_get(
IBUF_SPACE_ID, 0, FSP_IBUF_HEADER_PAGE_NO, RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_IBUF_HEADER);
return(buf_block_get_frame(block));
}
/******************************************************************//**
Gets the root page and x-latches it.
@return insert buffer tree root page */
static
page_t*
ibuf_tree_root_get(
/*===============*/
mtr_t* mtr) /*!< in: mtr */
{
buf_block_t* block;
ut_ad(ibuf_inside());
mtr_x_lock(dict_index_get_lock(ibuf->index), mtr);
block = buf_page_get(
IBUF_SPACE_ID, 0, FSP_IBUF_TREE_ROOT_PAGE_NO, RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
return(buf_block_get_frame(block));
}
#ifdef UNIV_IBUF_COUNT_DEBUG
/******************************************************************//**
Gets the ibuf count for a given page.
@return number of entries in the insert buffer currently buffered for
this page */
UNIV_INTERN
ulint
ibuf_count_get(
/*===========*/
ulint space, /*!< in: space id */
ulint page_no)/*!< in: page number */
{
ibuf_count_check(space, page_no);
return(ibuf_counts[space][page_no]);
}
/******************************************************************//**
Sets the ibuf count for a given page. */
static
void
ibuf_count_set(
/*===========*/
ulint space, /*!< in: space id */
ulint page_no,/*!< in: page number */
ulint val) /*!< in: value to set */
{
ibuf_count_check(space, page_no);
ut_a(val < UNIV_PAGE_SIZE);
ibuf_counts[space][page_no] = val;
}
#endif
/******************************************************************//**
Closes insert buffer and frees the data structures. */
UNIV_INTERN
void
ibuf_close(void)
/*============*/
{
mutex_free(&ibuf_pessimistic_insert_mutex);
memset(&ibuf_pessimistic_insert_mutex,
0x0, sizeof(ibuf_pessimistic_insert_mutex));
mutex_free(&ibuf_mutex);
memset(&ibuf_mutex, 0x0, sizeof(ibuf_mutex));
mutex_free(&ibuf_bitmap_mutex);
memset(&ibuf_bitmap_mutex, 0x0, sizeof(ibuf_mutex));
mem_free(ibuf);
ibuf = NULL;
}
/******************************************************************//**
Updates the size information of the ibuf, assuming the segment size has not
changed. */
static
void
ibuf_size_update(
/*=============*/
const page_t* root, /*!< in: ibuf tree root */
mtr_t* mtr) /*!< in: mtr */
{
ut_ad(mutex_own(&ibuf_mutex));
ibuf->free_list_len = flst_get_len(root + PAGE_HEADER
+ PAGE_BTR_IBUF_FREE_LIST, mtr);
ibuf->height = 1 + btr_page_get_level(root, mtr);
/* the '1 +' is the ibuf header page */
ibuf->size = ibuf->seg_size - (1 + ibuf->free_list_len);
ibuf->empty = page_get_n_recs(root) == 0;
}
/******************************************************************//**
Creates the insert buffer data structure at a database startup and initializes
the data structures for the insert buffer. */
UNIV_INTERN
void
ibuf_init_at_db_start(void)
/*=======================*/
{
page_t* root;
mtr_t mtr;
dict_table_t* table;
mem_heap_t* heap;
dict_index_t* index;
ulint n_used;
page_t* header_page;
ulint error;
ibuf = mem_alloc(sizeof(ibuf_t));
memset(ibuf, 0, sizeof(*ibuf));
/* Note that also a pessimistic delete can sometimes make a B-tree
grow in size, as the references on the upper levels of the tree can
change */
ibuf->max_size = buf_pool_get_curr_size() / UNIV_PAGE_SIZE
/ IBUF_POOL_SIZE_PER_MAX_SIZE;
mutex_create(ibuf_pessimistic_insert_mutex_key,
&ibuf_pessimistic_insert_mutex,
SYNC_IBUF_PESS_INSERT_MUTEX);
mutex_create(ibuf_mutex_key,
&ibuf_mutex, SYNC_IBUF_MUTEX);
mutex_create(ibuf_bitmap_mutex_key,
&ibuf_bitmap_mutex, SYNC_IBUF_BITMAP_MUTEX);
mtr_start(&mtr);
mutex_enter(&ibuf_mutex);
mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, NULL), &mtr);
header_page = ibuf_header_page_get(&mtr);
fseg_n_reserved_pages(header_page + IBUF_HEADER + IBUF_TREE_SEG_HEADER,
&n_used, &mtr);
ibuf_enter();
ut_ad(n_used >= 2);
ibuf->seg_size = n_used;
{
buf_block_t* block;
block = buf_page_get(
IBUF_SPACE_ID, 0, FSP_IBUF_TREE_ROOT_PAGE_NO,
RW_X_LATCH, &mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
root = buf_block_get_frame(block);
}
ibuf_size_update(root, &mtr);
mutex_exit(&ibuf_mutex);
mtr_commit(&mtr);
ibuf_exit();
heap = mem_heap_create(450);
/* Use old-style record format for the insert buffer. */
table = dict_mem_table_create(IBUF_TABLE_NAME, IBUF_SPACE_ID, 1, 0);
dict_mem_table_add_col(table, heap, "DUMMY_COLUMN", DATA_BINARY, 0, 0);
table->id = ut_dulint_add(DICT_IBUF_ID_MIN, IBUF_SPACE_ID);
dict_table_add_to_cache(table, heap);
mem_heap_free(heap);
index = dict_mem_index_create(
IBUF_TABLE_NAME, "CLUST_IND",
IBUF_SPACE_ID, DICT_CLUSTERED | DICT_UNIVERSAL | DICT_IBUF, 1);
dict_mem_index_add_field(index, "DUMMY_COLUMN", 0);
index->id = ut_dulint_add(DICT_IBUF_ID_MIN, IBUF_SPACE_ID);
error = dict_index_add_to_cache(table, index,
FSP_IBUF_TREE_ROOT_PAGE_NO, FALSE);
ut_a(error == DB_SUCCESS);
ibuf->index = dict_table_get_first_index(table);
}
#endif /* !UNIV_HOTBACKUP */
/*********************************************************************//**
Initializes an ibuf bitmap page. */
UNIV_INTERN
void
ibuf_bitmap_page_init(
/*==================*/
buf_block_t* block, /*!< in: bitmap page */
mtr_t* mtr) /*!< in: mtr */
{
page_t* page;
ulint byte_offset;
ulint zip_size = buf_block_get_zip_size(block);
ut_a(ut_is_2pow(zip_size));
page = buf_block_get_frame(block);
fil_page_set_type(page, FIL_PAGE_IBUF_BITMAP);
/* Write all zeros to the bitmap */
if (!zip_size) {
byte_offset = UT_BITS_IN_BYTES(UNIV_PAGE_SIZE
* IBUF_BITS_PER_PAGE);
} else {
byte_offset = UT_BITS_IN_BYTES(zip_size * IBUF_BITS_PER_PAGE);
}
memset(page + IBUF_BITMAP, 0, byte_offset);
/* The remaining area (up to the page trailer) is uninitialized. */
#ifndef UNIV_HOTBACKUP
mlog_write_initial_log_record(page, MLOG_IBUF_BITMAP_INIT, mtr);
#endif /* !UNIV_HOTBACKUP */
}
/*********************************************************************//**
Parses a redo log record of an ibuf bitmap page init.
@return end of log record or NULL */
UNIV_INTERN
byte*
ibuf_parse_bitmap_init(
/*===================*/
byte* ptr, /*!< in: buffer */
byte* end_ptr __attribute__((unused)), /*!< in: buffer end */
buf_block_t* block, /*!< in: block or NULL */
mtr_t* mtr) /*!< in: mtr or NULL */
{
ut_ad(ptr && end_ptr);
if (block) {
ibuf_bitmap_page_init(block, mtr);
}
return(ptr);
}
#ifndef UNIV_HOTBACKUP
/********************************************************************//**
Gets the desired bits for a given page from a bitmap page.
@return value of bits */
UNIV_INLINE
ulint
ibuf_bitmap_page_get_bits(
/*======================*/
const page_t* page, /*!< in: bitmap page */
ulint page_no,/*!< in: page whose bits to get */
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
ulint bit, /*!< in: IBUF_BITMAP_FREE,
IBUF_BITMAP_BUFFERED, ... */
mtr_t* mtr __attribute__((unused)))
/*!< in: mtr containing an
x-latch to the bitmap page */
{
ulint byte_offset;
ulint bit_offset;
ulint map_byte;
ulint value;
ut_ad(bit < IBUF_BITS_PER_PAGE);
#if IBUF_BITS_PER_PAGE % 2
# error "IBUF_BITS_PER_PAGE % 2 != 0"
#endif
ut_ad(ut_is_2pow(zip_size));
ut_ad(mtr_memo_contains_page(mtr, page, MTR_MEMO_PAGE_X_FIX));
if (!zip_size) {
bit_offset = (page_no % UNIV_PAGE_SIZE) * IBUF_BITS_PER_PAGE
+ bit;
} else {
bit_offset = (page_no & (zip_size - 1)) * IBUF_BITS_PER_PAGE
+ bit;
}
byte_offset = bit_offset / 8;
bit_offset = bit_offset % 8;
ut_ad(byte_offset + IBUF_BITMAP < UNIV_PAGE_SIZE);
map_byte = mach_read_from_1(page + IBUF_BITMAP + byte_offset);
value = ut_bit_get_nth(map_byte, bit_offset);
if (bit == IBUF_BITMAP_FREE) {
ut_ad(bit_offset + 1 < 8);
value = value * 2 + ut_bit_get_nth(map_byte, bit_offset + 1);
}
return(value);
}
/********************************************************************//**
Sets the desired bit for a given page in a bitmap page. */
static
void
ibuf_bitmap_page_set_bits(
/*======================*/
page_t* page, /*!< in: bitmap page */
ulint page_no,/*!< in: page whose bits to set */
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
ulint bit, /*!< in: IBUF_BITMAP_FREE, IBUF_BITMAP_BUFFERED, ... */
ulint val, /*!< in: value to set */
mtr_t* mtr) /*!< in: mtr containing an x-latch to the bitmap page */
{
ulint byte_offset;
ulint bit_offset;
ulint map_byte;
ut_ad(bit < IBUF_BITS_PER_PAGE);
#if IBUF_BITS_PER_PAGE % 2
# error "IBUF_BITS_PER_PAGE % 2 != 0"
#endif
ut_ad(ut_is_2pow(zip_size));
ut_ad(mtr_memo_contains_page(mtr, page, MTR_MEMO_PAGE_X_FIX));
#ifdef UNIV_IBUF_COUNT_DEBUG
ut_a((bit != IBUF_BITMAP_BUFFERED) || (val != FALSE)
|| (0 == ibuf_count_get(page_get_space_id(page),
page_no)));
#endif
if (!zip_size) {
bit_offset = (page_no % UNIV_PAGE_SIZE) * IBUF_BITS_PER_PAGE
+ bit;
} else {
bit_offset = (page_no & (zip_size - 1)) * IBUF_BITS_PER_PAGE
+ bit;
}
byte_offset = bit_offset / 8;
bit_offset = bit_offset % 8;
ut_ad(byte_offset + IBUF_BITMAP < UNIV_PAGE_SIZE);
map_byte = mach_read_from_1(page + IBUF_BITMAP + byte_offset);
if (bit == IBUF_BITMAP_FREE) {
ut_ad(bit_offset + 1 < 8);
ut_ad(val <= 3);
map_byte = ut_bit_set_nth(map_byte, bit_offset, val / 2);
map_byte = ut_bit_set_nth(map_byte, bit_offset + 1, val % 2);
} else {
ut_ad(val <= 1);
map_byte = ut_bit_set_nth(map_byte, bit_offset, val);
}
mlog_write_ulint(page + IBUF_BITMAP + byte_offset, map_byte,
MLOG_1BYTE, mtr);
}
/********************************************************************//**
Calculates the bitmap page number for a given page number.
@return the bitmap page number where the file page is mapped */
UNIV_INLINE
ulint
ibuf_bitmap_page_no_calc(
/*=====================*/
ulint zip_size, /*!< in: compressed page size in bytes;
0 for uncompressed pages */
ulint page_no) /*!< in: tablespace page number */
{
ut_ad(ut_is_2pow(zip_size));
if (!zip_size) {
return(FSP_IBUF_BITMAP_OFFSET
+ (page_no & ~(UNIV_PAGE_SIZE - 1)));
} else {
return(FSP_IBUF_BITMAP_OFFSET
+ (page_no & ~(zip_size - 1)));
}
}
/********************************************************************//**
Gets the ibuf bitmap page where the bits describing a given file page are
stored.
@return bitmap page where the file page is mapped, that is, the bitmap
page containing the descriptor bits for the file page; the bitmap page
is x-latched */
static
page_t*
ibuf_bitmap_get_map_page_func(
/*==========================*/
ulint space, /*!< in: space id of the file page */
ulint page_no,/*!< in: page number of the file page */
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
const char* file, /*!< in: file name */
ulint line, /*!< in: line where called */
mtr_t* mtr) /*!< in: mtr */
{
buf_block_t* block;
block = buf_page_get_gen(space, zip_size,
ibuf_bitmap_page_no_calc(zip_size, page_no),
RW_X_LATCH, NULL, BUF_GET,
file, line, mtr);
buf_block_dbg_add_level(block, SYNC_IBUF_BITMAP);
return(buf_block_get_frame(block));
}
/********************************************************************//**
Gets the ibuf bitmap page where the bits describing a given file page are
stored.
@return bitmap page where the file page is mapped, that is, the bitmap
page containing the descriptor bits for the file page; the bitmap page
is x-latched
@param space in: space id of the file page
@param page_no in: page number of the file page
@param zip_size in: compressed page size in bytes; 0 for uncompressed pages
@param mtr in: mini-transaction */
#define ibuf_bitmap_get_map_page(space, page_no, zip_size, mtr) \
ibuf_bitmap_get_map_page_func(space, page_no, zip_size, \
__FILE__, __LINE__, mtr)
/************************************************************************//**
Sets the free bits of the page in the ibuf bitmap. This is done in a separate
mini-transaction, hence this operation does not restrict further work to only
ibuf bitmap operations, which would result if the latch to the bitmap page
were kept. */
UNIV_INLINE
void
ibuf_set_free_bits_low(
/*===================*/
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
const buf_block_t* block, /*!< in: index page; free bits are set if
the index is non-clustered and page
level is 0 */
ulint val, /*!< in: value to set: < 4 */
mtr_t* mtr) /*!< in/out: mtr */
{
page_t* bitmap_page;
ulint space;
ulint page_no;
if (!page_is_leaf(buf_block_get_frame(block))) {
return;
}
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
bitmap_page = ibuf_bitmap_get_map_page(space, page_no, zip_size, mtr);
#ifdef UNIV_IBUF_DEBUG
# if 0
fprintf(stderr,
"Setting space %lu page %lu free bits to %lu should be %lu\n",
space, page_no, val,
ibuf_index_page_calc_free(zip_size, block));
# endif
ut_a(val <= ibuf_index_page_calc_free(zip_size, block));
#endif /* UNIV_IBUF_DEBUG */
ibuf_bitmap_page_set_bits(bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, val, mtr);
}
/************************************************************************//**
Sets the free bit of the page in the ibuf bitmap. This is done in a separate
mini-transaction, hence this operation does not restrict further work to only
ibuf bitmap operations, which would result if the latch to the bitmap page
were kept. */
UNIV_INTERN
void
ibuf_set_free_bits_func(
/*====================*/
buf_block_t* block, /*!< in: index page of a non-clustered index;
free bit is reset if page level is 0 */
#ifdef UNIV_IBUF_DEBUG
ulint max_val,/*!< in: ULINT_UNDEFINED or a maximum
value which the bits must have before
setting; this is for debugging */
#endif /* UNIV_IBUF_DEBUG */
ulint val) /*!< in: value to set: < 4 */
{
mtr_t mtr;
page_t* page;
page_t* bitmap_page;
ulint space;
ulint page_no;
ulint zip_size;
page = buf_block_get_frame(block);
if (!page_is_leaf(page)) {
return;
}
mtr_start(&mtr);
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
zip_size = buf_block_get_zip_size(block);
bitmap_page = ibuf_bitmap_get_map_page(space, page_no, zip_size, &mtr);
#ifdef UNIV_IBUF_DEBUG
if (max_val != ULINT_UNDEFINED) {
ulint old_val;
old_val = ibuf_bitmap_page_get_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, &mtr);
# if 0
if (old_val != max_val) {
fprintf(stderr,
"Ibuf: page %lu old val %lu max val %lu\n",
page_get_page_no(page),
old_val, max_val);
}
# endif
ut_a(old_val <= max_val);
}
# if 0
fprintf(stderr, "Setting page no %lu free bits to %lu should be %lu\n",
page_get_page_no(page), val,
ibuf_index_page_calc_free(zip_size, block));
# endif
ut_a(val <= ibuf_index_page_calc_free(zip_size, block));
#endif /* UNIV_IBUF_DEBUG */
ibuf_bitmap_page_set_bits(bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, val, &mtr);
mtr_commit(&mtr);
}
/************************************************************************//**
Resets the free bits of the page in the ibuf bitmap. This is done in a
separate mini-transaction, hence this operation does not restrict
further work to only ibuf bitmap operations, which would result if the
latch to the bitmap page were kept. NOTE: The free bits in the insert
buffer bitmap must never exceed the free space on a page. It is safe
to decrement or reset the bits in the bitmap in a mini-transaction
that is committed before the mini-transaction that affects the free
space. */
UNIV_INTERN
void
ibuf_reset_free_bits(
/*=================*/
buf_block_t* block) /*!< in: index page; free bits are set to 0
if the index is a non-clustered
non-unique, and page level is 0 */
{
ibuf_set_free_bits(block, 0, ULINT_UNDEFINED);
}
/**********************************************************************//**
Updates the free bits for an uncompressed page to reflect the present
state. Does this in the mtr given, which means that the latching
order rules virtually prevent any further operations for this OS
thread until mtr is committed. NOTE: The free bits in the insert
buffer bitmap must never exceed the free space on a page. It is safe
to set the free bits in the same mini-transaction that updated the
page. */
UNIV_INTERN
void
ibuf_update_free_bits_low(
/*======================*/
const buf_block_t* block, /*!< in: index page */
ulint max_ins_size, /*!< in: value of
maximum insert size
with reorganize before
the latest operation
performed to the page */
mtr_t* mtr) /*!< in/out: mtr */
{
ulint before;
ulint after;
ut_a(!buf_block_get_page_zip(block));
before = ibuf_index_page_calc_free_bits(0, max_ins_size);
after = ibuf_index_page_calc_free(0, block);
/* This approach cannot be used on compressed pages, since the
computed value of "before" often does not match the current
state of the bitmap. This is because the free space may
increase or decrease when a compressed page is reorganized. */
if (before != after) {
ibuf_set_free_bits_low(0, block, after, mtr);
}
}
/**********************************************************************//**
Updates the free bits for a compressed page to reflect the present
state. Does this in the mtr given, which means that the latching
order rules virtually prevent any further operations for this OS
thread until mtr is committed. NOTE: The free bits in the insert
buffer bitmap must never exceed the free space on a page. It is safe
to set the free bits in the same mini-transaction that updated the
page. */
UNIV_INTERN
void
ibuf_update_free_bits_zip(
/*======================*/
buf_block_t* block, /*!< in/out: index page */
mtr_t* mtr) /*!< in/out: mtr */
{
page_t* bitmap_page;
ulint space;
ulint page_no;
ulint zip_size;
ulint after;
space = buf_block_get_space(block);
page_no = buf_block_get_page_no(block);
zip_size = buf_block_get_zip_size(block);
ut_a(page_is_leaf(buf_block_get_frame(block)));
ut_a(zip_size);
bitmap_page = ibuf_bitmap_get_map_page(space, page_no, zip_size, mtr);
after = ibuf_index_page_calc_free_zip(zip_size, block);
if (after == 0) {
/* We move the page to the front of the buffer pool LRU list:
the purpose of this is to prevent those pages to which we
cannot make inserts using the insert buffer from slipping
out of the buffer pool */
buf_page_make_young(&block->page);
}
ibuf_bitmap_page_set_bits(bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, after, mtr);
}
/**********************************************************************//**
Updates the free bits for the two pages to reflect the present state.
Does this in the mtr given, which means that the latching order rules
virtually prevent any further operations until mtr is committed.
NOTE: The free bits in the insert buffer bitmap must never exceed the
free space on a page. It is safe to set the free bits in the same
mini-transaction that updated the pages. */
UNIV_INTERN
void
ibuf_update_free_bits_for_two_pages_low(
/*====================================*/
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
buf_block_t* block1, /*!< in: index page */
buf_block_t* block2, /*!< in: index page */
mtr_t* mtr) /*!< in: mtr */
{
ulint state;
/* As we have to x-latch two random bitmap pages, we have to acquire
the bitmap mutex to prevent a deadlock with a similar operation
performed by another OS thread. */
mutex_enter(&ibuf_bitmap_mutex);
state = ibuf_index_page_calc_free(zip_size, block1);
ibuf_set_free_bits_low(zip_size, block1, state, mtr);
state = ibuf_index_page_calc_free(zip_size, block2);
ibuf_set_free_bits_low(zip_size, block2, state, mtr);
mutex_exit(&ibuf_bitmap_mutex);
}
/**********************************************************************//**
Returns TRUE if the page is one of the fixed address ibuf pages.
@return TRUE if a fixed address ibuf i/o page */
UNIV_INLINE
ibool
ibuf_fixed_addr_page(
/*=================*/
ulint space, /*!< in: space id */
ulint zip_size,/*!< in: compressed page size in bytes;
0 for uncompressed pages */
ulint page_no)/*!< in: page number */
{
return((space == IBUF_SPACE_ID && page_no == IBUF_TREE_ROOT_PAGE_NO)
|| ibuf_bitmap_page(zip_size, page_no));
}
/***********************************************************************//**
Checks if a page is a level 2 or 3 page in the ibuf hierarchy of pages.
Must not be called when recv_no_ibuf_operations==TRUE.
@return TRUE if level 2 or level 3 page */
UNIV_INTERN
ibool
ibuf_page(
/*======*/
ulint space, /*!< in: space id */
ulint zip_size,/*!< in: compressed page size in bytes, or 0 */
ulint page_no,/*!< in: page number */
mtr_t* mtr) /*!< in: mtr which will contain an x-latch to the
bitmap page if the page is not one of the fixed
address ibuf pages, or NULL, in which case a new
transaction is created. */
{
ibool ret;
mtr_t local_mtr;
page_t* bitmap_page;
ut_ad(!recv_no_ibuf_operations);
if (ibuf_fixed_addr_page(space, zip_size, page_no)) {
return(TRUE);
} else if (space != IBUF_SPACE_ID) {
return(FALSE);
}
ut_ad(fil_space_get_type(IBUF_SPACE_ID) == FIL_TABLESPACE);
if (mtr == NULL) {
mtr = &local_mtr;
mtr_start(mtr);
}
bitmap_page = ibuf_bitmap_get_map_page(space, page_no, zip_size, mtr);
ret = ibuf_bitmap_page_get_bits(bitmap_page, page_no, zip_size,
IBUF_BITMAP_IBUF, mtr);
if (mtr == &local_mtr) {
mtr_commit(mtr);
}
return(ret);
}
/********************************************************************//**
Returns the page number field of an ibuf record.
@return page number */
static
ulint
ibuf_rec_get_page_no(
/*=================*/
const rec_t* rec) /*!< in: ibuf record */
{
const byte* field;
ulint len;
ut_ad(ibuf_inside());
ut_ad(rec_get_n_fields_old(rec) > 2);
field = rec_get_nth_field_old(rec, 1, &len);
if (len == 1) {
/* This is of the >= 4.1.x record format */
ut_a(trx_sys_multiple_tablespace_format);
field = rec_get_nth_field_old(rec, 2, &len);
} else {
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
field = rec_get_nth_field_old(rec, 0, &len);
}
ut_a(len == 4);
return(mach_read_from_4(field));
}
/********************************************************************//**
Returns the space id field of an ibuf record. For < 4.1.x format records
returns 0.
@return space id */
static
ulint
ibuf_rec_get_space(
/*===============*/
const rec_t* rec) /*!< in: ibuf record */
{
const byte* field;
ulint len;
ut_ad(ibuf_inside());
ut_ad(rec_get_n_fields_old(rec) > 2);
field = rec_get_nth_field_old(rec, 1, &len);
if (len == 1) {
/* This is of the >= 4.1.x record format */
ut_a(trx_sys_multiple_tablespace_format);
field = rec_get_nth_field_old(rec, 0, &len);
ut_a(len == 4);
return(mach_read_from_4(field));
}
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
return(0);
}
/****************************************************************//**
Get various information about an ibuf record in >= 4.1.x format. */
static
void
ibuf_rec_get_info(
/*==============*/
const rec_t* rec, /*!< in: ibuf record */
ibuf_op_t* op, /*!< out: operation type, or NULL */
ibool* comp, /*!< out: compact flag, or NULL */
ulint* info_len, /*!< out: length of info fields at the
start of the fourth field, or
NULL */
ulint* counter) /*!< in: counter value, or NULL */
{
const byte* types;
ulint fields;
ulint len;
/* Local variables to shadow arguments. */
ibuf_op_t op_local;
ibool comp_local;
ulint info_len_local;
ulint counter_local;
ut_ad(ibuf_inside());
fields = rec_get_n_fields_old(rec);
ut_a(fields > 4);
types = rec_get_nth_field_old(rec, 3, &len);
info_len_local = len % DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE;
switch (info_len_local) {
case 0:
case 1:
op_local = IBUF_OP_INSERT;
comp_local = info_len_local;
ut_ad(!counter);
counter_local = ULINT_UNDEFINED;
break;
case IBUF_REC_INFO_SIZE:
op_local = (ibuf_op_t)types[IBUF_REC_OFFSET_TYPE];
comp_local = types[IBUF_REC_OFFSET_FLAGS] & IBUF_REC_COMPACT;
counter_local = mach_read_from_2(
types + IBUF_REC_OFFSET_COUNTER);
break;
default:
ut_error;
}
ut_a(op_local < IBUF_OP_COUNT);
ut_a((len - info_len_local) ==
(fields - 4) * DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE);
if (op) {
*op = op_local;
}
if (comp) {
*comp = comp_local;
}
if (info_len) {
*info_len = info_len_local;
}
if (counter) {
*counter = counter_local;
}
}
/****************************************************************//**
Returns the operation type field of an ibuf record.
@return operation type */
static
ibuf_op_t
ibuf_rec_get_op_type(
/*=================*/
const rec_t* rec) /*!< in: ibuf record */
{
ulint len;
const byte* field;
ut_ad(ibuf_inside());
ut_ad(rec_get_n_fields_old(rec) > 2);
field = rec_get_nth_field_old(rec, 1, &len);
if (len > 1) {
/* This is a < 4.1.x format record */
return(IBUF_OP_INSERT);
} else {
ibuf_op_t op;
ibuf_rec_get_info(rec, &op, NULL, NULL, NULL);
return(op);
}
}
/****************************************************************//**
Read the first two bytes from a record's fourth field (counter field in new
records; something else in older records).
@return "counter" field, or ULINT_UNDEFINED if for some reason it
can't be read */
UNIV_INTERN
ulint
ibuf_rec_get_counter(
/*=================*/
const rec_t* rec) /*!< in: ibuf record */
{
const byte* ptr;
ulint len;
if (rec_get_n_fields_old(rec) < 4) {
return(ULINT_UNDEFINED);
}
ptr = rec_get_nth_field_old(rec, 3, &len);
if (len >= 2) {
return(mach_read_from_2(ptr));
} else {
return(ULINT_UNDEFINED);
}
}
/****************************************************************//**
Add accumulated operation counts to a permanent array. Both arrays must be
of size IBUF_OP_COUNT. */
static
void
ibuf_add_ops(
/*=========*/
ulint* arr, /*!< in/out: array to modify */
const ulint* ops) /*!< in: operation counts */
{
ulint i;
for (i = 0; i < IBUF_OP_COUNT; i++) {
arr[i] += ops[i];
}
}
/****************************************************************//**
Print operation counts. The array must be of size IBUF_OP_COUNT. */
static
void
ibuf_print_ops(
/*===========*/
const ulint* ops, /*!< in: operation counts */
FILE* file) /*!< in: file where to print */
{
static const char* op_names[] = {
"insert",
"delete mark",
"delete"
};
ulint i;
ut_a(UT_ARR_SIZE(op_names) == IBUF_OP_COUNT);
for (i = 0; i < IBUF_OP_COUNT; i++) {
fprintf(file, "%s %lu%s", op_names[i],
(ulong) ops[i], (i < (IBUF_OP_COUNT - 1)) ? ", " : "");
}
putc('\n', file);
}
/********************************************************************//**
Creates a dummy index for inserting a record to a non-clustered index.
@return dummy index */
static
dict_index_t*
ibuf_dummy_index_create(
/*====================*/
ulint n, /*!< in: number of fields */
ibool comp) /*!< in: TRUE=use compact record format */
{
dict_table_t* table;
dict_index_t* index;
table = dict_mem_table_create("IBUF_DUMMY",
DICT_HDR_SPACE, n,
comp ? DICT_TF_COMPACT : 0);
index = dict_mem_index_create("IBUF_DUMMY", "IBUF_DUMMY",
DICT_HDR_SPACE, 0, n);
index->table = table;
/* avoid ut_ad(index->cached) in dict_index_get_n_unique_in_tree */
index->cached = TRUE;
return(index);
}
/********************************************************************//**
Add a column to the dummy index */
static
void
ibuf_dummy_index_add_col(
/*=====================*/
dict_index_t* index, /*!< in: dummy index */
const dtype_t* type, /*!< in: the data type of the column */
ulint len) /*!< in: length of the column */
{
ulint i = index->table->n_def;
dict_mem_table_add_col(index->table, NULL, NULL,
dtype_get_mtype(type),
dtype_get_prtype(type),
dtype_get_len(type));
dict_index_add_col(index, index->table,
dict_table_get_nth_col(index->table, i), len);
}
/********************************************************************//**
Deallocates a dummy index for inserting a record to a non-clustered index. */
static
void
ibuf_dummy_index_free(
/*==================*/
dict_index_t* index) /*!< in, own: dummy index */
{
dict_table_t* table = index->table;
dict_mem_index_free(index);
dict_mem_table_free(table);
}
/*********************************************************************//**
Builds the entry to insert into a non-clustered index when we have the
corresponding record in an ibuf index.
NOTE that as we copy pointers to fields in ibuf_rec, the caller must
hold a latch to the ibuf_rec page as long as the entry is used!
@return own: entry to insert to a non-clustered index */
UNIV_INLINE
dtuple_t*
ibuf_build_entry_pre_4_1_x(
/*=======================*/
const rec_t* ibuf_rec, /*!< in: record in an insert buffer */
mem_heap_t* heap, /*!< in: heap where built */
dict_index_t** pindex) /*!< out, own: dummy index that
describes the entry */
{
ulint i;
ulint len;
const byte* types;
dtuple_t* tuple;
ulint n_fields;
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
n_fields = rec_get_n_fields_old(ibuf_rec) - 2;
tuple = dtuple_create(heap, n_fields);
types = rec_get_nth_field_old(ibuf_rec, 1, &len);
ut_a(len == n_fields * DATA_ORDER_NULL_TYPE_BUF_SIZE);
for (i = 0; i < n_fields; i++) {
const byte* data;
dfield_t* field;
field = dtuple_get_nth_field(tuple, i);
data = rec_get_nth_field_old(ibuf_rec, i + 2, &len);
dfield_set_data(field, data, len);
dtype_read_for_order_and_null_size(
dfield_get_type(field),
types + i * DATA_ORDER_NULL_TYPE_BUF_SIZE);
}
*pindex = ibuf_dummy_index_create(n_fields, FALSE);
return(tuple);
}
/*********************************************************************//**
Builds the entry used to
1) IBUF_OP_INSERT: insert into a non-clustered index
2) IBUF_OP_DELETE_MARK: find the record whose delete-mark flag we need to
activate
3) IBUF_OP_DELETE: find the record we need to delete
when we have the corresponding record in an ibuf index.
NOTE that as we copy pointers to fields in ibuf_rec, the caller must
hold a latch to the ibuf_rec page as long as the entry is used!
@return own: entry to insert to a non-clustered index */
static
dtuple_t*
ibuf_build_entry_from_ibuf_rec(
/*===========================*/
const rec_t* ibuf_rec, /*!< in: record in an insert buffer */
mem_heap_t* heap, /*!< in: heap where built */
dict_index_t** pindex) /*!< out, own: dummy index that
describes the entry */
{
dtuple_t* tuple;
dfield_t* field;
ulint n_fields;
const byte* types;
const byte* data;
ulint len;
ulint info_len;
ulint i;
ulint comp;
dict_index_t* index;
data = rec_get_nth_field_old(ibuf_rec, 1, &len);
if (len > 1) {
/* This a < 4.1.x format record */
return(ibuf_build_entry_pre_4_1_x(ibuf_rec, heap, pindex));
}
/* This a >= 4.1.x format record */
ut_a(trx_sys_multiple_tablespace_format);
ut_a(*data == 0);
ut_a(rec_get_n_fields_old(ibuf_rec) > 4);
n_fields = rec_get_n_fields_old(ibuf_rec) - 4;
tuple = dtuple_create(heap, n_fields);
types = rec_get_nth_field_old(ibuf_rec, 3, &len);
ibuf_rec_get_info(ibuf_rec, NULL, &comp, &info_len, NULL);
index = ibuf_dummy_index_create(n_fields, comp);
len -= info_len;
types += info_len;
ut_a(len == n_fields * DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE);
for (i = 0; i < n_fields; i++) {
field = dtuple_get_nth_field(tuple, i);
data = rec_get_nth_field_old(ibuf_rec, i + 4, &len);
dfield_set_data(field, data, len);
dtype_new_read_for_order_and_null_size(
dfield_get_type(field),
types + i * DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE);
ibuf_dummy_index_add_col(index, dfield_get_type(field), len);
}
/* Prevent an ut_ad() failure in page_zip_write_rec() by
adding system columns to the dummy table pointed to by the
dummy secondary index. The insert buffer is only used for
secondary indexes, whose records never contain any system
columns, such as DB_TRX_ID. */
ut_d(dict_table_add_system_columns(index->table, index->table->heap));
*pindex = index;
return(tuple);
}
/******************************************************************//**
Get the data size.
@return size of fields */
UNIV_INLINE
ulint
ibuf_rec_get_size(
/*==============*/
const rec_t* rec, /*!< in: ibuf record */
const byte* types, /*!< in: fields */
ulint n_fields, /*!< in: number of fields */
ibool pre_4_1, /*!< in: TRUE=pre-4.1 format,
FALSE=newer */
ulint comp) /*!< in: 0=ROW_FORMAT=REDUNDANT,
nonzero=ROW_FORMAT=COMPACT */
{
ulint i;
ulint field_offset;
ulint types_offset;
ulint size = 0;
if (pre_4_1) {
field_offset = 2;
types_offset = DATA_ORDER_NULL_TYPE_BUF_SIZE;
} else {
field_offset = 4;
types_offset = DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE;
}
for (i = 0; i < n_fields; i++) {
ulint len;
dtype_t dtype;
rec_get_nth_field_offs_old(rec, i + field_offset, &len);
if (len != UNIV_SQL_NULL) {
size += len;
} else if (pre_4_1) {
dtype_read_for_order_and_null_size(&dtype, types);
size += dtype_get_sql_null_size(&dtype, comp);
} else {
dtype_new_read_for_order_and_null_size(&dtype, types);
size += dtype_get_sql_null_size(&dtype, comp);
}
types += types_offset;
}
return(size);
}
/********************************************************************//**
Returns the space taken by a stored non-clustered index entry if converted to
an index record.
@return size of index record in bytes + an upper limit of the space
taken in the page directory */
static
ulint
ibuf_rec_get_volume(
/*================*/
const rec_t* ibuf_rec)/*!< in: ibuf record */
{
ulint len;
const byte* data;
const byte* types;
ulint n_fields;
ulint data_size;
ibool pre_4_1;
ulint comp;
ut_ad(ibuf_inside());
ut_ad(rec_get_n_fields_old(ibuf_rec) > 2);
data = rec_get_nth_field_old(ibuf_rec, 1, &len);
pre_4_1 = (len > 1);
if (pre_4_1) {
/* < 4.1.x format record */
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
n_fields = rec_get_n_fields_old(ibuf_rec) - 2;
types = rec_get_nth_field_old(ibuf_rec, 1, &len);
ut_ad(len == n_fields * DATA_ORDER_NULL_TYPE_BUF_SIZE);
comp = 0;
} else {
/* >= 4.1.x format record */
ibuf_op_t op;
ulint info_len;
ut_a(trx_sys_multiple_tablespace_format);
ut_a(*data == 0);
types = rec_get_nth_field_old(ibuf_rec, 3, &len);
ibuf_rec_get_info(ibuf_rec, &op, &comp, &info_len, NULL);
if (op == IBUF_OP_DELETE_MARK || op == IBUF_OP_DELETE) {
/* Delete-marking a record doesn't take any
additional space, and while deleting a record
actually frees up space, we have to play it safe and
pretend it takes no additional space (the record
might not exist, etc.). */
return(0);
} else if (comp) {
dtuple_t* entry;
ulint volume;
dict_index_t* dummy_index;
mem_heap_t* heap = mem_heap_create(500);
entry = ibuf_build_entry_from_ibuf_rec(
ibuf_rec, heap, &dummy_index);
volume = rec_get_converted_size(dummy_index, entry, 0);
ibuf_dummy_index_free(dummy_index);
mem_heap_free(heap);
return(volume + page_dir_calc_reserved_space(1));
}
types += info_len;
n_fields = rec_get_n_fields_old(ibuf_rec) - 4;
}
data_size = ibuf_rec_get_size(ibuf_rec, types, n_fields, pre_4_1, comp);
return(data_size + rec_get_converted_extra_size(data_size, n_fields, 0)
+ page_dir_calc_reserved_space(1));
}
/*********************************************************************//**
Builds the tuple to insert to an ibuf tree when we have an entry for a
non-clustered index.
NOTE that the original entry must be kept because we copy pointers to
its fields.
@return own: entry to insert into an ibuf index tree */
static
dtuple_t*
ibuf_entry_build(
/*=============*/
ibuf_op_t op, /*!< in: operation type */
dict_index_t* index, /*!< in: non-clustered index */
const dtuple_t* entry, /*!< in: entry for a non-clustered index */
ulint space, /*!< in: space id */
ulint page_no,/*!< in: index page number where entry should
be inserted */
ulint counter,/*!< in: counter value;
ULINT_UNDEFINED=not used */
mem_heap_t* heap) /*!< in: heap into which to build */
{
dtuple_t* tuple;
dfield_t* field;
const dfield_t* entry_field;
ulint n_fields;
byte* buf;
byte* ti;
byte* type_info;
ulint i;
ut_ad(counter != ULINT_UNDEFINED || op == IBUF_OP_INSERT);
ut_ad(counter == ULINT_UNDEFINED || counter <= 0xFFFF);
ut_ad(op < IBUF_OP_COUNT);
/* We have to build a tuple with the following fields:
1-4) These are described at the top of this file.
5) The rest of the fields are copied from the entry.
All fields in the tuple are ordered like the type binary in our
insert buffer tree. */
n_fields = dtuple_get_n_fields(entry);
tuple = dtuple_create(heap, n_fields + 4);
/* 1) Space Id */
field = dtuple_get_nth_field(tuple, 0);
buf = mem_heap_alloc(heap, 4);
mach_write_to_4(buf, space);
dfield_set_data(field, buf, 4);
/* 2) Marker byte */
field = dtuple_get_nth_field(tuple, 1);
buf = mem_heap_alloc(heap, 1);
/* We set the marker byte zero */
mach_write_to_1(buf, 0);
dfield_set_data(field, buf, 1);
/* 3) Page number */
field = dtuple_get_nth_field(tuple, 2);
buf = mem_heap_alloc(heap, 4);
mach_write_to_4(buf, page_no);
dfield_set_data(field, buf, 4);
/* 4) Type info, part #1 */
if (counter == ULINT_UNDEFINED) {
i = dict_table_is_comp(index->table) ? 1 : 0;
} else {
ut_ad(counter <= 0xFFFF);
i = IBUF_REC_INFO_SIZE;
}
ti = type_info = mem_heap_alloc(heap, i + n_fields
* DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE);
switch (i) {
default:
ut_error;
break;
case 1:
/* set the flag for ROW_FORMAT=COMPACT */
*ti++ = 0;
/* fall through */
case 0:
/* the old format does not allow delete buffering */
ut_ad(op == IBUF_OP_INSERT);
break;
case IBUF_REC_INFO_SIZE:
mach_write_to_2(ti + IBUF_REC_OFFSET_COUNTER, counter);
ti[IBUF_REC_OFFSET_TYPE] = (byte) op;
ti[IBUF_REC_OFFSET_FLAGS] = dict_table_is_comp(index->table)
? IBUF_REC_COMPACT : 0;
ti += IBUF_REC_INFO_SIZE;
break;
}
/* 5+) Fields from the entry */
for (i = 0; i < n_fields; i++) {
ulint fixed_len;
const dict_field_t* ifield;
/* We add 4 below because we have the 4 extra fields at the
start of an ibuf record */
field = dtuple_get_nth_field(tuple, i + 4);
entry_field = dtuple_get_nth_field(entry, i);
dfield_copy(field, entry_field);
ifield = dict_index_get_nth_field(index, i);
/* Prefix index columns of fixed-length columns are of
fixed length. However, in the function call below,
dfield_get_type(entry_field) contains the fixed length
of the column in the clustered index. Replace it with
the fixed length of the secondary index column. */
fixed_len = ifield->fixed_len;
#ifdef UNIV_DEBUG
if (fixed_len) {
/* dict_index_add_col() should guarantee these */
ut_ad(fixed_len <= (ulint)
dfield_get_type(entry_field)->len);
if (ifield->prefix_len) {
ut_ad(ifield->prefix_len == fixed_len);
} else {
ut_ad(fixed_len == (ulint)
dfield_get_type(entry_field)->len);
}
}
#endif /* UNIV_DEBUG */
dtype_new_store_for_order_and_null_size(
ti, dfield_get_type(entry_field), fixed_len);
ti += DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE;
}
/* 4) Type info, part #2 */
field = dtuple_get_nth_field(tuple, 3);
dfield_set_data(field, type_info, ti - type_info);
/* Set all the types in the new tuple binary */
dtuple_set_types_binary(tuple, n_fields + 4);
return(tuple);
}
/*********************************************************************//**
Builds a search tuple used to search buffered inserts for an index page.
This is for < 4.1.x format records
@return own: search tuple */
static
dtuple_t*
ibuf_search_tuple_build(
/*====================*/
ulint space, /*!< in: space id */
ulint page_no,/*!< in: index page number */
mem_heap_t* heap) /*!< in: heap into which to build */
{
dtuple_t* tuple;
dfield_t* field;
byte* buf;
ut_a(space == 0);
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
tuple = dtuple_create(heap, 1);
/* Store the page number in tuple */
field = dtuple_get_nth_field(tuple, 0);
buf = mem_heap_alloc(heap, 4);
mach_write_to_4(buf, page_no);
dfield_set_data(field, buf, 4);
dtuple_set_types_binary(tuple, 1);
return(tuple);
}
/*********************************************************************//**
Builds a search tuple used to search buffered inserts for an index page.
This is for >= 4.1.x format records.
@return own: search tuple */
static
dtuple_t*
ibuf_new_search_tuple_build(
/*========================*/
ulint space, /*!< in: space id */
ulint page_no,/*!< in: index page number */
mem_heap_t* heap) /*!< in: heap into which to build */
{
dtuple_t* tuple;
dfield_t* field;
byte* buf;
ut_a(trx_sys_multiple_tablespace_format);
tuple = dtuple_create(heap, 3);
/* Store the space id in tuple */
field = dtuple_get_nth_field(tuple, 0);
buf = mem_heap_alloc(heap, 4);
mach_write_to_4(buf, space);
dfield_set_data(field, buf, 4);
/* Store the new format record marker byte */
field = dtuple_get_nth_field(tuple, 1);
buf = mem_heap_alloc(heap, 1);
mach_write_to_1(buf, 0);
dfield_set_data(field, buf, 1);
/* Store the page number in tuple */
field = dtuple_get_nth_field(tuple, 2);
buf = mem_heap_alloc(heap, 4);
mach_write_to_4(buf, page_no);
dfield_set_data(field, buf, 4);
dtuple_set_types_binary(tuple, 3);
return(tuple);
}
/*********************************************************************//**
Checks if there are enough pages in the free list of the ibuf tree that we
dare to start a pessimistic insert to the insert buffer.
@return TRUE if enough free pages in list */
UNIV_INLINE
ibool
ibuf_data_enough_free_for_insert(void)
/*==================================*/
{
ut_ad(mutex_own(&ibuf_mutex));
/* We want a big margin of free pages, because a B-tree can sometimes
grow in size also if records are deleted from it, as the node pointers
can change, and we must make sure that we are able to delete the
inserts buffered for pages that we read to the buffer pool, without
any risk of running out of free space in the insert buffer. */
return(ibuf->free_list_len >= (ibuf->size / 2) + 3 * ibuf->height);
}
/*********************************************************************//**
Checks if there are enough pages in the free list of the ibuf tree that we
should remove them and free to the file space management.
@return TRUE if enough free pages in list */
UNIV_INLINE
ibool
ibuf_data_too_much_free(void)
/*=========================*/
{
ut_ad(mutex_own(&ibuf_mutex));
return(ibuf->free_list_len >= 3 + (ibuf->size / 2) + 3 * ibuf->height);
}
/*********************************************************************//**
Allocates a new page from the ibuf file segment and adds it to the free
list.
@return DB_SUCCESS, or DB_STRONG_FAIL if no space left */
static
ulint
ibuf_add_free_page(void)
/*====================*/
{
mtr_t mtr;
page_t* header_page;
ulint flags;
ulint zip_size;
ulint page_no;
page_t* page;
page_t* root;
page_t* bitmap_page;
mtr_start(&mtr);
/* Acquire the fsp latch before the ibuf header, obeying the latching
order */
mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &flags), &mtr);
zip_size = dict_table_flags_to_zip_size(flags);
header_page = ibuf_header_page_get(&mtr);
/* Allocate a new page: NOTE that if the page has been a part of a
non-clustered index which has subsequently been dropped, then the
page may have buffered inserts in the insert buffer, and these
should be deleted from there. These get deleted when the page
allocation creates the page in buffer. Thus the call below may end
up calling the insert buffer routines and, as we yet have no latches
to insert buffer tree pages, these routines can run without a risk
of a deadlock. This is the reason why we created a special ibuf
header page apart from the ibuf tree. */
page_no = fseg_alloc_free_page(
header_page + IBUF_HEADER + IBUF_TREE_SEG_HEADER, 0, FSP_UP,
&mtr);
if (page_no == FIL_NULL) {
mtr_commit(&mtr);
return(DB_STRONG_FAIL);
}
{
buf_block_t* block;
block = buf_page_get(
IBUF_SPACE_ID, 0, page_no, RW_X_LATCH, &mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE_NEW);
page = buf_block_get_frame(block);
}
ibuf_enter();
mutex_enter(&ibuf_mutex);
root = ibuf_tree_root_get(&mtr);
/* Add the page to the free list and update the ibuf size data */
flst_add_last(root + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST,
page + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST_NODE, &mtr);
mlog_write_ulint(page + FIL_PAGE_TYPE, FIL_PAGE_IBUF_FREE_LIST,
MLOG_2BYTES, &mtr);
ibuf->seg_size++;
ibuf->free_list_len++;
/* Set the bit indicating that this page is now an ibuf tree page
(level 2 page) */
bitmap_page = ibuf_bitmap_get_map_page(
IBUF_SPACE_ID, page_no, zip_size, &mtr);
ibuf_bitmap_page_set_bits(
bitmap_page, page_no, zip_size, IBUF_BITMAP_IBUF, TRUE, &mtr);
mtr_commit(&mtr);
mutex_exit(&ibuf_mutex);
ibuf_exit();
return(DB_SUCCESS);
}
/*********************************************************************//**
Removes a page from the free list and frees it to the fsp system. */
static
void
ibuf_remove_free_page(void)
/*=======================*/
{
mtr_t mtr;
mtr_t mtr2;
page_t* header_page;
ulint flags;
ulint zip_size;
ulint page_no;
page_t* page;
page_t* root;
page_t* bitmap_page;
mtr_start(&mtr);
/* Acquire the fsp latch before the ibuf header, obeying the latching
order */
mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &flags), &mtr);
zip_size = dict_table_flags_to_zip_size(flags);
header_page = ibuf_header_page_get(&mtr);
/* Prevent pessimistic inserts to insert buffer trees for a while */
mutex_enter(&ibuf_pessimistic_insert_mutex);
ibuf_enter();
mutex_enter(&ibuf_mutex);
if (!ibuf_data_too_much_free()) {
mutex_exit(&ibuf_mutex);
ibuf_exit();
mutex_exit(&ibuf_pessimistic_insert_mutex);
mtr_commit(&mtr);
return;
}
mtr_start(&mtr2);
root = ibuf_tree_root_get(&mtr2);
page_no = flst_get_last(root + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST,
&mtr2).page;
/* NOTE that we must release the latch on the ibuf tree root
because in fseg_free_page we access level 1 pages, and the root
is a level 2 page. */
mtr_commit(&mtr2);
mutex_exit(&ibuf_mutex);
ibuf_exit();
/* Since pessimistic inserts were prevented, we know that the
page is still in the free list. NOTE that also deletes may take
pages from the free list, but they take them from the start, and
the free list was so long that they cannot have taken the last
page from it. */
fseg_free_page(header_page + IBUF_HEADER + IBUF_TREE_SEG_HEADER,
IBUF_SPACE_ID, page_no, &mtr);
#ifdef UNIV_DEBUG_FILE_ACCESSES
buf_page_reset_file_page_was_freed(IBUF_SPACE_ID, page_no);
#endif
ibuf_enter();
mutex_enter(&ibuf_mutex);
root = ibuf_tree_root_get(&mtr);
ut_ad(page_no == flst_get_last(root + PAGE_HEADER
+ PAGE_BTR_IBUF_FREE_LIST, &mtr).page);
{
buf_block_t* block;
block = buf_page_get(
IBUF_SPACE_ID, 0, page_no, RW_X_LATCH, &mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
page = buf_block_get_frame(block);
}
/* Remove the page from the free list and update the ibuf size data */
flst_remove(root + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST,
page + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST_NODE, &mtr);
ibuf->seg_size--;
ibuf->free_list_len--;
mutex_exit(&ibuf_pessimistic_insert_mutex);
/* Set the bit indicating that this page is no more an ibuf tree page
(level 2 page) */
bitmap_page = ibuf_bitmap_get_map_page(
IBUF_SPACE_ID, page_no, zip_size, &mtr);
ibuf_bitmap_page_set_bits(
bitmap_page, page_no, zip_size, IBUF_BITMAP_IBUF, FALSE, &mtr);
#ifdef UNIV_DEBUG_FILE_ACCESSES
buf_page_set_file_page_was_freed(IBUF_SPACE_ID, page_no);
#endif
mtr_commit(&mtr);
mutex_exit(&ibuf_mutex);
ibuf_exit();
}
/***********************************************************************//**
Frees excess pages from the ibuf free list. This function is called when an OS
thread calls fsp services to allocate a new file segment, or a new page to a
file segment, and the thread did not own the fsp latch before this call. */
UNIV_INTERN
void
ibuf_free_excess_pages(void)
/*========================*/
{
ulint i;
#ifdef UNIV_SYNC_DEBUG
ut_ad(rw_lock_own(fil_space_get_latch(IBUF_SPACE_ID, NULL),
RW_LOCK_EX));
#endif /* UNIV_SYNC_DEBUG */
ut_ad(rw_lock_get_x_lock_count(
fil_space_get_latch(IBUF_SPACE_ID, NULL)) == 1);
ut_ad(!ibuf_inside());
/* NOTE: We require that the thread did not own the latch before,
because then we know that we can obey the correct latching order
for ibuf latches */
if (!ibuf) {
/* Not yet initialized; not sure if this is possible, but
does no harm to check for it. */
return;
}
/* Free at most a few pages at a time, so that we do not delay the
requested service too much */
for (i = 0; i < 4; i++) {
mutex_enter(&ibuf_mutex);
if (!ibuf_data_too_much_free()) {
mutex_exit(&ibuf_mutex);
return;
}
mutex_exit(&ibuf_mutex);
ibuf_remove_free_page();
}
}
/*********************************************************************//**
Reads page numbers from a leaf in an ibuf tree.
@return a lower limit for the combined volume of records which will be
merged */
static
ulint
ibuf_get_merge_page_nos(
/*====================*/
ibool contract,/*!< in: TRUE if this function is called to
contract the tree, FALSE if this is called
when a single page becomes full and we look
if it pays to read also nearby pages */
rec_t* rec, /*!< in: record from which we read up and down
in the chain of records */
ulint* space_ids,/*!< in/out: space id's of the pages */
ib_int64_t* space_versions,/*!< in/out: tablespace version
timestamps; used to prevent reading in old
pages after DISCARD + IMPORT tablespace */
ulint* page_nos,/*!< in/out: buffer for at least
IBUF_MAX_N_PAGES_MERGED many page numbers;
the page numbers are in an ascending order */
ulint* n_stored)/*!< out: number of page numbers stored to
page_nos in this function */
{
ulint prev_page_no;
ulint prev_space_id;
ulint first_page_no;
ulint first_space_id;
ulint rec_page_no;
ulint rec_space_id;
ulint sum_volumes;
ulint volume_for_page;
ulint rec_volume;
ulint limit;
ulint n_pages;
*n_stored = 0;
limit = ut_min(IBUF_MAX_N_PAGES_MERGED, buf_pool->curr_size / 4);
if (page_rec_is_supremum(rec)) {
rec = page_rec_get_prev(rec);
}
if (page_rec_is_infimum(rec)) {
rec = page_rec_get_next(rec);
}
if (page_rec_is_supremum(rec)) {
return(0);
}
first_page_no = ibuf_rec_get_page_no(rec);
first_space_id = ibuf_rec_get_space(rec);
n_pages = 0;
prev_page_no = 0;
prev_space_id = 0;
/* Go backwards from the first rec until we reach the border of the
'merge area', or the page start or the limit of storeable pages is
reached */
while (!page_rec_is_infimum(rec) && UNIV_LIKELY(n_pages < limit)) {
rec_page_no = ibuf_rec_get_page_no(rec);
rec_space_id = ibuf_rec_get_space(rec);
if (rec_space_id != first_space_id
|| (rec_page_no / IBUF_MERGE_AREA)
!= (first_page_no / IBUF_MERGE_AREA)) {
break;
}
if (rec_page_no != prev_page_no
|| rec_space_id != prev_space_id) {
n_pages++;
}
prev_page_no = rec_page_no;
prev_space_id = rec_space_id;
rec = page_rec_get_prev(rec);
}
rec = page_rec_get_next(rec);
/* At the loop start there is no prev page; we mark this with a pair
of space id, page no (0, 0) for which there can never be entries in
the insert buffer */
prev_page_no = 0;
prev_space_id = 0;
sum_volumes = 0;
volume_for_page = 0;
while (*n_stored < limit) {
if (page_rec_is_supremum(rec)) {
/* When no more records available, mark this with
another 'impossible' pair of space id, page no */
rec_page_no = 1;
rec_space_id = 0;
} else {
rec_page_no = ibuf_rec_get_page_no(rec);
rec_space_id = ibuf_rec_get_space(rec);
ut_ad(rec_page_no > IBUF_TREE_ROOT_PAGE_NO);
}
#ifdef UNIV_IBUF_DEBUG
ut_a(*n_stored < IBUF_MAX_N_PAGES_MERGED);
#endif
if ((rec_space_id != prev_space_id
|| rec_page_no != prev_page_no)
&& (prev_space_id != 0 || prev_page_no != 0)) {
if ((prev_page_no == first_page_no
&& prev_space_id == first_space_id)
|| contract
|| (volume_for_page
> ((IBUF_MERGE_THRESHOLD - 1)
* 4 * UNIV_PAGE_SIZE
/ IBUF_PAGE_SIZE_PER_FREE_SPACE)
/ IBUF_MERGE_THRESHOLD)) {
space_ids[*n_stored] = prev_space_id;
space_versions[*n_stored]
= fil_space_get_version(prev_space_id);
page_nos[*n_stored] = prev_page_no;
(*n_stored)++;
sum_volumes += volume_for_page;
}
if (rec_space_id != first_space_id
|| rec_page_no / IBUF_MERGE_AREA
!= first_page_no / IBUF_MERGE_AREA) {
break;
}
volume_for_page = 0;
}
if (rec_page_no == 1 && rec_space_id == 0) {
/* Supremum record */
break;
}
rec_volume = ibuf_rec_get_volume(rec);
volume_for_page += rec_volume;
prev_page_no = rec_page_no;
prev_space_id = rec_space_id;
rec = page_rec_get_next(rec);
}
#ifdef UNIV_IBUF_DEBUG
ut_a(*n_stored <= IBUF_MAX_N_PAGES_MERGED);
#endif
#if 0
fprintf(stderr, "Ibuf merge batch %lu pages %lu volume\n",
*n_stored, sum_volumes);
#endif
return(sum_volumes);
}
/*********************************************************************//**
Contracts insert buffer trees by reading pages to the buffer pool.
@return a lower limit for the combined size in bytes of entries which
will be merged from ibuf trees to the pages read, 0 if ibuf is
empty */
static
ulint
ibuf_contract_ext(
/*==============*/
ulint* n_pages,/*!< out: number of pages to which merged */
ibool sync) /*!< in: TRUE if the caller wants to wait for the
issued read with the highest tablespace address
to complete */
{
btr_pcur_t pcur;
ulint page_nos[IBUF_MAX_N_PAGES_MERGED];
ulint space_ids[IBUF_MAX_N_PAGES_MERGED];
ib_int64_t space_versions[IBUF_MAX_N_PAGES_MERGED];
ulint n_stored;
ulint sum_sizes;
mtr_t mtr;
*n_pages = 0;
ut_ad(!ibuf_inside());
mutex_enter(&ibuf_mutex);
if (ibuf->empty) {
ibuf_is_empty:
mutex_exit(&ibuf_mutex);
#if 0 /* TODO */
if (srv_shutdown_state) {
/* If the insert buffer becomes empty during
shutdown, note it in the system tablespace. */
trx_sys_set_ibuf_format(TRX_SYS_IBUF_EMPTY);
}
/* TO DO: call trx_sys_set_ibuf_format() at startup
and whenever ibuf_use is changed to allow buffered
delete-marking or deleting. Never downgrade the
stamped format except when the insert buffer becomes
empty. */
#endif
return(0);
}
mtr_start(&mtr);
ibuf_enter();
/* Open a cursor to a randomly chosen leaf of the tree, at a random
position within the leaf */
btr_pcur_open_at_rnd_pos(ibuf->index, BTR_SEARCH_LEAF, &pcur, &mtr);
ut_ad(page_validate(btr_pcur_get_page(&pcur), ibuf->index));
if (page_get_n_recs(btr_pcur_get_page(&pcur)) == 0) {
/* When the ibuf tree is emptied completely, the last record
is removed using an optimistic delete and ibuf_size_update
is not called, causing ibuf->empty to remain FALSE. If we do
not reset it to TRUE here then database shutdown will hang
in the loop in ibuf_contract_for_n_pages. */
ibuf->empty = TRUE;
ibuf_exit();
mtr_commit(&mtr);
btr_pcur_close(&pcur);
goto ibuf_is_empty;
}
mutex_exit(&ibuf_mutex);
sum_sizes = ibuf_get_merge_page_nos(TRUE, btr_pcur_get_rec(&pcur),
space_ids, space_versions,
page_nos, &n_stored);
#if 0 /* defined UNIV_IBUF_DEBUG */
fprintf(stderr, "Ibuf contract sync %lu pages %lu volume %lu\n",
sync, n_stored, sum_sizes);
#endif
ibuf_exit();
mtr_commit(&mtr);
btr_pcur_close(&pcur);
buf_read_ibuf_merge_pages(sync, space_ids, space_versions, page_nos,
n_stored);
*n_pages = n_stored;
return(sum_sizes + 1);
}
/*********************************************************************//**
Contracts insert buffer trees by reading pages to the buffer pool.
@return a lower limit for the combined size in bytes of entries which
will be merged from ibuf trees to the pages read, 0 if ibuf is
empty */
UNIV_INTERN
ulint
ibuf_contract(
/*==========*/
ibool sync) /*!< in: TRUE if the caller wants to wait for the
issued read with the highest tablespace address
to complete */
{
ulint n_pages;
return(ibuf_contract_ext(&n_pages, sync));
}
/*********************************************************************//**
Contracts insert buffer trees by reading pages to the buffer pool.
@return a lower limit for the combined size in bytes of entries which
will be merged from ibuf trees to the pages read, 0 if ibuf is
empty */
UNIV_INTERN
ulint
ibuf_contract_for_n_pages(
/*======================*/
ibool sync, /*!< in: TRUE if the caller wants to wait for the
issued read with the highest tablespace address
to complete */
ulint n_pages)/*!< in: try to read at least this many pages to
the buffer pool and merge the ibuf contents to
them */
{
ulint sum_bytes = 0;
ulint sum_pages = 0;
ulint n_bytes;
ulint n_pag2;
while (sum_pages < n_pages) {
n_bytes = ibuf_contract_ext(&n_pag2, sync);
if (n_bytes == 0) {
return(sum_bytes);
}
sum_bytes += n_bytes;
sum_pages += n_pag2;
}
return(sum_bytes);
}
/*********************************************************************//**
Contract insert buffer trees after insert if they are too big. */
UNIV_INLINE
void
ibuf_contract_after_insert(
/*=======================*/
ulint entry_size) /*!< in: size of a record which was inserted
into an ibuf tree */
{
ibool sync;
ulint sum_sizes;
ulint size;
mutex_enter(&ibuf_mutex);
if (ibuf->size < ibuf->max_size + IBUF_CONTRACT_ON_INSERT_NON_SYNC) {
mutex_exit(&ibuf_mutex);
return;
}
sync = FALSE;
if (ibuf->size >= ibuf->max_size + IBUF_CONTRACT_ON_INSERT_SYNC) {
sync = TRUE;
}
mutex_exit(&ibuf_mutex);
/* Contract at least entry_size many bytes */
sum_sizes = 0;
size = 1;
while ((size > 0) && (sum_sizes < entry_size)) {
size = ibuf_contract(sync);
sum_sizes += size;
}
}
/*********************************************************************//**
Determine if an insert buffer record has been encountered already.
@return TRUE if a new record, FALSE if possible duplicate */
static
ibool
ibuf_get_volume_buffered_hash(
/*==========================*/
const rec_t* rec, /*!< in: ibuf record in post-4.1 format */
const byte* types, /*!< in: fields */
const byte* data, /*!< in: start of user record data */
ulint comp, /*!< in: 0=ROW_FORMAT=REDUNDANT,
nonzero=ROW_FORMAT=COMPACT */
ulint* hash, /*!< in/out: hash array */
ulint size) /*!< in: number of elements in hash array */
{
ulint len;
ulint fold;
ulint bitmask;
len = ibuf_rec_get_size(rec, types, rec_get_n_fields_old(rec) - 4,
FALSE, comp);
fold = ut_fold_binary(data, len);
hash += (fold / (CHAR_BIT * sizeof *hash)) % size;
bitmask = 1 << (fold % (CHAR_BIT * sizeof *hash));
if (*hash & bitmask) {
return(FALSE);
}
/* We have not seen this record yet. Insert it. */
*hash |= bitmask;
return(TRUE);
}
/*********************************************************************//**
Update the estimate of the number of records on a page, and
get the space taken by merging the buffered record to the index page.
@return size of index record in bytes + an upper limit of the space
taken in the page directory */
static
ulint
ibuf_get_volume_buffered_count(
/*===========================*/
const rec_t* rec, /*!< in: insert buffer record */
ulint* hash, /*!< in/out: hash array */
ulint size, /*!< in: number of elements in hash array */
lint* n_recs) /*!< in/out: estimated number of records
on the page that rec points to */
{
ulint len;
ibuf_op_t ibuf_op;
const byte* types;
ulint n_fields = rec_get_n_fields_old(rec);
ut_ad(ibuf_inside());
ut_ad(n_fields > 4);
n_fields -= 4;
rec_get_nth_field_offs_old(rec, 1, &len);
/* This function is only invoked when buffering new
operations. All pre-4.1 records should have been merged
when the database was started up. */
ut_a(len == 1);
ut_ad(trx_sys_multiple_tablespace_format);
types = rec_get_nth_field_old(rec, 3, &len);
switch (UNIV_EXPECT(len % DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE,
IBUF_REC_INFO_SIZE)) {
default:
ut_error;
case 0:
/* This ROW_TYPE=REDUNDANT record does not include an
operation counter. Exclude it from the *n_recs,
because deletes cannot be buffered if there are
old-style inserts buffered for the page. */
len = ibuf_rec_get_size(rec, types, n_fields, FALSE, 0);
return(len
+ rec_get_converted_extra_size(len, n_fields, 0)
+ page_dir_calc_reserved_space(1));
case 1:
/* This ROW_TYPE=COMPACT record does not include an
operation counter. Exclude it from the *n_recs,
because deletes cannot be buffered if there are
old-style inserts buffered for the page. */
goto get_volume_comp;
case IBUF_REC_INFO_SIZE:
ibuf_op = (ibuf_op_t) types[IBUF_REC_OFFSET_TYPE];
break;
}
switch (ibuf_op) {
case IBUF_OP_INSERT:
/* Inserts can be done by
btr_cur_set_deleted_flag_for_ibuf(). Because
delete-mark and insert operations can be pointing to
the same records, we must not count duplicates. */
case IBUF_OP_DELETE_MARK:
/* There must be a record to delete-mark.
See if this record has been already buffered. */
if (n_recs && ibuf_get_volume_buffered_hash(
rec, types + IBUF_REC_INFO_SIZE,
types + len,
types[IBUF_REC_OFFSET_FLAGS] & IBUF_REC_COMPACT,
hash, size)) {
(*n_recs)++;
}
if (ibuf_op == IBUF_OP_DELETE_MARK) {
/* Setting the delete-mark flag does not
affect the available space on the page. */
return(0);
}
break;
case IBUF_OP_DELETE:
/* A record will be removed from the page. */
if (n_recs) {
(*n_recs)--;
}
/* While deleting a record actually frees up space,
we have to play it safe and pretend that it takes no
additional space (the record might not exist, etc.). */
return(0);
default:
ut_error;
}
ut_ad(ibuf_op == IBUF_OP_INSERT);
get_volume_comp:
{
dtuple_t* entry;
ulint volume;
dict_index_t* dummy_index;
mem_heap_t* heap = mem_heap_create(500);
entry = ibuf_build_entry_from_ibuf_rec(
rec, heap, &dummy_index);
volume = rec_get_converted_size(dummy_index, entry, 0);
ibuf_dummy_index_free(dummy_index);
mem_heap_free(heap);
return(volume + page_dir_calc_reserved_space(1));
}
}
/*********************************************************************//**
Gets an upper limit for the combined size of entries buffered in the insert
buffer for a given page.
@return upper limit for the volume of buffered inserts for the index
page, in bytes; UNIV_PAGE_SIZE, if the entries for the index page span
several pages in the insert buffer */
static
ulint
ibuf_get_volume_buffered(
/*=====================*/
btr_pcur_t* pcur, /*!< in: pcur positioned at a place in an
insert buffer tree where we would insert an
entry for the index page whose number is
page_no, latch mode has to be BTR_MODIFY_PREV
or BTR_MODIFY_TREE */
ulint space, /*!< in: space id */
ulint page_no,/*!< in: page number of an index page */
lint* n_recs, /*!< in/out: minimum number of records on the
page after the buffered changes have been
applied, or NULL to disable the counting */
mtr_t* mtr) /*!< in: mtr */
{
ulint volume;
rec_t* rec;
page_t* page;
ulint prev_page_no;
page_t* prev_page;
ulint next_page_no;
page_t* next_page;
ulint hash_bitmap[128 / sizeof(ulint)]; /* bitmap of buffered recs */
ut_a(trx_sys_multiple_tablespace_format);
ut_ad((pcur->latch_mode == BTR_MODIFY_PREV)
|| (pcur->latch_mode == BTR_MODIFY_TREE));
/* Count the volume of inserts earlier in the alphabetical order than
pcur */
volume = 0;
if (n_recs) {
memset(hash_bitmap, 0, sizeof hash_bitmap);
}
rec = btr_pcur_get_rec(pcur);
page = page_align(rec);
ut_ad(page_validate(page, ibuf->index));
if (page_rec_is_supremum(rec)) {
rec = page_rec_get_prev(rec);
}
for (;;) {
if (page_rec_is_infimum(rec)) {
break;
}
if (page_no != ibuf_rec_get_page_no(rec)
|| space != ibuf_rec_get_space(rec)) {
goto count_later;
}
volume += ibuf_get_volume_buffered_count(
rec, hash_bitmap, UT_ARR_SIZE(hash_bitmap), n_recs);
rec = page_rec_get_prev(rec);
ut_ad(page_align(rec) == page);
}
/* Look at the previous page */
prev_page_no = btr_page_get_prev(page, mtr);
if (prev_page_no == FIL_NULL) {
goto count_later;
}
{
buf_block_t* block;
block = buf_page_get(
IBUF_SPACE_ID, 0, prev_page_no, RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
prev_page = buf_block_get_frame(block);
ut_ad(page_validate(prev_page, ibuf->index));
}
#ifdef UNIV_BTR_DEBUG
ut_a(btr_page_get_next(prev_page, mtr)
== page_get_page_no(page));
#endif /* UNIV_BTR_DEBUG */
rec = page_get_supremum_rec(prev_page);
rec = page_rec_get_prev(rec);
for (;;) {
if (page_rec_is_infimum(rec)) {
/* We cannot go to yet a previous page, because we
do not have the x-latch on it, and cannot acquire one
because of the latching order: we have to give up */
return(UNIV_PAGE_SIZE);
}
if (page_no != ibuf_rec_get_page_no(rec)
|| space != ibuf_rec_get_space(rec)) {
goto count_later;
}
volume += ibuf_get_volume_buffered_count(
rec, hash_bitmap, UT_ARR_SIZE(hash_bitmap), n_recs);
rec = page_rec_get_prev(rec);
ut_ad(page_align(rec) == prev_page);
}
count_later:
rec = btr_pcur_get_rec(pcur);
if (!page_rec_is_supremum(rec)) {
rec = page_rec_get_next(rec);
}
for (;;) {
if (page_rec_is_supremum(rec)) {
break;
}
if (page_no != ibuf_rec_get_page_no(rec)
|| space != ibuf_rec_get_space(rec)) {
return(volume);
}
volume += ibuf_get_volume_buffered_count(
rec, hash_bitmap, UT_ARR_SIZE(hash_bitmap), n_recs);
rec = page_rec_get_next(rec);
}
/* Look at the next page */
next_page_no = btr_page_get_next(page, mtr);
if (next_page_no == FIL_NULL) {
return(volume);
}
{
buf_block_t* block;
block = buf_page_get(
IBUF_SPACE_ID, 0, next_page_no, RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
next_page = buf_block_get_frame(block);
ut_ad(page_validate(next_page, ibuf->index));
}
#ifdef UNIV_BTR_DEBUG
ut_a(btr_page_get_prev(next_page, mtr) == page_get_page_no(page));
#endif /* UNIV_BTR_DEBUG */
rec = page_get_infimum_rec(next_page);
rec = page_rec_get_next(rec);
for (;;) {
if (page_rec_is_supremum(rec)) {
/* We give up */
return(UNIV_PAGE_SIZE);
}
if (page_no != ibuf_rec_get_page_no(rec)
|| space != ibuf_rec_get_space(rec)) {
return(volume);
}
volume += ibuf_get_volume_buffered_count(
rec, hash_bitmap, UT_ARR_SIZE(hash_bitmap), n_recs);
rec = page_rec_get_next(rec);
ut_ad(page_align(rec) == next_page);
}
}
/*********************************************************************//**
Reads the biggest tablespace id from the high end of the insert buffer
tree and updates the counter in fil_system. */
UNIV_INTERN
void
ibuf_update_max_tablespace_id(void)
/*===============================*/
{
ulint max_space_id;
const rec_t* rec;
const byte* field;
ulint len;
btr_pcur_t pcur;
mtr_t mtr;
ut_a(!dict_table_is_comp(ibuf->index->table));
ibuf_enter();
mtr_start(&mtr);
btr_pcur_open_at_index_side(
FALSE, ibuf->index, BTR_SEARCH_LEAF, &pcur, TRUE, &mtr);
ut_ad(page_validate(btr_pcur_get_page(&pcur), ibuf->index));
btr_pcur_move_to_prev(&pcur, &mtr);
if (btr_pcur_is_before_first_on_page(&pcur)) {
/* The tree is empty */
max_space_id = 0;
} else {
rec = btr_pcur_get_rec(&pcur);
field = rec_get_nth_field_old(rec, 0, &len);
ut_a(len == 4);
max_space_id = mach_read_from_4(field);
}
mtr_commit(&mtr);
ibuf_exit();
/* printf("Maximum space id in insert buffer %lu\n", max_space_id); */
fil_set_max_space_id_if_bigger(max_space_id);
}
/****************************************************************//**
Helper function for ibuf_set_entry_counter. Checks if rec is for (space,
page_no), and if so, reads counter value from it and returns that + 1.
Otherwise, returns 0.
@return new counter value, or 0 */
static
ulint
ibuf_get_entry_counter_low(
/*=======================*/
const rec_t* rec, /*!< in: insert buffer record */
ulint space, /*!< in: space id */
ulint page_no) /*!< in: page number */
{
ulint counter;
const byte* field;
ulint len;
ut_ad(ibuf_inside());
ut_ad(rec_get_n_fields_old(rec) > 2);
field = rec_get_nth_field_old(rec, 1, &len);
if (UNIV_UNLIKELY(len != 1)) {
/* pre-4.1 format */
ut_a(trx_doublewrite_must_reset_space_ids);
ut_a(!trx_sys_multiple_tablespace_format);
return(ULINT_UNDEFINED);
}
ut_a(trx_sys_multiple_tablespace_format);
/* Check the tablespace identifier. */
field = rec_get_nth_field_old(rec, 0, &len);
ut_a(len == 4);
if (mach_read_from_4(field) != space) {
return(0);
}
/* Check the page offset. */
field = rec_get_nth_field_old(rec, 2, &len);
ut_a(len == 4);
if (mach_read_from_4(field) != page_no) {
return(0);
}
/* Check if the record contains a counter field. */
field = rec_get_nth_field_old(rec, 3, &len);
switch (len % DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE) {
default:
ut_error;
case 0: /* ROW_FORMAT=REDUNDANT */
case 1: /* ROW_FORMAT=COMPACT */
return(ULINT_UNDEFINED);
case IBUF_REC_INFO_SIZE:
counter = mach_read_from_2(field + IBUF_REC_OFFSET_COUNTER);
ut_a(counter < 0xFFFF);
return(counter + 1);
}
}
/****************************************************************//**
Set the counter field in entry to the correct value based on the current
last record in ibuf for (space, page_no).
@return FALSE if we should abort this insertion to ibuf */
static
ibool
ibuf_set_entry_counter(
/*===================*/
dtuple_t* entry, /*!< in/out: entry to patch */
ulint space, /*!< in: space id of entry */
ulint page_no, /*!< in: page number of entry */
btr_pcur_t* pcur, /*!< in: pcur positioned on the record
found by btr_pcur_open(.., entry,
PAGE_CUR_LE, ..., pcur, ...) */
ibool is_optimistic, /*!< in: is this an optimistic insert */
mtr_t* mtr) /*!< in: mtr */
{
ulint counter;
dfield_t* field;
byte* data;
/* pcur points to either a user rec or to a page's infimum record. */
ut_ad(page_validate(btr_pcur_get_page(pcur), ibuf->index));
if (btr_pcur_is_on_user_rec(pcur)) {
counter = ibuf_get_entry_counter_low(
btr_pcur_get_rec(pcur), space, page_no);
if (UNIV_UNLIKELY(counter == ULINT_UNDEFINED)) {
/* The record lacks a counter field.
Such old records must be merged before
new records can be buffered. */
return(FALSE);
}
} else if (btr_pcur_is_before_first_in_tree(pcur, mtr)) {
/* Ibuf tree is either completely empty, or the insert
position is at the very first record of a non-empty tree. In
either case we have no previous records for (space,
page_no). */
counter = 0;
} else if (btr_pcur_is_before_first_on_page(pcur)) {
btr_cur_t* cursor = btr_pcur_get_btr_cur(pcur);
if (cursor->low_match < 3) {
/* If low_match < 3, we know that the father node
pointer did not contain the searched for (space,
page_no), which means that the search ended on the
right page regardless of the counter value, and
since we're at the infimum record, there are no
existing records. */
counter = 0;
} else {
rec_t* rec;
const page_t* page;
buf_block_t* block;
page_t* prev_page;
ulint prev_page_no;
ut_a(cursor->ibuf_cnt != ULINT_UNDEFINED);
page = btr_pcur_get_page(pcur);
prev_page_no = btr_page_get_prev(page, mtr);
ut_a(prev_page_no != FIL_NULL);
block = buf_page_get(
IBUF_SPACE_ID, 0, prev_page_no,
RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
prev_page = buf_block_get_frame(block);
rec = page_rec_get_prev(
page_get_supremum_rec(prev_page));
ut_ad(page_rec_is_user_rec(rec));
counter = ibuf_get_entry_counter_low(
rec, space, page_no);
if (UNIV_UNLIKELY(counter == ULINT_UNDEFINED)) {
/* The record lacks a counter field.
Such old records must be merged before
new records can be buffered. */
return(FALSE);
}
if (counter < cursor->ibuf_cnt) {
/* Search ended on the wrong page. */
if (is_optimistic) {
/* In an optimistic insert, we can
shift the insert position to the left
page, since it only needs an X-latch
on the page itself, which the
original search acquired for us. */
btr_cur_position(
ibuf->index, rec, block,
btr_pcur_get_btr_cur(pcur));
} else {
/* We can't shift the insert
position to the left page in a
pessimistic insert since it would
require an X-latch on the left
page's left page, so we have to
abort. */
return(FALSE);
}
} else {
/* The counter field in the father node is
the same as we would insert; we don't know
whether the insert should go to this page or
the left page (the later fields can differ),
so refuse the insert. */
return(FALSE);
}
}
} else {
/* The cursor is not positioned at or before a user record. */
return(FALSE);
}
/* Patch counter value in already built entry. */
field = dtuple_get_nth_field(entry, 3);
data = dfield_get_data(field);
mach_write_to_2(data + IBUF_REC_OFFSET_COUNTER, counter);
return(TRUE);
}
/*********************************************************************//**
Buffer an operation in the insert/delete buffer, instead of doing it
directly to the disk page, if this is possible.
@return DB_SUCCESS, DB_FAIL, DB_STRONG_FAIL */
static
ulint
ibuf_insert_low(
/*============*/
ulint mode, /*!< in: BTR_MODIFY_PREV or BTR_MODIFY_TREE */
ibuf_op_t op, /*!< in: operation type */
ibool no_counter,
/*!< in: TRUE=use 5.0.3 format;
FALSE=allow delete buffering */
const dtuple_t* entry, /*!< in: index entry to insert */
ulint entry_size,
/*!< in: rec_get_converted_size(index, entry) */
dict_index_t* index, /*!< in: index where to insert; must not be
unique or clustered */
ulint space, /*!< in: space id where to insert */
ulint zip_size,/*!< in: compressed page size in bytes, or 0 */
ulint page_no,/*!< in: page number where to insert */
que_thr_t* thr) /*!< in: query thread */
{
big_rec_t* dummy_big_rec;
btr_pcur_t pcur;
btr_cur_t* cursor;
dtuple_t* ibuf_entry;
mem_heap_t* heap;
ulint buffered;
lint min_n_recs;
rec_t* ins_rec;
ibool old_bit_value;
page_t* bitmap_page;
page_t* root;
ulint err;
ibool do_merge;
ulint space_ids[IBUF_MAX_N_PAGES_MERGED];
ib_int64_t space_versions[IBUF_MAX_N_PAGES_MERGED];
ulint page_nos[IBUF_MAX_N_PAGES_MERGED];
ulint n_stored;
mtr_t mtr;
mtr_t bitmap_mtr;
ut_a(!dict_index_is_clust(index));
ut_ad(dtuple_check_typed(entry));
ut_ad(ut_is_2pow(zip_size));
ut_ad(!no_counter || op == IBUF_OP_INSERT);
ut_a(op < IBUF_OP_COUNT);
ut_a(trx_sys_multiple_tablespace_format);
do_merge = FALSE;
mutex_enter(&ibuf_mutex);
if (ibuf->size >= ibuf->max_size + IBUF_CONTRACT_DO_NOT_INSERT) {
/* Insert buffer is now too big, contract it but do not try
to insert */
mutex_exit(&ibuf_mutex);
#ifdef UNIV_IBUF_DEBUG
fputs("Ibuf too big\n", stderr);
#endif
/* Use synchronous contract (== TRUE) */
ibuf_contract(TRUE);
return(DB_STRONG_FAIL);
}
mutex_exit(&ibuf_mutex);
if (mode == BTR_MODIFY_TREE) {
mutex_enter(&ibuf_pessimistic_insert_mutex);
ibuf_enter();
mutex_enter(&ibuf_mutex);
while (!ibuf_data_enough_free_for_insert()) {
mutex_exit(&ibuf_mutex);
ibuf_exit();
mutex_exit(&ibuf_pessimistic_insert_mutex);
err = ibuf_add_free_page();
if (err == DB_STRONG_FAIL) {
return(err);
}
mutex_enter(&ibuf_pessimistic_insert_mutex);
ibuf_enter();
mutex_enter(&ibuf_mutex);
}
} else {
ibuf_enter();
}
heap = mem_heap_create(512);
/* Build the entry which contains the space id and the page number
as the first fields and the type information for other fields, and
which will be inserted to the insert buffer. Using a counter value
of 0xFFFF we find the last record for (space, page_no), from which
we can then read the counter value N and use N + 1 in the record we
insert. (We patch the ibuf_entry's counter field to the correct
value just before actually inserting the entry.) */
ibuf_entry = ibuf_entry_build(
op, index, entry, space, page_no,
no_counter ? ULINT_UNDEFINED : 0xFFFF, heap);
/* Open a cursor to the insert buffer tree to calculate if we can add
the new entry to it without exceeding the free space limit for the
page. */
mtr_start(&mtr);
btr_pcur_open(ibuf->index, ibuf_entry, PAGE_CUR_LE, mode, &pcur, &mtr);
ut_ad(page_validate(btr_pcur_get_page(&pcur), ibuf->index));
/* Find out the volume of already buffered inserts for the same index
page */
min_n_recs = 0;
buffered = ibuf_get_volume_buffered(&pcur, space, page_no,
op == IBUF_OP_DELETE
? &min_n_recs
: NULL, &mtr);
if (op == IBUF_OP_DELETE
&& (min_n_recs < 2
|| buf_pool_watch_occurred(space, page_no))) {
/* The page could become empty after the record is
deleted, or the page has been read in to the buffer
pool. Refuse to buffer the operation. */
/* The buffer pool watch is needed for IBUF_OP_DELETE
because of latching order considerations. We can
check buf_pool_watch_occurred() only after latching
the insert buffer B-tree pages that contain buffered
changes for the page. We never buffer IBUF_OP_DELETE,
unless some IBUF_OP_INSERT or IBUF_OP_DELETE_MARK have
been previously buffered for the page. Because there
are buffered operations for the page, the insert
buffer B-tree page latches held by mtr will guarantee
that no changes for the user page will be merged
before mtr_commit(&mtr). We must not mtr_commit(&mtr)
until after the IBUF_OP_DELETE has been buffered. */
err = DB_STRONG_FAIL;
goto function_exit;
}
/* After this point, the page could still be loaded to the
buffer pool, but we do not have to care about it, since we are
holding a latch on the insert buffer leaf page that contains
buffered changes for (space, page_no). If the page enters the
buffer pool, buf_page_io_complete() for (space, page_no) will
have to acquire a latch on the same insert buffer leaf page,
which it cannot do until we have buffered the IBUF_OP_DELETE
and done mtr_commit(&mtr) to release the latch. */
#ifdef UNIV_IBUF_COUNT_DEBUG
ut_a((buffered == 0) || ibuf_count_get(space, page_no));
#endif
mtr_start(&bitmap_mtr);
bitmap_page = ibuf_bitmap_get_map_page(space, page_no,
zip_size, &bitmap_mtr);
/* We check if the index page is suitable for buffered entries */
if (buf_page_peek(space, page_no)
|| lock_rec_expl_exist_on_page(space, page_no)) {
goto bitmap_fail;
}
if (op == IBUF_OP_INSERT) {
ulint bits = ibuf_bitmap_page_get_bits(
bitmap_page, page_no, zip_size, IBUF_BITMAP_FREE,
&bitmap_mtr);
if (buffered + entry_size + page_dir_calc_reserved_space(1)
> ibuf_index_page_calc_free_from_bits(zip_size, bits)) {
/* Release the bitmap page latch early. */
mtr_commit(&bitmap_mtr);
/* It may not fit */
do_merge = TRUE;
ibuf_get_merge_page_nos(
FALSE, btr_pcur_get_rec(&pcur),
space_ids, space_versions,
page_nos, &n_stored);
err = DB_STRONG_FAIL;
goto function_exit;
}
}
/* Patch correct counter value to the entry to insert. This can
change the insert position, which can result in the need to abort in
some cases. */
if (!no_counter
&& !ibuf_set_entry_counter(ibuf_entry, space, page_no, &pcur,
mode == BTR_MODIFY_PREV, &mtr)) {
bitmap_fail:
err = DB_STRONG_FAIL;
mtr_commit(&bitmap_mtr);
goto function_exit;
}
/* Set the bitmap bit denoting that the insert buffer contains
buffered entries for this index page, if the bit is not set yet */
old_bit_value = ibuf_bitmap_page_get_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_BUFFERED, &bitmap_mtr);
if (!old_bit_value) {
ibuf_bitmap_page_set_bits(bitmap_page, page_no, zip_size,
IBUF_BITMAP_BUFFERED, TRUE,
&bitmap_mtr);
}
mtr_commit(&bitmap_mtr);
cursor = btr_pcur_get_btr_cur(&pcur);
if (mode == BTR_MODIFY_PREV) {
err = btr_cur_optimistic_insert(BTR_NO_LOCKING_FLAG, cursor,
ibuf_entry, &ins_rec,
&dummy_big_rec, 0, thr, &mtr);
if (err == DB_SUCCESS && op != IBUF_OP_DELETE) {
/* Update the page max trx id field */
page_update_max_trx_id(btr_cur_get_block(cursor), NULL,
thr_get_trx(thr)->id, &mtr);
}
} else {
ut_ad(mode == BTR_MODIFY_TREE);
/* We acquire an x-latch to the root page before the insert,
because a pessimistic insert releases the tree x-latch,
which would cause the x-latching of the root after that to
break the latching order. */
root = ibuf_tree_root_get(&mtr);
err = btr_cur_pessimistic_insert(BTR_NO_LOCKING_FLAG
| BTR_NO_UNDO_LOG_FLAG,
cursor,
ibuf_entry, &ins_rec,
&dummy_big_rec, 0, thr, &mtr);
if (err == DB_SUCCESS && op != IBUF_OP_DELETE) {
/* Update the page max trx id field */
page_update_max_trx_id(btr_cur_get_block(cursor), NULL,
thr_get_trx(thr)->id, &mtr);
}
ibuf_size_update(root, &mtr);
}
function_exit:
#ifdef UNIV_IBUF_COUNT_DEBUG
if (err == DB_SUCCESS) {
fprintf(stderr,
"Incrementing ibuf count of space %lu page %lu\n"
"from %lu by 1\n", space, page_no,
ibuf_count_get(space, page_no));
ibuf_count_set(space, page_no,
ibuf_count_get(space, page_no) + 1);
}
#endif
if (mode == BTR_MODIFY_TREE) {
mutex_exit(&ibuf_mutex);
mutex_exit(&ibuf_pessimistic_insert_mutex);
}
mtr_commit(&mtr);
btr_pcur_close(&pcur);
ibuf_exit();
mem_heap_free(heap);
if (err == DB_SUCCESS) {
mutex_enter(&ibuf_mutex);
ibuf->empty = FALSE;
mutex_exit(&ibuf_mutex);
if (mode == BTR_MODIFY_TREE) {
ibuf_contract_after_insert(entry_size);
}
}
if (do_merge) {
#ifdef UNIV_IBUF_DEBUG
ut_a(n_stored <= IBUF_MAX_N_PAGES_MERGED);
#endif
buf_read_ibuf_merge_pages(FALSE, space_ids, space_versions,
page_nos, n_stored);
}
return(err);
}
/*********************************************************************//**
Buffer an operation in the insert/delete buffer, instead of doing it
directly to the disk page, if this is possible. Does not do it if the index
is clustered or unique.
@return TRUE if success */
UNIV_INTERN
ibool
ibuf_insert(
/*========*/
ibuf_op_t op, /*!< in: operation type */
const dtuple_t* entry, /*!< in: index entry to insert */
dict_index_t* index, /*!< in: index where to insert */
ulint space, /*!< in: space id where to insert */
ulint zip_size,/*!< in: compressed page size in bytes, or 0 */
ulint page_no,/*!< in: page number where to insert */
que_thr_t* thr) /*!< in: query thread */
{
ulint err;
ulint entry_size;
ibool no_counter;
/* Read the settable global variable ibuf_use only once in
this function, so that we will have a consistent view of it. */
ibuf_use_t use = ibuf_use;
ut_a(trx_sys_multiple_tablespace_format);
ut_ad(dtuple_check_typed(entry));
ut_ad(ut_is_2pow(zip_size));
ut_a(!dict_index_is_clust(index));
no_counter = use <= IBUF_USE_INSERT;
switch (op) {
case IBUF_OP_INSERT:
switch (use) {
case IBUF_USE_NONE:
case IBUF_USE_DELETE:
case IBUF_USE_DELETE_MARK:
return(FALSE);
case IBUF_USE_INSERT:
case IBUF_USE_INSERT_DELETE_MARK:
case IBUF_USE_ALL:
goto check_watch;
case IBUF_USE_COUNT:
break;
}
break;
case IBUF_OP_DELETE_MARK:
switch (use) {
case IBUF_USE_NONE:
case IBUF_USE_INSERT:
return(FALSE);
case IBUF_USE_DELETE_MARK:
case IBUF_USE_DELETE:
case IBUF_USE_INSERT_DELETE_MARK:
case IBUF_USE_ALL:
ut_ad(!no_counter);
goto check_watch;
case IBUF_USE_COUNT:
break;
}
break;
case IBUF_OP_DELETE:
switch (use) {
case IBUF_USE_NONE:
case IBUF_USE_INSERT:
case IBUF_USE_INSERT_DELETE_MARK:
return(FALSE);
case IBUF_USE_DELETE_MARK:
case IBUF_USE_DELETE:
case IBUF_USE_ALL:
ut_ad(!no_counter);
goto skip_watch;
case IBUF_USE_COUNT:
break;
}
break;
case IBUF_OP_COUNT:
break;
}
/* unknown op or use */
ut_error;
check_watch:
/* If a thread attempts to buffer an insert on a page while a
purge is in progress on the same page, the purge must not be
buffered, because it could remove a record that was
re-inserted later. For simplicity, we block the buffering of
all operations on a page that has a purge pending.
We do not check this in the IBUF_OP_DELETE case, because that
would always trigger the buffer pool watch during purge and
thus prevent the buffering of delete operations. We assume
that the issuer of IBUF_OP_DELETE has called
buf_pool_watch_set(space, page_no). */
{
buf_page_t* bpage;
ulint fold = buf_page_address_fold(space, page_no);
buf_pool_mutex_enter();
bpage = buf_page_hash_get_low(space, page_no, fold);
buf_pool_mutex_exit();
if (UNIV_LIKELY_NULL(bpage)) {
/* A buffer pool watch has been set or the
page has been read into the buffer pool.
Do not buffer the request. If a purge operation
is being buffered, have this request executed
directly on the page in the buffer pool after the
buffered entries for this page have been merged. */
return(FALSE);
}
}
skip_watch:
entry_size = rec_get_converted_size(index, entry, 0);
if (entry_size
>= page_get_free_space_of_empty(dict_table_is_comp(index->table))
/ 2) {
return(FALSE);
}
err = ibuf_insert_low(BTR_MODIFY_PREV, op, no_counter,
entry, entry_size,
index, space, zip_size, page_no, thr);
if (err == DB_FAIL) {
err = ibuf_insert_low(BTR_MODIFY_TREE, op, no_counter,
entry, entry_size,
index, space, zip_size, page_no, thr);
}
if (err == DB_SUCCESS) {
#ifdef UNIV_IBUF_DEBUG
/* fprintf(stderr, "Ibuf insert for page no %lu of index %s\n",
page_no, index->name); */
#endif
return(TRUE);
} else {
ut_a(err == DB_STRONG_FAIL);
return(FALSE);
}
}
/********************************************************************//**
During merge, inserts to an index page a secondary index entry extracted
from the insert buffer. */
static
void
ibuf_insert_to_index_page(
/*======================*/
dtuple_t* entry, /*!< in: buffered entry to insert */
buf_block_t* block, /*!< in/out: index page where the buffered entry
should be placed */
dict_index_t* index, /*!< in: record descriptor */
mtr_t* mtr) /*!< in: mtr */
{
page_cur_t page_cur;
ulint low_match;
page_t* page = buf_block_get_frame(block);
rec_t* rec;
page_t* bitmap_page;
ulint old_bits;
ut_ad(ibuf_inside());
ut_ad(dtuple_check_typed(entry));
if (UNIV_UNLIKELY(dict_table_is_comp(index->table)
!= (ibool)!!page_is_comp(page))) {
fputs("InnoDB: Trying to insert a record from"
" the insert buffer to an index page\n"
"InnoDB: but the 'compact' flag does not match!\n",
stderr);
goto dump;
}
rec = page_rec_get_next(page_get_infimum_rec(page));
if (page_rec_is_supremum(rec)) {
/* Empty pages can result from buffered delete operations.
The first record from the free list can be used to find the
father node. */
rec = page_header_get_ptr(page, PAGE_FREE);
if (UNIV_UNLIKELY(rec == NULL)) {
fputs("InnoDB: Trying to insert a record from"
" the insert buffer to an index page\n"
"InnoDB: but the index page is empty!\n",
stderr);
goto dump;
}
}
if (UNIV_UNLIKELY(rec_get_n_fields(rec, index)
!= dtuple_get_n_fields(entry))) {
fputs("InnoDB: Trying to insert a record from"
" the insert buffer to an index page\n"
"InnoDB: but the number of fields does not match!\n",
stderr);
dump:
buf_page_print(page, 0);
dtuple_print(stderr, entry);
fputs("InnoDB: The table where where"
" this index record belongs\n"
"InnoDB: is now probably corrupt."
" Please run CHECK TABLE on\n"
"InnoDB: your tables.\n"
"InnoDB: Submit a detailed bug report to"
" http://bugs.mysql.com!\n", stderr);
return;
}
low_match = page_cur_search(block, index, entry,
PAGE_CUR_LE, &page_cur);
if (low_match == dtuple_get_n_fields(entry)) {
page_zip_des_t* page_zip;
rec = page_cur_get_rec(&page_cur);
page_zip = buf_block_get_page_zip(block);
btr_cur_set_deleted_flag_for_ibuf(rec, page_zip, FALSE, mtr);
} else {
rec = page_cur_tuple_insert(&page_cur, entry, index, 0, mtr);
if (UNIV_LIKELY(rec != NULL)) {
return;
}
/* If the record did not fit, reorganize */
btr_page_reorganize(block, index, mtr);
page_cur_search(block, index, entry, PAGE_CUR_LE, &page_cur);
/* This time the record must fit */
if (UNIV_UNLIKELY
(!page_cur_tuple_insert(&page_cur, entry, index,
0, mtr))) {
ulint space;
ulint page_no;
ulint zip_size;
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Error: Insert buffer insert"
" fails; page free %lu,"
" dtuple size %lu\n",
(ulong) page_get_max_insert_size(
page, 1),
(ulong) rec_get_converted_size(
index, entry, 0));
fputs("InnoDB: Cannot insert index record ",
stderr);
dtuple_print(stderr, entry);
fputs("\nInnoDB: The table where"
" this index record belongs\n"
"InnoDB: is now probably corrupt."
" Please run CHECK TABLE on\n"
"InnoDB: that table.\n", stderr);
space = page_get_space_id(page);
zip_size = buf_block_get_zip_size(block);
page_no = page_get_page_no(page);
bitmap_page = ibuf_bitmap_get_map_page(
space, page_no, zip_size, mtr);
old_bits = ibuf_bitmap_page_get_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, mtr);
fprintf(stderr,
"InnoDB: space %lu, page %lu,"
" zip_size %lu, bitmap bits %lu\n",
(ulong) space, (ulong) page_no,
(ulong) zip_size, (ulong) old_bits);
fputs("InnoDB: Submit a detailed bug report"
" to http://bugs.mysql.com\n", stderr);
}
}
}
/****************************************************************//**
During merge, sets the delete mark on a record for a secondary index
entry. */
static
void
ibuf_set_del_mark(
/*==============*/
const dtuple_t* entry, /*!< in: entry */
buf_block_t* block, /*!< in/out: block */
const dict_index_t* index, /*!< in: record descriptor */
mtr_t* mtr) /*!< in: mtr */
{
page_cur_t page_cur;
ulint low_match;
ut_ad(ibuf_inside());
ut_ad(dtuple_check_typed(entry));
low_match = page_cur_search(
block, index, entry, PAGE_CUR_LE, &page_cur);
if (low_match == dtuple_get_n_fields(entry)) {
rec_t* rec;
page_zip_des_t* page_zip;
rec = page_cur_get_rec(&page_cur);
page_zip = page_cur_get_page_zip(&page_cur);
btr_cur_set_deleted_flag_for_ibuf(rec, page_zip, TRUE, mtr);
} else {
/* This can happen benignly in some situations. */
}
}
/****************************************************************//**
During merge, delete a record for a secondary index entry. */
static
void
ibuf_delete(
/*========*/
const dtuple_t* entry, /*!< in: entry */
buf_block_t* block, /*!< in/out: block */
dict_index_t* index, /*!< in: record descriptor */
mtr_t* mtr) /*!< in/out: mtr; must be committed
before latching any further pages */
{
page_cur_t page_cur;
ulint low_match;
ut_ad(ibuf_inside());
ut_ad(dtuple_check_typed(entry));
low_match = page_cur_search(
block, index, entry, PAGE_CUR_LE, &page_cur);
if (low_match == dtuple_get_n_fields(entry)) {
page_zip_des_t* page_zip= buf_block_get_page_zip(block);
page_t* page = buf_block_get_frame(block);
rec_t* rec = page_cur_get_rec(&page_cur);
/* TODO: the below should probably be a separate function,
it's a bastardized version of btr_cur_optimistic_delete. */
ulint offsets_[REC_OFFS_NORMAL_SIZE];
ulint* offsets = offsets_;
mem_heap_t* heap = NULL;
ulint max_ins_size;
rec_offs_init(offsets_);
offsets = rec_get_offsets(
rec, index, offsets, ULINT_UNDEFINED, &heap);
/* Refuse to delete the last record. */
ut_a(page_get_n_recs(page) > 1);
/* The record should have been marked for deletion. */
ut_ad(REC_INFO_DELETED_FLAG
& rec_get_info_bits(rec, page_is_comp(page)));
lock_update_delete(block, rec);
if (!page_zip) {
max_ins_size
= page_get_max_insert_size_after_reorganize(
page, 1);
}
#ifdef UNIV_ZIP_DEBUG
ut_a(!page_zip || page_zip_validate(page_zip, page));
#endif /* UNIV_ZIP_DEBUG */
page_cur_delete_rec(&page_cur, index, offsets, mtr);
#ifdef UNIV_ZIP_DEBUG
ut_a(!page_zip || page_zip_validate(page_zip, page));
#endif /* UNIV_ZIP_DEBUG */
if (page_zip) {
ibuf_update_free_bits_zip(block, mtr);
} else {
ibuf_update_free_bits_low(block, max_ins_size, mtr);
}
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
} else {
/* This can happen benignly in some situations: either when
we crashed at just the right time, or on database startup
when we redo some old log entries (due to worse stored
position granularity on disk than in memory). */
}
}
/*********************************************************************//**
Restores insert buffer tree cursor position
@return TRUE if the position was restored; FALSE if not */
static __attribute__((nonnull))
ibool
ibuf_restore_pos(
/*=============*/
ulint space, /*!< in: space id */
ulint page_no,/*!< in: index page number where the record
should belong */
const dtuple_t* search_tuple,
/*!< in: search tuple for entries of page_no */
ulint mode, /*!< in: BTR_MODIFY_LEAF or BTR_MODIFY_TREE */
btr_pcur_t* pcur, /*!< in/out: persistent cursor whose
position is to be restored */
mtr_t* mtr) /*!< in/out: mini-transaction */
{
ut_ad(mode == BTR_MODIFY_LEAF || mode == BTR_MODIFY_TREE);
if (btr_pcur_restore_position(mode, pcur, mtr)) {
return(TRUE);
}
if (fil_space_get_flags(space) == ULINT_UNDEFINED) {
/* The tablespace has been dropped. It is possible
that another thread has deleted the insert buffer
entry. Do not complain. */
btr_pcur_commit_specify_mtr(pcur, mtr);
} else {
fprintf(stderr,
"InnoDB: ERROR: Submit the output to"
" http://bugs.mysql.com\n"
"InnoDB: ibuf cursor restoration fails!\n"
"InnoDB: ibuf record inserted to page %lu:%lu\n",
(ulong) space, (ulong) page_no);
fflush(stderr);
rec_print_old(stderr, btr_pcur_get_rec(pcur));
rec_print_old(stderr, pcur->old_rec);
dtuple_print(stderr, search_tuple);
rec_print_old(stderr,
page_rec_get_next(btr_pcur_get_rec(pcur)));
fflush(stderr);
btr_pcur_commit_specify_mtr(pcur, mtr);
fputs("InnoDB: Validating insert buffer tree:\n", stderr);
if (!btr_validate_index(ibuf->index, NULL)) {
ut_error;
}
fprintf(stderr, "InnoDB: ibuf tree ok\n");
fflush(stderr);
}
return(FALSE);
}
/*********************************************************************//**
Deletes from ibuf the record on which pcur is positioned. If we have to
resort to a pessimistic delete, this function commits mtr and closes
the cursor.
@return TRUE if mtr was committed and pcur closed in this operation */
static
ibool
ibuf_delete_rec(
/*============*/
ulint space, /*!< in: space id */
ulint page_no,/*!< in: index page number where the record
should belong */
btr_pcur_t* pcur, /*!< in: pcur positioned on the record to
delete, having latch mode BTR_MODIFY_LEAF */
const dtuple_t* search_tuple,
/*!< in: search tuple for entries of page_no */
mtr_t* mtr) /*!< in: mtr */
{
ibool success;
page_t* root;
ulint err;
ut_ad(ibuf_inside());
ut_ad(page_rec_is_user_rec(btr_pcur_get_rec(pcur)));
ut_ad(ibuf_rec_get_page_no(btr_pcur_get_rec(pcur)) == page_no);
ut_ad(ibuf_rec_get_space(btr_pcur_get_rec(pcur)) == space);
success = btr_cur_optimistic_delete(btr_pcur_get_btr_cur(pcur), mtr);
if (success) {
#ifdef UNIV_IBUF_COUNT_DEBUG
fprintf(stderr,
"Decrementing ibuf count of space %lu page %lu\n"
"from %lu by 1\n", space, page_no,
ibuf_count_get(space, page_no));
ibuf_count_set(space, page_no,
ibuf_count_get(space, page_no) - 1);
#endif
return(FALSE);
}
ut_ad(page_rec_is_user_rec(btr_pcur_get_rec(pcur)));
ut_ad(ibuf_rec_get_page_no(btr_pcur_get_rec(pcur)) == page_no);
ut_ad(ibuf_rec_get_space(btr_pcur_get_rec(pcur)) == space);
/* We have to resort to a pessimistic delete from ibuf */
btr_pcur_store_position(pcur, mtr);
btr_pcur_commit_specify_mtr(pcur, mtr);
mutex_enter(&ibuf_mutex);
mtr_start(mtr);
if (!ibuf_restore_pos(space, page_no, search_tuple,
BTR_MODIFY_TREE, pcur, mtr)) {
goto func_exit;
}
root = ibuf_tree_root_get(mtr);
btr_cur_pessimistic_delete(&err, TRUE, btr_pcur_get_btr_cur(pcur),
RB_NONE, mtr);
ut_a(err == DB_SUCCESS);
#ifdef UNIV_IBUF_COUNT_DEBUG
ibuf_count_set(space, page_no, ibuf_count_get(space, page_no) - 1);
#endif
ibuf_size_update(root, mtr);
btr_pcur_commit_specify_mtr(pcur, mtr);
func_exit:
btr_pcur_close(pcur);
mutex_exit(&ibuf_mutex);
return(TRUE);
}
/*********************************************************************//**
When an index page is read from a disk to the buffer pool, this function
applies any buffered operations to the page and deletes the entries from the
insert buffer. If the page is not read, but created in the buffer pool, this
function deletes its buffered entries from the insert buffer; there can
exist entries for such a page if the page belonged to an index which
subsequently was dropped. */
UNIV_INTERN
void
ibuf_merge_or_delete_for_page(
/*==========================*/
buf_block_t* block, /*!< in: if page has been read from
disk, pointer to the page x-latched,
else NULL */
ulint space, /*!< in: space id of the index page */
ulint page_no,/*!< in: page number of the index page */
ulint zip_size,/*!< in: compressed page size in bytes,
or 0 */
ibool update_ibuf_bitmap)/*!< in: normally this is set
to TRUE, but if we have deleted or are
deleting the tablespace, then we
naturally do not want to update a
non-existent bitmap page */
{
mem_heap_t* heap;
btr_pcur_t pcur;
dtuple_t* search_tuple;
#ifdef UNIV_IBUF_DEBUG
ulint volume = 0;
#endif
page_zip_des_t* page_zip = NULL;
ibool tablespace_being_deleted = FALSE;
ibool corruption_noticed = FALSE;
mtr_t mtr;
/* Counts for merged & discarded operations. */
ulint mops[IBUF_OP_COUNT];
ulint dops[IBUF_OP_COUNT];
ut_ad(!block || buf_block_get_space(block) == space);
ut_ad(!block || buf_block_get_page_no(block) == page_no);
ut_ad(!block || buf_block_get_zip_size(block) == zip_size);
if (srv_force_recovery >= SRV_FORCE_NO_IBUF_MERGE
|| trx_sys_hdr_page(space, page_no)) {
return;
}
/* We cannot refer to zip_size in the following, because
zip_size is passed as ULINT_UNDEFINED (it is unknown) when
buf_read_ibuf_merge_pages() is merging (discarding) changes
for a dropped tablespace. When block != NULL or
update_ibuf_bitmap is specified, the zip_size must be known.
That is why we will repeat the check below, with zip_size in
place of 0. Passing zip_size as 0 assumes that the
uncompressed page size always is a power-of-2 multiple of the
compressed page size. */
if (ibuf_fixed_addr_page(space, 0, page_no)
|| fsp_descr_page(0, page_no)) {
return;
}
if (UNIV_LIKELY(update_ibuf_bitmap)) {
ut_a(ut_is_2pow(zip_size));
if (ibuf_fixed_addr_page(space, zip_size, page_no)
|| fsp_descr_page(zip_size, page_no)) {
return;
}
/* If the following returns FALSE, we get the counter
incremented, and must decrement it when we leave this
function. When the counter is > 0, that prevents tablespace
from being dropped. */
tablespace_being_deleted = fil_inc_pending_ibuf_merges(space);
if (UNIV_UNLIKELY(tablespace_being_deleted)) {
/* Do not try to read the bitmap page from space;
just delete the ibuf records for the page */
block = NULL;
update_ibuf_bitmap = FALSE;
} else {
page_t* bitmap_page;
mtr_start(&mtr);
bitmap_page = ibuf_bitmap_get_map_page(
space, page_no, zip_size, &mtr);
if (!ibuf_bitmap_page_get_bits(bitmap_page, page_no,
zip_size,
IBUF_BITMAP_BUFFERED,
&mtr)) {
/* No inserts buffered for this page */
mtr_commit(&mtr);
if (!tablespace_being_deleted) {
fil_decr_pending_ibuf_merges(space);
}
return;
}
mtr_commit(&mtr);
}
} else if (block
&& (ibuf_fixed_addr_page(space, zip_size, page_no)
|| fsp_descr_page(zip_size, page_no))) {
return;
}
ibuf_enter();
heap = mem_heap_create(512);
if (!trx_sys_multiple_tablespace_format) {
ut_a(trx_doublewrite_must_reset_space_ids);
search_tuple = ibuf_search_tuple_build(space, page_no, heap);
} else {
search_tuple = ibuf_new_search_tuple_build(space, page_no,
heap);
}
if (block) {
/* Move the ownership of the x-latch on the page to this OS
thread, so that we can acquire a second x-latch on it. This
is needed for the insert operations to the index page to pass
the debug checks. */
rw_lock_x_lock_move_ownership(&(block->lock));
page_zip = buf_block_get_page_zip(block);
if (UNIV_UNLIKELY(fil_page_get_type(block->frame)
!= FIL_PAGE_INDEX)
|| UNIV_UNLIKELY(!page_is_leaf(block->frame))) {
page_t* bitmap_page;
corruption_noticed = TRUE;
ut_print_timestamp(stderr);
mtr_start(&mtr);
fputs(" InnoDB: Dump of the ibuf bitmap page:\n",
stderr);
bitmap_page = ibuf_bitmap_get_map_page(space, page_no,
zip_size, &mtr);
buf_page_print(bitmap_page, 0);
mtr_commit(&mtr);
fputs("\nInnoDB: Dump of the page:\n", stderr);
buf_page_print(block->frame, 0);
fprintf(stderr,
"InnoDB: Error: corruption in the tablespace."
" Bitmap shows insert\n"
"InnoDB: buffer records to page n:o %lu"
" though the page\n"
"InnoDB: type is %lu, which is"
" not an index leaf page!\n"
"InnoDB: We try to resolve the problem"
" by skipping the insert buffer\n"
"InnoDB: merge for this page."
" Please run CHECK TABLE on your tables\n"
"InnoDB: to determine if they are corrupt"
" after this.\n\n"
"InnoDB: Please submit a detailed bug report"
" to http://bugs.mysql.com\n\n",
(ulong) page_no,
(ulong)
fil_page_get_type(block->frame));
}
}
memset(mops, 0, sizeof(mops));
memset(dops, 0, sizeof(dops));
loop:
mtr_start(&mtr);
if (block) {
ibool success;
success = buf_page_get_known_nowait(
RW_X_LATCH, block,
BUF_KEEP_OLD, __FILE__, __LINE__, &mtr);
ut_a(success);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
}
/* Position pcur in the insert buffer at the first entry for this
index page */
btr_pcur_open_on_user_rec(
ibuf->index, search_tuple, PAGE_CUR_GE, BTR_MODIFY_LEAF,
&pcur, &mtr);
if (!btr_pcur_is_on_user_rec(&pcur)) {
ut_ad(btr_pcur_is_after_last_in_tree(&pcur, &mtr));
goto reset_bit;
}
for (;;) {
rec_t* rec;
ut_ad(btr_pcur_is_on_user_rec(&pcur));
rec = btr_pcur_get_rec(&pcur);
/* Check if the entry is for this index page */
if (ibuf_rec_get_page_no(rec) != page_no
|| ibuf_rec_get_space(rec) != space) {
if (block) {
page_header_reset_last_insert(
block->frame, page_zip, &mtr);
}
goto reset_bit;
}
if (UNIV_UNLIKELY(corruption_noticed)) {
fputs("InnoDB: Discarding record\n ", stderr);
rec_print_old(stderr, rec);
fputs("\nInnoDB: from the insert buffer!\n\n", stderr);
} else if (block) {
/* Now we have at pcur a record which should be
applied on the index page; NOTE that the call below
copies pointers to fields in rec, and we must
keep the latch to the rec page until the
insertion is finished! */
dtuple_t* entry;
trx_id_t max_trx_id;
dict_index_t* dummy_index;
ibuf_op_t op = ibuf_rec_get_op_type(rec);
max_trx_id = page_get_max_trx_id(page_align(rec));
page_update_max_trx_id(block, page_zip, max_trx_id,
&mtr);
ut_ad(page_validate(page_align(rec), ibuf->index));
entry = ibuf_build_entry_from_ibuf_rec(
rec, heap, &dummy_index);
ut_ad(page_validate(block->frame, dummy_index));
switch (op) {
ibool success;
case IBUF_OP_INSERT:
#ifdef UNIV_IBUF_DEBUG
volume += rec_get_converted_size(
dummy_index, entry, 0);
volume += page_dir_calc_reserved_space(1);
ut_a(volume <= 4 * UNIV_PAGE_SIZE
/ IBUF_PAGE_SIZE_PER_FREE_SPACE);
#endif
ibuf_insert_to_index_page(
entry, block, dummy_index, &mtr);
break;
case IBUF_OP_DELETE_MARK:
ibuf_set_del_mark(
entry, block, dummy_index, &mtr);
break;
case IBUF_OP_DELETE:
ibuf_delete(entry, block, dummy_index, &mtr);
/* Because ibuf_delete() will latch an
insert buffer bitmap page, commit mtr
before latching any further pages.
Store and restore the cursor position. */
ut_ad(rec == btr_pcur_get_rec(&pcur));
ut_ad(page_rec_is_user_rec(rec));
ut_ad(ibuf_rec_get_page_no(rec) == page_no);
ut_ad(ibuf_rec_get_space(rec) == space);
btr_pcur_store_position(&pcur, &mtr);
btr_pcur_commit_specify_mtr(&pcur, &mtr);
mtr_start(&mtr);
success = buf_page_get_known_nowait(
RW_X_LATCH, block,
BUF_KEEP_OLD,
__FILE__, __LINE__, &mtr);
ut_a(success);
buf_block_dbg_add_level(block, SYNC_TREE_NODE);
if (!ibuf_restore_pos(space, page_no,
search_tuple,
BTR_MODIFY_LEAF,
&pcur, &mtr)) {
mtr_commit(&mtr);
mops[op]++;
ibuf_dummy_index_free(dummy_index);
goto loop;
}
break;
default:
ut_error;
}
mops[op]++;
ibuf_dummy_index_free(dummy_index);
} else {
dops[ibuf_rec_get_op_type(rec)]++;
}
/* Delete the record from ibuf */
if (ibuf_delete_rec(space, page_no, &pcur, search_tuple,
&mtr)) {
/* Deletion was pessimistic and mtr was committed:
we start from the beginning again */
goto loop;
} else if (btr_pcur_is_after_last_on_page(&pcur)) {
mtr_commit(&mtr);
btr_pcur_close(&pcur);
goto loop;
}
}
reset_bit:
if (UNIV_LIKELY(update_ibuf_bitmap)) {
page_t* bitmap_page;
bitmap_page = ibuf_bitmap_get_map_page(
space, page_no, zip_size, &mtr);
ibuf_bitmap_page_set_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_BUFFERED, FALSE, &mtr);
if (block) {
ulint old_bits = ibuf_bitmap_page_get_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, &mtr);
ulint new_bits = ibuf_index_page_calc_free(
zip_size, block);
if (old_bits != new_bits) {
ibuf_bitmap_page_set_bits(
bitmap_page, page_no, zip_size,
IBUF_BITMAP_FREE, new_bits, &mtr);
}
}
}
mtr_commit(&mtr);
btr_pcur_close(&pcur);
mem_heap_free(heap);
/* Protect our statistics keeping from race conditions */
mutex_enter(&ibuf_mutex);
ibuf->n_merges++;
ibuf_add_ops(ibuf->n_merged_ops, mops);
ibuf_add_ops(ibuf->n_discarded_ops, dops);
mutex_exit(&ibuf_mutex);
if (update_ibuf_bitmap && !tablespace_being_deleted) {
fil_decr_pending_ibuf_merges(space);
}
ibuf_exit();
#ifdef UNIV_IBUF_COUNT_DEBUG
ut_a(ibuf_count_get(space, page_no) == 0);
#endif
}
/*********************************************************************//**
Deletes all entries in the insert buffer for a given space id. This is used
in DISCARD TABLESPACE and IMPORT TABLESPACE.
NOTE: this does not update the page free bitmaps in the space. The space will
become CORRUPT when you call this function! */
UNIV_INTERN
void
ibuf_delete_for_discarded_space(
/*============================*/
ulint space) /*!< in: space id */
{
mem_heap_t* heap;
btr_pcur_t pcur;
dtuple_t* search_tuple;
rec_t* ibuf_rec;
ulint page_no;
ibool closed;
mtr_t mtr;
/* Counts for discarded operations. */
ulint dops[IBUF_OP_COUNT];
heap = mem_heap_create(512);
/* Use page number 0 to build the search tuple so that we get the
cursor positioned at the first entry for this space id */
search_tuple = ibuf_new_search_tuple_build(space, 0, heap);
memset(dops, 0, sizeof(dops));
loop:
ibuf_enter();
mtr_start(&mtr);
/* Position pcur in the insert buffer at the first entry for the
space */
btr_pcur_open_on_user_rec(
ibuf->index, search_tuple, PAGE_CUR_GE, BTR_MODIFY_LEAF,
&pcur, &mtr);
if (!btr_pcur_is_on_user_rec(&pcur)) {
ut_ad(btr_pcur_is_after_last_in_tree(&pcur, &mtr));
goto leave_loop;
}
for (;;) {
ut_ad(btr_pcur_is_on_user_rec(&pcur));
ibuf_rec = btr_pcur_get_rec(&pcur);
/* Check if the entry is for this space */
if (ibuf_rec_get_space(ibuf_rec) != space) {
goto leave_loop;
}
page_no = ibuf_rec_get_page_no(ibuf_rec);
dops[ibuf_rec_get_op_type(ibuf_rec)]++;
/* Delete the record from ibuf */
closed = ibuf_delete_rec(space, page_no, &pcur, search_tuple,
&mtr);
if (closed) {
/* Deletion was pessimistic and mtr was committed:
we start from the beginning again */
ibuf_exit();
goto loop;
}
if (btr_pcur_is_after_last_on_page(&pcur)) {
mtr_commit(&mtr);
btr_pcur_close(&pcur);
ibuf_exit();
goto loop;
}
}
leave_loop:
mtr_commit(&mtr);
btr_pcur_close(&pcur);
/* Protect our statistics keeping from race conditions */
mutex_enter(&ibuf_mutex);
ibuf_add_ops(ibuf->n_discarded_ops, dops);
mutex_exit(&ibuf_mutex);
ibuf_exit();
mem_heap_free(heap);
}
/******************************************************************//**
Looks if the insert buffer is empty.
@return TRUE if empty */
UNIV_INTERN
ibool
ibuf_is_empty(void)
/*===============*/
{
ibool is_empty;
const page_t* root;
mtr_t mtr;
ibuf_enter();
mutex_enter(&ibuf_mutex);
mtr_start(&mtr);
root = ibuf_tree_root_get(&mtr);
if (page_get_n_recs(root) == 0) {
is_empty = TRUE;
if (ibuf->empty == FALSE) {
fprintf(stderr,
"InnoDB: Warning: insert buffer tree is empty"
" but the data struct does not\n"
"InnoDB: know it. This condition is legal"
" if the master thread has not yet\n"
"InnoDB: run to completion.\n");
}
} else {
ut_a(ibuf->empty == FALSE);
is_empty = FALSE;
}
mtr_commit(&mtr);
mutex_exit(&ibuf_mutex);
ibuf_exit();
return(is_empty);
}
/******************************************************************//**
Prints info of ibuf. */
UNIV_INTERN
void
ibuf_print(
/*=======*/
FILE* file) /*!< in: file where to print */
{
#ifdef UNIV_IBUF_COUNT_DEBUG
ulint i;
ulint j;
#endif
mutex_enter(&ibuf_mutex);
fprintf(file,
"Ibuf: size %lu, free list len %lu,"
" seg size %lu, %lu merges\n",
(ulong) ibuf->size,
(ulong) ibuf->free_list_len,
(ulong) ibuf->seg_size,
(ulong) ibuf->n_merges);
fputs("merged operations:\n ", file);
ibuf_print_ops(ibuf->n_merged_ops, file);
fputs("discarded operations:\n ", file);
ibuf_print_ops(ibuf->n_discarded_ops, file);
#ifdef UNIV_IBUF_COUNT_DEBUG
for (i = 0; i < IBUF_COUNT_N_SPACES; i++) {
for (j = 0; j < IBUF_COUNT_N_PAGES; j++) {
ulint count = ibuf_count_get(i, j);
if (count > 0) {
fprintf(stderr,
"Ibuf count for space/page %lu/%lu"
" is %lu\n",
(ulong) i, (ulong) j, (ulong) count);
}
}
}
#endif /* UNIV_IBUF_COUNT_DEBUG */
mutex_exit(&ibuf_mutex);
}
#endif /* !UNIV_HOTBACKUP */