1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
|
/* Copyright (C) 2000-2008 MySQL AB, 2008-2011 Monty Program Ab
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111-1301 USA */
/*
These functions handle page caching for Maria tables.
One cache can handle many files.
It must contain buffers of the same blocksize.
init_pagecache() should be used to init cache handler.
The free list (free_block_list) is a stack like structure.
When a block is freed by free_block(), it is pushed onto the stack.
When a new block is required it is first tried to pop one from the stack.
If the stack is empty, it is tried to get a never-used block from the pool.
If this is empty too, then a block is taken from the LRU ring, flushing it
to disk, if necessary. This is handled in find_block().
With the new free list, the blocks can have three temperatures:
hot, warm and cold (which is free). This is remembered in the block header
by the enum PCBLOCK_TEMPERATURE temperature variable. Remembering the
temperature is necessary to correctly count the number of warm blocks,
which is required to decide when blocks are allowed to become hot. Whenever
a block is inserted to another (sub-)chain, we take the old and new
temperature into account to decide if we got one more or less warm block.
blocks_unused is the sum of never used blocks in the pool and of currently
free blocks. blocks_used is the number of blocks fetched from the pool and
as such gives the maximum number of in-use blocks at any time.
TODO: Write operation locks whole cache till the end of the operation.
Should be fixed.
*/
#include "maria_def.h"
#include <m_string.h>
#include "ma_pagecache.h"
#include "ma_blockrec.h"
#include <my_bit.h>
#include <errno.h>
/*
Some compilation flags have been added specifically for this module
to control the following:
- not to let a thread to yield the control when reading directly
from page cache, which might improve performance in many cases;
to enable this add:
#define SERIALIZED_READ_FROM_CACHE
- to set an upper bound for number of threads simultaneously
using the page cache; this setting helps to determine an optimal
size for hash table and improve performance when the number of
blocks in the page cache much less than the number of threads
accessing it;
to set this number equal to <N> add
#define MAX_THREADS <N>
- to substitute calls of mysql_cond_wait for calls of
mysql_cond_timedwait (wait with timeout set up);
this setting should be used only when you want to trap a deadlock
situation, which theoretically should not happen;
to set timeout equal to <T> seconds add
#define PAGECACHE_TIMEOUT <T>
- to enable the module traps and to send debug information from
page cache module to a special debug log add:
#define PAGECACHE_DEBUG
the name of this debug log file <LOG NAME> can be set through:
#define PAGECACHE_DEBUG_LOG <LOG NAME>
if the name is not defined, it's set by default;
if the PAGECACHE_DEBUG flag is not set up and we are in a debug
mode, i.e. when ! defined(DBUG_OFF), the debug information from the
module is sent to the regular debug log.
Example of the settings:
#define SERIALIZED_READ_FROM_CACHE
#define MAX_THREADS 100
#define PAGECACHE_TIMEOUT 1
#define PAGECACHE_DEBUG
#define PAGECACHE_DEBUG_LOG "my_pagecache_debug.log"
*/
/*
In key cache we have external raw locking here we use
SERIALIZED_READ_FROM_CACHE to avoid problem of reading
not consistent data from the page.
(keycache functions (key_cache_read(), key_cache_insert() and
key_cache_write()) rely on external MyISAM lock, we don't)
*/
#define SERIALIZED_READ_FROM_CACHE yes
#define PCBLOCK_INFO(B) \
DBUG_PRINT("info", \
("block: 0x%lx fd: %lu page: %lu status: 0x%x " \
"hshL: 0x%lx requests: %u/%u wrlocks: %u rdlocks: %u " \
"rdlocks_q: %u pins: %u type: %s", \
(ulong)(B), \
(ulong)((B)->hash_link ? \
(B)->hash_link->file.file : \
0), \
(ulong)((B)->hash_link ? \
(B)->hash_link->pageno : \
0), \
(uint) (B)->status, \
(ulong)(B)->hash_link, \
(uint) (B)->requests, \
(uint)((B)->hash_link ? \
(B)->hash_link->requests : \
0), \
(B)->wlocks, (B)->rlocks, (B)->rlocks_queue, \
(uint)(B)->pins, \
page_cache_page_type_str[(B)->type]))
/* TODO: put it to my_static.c */
my_bool my_disable_flush_pagecache_blocks= 0;
#define STRUCT_PTR(TYPE, MEMBER, a) \
(TYPE *) ((char *) (a) - offsetof(TYPE, MEMBER))
/* types of condition variables */
#define COND_FOR_REQUESTED 0 /* queue of thread waiting for read operation */
#define COND_FOR_SAVED 1 /* queue of thread waiting for flush */
#define COND_FOR_WRLOCK 2 /* queue of write lock */
#define COND_SIZE 3 /* number of COND_* queues */
typedef mysql_cond_t KEYCACHE_CONDVAR;
/* descriptor of the page in the page cache block buffer */
struct st_pagecache_page
{
PAGECACHE_FILE file; /* file to which the page belongs to */
pgcache_page_no_t pageno; /* number of the page in the file */
};
/* element in the chain of a hash table bucket */
struct st_pagecache_hash_link
{
struct st_pagecache_hash_link
*next, **prev; /* to connect links in the same bucket */
struct st_pagecache_block_link
*block; /* reference to the block for the page: */
PAGECACHE_FILE file; /* from such a file */
pgcache_page_no_t pageno; /* this page */
uint requests; /* number of requests for the page */
};
/* simple states of a block */
#define PCBLOCK_ERROR 1 /* an error occurred when performing disk i/o */
#define PCBLOCK_READ 2 /* the is page in the block buffer */
/*
A tread is reading the data to the page.
If the page contained old changed data, it will be written out with
this state set on the block.
The page is not yet ready to be used for reading.
*/
#define PCBLOCK_IN_SWITCH 4
/*
Block does not accept new requests for old page that would cause
the page to be pinned or written to.
(Reads that copies the block can still continue).
This state happens when another thread is waiting for readers to finish
to read data to the block (after the block, if it was changed, has been
flushed out to disk).
*/
#define PCBLOCK_REASSIGNED 8
#define PCBLOCK_IN_FLUSH 16 /* block is in flush operation */
#define PCBLOCK_CHANGED 32 /* block buffer contains a dirty page */
#define PCBLOCK_DIRECT_W 64 /* possible direct write to the block */
#define PCBLOCK_DEL_WRITE 128 /* should be written on delete */
/* page status, returned by find_block */
#define PAGE_READ 0
#define PAGE_TO_BE_READ 1
#define PAGE_WAIT_TO_BE_READ 2
/* block temperature determines in which (sub-)chain the block currently is */
enum PCBLOCK_TEMPERATURE { PCBLOCK_COLD /*free*/ , PCBLOCK_WARM , PCBLOCK_HOT };
/* debug info */
#ifndef DBUG_OFF
static const char *page_cache_page_type_str[]=
{
/* used only for control page type changing during debugging */
"EMPTY",
"PLAIN",
"LSN",
"READ_UNKNOWN"
};
static const char *page_cache_page_write_mode_str[]=
{
"DELAY",
"DONE"
};
static const char *page_cache_page_lock_str[]=
{
"free -> free",
"read -> read",
"write -> write",
"free -> read",
"free -> write",
"read -> free",
"write -> free",
"write -> read"
};
static const char *page_cache_page_pin_str[]=
{
"pinned -> pinned",
"unpinned -> unpinned",
"unpinned -> pinned",
"pinned -> unpinned"
};
typedef struct st_pagecache_pin_info
{
struct st_pagecache_pin_info *next, **prev;
struct st_my_thread_var *thread;
} PAGECACHE_PIN_INFO;
/*
st_pagecache_lock_info structure should be kept in next, prev, thread part
compatible with st_pagecache_pin_info to be compatible in functions.
*/
typedef struct st_pagecache_lock_info
{
struct st_pagecache_lock_info *next, **prev;
struct st_my_thread_var *thread;
my_bool write_lock;
} PAGECACHE_LOCK_INFO;
/* service functions maintain debugging info about pin & lock */
/*
Links information about thread pinned/locked the block to the list
SYNOPSIS
info_link()
list the list to link in
node the node which should be linked
*/
static void info_link(PAGECACHE_PIN_INFO **list, PAGECACHE_PIN_INFO *node)
{
if ((node->next= *list))
node->next->prev= &(node->next);
*list= node;
node->prev= list;
}
/*
Unlinks information about thread pinned/locked the block from the list
SYNOPSIS
info_unlink()
node the node which should be unlinked
*/
static void info_unlink(PAGECACHE_PIN_INFO *node)
{
if ((*node->prev= node->next))
node->next->prev= node->prev;
}
/*
Finds information about given thread in the list of threads which
pinned/locked this block.
SYNOPSIS
info_find()
list the list where to find the thread
thread thread ID (reference to the st_my_thread_var
of the thread)
any return any thread of the list
RETURN
0 - the thread was not found
pointer to the information node of the thread in the list, or, if 'any',
to any thread of the list.
*/
static PAGECACHE_PIN_INFO *info_find(PAGECACHE_PIN_INFO *list,
struct st_my_thread_var *thread,
my_bool any)
{
register PAGECACHE_PIN_INFO *i= list;
if (any)
return i;
for(; i != 0; i= i->next)
if (i->thread == thread)
return i;
return 0;
}
#endif /* !DBUG_OFF */
/* page cache block */
struct st_pagecache_block_link
{
struct st_pagecache_block_link
*next_used, **prev_used; /* to connect links in the LRU chain (ring) */
struct st_pagecache_block_link
*next_changed, **prev_changed; /* for lists of file dirty/clean blocks */
struct st_pagecache_hash_link
*hash_link; /* backward ptr to referring hash_link */
#ifndef DBUG_OFF
PAGECACHE_PIN_INFO *pin_list;
PAGECACHE_LOCK_INFO *lock_list;
#endif
KEYCACHE_CONDVAR *condvar; /* condition variable for 'no readers' event */
uchar *buffer; /* buffer for the block page */
pthread_t write_locker;
ulonglong last_hit_time; /* timestamp of the last hit */
WQUEUE
wqueue[COND_SIZE]; /* queues on waiting requests for new/old pages */
uint32 requests; /* number of requests for the block */
uint32 pins; /* pin counter */
uint32 wlocks; /* write locks counter */
uint32 rlocks; /* read locks counter */
uint32 rlocks_queue; /* rd. locks waiting wr. lock of this thread */
uint16 status; /* state of the block */
int16 error; /* error code for block in case of error */
enum PCBLOCK_TEMPERATURE temperature; /* block temperature: cold, warm, hot*/
enum pagecache_page_type type; /* type of the block */
uint hits_left; /* number of hits left until promotion */
/** @brief LSN when first became dirty; LSN_MAX means "not yet set" */
LSN rec_lsn;
};
/** @brief information describing a run of flush_pagecache_blocks_int() */
struct st_file_in_flush
{
File file;
/**
@brief threads waiting for the thread currently flushing this file to be
done
*/
WQUEUE flush_queue;
/**
@brief if the thread currently flushing the file has a non-empty
first_in_switch list.
*/
my_bool first_in_switch;
};
#ifndef DBUG_OFF
/* debug checks */
#ifdef NOT_USED
static my_bool info_check_pin(PAGECACHE_BLOCK_LINK *block,
enum pagecache_page_pin mode
__attribute__((unused)))
{
struct st_my_thread_var *thread= my_thread_var;
PAGECACHE_PIN_INFO *info= info_find(block->pin_list, thread);
DBUG_ENTER("info_check_pin");
DBUG_PRINT("enter", ("thread: 0x%lx pin: %s",
(ulong) thread, page_cache_page_pin_str[mode]));
if (info)
{
if (mode == PAGECACHE_PIN_LEFT_UNPINNED)
{
DBUG_PRINT("info",
("info_check_pin: thread: 0x%lx block: 0x%lx ; LEFT_UNPINNED!!!",
(ulong)thread, (ulong)block));
DBUG_RETURN(1);
}
else if (mode == PAGECACHE_PIN)
{
DBUG_PRINT("info",
("info_check_pin: thread: 0x%lx block: 0x%lx ; PIN!!!",
(ulong)thread, (ulong)block));
DBUG_RETURN(1);
}
}
else
{
if (mode == PAGECACHE_PIN_LEFT_PINNED)
{
DBUG_PRINT("info",
("info_check_pin: thread: 0x%lx block: 0x%lx ; LEFT_PINNED!!!",
(ulong)thread, (ulong)block));
DBUG_RETURN(1);
}
else if (mode == PAGECACHE_UNPIN)
{
DBUG_PRINT("info",
("info_check_pin: thread: 0x%lx block: 0x%lx ; UNPIN!!!",
(ulong)thread, (ulong)block));
DBUG_RETURN(1);
}
}
DBUG_RETURN(0);
}
/*
Debug function which checks current lock/pin state and requested changes
SYNOPSIS
info_check_lock()
lock requested lock changes
pin requested pin changes
RETURN
0 - OK
1 - Error
*/
static my_bool info_check_lock(PAGECACHE_BLOCK_LINK *block,
enum pagecache_page_lock lock,
enum pagecache_page_pin pin)
{
struct st_my_thread_var *thread= my_thread_var;
PAGECACHE_LOCK_INFO *info=
(PAGECACHE_LOCK_INFO *) info_find((PAGECACHE_PIN_INFO *) block->lock_list,
thread);
DBUG_ENTER("info_check_lock");
switch(lock) {
case PAGECACHE_LOCK_LEFT_UNLOCKED:
if (pin != PAGECACHE_PIN_LEFT_UNPINNED ||
info)
goto error;
break;
case PAGECACHE_LOCK_LEFT_READLOCKED:
if ((pin != PAGECACHE_PIN_LEFT_UNPINNED &&
pin != PAGECACHE_PIN_LEFT_PINNED) ||
info == 0 || info->write_lock)
goto error;
break;
case PAGECACHE_LOCK_LEFT_WRITELOCKED:
if (pin != PAGECACHE_PIN_LEFT_PINNED ||
info == 0 || !info->write_lock)
goto error;
break;
case PAGECACHE_LOCK_READ:
if ((pin != PAGECACHE_PIN_LEFT_UNPINNED &&
pin != PAGECACHE_PIN) ||
info != 0)
goto error;
break;
case PAGECACHE_LOCK_WRITE:
if (pin != PAGECACHE_PIN ||
info != 0)
goto error;
break;
case PAGECACHE_LOCK_READ_UNLOCK:
if ((pin != PAGECACHE_PIN_LEFT_UNPINNED &&
pin != PAGECACHE_UNPIN) ||
info == 0 || info->write_lock)
goto error;
break;
case PAGECACHE_LOCK_WRITE_UNLOCK:
if (pin != PAGECACHE_UNPIN ||
info == 0 || !info->write_lock)
goto error;
break;
case PAGECACHE_LOCK_WRITE_TO_READ:
if ((pin != PAGECACHE_PIN_LEFT_PINNED &&
pin != PAGECACHE_UNPIN) ||
info == 0 || !info->write_lock)
goto error;
break;
}
DBUG_RETURN(0);
error:
DBUG_PRINT("info",
("info_check_lock: thread: 0x%lx block 0x%lx: info: %d wrt: %d,"
"to lock: %s, to pin: %s",
(ulong) thread, (ulong) block, MY_TEST(info),
(info ? info->write_lock : 0),
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin]));
DBUG_RETURN(1);
}
#endif /* NOT_USED */
#endif /* !DBUG_OFF */
#define FLUSH_CACHE 2000 /* sort this many blocks at once */
static my_bool free_block(PAGECACHE *pagecache, PAGECACHE_BLOCK_LINK *block,
my_bool abort_if_pinned);
static void unlink_hash(PAGECACHE *pagecache, PAGECACHE_HASH_LINK *hash_link);
#ifndef DBUG_OFF
static void test_key_cache(PAGECACHE *pagecache,
const char *where, my_bool lock);
#endif
#define PAGECACHE_HASH(p, f, pos) (((size_t) (pos) + \
(size_t) (f).file) & (p->hash_entries-1))
#define FILE_HASH(f,cache) ((uint) (f).file & (cache->changed_blocks_hash_size-1))
#define DEFAULT_PAGECACHE_DEBUG_LOG "pagecache_debug.log"
#if defined(PAGECACHE_DEBUG) && ! defined(PAGECACHE_DEBUG_LOG)
#define PAGECACHE_DEBUG_LOG DEFAULT_PAGECACHE_DEBUG_LOG
#endif
#if defined(PAGECACHE_DEBUG_LOG)
static FILE *pagecache_debug_log= NULL;
static void pagecache_debug_print _VARARGS((const char *fmt, ...));
#define PAGECACHE_DEBUG_OPEN \
if (!pagecache_debug_log) \
{ \
pagecache_debug_log= fopen(PAGECACHE_DEBUG_LOG, "w"); \
(void) setvbuf(pagecache_debug_log, NULL, _IOLBF, BUFSIZ); \
}
#define PAGECACHE_DEBUG_CLOSE \
if (pagecache_debug_log) \
{ \
fclose(pagecache_debug_log); \
pagecache_debug_log= 0; \
}
#else
#define PAGECACHE_DEBUG_OPEN
#define PAGECACHE_DEBUG_CLOSE
#endif /* defined(PAGECACHE_DEBUG_LOG) */
#if defined(PAGECACHE_DEBUG_LOG) && defined(PAGECACHE_DEBUG)
#define KEYCACHE_PRINT(l, m) KEYCACHE_DBUG_PRINT(l,m)
#define KEYCACHE_DBUG_PRINT(l, m) \
{ if (pagecache_debug_log) \
fprintf(pagecache_debug_log, "%s: ", l); \
pagecache_debug_print m; }
#define KEYCACHE_DBUG_ASSERT(a) \
{ if (! (a) && pagecache_debug_log) \
fclose(pagecache_debug_log); \
DBUG_ASSERT(a); }
#else
#define KEYCACHE_PRINT(l, m)
#define KEYCACHE_DBUG_PRINT(l, m) DBUG_PRINT(l, m)
#define KEYCACHE_DBUG_ASSERT(a) DBUG_ASSERT(a)
#endif /* defined(PAGECACHE_DEBUG_LOG) && defined(PAGECACHE_DEBUG) */
#if defined(PAGECACHE_DEBUG) || !defined(DBUG_OFF)
static long pagecache_thread_id;
#define KEYCACHE_THREAD_TRACE(l) \
KEYCACHE_DBUG_PRINT(l,("|thread %ld",pagecache_thread_id))
#define KEYCACHE_THREAD_TRACE_BEGIN(l) \
{ struct st_my_thread_var *thread_var= my_thread_var; \
pagecache_thread_id= thread_var->id; \
KEYCACHE_DBUG_PRINT(l,("[thread %ld",pagecache_thread_id)) }
#define KEYCACHE_THREAD_TRACE_END(l) \
KEYCACHE_DBUG_PRINT(l,("]thread %ld",pagecache_thread_id))
#else
#define KEYCACHE_PRINT(l,m)
#define KEYCACHE_THREAD_TRACE_BEGIN(l)
#define KEYCACHE_THREAD_TRACE_END(l)
#define KEYCACHE_THREAD_TRACE(l)
#endif /* defined(PAGECACHE_DEBUG) || !defined(DBUG_OFF) */
#define PCBLOCK_NUMBER(p, b) \
((uint) (((char*)(b)-(char *) p->block_root)/sizeof(PAGECACHE_BLOCK_LINK)))
#define PAGECACHE_HASH_LINK_NUMBER(p, h) \
((uint) (((char*)(h)-(char *) p->hash_link_root)/ \
sizeof(PAGECACHE_HASH_LINK)))
#if (defined(PAGECACHE_TIMEOUT) && !defined(__WIN__)) || defined(PAGECACHE_DEBUG)
static int pagecache_pthread_cond_wait(mysql_cond_t *cond,
mysql_mutex_t *mutex);
#else
#define pagecache_pthread_cond_wait mysql_cond_wait
#endif
#if defined(PAGECACHE_DEBUG)
static int ___pagecache_pthread_mutex_lock(mysql_mutex_t *mutex);
static void ___pagecache_pthread_mutex_unlock(mysql_mutex_t *mutex);
static int ___pagecache_pthread_cond_signal(mysql_cond_t *cond);
#define pagecache_pthread_mutex_lock(M) \
{ DBUG_PRINT("lock", ("mutex lock 0x%lx %u", (ulong)(M), __LINE__)); \
___pagecache_pthread_mutex_lock(M);}
#define pagecache_pthread_mutex_unlock(M) \
{ DBUG_PRINT("lock", ("mutex unlock 0x%lx %u", (ulong)(M), __LINE__)); \
___pagecache_pthread_mutex_unlock(M);}
#define pagecache_pthread_cond_signal(M) \
{ DBUG_PRINT("lock", ("signal 0x%lx %u", (ulong)(M), __LINE__)); \
___pagecache_pthread_cond_signal(M);}
#else
#define pagecache_pthread_mutex_lock mysql_mutex_lock
#define pagecache_pthread_mutex_unlock mysql_mutex_unlock
#define pagecache_pthread_cond_signal mysql_cond_signal
#endif /* defined(PAGECACHE_DEBUG) */
extern my_bool translog_flush(TRANSLOG_ADDRESS lsn);
/*
Write page to the disk
SYNOPSIS
pagecache_fwrite()
pagecache - page cache pointer
filedesc - pagecache file descriptor structure
buffer - buffer which we will write
type - page type (plain or with LSN)
flags - MYF() flags
RETURN
0 - OK
1 - Error
*/
static my_bool pagecache_fwrite(PAGECACHE *pagecache,
PAGECACHE_FILE *filedesc,
uchar *buffer,
pgcache_page_no_t pageno,
enum pagecache_page_type type
__attribute__((unused)),
myf flags)
{
int res;
PAGECACHE_IO_HOOK_ARGS args;
DBUG_ENTER("pagecache_fwrite");
DBUG_ASSERT(type != PAGECACHE_READ_UNKNOWN_PAGE);
#ifdef EXTRA_DEBUG_BITMAP
/*
This code is very good when debugging changes in bitmaps or dirty lists
The above define should be defined for all Aria files if you want to
debug either of the above issues.
*/
if (pagecache->extra_debug)
{
char buff[80];
uint len= my_sprintf(buff,
(buff, "fwrite: fd: %d id: %u page: %llu",
filedesc->file,
_ma_file_callback_to_id(filedesc->callback_data),
pageno));
(void) translog_log_debug_info(0, LOGREC_DEBUG_INFO_QUERY,
(uchar*) buff, len);
}
#endif
/* initialize hooks args */
args.page= buffer;
args.pageno= pageno;
args.data= filedesc->callback_data;
/* Todo: Integrate this with write_callback so we have only one callback */
if ((*filedesc->flush_log_callback)(&args))
DBUG_RETURN(1);
DBUG_PRINT("info", ("pre_write_hook: 0x%lx data: 0x%lx",
(ulong) filedesc->pre_write_hook,
(ulong) filedesc->callback_data));
if ((*filedesc->pre_write_hook)(&args))
{
DBUG_PRINT("error", ("write callback problem"));
DBUG_RETURN(1);
}
res= my_pwrite(filedesc->file, args.page, pagecache->block_size,
((my_off_t) pageno << pagecache->shift), flags);
(*filedesc->post_write_hook)(res, &args);
DBUG_RETURN(res);
}
/*
Read page from the disk
SYNOPSIS
pagecache_fread()
pagecache - page cache pointer
filedesc - pagecache file descriptor structure
buffer - buffer in which we will read
pageno - page number
flags - MYF() flags
*/
#define pagecache_fread(pagecache, filedesc, buffer, pageno, flags) \
mysql_file_pread((filedesc)->file, buffer, pagecache->block_size, \
((my_off_t) pageno << pagecache->shift), flags)
/**
@brief set rec_lsn of pagecache block (if it is needed)
@param block block where to set rec_lsn
@param first_REDO_LSN_for_page the LSN to set
*/
static inline void pagecache_set_block_rec_lsn(PAGECACHE_BLOCK_LINK *block,
LSN first_REDO_LSN_for_page)
{
if (block->rec_lsn == LSN_MAX)
block->rec_lsn= first_REDO_LSN_for_page;
else
DBUG_ASSERT(cmp_translog_addr(block->rec_lsn,
first_REDO_LSN_for_page) <= 0);
}
/*
next_power(value) is 2 at the power of (1+floor(log2(value)));
e.g. next_power(2)=4, next_power(3)=4.
*/
static inline uint next_power(uint value)
{
return (uint) my_round_up_to_next_power((uint32) value) << 1;
}
/*
Initialize a page cache
SYNOPSIS
init_pagecache()
pagecache pointer to a page cache data structure
key_cache_block_size size of blocks to keep cached data
use_mem total memory to use for the key cache
division_limit division limit (may be zero)
age_threshold age threshold (may be zero)
block_size size of block (should be power of 2)
my_read_flags Flags used for all pread/pwrite calls
Usually MY_WME in case of recovery
RETURN VALUE
number of blocks in the key cache, if successful,
0 - otherwise.
NOTES.
if pagecache->inited != 0 we assume that the key cache
is already initialized. This is for now used by myisamchk, but shouldn't
be something that a program should rely on!
It's assumed that no two threads call this function simultaneously
referring to the same key cache handle.
*/
size_t init_pagecache(PAGECACHE *pagecache, size_t use_mem,
uint division_limit, uint age_threshold,
uint block_size, uint changed_blocks_hash_size,
myf my_readwrite_flags)
{
size_t blocks, hash_links, length;
int error;
DBUG_ENTER("init_pagecache");
DBUG_ASSERT(block_size >= 512);
PAGECACHE_DEBUG_OPEN;
if (pagecache->inited && pagecache->disk_blocks > 0)
{
DBUG_PRINT("warning",("key cache already in use"));
DBUG_RETURN(0);
}
pagecache->global_cache_w_requests= pagecache->global_cache_r_requests= 0;
pagecache->global_cache_read= pagecache->global_cache_write= 0;
pagecache->disk_blocks= -1;
if (! pagecache->inited)
{
if (mysql_mutex_init(key_PAGECACHE_cache_lock,
&pagecache->cache_lock, MY_MUTEX_INIT_FAST) ||
my_hash_init(&pagecache->files_in_flush, &my_charset_bin, 32,
offsetof(struct st_file_in_flush, file),
sizeof(((struct st_file_in_flush *)NULL)->file),
NULL, NULL, 0))
goto err;
pagecache->inited= 1;
pagecache->in_init= 0;
pagecache->resize_queue.last_thread= NULL;
}
pagecache->mem_size= use_mem;
pagecache->block_size= block_size;
pagecache->shift= my_bit_log2(block_size);
pagecache->readwrite_flags= my_readwrite_flags | MY_NABP | MY_WAIT_IF_FULL;
pagecache->org_readwrite_flags= pagecache->readwrite_flags;
DBUG_PRINT("info", ("block_size: %u", block_size));
DBUG_ASSERT(((uint)(1 << pagecache->shift)) == block_size);
blocks= use_mem / (sizeof(PAGECACHE_BLOCK_LINK) +
2 * sizeof(PAGECACHE_HASH_LINK) +
sizeof(PAGECACHE_HASH_LINK*) *
5/4 + block_size);
/* Changed blocks hash needs to be a power of 2 */
changed_blocks_hash_size= my_round_up_to_next_power(MY_MAX(changed_blocks_hash_size,
MIN_PAGECACHE_CHANGED_BLOCKS_HASH_SIZE));
/*
We need to support page cache with just one block to be able to do
scanning of rows-in-block files
*/
for ( ; ; )
{
if (blocks < 8)
{
my_message(ENOMEM, "Not enough memory to allocate 8 pagecache pages",
MYF(0));
my_errno= ENOMEM;
goto err;
}
/* Set my_hash_entries to the next bigger 2 power */
if ((pagecache->hash_entries= next_power(blocks)) <
(blocks) * 5/4)
pagecache->hash_entries<<= 1;
hash_links= 2 * blocks;
#if defined(MAX_THREADS)
if (hash_links < MAX_THREADS + blocks - 1)
hash_links= MAX_THREADS + blocks - 1;
#endif
while ((length= (ALIGN_SIZE(blocks * sizeof(PAGECACHE_BLOCK_LINK)) +
ALIGN_SIZE(sizeof(PAGECACHE_HASH_LINK*) *
pagecache->hash_entries) +
ALIGN_SIZE(hash_links * sizeof(PAGECACHE_HASH_LINK)) +
sizeof(PAGECACHE_BLOCK_LINK*)* (changed_blocks_hash_size*2))) +
(blocks << pagecache->shift) > use_mem && blocks > 8)
blocks--;
/* Allocate memory for cache page buffers */
if ((pagecache->block_mem=
my_large_malloc(blocks * pagecache->block_size,
MYF(MY_WME))))
{
/*
Allocate memory for blocks, hash_links and hash entries;
For each block 2 hash links are allocated
*/
if (my_multi_malloc_large(MYF(MY_ZEROFILL),
&pagecache->block_root,
(ulonglong) (blocks *
sizeof(PAGECACHE_BLOCK_LINK)),
&pagecache->hash_root,
(ulonglong) (sizeof(PAGECACHE_HASH_LINK*) *
pagecache->hash_entries),
&pagecache->hash_link_root,
(ulonglong) (hash_links *
sizeof(PAGECACHE_HASH_LINK)),
&pagecache->changed_blocks,
(ulonglong) (sizeof(PAGECACHE_BLOCK_LINK*) *
changed_blocks_hash_size),
&pagecache->file_blocks,
(ulonglong) (sizeof(PAGECACHE_BLOCK_LINK*) *
changed_blocks_hash_size),
NullS))
break;
my_large_free(pagecache->block_mem);
pagecache->block_mem= 0;
}
blocks= blocks / 4*3;
}
pagecache->blocks_unused= blocks;
pagecache->disk_blocks= blocks;
pagecache->hash_links= hash_links;
pagecache->hash_links_used= 0;
pagecache->free_hash_list= NULL;
pagecache->blocks_used= pagecache->blocks_changed= 0;
pagecache->global_blocks_changed= 0;
pagecache->blocks_available=0; /* For debugging */
/* The LRU chain is empty after initialization */
pagecache->used_last= NULL;
pagecache->used_ins= NULL;
pagecache->free_block_list= NULL;
pagecache->time= 0;
pagecache->warm_blocks= 0;
pagecache->min_warm_blocks= (division_limit ?
blocks * division_limit / 100 + 1 :
blocks);
pagecache->age_threshold= (age_threshold ?
blocks * age_threshold / 100 :
blocks);
pagecache->changed_blocks_hash_size= changed_blocks_hash_size;
pagecache->cnt_for_resize_op= 0;
pagecache->resize_in_flush= 0;
pagecache->can_be_used= 1;
pagecache->waiting_for_hash_link.last_thread= NULL;
pagecache->waiting_for_block.last_thread= NULL;
DBUG_PRINT("exit",
("disk_blocks: %zu block_root: %p hash_entries: %zu\
hash_root: %p hash_links: %zu hash_link_root: %p",
pagecache->disk_blocks, pagecache->block_root,
pagecache->hash_entries, pagecache->hash_root,
pagecache->hash_links, pagecache->hash_link_root));
pagecache->blocks= pagecache->disk_blocks > 0 ? pagecache->disk_blocks : 0;
DBUG_RETURN((size_t)pagecache->disk_blocks);
err:
error= my_errno;
pagecache->disk_blocks= 0;
pagecache->blocks= 0;
if (pagecache->block_mem)
{
my_large_free(pagecache->block_mem);
pagecache->block_mem= NULL;
}
if (pagecache->block_root)
{
my_free(pagecache->block_root);
pagecache->block_root= NULL;
}
my_errno= error;
pagecache->can_be_used= 0;
DBUG_RETURN(0);
}
/*
Flush all blocks in the key cache to disk
*/
#ifdef NOT_USED
static int flush_all_key_blocks(PAGECACHE *pagecache)
{
#if defined(PAGECACHE_DEBUG)
uint cnt=0;
#endif
while (pagecache->blocks_changed > 0)
{
PAGECACHE_BLOCK_LINK *block;
for (block= pagecache->used_last->next_used ; ; block=block->next_used)
{
if (block->hash_link)
{
#if defined(PAGECACHE_DEBUG)
cnt++;
KEYCACHE_DBUG_ASSERT(cnt <= pagecache->blocks_used);
#endif
if (flush_pagecache_blocks_int(pagecache, &block->hash_link->file,
FLUSH_RELEASE, NULL, NULL))
return 1;
break;
}
if (block == pagecache->used_last)
break;
}
}
return 0;
}
#endif /* NOT_USED */
/*
Resize a key cache
SYNOPSIS
resize_pagecache()
pagecache pointer to a page cache data structure
use_mem total memory to use for the new key cache
division_limit new division limit (if not zero)
age_threshold new age threshold (if not zero)
RETURN VALUE
number of blocks in the key cache, if successful,
0 - otherwise.
NOTES.
The function first compares the memory size parameter
with the key cache value.
If they differ the function free the the memory allocated for the
old key cache blocks by calling the end_pagecache function and
then rebuilds the key cache with new blocks by calling
init_key_cache.
The function starts the operation only when all other threads
performing operations with the key cache let her to proceed
(when cnt_for_resize=0).
Before being usable, this function needs:
- to receive fixes for BUG#17332 "changing key_buffer_size on a running
server can crash under load" similar to those done to the key cache
- to have us (Sanja) look at the additional constraints placed on
resizing, due to the page locking specific to this page cache.
So we disable it for now.
*/
#if NOT_USED /* keep disabled until code is fixed see above !! */
size_t resize_pagecache(PAGECACHE *pagecache,
size_t use_mem, uint division_limit,
uint age_threshold, uint changed_blocks_hash_size)
{
size_t blocks;
struct st_my_thread_var *thread;
WQUEUE *wqueue;
DBUG_ENTER("resize_pagecache");
if (!pagecache->inited)
DBUG_RETURN(pagecache->disk_blocks);
if(use_mem == pagecache->mem_size)
{
change_pagecache_param(pagecache, division_limit, age_threshold);
DBUG_RETURN(pagecache->disk_blocks);
}
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
wqueue= &pagecache->resize_queue;
thread= my_thread_var;
wqueue_link_into_queue(wqueue, thread);
while (wqueue->last_thread->next != thread)
{
pagecache_pthread_cond_wait(&thread->suspend, &pagecache->cache_lock);
}
pagecache->resize_in_flush= 1;
if (flush_all_key_blocks(pagecache))
{
/* TODO: if this happens, we should write a warning in the log file ! */
pagecache->resize_in_flush= 0;
blocks= 0;
pagecache->can_be_used= 0;
goto finish;
}
pagecache->resize_in_flush= 0;
pagecache->can_be_used= 0;
while (pagecache->cnt_for_resize_op)
{
DBUG_PRINT("wait", ("suspend thread %s %ld", thread->name, thread->id));
pagecache_pthread_cond_wait(&thread->suspend, &pagecache->cache_lock);
}
end_pagecache(pagecache, 0); /* Don't free mutex */
/* The following will work even if use_mem is 0 */
blocks= init_pagecache(pagecache, pagecache->block_size, use_mem,
division_limit, age_threshold, changed_blocks_hash_size,
pagecache->readwrite_flags);
finish:
wqueue_unlink_from_queue(wqueue, thread);
/* Signal for the next resize request to proceeed if any */
if (wqueue->last_thread)
{
DBUG_PRINT("signal",
("thread %s %ld", wqueue->last_thread->next->name,
wqueue->last_thread->next->id));
pagecache_pthread_cond_signal(&wqueue->last_thread->next->suspend);
}
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_RETURN(blocks);
}
#endif /* 0 */
/*
Increment counter blocking resize key cache operation
*/
static inline void inc_counter_for_resize_op(PAGECACHE *pagecache)
{
mysql_mutex_assert_owner(&pagecache->cache_lock);
pagecache->cnt_for_resize_op++;
}
/*
Decrement counter blocking resize key cache operation;
Signal the operation to proceed when counter becomes equal zero
*/
static inline void dec_counter_for_resize_op(PAGECACHE *pagecache)
{
struct st_my_thread_var *last_thread;
mysql_mutex_assert_owner(&pagecache->cache_lock);
if (!--pagecache->cnt_for_resize_op &&
(last_thread= pagecache->resize_queue.last_thread))
{
DBUG_PRINT("signal",
("thread %s %ld", last_thread->next->name,
(ulong) last_thread->next->id));
pagecache_pthread_cond_signal(&last_thread->next->suspend);
}
}
/*
Change the page cache parameters
SYNOPSIS
change_pagecache_param()
pagecache pointer to a page cache data structure
division_limit new division limit (if not zero)
age_threshold new age threshold (if not zero)
RETURN VALUE
none
NOTES.
Presently the function resets the key cache parameters
concerning midpoint insertion strategy - division_limit and
age_threshold.
*/
void change_pagecache_param(PAGECACHE *pagecache, uint division_limit,
uint age_threshold)
{
DBUG_ENTER("change_pagecache_param");
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (division_limit)
pagecache->min_warm_blocks= (pagecache->disk_blocks *
division_limit / 100 + 1);
if (age_threshold)
pagecache->age_threshold= (pagecache->disk_blocks *
age_threshold / 100);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/*
Check that pagecache was used and cleaned up properly.
*/
#ifndef DBUG_OFF
void check_pagecache_is_cleaned_up(PAGECACHE *pagecache)
{
DBUG_ENTER("check_pagecache_is_cleaned_up");
/*
Ensure we called inc_counter_for_resize_op and dec_counter_for_resize_op
the same number of times. (If not, a resize() could never happen.
*/
DBUG_ASSERT(pagecache->cnt_for_resize_op == 0);
if (pagecache->disk_blocks > 0)
{
if (pagecache->block_mem)
{
uint i;
for (i=0 ; i < pagecache->blocks_used ; i++)
{
DBUG_ASSERT(pagecache->block_root[i].status == 0);
DBUG_ASSERT(pagecache->block_root[i].type == PAGECACHE_EMPTY_PAGE);
}
}
}
DBUG_VOID_RETURN;
}
#endif
/*
Removes page cache from memory. Does NOT flush pages to disk.
SYNOPSIS
end_pagecache()
pagecache page cache handle
cleanup Complete free (Free also mutex for key cache)
RETURN VALUE
none
*/
void end_pagecache(PAGECACHE *pagecache, my_bool cleanup)
{
DBUG_ENTER("end_pagecache");
DBUG_PRINT("enter", ("key_cache: %p", pagecache));
if (!pagecache->inited)
DBUG_VOID_RETURN;
if (pagecache->disk_blocks > 0)
{
#ifndef DBUG_OFF
check_pagecache_is_cleaned_up(pagecache);
#endif
if (pagecache->block_mem)
{
my_large_free(pagecache->block_mem);
pagecache->block_mem= NULL;
my_free(pagecache->block_root);
pagecache->block_root= NULL;
}
pagecache->disk_blocks= -1;
/* Reset blocks_changed to be safe if flush_all_key_blocks is called */
pagecache->blocks_changed= 0;
}
DBUG_PRINT("status", ("used: %zu changed: %zu w_requests: %llu "
"writes: %llu r_requests: %llu reads: %llu",
pagecache->blocks_used,
pagecache->global_blocks_changed,
pagecache->global_cache_w_requests,
pagecache->global_cache_write,
pagecache->global_cache_r_requests,
pagecache->global_cache_read));
if (cleanup)
{
my_hash_free(&pagecache->files_in_flush);
mysql_mutex_destroy(&pagecache->cache_lock);
pagecache->inited= pagecache->can_be_used= 0;
PAGECACHE_DEBUG_CLOSE;
}
DBUG_VOID_RETURN;
} /* end_pagecache */
/*
Unlink a block from the chain of dirty/clean blocks
*/
static inline void unlink_changed(PAGECACHE_BLOCK_LINK *block)
{
if (block->next_changed)
block->next_changed->prev_changed= block->prev_changed;
*block->prev_changed= block->next_changed;
}
/*
Link a block into the chain of dirty/clean blocks
*/
static inline void link_changed(PAGECACHE_BLOCK_LINK *block,
PAGECACHE_BLOCK_LINK **phead)
{
block->prev_changed= phead;
if ((block->next_changed= *phead))
(*phead)->prev_changed= &block->next_changed;
*phead= block;
}
/*
Unlink a block from the chain of dirty/clean blocks, if it's asked for,
and link it to the chain of clean blocks for the specified file
*/
static void link_to_file_list(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
PAGECACHE_FILE *file, my_bool unlink_flag)
{
if (unlink_flag)
unlink_changed(block);
link_changed(block, &pagecache->file_blocks[FILE_HASH(*file, pagecache)]);
if (block->status & PCBLOCK_CHANGED)
{
block->status&= ~(PCBLOCK_CHANGED | PCBLOCK_DEL_WRITE);
block->rec_lsn= LSN_MAX;
pagecache->blocks_changed--;
pagecache->global_blocks_changed--;
}
}
/*
Unlink a block from the chain of clean blocks for the specified
file and link it to the chain of dirty blocks for this file
*/
static inline void link_to_changed_list(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block)
{
unlink_changed(block);
link_changed(block,
&pagecache->changed_blocks[FILE_HASH(block->hash_link->file, pagecache)]);
block->status|=PCBLOCK_CHANGED;
pagecache->blocks_changed++;
pagecache->global_blocks_changed++;
}
/*
Link a block to the LRU chain at the beginning or at the end of
one of two parts.
SYNOPSIS
link_block()
pagecache pointer to a page cache data structure
block pointer to the block to link to the LRU chain
hot <-> to link the block into the hot subchain
at_end <-> to link the block at the end of the subchain
RETURN VALUE
none
NOTES.
The LRU chain is represented by a circular list of block structures.
The list is double-linked of the type (**prev,*next) type.
The LRU chain is divided into two parts - hot and warm.
There are two pointers to access the last blocks of these two
parts. The beginning of the warm part follows right after the
end of the hot part.
Only blocks of the warm part can be used for replacement.
The first block from the beginning of this subchain is always
taken for eviction (pagecache->last_used->next)
LRU chain: +------+ H O T +------+
+----| end |----...<----| beg |----+
| +------+last +------+ |
v<-link in latest hot (new end) |
| link in latest warm (new end)->^
| +------+ W A R M +------+ |
+----| beg |---->...----| end |----+
+------+ +------+ins
first for eviction
*/
static void link_block(PAGECACHE *pagecache, PAGECACHE_BLOCK_LINK *block,
my_bool hot, my_bool at_end)
{
PAGECACHE_BLOCK_LINK *ins;
PAGECACHE_BLOCK_LINK **ptr_ins;
DBUG_ENTER("link_block");
PCBLOCK_INFO(block);
KEYCACHE_DBUG_ASSERT(! (block->hash_link && block->hash_link->requests));
if (!hot && pagecache->waiting_for_block.last_thread)
{
/* Signal that in the LRU warm sub-chain an available block has appeared */
struct st_my_thread_var *last_thread=
pagecache->waiting_for_block.last_thread;
struct st_my_thread_var *first_thread= last_thread->next;
struct st_my_thread_var *next_thread= first_thread;
PAGECACHE_HASH_LINK *hash_link=
(PAGECACHE_HASH_LINK *) first_thread->keycache_link;
struct st_my_thread_var *thread;
DBUG_ASSERT(block->requests + block->wlocks + block->rlocks +
block->pins == 0);
DBUG_ASSERT(block->next_used == NULL);
do
{
thread= next_thread;
next_thread= thread->next;
/*
We notify about the event all threads that ask
for the same page as the first thread in the queue
*/
if ((PAGECACHE_HASH_LINK *) thread->keycache_link == hash_link)
{
DBUG_PRINT("signal", ("thread: %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_signal(&thread->suspend);
wqueue_unlink_from_queue(&pagecache->waiting_for_block, thread);
block->requests++;
}
}
while (thread != last_thread);
hash_link->block= block;
/* Ensure that no other thread tries to use this block */
block->status|= PCBLOCK_REASSIGNED;
DBUG_PRINT("signal", ("after signal"));
#if defined(PAGECACHE_DEBUG)
KEYCACHE_DBUG_PRINT("link_block",
("linked,unlinked block: %u status: %x #requests: %u #available: %u",
PCBLOCK_NUMBER(pagecache, block), block->status,
block->requests, pagecache->blocks_available));
#endif
DBUG_VOID_RETURN;
}
ptr_ins= hot ? &pagecache->used_ins : &pagecache->used_last;
ins= *ptr_ins;
if (ins)
{
ins->next_used->prev_used= &block->next_used;
block->next_used= ins->next_used;
block->prev_used= &ins->next_used;
ins->next_used= block;
if (at_end)
*ptr_ins= block;
}
else
{
/* The LRU chain is empty */
pagecache->used_last= pagecache->used_ins= block->next_used= block;
block->prev_used= &block->next_used;
}
KEYCACHE_THREAD_TRACE("link_block");
#if defined(PAGECACHE_DEBUG)
pagecache->blocks_available++;
KEYCACHE_DBUG_PRINT("link_block",
("linked block: %u:%1u status: %x #requests: %u #available: %u",
PCBLOCK_NUMBER(pagecache, block), at_end, block->status,
block->requests, pagecache->blocks_available));
KEYCACHE_DBUG_ASSERT(pagecache->blocks_available <=
pagecache->blocks_used);
#endif
DBUG_VOID_RETURN;
}
/*
Unlink a block from the LRU chain
SYNOPSIS
unlink_block()
pagecache pointer to a page cache data structure
block pointer to the block to unlink from the LRU chain
RETURN VALUE
none
NOTES.
See NOTES for link_block
*/
static void unlink_block(PAGECACHE *pagecache, PAGECACHE_BLOCK_LINK *block)
{
DBUG_ENTER("unlink_block");
DBUG_PRINT("pagecache", ("unlink %p", block));
DBUG_ASSERT(block->next_used != NULL);
if (block->next_used == block)
{
/* The list contains only one member */
pagecache->used_last= pagecache->used_ins= NULL;
}
else
{
block->next_used->prev_used= block->prev_used;
*block->prev_used= block->next_used;
if (pagecache->used_last == block)
pagecache->used_last= STRUCT_PTR(PAGECACHE_BLOCK_LINK,
next_used, block->prev_used);
if (pagecache->used_ins == block)
pagecache->used_ins= STRUCT_PTR(PAGECACHE_BLOCK_LINK,
next_used, block->prev_used);
}
block->next_used= NULL;
KEYCACHE_THREAD_TRACE("unlink_block");
#if defined(PAGECACHE_DEBUG)
KEYCACHE_DBUG_ASSERT(pagecache->blocks_available != 0);
pagecache->blocks_available--;
KEYCACHE_DBUG_PRINT("pagecache",
("unlinked block: %p (%u) status: %x #requests: %u #available: %u",
block, PCBLOCK_NUMBER(pagecache, block),
block->status,
block->requests, pagecache->blocks_available));
PCBLOCK_INFO(block);
#endif
DBUG_VOID_RETURN;
}
/*
Register requests for a block
SYNOPSIS
reg_requests()
pagecache this page cache reference
block the block we request reference
count how many requests we register (it is 1 everywhere)
NOTE
Registration of request means we are going to use this block so we exclude
it from the LRU if it is first request
*/
static void reg_requests(PAGECACHE *pagecache, PAGECACHE_BLOCK_LINK *block,
int count)
{
DBUG_ENTER("reg_requests");
PCBLOCK_INFO(block);
if (! block->requests)
/* First request for the block unlinks it */
unlink_block(pagecache, block);
block->requests+= count;
DBUG_VOID_RETURN;
}
/*
Unregister request for a block
linking it to the LRU chain if it's the last request
SYNOPSIS
unreg_request()
pagecache pointer to a page cache data structure
block pointer to the block to link to the LRU chain
at_end <-> to link the block at the end of the LRU chain
RETURN VALUE
none
NOTES.
Every linking to the LRU chain decrements by one a special block
counter (if it's positive). If the at_end parameter is TRUE the block is
added either at the end of warm sub-chain or at the end of hot sub-chain.
It is added to the hot subchain if its counter is zero and number of
blocks in warm sub-chain is not less than some low limit (determined by
the division_limit parameter). Otherwise the block is added to the warm
sub-chain. If the at_end parameter is FALSE the block is always added
at beginning of the warm sub-chain.
Thus a warm block can be promoted to the hot sub-chain when its counter
becomes zero for the first time.
At the same time the block at the very beginning of the hot subchain
might be moved to the beginning of the warm subchain if it stays untouched
for a too long time (this time is determined by parameter age_threshold).
*/
static void unreg_request(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block, int at_end)
{
DBUG_ENTER("unreg_request");
DBUG_PRINT("enter", ("block %p (%u) status: %x requests: %u",
block, PCBLOCK_NUMBER(pagecache, block),
block->status, block->requests));
PCBLOCK_INFO(block);
DBUG_ASSERT(block->requests > 0);
if (! --block->requests)
{
my_bool hot;
if (block->hits_left)
block->hits_left--;
hot= !block->hits_left && at_end &&
pagecache->warm_blocks > pagecache->min_warm_blocks;
if (hot)
{
if (block->temperature == PCBLOCK_WARM)
pagecache->warm_blocks--;
block->temperature= PCBLOCK_HOT;
KEYCACHE_DBUG_PRINT("unreg_request", ("#warm_blocks: %zu",
pagecache->warm_blocks));
}
link_block(pagecache, block, hot, (my_bool)at_end);
block->last_hit_time= pagecache->time;
pagecache->time++;
block= pagecache->used_ins;
/* Check if we should link a hot block to the warm block */
if (block && pagecache->time - block->last_hit_time >
pagecache->age_threshold)
{
unlink_block(pagecache, block);
link_block(pagecache, block, 0, 0);
if (block->temperature != PCBLOCK_WARM)
{
pagecache->warm_blocks++;
block->temperature= PCBLOCK_WARM;
}
KEYCACHE_DBUG_PRINT("unreg_request", ("#warm_blocks: %zu",
pagecache->warm_blocks));
}
}
DBUG_VOID_RETURN;
}
/*
Remove a reader of the page in block
*/
static inline void remove_reader(PAGECACHE_BLOCK_LINK *block)
{
DBUG_ENTER("remove_reader");
PCBLOCK_INFO(block);
DBUG_ASSERT(block->hash_link->requests > 0);
if (! --block->hash_link->requests && block->condvar)
pagecache_pthread_cond_signal(block->condvar);
DBUG_VOID_RETURN;
}
/*
Wait until the last reader of the page in block
signals on its termination
*/
static inline void wait_for_readers(PAGECACHE *pagecache
__attribute__((unused)),
PAGECACHE_BLOCK_LINK *block
__attribute__((unused)))
{
struct st_my_thread_var *thread= my_thread_var;
DBUG_ASSERT(block->condvar == NULL);
while (block->hash_link->requests)
{
DBUG_ENTER("wait_for_readers");
DBUG_PRINT("wait",
("suspend thread: %s %ld block: %u",
thread->name, (ulong) thread->id,
PCBLOCK_NUMBER(pagecache, block)));
block->condvar= &thread->suspend;
pagecache_pthread_cond_wait(&thread->suspend, &pagecache->cache_lock);
block->condvar= NULL;
DBUG_VOID_RETURN;
}
}
/*
Wait until the flush of the page is done.
*/
static void wait_for_flush(PAGECACHE *pagecache
__attribute__((unused)),
PAGECACHE_BLOCK_LINK *block
__attribute__((unused)))
{
struct st_my_thread_var *thread= my_thread_var;
DBUG_ENTER("wait_for_flush");
wqueue_add_to_queue(&block->wqueue[COND_FOR_SAVED], thread);
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name, (ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while(thread->next);
DBUG_VOID_RETURN;
}
/*
Add a hash link to a bucket in the hash_table
*/
static inline void link_hash(PAGECACHE_HASH_LINK **start,
PAGECACHE_HASH_LINK *hash_link)
{
if (*start)
(*start)->prev= &hash_link->next;
hash_link->next= *start;
hash_link->prev= start;
*start= hash_link;
}
/*
Remove a hash link from the hash table
*/
static void unlink_hash(PAGECACHE *pagecache, PAGECACHE_HASH_LINK *hash_link)
{
DBUG_ENTER("unlink_hash");
DBUG_PRINT("enter", ("hash_link: %p fd: %u pos: %lu requests: %u",
hash_link, (uint) hash_link->file.file,
(ulong) hash_link->pageno,
hash_link->requests));
DBUG_ASSERT(hash_link->requests == 0);
DBUG_ASSERT(!hash_link->block || hash_link->block->pins == 0);
if ((*hash_link->prev= hash_link->next))
hash_link->next->prev= hash_link->prev;
hash_link->block= NULL;
if (pagecache->waiting_for_hash_link.last_thread)
{
/* Signal that a free hash link has appeared */
struct st_my_thread_var *last_thread=
pagecache->waiting_for_hash_link.last_thread;
struct st_my_thread_var *first_thread= last_thread->next;
struct st_my_thread_var *next_thread= first_thread;
PAGECACHE_PAGE *first_page= (PAGECACHE_PAGE *) (first_thread->keycache_link);
struct st_my_thread_var *thread;
hash_link->file= first_page->file;
DBUG_ASSERT(first_page->pageno < ((1ULL) << 40));
hash_link->pageno= first_page->pageno;
do
{
PAGECACHE_PAGE *page;
thread= next_thread;
page= (PAGECACHE_PAGE *) thread->keycache_link;
next_thread= thread->next;
/*
We notify about the event all threads that ask
for the same page as the first thread in the queue
*/
if (page->file.file == hash_link->file.file &&
page->pageno == hash_link->pageno)
{
DBUG_PRINT("signal", ("thread %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_signal(&thread->suspend);
wqueue_unlink_from_queue(&pagecache->waiting_for_hash_link, thread);
}
}
while (thread != last_thread);
/*
Add this to the hash, so that the waiting threads can find it
when they retry the call to get_hash_link(). This entry is special
in that it has no associated block.
*/
link_hash(&pagecache->hash_root[PAGECACHE_HASH(pagecache,
hash_link->file,
hash_link->pageno)],
hash_link);
DBUG_VOID_RETURN;
}
/* Add hash to free hash list */
hash_link->next= pagecache->free_hash_list;
pagecache->free_hash_list= hash_link;
DBUG_VOID_RETURN;
}
/*
Get the hash link for the page if it is in the cache (do not put the
page in the cache if it is absent there)
SYNOPSIS
get_present_hash_link()
pagecache Pagecache reference
file file ID
pageno page number in the file
start where to put pointer to found hash bucket (for
direct referring it)
RETURN
found hashlink pointer
*/
static PAGECACHE_HASH_LINK *get_present_hash_link(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
PAGECACHE_HASH_LINK ***start)
{
reg1 PAGECACHE_HASH_LINK *hash_link;
#if defined(PAGECACHE_DEBUG)
int cnt;
#endif
DBUG_ENTER("get_present_hash_link");
DBUG_PRINT("enter", ("fd: %u pos: %lu", (uint) file->file, (ulong) pageno));
/*
Find the bucket in the hash table for the pair (file, pageno);
start contains the head of the bucket list,
hash_link points to the first member of the list
*/
hash_link= *(*start= &pagecache->hash_root[PAGECACHE_HASH(pagecache,
*file, pageno)]);
#if defined(PAGECACHE_DEBUG)
cnt= 0;
#endif
/* Look for an element for the pair (file, pageno) in the bucket chain */
while (hash_link &&
(hash_link->pageno != pageno ||
hash_link->file.file != file->file))
{
hash_link= hash_link->next;
#if defined(PAGECACHE_DEBUG)
cnt++;
if (! (cnt <= pagecache->hash_links_used))
{
int i;
for (i=0, hash_link= **start ;
i < cnt ; i++, hash_link= hash_link->next)
{
KEYCACHE_DBUG_PRINT("get_present_hash_link", ("fd: %u pos: %lu",
(uint) hash_link->file.file, (ulong) hash_link->pageno));
}
}
KEYCACHE_DBUG_ASSERT(cnt <= pagecache->hash_links_used);
#endif
}
if (hash_link)
{
DBUG_PRINT("exit", ("hash_link: %p", hash_link));
/* Register the request for the page */
hash_link->requests++;
}
/*
As soon as the caller will release the page cache's lock, "hash_link"
will be potentially obsolete (unusable) information.
*/
DBUG_RETURN(hash_link);
}
/*
Get the hash link for a page
*/
static PAGECACHE_HASH_LINK *get_hash_link(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno)
{
reg1 PAGECACHE_HASH_LINK *hash_link;
PAGECACHE_HASH_LINK **start;
DBUG_ENTER("get_hash_link");
restart:
/* try to find the page in the cache */
hash_link= get_present_hash_link(pagecache, file, pageno,
&start);
if (!hash_link)
{
/* There is no hash link in the hash table for the pair (file, pageno) */
if (pagecache->free_hash_list)
{
DBUG_PRINT("info", ("free_hash_list: %p free_hash_list->next: %p",
pagecache->free_hash_list,
pagecache->free_hash_list->next));
hash_link= pagecache->free_hash_list;
pagecache->free_hash_list= hash_link->next;
}
else if (pagecache->hash_links_used < pagecache->hash_links)
{
hash_link= &pagecache->hash_link_root[pagecache->hash_links_used++];
}
else
{
/* Wait for a free hash link */
struct st_my_thread_var *thread= my_thread_var;
PAGECACHE_PAGE page;
page.file= *file;
page.pageno= pageno;
thread->keycache_link= (void *) &page;
wqueue_link_into_queue(&pagecache->waiting_for_hash_link, thread);
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name, (ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
thread->keycache_link= NULL;
DBUG_PRINT("thread", ("restarting..."));
goto restart;
}
hash_link->file= *file;
DBUG_ASSERT(pageno < ((1ULL) << 40));
hash_link->pageno= pageno;
link_hash(start, hash_link);
/* Register the request for the page */
hash_link->requests++;
DBUG_ASSERT(hash_link->block == 0);
DBUG_ASSERT(hash_link->requests == 1);
}
else
{
/*
We have to copy the flush_log callback, as it may change if the table
goes from non_transactional to transactional during recovery
*/
hash_link->file.flush_log_callback= file->flush_log_callback;
}
DBUG_PRINT("exit", ("hash_link: %p block: %p", hash_link,
hash_link->block));
DBUG_RETURN(hash_link);
}
/*
Get a block for the file page requested by a pagecache read/write operation;
If the page is not in the cache return a free block, if there is none
return the lru block after saving its buffer if the page is dirty.
SYNOPSIS
find_block()
pagecache pointer to a page cache data structure
file handler for the file to read page from
pageno number of the page in the file
init_hits_left how initialize the block counter for the page
wrmode <-> get for writing
block_is_copied 1 if block will be copied from page cache under
the pagelock mutex.
reg_req Register request to the page. Normally all pages
should be registered; The only time it's ok to
not register a page is when the page is already
pinned (and thus registered) by the same thread.
page_st out {PAGE_READ,PAGE_TO_BE_READ,PAGE_WAIT_TO_BE_READ}
RETURN VALUE
Pointer to the found block if successful, 0 - otherwise
NOTES.
For the page from file positioned at pageno the function checks whether
the page is in the key cache specified by the first parameter.
If this is the case it immediately returns the block.
If not, the function first chooses a block for this page. If there is
no not used blocks in the key cache yet, the function takes the block
at the very beginning of the warm sub-chain. It saves the page in that
block if it's dirty before returning the pointer to it.
The function returns in the page_st parameter the following values:
PAGE_READ - if page already in the block,
PAGE_TO_BE_READ - if it is to be read yet by the current thread
WAIT_TO_BE_READ - if it is to be read by another thread
If an error occurs THE PCBLOCK_ERROR bit is set in the block status.
It might happen that there are no blocks in LRU chain (in warm part) -
all blocks are unlinked for some read/write operations. Then the function
waits until first of this operations links any block back.
*/
static PAGECACHE_BLOCK_LINK *find_block(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
int init_hits_left,
my_bool wrmode,
my_bool block_is_copied,
my_bool reg_req,
int *page_st)
{
PAGECACHE_HASH_LINK *hash_link;
PAGECACHE_BLOCK_LINK *block;
int error= 0;
int page_status;
DBUG_ENTER("find_block");
DBUG_PRINT("enter", ("fd: %d pos: %lu wrmode: %d block_is_copied: %d",
file->file, (ulong) pageno, wrmode, block_is_copied));
KEYCACHE_PRINT("find_block", ("fd: %d pos: %lu wrmode: %d",
file->file, (ulong) pageno,
wrmode));
#if !defined(DBUG_OFF) && defined(EXTRA_DEBUG)
DBUG_EXECUTE("check_pagecache",
test_key_cache(pagecache, "start of find_block", 0););
#endif
restart:
/* Find the hash link for the requested page (file, pageno) */
hash_link= get_hash_link(pagecache, file, pageno);
page_status= -1;
if ((block= hash_link->block) &&
block->hash_link == hash_link && (block->status & PCBLOCK_READ))
page_status= PAGE_READ;
if (wrmode && pagecache->resize_in_flush)
{
/* This is a write request during the flush phase of a resize operation */
if (page_status != PAGE_READ)
{
/* We don't need the page in the cache: we are going to write on disk */
DBUG_ASSERT(hash_link->requests > 0);
hash_link->requests--;
unlink_hash(pagecache, hash_link);
return 0;
}
if (!(block->status & PCBLOCK_IN_FLUSH))
{
DBUG_ASSERT(hash_link->requests > 0);
hash_link->requests--;
/*
Remove block to invalidate the page in the block buffer
as we are going to write directly on disk.
Although we have an exclusive lock for the updated key part
the control can be yielded by the current thread as we might
have unfinished readers of other key parts in the block
buffer. Still we are guaranteed not to have any readers
of the key part we are writing into until the block is
removed from the cache as we set the PCBLOCK_REASSIGNED
flag (see the code below that handles reading requests).
*/
free_block(pagecache, block, 0);
return 0;
}
/* Wait until the page is flushed on disk */
DBUG_ASSERT(hash_link->requests > 0);
hash_link->requests--;
wait_for_flush(pagecache, block);
/* Invalidate page in the block if it has not been done yet */
DBUG_ASSERT(block->status); /* Should always be true */
if (block->status)
free_block(pagecache, block, 0);
return 0;
}
if (page_status == PAGE_READ &&
(block->status & (PCBLOCK_IN_SWITCH | PCBLOCK_REASSIGNED)))
{
/* This is a request for a page to be removed from cache */
KEYCACHE_DBUG_PRINT("find_block",
("request for old page in block: %u "
"wrmode: %d block->status: %d",
PCBLOCK_NUMBER(pagecache, block), wrmode,
block->status));
/*
Only reading requests can proceed until the old dirty page is flushed,
all others are to be suspended, then resubmitted
*/
if (!wrmode && block_is_copied && !(block->status & PCBLOCK_REASSIGNED))
{
if (reg_req)
reg_requests(pagecache, block, 1);
}
else
{
/*
When we come here either PCBLOCK_REASSIGNED or PCBLOCK_IN_SWITCH are
active. In both cases wqueue_release_queue() is called when the
state changes.
*/
DBUG_ASSERT(block->hash_link == hash_link);
remove_reader(block);
KEYCACHE_DBUG_PRINT("find_block",
("request waiting for old page to be saved"));
{
struct st_my_thread_var *thread= my_thread_var;
/* Put the request into the queue of those waiting for the old page */
wqueue_add_to_queue(&block->wqueue[COND_FOR_SAVED], thread);
/* Wait until the request can be resubmitted */
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while(thread->next);
}
KEYCACHE_DBUG_PRINT("find_block",
("request for old page resubmitted"));
DBUG_PRINT("info", ("restarting..."));
/* Resubmit the request */
goto restart;
}
}
else
{
/* This is a request for a new page or for a page not to be removed */
if (! block)
{
/* No block is assigned for the page yet */
if (pagecache->blocks_unused)
{
if (pagecache->free_block_list)
{
/* There is a block in the free list. */
block= pagecache->free_block_list;
pagecache->free_block_list= block->next_used;
block->next_used= NULL;
}
else
{
/* There are some never used blocks, take first of them */
block= &pagecache->block_root[pagecache->blocks_used];
block->buffer= ADD_TO_PTR(pagecache->block_mem,
(pagecache->blocks_used*
pagecache->block_size),
uchar*);
pagecache->blocks_used++;
}
pagecache->blocks_unused--;
DBUG_ASSERT(block->wlocks == 0);
DBUG_ASSERT(block->rlocks == 0);
DBUG_ASSERT(block->rlocks_queue == 0);
DBUG_ASSERT(block->pins == 0);
block->status= 0;
#ifndef DBUG_OFF
block->type= PAGECACHE_EMPTY_PAGE;
#endif
DBUG_ASSERT(reg_req);
block->requests= 1;
block->temperature= PCBLOCK_COLD;
block->hits_left= init_hits_left;
block->last_hit_time= 0;
block->rec_lsn= LSN_MAX;
link_to_file_list(pagecache, block, file, 0);
block->hash_link= hash_link;
hash_link->block= block;
page_status= PAGE_TO_BE_READ;
DBUG_PRINT("info", ("page to be read set for page %p (%u)",
block, PCBLOCK_NUMBER(pagecache, block)));
KEYCACHE_PRINT("find_block",
("got free or never used block %u",
PCBLOCK_NUMBER(pagecache, block)));
}
else
{
/* There are no never used blocks, use a block from the LRU chain */
/*
Ensure that we are going to register the block.
(This should be true as a new block could not have been
pinned by caller).
*/
DBUG_ASSERT(reg_req);
if (! pagecache->used_last)
{
/*
Wait until a new block is added to the LRU chain;
several threads might wait here for the same page,
all of them must get the same block.
The block is given to us by the next thread executing
link_block().
*/
struct st_my_thread_var *thread= my_thread_var;
thread->keycache_link= (void *) hash_link;
wqueue_link_into_queue(&pagecache->waiting_for_block, thread);
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while (thread->next);
thread->keycache_link= NULL;
block= hash_link->block;
/* Ensure that the block is registered */
DBUG_ASSERT(block->requests >= 1);
}
else
{
/*
Take the first block from the LRU chain
unlinking it from the chain
*/
block= pagecache->used_last->next_used;
if (reg_req)
reg_requests(pagecache, block, 1);
hash_link->block= block;
DBUG_ASSERT(block->requests == 1);
}
PCBLOCK_INFO(block);
DBUG_ASSERT(block->hash_link == hash_link ||
!(block->status & PCBLOCK_IN_SWITCH));
if (block->hash_link != hash_link &&
! (block->status & PCBLOCK_IN_SWITCH) )
{
/* If another thread is flushing the block, wait for it. */
if (block->status & PCBLOCK_IN_FLUSH)
wait_for_flush(pagecache, block);
/* this is a primary request for a new page */
DBUG_ASSERT(block->wlocks == 0);
DBUG_ASSERT(block->rlocks == 0);
DBUG_ASSERT(block->rlocks_queue == 0);
DBUG_ASSERT(block->pins == 0);
block->status|= PCBLOCK_IN_SWITCH;
KEYCACHE_DBUG_PRINT("find_block",
("got block %u for new page",
PCBLOCK_NUMBER(pagecache, block)));
if (block->status & PCBLOCK_CHANGED)
{
/* The block contains a dirty page - push it out of the cache */
KEYCACHE_DBUG_PRINT("find_block", ("block is dirty"));
/*
The call is thread safe because only the current
thread might change the block->hash_link value
*/
DBUG_ASSERT(block->pins == 0);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
error= pagecache_fwrite(pagecache,
&block->hash_link->file,
block->buffer,
block->hash_link->pageno,
block->type,
pagecache->readwrite_flags);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
pagecache->global_cache_write++;
}
block->status|= PCBLOCK_REASSIGNED;
if (block->hash_link)
{
/*
Wait until all pending read requests
for this page are executed
(we could have avoided this waiting, if we had read
a page in the cache in a sweep, without yielding control)
*/
wait_for_readers(pagecache, block);
/* Remove the hash link for this page from the hash table */
unlink_hash(pagecache, block->hash_link);
/* All pending requests for this page must be resubmitted */
if (block->wqueue[COND_FOR_SAVED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_SAVED]);
}
link_to_file_list(pagecache, block, file,
(my_bool)(block->hash_link ? 1 : 0));
block->hash_link= hash_link;
PCBLOCK_INFO(block);
block->hits_left= init_hits_left;
block->last_hit_time= 0;
block->status= error ? PCBLOCK_ERROR : 0;
block->error= error ? (int16) my_errno : 0;
#ifndef DBUG_OFF
block->type= PAGECACHE_EMPTY_PAGE;
if (error)
my_debug_put_break_here();
#endif
page_status= PAGE_TO_BE_READ;
DBUG_PRINT("info", ("page to be read set for page %p", block));
KEYCACHE_DBUG_ASSERT(block->hash_link->block == block);
KEYCACHE_DBUG_ASSERT(hash_link->block->hash_link == hash_link);
}
else
{
/* This is for secondary requests for a new page only */
KEYCACHE_DBUG_PRINT("find_block",
("block->hash_link: %p hash_link: %p "
"block->status: %u", block->hash_link,
hash_link, block->status ));
page_status= (((block->hash_link == hash_link) &&
(block->status & PCBLOCK_READ)) ?
PAGE_READ : PAGE_WAIT_TO_BE_READ);
}
}
}
else
{
/*
The block was found in the cache. It's either a already read
block or a block waiting to be read by another thread.
*/
if (reg_req)
reg_requests(pagecache, block, 1);
KEYCACHE_DBUG_PRINT("find_block",
("block->hash_link: %p hash_link: %p "
"block->status: %u", block->hash_link,
hash_link, block->status ));
/*
block->hash_link != hash_link can only happen when
the block is in PCBLOCK_IN_SWITCH above (is flushed out
to be replaced by another block). The SWITCH code will change
block->hash_link to point to hash_link.
*/
KEYCACHE_DBUG_ASSERT(block->hash_link == hash_link ||
block->status & PCBLOCK_IN_SWITCH);
page_status= (((block->hash_link == hash_link) &&
(block->status & PCBLOCK_READ)) ?
PAGE_READ : PAGE_WAIT_TO_BE_READ);
}
}
KEYCACHE_DBUG_ASSERT(page_status != -1);
*page_st= page_status;
DBUG_PRINT("info",
("block: %p fd: %u pos: %lu block->status: %u page_status: %u",
block, (uint) file->file,
(ulong) pageno, block->status, (uint) page_status));
KEYCACHE_PRINT("find_block",
("block: %p fd: %d pos: %lu block->status: %u page_status: %d",
block, file->file, (ulong) pageno, block->status,
page_status));
#if !defined(DBUG_OFF) && defined(EXTRA_DEBUG)
DBUG_EXECUTE("check_pagecache",
test_key_cache(pagecache, "end of find_block",0););
#endif
KEYCACHE_THREAD_TRACE("find_block:end");
DBUG_RETURN(block);
}
static void add_pin(PAGECACHE_BLOCK_LINK *block)
{
DBUG_ENTER("add_pin");
DBUG_PRINT("enter", ("block: %p pins: %u", block, block->pins));
PCBLOCK_INFO(block);
block->pins++;
#ifndef DBUG_OFF
{
PAGECACHE_PIN_INFO *info=
(PAGECACHE_PIN_INFO *)my_malloc(sizeof(PAGECACHE_PIN_INFO), MYF(0));
info->thread= my_thread_var;
info_link(&block->pin_list, info);
}
#endif
DBUG_VOID_RETURN;
}
static void remove_pin(PAGECACHE_BLOCK_LINK *block, my_bool any
#ifdef DBUG_OFF
__attribute__((unused))
#endif
)
{
DBUG_ENTER("remove_pin");
DBUG_PRINT("enter", ("block: %p pins: %u any: %d", block, block->pins,
(int)any));
PCBLOCK_INFO(block);
DBUG_ASSERT(block->pins > 0);
block->pins--;
#ifndef DBUG_OFF
{
PAGECACHE_PIN_INFO *info= info_find(block->pin_list, my_thread_var, any);
DBUG_ASSERT(info != 0);
info_unlink(info);
my_free(info);
}
#endif
DBUG_VOID_RETURN;
}
#ifndef DBUG_OFF
static void info_add_lock(PAGECACHE_BLOCK_LINK *block, my_bool wl)
{
PAGECACHE_LOCK_INFO *info=
(PAGECACHE_LOCK_INFO *)my_malloc(sizeof(PAGECACHE_LOCK_INFO), MYF(0));
info->thread= my_thread_var;
info->write_lock= wl;
info_link((PAGECACHE_PIN_INFO **)&block->lock_list,
(PAGECACHE_PIN_INFO *)info);
}
static void info_remove_lock(PAGECACHE_BLOCK_LINK *block)
{
PAGECACHE_LOCK_INFO *info=
(PAGECACHE_LOCK_INFO *)info_find((PAGECACHE_PIN_INFO *)block->lock_list,
my_thread_var, FALSE);
DBUG_ASSERT(info != 0);
info_unlink((PAGECACHE_PIN_INFO *)info);
my_free(info);
}
static void info_change_lock(PAGECACHE_BLOCK_LINK *block, my_bool wl)
{
PAGECACHE_LOCK_INFO *info=
(PAGECACHE_LOCK_INFO *)info_find((PAGECACHE_PIN_INFO *)block->lock_list,
my_thread_var, FALSE);
DBUG_ASSERT(info != 0);
DBUG_ASSERT(info->write_lock != wl);
info->write_lock= wl;
}
#else
#define info_add_lock(B,W)
#define info_remove_lock(B)
#define info_change_lock(B,W)
#endif
/**
@brief waiting for lock for read and write lock
@parem pagecache pointer to a page cache data structure
@parem block the block to work with
@param file file of the block when it was locked
@param pageno page number of the block when it was locked
@param lock_type MY_PTHREAD_LOCK_READ or MY_PTHREAD_LOCK_WRITE
@retval 0 OK
@retval 1 Can't lock this block, need retry
*/
static my_bool pagecache_wait_lock(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
PAGECACHE_FILE file,
pgcache_page_no_t pageno,
uint lock_type)
{
/* Lock failed we will wait */
struct st_my_thread_var *thread= my_thread_var;
DBUG_ENTER("pagecache_wait_lock");
DBUG_PRINT("info", ("fail to lock, waiting... %p", block));
thread->lock_type= lock_type;
wqueue_add_to_queue(&block->wqueue[COND_FOR_WRLOCK], thread);
dec_counter_for_resize_op(pagecache);
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name, (ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while(thread->next);
inc_counter_for_resize_op(pagecache);
PCBLOCK_INFO(block);
if ((block->status & (PCBLOCK_REASSIGNED | PCBLOCK_IN_SWITCH)) ||
!block->hash_link ||
file.file != block->hash_link->file.file ||
pageno != block->hash_link->pageno)
{
DBUG_PRINT("info", ("the block %p changed => need retry "
"status: %x files %d != %d or pages %lu != %lu",
block, block->status, file.file,
block->hash_link ? block->hash_link->file.file : -1,
(ulong) pageno,
(ulong) (block->hash_link ? block->hash_link->pageno : 0)));
DBUG_RETURN(1);
}
DBUG_RETURN(0);
}
/**
@brief Put on the block write lock
@parem pagecache pointer to a page cache data structure
@parem block the block to work with
@note We have loose scheme for locking by the same thread:
* Downgrade to read lock if no other locks are taken
* Our scheme of locking allow for the same thread
- the same kind of lock
- taking read lock if write lock present
- downgrading to read lock if still other place the same
thread keep write lock
* But unlock operation number should be the same to lock operation.
* If we try to get read lock having active write locks we put read
locks to queue, and as soon as write lock(s) gone the read locks
from queue came in force.
* If read lock is unlocked earlier then it came to force it
just removed from the queue
@retval 0 OK
@retval 1 Can't lock this block, need retry
*/
static my_bool get_wrlock(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block)
{
PAGECACHE_FILE file= block->hash_link->file;
pgcache_page_no_t pageno= block->hash_link->pageno;
pthread_t locker= pthread_self();
DBUG_ENTER("get_wrlock");
DBUG_PRINT("info", ("the block %p "
"files %d(%d) pages %lu(%lu)",
block, file.file, block->hash_link->file.file,
(ulong) pageno, (ulong) block->hash_link->pageno));
PCBLOCK_INFO(block);
/*
We assume that the same thread will try write lock on block on which it
has already read lock.
*/
while ((block->wlocks && !pthread_equal(block->write_locker, locker)) ||
block->rlocks)
{
/* Lock failed we will wait */
if (pagecache_wait_lock(pagecache, block, file, pageno,
MY_PTHREAD_LOCK_WRITE))
DBUG_RETURN(1);
}
/* we are doing it by global cache mutex protection, so it is OK */
block->wlocks++;
block->write_locker= locker;
DBUG_PRINT("info", ("WR lock set, block %p", block));
DBUG_RETURN(0);
}
/*
@brief Put on the block read lock
@param pagecache pointer to a page cache data structure
@param block the block to work with
@param user_file Unique handler per handler file. Used to check if
we request many write locks withing the same
statement
@note see note for get_wrlock().
@retvalue 0 OK
@retvalue 1 Can't lock this block, need retry
*/
static my_bool get_rdlock(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block)
{
PAGECACHE_FILE file= block->hash_link->file;
pgcache_page_no_t pageno= block->hash_link->pageno;
pthread_t locker= pthread_self();
DBUG_ENTER("get_rdlock");
DBUG_PRINT("info", ("the block %p "
"files %d(%d) pages %lu(%lu)",
block, file.file, block->hash_link->file.file,
(ulong) pageno, (ulong) block->hash_link->pageno));
PCBLOCK_INFO(block);
while (block->wlocks && !pthread_equal(block->write_locker, locker))
{
/* Lock failed we will wait */
if (pagecache_wait_lock(pagecache, block, file, pageno,
MY_PTHREAD_LOCK_READ))
DBUG_RETURN(1);
}
/* we are doing it by global cache mutex protection, so it is OK */
if (block->wlocks)
{
DBUG_ASSERT(pthread_equal(block->write_locker, locker));
block->rlocks_queue++;
DBUG_PRINT("info", ("RD lock put into queue, block %p", block));
}
else
{
block->rlocks++;
DBUG_PRINT("info", ("RD lock set, block %p", block));
}
DBUG_RETURN(0);
}
/*
@brief Remove write lock from the block
@param pagecache pointer to a page cache data structure
@param block the block to work with
@param read_lock downgrade to read lock
@note see note for get_wrlock().
*/
static void release_wrlock(PAGECACHE_BLOCK_LINK *block, my_bool read_lock)
{
DBUG_ENTER("release_wrlock");
PCBLOCK_INFO(block);
DBUG_ASSERT(block->wlocks > 0);
DBUG_ASSERT(block->rlocks == 0);
DBUG_ASSERT(block->pins > 0);
if (read_lock)
block->rlocks_queue++;
if (block->wlocks == 1)
{
block->rlocks= block->rlocks_queue;
block->rlocks_queue= 0;
}
block->wlocks--;
if (block->wlocks > 0)
DBUG_VOID_RETURN; /* Multiple write locked */
DBUG_PRINT("info", ("WR lock reset, block %p", block));
/* release all threads waiting for read lock or one waiting for write */
if (block->wqueue[COND_FOR_WRLOCK].last_thread)
wqueue_release_one_locktype_from_queue(&block->wqueue[COND_FOR_WRLOCK]);
PCBLOCK_INFO(block);
DBUG_VOID_RETURN;
}
/*
@brief Remove read lock from the block
@param pagecache pointer to a page cache data structure
@param block the block to work with
@note see note for get_wrlock().
*/
static void release_rdlock(PAGECACHE_BLOCK_LINK *block)
{
DBUG_ENTER("release_wrlock");
PCBLOCK_INFO(block);
if (block->wlocks)
{
DBUG_ASSERT(pthread_equal(block->write_locker, pthread_self()));
DBUG_ASSERT(block->rlocks == 0);
DBUG_ASSERT(block->rlocks_queue > 0);
block->rlocks_queue--;
DBUG_PRINT("info", ("RD lock queue decreased, block %p", block));
DBUG_VOID_RETURN;
}
DBUG_ASSERT(block->rlocks > 0);
DBUG_ASSERT(block->rlocks_queue == 0);
block->rlocks--;
DBUG_PRINT("info", ("RD lock decreased, block %p", block));
if (block->rlocks > 0)
DBUG_VOID_RETURN; /* Multiple write locked */
DBUG_PRINT("info", ("RD lock reset, block %p", block));
/* release all threads waiting for read lock or one waiting for write */
if (block->wqueue[COND_FOR_WRLOCK].last_thread)
wqueue_release_one_locktype_from_queue(&block->wqueue[COND_FOR_WRLOCK]);
PCBLOCK_INFO(block);
DBUG_VOID_RETURN;
}
/**
@brief Try to lock/unlock and pin/unpin the block
@param pagecache pointer to a page cache data structure
@param block the block to work with
@param lock lock change mode
@param pin pinchange mode
@param file File handler requesting pin
@param any allow unpinning block pinned by any thread; possible
only if not locked, see pagecache_unlock_by_link()
@retval 0 OK
@retval 1 Try to lock the block failed
*/
static my_bool make_lock_and_pin(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
enum pagecache_page_lock lock,
enum pagecache_page_pin pin,
my_bool any)
{
DBUG_ENTER("make_lock_and_pin");
DBUG_PRINT("enter", ("block: %p (%u) lock: %s pin: %s any %d",
block, PCBLOCK_NUMBER(pagecache, block),
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin], (int)any));
PCBLOCK_INFO(block);
DBUG_ASSERT(block);
DBUG_ASSERT(!any ||
((lock == PAGECACHE_LOCK_LEFT_UNLOCKED) &&
(pin == PAGECACHE_UNPIN)));
DBUG_ASSERT(block->hash_link->block == block);
switch (lock) {
case PAGECACHE_LOCK_WRITE: /* free -> write */
/* Writelock and pin the buffer */
if (get_wrlock(pagecache, block))
{
/* Couldn't lock because block changed status => need retry */
goto retry;
}
/* The cache is locked so nothing afraid of */
add_pin(block);
info_add_lock(block, 1);
break;
case PAGECACHE_LOCK_WRITE_TO_READ: /* write -> read */
case PAGECACHE_LOCK_WRITE_UNLOCK: /* write -> free */
/* Removes write lock and puts read lock */
release_wrlock(block, lock == PAGECACHE_LOCK_WRITE_TO_READ);
/* fall through */
case PAGECACHE_LOCK_READ_UNLOCK: /* read -> free */
if (lock == PAGECACHE_LOCK_READ_UNLOCK)
release_rdlock(block);
/* fall through */
case PAGECACHE_LOCK_LEFT_READLOCKED: /* read -> read */
if (pin == PAGECACHE_UNPIN)
{
remove_pin(block, FALSE);
}
if (lock == PAGECACHE_LOCK_WRITE_TO_READ)
{
info_change_lock(block, 0);
}
else if (lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_READ_UNLOCK)
{
info_remove_lock(block);
}
break;
case PAGECACHE_LOCK_READ: /* free -> read */
if (get_rdlock(pagecache, block))
{
/* Couldn't lock because block changed status => need retry */
goto retry;
}
if (pin == PAGECACHE_PIN)
{
/* The cache is locked so nothing afraid off */
add_pin(block);
}
info_add_lock(block, 0);
break;
case PAGECACHE_LOCK_LEFT_UNLOCKED: /* free -> free */
if (pin == PAGECACHE_UNPIN)
{
remove_pin(block, any);
}
/* fall through */
case PAGECACHE_LOCK_LEFT_WRITELOCKED: /* write -> write */
break; /* do nothing */
default:
DBUG_ASSERT(0); /* Never should happened */
}
PCBLOCK_INFO(block);
DBUG_RETURN(0);
retry:
DBUG_PRINT("INFO", ("Retry block %p", block));
PCBLOCK_INFO(block);
DBUG_ASSERT(block->hash_link->requests > 0);
block->hash_link->requests--;
DBUG_RETURN(1);
}
/*
Read into a key cache block buffer from disk.
SYNOPSIS
read_block()
pagecache pointer to a page cache data structure
block block to which buffer the data is to be read
primary <-> the current thread will read the data
RETURN VALUE
None
NOTES.
The function either reads a page data from file to the block buffer,
or waits until another thread reads it. What page to read is determined
by a block parameter - reference to a hash link for this page.
If an error occurs THE PCBLOCK_ERROR bit is set in the block status.
On entry cache_lock is locked
*/
static void read_block(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
my_bool primary)
{
DBUG_ENTER("read_block");
DBUG_PRINT("enter", ("read block: %p primary: %d", block, primary));
if (primary)
{
size_t error;
PAGECACHE_IO_HOOK_ARGS args;
/*
This code is executed only by threads
that submitted primary requests
*/
pagecache->global_cache_read++;
/*
Page is not in buffer yet, is to be read from disk
Here other threads may step in and register as secondary readers.
They will register in block->wqueue[COND_FOR_REQUESTED].
*/
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
args.page= block->buffer;
args.pageno= block->hash_link->pageno;
args.data= block->hash_link->file.callback_data;
error= (*block->hash_link->file.pre_read_hook)(&args);
if (!error)
{
error= pagecache_fread(pagecache, &block->hash_link->file,
args.page,
block->hash_link->pageno,
pagecache->readwrite_flags);
}
error= (*block->hash_link->file.post_read_hook)(error != 0, &args);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (error)
{
DBUG_ASSERT(maria_in_recovery || !maria_assert_if_crashed_table);
block->status|= PCBLOCK_ERROR;
block->error= (int16) my_errno;
my_debug_put_break_here();
}
else
{
block->status|= PCBLOCK_READ;
}
DBUG_PRINT("read_block",
("primary request: new page in cache"));
/* Signal that all pending requests for this page now can be processed */
if (block->wqueue[COND_FOR_REQUESTED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_REQUESTED]);
}
else
{
/*
This code is executed only by threads
that submitted secondary requests
*/
struct st_my_thread_var *thread= my_thread_var;
/* Put the request into a queue and wait until it can be processed */
wqueue_add_to_queue(&block->wqueue[COND_FOR_REQUESTED], thread);
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while (thread->next);
DBUG_PRINT("read_block",
("secondary request: new page in cache"));
}
DBUG_VOID_RETURN;
}
/**
@brief Set LSN on the page to the given one if the given LSN is bigger
@param pagecache pointer to a page cache data structure
@param lsn LSN to set
@param block block to check and set
*/
static void check_and_set_lsn(PAGECACHE *pagecache,
LSN lsn, PAGECACHE_BLOCK_LINK *block)
{
LSN old;
DBUG_ENTER("check_and_set_lsn");
/*
In recovery, we can _ma_unpin_all_pages() to put a LSN on page, though
page would be PAGECACHE_PLAIN_PAGE (transactionality temporarily disabled
to not log REDOs).
*/
DBUG_ASSERT((block->type == PAGECACHE_LSN_PAGE) || maria_in_recovery);
old= lsn_korr(block->buffer);
DBUG_PRINT("info", ("old lsn: (%lu, 0x%lx) new lsn: (%lu, 0x%lx)",
LSN_IN_PARTS(old), LSN_IN_PARTS(lsn)));
if (cmp_translog_addr(lsn, old) > 0)
{
DBUG_ASSERT(block->type != PAGECACHE_READ_UNKNOWN_PAGE);
lsn_store(block->buffer, lsn);
/* we stored LSN in page so we dirtied it */
if (!(block->status & PCBLOCK_CHANGED))
link_to_changed_list(pagecache, block);
}
DBUG_VOID_RETURN;
}
/**
@brief Unlock/unpin page and put LSN stamp if it need
@param pagecache pointer to a page cache data structure
@pagam file handler for the file for the block of data to be read
@param pageno number of the block of data in the file
@param lock lock change
@param pin pin page
@param first_REDO_LSN_for_page do not set it if it is zero
@param lsn if it is not LSN_IMPOSSIBLE (0) and it
is bigger then LSN on the page it will be written on
the page
@param was_changed should be true if the page was write locked with
direct link giving and the page was changed
@note
Pininig uses requests registration mechanism it works following way:
| beginnig | ending |
| of func. | of func. |
----------------------------+-------------+---------------+
PAGECACHE_PIN_LEFT_PINNED | - | - |
PAGECACHE_PIN_LEFT_UNPINNED | reg request | unreg request |
PAGECACHE_PIN | reg request | - |
PAGECACHE_UNPIN | - | unreg request |
*/
void pagecache_unlock(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
enum pagecache_page_lock lock,
enum pagecache_page_pin pin,
LSN first_REDO_LSN_for_page,
LSN lsn, my_bool was_changed)
{
PAGECACHE_BLOCK_LINK *block;
int page_st;
DBUG_ENTER("pagecache_unlock");
DBUG_PRINT("enter", ("fd: %u page: %lu %s %s",
(uint) file->file, (ulong) pageno,
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin]));
/* we do not allow any lock/pin increasing here */
DBUG_ASSERT(pin != PAGECACHE_PIN);
DBUG_ASSERT(lock != PAGECACHE_LOCK_READ && lock != PAGECACHE_LOCK_WRITE);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
/*
As soon as we keep lock cache can be used, and we have lock because want
to unlock.
*/
DBUG_ASSERT(pagecache->can_be_used);
inc_counter_for_resize_op(pagecache);
/* See NOTE for pagecache_unlock about registering requests */
block= find_block(pagecache, file, pageno, 0, 0, 0,
pin == PAGECACHE_PIN_LEFT_UNPINNED, &page_st);
PCBLOCK_INFO(block);
DBUG_ASSERT(block != 0 && page_st == PAGE_READ);
if (first_REDO_LSN_for_page)
{
DBUG_ASSERT(lock == PAGECACHE_LOCK_WRITE_UNLOCK);
DBUG_ASSERT(pin == PAGECACHE_UNPIN);
pagecache_set_block_rec_lsn(block, first_REDO_LSN_for_page);
}
if (lsn != LSN_IMPOSSIBLE)
check_and_set_lsn(pagecache, lsn, block);
/* if we lock for write we must link the block to changed blocks */
DBUG_ASSERT((block->status & PCBLOCK_DIRECT_W) == 0 ||
(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_WRITE_TO_READ ||
lock == PAGECACHE_LOCK_LEFT_WRITELOCKED));
/*
if was_changed then status should be PCBLOCK_DIRECT_W or marked
as dirty
*/
DBUG_ASSERT(!was_changed || (block->status & PCBLOCK_DIRECT_W) ||
(block->status & PCBLOCK_CHANGED));
if ((block->status & PCBLOCK_DIRECT_W) &&
(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_WRITE_TO_READ))
{
if (!(block->status & PCBLOCK_CHANGED) && was_changed)
link_to_changed_list(pagecache, block);
block->status&= ~PCBLOCK_DIRECT_W;
DBUG_PRINT("info", ("Drop PCBLOCK_DIRECT_W for block: %p", block));
}
if (make_lock_and_pin(pagecache, block, lock, pin, FALSE))
{
DBUG_ASSERT(0); /* should not happend */
}
remove_reader(block);
/*
Link the block into the LRU chain if it's the last submitted request
for the block and block will not be pinned.
See NOTE for pagecache_unlock about registering requests.
*/
if (pin != PAGECACHE_PIN_LEFT_PINNED)
unreg_request(pagecache, block, 1);
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/*
Unpin page
SYNOPSIS
pagecache_unpin()
pagecache pointer to a page cache data structure
file handler for the file for the block of data to be read
pageno number of the block of data in the file
lsn if it is not LSN_IMPOSSIBLE (0) and it
is bigger then LSN on the page it will be written on
the page
*/
void pagecache_unpin(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
LSN lsn)
{
PAGECACHE_BLOCK_LINK *block;
int page_st;
DBUG_ENTER("pagecache_unpin");
DBUG_PRINT("enter", ("fd: %u page: %lu",
(uint) file->file, (ulong) pageno));
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
/*
As soon as we keep lock cache can be used, and we have lock bacause want
aunlock.
*/
DBUG_ASSERT(pagecache->can_be_used);
inc_counter_for_resize_op(pagecache);
/* See NOTE for pagecache_unlock about registering requests */
block= find_block(pagecache, file, pageno, 0, 0, 0, 0, &page_st);
DBUG_ASSERT(block != 0);
DBUG_ASSERT(page_st == PAGE_READ);
/* we can't unpin such page without unlock */
DBUG_ASSERT((block->status & PCBLOCK_DIRECT_W) == 0);
if (lsn != LSN_IMPOSSIBLE)
check_and_set_lsn(pagecache, lsn, block);
/*
we can just unpin only with keeping read lock because:
a) we can't pin without any lock
b) we can't unpin keeping write lock
*/
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_LEFT_READLOCKED,
PAGECACHE_UNPIN, FALSE))
DBUG_ASSERT(0); /* should not happend */
remove_reader(block);
/*
Link the block into the LRU chain if it's the last submitted request
for the block and block will not be pinned.
See NOTE for pagecache_unlock about registering requests
*/
unreg_request(pagecache, block, 1);
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/**
@brief Unlock/unpin page and put LSN stamp if it need
(uses direct block/page pointer)
@param pagecache pointer to a page cache data structure
@param link direct link to page (returned by read or write)
@param lock lock change
@param pin pin page
@param first_REDO_LSN_for_page do not set it if it is LSN_IMPOSSIBLE (0)
@param lsn if it is not LSN_IMPOSSIBLE and it is bigger then
LSN on the page it will be written on the page
@param was_changed should be true if the page was write locked with
direct link giving and the page was changed
@param any allow unpinning block pinned by any thread; possible
only if not locked
@note 'any' is a hack so that _ma_bitmap_unpin_all() is allowed to unpin
non-locked bitmap pages pinned by other threads. Because it always uses
PAGECACHE_LOCK_LEFT_UNLOCKED and PAGECACHE_UNPIN
(see write_changed_bitmap()), the hack is limited to these conditions.
*/
void pagecache_unlock_by_link(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
enum pagecache_page_lock lock,
enum pagecache_page_pin pin,
LSN first_REDO_LSN_for_page,
LSN lsn, my_bool was_changed,
my_bool any)
{
DBUG_ENTER("pagecache_unlock_by_link");
DBUG_PRINT("enter", ("block: %p fd: %u page: %lu changed: %d %s %s",
block, (uint) block->hash_link->file.file,
(ulong) block->hash_link->pageno, was_changed,
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin]));
/*
We do not allow any lock/pin increasing here and page can't be
unpinned because we use direct link.
*/
DBUG_ASSERT(pin != PAGECACHE_PIN);
DBUG_ASSERT(pin != PAGECACHE_PIN_LEFT_UNPINNED);
DBUG_ASSERT(lock != PAGECACHE_LOCK_READ);
DBUG_ASSERT(lock != PAGECACHE_LOCK_WRITE);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (pin == PAGECACHE_PIN_LEFT_UNPINNED &&
lock == PAGECACHE_LOCK_READ_UNLOCK)
{
if (make_lock_and_pin(pagecache, block, lock, pin, FALSE))
DBUG_ASSERT(0); /* should not happend */
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/*
As soon as we keep lock cache can be used, and we have lock because want
unlock.
*/
DBUG_ASSERT(pagecache->can_be_used);
inc_counter_for_resize_op(pagecache);
if (was_changed)
{
if (first_REDO_LSN_for_page != LSN_IMPOSSIBLE)
{
/*
LOCK_READ_UNLOCK is ok here as the page may have first locked
with WRITE lock that was temporarly converted to READ lock before
it's unpinned
*/
DBUG_ASSERT(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_READ_UNLOCK);
DBUG_ASSERT(pin == PAGECACHE_UNPIN);
pagecache_set_block_rec_lsn(block, first_REDO_LSN_for_page);
}
if (lsn != LSN_IMPOSSIBLE)
check_and_set_lsn(pagecache, lsn, block);
/*
Reset error flag. Mark also that page is active; This may not have
been the case if there was an error reading the page
*/
block->status= (block->status & ~PCBLOCK_ERROR) | PCBLOCK_READ;
}
/* if we lock for write we must link the block to changed blocks */
DBUG_ASSERT((block->status & PCBLOCK_DIRECT_W) == 0 ||
(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_WRITE_TO_READ ||
lock == PAGECACHE_LOCK_LEFT_WRITELOCKED));
/*
If was_changed then status should be PCBLOCK_DIRECT_W or marked
as dirty
*/
DBUG_ASSERT(!was_changed || (block->status & PCBLOCK_DIRECT_W) ||
(block->status & PCBLOCK_CHANGED));
if ((block->status & PCBLOCK_DIRECT_W) &&
(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_WRITE_TO_READ))
{
if (!(block->status & PCBLOCK_CHANGED) && was_changed)
link_to_changed_list(pagecache, block);
block->status&= ~PCBLOCK_DIRECT_W;
DBUG_PRINT("info", ("Drop PCBLOCK_DIRECT_W for block: %p", block));
}
if (make_lock_and_pin(pagecache, block, lock, pin, any))
DBUG_ASSERT(0); /* should not happend */
/*
Link the block into the LRU chain if it's the last submitted request
for the block and block will not be pinned.
See NOTE for pagecache_unlock about registering requests.
*/
if (pin != PAGECACHE_PIN_LEFT_PINNED)
unreg_request(pagecache, block, 1);
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/*
Unpin page
(uses direct block/page pointer)
SYNOPSIS
pagecache_unpin_by_link()
pagecache pointer to a page cache data structure
link direct link to page (returned by read or write)
lsn if it is not LSN_IMPOSSIBLE (0) and it
is bigger then LSN on the page it will be written on
the page
*/
void pagecache_unpin_by_link(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
LSN lsn)
{
DBUG_ENTER("pagecache_unpin_by_link");
DBUG_PRINT("enter", ("block: %p fd: %u page: %lu",
block, (uint) block->hash_link->file.file,
(ulong) block->hash_link->pageno));
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
/*
As soon as we keep lock cache can be used, and we have lock because want
unlock.
*/
DBUG_ASSERT(pagecache->can_be_used);
/* we can't unpin such page without unlock */
DBUG_ASSERT((block->status & PCBLOCK_DIRECT_W) == 0);
inc_counter_for_resize_op(pagecache);
if (lsn != LSN_IMPOSSIBLE)
check_and_set_lsn(pagecache, lsn, block);
/*
We can just unpin only with keeping read lock because:
a) we can't pin without any lock
b) we can't unpin keeping write lock
*/
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_LEFT_READLOCKED,
PAGECACHE_UNPIN, FALSE))
DBUG_ASSERT(0); /* should not happend */
/*
Link the block into the LRU chain if it's the last submitted request
for the block and block will not be pinned.
See NOTE for pagecache_unlock about registering requests.
*/
unreg_request(pagecache, block, 1);
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_VOID_RETURN;
}
/* description of how to change lock before and after read/write */
struct rw_lock_change
{
my_bool need_lock_change; /* need changing of lock at the end */
enum pagecache_page_lock new_lock; /* lock at the beginning */
enum pagecache_page_lock unlock_lock; /* lock at the end */
};
/* description of how to change pin before and after read/write */
struct rw_pin_change
{
enum pagecache_page_pin new_pin; /* pin status at the beginning */
enum pagecache_page_pin unlock_pin; /* pin status at the end */
};
/**
Depending on the lock which the user wants in pagecache_read(), we
need to acquire a first type of lock at start of pagecache_read(), and
downgrade it to a second type of lock at end. For example, if user
asked for no lock (PAGECACHE_LOCK_LEFT_UNLOCKED) this translates into
taking first a read lock PAGECACHE_LOCK_READ (to rightfully block on
existing write locks) then read then unlock the lock i.e. change lock
to PAGECACHE_LOCK_READ_UNLOCK (the "1" below tells that a change is
needed).
*/
static struct rw_lock_change lock_to_read[8]=
{
{ /*PAGECACHE_LOCK_LEFT_UNLOCKED*/
1,
PAGECACHE_LOCK_READ, PAGECACHE_LOCK_READ_UNLOCK
},
{ /*PAGECACHE_LOCK_LEFT_READLOCKED*/
0,
PAGECACHE_LOCK_LEFT_READLOCKED, PAGECACHE_LOCK_LEFT_READLOCKED
},
{ /*PAGECACHE_LOCK_LEFT_WRITELOCKED*/
0,
PAGECACHE_LOCK_LEFT_WRITELOCKED, PAGECACHE_LOCK_LEFT_WRITELOCKED
},
{ /*PAGECACHE_LOCK_READ*/
1,
PAGECACHE_LOCK_READ, PAGECACHE_LOCK_LEFT_READLOCKED
},
{ /*PAGECACHE_LOCK_WRITE*/
1,
PAGECACHE_LOCK_WRITE, PAGECACHE_LOCK_LEFT_WRITELOCKED
},
{ /*PAGECACHE_LOCK_READ_UNLOCK*/
1,
PAGECACHE_LOCK_LEFT_READLOCKED, PAGECACHE_LOCK_READ_UNLOCK
},
{ /*PAGECACHE_LOCK_WRITE_UNLOCK*/
1,
PAGECACHE_LOCK_LEFT_WRITELOCKED, PAGECACHE_LOCK_WRITE_UNLOCK
},
{ /*PAGECACHE_LOCK_WRITE_TO_READ*/
1,
PAGECACHE_LOCK_LEFT_WRITELOCKED, PAGECACHE_LOCK_WRITE_TO_READ
}
};
/**
Two sets of pin modes (every as for lock upper but for pinning). The
difference between sets if whether we are going to provide caller with
reference on the block or not
*/
static struct rw_pin_change lock_to_pin[2][8]=
{
{
{ /*PAGECACHE_LOCK_LEFT_UNLOCKED*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED
},
{ /*PAGECACHE_LOCK_LEFT_READLOCKED*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED,
},
{ /*PAGECACHE_LOCK_LEFT_WRITELOCKED*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_PIN_LEFT_PINNED
},
{ /*PAGECACHE_LOCK_READ*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED
},
{ /*PAGECACHE_LOCK_WRITE*/
PAGECACHE_PIN,
PAGECACHE_PIN_LEFT_PINNED
},
{ /*PAGECACHE_LOCK_READ_UNLOCK*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED
},
{ /*PAGECACHE_LOCK_WRITE_UNLOCK*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_UNPIN
},
{ /*PAGECACHE_LOCK_WRITE_TO_READ*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_UNPIN
}
},
{
{ /*PAGECACHE_LOCK_LEFT_UNLOCKED*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED
},
{ /*PAGECACHE_LOCK_LEFT_READLOCKED*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED,
},
{ /*PAGECACHE_LOCK_LEFT_WRITELOCKED*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_PIN_LEFT_PINNED
},
{ /*PAGECACHE_LOCK_READ*/
PAGECACHE_PIN,
PAGECACHE_PIN_LEFT_PINNED
},
{ /*PAGECACHE_LOCK_WRITE*/
PAGECACHE_PIN,
PAGECACHE_PIN_LEFT_PINNED
},
{ /*PAGECACHE_LOCK_READ_UNLOCK*/
PAGECACHE_PIN_LEFT_UNPINNED,
PAGECACHE_PIN_LEFT_UNPINNED
},
{ /*PAGECACHE_LOCK_WRITE_UNLOCK*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_UNPIN
},
{ /*PAGECACHE_LOCK_WRITE_TO_READ*/
PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_PIN_LEFT_PINNED,
}
}
};
/*
@brief Read a block of data from a cached file into a buffer;
@param pagecache pointer to a page cache data structure
@param file handler for the file for the block of data to be read
@param pageno number of the block of data in the file
@param level determines the weight of the data
@param buff buffer to where the data must be placed
@param type type of the page
@param lock lock change
@param link link to the page if we pin it
@return address from where the data is placed if successful, 0 - otherwise.
@note Pin will be chosen according to lock parameter (see lock_to_pin)
@note 'buff', if not NULL, must be long-aligned.
@note If buff==0 then we provide reference on the page so should keep the
page pinned.
*/
uchar *pagecache_read(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
uint level,
uchar *buff,
enum pagecache_page_type type,
enum pagecache_page_lock lock,
PAGECACHE_BLOCK_LINK **page_link)
{
my_bool error= 0;
enum pagecache_page_pin
new_pin= lock_to_pin[buff==0][lock].new_pin,
unlock_pin= lock_to_pin[buff==0][lock].unlock_pin;
PAGECACHE_BLOCK_LINK *fake_link;
my_bool reg_request;
#ifndef DBUG_OFF
char llbuf[22];
DBUG_ENTER("pagecache_read");
DBUG_PRINT("enter", ("fd: %u page: %s buffer: %p level: %u "
"t:%s (%d)%s->%s %s->%s",
(uint) file->file, ullstr(pageno, llbuf),
buff, level,
page_cache_page_type_str[type],
lock_to_read[lock].need_lock_change,
page_cache_page_lock_str[lock_to_read[lock].new_lock],
page_cache_page_lock_str[lock_to_read[lock].unlock_lock],
page_cache_page_pin_str[new_pin],
page_cache_page_pin_str[unlock_pin]));
DBUG_ASSERT(buff != 0 || (buff == 0 && (unlock_pin == PAGECACHE_PIN ||
unlock_pin == PAGECACHE_PIN_LEFT_PINNED)));
DBUG_ASSERT(pageno < ((1ULL) << 40));
#endif
if (!page_link)
page_link= &fake_link;
*page_link= 0; /* Catch errors */
restart:
if (pagecache->can_be_used)
{
/* Key cache is used */
PAGECACHE_BLOCK_LINK *block;
uint status;
int UNINIT_VAR(page_st);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (!pagecache->can_be_used)
{
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
goto no_key_cache;
}
inc_counter_for_resize_op(pagecache);
pagecache->global_cache_r_requests++;
/* See NOTE for pagecache_unlock about registering requests. */
reg_request= ((new_pin == PAGECACHE_PIN_LEFT_UNPINNED) ||
(new_pin == PAGECACHE_PIN));
block= find_block(pagecache, file, pageno, level,
lock == PAGECACHE_LOCK_WRITE, buff != 0,
reg_request, &page_st);
DBUG_PRINT("info", ("Block type: %s current type %s",
page_cache_page_type_str[block->type],
page_cache_page_type_str[type]));
if (((block->status & PCBLOCK_ERROR) == 0) && (page_st != PAGE_READ))
{
/* The requested page is to be read into the block buffer */
read_block(pagecache, block,
(my_bool)(page_st == PAGE_TO_BE_READ));
DBUG_PRINT("info", ("read is done"));
}
/*
Assert after block is read. Imagine two concurrent SELECTs on same
table (thread1 and 2), which want to pagecache_read() the same
pageno/fileno. Thread1 calls find_block(), decides to evict a dirty
page from LRU; while it's writing this dirty page to disk, it is
pre-empted and thread2 runs its find_block(), gets the block (in
PAGE_TO_BE_READ state). This block is still containing the in-eviction
dirty page so has an its type, which cannot be tested.
So thread2 has to wait for read_block() to finish (when it wakes up in
read_block(), it's woken up by read_block() of thread1, which implies
that block's type was set to EMPTY by thread1 as part of find_block()).
*/
DBUG_ASSERT(block->type == PAGECACHE_EMPTY_PAGE ||
block->type == type ||
type == PAGECACHE_LSN_PAGE ||
type == PAGECACHE_READ_UNKNOWN_PAGE ||
block->type == PAGECACHE_READ_UNKNOWN_PAGE);
if (type != PAGECACHE_READ_UNKNOWN_PAGE ||
block->type == PAGECACHE_EMPTY_PAGE)
block->type= type;
if (make_lock_and_pin(pagecache, block, lock_to_read[lock].new_lock,
new_pin, FALSE))
{
/*
We failed to write lock the block, cache is unlocked,
we will try to get the block again.
*/
if (reg_request)
unreg_request(pagecache, block, 1);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_PRINT("info", ("restarting..."));
goto restart;
}
status= block->status;
if (!buff)
{
buff= block->buffer;
/* possibly we will write here (resolved on unlock) */
if ((lock == PAGECACHE_LOCK_WRITE ||
lock == PAGECACHE_LOCK_LEFT_WRITELOCKED))
{
block->status|= PCBLOCK_DIRECT_W;
DBUG_PRINT("info", ("Set PCBLOCK_DIRECT_W for block: %p", block));
}
}
else
{
if (status & PCBLOCK_READ)
{
#if !defined(SERIALIZED_READ_FROM_CACHE)
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
#endif
DBUG_ASSERT((pagecache->block_size & 511) == 0);
/* Copy data from the cache buffer */
memcpy(buff, block->buffer, pagecache->block_size);
#if !defined(SERIALIZED_READ_FROM_CACHE)
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
#endif
}
if (status & PCBLOCK_ERROR)
my_errno= block->error;
}
remove_reader(block);
if (lock_to_read[lock].need_lock_change)
{
if (make_lock_and_pin(pagecache, block,
lock_to_read[lock].unlock_lock,
unlock_pin, FALSE))
{
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_ASSERT(0);
return (uchar*) 0;
}
}
/*
Link the block into the LRU chain if it's the last submitted request
for the block and block will not be pinned.
See NOTE for pagecache_unlock about registering requests.
*/
if (unlock_pin == PAGECACHE_PIN_LEFT_UNPINNED ||
unlock_pin == PAGECACHE_UNPIN)
unreg_request(pagecache, block, 1);
else
*page_link= block;
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
if (status & PCBLOCK_ERROR)
{
DBUG_ASSERT(my_errno != 0);
DBUG_PRINT("error", ("Got error %d when doing page read", my_errno));
DBUG_RETURN((uchar *) 0);
}
DBUG_RETURN(buff);
}
no_key_cache: /* Key cache is not used */
/* We can't use mutex here as the key cache may not be initialized */
pagecache->global_cache_r_requests++;
pagecache->global_cache_read++;
{
PAGECACHE_IO_HOOK_ARGS args;
args.page= buff;
args.pageno= pageno;
args.data= file->callback_data;
error= (* file->pre_read_hook)(&args);
if (!error)
{
error= pagecache_fread(pagecache, file, args.page, pageno,
pagecache->readwrite_flags) != 0;
}
error= (* file->post_read_hook)(error, &args);
}
DBUG_RETURN(error ? (uchar*) 0 : buff);
}
/*
@brief Set/reset flag that page always should be flushed on delete
@param pagecache pointer to a page cache data structure
@param link direct link to page (returned by read or write)
@param write write on delete flag value
*/
void pagecache_set_write_on_delete_by_link(PAGECACHE_BLOCK_LINK *block)
{
DBUG_ENTER("pagecache_set_write_on_delete_by_link");
DBUG_PRINT("enter", ("fd: %d block %p %d -> TRUE",
block->hash_link->file.file,
block, (int) block->status & PCBLOCK_DEL_WRITE));
DBUG_ASSERT(block->pins); /* should be pinned */
DBUG_ASSERT(block->wlocks); /* should be write locked */
block->status|= PCBLOCK_DEL_WRITE;
DBUG_VOID_RETURN;
}
/*
@brief Delete page from the buffer (common part for link and file/page)
@param pagecache pointer to a page cache data structure
@param block direct link to page (returned by read or write)
@param page_link hash link of the block
@param flush flush page if it is dirty
@retval 0 deleted or was not present at all
@retval 1 error
*/
static my_bool pagecache_delete_internal(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
PAGECACHE_HASH_LINK *page_link,
my_bool flush)
{
my_bool error= 0;
if (block->status & PCBLOCK_IN_FLUSH)
{
/*
this call is just 'hint' for the cache to free the page so we will
not interferes with flushing process but must return success
*/
goto out;
}
if (block->status & PCBLOCK_CHANGED)
{
flush= (flush || (block->status & PCBLOCK_DEL_WRITE));
if (flush)
{
/* The block contains a dirty page - push it out of the cache */
KEYCACHE_DBUG_PRINT("find_block", ("block is dirty"));
/*
The call is thread safe because only the current
thread might change the block->hash_link value
*/
DBUG_ASSERT(block->pins == 1);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
error= pagecache_fwrite(pagecache,
&block->hash_link->file,
block->buffer,
block->hash_link->pageno,
block->type,
pagecache->readwrite_flags);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
pagecache->global_cache_write++;
if (error)
{
block->status|= PCBLOCK_ERROR;
block->error= (int16) my_errno;
my_debug_put_break_here();
goto out;
}
}
else
{
PAGECACHE_IO_HOOK_ARGS args;
PAGECACHE_FILE *filedesc= &block->hash_link->file;
args.page= block->buffer;
args.pageno= block->hash_link->pageno;
args.data= filedesc->callback_data;
/* We are not going to write the page but have to call callbacks */
DBUG_PRINT("info", ("flush_callback: %p data: %p",
filedesc->flush_log_callback,
filedesc->callback_data));
if ((*filedesc->flush_log_callback)(&args))
{
DBUG_PRINT("error", ("flush or write callback problem"));
error= 1;
goto out;
}
}
pagecache->blocks_changed--;
pagecache->global_blocks_changed--;
/*
free_block() will change the status and rec_lsn of the block so no
need to change them here.
*/
}
/* Cache is locked, so we can relese page before freeing it */
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_WRITE_UNLOCK,
PAGECACHE_UNPIN, FALSE))
DBUG_ASSERT(0);
DBUG_ASSERT(block->hash_link->requests > 0);
page_link->requests--;
/* See NOTE for pagecache_unlock() about registering requests. */
free_block(pagecache, block, 0);
dec_counter_for_resize_op(pagecache);
return 0;
out:
/* Cache is locked, so we can relese page before freeing it */
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_WRITE_UNLOCK,
PAGECACHE_UNPIN, FALSE))
DBUG_ASSERT(0);
page_link->requests--;
unreg_request(pagecache, block, 1);
dec_counter_for_resize_op(pagecache);
return error;
}
/*
@brief Delete page from the buffer by link
@param pagecache pointer to a page cache data structure
@param link direct link to page (returned by read or write)
@param lock lock change
@param flush flush page if it is dirty
@retval 0 deleted or was not present at all
@retval 1 error
@note lock can be only PAGECACHE_LOCK_LEFT_WRITELOCKED (page was
write locked before) or PAGECACHE_LOCK_WRITE (delete will write
lock page before delete)
*/
my_bool pagecache_delete_by_link(PAGECACHE *pagecache,
PAGECACHE_BLOCK_LINK *block,
enum pagecache_page_lock lock,
my_bool flush)
{
my_bool error= 0;
enum pagecache_page_pin pin= PAGECACHE_PIN_LEFT_PINNED;
DBUG_ENTER("pagecache_delete_by_link");
DBUG_PRINT("enter", ("fd: %d block %p %s %s",
block->hash_link->file.file,
block,
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin]));
DBUG_ASSERT(lock == PAGECACHE_LOCK_WRITE ||
lock == PAGECACHE_LOCK_LEFT_WRITELOCKED);
DBUG_ASSERT(block->pins != 0); /* should be pinned */
if (pagecache->can_be_used)
{
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (!pagecache->can_be_used)
goto end;
/*
This block should be pinned (i.e. has not zero request counter) =>
Such block can't be chosen for eviction.
*/
DBUG_ASSERT((block->status &
(PCBLOCK_IN_SWITCH | PCBLOCK_REASSIGNED)) == 0);
inc_counter_for_resize_op(pagecache);
/*
make_lock_and_pin() can't fail here, because we are keeping pin on the
block and it can't be evicted (which is cause of lock fail and retry)
*/
if (make_lock_and_pin(pagecache, block, lock, pin, FALSE))
DBUG_ASSERT(0);
/*
get_present_hash_link() side effect emulation before call
pagecache_delete_internal()
*/
block->hash_link->requests++;
error= pagecache_delete_internal(pagecache, block, block->hash_link,
flush);
end:
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
}
DBUG_RETURN(error);
}
/**
@brief Returns "hits" for promotion
@return "hits" for promotion
*/
uint pagecache_pagelevel(PAGECACHE_BLOCK_LINK *block)
{
return block->hits_left;
}
/*
@brief Adds "hits" to the page
@param link direct link to page (returned by read or write)
@param level number of "hits" which we add to the page
*/
void pagecache_add_level_by_link(PAGECACHE_BLOCK_LINK *block,
uint level)
{
DBUG_ASSERT(block->pins != 0); /* should be pinned */
/*
Operation is just for statistics so it is not really important
if it interfere with other hit increasing => we are doing it without
locking the pagecache.
*/
block->hits_left+= level;
}
/*
@brief Delete page from the buffer
@param pagecache pointer to a page cache data structure
@param file handler for the file for the block of data to be read
@param pageno number of the block of data in the file
@param lock lock change
@param flush flush page if it is dirty
@retval 0 deleted or was not present at all
@retval 1 error
@note lock can be only PAGECACHE_LOCK_LEFT_WRITELOCKED (page was
write locked before) or PAGECACHE_LOCK_WRITE (delete will write
lock page before delete)
*/
static enum pagecache_page_pin lock_to_pin_one_phase[8]=
{
PAGECACHE_PIN_LEFT_UNPINNED /*PAGECACHE_LOCK_LEFT_UNLOCKED*/,
PAGECACHE_PIN_LEFT_UNPINNED /*PAGECACHE_LOCK_LEFT_READLOCKED*/,
PAGECACHE_PIN_LEFT_PINNED /*PAGECACHE_LOCK_LEFT_WRITELOCKED*/,
PAGECACHE_PIN_LEFT_UNPINNED /*PAGECACHE_LOCK_READ*/,
PAGECACHE_PIN /*PAGECACHE_LOCK_WRITE*/,
PAGECACHE_PIN_LEFT_UNPINNED /*PAGECACHE_LOCK_READ_UNLOCK*/,
PAGECACHE_UNPIN /*PAGECACHE_LOCK_WRITE_UNLOCK*/,
PAGECACHE_UNPIN /*PAGECACHE_LOCK_WRITE_TO_READ*/
};
my_bool pagecache_delete(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
enum pagecache_page_lock lock,
my_bool flush)
{
my_bool error= 0;
enum pagecache_page_pin pin= lock_to_pin_one_phase[lock];
DBUG_ENTER("pagecache_delete");
DBUG_PRINT("enter", ("fd: %u page: %lu %s %s",
(uint) file->file, (ulong) pageno,
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin]));
DBUG_ASSERT(lock == PAGECACHE_LOCK_WRITE ||
lock == PAGECACHE_LOCK_LEFT_WRITELOCKED);
DBUG_ASSERT(pin == PAGECACHE_PIN ||
pin == PAGECACHE_PIN_LEFT_PINNED);
restart:
DBUG_ASSERT(pageno < ((1ULL) << 40));
if (pagecache->can_be_used)
{
/* Key cache is used */
reg1 PAGECACHE_BLOCK_LINK *block;
PAGECACHE_HASH_LINK **unused_start, *page_link;
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (!pagecache->can_be_used)
goto end;
inc_counter_for_resize_op(pagecache);
page_link= get_present_hash_link(pagecache, file, pageno, &unused_start);
if (!page_link)
{
DBUG_PRINT("info", ("There is no such page in the cache"));
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_RETURN(0);
}
block= page_link->block;
if (block->status & (PCBLOCK_REASSIGNED | PCBLOCK_IN_SWITCH))
{
DBUG_PRINT("info", ("Block 0x%0lx already is %s",
(ulong) block,
((block->status & PCBLOCK_REASSIGNED) ?
"reassigned" : "in switch")));
PCBLOCK_INFO(block);
page_link->requests--;
dec_counter_for_resize_op(pagecache);
goto end;
}
/* See NOTE for pagecache_unlock about registering requests. */
if (pin == PAGECACHE_PIN)
reg_requests(pagecache, block, 1);
if (make_lock_and_pin(pagecache, block, lock, pin, FALSE))
{
/*
We failed to writelock the block, cache is unlocked, and last write
lock is released, we will try to get the block again.
*/
if (pin == PAGECACHE_PIN)
unreg_request(pagecache, block, 1);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_PRINT("info", ("restarting..."));
goto restart;
}
/* we can't delete with opened direct link for write */
DBUG_ASSERT((block->status & PCBLOCK_DIRECT_W) == 0);
error= pagecache_delete_internal(pagecache, block, page_link, flush);
end:
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
}
DBUG_RETURN(error);
}
my_bool pagecache_delete_pages(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
uint page_count,
enum pagecache_page_lock lock,
my_bool flush)
{
pgcache_page_no_t page_end;
DBUG_ENTER("pagecache_delete_pages");
DBUG_ASSERT(page_count > 0);
page_end= pageno + page_count;
do
{
if (pagecache_delete(pagecache, file, pageno,
lock, flush))
DBUG_RETURN(1);
} while (++pageno != page_end);
DBUG_RETURN(0);
}
/**
@brief Writes a buffer into a cached file.
@param pagecache pointer to a page cache data structure
@param file handler for the file to write data to
@param pageno number of the block of data in the file
@param level determines the weight of the data
@param buff buffer with the data
@param type type of the page
@param lock lock change
@param pin pin page
@param write_mode how to write page
@param link link to the page if we pin it
@param first_REDO_LSN_for_page the lsn to set rec_lsn
@param offset offset in the page
@param size size of data
@param validator read page validator
@param validator_data the validator data
@retval 0 if a success.
@retval 1 Error.
*/
static struct rw_lock_change write_lock_change_table[]=
{
{1,
PAGECACHE_LOCK_WRITE,
PAGECACHE_LOCK_WRITE_UNLOCK} /*PAGECACHE_LOCK_LEFT_UNLOCKED*/,
{0, /*unsupported (we can't write having the block read locked) */
PAGECACHE_LOCK_LEFT_UNLOCKED,
PAGECACHE_LOCK_LEFT_UNLOCKED} /*PAGECACHE_LOCK_LEFT_READLOCKED*/,
{0, PAGECACHE_LOCK_LEFT_WRITELOCKED, 0} /*PAGECACHE_LOCK_LEFT_WRITELOCKED*/,
{1,
PAGECACHE_LOCK_WRITE,
PAGECACHE_LOCK_WRITE_TO_READ} /*PAGECACHE_LOCK_READ*/,
{0, PAGECACHE_LOCK_WRITE, 0} /*PAGECACHE_LOCK_WRITE*/,
{0, /*unsupported (we can't write having the block read locked) */
PAGECACHE_LOCK_LEFT_UNLOCKED,
PAGECACHE_LOCK_LEFT_UNLOCKED} /*PAGECACHE_LOCK_READ_UNLOCK*/,
{1,
PAGECACHE_LOCK_LEFT_WRITELOCKED,
PAGECACHE_LOCK_WRITE_UNLOCK } /*PAGECACHE_LOCK_WRITE_UNLOCK*/,
{1,
PAGECACHE_LOCK_LEFT_WRITELOCKED,
PAGECACHE_LOCK_WRITE_TO_READ} /*PAGECACHE_LOCK_WRITE_TO_READ*/
};
static struct rw_pin_change write_pin_change_table[]=
{
{PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_PIN_LEFT_PINNED} /*PAGECACHE_PIN_LEFT_PINNED*/,
{PAGECACHE_PIN,
PAGECACHE_UNPIN} /*PAGECACHE_PIN_LEFT_UNPINNED*/,
{PAGECACHE_PIN,
PAGECACHE_PIN_LEFT_PINNED} /*PAGECACHE_PIN*/,
{PAGECACHE_PIN_LEFT_PINNED,
PAGECACHE_UNPIN} /*PAGECACHE_UNPIN*/
};
/**
@note 'buff', if not NULL, must be long-aligned.
*/
my_bool pagecache_write_part(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
pgcache_page_no_t pageno,
uint level,
uchar *buff,
enum pagecache_page_type type,
enum pagecache_page_lock lock,
enum pagecache_page_pin pin,
enum pagecache_write_mode write_mode,
PAGECACHE_BLOCK_LINK **page_link,
LSN first_REDO_LSN_for_page,
uint offset, uint size)
{
PAGECACHE_BLOCK_LINK *block= NULL;
PAGECACHE_BLOCK_LINK *fake_link;
my_bool error= 0;
int need_lock_change= write_lock_change_table[lock].need_lock_change;
my_bool reg_request;
#ifndef DBUG_OFF
char llbuf[22];
DBUG_ENTER("pagecache_write_part");
DBUG_PRINT("enter", ("fd: %u page: %s level: %u type: %s lock: %s "
"pin: %s mode: %s offset: %u size %u",
(uint) file->file, ullstr(pageno, llbuf), level,
page_cache_page_type_str[type],
page_cache_page_lock_str[lock],
page_cache_page_pin_str[pin],
page_cache_page_write_mode_str[write_mode],
offset, size));
DBUG_ASSERT(type != PAGECACHE_READ_UNKNOWN_PAGE);
DBUG_ASSERT(lock != PAGECACHE_LOCK_LEFT_READLOCKED);
DBUG_ASSERT(lock != PAGECACHE_LOCK_READ_UNLOCK);
DBUG_ASSERT(offset + size <= pagecache->block_size);
DBUG_ASSERT(pageno < ((1ULL) << 40));
#endif
if (!page_link)
page_link= &fake_link;
*page_link= 0;
restart:
#if !defined(DBUG_OFF) && defined(EXTRA_DEBUG)
DBUG_EXECUTE("check_pagecache",
test_key_cache(pagecache, "start of key_cache_write", 1););
#endif
if (pagecache->can_be_used)
{
/* Key cache is used */
int page_st;
my_bool need_page_ready_signal= FALSE;
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (!pagecache->can_be_used)
{
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
goto no_key_cache;
}
inc_counter_for_resize_op(pagecache);
pagecache->global_cache_w_requests++;
/*
Here we register a request if the page was not already pinned.
See NOTE for pagecache_unlock about registering requests.
*/
reg_request= ((pin == PAGECACHE_PIN_LEFT_UNPINNED) ||
(pin == PAGECACHE_PIN));
block= find_block(pagecache, file, pageno, level,
TRUE, FALSE,
reg_request, &page_st);
if (!block)
{
DBUG_ASSERT(write_mode != PAGECACHE_WRITE_DONE);
/* It happens only for requests submitted during resize operation */
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
/* Write to the disk key cache is in resize at the moment*/
goto no_key_cache;
}
DBUG_PRINT("info", ("page status: %d", page_st));
if (!(block->status & PCBLOCK_ERROR) &&
((page_st == PAGE_TO_BE_READ &&
(offset || size < pagecache->block_size)) ||
(page_st == PAGE_WAIT_TO_BE_READ)))
{
/* The requested page is to be read into the block buffer */
read_block(pagecache, block,
(my_bool)(page_st == PAGE_TO_BE_READ));
DBUG_PRINT("info", ("read is done"));
}
else if (page_st == PAGE_TO_BE_READ)
{
need_page_ready_signal= TRUE;
}
DBUG_ASSERT(block->type == PAGECACHE_EMPTY_PAGE ||
block->type == PAGECACHE_READ_UNKNOWN_PAGE ||
block->type == type ||
/* this is for when going to non-trans to trans */
(block->type == PAGECACHE_PLAIN_PAGE &&
type == PAGECACHE_LSN_PAGE));
block->type= type;
/* we write to the page so it has no sense to keep the flag */
block->status&= ~PCBLOCK_DIRECT_W;
DBUG_PRINT("info", ("Drop PCBLOCK_DIRECT_W for block: %p", block));
if (make_lock_and_pin(pagecache, block,
write_lock_change_table[lock].new_lock,
(need_lock_change ?
write_pin_change_table[pin].new_pin :
pin), FALSE))
{
/*
We failed to writelock the block, cache is unlocked, and last write
lock is released, we will try to get the block again.
*/
if (reg_request)
unreg_request(pagecache, block, 1);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_PRINT("info", ("restarting..."));
goto restart;
}
if (write_mode == PAGECACHE_WRITE_DONE)
{
if (block->status & PCBLOCK_ERROR)
{
my_debug_put_break_here();
DBUG_PRINT("warning", ("Writing on page with error"));
}
else
{
/* Copy data from buff */
memcpy(block->buffer + offset, buff, size);
block->status= PCBLOCK_READ;
KEYCACHE_DBUG_PRINT("key_cache_insert",
("Page injection"));
/* Signal that all pending requests for this now can be processed. */
if (block->wqueue[COND_FOR_REQUESTED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_REQUESTED]);
}
}
else
{
if (! (block->status & PCBLOCK_CHANGED))
link_to_changed_list(pagecache, block);
memcpy(block->buffer + offset, buff, size);
block->status|= PCBLOCK_READ;
/* Page is correct again if we made a full write in it */
if (size == pagecache->block_size)
block->status&= ~PCBLOCK_ERROR;
}
if (need_page_ready_signal &&
block->wqueue[COND_FOR_REQUESTED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_REQUESTED]);
if (first_REDO_LSN_for_page)
{
/* single write action of the last write action */
DBUG_ASSERT(lock == PAGECACHE_LOCK_WRITE_UNLOCK ||
lock == PAGECACHE_LOCK_LEFT_UNLOCKED);
DBUG_ASSERT(pin == PAGECACHE_UNPIN ||
pin == PAGECACHE_PIN_LEFT_UNPINNED);
pagecache_set_block_rec_lsn(block, first_REDO_LSN_for_page);
}
if (need_lock_change)
{
/*
We don't set rec_lsn of the block; this is ok as for the
Maria-block-record's pages, we always keep pages pinned here.
*/
if (make_lock_and_pin(pagecache, block,
write_lock_change_table[lock].unlock_lock,
write_pin_change_table[pin].unlock_pin, FALSE))
DBUG_ASSERT(0);
}
/* Unregister the request */
DBUG_ASSERT(block->hash_link->requests > 0);
block->hash_link->requests--;
/* See NOTE for pagecache_unlock about registering requests. */
if (pin == PAGECACHE_PIN_LEFT_UNPINNED || pin == PAGECACHE_UNPIN)
{
unreg_request(pagecache, block, 1);
DBUG_ASSERT(page_link == &fake_link);
}
else
*page_link= block;
if (block->status & PCBLOCK_ERROR)
{
error= 1;
my_debug_put_break_here();
}
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
goto end;
}
no_key_cache:
/*
We can't by pass the normal page cache operations because need
whole page for calling callbacks & so on.
This branch should not be used for now (but it is fixed as it
should be just to avoid confusing)
*/
DBUG_ASSERT(0);
/* Key cache is not used */
if (write_mode == PAGECACHE_WRITE_DELAY)
{
/* We can't use mutex here as the key cache may not be initialized */
pagecache->global_cache_w_requests++;
pagecache->global_cache_write++;
if (offset != 0 || size != pagecache->block_size)
{
uchar *page_buffer= (uchar *) alloca(pagecache->block_size);
PAGECACHE_IO_HOOK_ARGS args;
args.page= page_buffer;
args.pageno= pageno;
args.data= file->callback_data;
pagecache->global_cache_read++;
error= (*file->pre_read_hook)(&args);
if (!error)
{
error= pagecache_fread(pagecache, file,
page_buffer,
pageno,
pagecache->readwrite_flags) != 0;
}
if ((*file->post_read_hook)(error, &args))
{
DBUG_PRINT("error", ("read callback problem"));
error= 1;
goto end;
}
memcpy((char *)page_buffer + offset, buff, size);
buff= page_buffer;
}
if (pagecache_fwrite(pagecache, file, buff, pageno, type,
pagecache->readwrite_flags))
error= 1;
}
end:
#if !defined(DBUG_OFF) && defined(EXTRA_DEBUG)
DBUG_EXECUTE("exec",
test_key_cache(pagecache, "end of key_cache_write", 1););
#endif
if (block)
PCBLOCK_INFO(block);
else
DBUG_PRINT("info", ("No block"));
DBUG_RETURN(error);
}
/*
Free block: remove reference to it from hash table,
remove it from the chain file of dirty/clean blocks
and add it to the free list.
*/
static my_bool free_block(PAGECACHE *pagecache, PAGECACHE_BLOCK_LINK *block,
my_bool abort_if_pinned)
{
uint status= block->status;
KEYCACHE_THREAD_TRACE("free block");
KEYCACHE_DBUG_PRINT("free_block",
("block: %u hash_link %p",
PCBLOCK_NUMBER(pagecache, block),
block->hash_link));
mysql_mutex_assert_owner(&pagecache->cache_lock);
if (block->hash_link)
{
/*
While waiting for readers to finish, new readers might request the
block. But since we set block->status|= PCBLOCK_REASSIGNED, they
will wait on block->wqueue[COND_FOR_SAVED]. They must be signaled
later.
*/
block->status|= PCBLOCK_REASSIGNED;
wait_for_readers(pagecache, block);
if (unlikely(abort_if_pinned) && unlikely(block->pins))
{
/*
Block got pinned while waiting for readers.
This can only happens when called from flush_pagecache_blocks_int()
when flushing blocks as part of prepare for maria_close() or from
flush_cached_blocks()
*/
block->status&= ~PCBLOCK_REASSIGNED;
unreg_request(pagecache, block, 0);
/* All pending requests for this page must be resubmitted. */
if (block->wqueue[COND_FOR_SAVED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_SAVED]);
return 1;
}
unlink_hash(pagecache, block->hash_link);
}
unlink_changed(block);
DBUG_ASSERT(block->wlocks == 0);
DBUG_ASSERT(block->rlocks == 0);
DBUG_ASSERT(block->rlocks_queue == 0);
DBUG_ASSERT(block->pins == 0);
DBUG_ASSERT((block->status & ~(PCBLOCK_ERROR | PCBLOCK_READ | PCBLOCK_IN_FLUSH | PCBLOCK_CHANGED | PCBLOCK_REASSIGNED | PCBLOCK_DEL_WRITE)) == 0);
DBUG_ASSERT(block->requests >= 1);
DBUG_ASSERT(block->next_used == NULL);
block->status= 0;
#ifndef DBUG_OFF
block->type= PAGECACHE_EMPTY_PAGE;
#endif
block->rec_lsn= LSN_MAX;
block->hash_link= NULL;
if (block->temperature == PCBLOCK_WARM)
pagecache->warm_blocks--;
block->temperature= PCBLOCK_COLD;
KEYCACHE_THREAD_TRACE("free block");
KEYCACHE_DBUG_PRINT("free_block",
("block is freed"));
unreg_request(pagecache, block, 0);
/*
Block->requests is != 0 if unreg_requests()/link_block() gave the block
to a waiting thread
*/
if (!block->requests)
{
DBUG_ASSERT(block->next_used != 0);
/* Remove the free block from the LRU ring. */
unlink_block(pagecache, block);
/* Insert the free block in the free list. */
block->next_used= pagecache->free_block_list;
pagecache->free_block_list= block;
/* Keep track of the number of currently unused blocks. */
pagecache->blocks_unused++;
}
else
{
/* keep flag set by link_block() */
block->status= status & PCBLOCK_REASSIGNED;
}
/* All pending requests for this page must be resubmitted. */
if (block->wqueue[COND_FOR_SAVED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_SAVED]);
return 0;
}
static int cmp_sec_link(PAGECACHE_BLOCK_LINK **a, PAGECACHE_BLOCK_LINK **b)
{
return (((*a)->hash_link->pageno < (*b)->hash_link->pageno) ? -1 :
((*a)->hash_link->pageno > (*b)->hash_link->pageno) ? 1 : 0);
}
/**
@brief Flush a portion of changed blocks to disk, free used blocks
if requested
@param pagecache This page cache reference.
@param file File which should be flushed
@param cache Beginning of array of the block.
@param end Reference to the block after last in the array.
@param flush_type Type of the flush.
@param first_errno Where to store first errno of the flush.
@return Operation status
@retval PCFLUSH_OK OK
@retval PCFLUSH_ERROR There was errors during the flush process.
@retval PCFLUSH_PINNED Pinned blocks was met and skipped.
@retval PCFLUSH_PINNED_AND_ERROR PCFLUSH_ERROR and PCFLUSH_PINNED.
*/
static int flush_cached_blocks(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
PAGECACHE_BLOCK_LINK **cache,
PAGECACHE_BLOCK_LINK **end,
enum flush_type type,
int *first_errno)
{
int rc= PCFLUSH_OK;
my_bool error;
uint count= (uint) (end-cache);
DBUG_ENTER("flush_cached_blocks");
*first_errno= 0;
/* Don't lock the cache during the flush */
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
/*
As all blocks referred in 'cache' are marked by PCBLOCK_IN_FLUSH
we are guaranteed that no thread will change them
*/
qsort((uchar*) cache, count, sizeof(*cache), (qsort_cmp) cmp_sec_link);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
for (; cache != end; cache++)
{
PAGECACHE_BLOCK_LINK *block= *cache;
/*
In the case of non_transactional tables we want to flush also
block pinned with reads. This is becasue we may have other
threads reading the block during flush, as non transactional
tables can have many readers while the one writer is doing the
flush.
We don't want to do flush pinned blocks during checkpoint.
We detect the checkpoint case by checking if type is LAZY.
*/
if ((type == FLUSH_KEEP_LAZY && block->pins) || block->wlocks)
{
KEYCACHE_DBUG_PRINT("flush_cached_blocks",
("block: %u (%p) pinned",
PCBLOCK_NUMBER(pagecache, block), block));
DBUG_PRINT("info", ("block: %u (%p) pinned",
PCBLOCK_NUMBER(pagecache, block), block));
PCBLOCK_INFO(block);
/* undo the mark put by flush_pagecache_blocks_int(): */
block->status&= ~PCBLOCK_IN_FLUSH;
rc|= PCFLUSH_PINNED;
DBUG_PRINT("warning", ("Page pinned"));
unreg_request(pagecache, block, 1);
if (!*first_errno)
*first_errno= HA_ERR_INTERNAL_ERROR;
continue;
}
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_READ, PAGECACHE_PIN, FALSE))
DBUG_ASSERT(0);
KEYCACHE_PRINT("flush_cached_blocks",
("block: %u (%p) to be flushed",
PCBLOCK_NUMBER(pagecache, block), block));
DBUG_PRINT("info", ("block: %u (%p) to be flushed",
PCBLOCK_NUMBER(pagecache, block), block));
PCBLOCK_INFO(block);
/**
@todo IO If page is contiguous with next page to flush, group flushes
in one single my_pwrite().
*/
/**
It is important to use block->hash_link->file below and not 'file', as
the first one is right and the second may have different out-of-date
content (see StaleFilePointersInFlush in ma_checkpoint.c).
@todo change argument of functions to be File.
*/
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
error= pagecache_fwrite(pagecache, &block->hash_link->file,
block->buffer,
block->hash_link->pageno,
block->type,
pagecache->readwrite_flags);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
if (make_lock_and_pin(pagecache, block,
PAGECACHE_LOCK_READ_UNLOCK,
PAGECACHE_UNPIN, FALSE))
DBUG_ASSERT(0);
pagecache->global_cache_write++;
if (error)
{
block->status|= PCBLOCK_ERROR;
block->error= (int16) my_errno;
my_debug_put_break_here();
if (!*first_errno)
*first_errno= my_errno ? my_errno : -1;
rc|= PCFLUSH_ERROR;
}
/*
Let to proceed for possible waiting requests to write to the block page.
It might happen only during an operation to resize the key cache.
*/
if (block->wqueue[COND_FOR_SAVED].last_thread)
wqueue_release_queue(&block->wqueue[COND_FOR_SAVED]);
/* type will never be FLUSH_IGNORE_CHANGED here */
if (! (type == FLUSH_KEEP || type == FLUSH_KEEP_LAZY ||
type == FLUSH_FORCE_WRITE))
{
if (!free_block(pagecache, block, 1))
{
pagecache->blocks_changed--;
pagecache->global_blocks_changed--;
}
else
{
block->status&= ~PCBLOCK_IN_FLUSH;
link_to_file_list(pagecache, block, file, 1);
}
}
else
{
block->status&= ~PCBLOCK_IN_FLUSH;
link_to_file_list(pagecache, block, file, 1);
unreg_request(pagecache, block, 1);
}
}
DBUG_RETURN(rc);
}
/**
@brief flush all blocks for a file to disk but don't do any mutex locks
@param pagecache pointer to a pagecache data structure
@param file handler for the file to flush to
@param flush_type type of the flush
@param filter optional function which tells what blocks to flush;
can be non-NULL only if FLUSH_KEEP, FLUSH_KEEP_LAZY
or FLUSH_FORCE_WRITE.
@param filter_arg an argument to pass to 'filter'. Information about
the block will be passed too.
@note
Flushes all blocks having the same OS file descriptor as 'file->file', so
can flush blocks having '*block->hash_link->file' != '*file'.
@note
This function doesn't do any mutex locks because it needs to be called
both from flush_pagecache_blocks and flush_all_key_blocks (the later one
does the mutex lock in the resize_pagecache() function).
@note
This function can cause problems if two threads call it
concurrently on the same file (look for "PageCacheFlushConcurrencyBugs"
in ma_checkpoint.c); to avoid them, it has internal logic to serialize in
this situation.
@return Operation status
@retval PCFLUSH_OK OK
@retval PCFLUSH_ERROR There was errors during the flush process.
@retval PCFLUSH_PINNED Pinned blocks was met and skipped.
@retval PCFLUSH_PINNED_AND_ERROR PCFLUSH_ERROR and PCFLUSH_PINNED.
*/
static int flush_pagecache_blocks_int(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
enum flush_type type,
PAGECACHE_FLUSH_FILTER filter,
void *filter_arg)
{
PAGECACHE_BLOCK_LINK *cache_buff[FLUSH_CACHE],**cache;
int last_errno= 0;
int rc= PCFLUSH_OK;
DBUG_ENTER("flush_pagecache_blocks_int");
DBUG_PRINT("enter",
("fd: %d blocks_used: %zu blocks_changed: %zu type: %d",
file->file, pagecache->blocks_used, pagecache->blocks_changed,
type));
#if !defined(DBUG_OFF) && defined(EXTRA_DEBUG)
DBUG_EXECUTE("check_pagecache",
test_key_cache(pagecache,
"start of flush_pagecache_blocks", 0););
#endif
cache= cache_buff;
if (pagecache->disk_blocks > 0 &&
(!my_disable_flush_pagecache_blocks ||
(type != FLUSH_KEEP && type != FLUSH_KEEP_LAZY)))
{
/*
Key cache exists. If my_disable_flush_pagecache_blocks is true it
disables the operation but only FLUSH_KEEP[_LAZY]: other flushes still
need to be allowed: FLUSH_RELEASE has to free blocks, and
FLUSH_FORCE_WRITE is to overrule my_disable_flush_pagecache_blocks.
*/
int error= 0;
uint count= 0;
PAGECACHE_BLOCK_LINK **pos, **end;
PAGECACHE_BLOCK_LINK *first_in_switch= NULL;
PAGECACHE_BLOCK_LINK *block, *next;
#if defined(PAGECACHE_DEBUG)
uint cnt= 0;
#endif
struct st_file_in_flush us_flusher, *other_flusher;
us_flusher.file= file->file;
us_flusher.flush_queue.last_thread= NULL;
us_flusher.first_in_switch= FALSE;
while ((other_flusher= (struct st_file_in_flush *)
my_hash_search(&pagecache->files_in_flush, (uchar *)&file->file,
sizeof(file->file))))
{
/*
File is in flush already: wait, unless FLUSH_KEEP_LAZY. "Flusher"
means "who can mark PCBLOCK_IN_FLUSH", i.e. caller of
flush_pagecache_blocks_int().
*/
struct st_my_thread_var *thread;
if (type == FLUSH_KEEP_LAZY)
{
DBUG_PRINT("info",("FLUSH_KEEP_LAZY skips"));
DBUG_RETURN(0);
}
thread= my_thread_var;
wqueue_add_to_queue(&other_flusher->flush_queue, thread);
do
{
DBUG_PRINT("wait",
("(1) suspend thread %s %ld",
thread->name, (ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while (thread->next);
}
/* we are the only flusher of this file now */
while (my_hash_insert(&pagecache->files_in_flush, (uchar *)&us_flusher))
{
/*
Out of memory, wait for flushers to empty the hash and retry; should
rarely happen. Other threads are flushing the file; when done, they
are going to remove themselves from the hash, and thus memory will
appear again. However, this memory may be stolen by yet another thread
(for a purpose unrelated to page cache), before we retry
my_hash_insert(). So the loop may run for long. Only if the thread was
killed do we abort the loop, returning 1 (error) which can cause the
table to be marked as corrupted (cf maria_chk_size(), maria_close())
and thus require a table check.
*/
DBUG_ASSERT(0);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
if (my_thread_var->abort)
DBUG_RETURN(1); /* End if aborted by user */
sleep(10);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
}
if (type != FLUSH_IGNORE_CHANGED)
{
/*
Count how many key blocks we have to cache to be able
to flush all dirty pages with minimum seek moves.
*/
for (block= pagecache->changed_blocks[FILE_HASH(*file, pagecache)] ;
block;
block= block->next_changed)
{
if (block->hash_link->file.file == file->file)
{
count++;
KEYCACHE_DBUG_ASSERT(count<= pagecache->blocks_used);
}
}
count++; /* Allocate one extra for easy end-of-buffer test */
/* Allocate a new buffer only if its bigger than the one we have */
if (count > FLUSH_CACHE &&
!(cache=
(PAGECACHE_BLOCK_LINK**)
my_malloc(sizeof(PAGECACHE_BLOCK_LINK*)*count, MYF(0))))
{
cache= cache_buff;
count= FLUSH_CACHE;
}
}
/* Retrieve the blocks and write them to a buffer to be flushed */
restart:
end= (pos= cache)+count;
for (block= pagecache->changed_blocks[FILE_HASH(*file, pagecache)] ;
block;
block= next)
{
#if defined(PAGECACHE_DEBUG)
cnt++;
KEYCACHE_DBUG_ASSERT(cnt <= pagecache->blocks_used);
#endif
next= block->next_changed;
if (block->hash_link->file.file != file->file)
continue;
if (filter != NULL)
{
int filter_res= (*filter)(block->type, block->hash_link->pageno,
block->rec_lsn, filter_arg);
DBUG_PRINT("info",("filter returned %d", filter_res));
if (filter_res == FLUSH_FILTER_SKIP_TRY_NEXT)
continue;
if (filter_res == FLUSH_FILTER_SKIP_ALL)
break;
DBUG_ASSERT(filter_res == FLUSH_FILTER_OK);
}
{
DBUG_ASSERT(!(block->status & PCBLOCK_IN_FLUSH));
/*
We care only for the blocks for which flushing was not
initiated by other threads as a result of page swapping
*/
if (! (block->status & PCBLOCK_IN_SWITCH))
{
/*
Mark the block with BLOCK_IN_FLUSH in order not to let
other threads to use it for new pages and interfere with
our sequence of flushing dirty file pages
*/
block->status|= PCBLOCK_IN_FLUSH;
reg_requests(pagecache, block, 1);
if (type != FLUSH_IGNORE_CHANGED)
{
*pos++= block;
/* It's not a temporary file */
if (pos == end)
{
/*
This happens only if there is not enough
memory for the big block
*/
if ((rc|= flush_cached_blocks(pagecache, file, cache,
end, type, &error)) &
(PCFLUSH_ERROR | PCFLUSH_PINNED))
last_errno=error;
DBUG_PRINT("info", ("restarting..."));
/*
Restart the scan as some other thread might have changed
the changed blocks chain: the blocks that were in switch
state before the flush started have to be excluded
*/
goto restart;
}
}
else
{
/* It's a temporary file */
pagecache->blocks_changed--;
pagecache->global_blocks_changed--;
free_block(pagecache, block, 0);
}
}
else if (type != FLUSH_KEEP_LAZY)
{
/*
Link the block into a list of blocks 'in switch', and then we will
wait for this list to be empty, which means they have been flushed
*/
unlink_changed(block);
link_changed(block, &first_in_switch);
us_flusher.first_in_switch= TRUE;
}
}
}
if (pos != cache)
{
if ((rc|= flush_cached_blocks(pagecache, file, cache, pos, type,
&error)) &
(PCFLUSH_ERROR | PCFLUSH_PINNED))
last_errno= error;
}
/* Wait until list of blocks in switch is empty */
while (first_in_switch)
{
#if defined(PAGECACHE_DEBUG)
cnt= 0;
#endif
block= first_in_switch;
{
struct st_my_thread_var *thread= my_thread_var;
wqueue_add_to_queue(&block->wqueue[COND_FOR_SAVED], thread);
do
{
DBUG_PRINT("wait",
("(2) suspend thread %s %ld",
thread->name, (ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while (thread->next);
}
#if defined(PAGECACHE_DEBUG)
cnt++;
KEYCACHE_DBUG_ASSERT(cnt <= pagecache->blocks_used);
#endif
}
us_flusher.first_in_switch= FALSE;
/* The following happens very seldom */
if (! (type == FLUSH_KEEP || type == FLUSH_KEEP_LAZY ||
type == FLUSH_FORCE_WRITE))
{
/*
this code would free all blocks while filter maybe handled only a
few, that is not possible.
*/
DBUG_ASSERT(filter == NULL);
#if defined(PAGECACHE_DEBUG)
cnt=0;
#endif
for (block= pagecache->file_blocks[FILE_HASH(*file, pagecache)] ;
block;
block= next)
{
#if defined(PAGECACHE_DEBUG)
cnt++;
KEYCACHE_DBUG_ASSERT(cnt <= pagecache->blocks_used);
#endif
next= block->next_changed;
if (block->hash_link->file.file == file->file &&
!block->pins &&
(! (block->status & PCBLOCK_CHANGED)
|| type == FLUSH_IGNORE_CHANGED))
{
reg_requests(pagecache, block, 1);
free_block(pagecache, block, 1);
}
}
}
/* wake up others waiting to flush this file */
my_hash_delete(&pagecache->files_in_flush, (uchar *)&us_flusher);
if (us_flusher.flush_queue.last_thread)
wqueue_release_queue(&us_flusher.flush_queue);
}
#ifndef DBUG_OFF
DBUG_EXECUTE("check_pagecache",
test_key_cache(pagecache, "end of flush_pagecache_blocks", 0););
#endif
if (cache != cache_buff)
my_free(cache);
if (rc != 0)
{
if (last_errno)
my_errno= last_errno; /* Return first error */
DBUG_PRINT("error", ("Got error: %d", my_errno));
}
DBUG_RETURN(rc);
}
/**
@brief flush all blocks for a file to disk
@param pagecache pointer to a pagecache data structure
@param file handler for the file to flush to
@param flush_type type of the flush
@param filter optional function which tells what blocks to flush;
can be non-NULL only if FLUSH_KEEP, FLUSH_KEEP_LAZY
or FLUSH_FORCE_WRITE.
@param filter_arg an argument to pass to 'filter'. Information about
the block will be passed too.
@return Operation status
@retval PCFLUSH_OK OK
@retval PCFLUSH_ERROR There was errors during the flush process.
@retval PCFLUSH_PINNED Pinned blocks was met and skipped.
@retval PCFLUSH_PINNED_AND_ERROR PCFLUSH_ERROR and PCFLUSH_PINNED.
*/
int flush_pagecache_blocks_with_filter(PAGECACHE *pagecache,
PAGECACHE_FILE *file,
enum flush_type type,
PAGECACHE_FLUSH_FILTER filter,
void *filter_arg)
{
int res;
DBUG_ENTER("flush_pagecache_blocks_with_filter");
DBUG_PRINT("enter", ("pagecache: %p", pagecache));
if (pagecache->disk_blocks <= 0)
DBUG_RETURN(0);
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
inc_counter_for_resize_op(pagecache);
res= flush_pagecache_blocks_int(pagecache, file, type, filter, filter_arg);
dec_counter_for_resize_op(pagecache);
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
DBUG_RETURN(res);
}
/*
Reset the counters of a key cache.
SYNOPSIS
reset_pagecache_counters()
name the name of a key cache
pagecache pointer to the pagecache to be reset
DESCRIPTION
This procedure is used to reset the counters of all currently used key
caches, both the default one and the named ones.
RETURN
0 on success (always because it can't fail)
*/
int reset_pagecache_counters(const char *name __attribute__((unused)),
PAGECACHE *pagecache)
{
DBUG_ENTER("reset_pagecache_counters");
if (!pagecache->inited)
{
DBUG_PRINT("info", ("Key cache %s not initialized.", name));
DBUG_RETURN(0);
}
DBUG_PRINT("info", ("Resetting counters for key cache %s.", name));
pagecache->global_blocks_changed= 0; /* Key_blocks_not_flushed */
pagecache->global_cache_r_requests= 0; /* Key_read_requests */
pagecache->global_cache_read= 0; /* Key_reads */
pagecache->global_cache_w_requests= 0; /* Key_write_requests */
pagecache->global_cache_write= 0; /* Key_writes */
DBUG_RETURN(0);
}
/**
@brief Allocates a buffer and stores in it some info about all dirty pages
Does the allocation because the caller cannot know the size itself.
Memory freeing is to be done by the caller (if the "str" member of the
LEX_STRING is not NULL).
Ignores all pages of another type than PAGECACHE_LSN_PAGE, because they
are not interesting for a checkpoint record.
The caller has the intention of doing checkpoints.
@param pagecache pointer to the page cache
@param[out] str pointer to where the allocated buffer, and
its size, will be put
@param[out] min_rec_lsn pointer to where the minimum rec_lsn of all
relevant dirty pages will be put
@return Operation status
@retval 0 OK
@retval 1 Error
*/
my_bool pagecache_collect_changed_blocks_with_lsn(PAGECACHE *pagecache,
LEX_STRING *str,
LSN *min_rec_lsn)
{
my_bool error= 0;
size_t stored_list_size= 0;
uint file_hash;
char *ptr;
LSN minimum_rec_lsn= LSN_MAX;
DBUG_ENTER("pagecache_collect_changed_blocks_with_LSN");
DBUG_ASSERT(NULL == str->str);
/*
We lock the entire cache but will be quick, just reading/writing a few MBs
of memory at most.
*/
pagecache_pthread_mutex_lock(&pagecache->cache_lock);
for (;;)
{
struct st_file_in_flush *other_flusher;
for (file_hash= 0;
(other_flusher= (struct st_file_in_flush *)
my_hash_element(&pagecache->files_in_flush, file_hash)) != NULL &&
!other_flusher->first_in_switch;
file_hash++)
{}
if (other_flusher == NULL)
break;
/*
other_flusher.first_in_switch is true: some thread is flushing a file
and has removed dirty blocks from changed_blocks[] while they were still
dirty (they were being evicted (=>flushed) by yet another thread, which
may not have flushed the block yet so it may still be dirty).
If Checkpoint proceeds now, it will not see the page. If there is a
crash right after writing the checkpoint record, before the page is
flushed, at recovery the page will be wrongly ignored because it won't
be in the dirty pages list in the checkpoint record. So wait.
*/
{
struct st_my_thread_var *thread= my_thread_var;
wqueue_add_to_queue(&other_flusher->flush_queue, thread);
do
{
DBUG_PRINT("wait",
("suspend thread %s %ld", thread->name,
(ulong) thread->id));
pagecache_pthread_cond_wait(&thread->suspend,
&pagecache->cache_lock);
}
while (thread->next);
}
}
/* Count how many dirty pages are interesting */
for (file_hash= 0; file_hash < pagecache->changed_blocks_hash_size; file_hash++)
{
PAGECACHE_BLOCK_LINK *block;
for (block= pagecache->changed_blocks[file_hash] ;
block;
block= block->next_changed)
{
/*
Q: is there something subtle with block->hash_link: can it be NULL?
does it have to be == hash_link->block... ?
*/
DBUG_ASSERT(block->hash_link != NULL);
DBUG_ASSERT(block->status & PCBLOCK_CHANGED);
/*
Note that we don't store bitmap pages, or pages from non-transactional
(like temporary) tables. Don't checkpoint during Recovery which uses
PAGECACHE_PLAIN_PAGE.
*/
if (block->type != PAGECACHE_LSN_PAGE)
continue; /* no need to store it */
stored_list_size++;
}
}
compile_time_assert(sizeof(pagecache->blocks) <= 8);
str->length= 8 + /* number of dirty pages */
(2 + /* table id */
1 + /* data or index file */
5 + /* pageno */
LSN_STORE_SIZE /* rec_lsn */
) * stored_list_size;
if (NULL == (str->str= my_malloc(str->length, MYF(MY_WME))))
goto err;
ptr= str->str;
int8store(ptr, (ulonglong)stored_list_size);
ptr+= 8;
DBUG_PRINT("info", ("found %zu dirty pages", stored_list_size));
if (stored_list_size == 0)
goto end;
for (file_hash= 0; file_hash < pagecache->changed_blocks_hash_size; file_hash++)
{
PAGECACHE_BLOCK_LINK *block;
for (block= pagecache->changed_blocks[file_hash] ;
block;
block= block->next_changed)
{
uint16 table_id;
MARIA_SHARE *share;
if (block->type != PAGECACHE_LSN_PAGE)
continue; /* no need to store it in the checkpoint record */
share= (MARIA_SHARE *)(block->hash_link->file.callback_data);
table_id= share->id;
int2store(ptr, table_id);
ptr+= 2;
ptr[0]= (share->kfile.file == block->hash_link->file.file);
ptr++;
DBUG_ASSERT(block->hash_link->pageno < ((1ULL) << 40));
page_store(ptr, block->hash_link->pageno);
ptr+= PAGE_STORE_SIZE;
lsn_store(ptr, block->rec_lsn);
ptr+= LSN_STORE_SIZE;
if (block->rec_lsn != LSN_MAX)
{
DBUG_ASSERT(LSN_VALID(block->rec_lsn));
if (cmp_translog_addr(block->rec_lsn, minimum_rec_lsn) < 0)
minimum_rec_lsn= block->rec_lsn;
} /* otherwise, some trn->rec_lsn should hold the correct info */
}
}
end:
pagecache_pthread_mutex_unlock(&pagecache->cache_lock);
*min_rec_lsn= minimum_rec_lsn;
DBUG_RETURN(error);
err:
error= 1;
goto end;
}
#ifndef DBUG_OFF
/**
Verifies that a file has no dirty pages.
*/
void pagecache_file_no_dirty_page(PAGECACHE *pagecache, PAGECACHE_FILE *file)
{
File fd= file->file;
PAGECACHE_BLOCK_LINK *block;
for (block= pagecache->changed_blocks[FILE_HASH(*file, pagecache)];
block != NULL;
block= block->next_changed)
if (block->hash_link->file.file == fd)
{
DBUG_PRINT("info", ("pagecache_file_not_in error"));
PCBLOCK_INFO(block);
DBUG_ASSERT(0);
}
}
/*
Test if disk-cache is ok
*/
static void test_key_cache(PAGECACHE *pagecache __attribute__((unused)),
const char *where __attribute__((unused)),
my_bool lock __attribute__((unused)))
{
/* TODO */
}
#endif
uchar *pagecache_block_link_to_buffer(PAGECACHE_BLOCK_LINK *block)
{
return block->buffer;
}
#if defined(PAGECACHE_TIMEOUT)
#define KEYCACHE_DUMP_FILE "pagecache_dump.txt"
#define MAX_QUEUE_LEN 100
static void pagecache_dump(PAGECACHE *pagecache)
{
FILE *pagecache_dump_file=fopen(KEYCACHE_DUMP_FILE, "w");
struct st_my_thread_var *last;
struct st_my_thread_var *thread;
PAGECACHE_BLOCK_LINK *block;
PAGECACHE_HASH_LINK *hash_link;
PAGECACHE_PAGE *page;
uint i;
fprintf(pagecache_dump_file, "thread: %s %ld\n", thread->name,
(ulong) thread->id);
i=0;
thread=last=waiting_for_hash_link.last_thread;
fprintf(pagecache_dump_file, "queue of threads waiting for hash link\n");
if (thread)
do
{
thread= thread->next;
page= (PAGECACHE_PAGE *) thread->keycache_link;
fprintf(pagecache_dump_file,
"thread: %s %ld, (file,pageno)=(%u,%lu)\n",
thread->name, (ulong) thread->id,
(uint) page->file.file,(ulong) page->pageno);
if (++i == MAX_QUEUE_LEN)
break;
}
while (thread != last);
i=0;
thread=last=waiting_for_block.last_thread;
fprintf(pagecache_dump_file, "queue of threads waiting for block\n");
if (thread)
do
{
thread=thread->next;
hash_link= (PAGECACHE_HASH_LINK *) thread->keycache_link;
fprintf(pagecache_dump_file,
"thread: %s %u hash_link:%u (file,pageno)=(%u,%lu)\n",
thread->name, (ulong) thread->id,
(uint) PAGECACHE_HASH_LINK_NUMBER(pagecache, hash_link),
(uint) hash_link->file.file,(ulong) hash_link->pageno);
if (++i == MAX_QUEUE_LEN)
break;
}
while (thread != last);
for (i=0 ; i < pagecache->blocks_used ; i++)
{
int j;
block= &pagecache->block_root[i];
hash_link= block->hash_link;
fprintf(pagecache_dump_file,
"block:%u hash_link:%d status:%x #requests=%u waiting_for_readers:%d\n",
i, (int) (hash_link ?
PAGECACHE_HASH_LINK_NUMBER(pagecache, hash_link) :
-1),
block->status, block->requests, block->condvar ? 1 : 0);
for (j=0 ; j < COND_SIZE; j++)
{
PAGECACHE_WQUEUE *wqueue=&block->wqueue[j];
thread= last= wqueue->last_thread;
fprintf(pagecache_dump_file, "queue #%d\n", j);
if (thread)
{
do
{
thread=thread->next;
fprintf(pagecache_dump_file,
"thread: %s %ld\n", thread->name, (ulong) thread->id);
if (++i == MAX_QUEUE_LEN)
break;
}
while (thread != last);
}
}
}
fprintf(pagecache_dump_file, "LRU chain:");
block= pagecache= used_last;
if (block)
{
do
{
block= block->next_used;
fprintf(pagecache_dump_file,
"block:%u, ", PCBLOCK_NUMBER(pagecache, block));
}
while (block != pagecache->used_last);
}
fprintf(pagecache_dump_file, "\n");
fclose(pagecache_dump_file);
}
#endif /* defined(PAGECACHE_TIMEOUT) */
#if defined(PAGECACHE_TIMEOUT) && !defined(__WIN__)
static int pagecache_pthread_cond_wait(mysql_cond_t *cond,
mysql_mutex_t *mutex)
{
int rc;
struct timeval now; /* time when we started waiting */
struct timespec timeout; /* timeout value for the wait function */
struct timezone tz;
#if defined(PAGECACHE_DEBUG)
int cnt=0;
#endif
/* Get current time */
gettimeofday(&now, &tz);
/* Prepare timeout value */
timeout.tv_sec= now.tv_sec + PAGECACHE_TIMEOUT;
/*
timeval uses microseconds.
timespec uses nanoseconds.
1 nanosecond = 1000 micro seconds
*/
timeout.tv_nsec= now.tv_usec * 1000;
KEYCACHE_THREAD_TRACE_END("started waiting");
#if defined(PAGECACHE_DEBUG)
cnt++;
if (cnt % 100 == 0)
fprintf(pagecache_debug_log, "waiting...\n");
fflush(pagecache_debug_log);
#endif
rc= mysql_cond_timedwait(cond, mutex, &timeout);
KEYCACHE_THREAD_TRACE_BEGIN("finished waiting");
if (rc == ETIMEDOUT || rc == ETIME)
{
#if defined(PAGECACHE_DEBUG)
fprintf(pagecache_debug_log,"aborted by pagecache timeout\n");
fclose(pagecache_debug_log);
abort();
#endif
pagecache_dump();
}
#if defined(PAGECACHE_DEBUG)
KEYCACHE_DBUG_ASSERT(rc != ETIMEDOUT);
#else
assert(rc != ETIMEDOUT);
#endif
return rc;
}
#else
#if defined(PAGECACHE_DEBUG)
static int pagecache_pthread_cond_wait(mysql_cond_t *cond,
mysql_mutex_t *mutex)
{
int rc;
KEYCACHE_THREAD_TRACE_END("started waiting");
rc= mysql_cond_wait(cond, mutex);
KEYCACHE_THREAD_TRACE_BEGIN("finished waiting");
return rc;
}
#endif
#endif /* defined(PAGECACHE_TIMEOUT) && !defined(__WIN__) */
#if defined(PAGECACHE_DEBUG)
static int ___pagecache_pthread_mutex_lock(mysql_mutex_t *mutex)
{
int rc;
rc= mysql_mutex_lock(mutex);
KEYCACHE_THREAD_TRACE_BEGIN("");
return rc;
}
static void ___pagecache_pthread_mutex_unlock(mysql_mutex_t *mutex)
{
KEYCACHE_THREAD_TRACE_END("");
mysql_mutex_unlock(mutex);
}
static int ___pagecache_pthread_cond_signal(mysql_cond_t *cond)
{
int rc;
KEYCACHE_THREAD_TRACE("signal");
rc= mysql_cond_signal(cond);
return rc;
}
#if defined(PAGECACHE_DEBUG_LOG)
static void pagecache_debug_print(const char * fmt, ...)
{
va_list args;
va_start(args,fmt);
if (pagecache_debug_log)
{
VOID(vfprintf(pagecache_debug_log, fmt, args));
VOID(fputc('\n',pagecache_debug_log));
}
va_end(args);
}
#endif /* defined(PAGECACHE_DEBUG_LOG) */
#if defined(PAGECACHE_DEBUG_LOG)
void pagecache_debug_log_close(void)
{
if (pagecache_debug_log)
fclose(pagecache_debug_log);
}
#endif /* defined(PAGECACHE_DEBUG_LOG) */
#endif /* defined(PAGECACHE_DEBUG) */
/**
@brief null hooks
*/
static my_bool null_pre_hook(PAGECACHE_IO_HOOK_ARGS *args
__attribute__((unused)))
{
return 0;
}
static my_bool null_post_read_hook(int res, PAGECACHE_IO_HOOK_ARGS *args
__attribute__((unused)))
{
return res != 0;
}
static void null_post_write_hook(int res __attribute__((unused)),
PAGECACHE_IO_HOOK_ARGS *args
__attribute__((unused)))
{
return;
}
void
pagecache_file_set_null_hooks(PAGECACHE_FILE *file)
{
file->pre_read_hook= null_pre_hook;
file->post_read_hook= null_post_read_hook;
file->pre_write_hook= null_pre_hook;
file->post_write_hook= null_post_write_hook;
file->flush_log_callback= null_pre_hook;
file->callback_data= NULL;
}
|