1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637 6638 6639 6640 6641 6642 6643 6644 6645 6646 6647 6648 6649 6650 6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733 6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762 6763 6764 6765 6766 6767 6768 6769 6770 6771 6772 6773 6774 6775 6776 6777 6778 6779 6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791 6792 6793 6794 6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948 6949 6950 6951 6952 6953 6954 6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970 6971 6972 6973 6974 6975 6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994 6995 6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173 7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212 7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223 7224 7225 7226 7227 7228 7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277 7278 7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356 7357 7358 7359 7360 7361 7362 7363 7364 7365 7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376 7377 7378 7379 7380 7381 7382 7383 7384 7385 7386 7387 7388 7389 7390 7391 7392 7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403 7404 7405 7406 7407 7408 7409 7410 7411 7412 7413 7414 7415 7416 7417 7418 7419 7420 7421 7422 7423 7424 7425 7426 7427 7428 7429 7430 7431 7432 7433 7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457 7458 7459 7460 7461 7462 7463 7464 7465 7466 7467 7468 7469 7470 7471 7472 7473 7474 7475 7476 7477 7478 7479 7480 7481 7482
|
#ifndef HANDLER_INCLUDED
#define HANDLER_INCLUDED
/*
Copyright (c) 2000, 2025, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License, version 2.0,
as published by the Free Software Foundation.
This program is designed to work with certain software (including
but not limited to OpenSSL) that is licensed under separate terms,
as designated in a particular file or component or in included license
documentation. The authors of MySQL hereby grant you an additional
permission to link the program and your derivative works with the
separately licensed software that they have either included with
the program or referenced in the documentation.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License, version 2.0, for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
/* Definitions for parameters to do with handler-routines */
#include <fcntl.h>
#include <float.h>
#include <string.h>
#include <sys/types.h>
#include <time.h>
#include <algorithm>
#include <bitset>
#include <functional>
#include <map>
#include <optional>
#include <random> // std::mt19937
#include <set>
#include <string>
#include <string_view>
#include <mysql/components/services/page_track_service.h>
#include "ft_global.h" // ft_hints
#include "lex_string.h"
#include "m_ctype.h"
#include "map_helpers.h"
#include "my_alloc.h"
#include "my_base.h"
#include "my_bitmap.h"
#include "my_checksum.h" // ha_checksum
#include "my_compiler.h"
#include "my_dbug.h"
#include "my_double2ulonglong.h"
#include "my_inttypes.h"
#include "my_io.h"
#include "my_sys.h"
#include "my_table_map.h"
#include "my_thread_local.h" // my_errno
#include "mysql/components/services/bits/psi_table_bits.h"
#include "sql/dd/object_id.h" // dd::Object_id
#include "sql/dd/string_type.h"
#include "sql/dd/types/object_table.h" // dd::Object_table
#include "sql/discrete_interval.h" // Discrete_interval
#include "sql/key.h"
#include "sql/sql_const.h" // SHOW_COMP_OPTION
#include "sql/sql_list.h" // SQL_I_List
#include "sql/sql_plugin_ref.h" // plugin_ref
#include "thr_lock.h" // thr_lock_type
#include "typelib.h"
class Alter_info;
class Create_field;
class Field;
class Item;
class JOIN;
class Json_dom;
class Partition_handler;
class Plugin_table;
class Plugin_tablespace;
class Record_buffer;
class SE_cost_constants; // see opt_costconstants.h
class String;
class THD;
class handler;
class partition_info;
struct System_status_var;
namespace dd {
class Properties;
} // namespace dd
struct AccessPath;
struct JoinHypergraph;
struct KEY_CACHE;
struct LEX;
struct MY_BITMAP;
struct SAVEPOINT;
struct TABLE;
class Table_ref;
struct TABLE_SHARE;
struct Tablespace_options;
struct handlerton;
typedef struct xid_t XID;
typedef struct st_xarecover_txn XA_recover_txn;
struct MDL_key;
namespace dd {
enum class enum_column_types;
class Table;
class Tablespace;
} // namespace dd
constexpr const ha_rows EXTRA_RECORDS{10};
/** Id for identifying Table SDIs */
constexpr const uint32 SDI_TYPE_TABLE = 1;
/** Id for identifying Tablespace SDIs */
constexpr const uint32 SDI_TYPE_TABLESPACE = 2;
/** Key to identify a dictionary object */
struct sdi_key_t {
/** Type of Object, For ex: column, index, etc */
uint32 type;
/** Object id which should be unique in tablespsace */
uint64 id;
};
using sdi_container = std::vector<sdi_key_t>;
struct sdi_vector_t {
sdi_container m_vec;
};
typedef bool (*qc_engine_callback)(THD *thd, const char *table_key,
uint key_length, ulonglong *engine_data);
typedef bool(stat_print_fn)(THD *thd, const char *type, size_t type_len,
const char *file, size_t file_len,
const char *status, size_t status_len);
class ha_statistics;
class ha_tablespace_statistics;
class Unique_on_insert;
extern ulong savepoint_alloc_size;
/// Maps from slot to plugin. May return NULL if plugin has been unloaded.
st_plugin_int *hton2plugin(uint slot);
/// Returns the size of the array holding pointers to plugins.
size_t num_hton2plugins();
/**
For unit testing.
Insert plugin into arbitrary slot in array.
Remove plugin from arbitrary slot in array.
*/
st_plugin_int *insert_hton2plugin(uint slot, st_plugin_int *plugin);
st_plugin_int *remove_hton2plugin(uint slot);
extern const char *ha_row_type[];
extern const char *tx_isolation_names[];
extern const char *binlog_format_names[];
extern TYPELIB tx_isolation_typelib;
extern ulong total_ha_2pc;
// the following is for checking tables
#define HA_ADMIN_ALREADY_DONE 1
#define HA_ADMIN_OK 0
#define HA_ADMIN_NOT_IMPLEMENTED -1
#define HA_ADMIN_FAILED -2
#define HA_ADMIN_CORRUPT -3
#define HA_ADMIN_INTERNAL_ERROR -4
#define HA_ADMIN_INVALID -5
#define HA_ADMIN_REJECT -6
#define HA_ADMIN_TRY_ALTER -7
#define HA_ADMIN_WRONG_CHECKSUM -8
#define HA_ADMIN_NOT_BASE_TABLE -9
#define HA_ADMIN_NEEDS_UPGRADE -10
#define HA_ADMIN_NEEDS_ALTER -11
#define HA_ADMIN_NEEDS_CHECK -12
#define HA_ADMIN_STATS_UPD_ERR -13
/** User needs to dump and re-create table to fix pre 5.0 decimal types */
#define HA_ADMIN_NEEDS_DUMP_UPGRADE -14
/**
Return values for check_if_supported_inplace_alter().
@see check_if_supported_inplace_alter() for description of
the individual values.
*/
enum enum_alter_inplace_result {
HA_ALTER_ERROR,
HA_ALTER_INPLACE_NOT_SUPPORTED,
HA_ALTER_INPLACE_EXCLUSIVE_LOCK,
HA_ALTER_INPLACE_SHARED_LOCK_AFTER_PREPARE,
HA_ALTER_INPLACE_SHARED_LOCK,
HA_ALTER_INPLACE_NO_LOCK_AFTER_PREPARE,
HA_ALTER_INPLACE_NO_LOCK,
HA_ALTER_INPLACE_INSTANT
};
/* Bits in table_flags() to show what database can do */
#define HA_NO_TRANSACTIONS (1 << 0) /* Doesn't support transactions */
#define HA_PARTIAL_COLUMN_READ (1 << 1) /* read may not return all columns */
/*
Used to avoid scanning full tables on an index. If this flag is set then
the handler always has a primary key (hidden if not defined) and this
index is used for scanning rather than a full table scan in all
situations. No separate data/index file.
*/
#define HA_TABLE_SCAN_ON_INDEX (1 << 2)
/// Not in use.
#define HA_UNUSED3 (1 << 3)
/*
Can the storage engine handle spatial data.
Used to check that no spatial attributes are declared unless
the storage engine is capable of handling it.
*/
#define HA_CAN_GEOMETRY (1 << 4)
/*
Reading keys in random order is as fast as reading keys in sort order
(Used by filesort to decide if we should sort key + data or key +
pointer-to-row.)
*/
#define HA_FAST_KEY_READ (1 << 5)
/*
Set the following flag if we on delete should force all key to be read
and on update read all keys that changes
*/
#define HA_REQUIRES_KEY_COLUMNS_FOR_DELETE (1 << 6)
/*
Is NULL values allowed in indexes.
If this is not allowed then it is not possible to use an index on a
NULLable field.
*/
#define HA_NULL_IN_KEY (1 << 7)
/*
Tells that we can the position for the conflicting duplicate key
record is stored in table->file->dupp_ref. (insert uses rnd_pos() on
this to find the duplicated row)
*/
#define HA_DUPLICATE_POS (1 << 8)
#define HA_NO_BLOBS (1 << 9) /* Doesn't support blobs */
/*
Is the storage engine capable of defining an index of a prefix on
a BLOB attribute.
*/
#define HA_CAN_INDEX_BLOBS (1 << 10)
/*
Auto increment fields can be part of a multi-part key. For second part
auto-increment keys, the auto_incrementing is done in handler.cc
*/
#define HA_AUTO_PART_KEY (1 << 11)
/*
Can't define a table without primary key (and cannot handle a table
with hidden primary key)
*/
#define HA_REQUIRE_PRIMARY_KEY (1 << 12)
/*
Does the counter of records after the info call specify an exact
value or not. If it does this flag is set.
*/
#define HA_STATS_RECORDS_IS_EXACT (1 << 13)
/// Not in use.
#define HA_UNUSED14 (1 << 14)
/*
This parameter is set when the handler will also return the primary key
when doing read-only-key on another index, i.e., if we get the primary
key columns for free when we do an index read (usually, it also implies
that HA_PRIMARY_KEY_REQUIRED_FOR_POSITION flag is set).
*/
#define HA_PRIMARY_KEY_IN_READ_INDEX (1 << 15)
/*
If HA_PRIMARY_KEY_REQUIRED_FOR_POSITION is set, it means that to position()
uses a primary key given by the record argument.
Without primary key, we can't call position().
If not set, the position is returned as the current rows position
regardless of what argument is given.
*/
#define HA_PRIMARY_KEY_REQUIRED_FOR_POSITION (1 << 16)
#define HA_CAN_RTREEKEYS (1 << 17)
/// Not in use.
#define HA_UNUSED18
/*
The following is we need to a primary key to delete (and update) a row.
If there is no primary key, all columns needs to be read on update and delete
*/
#define HA_PRIMARY_KEY_REQUIRED_FOR_DELETE (1 << 19)
/*
Indexes on prefixes of character fields are not allowed.
*/
#define HA_NO_PREFIX_CHAR_KEYS (1 << 20)
/*
Does the storage engine support fulltext indexes.
*/
#define HA_CAN_FULLTEXT (1 << 21)
/*
Can the HANDLER interface in the MySQL API be used towards this
storage engine.
*/
#define HA_CAN_SQL_HANDLER (1 << 22)
/*
Set if the storage engine does not support auto increment fields.
*/
#define HA_NO_AUTO_INCREMENT (1 << 23)
/*
Supports CHECKSUM option in CREATE TABLE (MyISAM feature).
*/
#define HA_HAS_CHECKSUM (1 << 24)
/*
Table data are stored in separate files (for lower_case_table_names).
Should file names always be in lower case (used by engines that map
table names to file names.
*/
#define HA_FILE_BASED (1 << 26)
#define HA_NO_VARCHAR (1 << 27)
/*
Is the storage engine capable of handling bit fields.
*/
#define HA_CAN_BIT_FIELD (1 << 28)
#define HA_ANY_INDEX_MAY_BE_UNIQUE (1 << 30)
#define HA_NO_COPY_ON_ALTER (1LL << 31)
#define HA_COUNT_ROWS_INSTANT (1LL << 32) /* records() gives exact count*/
/* Has it's own method of binlog logging */
#define HA_HAS_OWN_BINLOGGING (1LL << 33)
/*
Engine is capable of row-format and statement-format logging,
respectively
*/
#define HA_BINLOG_ROW_CAPABLE (1LL << 34)
#define HA_BINLOG_STMT_CAPABLE (1LL << 35)
/*
When a multiple key conflict happens in a REPLACE command mysql
expects the conflicts to be reported in the ascending order of
key names.
For e.g.
CREATE TABLE t1 (a INT, UNIQUE (a), b INT NOT NULL, UNIQUE (b), c INT NOT
NULL, INDEX(c));
REPLACE INTO t1 VALUES (1,1,1),(2,2,2),(2,1,3);
MySQL expects the conflict with 'a' to be reported before the conflict with
'b'.
If the underlying storage engine does not report the conflicting keys in
ascending order, it causes unexpected errors when the REPLACE command is
executed.
This flag helps the underlying SE to inform the server that the keys are not
ordered.
*/
#define HA_DUPLICATE_KEY_NOT_IN_ORDER (1LL << 36)
/*
Engine supports REPAIR TABLE. Used by CHECK TABLE FOR UPGRADE if an
incompatible table is detected. If this flag is set, CHECK TABLE FOR UPGRADE
will report ER_TABLE_NEEDS_UPGRADE, otherwise ER_TABLE_NEED_REBUILD.
*/
#define HA_CAN_REPAIR (1LL << 37)
/*
Set of all binlog flags. Currently only contain the capabilities
flags.
*/
#define HA_BINLOG_FLAGS (HA_BINLOG_ROW_CAPABLE | HA_BINLOG_STMT_CAPABLE)
/**
The handler supports read before write removal optimization
Read before write removal may be used for storage engines which support
write without previous read of the row to be updated. Handler returning
this flag must implement start_read_removal() and end_read_removal().
The handler may return "fake" rows constructed from the key of the row
asked for. This is used to optimize UPDATE and DELETE by reducing the
number of round-trips between handler and storage engine.
Example:
UPDATE a=1 WHERE pk IN (@<keys@>)
@verbatim
mysql_update()
{
if (<conditions for starting read removal>)
start_read_removal()
-> handler returns true if read removal supported for this table/query
while(read_record("pk=<key>"))
-> handler returns fake row with column "pk" set to <key>
ha_update_row()
-> handler sends write "a=1" for row with "pk=<key>"
end_read_removal()
-> handler returns the number of rows actually written
}
@endverbatim
@note This optimization in combination with batching may be used to
remove even more round-trips.
*/
#define HA_READ_BEFORE_WRITE_REMOVAL (1LL << 38)
/*
Engine supports extended fulltext API
*/
#define HA_CAN_FULLTEXT_EXT (1LL << 39)
/*
Storage engine doesn't synchronize result set with expected table contents.
Used by replication slave to check if it is possible to retrieve rows from
the table when deciding whether to do a full table scan, index scan or hash
scan while applying a row event.
*/
#define HA_READ_OUT_OF_SYNC (1LL << 40)
/*
Storage engine supports table export using the
FLUSH TABLE <table_list> FOR EXPORT statement.
*/
#define HA_CAN_EXPORT (1LL << 41)
/*
The handler don't want accesses to this table to
be const-table optimized
*/
#define HA_BLOCK_CONST_TABLE (1LL << 42)
/*
Handler supports FULLTEXT hints
*/
#define HA_CAN_FULLTEXT_HINTS (1LL << 43)
/**
Storage engine doesn't support LOCK TABLE ... READ LOCAL locks
but doesn't want to use handler::store_lock() API for upgrading
them to LOCK TABLE ... READ locks, for example, because it doesn't
use THR_LOCK locks at all.
*/
#define HA_NO_READ_LOCAL_LOCK (1LL << 44)
/**
A storage engine is compatible with the attachable transaction requirements
means that
- either SE detects the fact that THD::ha_data was reset and starts a new
attachable transaction, closes attachable transaction on close_connection
and resumes regular (outer) transaction when THD::ha_data is restored;
- or SE completely ignores THD::ha_data and close_connection like MyISAM
does.
*/
#define HA_ATTACHABLE_TRX_COMPATIBLE (1LL << 45)
/**
Handler supports Generated Columns
*/
#define HA_GENERATED_COLUMNS (1LL << 46)
/**
Supports index on virtual generated column
*/
#define HA_CAN_INDEX_VIRTUAL_GENERATED_COLUMN (1LL << 47)
/**
Supports descending indexes
*/
#define HA_DESCENDING_INDEX (1LL << 48)
/**
Supports partial update of BLOB columns.
*/
#define HA_BLOB_PARTIAL_UPDATE (1LL << 49)
/**
If this isn't defined, only columns/indexes with Cartesian coordinate systems
(projected SRS or SRID 0) is supported. Columns/indexes without SRID
restriction is also supported if this isn't defined.
*/
#define HA_SUPPORTS_GEOGRAPHIC_GEOMETRY_COLUMN (1LL << 50)
/**
Handler supports expressions as DEFAULT for a column.
*/
#define HA_SUPPORTS_DEFAULT_EXPRESSION (1LL << 51)
/**
Handlers with this flag set do not support UPDATE operations.
*/
#define HA_UPDATE_NOT_SUPPORTED (1LL << 52)
/**
Handlers with this flag set do not support DELETE operations.
*/
#define HA_DELETE_NOT_SUPPORTED (1LL << 53)
/**
The storage engine does not support using indexes for access. Indexes can only
be used for estimating cost.
*/
#define HA_NO_INDEX_ACCESS (1LL << 54)
/**
Supports multi-valued index
*/
#define HA_MULTI_VALUED_KEY_SUPPORT (1LL << 55)
/*
Bits in index_flags(index_number) for what you can do with index.
If you do not implement indexes, just return zero here.
*/
/*
Does the index support read next, this is assumed in the server
code and never checked so all indexes must support this.
Note that the handler can be used even if it doesn't have any index.
*/
#define HA_READ_NEXT 1 /* TODO really use this flag */
/*
Can the index be used to scan backwards (supports ::index_prev).
*/
#define HA_READ_PREV 2
/*
Can the index deliver its record in index order. Typically true for
all ordered indexes and not true for hash indexes. Used to set keymap
part_of_sortkey.
This keymap is only used to find indexes usable for resolving an ORDER BY
in the query. Thus in most cases index_read will work just fine without
order in result production. When this flag is set it is however safe to
order all output started by index_read since most engines do this. With
read_multi_range calls there is a specific flag setting order or not
order so in those cases ordering of index output can be avoided.
*/
#define HA_READ_ORDER 4
/*
Specify whether index can handle ranges, typically true for all
ordered indexes and not true for hash indexes.
Used by optimiser to check if ranges (as key >= 5) can be optimised
by index.
*/
#define HA_READ_RANGE 8
/*
Can't use part key searches. This is typically true for hash indexes
and typically not true for ordered indexes.
*/
#define HA_ONLY_WHOLE_INDEX 16
/*
Index does not store NULL values, even if the column is nullable.
(KEY::flags may still contain HA_NULL_PART_KEY)
If the key has a NULL-value, the handler need to do a full table scan
instead of using this key. This is typically true for NDB hash indexes.
*/
#define HA_TABLE_SCAN_ON_NULL 32
/*
Does the storage engine support index-only scans on this index.
Enables use of HA_EXTRA_KEYREAD and HA_EXTRA_NO_KEYREAD
Used to set Key_map keys_for_keyread and to check in optimiser for
index-only scans. When doing a read under HA_EXTRA_KEYREAD the handler
only have to fill in the columns the key covers. If
HA_PRIMARY_KEY_IN_READ_INDEX is set then also the PRIMARY KEY columns
must be updated in the row.
*/
#define HA_KEYREAD_ONLY 64
/*
Index scan will not return records in rowid order. Not guaranteed to be
set for unordered (e.g. HASH) indexes.
*/
#define HA_KEY_SCAN_NOT_ROR 128
#define HA_DO_INDEX_COND_PUSHDOWN 256 /* Supports Index Condition Pushdown */
/* operations for disable/enable indexes */
#define HA_KEY_SWITCH_NONUNIQ 0
#define HA_KEY_SWITCH_ALL 1
#define HA_KEY_SWITCH_NONUNIQ_SAVE 2
#define HA_KEY_SWITCH_ALL_SAVE 3
/*
Use this instead of 0 as the initial value for the slot number of
handlerton, so that we can distinguish uninitialized slot number
from slot 0.
*/
#define HA_SLOT_UNDEF ((uint)-1)
/*
Parameters for open() (in register form->filestat)
HA_GET_INFO does an implicit HA_ABORT_IF_LOCKED
*/
#define HA_OPEN_KEYFILE 1
#define HA_OPEN_RNDFILE 2
#define HA_GET_INDEX 4
#define HA_GET_INFO 8 /* do a handler::info() after open */
#define HA_READ_ONLY 16 /* File opened as readonly */
/* Try readonly if can't open with read and write */
#define HA_TRY_READ_ONLY 32
#define HA_WAIT_IF_LOCKED 64 /* Wait if locked on open */
#define HA_ABORT_IF_LOCKED 128 /* skip if locked on open.*/
#define HA_BLOCK_LOCK 256 /* unlock when reading some records */
#define HA_OPEN_TEMPORARY 512
/* Some key definitions */
#define HA_KEY_NULL_LENGTH 1
#define HA_KEY_BLOB_LENGTH 2
#define HA_LEX_CREATE_TMP_TABLE 1
#define HA_LEX_CREATE_IF_NOT_EXISTS 2
#define HA_LEX_CREATE_TABLE_LIKE 4
#define HA_LEX_CREATE_INTERNAL_TMP_TABLE 8
#define HA_MAX_REC_LENGTH 65535U
/**
Options for the START TRANSACTION statement.
Note that READ ONLY and READ WRITE are logically mutually exclusive.
This is enforced by the parser and depended upon by trans_begin().
We need two flags instead of one in order to differentiate between
situation when no READ WRITE/ONLY clause were given and thus transaction
is implicitly READ WRITE and the case when READ WRITE clause was used
explicitly.
*/
// WITH CONSISTENT SNAPSHOT option
static const uint MYSQL_START_TRANS_OPT_WITH_CONS_SNAPSHOT = 1;
// READ ONLY option
static const uint MYSQL_START_TRANS_OPT_READ_ONLY = 2;
// READ WRITE option
static const uint MYSQL_START_TRANS_OPT_READ_WRITE = 4;
// HIGH PRIORITY option
static const uint MYSQL_START_TRANS_OPT_HIGH_PRIORITY = 8;
enum legacy_db_type {
DB_TYPE_UNKNOWN = 0,
DB_TYPE_DIAB_ISAM = 1,
DB_TYPE_HASH,
DB_TYPE_MISAM,
DB_TYPE_PISAM,
DB_TYPE_RMS_ISAM,
DB_TYPE_HEAP,
DB_TYPE_ISAM,
DB_TYPE_MRG_ISAM,
DB_TYPE_MYISAM,
DB_TYPE_MRG_MYISAM,
DB_TYPE_BERKELEY_DB,
DB_TYPE_INNODB,
DB_TYPE_GEMINI,
DB_TYPE_NDBCLUSTER,
DB_TYPE_EXAMPLE_DB,
DB_TYPE_ARCHIVE_DB,
DB_TYPE_CSV_DB,
DB_TYPE_FEDERATED_DB,
DB_TYPE_BLACKHOLE_DB,
DB_TYPE_PARTITION_DB, // No longer used.
DB_TYPE_BINLOG,
DB_TYPE_SOLID,
DB_TYPE_PBXT,
DB_TYPE_TABLE_FUNCTION,
DB_TYPE_MEMCACHE,
DB_TYPE_FALCON,
DB_TYPE_MARIA,
/** Performance schema engine. */
DB_TYPE_PERFORMANCE_SCHEMA,
DB_TYPE_TEMPTABLE,
DB_TYPE_FIRST_DYNAMIC = 42,
DB_TYPE_DEFAULT = 127 // Must be last
};
enum row_type : int {
ROW_TYPE_NOT_USED = -1,
ROW_TYPE_DEFAULT,
ROW_TYPE_FIXED,
ROW_TYPE_DYNAMIC,
ROW_TYPE_COMPRESSED,
ROW_TYPE_REDUNDANT,
ROW_TYPE_COMPACT,
/** Unused. Reserved for future versions. */
ROW_TYPE_PAGED
};
enum enum_binlog_func {
BFN_RESET_LOGS = 1,
BFN_RESET_SLAVE = 2,
BFN_BINLOG_WAIT = 3,
BFN_BINLOG_END = 4,
BFN_BINLOG_PURGE_FILE = 5,
BFN_BINLOG_PURGE_WAIT = 6
};
enum enum_binlog_command {
LOGCOM_CREATE_TABLE,
LOGCOM_ALTER_TABLE,
LOGCOM_RENAME_TABLE,
LOGCOM_DROP_TABLE,
LOGCOM_CREATE_DB,
LOGCOM_ALTER_DB,
LOGCOM_DROP_DB,
};
enum class enum_sampling_method { SYSTEM, NONE };
/* Bits in used_fields */
#define HA_CREATE_USED_AUTO (1L << 0)
#define HA_CREATE_USED_RAID (1L << 1) // RAID is no longer available
#define HA_CREATE_USED_UNION (1L << 2)
#define HA_CREATE_USED_INSERT_METHOD (1L << 3)
#define HA_CREATE_USED_MIN_ROWS (1L << 4)
#define HA_CREATE_USED_MAX_ROWS (1L << 5)
#define HA_CREATE_USED_AVG_ROW_LENGTH (1L << 6)
#define HA_CREATE_USED_PACK_KEYS (1L << 7)
#define HA_CREATE_USED_CHARSET (1L << 8)
#define HA_CREATE_USED_DEFAULT_CHARSET (1L << 9)
#define HA_CREATE_USED_DATADIR (1L << 10)
#define HA_CREATE_USED_INDEXDIR (1L << 11)
#define HA_CREATE_USED_ENGINE (1L << 12)
#define HA_CREATE_USED_CHECKSUM (1L << 13)
#define HA_CREATE_USED_DELAY_KEY_WRITE (1L << 14)
#define HA_CREATE_USED_ROW_FORMAT (1L << 15)
#define HA_CREATE_USED_COMMENT (1L << 16)
#define HA_CREATE_USED_PASSWORD (1L << 17)
#define HA_CREATE_USED_CONNECTION (1L << 18)
#define HA_CREATE_USED_KEY_BLOCK_SIZE (1L << 19)
/** Unused. Reserved for future versions. */
#define HA_CREATE_USED_TRANSACTIONAL (1L << 20)
/** Unused. Reserved for future versions. */
#define HA_CREATE_USED_PAGE_CHECKSUM (1L << 21)
/** This is set whenever STATS_PERSISTENT=0|1|default has been
specified in CREATE/ALTER TABLE. See also HA_OPTION_STATS_PERSISTENT in
include/my_base.h. It is possible to distinguish whether
STATS_PERSISTENT=default has been specified or no STATS_PERSISTENT= is
given at all. */
#define HA_CREATE_USED_STATS_PERSISTENT (1L << 22)
/**
This is set whenever STATS_AUTO_RECALC=0|1|default has been
specified in CREATE/ALTER TABLE. See enum_stats_auto_recalc.
It is possible to distinguish whether STATS_AUTO_RECALC=default
has been specified or no STATS_AUTO_RECALC= is given at all.
*/
#define HA_CREATE_USED_STATS_AUTO_RECALC (1L << 23)
/**
This is set whenever STATS_SAMPLE_PAGES=N|default has been
specified in CREATE/ALTER TABLE. It is possible to distinguish whether
STATS_SAMPLE_PAGES=default has been specified or no STATS_SAMPLE_PAGES= is
given at all.
*/
#define HA_CREATE_USED_STATS_SAMPLE_PAGES (1L << 24)
/**
This is set whenever a 'TABLESPACE=...' phrase is used on CREATE TABLE
*/
#define HA_CREATE_USED_TABLESPACE (1L << 25)
/** COMPRESSION="zlib|lz4|none" used during table create. */
#define HA_CREATE_USED_COMPRESS (1L << 26)
/** ENCRYPTION="Y" used during table create. */
#define HA_CREATE_USED_ENCRYPT (1L << 27)
/**
CREATE|ALTER SCHEMA|DATABASE|TABLE has an explicit COLLATE clause.
Implies HA_CREATE_USED_DEFAULT_CHARSET.
*/
#define HA_CREATE_USED_DEFAULT_COLLATE (1L << 28)
/** SECONDARY_ENGINE used during table create. */
#define HA_CREATE_USED_SECONDARY_ENGINE (1L << 29)
/**
CREATE|ALTER SCHEMA|DATABASE has an explicit ENCRYPTION clause.
Implies HA_CREATE_USED_DEFAULT_ENCRYPTION.
*/
#define HA_CREATE_USED_DEFAULT_ENCRYPTION (1L << 30)
/**
This option is used to convey that the create table should not
commit the operation and keep the transaction started.
*/
constexpr const uint64_t HA_CREATE_USED_START_TRANSACTION{1ULL << 31};
constexpr const uint64_t HA_CREATE_USED_ENGINE_ATTRIBUTE{1ULL << 32};
constexpr const uint64_t HA_CREATE_USED_SECONDARY_ENGINE_ATTRIBUTE{1ULL << 33};
/**
ALTER SCHEMA|DATABASE has an explicit READ_ONLY clause.
Implies HA_CREATE_USED_READ_ONLY.
*/
constexpr const uint64_t HA_CREATE_USED_READ_ONLY{1ULL << 34};
/**
These flags convey that the options AUTOEXTEND_SIZE has been
specified in the CREATE TABLE statement
*/
constexpr const uint64_t HA_CREATE_USED_AUTOEXTEND_SIZE{1ULL << 35};
/*
End of bits used in used_fields
*/
/*
Structure to hold list of database_name.table_name.
This is used at both mysqld and storage engine layer.
*/
struct st_handler_tablename {
const char *db;
const char *tablename;
};
#define MAXGTRIDSIZE 64
#define MAXBQUALSIZE 64
#define COMPATIBLE_DATA_YES 0
#define COMPATIBLE_DATA_NO 1
/*
These structures are used to pass information from a set of SQL commands
on add/drop/change tablespace definitions to the proper hton.
*/
#define UNDEF_NODEGROUP 65535
// FUTURE: Combine these two enums into one enum class
enum ts_command_type {
TS_CMD_NOT_DEFINED = -1,
CREATE_TABLESPACE = 0,
ALTER_TABLESPACE = 1,
CREATE_LOGFILE_GROUP = 2,
ALTER_LOGFILE_GROUP = 3,
DROP_TABLESPACE = 4,
DROP_LOGFILE_GROUP = 5,
CHANGE_FILE_TABLESPACE = 6,
ALTER_ACCESS_MODE_TABLESPACE = 7,
CREATE_UNDO_TABLESPACE = 8,
ALTER_UNDO_TABLESPACE = 9,
DROP_UNDO_TABLESPACE = 10
};
enum ts_alter_tablespace_type {
TS_ALTER_TABLESPACE_TYPE_NOT_DEFINED = -1,
ALTER_TABLESPACE_ADD_FILE = 1,
ALTER_TABLESPACE_DROP_FILE = 2,
ALTER_TABLESPACE_RENAME = 3,
ALTER_TABLESPACE_OPTIONS = 4,
ALTER_UNDO_TABLESPACE_SET_ACTIVE = 5,
ALTER_UNDO_TABLESPACE_SET_INACTIVE = 6
};
/**
Legacy struct for passing tablespace information to SEs.
FUTURE: Pass all info through dd objects
*/
class st_alter_tablespace {
public:
const char *tablespace_name = nullptr;
const char *logfile_group_name = nullptr;
ts_command_type ts_cmd_type = TS_CMD_NOT_DEFINED;
enum ts_alter_tablespace_type ts_alter_tablespace_type =
TS_ALTER_TABLESPACE_TYPE_NOT_DEFINED;
const char *data_file_name = nullptr;
const char *undo_file_name = nullptr;
ulonglong extent_size = 1024 * 1024; // Default 1 MByte
ulonglong undo_buffer_size = 8 * 1024 * 1024; // Default 8 MByte
ulonglong redo_buffer_size = 8 * 1024 * 1024; // Default 8 MByte
ulonglong initial_size = 128 * 1024 * 1024; // Default 128 MByte
std::optional<ulonglong> autoextend_size; // No autoextension as default
ulonglong max_size = 0; // Max size == initial size => no extension
ulonglong file_block_size = 0; // 0=default or must be a valid Page Size
uint nodegroup_id = UNDEF_NODEGROUP;
bool wait_until_completed = true;
const char *ts_comment = nullptr;
const char *encryption = nullptr;
bool is_tablespace_command() {
return ts_cmd_type == CREATE_TABLESPACE ||
ts_cmd_type == ALTER_TABLESPACE || ts_cmd_type == DROP_TABLESPACE ||
ts_cmd_type == CHANGE_FILE_TABLESPACE ||
ts_cmd_type == ALTER_ACCESS_MODE_TABLESPACE;
}
/**
Proper constructor even for all-public class simplifies initialization and
allows members to be const.
FUTURE: With constructor all members can be made const, and do not need
default initializers.
@param tablespace name of tabelspace (nullptr for logfile group statements)
@param logfile_group name of logfile group or nullptr
@param cmd main statement type
@param alter_tablespace_cmd subcommand type for ALTER TABLESPACE
@param datafile tablespace file for CREATE and ALTER ... ADD ...
@param undofile only applies to logfile group statements. nullptr otherwise.
@param opts options provided by parser
*/
st_alter_tablespace(const char *tablespace, const char *logfile_group,
ts_command_type cmd,
enum ts_alter_tablespace_type alter_tablespace_cmd,
const char *datafile, const char *undofile,
const Tablespace_options &opts);
};
/*
Make sure that the order of schema_tables and enum_schema_tables are the same.
*/
enum enum_schema_tables : int {
SCH_FIRST = 0,
SCH_COLUMN_PRIVILEGES = SCH_FIRST,
SCH_ENGINES,
SCH_OPEN_TABLES,
SCH_OPTIMIZER_TRACE,
SCH_PLUGINS,
SCH_PROCESSLIST,
SCH_PROFILES,
SCH_SCHEMA_PRIVILEGES,
SCH_TABLESPACES,
SCH_TABLE_PRIVILEGES,
SCH_USER_PRIVILEGES,
SCH_TMP_TABLE_COLUMNS,
SCH_TMP_TABLE_KEYS,
SCH_LAST = SCH_TMP_TABLE_KEYS
};
enum ha_stat_type { HA_ENGINE_STATUS, HA_ENGINE_LOGS, HA_ENGINE_MUTEX };
enum ha_notification_type : int { HA_NOTIFY_PRE_EVENT, HA_NOTIFY_POST_EVENT };
enum ha_ddl_type : int {
HA_INVALID_DDL,
HA_ALTER_DDL,
HA_TRUNCATE_DDL,
HA_RENAME_DDL
};
/** Clone start operation mode */
enum Ha_clone_mode {
/** Start a new clone operation */
HA_CLONE_MODE_START,
/** Re-start a clone operation after failure */
HA_CLONE_MODE_RESTART,
/** Add a new task to a running clone operation */
HA_CLONE_MODE_ADD_TASK,
/** Get version for transfer data format */
HA_CLONE_MODE_VERSION,
/** Max value for clone mode */
HA_CLONE_MODE_MAX
};
/** Clone operation types. */
enum Ha_clone_type : size_t {
/** Caller must block all write operation to the SE. */
HA_CLONE_BLOCKING,
/** For transactional SE, archive redo to support concurrent dml */
HA_CLONE_REDO,
/** For transactional SE, track page changes to support concurrent dml */
HA_CLONE_PAGE,
/** For transactional SE, use both page tracking and redo to optimize
clone with concurrent dml. Currently supported by Innodb. */
HA_CLONE_HYBRID,
/** SE supports multiple threads for clone */
HA_CLONE_MULTI_TASK,
/** SE supports restarting clone after network failure */
HA_CLONE_RESTART,
/** Maximum value of clone type */
HA_CLONE_TYPE_MAX
};
using Ha_clone_flagset = std::bitset<HA_CLONE_TYPE_MAX>;
void set_externally_disabled_storage_engine_names(const char *);
bool ha_is_externally_disabled(const handlerton &);
/** File reference for clone */
struct Ha_clone_file {
/** File reference type */
enum {
/** File handle */
FILE_HANDLE,
/** File descriptor */
FILE_DESC
} type;
/** File reference */
union {
/** File descriptor */
int file_desc;
/** File handle for windows */
void *file_handle;
};
};
/* Abstract callback interface to stream data back to the caller. */
class Ha_clone_cbk {
protected:
/** Constructor to initialize members. */
Ha_clone_cbk()
: m_hton(),
m_loc_idx(),
m_client_buff_size(),
m_data_desc(),
m_desc_len(),
m_src_name(),
m_dest_name(),
m_state_estimate(),
m_flag() {}
public:
/** Callback providing data from current position of a
file descriptor of specific length.
@param[in] from_file source file to read from
@param[in] len data length
@return error code */
virtual int file_cbk(Ha_clone_file from_file, uint len) = 0;
/** Callback providing data in buffer of specific length.
@param[in] from_buffer source buffer to read from
@param[in] len data length
@return error code */
virtual int buffer_cbk(uchar *from_buffer, uint len) = 0;
/** Callback providing a file descriptor to write data starting
from current position.
@param[in] to_file destination file to write data
@return error code */
virtual int apply_file_cbk(Ha_clone_file to_file) = 0;
/** Callback to get data in buffer.
@param[out] to_buffer data buffer
@param[out] len data length
@return error code */
virtual int apply_buffer_cbk(uchar *&to_buffer, uint &len) = 0;
/** virtual destructor. */
virtual ~Ha_clone_cbk() = default;
/** Set current storage engine handlerton.
@param[in] hton SE handlerton */
void set_hton(handlerton *hton) { m_hton = hton; }
/** Get current storage engine handlerton.
@return SE handlerton */
handlerton *get_hton() { return (m_hton); }
/** Set caller's transfer buffer size. SE can adjust the data chunk size
based on this parameter.
@param[in] size buffer size in bytes */
void set_client_buffer_size(uint size) { m_client_buff_size = size; }
/** Get caller's transfer buffer size.
@return buffer size in bytes */
uint get_client_buffer_size() { return (m_client_buff_size); }
/** Set current SE index.
@param[in] idx SE index in locator array */
void set_loc_index(uint idx) { m_loc_idx = idx; }
/** Get current SE index.
@return SE index in locator array */
uint get_loc_index() { return (m_loc_idx); }
/** Set data descriptor. SE specific descriptor for the
data transferred by the callbacks.
@param[in] desc serialized data descriptor
@param[in] len length of the descriptor byte stream */
void set_data_desc(const uchar *desc, uint len) {
m_data_desc = desc;
m_desc_len = len;
}
/** Get data descriptor. SE specific descriptor for the
data transferred by the callbacks.
@param[out] lenp length of the descriptor byte stream
@return pointer to the serialized data descriptor */
const uchar *get_data_desc(uint *lenp) {
if (lenp != nullptr) {
*lenp = m_desc_len;
}
return (m_data_desc);
}
/** Get SE source file name. Used for debug printing and error message.
@return null terminated string for source file name */
const char *get_source_name() { return (m_src_name); }
/** Set SE source file name.
@param[in] name null terminated string for source file name */
void set_source_name(const char *name) { m_src_name = name; }
/** Get SE destination file name. Used for debug printing and error message.
@return null terminated string for destination file name */
const char *get_dest_name() { return (m_dest_name); }
/** Set SE destination file name.
@param[in] name null terminated string for destination file name */
void set_dest_name(const char *name) { m_dest_name = name; }
/** Clear all flags set by SE */
void clear_flags() { m_flag = 0; }
/** Mark that ACK is needed for the data transfer before returning
from callback. Set by SE. */
void set_ack() { m_flag |= HA_CLONE_ACK; }
/** Check if ACK is needed for the data transfer
@return true if ACK is needed */
bool is_ack_needed() const { return (m_flag & HA_CLONE_ACK); }
/** Mark that the file descriptor is opened for read/write
with OS buffer cache. For O_DIRECT, the flag is not set. */
void set_os_buffer_cache() { m_flag |= HA_CLONE_FILE_CACHE; }
/** Check if the file descriptor is opened for read/write with OS
buffer cache. Currently clone avoids using zero copy (sendfile on linux),
if SE is using O_DIRECT. This improves data copy performance.
@return true if O_DIRECT is not used */
bool is_os_buffer_cache() const { return (m_flag & HA_CLONE_FILE_CACHE); }
/** Mark that the file can be transferred with zero copy. */
void set_zero_copy() { m_flag |= HA_CLONE_ZERO_COPY; }
/** Check if zero copy optimization is suggested. */
bool is_zero_copy() const { return (m_flag & HA_CLONE_ZERO_COPY); }
/** Mark that data needs secure transfer. */
void set_secure() { m_flag |= HA_CLONE_SECURE; }
/** Check if data needs secure transfer. */
bool is_secure() const { return (m_flag & HA_CLONE_SECURE); }
/** Set state information and notify state change.
@param[in] estimate estimated bytes for current state. */
void mark_state_change(uint64_t estimate) {
m_flag |= HA_CLONE_STATE_CHANGE;
m_state_estimate = estimate;
}
/** Check if SE notified state change. */
bool is_state_change(uint64_t &estimate) {
estimate = m_state_estimate;
return (m_flag & HA_CLONE_STATE_CHANGE);
}
private:
/** Handlerton for the SE */
handlerton *m_hton;
/** SE index in caller's locator array */
uint m_loc_idx;
/** Caller's transfer buffer size. */
uint m_client_buff_size;
/** SE's Serialized data descriptor */
const uchar *m_data_desc;
/** SE's Serialized descriptor length. */
uint m_desc_len;
/** Current source file name */
const char *m_src_name;
/** Current destination file name */
const char *m_dest_name;
/** Estimated bytes to be transferred. */
uint64_t m_state_estimate;
/** Flag storing data related options */
int m_flag;
/** Acknowledgement is needed for the data transfer. */
const int HA_CLONE_ACK = 0x01;
/** Data file is opened for read/write with OS buffer cache. */
const int HA_CLONE_FILE_CACHE = 0x02;
/** Data file can be transferred with zero copy. */
const int HA_CLONE_ZERO_COPY = 0x04;
/** Data needs to be transferred securely over SSL connection. */
const int HA_CLONE_SECURE = 0x08;
/** State change notification by SE. */
const int HA_CLONE_STATE_CHANGE = 0x10;
};
/**
Column type description for foreign key columns compatibility check.
Contains subset of information from dd::Column class. It is inconvenient
to use dd::Column class directly for such checks because it requires valid
dd::Table object and in some cases we want to produce Ha_fk_column_type
right from column description in Create_field format.
*/
struct Ha_fk_column_type {
dd::enum_column_types type;
/*
Note that both dd::Column::char_length() and length here are really
in bytes.
*/
size_t char_length;
const CHARSET_INFO *field_charset;
size_t elements_count;
uint numeric_scale;
bool is_unsigned;
};
typedef ulonglong my_xid; // this line is the same as in log_event.h
/**
Enumeration of possible states for externally coordinated transactions (XA).
*/
enum class enum_ha_recover_xa_state : int {
NOT_FOUND = -1, // Trnasaction not found
PREPARED_IN_SE = 0, // Transaction is prepared in SEs
PREPARED_IN_TC = 1, // Transaction is prepared in SEs and TC
COMMITTED_WITH_ONEPHASE = 2, // Transaction was one-phase committed
COMMITTED = 3, // Transaction was committed
ROLLEDBACK = 4 // Transaction was rolled back
};
/**
Single occurrence set of XIDs of internally coordinated transactions
found as been committed in the transaction coordinator state.
*/
using Xid_commit_list =
std::unordered_set<my_xid, std::hash<my_xid>, std::equal_to<my_xid>,
Mem_root_allocator<my_xid>>;
/**
Class to maintain list of externally coordinated transactions and their
current state at recovery.
*/
class Xa_state_list {
public:
using pair = std::pair<const XID, enum_ha_recover_xa_state>;
using allocator = Mem_root_allocator<Xa_state_list::pair>;
using list = std::map<XID, enum_ha_recover_xa_state, std::less<XID>,
Xa_state_list::allocator>;
using iterator = std::map<XID, enum_ha_recover_xa_state, std::less<XID>,
Xa_state_list::allocator>::iterator;
using instantiation_tuple = std::tuple<
std::unique_ptr<MEM_ROOT>, std::unique_ptr<Xa_state_list::allocator>,
std::unique_ptr<Xa_state_list::list>, std::unique_ptr<Xa_state_list>>;
/**
Class constructor.
@param populated_by_tc The underlying list of XIDs and transaction
states, after being populated by the transaction
coodinator.
*/
Xa_state_list(Xa_state_list::list &populated_by_tc);
virtual ~Xa_state_list() = default;
/**
Searches the underlying map to find an key that corresponds to the
parameter.
@param to_find The XID to find within the underlying map.
@return Ha_recover_states::NOT_FOUND if the transaction wasn't found,
the state of the transaction, otherwise.
*/
enum_ha_recover_xa_state find(XID const &to_find);
/**
Adds a transaction and state to the underlying map. If the given XID
already exists in the underlying map, the associated state changes according
to the following rules:
- If the parameter state is `PREPARED_IN_SE` it means that the
transaction didn't reach PREPARED_IN_TC, COMMIT or ROLLBACK for
sure. In that case:
. If other participants state is `COMMITTED`/`ROLLEDBACK`, it would
mean that it's a state inherited from a previous execution with the
same XID and we should set the state to `PREPARED_IN_SE`.
. If other participants state is `PREPARED_IN_TC`/
`COMMITTED_WITH_ONEPHASE` it means that the current participant
didn't reach it but some other did so, keep the state as
`PREPARED_IN_TC`/`COMMITTED_WITH_ONEPHASE`.
- If the parameter state is `PREPARED_IN_TC`, it means that other
participants must have persisted either the PREPARE, the COMMIT or
the ROLLBACK. In that case, keep whatever state is already there and
ensure that is not `PREPARED_IN_SE`.
- If the parameter state is `COMMITTED_WITH_ONEPHASE`, `COMMITTED` or
`ROLLEDBACK`, do nothing, only the active transaction coordinator has
the ability, for now, to set the transaction state to those values.
@param xid The XID to be added (the key).
@param xid The state to be added (the value).
@return The current value of the transaction state if the XID has
already been added, Ha_recover_states::NOT_FOUND otherwise.
*/
enum_ha_recover_xa_state add(XID const &xid, enum_ha_recover_xa_state state);
/**
Factory like method to instantiate all the infra-structure needed to
create an `Xa_state_list`. Since such infra-structuer is dependent on
`MEM_ROOT` and `Mem_root_allocator`, the method returns a tuple
containing unique pointers to all 4 objects needed: MEM_ROOT;
Mem_root_allocator; Xa_state_list::list; Xa_state_list.
@return An std::tuple containing unique pointers to all 4 objects
needed: MEM_ROOT; Mem_root_allocator; Xa_state_list::list;
Xa_state_list.
*/
static Xa_state_list::instantiation_tuple new_instance();
private:
/** The underlying map holding the trx and states*/
Xa_state_list::list &m_underlying;
};
/* handlerton methods */
/**
close_connection is only called if
thd->ha_data[xxx_hton.slot] is non-zero, so even if you don't need
this storage area - set it to something, so that MySQL would know
this storage engine was accessed in this connection
*/
typedef int (*close_connection_t)(handlerton *hton, THD *thd);
/** Terminate connection/statement notification. */
typedef void (*kill_connection_t)(handlerton *hton, THD *thd);
/**
Shut down all storage engine background tasks that might access
the data dictionary, before the main shutdown.
*/
typedef void (*pre_dd_shutdown_t)(handlerton *hton);
/**
sv points to a storage area, that was earlier passed
to the savepoint_set call
*/
typedef int (*savepoint_rollback_t)(handlerton *hton, THD *thd, void *sv);
/**
sv points to an uninitialized storage area of requested size
(see savepoint_offset description)
*/
typedef int (*savepoint_set_t)(handlerton *hton, THD *thd, void *sv);
/**
Check if storage engine allows to release metadata locks which were
acquired after the savepoint if rollback to savepoint is done.
@return true - If it is safe to release MDL locks.
false - If it is not.
*/
typedef bool (*savepoint_rollback_can_release_mdl_t)(handlerton *hton,
THD *thd);
typedef int (*savepoint_release_t)(handlerton *hton, THD *thd, void *sv);
/**
'all' is true if it's a real commit, that makes persistent changes
'all' is false if it's not in fact a commit but an end of the
statement that is part of the transaction.
NOTE 'all' is also false in auto-commit mode where 'end of statement'
and 'real commit' mean the same event.
*/
typedef int (*commit_t)(handlerton *hton, THD *thd, bool all);
typedef int (*rollback_t)(handlerton *hton, THD *thd, bool all);
typedef int (*prepare_t)(handlerton *hton, THD *thd, bool all);
typedef int (*recover_t)(handlerton *hton, XA_recover_txn *xid_list, uint len,
MEM_ROOT *mem_root);
/**
Retrieves information about externally coordinated transactions for which
the two-phase prepare was finished and transactions were prepared in the
server TC.
*/
using recover_prepared_in_tc_t = int (*)(handlerton *hton,
Xa_state_list &xa_list);
/**
Instructs the storage engine to mark the externally coordinated
transactions held by the THD parameters as prepared in the server TC.
*/
using set_prepared_in_tc_t = int (*)(handlerton *hton, THD *thd);
/** X/Open XA distributed transaction status codes */
enum xa_status_code {
/**
normal execution
*/
XA_OK = 0,
/**
asynchronous operation already outstanding
*/
XAER_ASYNC = -2,
/**
a resource manager error occurred in the transaction branch
*/
XAER_RMERR = -3,
/**
the XID is not valid
*/
XAER_NOTA = -4,
/**
invalid arguments were given
*/
XAER_INVAL = -5,
/**
routine invoked in an improper context
*/
XAER_PROTO = -6,
/**
resource manager unavailable
*/
XAER_RMFAIL = -7,
/**
the XID already exists
*/
XAER_DUPID = -8,
/**
resource manager doing work outside transaction
*/
XAER_OUTSIDE = -9
};
typedef xa_status_code (*commit_by_xid_t)(handlerton *hton, XID *xid);
typedef xa_status_code (*rollback_by_xid_t)(handlerton *hton, XID *xid);
/**
Instructs the storage engine to mark the externally coordinated
transactions identified by the XID parameters as prepared in the server
TC.
*/
using set_prepared_in_tc_by_xid_t = xa_status_code (*)(handlerton *hton,
XID *xid);
/**
Create handler object for the table in the storage engine.
@param hton Handlerton object for the storage engine.
@param table TABLE_SHARE for the table, can be NULL if caller
didn't perform full-blown open of table definition.
@param partitioned Indicates whether table is partitioned.
@param mem_root Memory root to be used for allocating handler
object.
*/
typedef handler *(*create_t)(handlerton *hton, TABLE_SHARE *table,
bool partitioned, MEM_ROOT *mem_root);
typedef void (*drop_database_t)(handlerton *hton, char *path);
typedef int (*panic_t)(handlerton *hton, enum ha_panic_function flag);
typedef int (*start_consistent_snapshot_t)(handlerton *hton, THD *thd);
/**
Flush the log(s) of storage engine(s).
@param hton Handlerton of storage engine.
@param binlog_group_flush true if we got invoked by binlog group
commit during flush stage, false in other cases.
@retval false Succeed
@retval true Error
*/
typedef bool (*flush_logs_t)(handlerton *hton, bool binlog_group_flush);
typedef bool (*show_status_t)(handlerton *hton, THD *thd, stat_print_fn *print,
enum ha_stat_type stat);
/**
The flag values are defined in sql_partition.h.
If this function is set, then it implies that the handler supports
partitioned tables.
If this function exists, then handler::get_partition_handler must also be
implemented.
*/
typedef uint (*partition_flags_t)();
/**
SE specific validation of the tablespace name.
This function will ask the relevant SE whether the submitted tablespace
name is valid.
@param ts_cmd Purpose of usage - is this tablespace DDL?
@param tablespace_name Name of the tablespace.
@return Tablespace name validity.
@retval Whether the tablespace name is valid.
*/
typedef bool (*is_valid_tablespace_name_t)(ts_command_type ts_cmd,
const char *tablespace_name);
/**
Get the tablespace name from the SE for the given schema and table.
@param thd Thread context.
@param db_name Name of the relevant schema.
@param table_name Name of the relevant table.
@param [out] tablespace_name Name of the tablespace containing the table.
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned).
*/
typedef int (*get_tablespace_t)(THD *thd, LEX_CSTRING db_name,
LEX_CSTRING table_name,
LEX_CSTRING *tablespace_name);
/**
Create/drop or alter tablespace in the storage engine.
@param hton Hadlerton of the SE.
@param thd Thread context.
@param ts_info Description of tablespace and specific
operation on it.
@param old_ts_def dd::Tablespace object describing old version
of tablespace.
@param [in,out] new_ts_def dd::Tablespace object describing new version
of tablespace. Engines which support atomic DDL
can adjust this object. The updated information
will be saved to the data-dictionary.
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned).
*/
typedef int (*alter_tablespace_t)(handlerton *hton, THD *thd,
st_alter_tablespace *ts_info,
const dd::Tablespace *old_ts_def,
dd::Tablespace *new_ts_def);
/**
SE interface for getting tablespace extension.
@return Extension of tablespace datafile name.
*/
typedef const char *(*get_tablespace_filename_ext_t)();
/**
Get the tablespace data from SE and insert it into Data dictionary
@param thd Thread context
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned)
*/
typedef int (*upgrade_tablespace_t)(THD *thd);
/**
Get the tablespace data from SE and insert it into Data dictionary
@param[in] tablespace tablespace object
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned)
*/
typedef bool (*upgrade_space_version_t)(dd::Tablespace *tablespace);
/**
Finish upgrade process inside storage engines.
This includes resetting flags to indicate upgrade process
and cleanup after upgrade.
@param thd Thread context
@param failed_upgrade True if the upgrade failed.
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned)
*/
typedef int (*finish_upgrade_t)(THD *thd, bool failed_upgrade);
/**
Upgrade logs after the checkpoint from where upgrade
process can only roll forward.
@param thd Thread context
@return Operation status.
@retval == 0 Success.
@retval != 0 Error (handler error code returned)
*/
typedef int (*upgrade_logs_t)(THD *thd);
enum class Tablespace_type {
SPACE_TYPE_DICTIONARY,
SPACE_TYPE_SYSTEM,
SPACE_TYPE_UNDO,
SPACE_TYPE_TEMPORARY,
SPACE_TYPE_SHARED,
SPACE_TYPE_IMPLICIT
};
/**
Get the tablespace type from the SE.
@param[in] space tablespace object
@param[out] space_type type of space
@return Operation status.
@retval false on success and true for failure.
*/
typedef bool (*get_tablespace_type_t)(const dd::Tablespace &space,
Tablespace_type *space_type);
/**
Get the tablespace type given the name, from the SE.
@param[in] tablespace_name tablespace name
@param[out] space_type type of space
@return Operation status.
@retval false on success and true for failure.
*/
typedef bool (*get_tablespace_type_by_name_t)(const char *tablespace_name,
Tablespace_type *space_type);
typedef int (*fill_is_table_t)(handlerton *hton, THD *thd, Table_ref *tables,
class Item *cond, enum enum_schema_tables);
typedef int (*binlog_func_t)(handlerton *hton, THD *thd, enum_binlog_func fn,
void *arg);
typedef void (*binlog_log_query_t)(handlerton *hton, THD *thd,
enum_binlog_command binlog_command,
const char *query, uint query_length,
const char *db, const char *table_name);
typedef void (*acl_notify_t)(THD *thd,
const class Acl_change_notification *notice);
typedef int (*discover_t)(handlerton *hton, THD *thd, const char *db,
const char *name, uchar **frmblob, size_t *frmlen);
typedef int (*find_files_t)(handlerton *hton, THD *thd, const char *db,
const char *path, const char *wild, bool dir,
List<LEX_STRING> *files);
typedef int (*table_exists_in_engine_t)(handlerton *hton, THD *thd,
const char *db, const char *name);
/**
Let storage engine inspect the query Accesspath and pick whatever
it like for being pushed down to the engine. (Join, conditions, ..)
The handler implementation should itself keep track of what it 'pushed',
such that later calls to the handlers access methods should
activate the pushed parts of the execution plan on the storage
engines.
@param thd Thread context
@param query The AccessPath for the entire query.
@param join The JOIN to be pushed
@returns
0 on success
error otherwise
*/
using push_to_engine_t = int (*)(THD *thd, AccessPath *query, JOIN *);
/**
Check if the given db.tablename is a system table for this SE.
@param db Database name to check.
@param table_name table name to check.
@param is_sql_layer_system_table if the supplied db.table_name is a SQL
layer system table.
@see example_is_supported_system_table in ha_example.cc
is_sql_layer_system_table is supplied to make more efficient
checks possible for SEs that support all SQL layer tables.
This interface is optional, so every SE need not implement it.
*/
typedef bool (*is_supported_system_table_t)(const char *db,
const char *table_name,
bool is_sql_layer_system_table);
/**
Create SDI in a tablespace. This API should be used when upgrading
a tablespace with no SDI or after invoking sdi_drop().
@param[in] tablespace tablespace object
@retval false success
@retval true failure
*/
typedef bool (*sdi_create_t)(dd::Tablespace *tablespace);
/**
Drop SDI in a tablespace. This API should be used only when
SDI is corrupted.
@param[in] tablespace tablespace object
@retval false success
@retval true failure
*/
typedef bool (*sdi_drop_t)(dd::Tablespace *tablespace);
/**
Get the SDI keys in a tablespace into vector.
@param[in] tablespace tablespace object
@param[in,out] vector vector of SDI Keys
@retval false success
@retval true failure
*/
typedef bool (*sdi_get_keys_t)(const dd::Tablespace &tablespace,
sdi_vector_t &vector);
/**
Retrieve SDI for a given SDI key.
Since the caller of this api will not know the SDI length, SDI retrieval
should be done in the following way.
i. Allocate initial memory of some size (Lets say 64KB)
ii. Pass the allocated memory to the below api.
iii. If passed buffer is sufficient, sdi_get_by_id() copies the sdi
to the buffer passed and returns success, else sdi_len is modified
with the actual length of the SDI (and returns false on failure).
For genuine errors, sdi_len is returned as UINT64_MAX
iv. If sdi_len != UINT64_MAX, retry the call after allocating the memory
of sdi_len
v. Free the memory after using SDI (responsibility of caller)
@param[in] tablespace tablespace object
@param[in] sdi_key SDI key to uniquely identify SDI obj
@param[in,out] sdi SDI retrieved from tablespace
A non-null pointer must be passed in
@param[in,out] sdi_len in: length of the memory allocated
out: actual length of SDI
@retval false success
@retval true failure
*/
typedef bool (*sdi_get_t)(const dd::Tablespace &tablespace,
const sdi_key_t *sdi_key, void *sdi, uint64 *sdi_len);
/**
Insert/Update SDI for a given SDI key.
@param[in] hton handlerton object
@param[in] tablespace tablespace object
@param[in] table table object
@param[in] sdi_key SDI key to uniquely identify SDI obj
@param[in] sdi SDI to write into the tablespace
@param[in] sdi_len length of SDI BLOB returned
@retval false success
@retval true failure, my_error() should be called
by SE
*/
typedef bool (*sdi_set_t)(handlerton *hton, const dd::Tablespace &tablespace,
const dd::Table *table, const sdi_key_t *sdi_key,
const void *sdi, uint64 sdi_len);
/**
Delete SDI for a given SDI key.
@param[in] tablespace tablespace object
@param[in] table table object
@param[in] sdi_key SDI key to uniquely identify SDI obj
@retval false success
@retval true failure, my_error() should be called
by SE
*/
typedef bool (*sdi_delete_t)(const dd::Tablespace &tablespace,
const dd::Table *table, const sdi_key_t *sdi_key);
/**
Check if the DDSE is started in a way that leaves thd DD being read only.
@retval true The data dictionary can only be read.
@retval false The data dictionary can be read and written.
*/
typedef bool (*is_dict_readonly_t)();
/**
Drop all temporary tables which have been left from previous server
run belonging to this SE. Used on server start-up.
@param[in] hton Handlerton for storage engine.
@param[in] thd Thread context.
@param[in,out] files List of files in directories for temporary files
which match tmp_file_prefix and thus can belong to
temporary tables (but not necessarily in this SE).
It is recommended to remove file from the list if
SE recognizes it as belonging to temporary table
in this SE and deletes it.
*/
typedef bool (*rm_tmp_tables_t)(handlerton *hton, THD *thd,
List<LEX_STRING> *files);
/**
Retrieve cost constants to be used for this storage engine.
A storage engine that wants to provide its own cost constants to
be used in the optimizer cost model, should implement this function.
The server will call this function to get a cost constant object
that will be used for tables stored in this storage engine instead
of using the default cost constants.
Life cycle for the cost constant object: The storage engine must
allocate the cost constant object on the heap. After the function
returns, the server takes over the ownership of this object.
The server will eventually delete the object by calling delete.
@note In the initial version the storage_category parameter will
not be used. The only valid value this will have is DEFAULT_STORAGE_CLASS
(see declaration in opt_costconstants.h).
@param storage_category the storage type that the cost constants will
be used for
@return a pointer to the cost constant object, if NULL is returned
the default cost constants will be used
*/
typedef SE_cost_constants *(*get_cost_constants_t)(uint storage_category);
/**
@param[in,out] thd pointer to THD
@param[in] new_trx_arg pointer to replacement transaction
@param[out] ptr_trx_arg double pointer to being replaced transaction
Associated with THD engine's native transaction is replaced
with @c new_trx_arg. The old value is returned through a buffer if non-null
pointer is provided with @c ptr_trx_arg.
The method is adapted by XA start and XA prepare handlers to
handle XA transaction that is logged as two parts by slave applier.
This interface concerns engines that are aware of XA transaction.
*/
typedef void (*replace_native_transaction_in_thd_t)(THD *thd, void *new_trx_arg,
void **ptr_trx_arg);
/** Mode for initializing the data dictionary. */
enum dict_init_mode_t {
DICT_INIT_CREATE_FILES, ///< Create all required SE files
DICT_INIT_CHECK_FILES, ///< Verify existence of expected files
DICT_INIT_UPGRADE_57_FILES, ///< Used for upgrade from mysql-5.7
DICT_INIT_IGNORE_FILES ///< Don't care about files at all
};
/**
Initialize the SE for being used to store the DD tables. Create
the required files according to the dict_init_mode. Create strings
representing the required DDSE tables, i.e., tables that the DDSE
expects to exist in the DD, and add them to the appropriate out
parameter.
@note There are two variants of this function type, one is to be
used by the DDSE, and has a different type of output parameters
because the SQL layer needs more information about the DDSE tables
in order to support upgrade.
@param dict_init_mode How to initialize files
@param version Target DD version if a new
server is being installed.
0 if restarting an existing
server.
@param [out] DDSE_tables List of SQL DDL statements
for creating DD tables that
are needed by the DDSE.
@param [out] DDSE_tablespaces List of meta data for predefined
tablespaces created by the DDSE.
@retval true An error occurred.
@retval false Success - no errors.
*/
typedef bool (*dict_init_t)(dict_init_mode_t dict_init_mode, uint version,
List<const Plugin_table> *DDSE_tables,
List<const Plugin_tablespace> *DDSE_tablespaces);
typedef bool (*ddse_dict_init_t)(
dict_init_mode_t dict_init_mode, uint version,
List<const dd::Object_table> *DDSE_tables,
List<const Plugin_tablespace> *DDSE_tablespaces);
/**
Initialize the set of hard coded DD table ids.
*/
typedef void (*dict_register_dd_table_id_t)(dd::Object_id hard_coded_tables);
/**
Invalidate an entry in the local dictionary cache.
Needed during bootstrap to make sure the contents in the DDSE
dictionary cache is in sync with the global DD.
@param schema_name Schema name.
@param table_name Table name.
*/
typedef void (*dict_cache_reset_t)(const char *schema_name,
const char *table_name);
/**
Invalidate all table and tablespace entries in the local dictionary cache.
Needed for recovery during server restart.
*/
typedef void (*dict_cache_reset_tables_and_tablespaces_t)();
/** Mode for data dictionary recovery. */
enum dict_recovery_mode_t {
DICT_RECOVERY_INITIALIZE_SERVER, ///< First start of a new server
DICT_RECOVERY_INITIALIZE_TABLESPACES, ///< First start, create tablespaces
DICT_RECOVERY_RESTART_SERVER ///< Restart of an existing server
};
/**
Do recovery in the DDSE as part of initializing the data dictionary.
The dict_recovery_mode indicates what kind of recovery should be
done.
@param dict_recovery_mode How to do recovery
@param version Target DD version if a new
server is being installed.
Actual DD version if restarting
an existing server.
@retval true An error occurred.
@retval false Success - no errors.
*/
typedef bool (*dict_recover_t)(dict_recovery_mode_t dict_recovery_mode,
uint version);
/**
Get the server version id stored in the header of the
dictionary tablespace.
@param [out] version Version number from the DD
tablespace header.
@retval Operation outcome, false if no error, otherwise true.
*/
typedef bool (*dict_get_server_version_t)(uint *version);
/**
Store the current server version number into the
header of the dictionary tablespace.
@retval Operation outcome, false if no error, otherwise true.
*/
typedef bool (*dict_set_server_version_t)();
/**
Notify/get permission from storage engine before acquisition or after
release of exclusive metadata lock on object represented by key.
@param thd Thread context.
@param mdl_key MDL key identifying object on which exclusive
lock is to be acquired/was released.
@param notification_type Indicates whether this is pre-acquire or
post-release notification.
@param victimized 'true' if locking failed as we were selected
as a victim in order to avoid possible deadlocks.
@note Notification is done only for objects from TABLESPACE, SCHEMA,
TABLE, FUNCTION, PROCEDURE, TRIGGER and EVENT namespaces.
@note Problems during notification are to be reported as warnings, MDL
subsystem will report generic error if pre-acquire notification
fails/SE refuses lock acquisition.
@note Return value is ignored/error is not reported in case of
post-release notification.
@note In some cases post-release notification might happen even if
there were no prior pre-acquire notification. For example,
when SE was loaded after exclusive lock acquisition, or when
we need notify SEs which permitted lock acquisition that it
didn't happen because one of SEs didn't allow it (in such case
we will do post-release notification for all SEs for simplicity).
@return False - if notification was successful/lock can be acquired,
True - if it has failed/lock should not be acquired.
*/
typedef bool (*notify_exclusive_mdl_t)(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type,
bool *victimized);
/**
Notify/get permission from storage engine before or after execution of
ALTER TABLE operation on the table identified by the MDL key.
@param thd Thread context.
@param mdl_key MDL key identifying table which is going to be
or was ALTERed.
@param notification_type Indicates whether this is pre-ALTER TABLE or
post-ALTER TABLE notification.
@note This hook is necessary because for ALTER TABLE upgrade to X
metadata lock happens fairly late during the execution process,
so it can be expensive to abort ALTER TABLE operation at this
stage by returning failure from notify_exclusive_mdl() hook.
@note This hook follows the same error reporting convention as
@see notify_exclusive_mdl().
@note Similarly to notify_exclusive_mdl() in some cases post-ALTER
notification might happen even if there were no prior pre-ALTER
notification.
@note Post-ALTER notification can happen before post-release notification
for exclusive metadata lock acquired by this ALTER TABLE.
@return False - if notification was successful/ALTER TABLE can proceed.
True - if it has failed/ALTER TABLE should be aborted.
*/
typedef bool (*notify_alter_table_t)(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type);
/**
Notify/get permission from storage engine before or after execution of
RENAME TABLE operation on the table identified by the MDL key.
@param thd Thread context.
@param mdl_key MDL key identifying table which is going to be
or was RENAMEd.
@param notification_type Indicates whether this is pre-RENAME TABLE or
post-RENAME TABLE notification.
@param old_db_name
@param old_table_name
@param new_db_name
@param new_table_name
*/
typedef bool (*notify_rename_table_t)(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type,
const char *old_db_name,
const char *old_table_name,
const char *new_db_name,
const char *new_table_name);
/**
Notify/get permission from storage engine before or after execution of
TRUNCATE TABLE operation on the table identified by the MDL key.
@param thd Thread context.
@param mdl_key MDL key identifying table which is going to be
or was TRUNCATEd.
@param notification_type Indicates whether this is pre-TRUNCATE TABLE or
post-TRUNCATE TABLE notification.
*/
typedef bool (*notify_truncate_table_t)(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type);
/**
@brief
Initiate master key rotation
@returns false on success,
true on failure
*/
typedef bool (*rotate_encryption_master_key_t)(void);
/**
@brief
Enable or Disable SE write ahead logging.
@param[in] thd server thread handle
@param[in] enable enable/disable redo logging
@return true iff failed.
*/
typedef bool (*redo_log_set_state_t)(THD *thd, bool enable);
/**
@brief
Retrieve ha_statistics from SE.
@param db_name Name of schema
@param table_name Name of table
@param se_private_id SE private id of the table.
@param ts_se_private_data Tablespace SE private data.
@param tbl_se_private_data Table SE private data.
@param flags Type of statistics to retrieve.
@param[out] stats Contains statistics read from SE.
@note Handlers that implement this callback/API should adhere
to servers expectation that, the implementation would invoke
my_error() before returning 'true'/failure from this function.
@returns false on success,
true on failure
*/
typedef bool (*get_table_statistics_t)(
const char *db_name, const char *table_name, dd::Object_id se_private_id,
const dd::Properties &ts_se_private_data,
const dd::Properties &tbl_se_private_data, uint flags,
ha_statistics *stats);
/**
@brief
Retrieve index column cardinality from SE.
@param db_name Name of schema
@param table_name Name of table
@param index_name Name of index
@param index_ordinal_position Position of index.
@param column_ordinal_position Position of column in index.
@param se_private_id SE private id of the table.
@param[out] cardinality cardinality being returned by SE.
@note Handlers that implement this callback/API should adhere
to servers expectation that, the implementation would invoke
my_error() before returning 'true'/failure from this function.
@returns false on success,
true on failure
*/
typedef bool (*get_index_column_cardinality_t)(
const char *db_name, const char *table_name, const char *index_name,
uint index_ordinal_position, uint column_ordinal_position,
dd::Object_id se_private_id, ulonglong *cardinality);
/**
Retrieve ha_tablespace_statistics from SE.
@param tablespace_name Tablespace_name
@param file_name Tablespace file name.
@param ts_se_private_data Tablespace SE private data.
@param[out] stats Contains tablespace
statistics read from SE.
@note Handlers that implement this callback/API should adhere
to servers expectation that, the implementation would invoke
my_error() before returning 'true'/failure from this function.
@returns false on success, true on failure
*/
typedef bool (*get_tablespace_statistics_t)(
const char *tablespace_name, const char *file_name,
const dd::Properties &ts_se_private_data, ha_tablespace_statistics *stats);
/* Database physical clone interfaces */
/** Get capability flags for clone operation
@param[out] flags capability flag */
using Clone_capability_t = void (*)(Ha_clone_flagset &flags);
/** Begin copy from source database
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in,out] loc locator
@param[in,out] loc_len locator length
@param[out] task_id task identifier
@param[in] type clone type
@param[in] mode mode for starting clone
@return error code */
using Clone_begin_t = int (*)(handlerton *hton, THD *thd, const uchar *&loc,
uint &loc_len, uint &task_id, Ha_clone_type type,
Ha_clone_mode mode);
/** Copy data from source database in chunks via callback
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in] loc locator
@param[in] loc_len locator length in bytes
@param[in] task_id task identifier
@param[in] cbk callback interface for sending data
@return error code */
using Clone_copy_t = int (*)(handlerton *hton, THD *thd, const uchar *loc,
uint loc_len, uint task_id, Ha_clone_cbk *cbk);
/** Acknowledge data transfer to source database
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in] loc locator
@param[in] loc_len locator length in bytes
@param[in] task_id task identifier
@param[in] in_err inform any error occurred
@param[in] cbk callback interface
@return error code */
using Clone_ack_t = int (*)(handlerton *hton, THD *thd, const uchar *loc,
uint loc_len, uint task_id, int in_err,
Ha_clone_cbk *cbk);
/** End copy from source database
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in] loc locator
@param[in] loc_len locator length in bytes
@param[in] task_id task identifier
@param[in] in_err error code when ending after error
@return error code */
using Clone_end_t = int (*)(handlerton *hton, THD *thd, const uchar *loc,
uint loc_len, uint task_id, int in_err);
/** Begin apply to destination database
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in,out] loc locator
@param[in,out] loc_len locator length
@param[in] task_id task identifier
@param[in] mode mode for starting clone
@param[in] data_dir target data directory
@return error code */
using Clone_apply_begin_t = int (*)(handlerton *hton, THD *thd,
const uchar *&loc, uint &loc_len,
uint &task_id, Ha_clone_mode mode,
const char *data_dir);
/** Apply data to destination database in chunks via callback
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in] loc locator
@param[in] loc_len locator length in bytes
@param[in] task_id task identifier
@param[in] in_err inform any error occurred
@param[in] cbk callback interface for receiving data
@return error code */
using Clone_apply_t = int (*)(handlerton *hton, THD *thd, const uchar *loc,
uint loc_len, uint task_id, int in_err,
Ha_clone_cbk *cbk);
/** End apply to destination database
@param[in] hton handlerton for SE
@param[in] thd server thread handle
@param[in] loc locator
@param[in] loc_len locator length in bytes
@param[in] task_id task identifier
@param[in] in_err error code when ending after error
@return error code */
using Clone_apply_end_t = int (*)(handlerton *hton, THD *thd, const uchar *loc,
uint loc_len, uint task_id, int in_err);
struct Clone_interface_t {
/* Get clone capabilities of an SE */
Clone_capability_t clone_capability;
/* Interfaces to copy data. */
Clone_begin_t clone_begin;
Clone_copy_t clone_copy;
Clone_ack_t clone_ack;
Clone_end_t clone_end;
/* Interfaces to apply data. */
Clone_apply_begin_t clone_apply_begin;
Clone_apply_t clone_apply;
Clone_apply_end_t clone_apply_end;
};
/**
Perform post-commit/rollback cleanup after DDL statement (e.g. in
case of DROP TABLES really remove table files from disk).
@note This hook will be invoked after DDL commit or rollback only
for storage engines supporting atomic DDL.
@note Problems during execution of this method should be reported to
error log and as warnings/notes to user. Since this method is
called after successful commit of the statement we can't fail
statement with error.
*/
typedef void (*post_ddl_t)(THD *thd);
/**
Perform SE-specific cleanup after recovery of transactions.
@note Particularly SEs supporting atomic DDL can use this call
to perform post-DDL actions for DDL statements which were
committed or rolled back during recovery stage.
*/
typedef void (*post_recover_t)(void);
/**
Lock a handlerton (resource) log to collect log information.
*/
typedef bool (*lock_hton_log_t)(handlerton *hton);
/**
Unlock a handlerton (resource) log after collecting log information.
*/
typedef bool (*unlock_hton_log_t)(handlerton *hton);
/**
Collect a handlerton (resource) log information.
*/
typedef bool (*collect_hton_log_info_t)(handlerton *hton, Json_dom *json);
/**
Check SE considers types of child and parent columns in foreign key
to be compatible.
@param child_column_type Child column type description.
@param parent_column_type Parent column type description.
@param check_charsets Indicates whether we need to check
that charsets of string columns
match. Which is true in most cases.
@returns True if types are compatible, False if not.
*/
typedef bool (*check_fk_column_compat_t)(
const Ha_fk_column_type *child_column_type,
const Ha_fk_column_type *parent_column_type, bool check_charsets);
typedef bool (*is_reserved_db_name_t)(handlerton *hton, const char *name);
/**
Prepare the secondary engine for executing a statement. This function is
called right after the secondary engine TABLE objects have been opened by
open_secondary_engine_tables(), before the statement is optimized and
executed. Secondary engines will typically create a context object in this
function, which they can use to store state that is needed during the
optimization and execution phases.
@param thd thread context
@param lex the statement to execute
@return true on error, false on success
*/
using prepare_secondary_engine_t = bool (*)(THD *thd, LEX *lex);
/**
Optimize a statement for execution on a secondary storage engine. This
function is called when the optimization of a statement has completed, just
before the statement is executed. Secondary engines can use this function to
apply engine-specific optimizations to the execution plan. They can also
reject executing the query by raising an error, in which case the query will
be reprepared and executed by the primary storage engine.
@param thd thread context
@param lex the statement being optimized
@return true on error, false on success
*/
using optimize_secondary_engine_t = bool (*)(THD *thd, LEX *lex);
/**
Compares the cost of two join plans in the secondary storage engine. The cost
of the current candidate is compared with the cost of the best plan seen so
far.
@param thd thread context
@param join the candidate plan to evaluate
@param optimizer_cost the cost estimate calculated by the optimizer
@param[out] use_best_so_far true if the optimizer should stop searching for
a better plan and use the best plan it has seen so far
@param[out] cheaper true if the candidate is the best plan seen so far for
this JOIN (must be true if it is the first plan seen),
false otherwise
@param[out] secondary_engine_cost the cost estimated by the secondary engine
@return false on success, or true if an error has been raised
*/
using compare_secondary_engine_cost_t = bool (*)(THD *thd, const JOIN &join,
double optimizer_cost,
bool *use_best_so_far,
bool *cheaper,
double *secondary_engine_cost);
/**
Evaluates the cost of executing the given access path in this secondary
storage engine, and potentially modifies the cost estimates that are in the
access path. This function is only called from the hypergraph join optimizer.
The function is called on every access path that the join optimizer might
compare to an alternative access path. This includes both paths that represent
complete execution plans and paths that represent partial plans. It is not
guaranteed to be called on every child path. For example, if GROUP BY is done
by sorting first and then aggregating the sorted results, the function will
only be called on the aggregation path, and not on the sort path, because only
the aggregation path will be compared to other paths.
The secondary engine is allowed to modify the estimates in the access path to
better match the costs of the access path in the secondary engine. It can
change any of the following AccessPath members:
- init_once_cost
- init_cost
- cost
- cost_before_filter
- num_output_rows
- num_output_rows_before_filter
- secondary_engine_data
Any other members should be left unchanged. The AccessPath must be in an
internally consistent state when the function returns, and satisfy invariants
expected by the hypergraph join optimizer, such as:
- init_cost <= cost_before_filter <= cost
- num_output_rows <= num_output_rows_before_filter
The secondary engine can also reject an access path altogether, by returning
true, in which case the join optimizer will not use that path in the final
plan. Since the secondary engine can reject any partial or complete plan, it
is possible that the join optimizer does not find any valid plan that is
accepted. In this case, the join optimizer will raise an error.
If the secondary encounters an error when evaluating the cost of the path, it
can signal an error by calling my_error() and return true, in which case the
join optimizer will not suggest any plan for the query.
@param thd The thread context.
@param hypergraph The hypergraph that represents the search space.
@param[in,out] access_path The AccessPath to evaluate.
@retval false on success.
@retval true if the plan is to be rejected, or if an error was raised.
*/
using secondary_engine_modify_access_path_cost_t = bool (*)(
THD *thd, const JoinHypergraph &hypergraph, AccessPath *access_path);
/**
Looks up and returns a specific secondary engine query offload or exec
failure reason as a string given a thread context (representing the query)
when the offloaded query fails in the secondary storage engine.
@param thd thread context.
@retval const char * as the offload failure reason.
The memory pointed to is managed by the handlerton and may be freed
when the statement completes.
*/
using get_secondary_engine_offload_or_exec_fail_reason_t =
const char *(*)(THD *thd);
/**
Sets a specific secondary engine offload failure reason for a query
represented by the thread context when the offloaded query fails in
the secondary storage engine.
@param thd thread context.
@param const char * as the offload failure reason.
*/
using set_secondary_engine_offload_fail_reason_t = void (*)(THD *thd,
const char *);
// Capabilities (bit flags) for secondary engines.
using SecondaryEngineFlags = uint64_t;
enum class SecondaryEngineFlag : SecondaryEngineFlags {
SUPPORTS_HASH_JOIN = 0,
SUPPORTS_NESTED_LOOP_JOIN = 1,
// If this flag is set, aggregation (GROUP BY and DISTINCT) do not require
// ordered inputs and create unordered outputs. This is typically the case
// if they are implemented using hash-based techniques.
AGGREGATION_IS_UNORDERED = 2,
/// This flag can be set to signal that a secondary storage engine will not
/// use MySQL's executor (see JOIN::override_executor_func). In this case, it
/// doesn't need MySQL's execution data structures, like internal temporary
/// tables, filesort objects or iterators. If the flag is set,
/// FinalizePlanForQueryBlock() will not make any changes to the plan, and
/// CreateIteratorFromAccessPath() will not be called.
USE_EXTERNAL_EXECUTOR = 3,
};
/// Creates an empty bitmap of access path types. This is the base
/// case for the function template with the same name below.
inline constexpr SecondaryEngineFlags MakeSecondaryEngineFlags() { return 0; }
/// Creates a bitmap representing a set of access path types.
template <typename... Args>
constexpr SecondaryEngineFlags MakeSecondaryEngineFlags(
SecondaryEngineFlag flag1, Args... rest) {
return (uint64_t{1} << static_cast<int>(flag1)) |
MakeSecondaryEngineFlags(rest...);
}
/// Returns the handlerton of the secondary engine that is used in the session,
/// or nullptr if a secondary engine is not used.
const handlerton *SecondaryEngineHandlerton(const THD *thd);
// FIXME: Temporary workaround to enable storage engine plugins to use the
// before_commit hook. Remove after WL#11320 has been completed.
typedef void (*se_before_commit_t)(void *arg);
// FIXME: Temporary workaround to enable storage engine plugins to use the
// after_commit hook. Remove after WL#11320 has been completed.
typedef void (*se_after_commit_t)(void *arg);
// FIXME: Temporary workaround to enable storage engine plugins to use the
// before_rollback hook. Remove after WL#11320 has been completed.
typedef void (*se_before_rollback_t)(void *arg);
/*
Page Tracking : interfaces to handlerton functions which starts/stops page
tracking, and purges/fetches page tracking information.
*/
/**
Start page tracking.
@param[out] start_id SE specific sequence number [LSN for InnoDB]
indicating when the tracking was started
@return Operation status.
@retval 0 Success
@retval other ER_* mysql error. Get error details from THD.
*/
using page_track_start_t = int (*)(uint64_t *start_id);
/**
Stop page tracking.
@param[out] stop_id SE specific sequence number [LSN for InnoDB]
indicating when the tracking was stopped
@return Operation status.
@retval 0 Success
@retval other ER_* mysql error. Get error details from THD.
*/
using page_track_stop_t = int (*)(uint64_t *stop_id);
/**
Purge page tracking data.
@param[in,out] purge_id SE specific sequence number [LSN for InnoDB]
initially indicating till where the data needs to be purged and finally
updated to until where it was actually purged
@return Operation status.
@retval 0 Success
@retval other ER_* mysql error. Get error details from THD.
*/
using page_track_purge_t = int (*)(uint64_t *purge_id);
/**
Fetch tracked pages.
@param[in] cbk_func callback function return page IDs
@param[in] cbk_ctx caller's context for callback
@param[in,out] start_id SE specific sequence number [LSN for InnoDB] from
where the pages tracked would be returned.
@note The range might get expanded and the actual start_id used for the
querying will be updated.
@param[in,out] stop_id SE specific sequence number [LSN for InnoDB]
until where the pages tracked would be returned.
@note The range might get expanded and the actual stop_id used for the
querying will be updated.
@param[out] buffer allocated buffer to copy page IDs
@param[in] buffer_len length of buffer in bytes
@return Operation status.
@retval 0 Success
@retval other ER_* mysql error. Get error details from THD.
*/
using page_track_get_page_ids_t = int (*)(Page_Track_Callback cbk_func,
void *cbk_ctx, uint64_t *start_id,
uint64_t *stop_id,
unsigned char *buffer,
size_t buffer_len);
/**
Fetch approximate number of tracked pages in the given range.
@param[in,out] start_id SE specific sequence number [LSN for InnoDB] from
where the pages tracked would be returned.
@note the range might get expanded and the actual start_id used for the
querying will be updated.
@param[in,out] stop_id SE specific sequence number [LSN for InnoDB]
until where the pages tracked would be returned.
@note the range might get expanded and the actual stop_id used for the
querying will be updated.
@param[out] num_pages number of pages tracked
@return Operation status.
@retval 0 Success
@retval other ER_* mysql error. Get error details from THD.
*/
using page_track_get_num_page_ids_t = int (*)(uint64_t *start_id,
uint64_t *stop_id,
uint64_t *num_pages);
/** Fetch the status of the page tracking system.
@param[out] status vector of a pair of (ID, bool) where ID is the
start/stop point and bool is true if the ID is a start point else false */
using page_track_get_status_t =
void (*)(std::vector<std::pair<uint64_t, bool>> &status);
/** Page track interface */
struct Page_track_t {
page_track_start_t start;
page_track_stop_t stop;
page_track_purge_t purge;
page_track_get_page_ids_t get_page_ids;
page_track_get_num_page_ids_t get_num_page_ids;
page_track_get_status_t get_status;
};
/**
handlerton is a singleton structure - one instance per storage engine -
to provide access to storage engine functionality that works on the
"global" level (unlike handler class that works on a per-table basis).
usually handlerton instance is defined statically in ha_xxx.cc as
static handlerton { ... } xxx_hton;
savepoint_*, prepare, recover, and *_by_xid pointers can be 0.
*/
struct handlerton {
/**
Historical marker for if the engine is available or not.
*/
SHOW_COMP_OPTION state;
/**
Historical number used for frm file to determine the correct storage engine.
This is going away and new engines will just use "name" for this.
*/
enum legacy_db_type db_type;
/**
Each storage engine has it's own memory area (actually a pointer)
in the thd, for storing per-connection information.
It is accessed as
thd->ha_data[xxx_hton.slot]
slot number is initialized by MySQL after xxx_init() is called.
*/
uint slot;
/**
To store per-savepoint data storage engine is provided with an area
of a requested size (0 is ok here).
savepoint_offset must be initialized statically to the size of
the needed memory to store per-savepoint information.
After xxx_init it is changed to be an offset to savepoint storage
area and need not be used by storage engine.
see binlog_hton and binlog_savepoint_set/rollback for an example.
*/
uint savepoint_offset;
/* handlerton methods */
close_connection_t close_connection;
kill_connection_t kill_connection;
pre_dd_shutdown_t pre_dd_shutdown;
savepoint_set_t savepoint_set;
savepoint_rollback_t savepoint_rollback;
savepoint_rollback_can_release_mdl_t savepoint_rollback_can_release_mdl;
savepoint_release_t savepoint_release;
commit_t commit;
rollback_t rollback;
prepare_t prepare;
recover_t recover;
recover_prepared_in_tc_t recover_prepared_in_tc;
commit_by_xid_t commit_by_xid;
rollback_by_xid_t rollback_by_xid;
set_prepared_in_tc_t set_prepared_in_tc;
set_prepared_in_tc_by_xid_t set_prepared_in_tc_by_xid;
create_t create;
drop_database_t drop_database;
panic_t panic;
start_consistent_snapshot_t start_consistent_snapshot;
flush_logs_t flush_logs;
show_status_t show_status;
partition_flags_t partition_flags;
is_valid_tablespace_name_t is_valid_tablespace_name;
get_tablespace_t get_tablespace;
alter_tablespace_t alter_tablespace;
get_tablespace_filename_ext_t get_tablespace_filename_ext;
upgrade_tablespace_t upgrade_tablespace;
upgrade_space_version_t upgrade_space_version;
get_tablespace_type_t get_tablespace_type;
get_tablespace_type_by_name_t get_tablespace_type_by_name;
upgrade_logs_t upgrade_logs;
finish_upgrade_t finish_upgrade;
fill_is_table_t fill_is_table;
dict_init_t dict_init;
ddse_dict_init_t ddse_dict_init;
dict_register_dd_table_id_t dict_register_dd_table_id;
dict_cache_reset_t dict_cache_reset;
dict_cache_reset_tables_and_tablespaces_t
dict_cache_reset_tables_and_tablespaces;
dict_recover_t dict_recover;
dict_get_server_version_t dict_get_server_version;
dict_set_server_version_t dict_set_server_version;
is_reserved_db_name_t is_reserved_db_name;
/** Global handler flags. */
uint32 flags{0};
/*
Those handlerton functions below are properly initialized at handler
init.
*/
binlog_func_t binlog_func;
binlog_log_query_t binlog_log_query;
acl_notify_t acl_notify;
discover_t discover;
find_files_t find_files;
table_exists_in_engine_t table_exists_in_engine;
push_to_engine_t push_to_engine;
is_supported_system_table_t is_supported_system_table;
/*
APIs for retrieving Serialized Dictionary Information by tablespace id
*/
sdi_create_t sdi_create;
sdi_drop_t sdi_drop;
sdi_get_keys_t sdi_get_keys;
sdi_get_t sdi_get;
sdi_set_t sdi_set;
sdi_delete_t sdi_delete;
/**
Null-ended array of file extensions that exist for the storage engine.
Used by frm_error() and the default handler::rename_table and delete_table
methods in handler.cc.
For engines that have two file name extensions (separate meta/index file
and data file), the order of elements is relevant. First element of engine
file name extensions array should be meta/index file extension. Second
element - data file extension. This order is assumed by
prepare_for_repair() when REPAIR TABLE ... USE_FRM is issued.
For engines that don't have files, file_extensions is NULL.
Currently, the following alternatives are used:
- file_extensions == NULL;
- file_extensions[0] != NULL, file_extensions[1] == NULL;
- file_extensions[0] != NULL, file_extensions[1] != NULL,
file_extensions[2] == NULL;
*/
const char **file_extensions;
is_dict_readonly_t is_dict_readonly;
rm_tmp_tables_t rm_tmp_tables;
get_cost_constants_t get_cost_constants;
replace_native_transaction_in_thd_t replace_native_transaction_in_thd;
notify_exclusive_mdl_t notify_exclusive_mdl;
notify_alter_table_t notify_alter_table;
notify_rename_table_t notify_rename_table;
notify_truncate_table_t notify_truncate_table;
rotate_encryption_master_key_t rotate_encryption_master_key;
redo_log_set_state_t redo_log_set_state;
get_table_statistics_t get_table_statistics;
get_index_column_cardinality_t get_index_column_cardinality;
get_tablespace_statistics_t get_tablespace_statistics;
post_ddl_t post_ddl;
post_recover_t post_recover;
/** Clone data transfer interfaces */
Clone_interface_t clone_interface;
/** Flag for Engine License. */
uint32 license;
/** Location for engines to keep personal structures. */
void *data;
/*
Log_resource functions that must be supported by storage engines
with relevant log information to be collected.
*/
lock_hton_log_t lock_hton_log;
unlock_hton_log_t unlock_hton_log;
collect_hton_log_info_t collect_hton_log_info;
/** Flags describing details of foreign key support by storage engine. */
uint32 foreign_keys_flags;
check_fk_column_compat_t check_fk_column_compat;
/**
Suffix for auto-generated foreign key names for tables using this storage
engine. If such suffix is specified by SE then its generated foreign key
names follow (table name)(SE-specific FK name suffix)(FK number) pattern.
Length of such suffix should not exceed MAX_FK_NAME_SUFFIX_LENGTH bytes.
If no suffix is specified then FK_NAME_DEFAULT_SUFFIX is used as
default.
*/
LEX_CSTRING fk_name_suffix;
/**
Pointer to a function that prepares a secondary engine for executing a
statement.
@see prepare_secondary_engine_t for function signature.
*/
prepare_secondary_engine_t prepare_secondary_engine;
/**
Pointer to a function that optimizes the current statement for
execution on the secondary storage engine represented by this
handlerton.
@see optimize_secondary_engine_t for function signature.
*/
optimize_secondary_engine_t optimize_secondary_engine;
/**
Pointer to a function that estimates the cost of executing a join in a
secondary storage engine.
@see compare_secondary_engine_cost_t for function signature.
*/
compare_secondary_engine_cost_t compare_secondary_engine_cost;
/// Bitmap which contains the supported join types and other flags
/// for a secondary storage engine when used with the hypergraph join
/// optimizer. If it is empty, it means that the secondary engine
/// does not support the hypergraph join optimizer.
SecondaryEngineFlags secondary_engine_flags;
/// Pointer to a function that evaluates the cost of executing an access path
/// in a secondary storage engine.
///
/// @see secondary_engine_modify_access_path_cost_t for function signature.
secondary_engine_modify_access_path_cost_t
secondary_engine_modify_access_path_cost;
/// Pointer to a function that returns the query offload or exec failure
/// reason as a string given a thread context (representing the query) when
/// the offloaded query failed in a secondary storage engine.
///
/// @see get_secondary_engine_offload_or_exec_fail_reason_t for function
/// signature.
get_secondary_engine_offload_or_exec_fail_reason_t
get_secondary_engine_offload_or_exec_fail_reason;
/// Pointer to a function that sets the offload failure reason as a string
/// for a thread context (representing the query) when the offloaded query
/// failed in a secondary storage engine.
///
/// @see set_secondary_engine_offload_fail_reason_t for function signature.
set_secondary_engine_offload_fail_reason_t
set_secondary_engine_offload_fail_reason;
se_before_commit_t se_before_commit;
se_after_commit_t se_after_commit;
se_before_rollback_t se_before_rollback;
/** Page tracking interface */
Page_track_t page_track;
};
/* Possible flags of a handlerton (there can be 32 of them) */
#define HTON_NO_FLAGS 0
#define HTON_CLOSE_CURSORS_AT_COMMIT (1 << 0)
#define HTON_ALTER_NOT_SUPPORTED (1 << 1) // Engine does not support alter
#define HTON_CAN_RECREATE (1 << 2) // Delete all is used for truncate
#define HTON_HIDDEN (1 << 3) // Engine does not appear in lists
/*
Bit 4 was occupied by BDB-specific HTON_FLUSH_AFTER_RENAME flag and is no
longer used.
*/
#define HTON_NOT_USER_SELECTABLE (1 << 5)
#define HTON_TEMPORARY_NOT_SUPPORTED \
(1 << 6) // Having temporary tables not supported
#define HTON_SUPPORT_LOG_TABLES (1 << 7) // Engine supports log tables
#define HTON_NO_PARTITION (1 << 8) // You can not partition these tables
/*
This flag should be set when deciding that the engine does not allow row based
binary logging (RBL) optimizations.
Currently, setting this flag, means that table's read/write_set will be left
untouched when logging changes to tables in this engine. In practice this
means that the server will not mess around with table->write_set and/or
table->read_set when using RBL and deciding whether to log full or minimal
rows.
It's valuable for instance for virtual tables, eg: Performance Schema which
have no meaning for replication.
*/
#define HTON_NO_BINLOG_ROW_OPT (1 << 9)
/**
Engine supports extended keys. The flag allows to
use 'extended key' feature if the engine is able to
do it (has primary key values in the secondary key).
Note that handler flag HA_PRIMARY_KEY_IN_READ_INDEX is
actually partial case of HTON_SUPPORTS_EXTENDED_KEYS.
*/
#define HTON_SUPPORTS_EXTENDED_KEYS (1 << 10)
// Engine support foreign key constraint.
#define HTON_SUPPORTS_FOREIGN_KEYS (1 << 11)
/**
Engine supports atomic DDL. That is rollback of transaction for DDL
statement will also rollback all changes in SE, commit of transaction
of DDL statement will make it durable.
*/
#define HTON_SUPPORTS_ATOMIC_DDL (1 << 12)
/* Engine supports packed keys. */
#define HTON_SUPPORTS_PACKED_KEYS (1 << 13)
/** Engine is a secondary storage engine. */
#define HTON_IS_SECONDARY_ENGINE (1 << 14)
/** Engine supports secondary storage engines. */
#define HTON_SUPPORTS_SECONDARY_ENGINE (1 << 15)
/** Engine supports table or tablespace encryption . */
#define HTON_SUPPORTS_TABLE_ENCRYPTION (1 << 16)
constexpr const decltype(handlerton::flags) HTON_SUPPORTS_ENGINE_ATTRIBUTE{
1 << 17};
/** Engine supports Generated invisible primary key. */
constexpr const decltype(handlerton::flags)
HTON_SUPPORTS_GENERATED_INVISIBLE_PK{1 << 18};
/** Whether the secondary engine supports DDLs. No meaning if the engine is not
* secondary. */
#define HTON_SECONDARY_ENGINE_SUPPORTS_DDL (1 << 19)
/** Whether the engine does not support triggers. */
#define HTON_NO_TRIGGER_SUPPORT (1 << 20)
/** Whether the primary engine supports external data sources. This case refers
to having tables with data in object store and the engine does not store any
of those data, only metadata. Table contents can be accessed only after
loading the table in the secondary storage engine. The flag is used for
a primary engine only.
*/
#define HTON_SUPPORTS_EXTERNAL_SOURCE (1 << 21)
inline bool secondary_engine_supports_ddl(const handlerton *hton) {
assert(hton->flags & HTON_IS_SECONDARY_ENGINE);
return (hton->flags & HTON_SECONDARY_ENGINE_SUPPORTS_DDL) != 0;
}
inline bool ddl_is_atomic(const handlerton *hton) {
return (hton->flags & HTON_SUPPORTS_ATOMIC_DDL) != 0;
}
/* Bits for handlerton::foreign_keys_flags bitmap. */
/**
Engine supports both unique and non-unique parent keys for
foreign keys which contain full foreign key as its prefix.
Storage engines which support foreign keys but do not have
this flag set are assumed to support only parent keys which
are primary/unique and contain exactly the same columns as
the foreign key, possibly, in different order.
*/
static const uint32 HTON_FKS_WITH_PREFIX_PARENT_KEYS = (1 << 0);
/**
Storage engine supports hash keys as supporting keys for foreign
keys. Hash key should contain all foreign key columns and only
them (although in any order).
Storage engines which support foreign keys but do not have this
flag set are assumed to not allow hash keys as supporting keys.
*/
static const uint32 HTON_FKS_WITH_SUPPORTING_HASH_KEYS = (1 << 1);
/**
Storage engine supports non-hash keys which have common prefix
with the foreign key as supporting keys for it. If there are
several such keys, one which shares biggest prefix with FK is
chosen.
Storage engines which support foreign keys but do not have this
flag set are assumed to require that supporting key contains full
foreign key as its prefix.
*/
static const uint32 HTON_FKS_WITH_ANY_PREFIX_SUPPORTING_KEYS = (1 << 2);
/**
Storage engine does not support using the same key for both parent
and supporting key, but requires the two to be different.
*/
static const uint32 HTON_FKS_NEED_DIFFERENT_PARENT_AND_SUPPORTING_KEYS =
(1 << 3);
/**
Engine takes into account hidden part of key (coming from primary key)
when determines if it can serve as parent key for a foreign key.
Implies HTON_FKS_WITH_PREFIX_PARENT_KEYS and is related to
HTON_SUPPORTS_EXTENDED_KEYS.
*/
static const uint32 HTON_FKS_WITH_EXTENDED_PARENT_KEYS = (1 << 4);
/**
Maximum possible length of SE-specific suffixes for auto-generated
foreign key names.
*/
static const size_t MAX_FK_NAME_SUFFIX_LENGTH = 16;
/**
Suffix for auto-generated foreign key names for tables in SE's which
don't specify own suffix. I.e. for foreign keys on tables in such
SE's generated names follow (table name)FK_NAME_DEFAULT_SUFFIX(FK number)
pattern.
*/
static const LEX_CSTRING FK_NAME_DEFAULT_SUFFIX = {STRING_WITH_LEN("_fk_")};
enum enum_tx_isolation : int {
ISO_READ_UNCOMMITTED,
ISO_READ_COMMITTED,
ISO_REPEATABLE_READ,
ISO_SERIALIZABLE
};
enum enum_stats_auto_recalc : int {
HA_STATS_AUTO_RECALC_DEFAULT = 0,
HA_STATS_AUTO_RECALC_ON,
HA_STATS_AUTO_RECALC_OFF
};
/**
Struct to hold information about the table that should be created.
*/
struct HA_CREATE_INFO {
const CHARSET_INFO *table_charset{nullptr};
const CHARSET_INFO *default_table_charset{nullptr};
bool schema_read_only{false};
LEX_STRING connect_string{nullptr, 0};
const char *password{nullptr};
const char *tablespace{nullptr};
LEX_STRING comment{nullptr, 0};
/**
Algorithm (and possible options) to be used for InnoDB's transparent
page compression. If this attribute is set then it is hint to the
storage engine to try and compress the data using the specified algorithm
where possible. Note: this value is interpreted by the storage engine only.
and ignored by the Server layer. */
LEX_STRING compress{nullptr, 0};
/**
This attribute is used for InnoDB's transparent page encryption.
If this attribute is set then it is hint to the storage engine to encrypt
the data. Note: this value is interpreted by the storage engine only.
and ignored by the Server layer. */
LEX_STRING encrypt_type{nullptr, 0};
/**
* Secondary engine of the table.
* Is nullptr if no secondary engine defined.
*/
LEX_CSTRING secondary_engine{nullptr, 0};
/** Secondary engine load status */
bool secondary_load{false};
const char *data_file_name{nullptr};
const char *index_file_name{nullptr};
const char *alias{nullptr};
ulonglong max_rows{0};
ulonglong min_rows{0};
ulonglong auto_increment_value{0};
ulong table_options{0};
ulong avg_row_length{0};
uint64_t used_fields{0};
// Can only be 1,2,4,8 or 16, but use uint32_t since that how it is
// represented in InnoDB
std::uint32_t key_block_size{0};
uint stats_sample_pages{0}; /* number of pages to sample during
stats estimation, if used, otherwise 0. */
enum_stats_auto_recalc stats_auto_recalc{HA_STATS_AUTO_RECALC_DEFAULT};
SQL_I_List<Table_ref> merge_list;
handlerton *db_type{nullptr};
/**
Row type of the table definition.
Defaults to ROW_TYPE_DEFAULT for all non-ALTER statements.
For ALTER TABLE defaults to ROW_TYPE_NOT_USED (means "keep the current").
Can be changed either explicitly by the parser.
If nothing specified inherits the value of the original table (if present).
*/
enum row_type row_type = ROW_TYPE_DEFAULT;
uint null_bits{0}; /* NULL bits at start of record */
uint options{0}; /* OR of HA_CREATE_ options */
uint merge_insert_method{0};
ha_storage_media storage_media{HA_SM_DEFAULT}; /* DEFAULT, DISK or MEMORY */
/*
A flag to indicate if this table should be marked as a hidden table in
the data dictionary. One use case is to mark the temporary tables
created by ALTER to be marked as hidden.
*/
bool m_hidden{false};
/*
A flag to indicate if this table should be created but not committed at
the end of statement.
*/
bool m_transactional_ddl{false};
LEX_CSTRING engine_attribute = NULL_CSTR;
LEX_CSTRING secondary_engine_attribute = NULL_CSTR;
ulonglong m_implicit_tablespace_autoextend_size{0};
bool m_implicit_tablespace_autoextend_size_change{true};
/**
Fill HA_CREATE_INFO to be used by ALTER as well as upgrade code.
This function separates code from mysql_prepare_alter_table() to be
used by upgrade code as well to reduce code duplication.
For ALTER code path, this lets new create options override the old
ones.
@param[in] share TABLE_SHARE object
@param[in] used_fields If a given create option is not flagged, old
value be copied from the TABLE_SHARE.
*/
void init_create_options_from_share(const TABLE_SHARE *share,
uint64_t used_fields);
};
/**
Structure describing changes to an index to be caused by ALTER TABLE.
*/
struct KEY_PAIR {
/**
Pointer to KEY object describing old version of index in
TABLE::key_info array for TABLE instance representing old
version of table.
*/
KEY *old_key;
/**
Pointer to KEY object describing new version of index in
Alter_inplace_info::key_info_buffer array.
*/
KEY *new_key;
};
/**
In-place alter handler context.
This is a superclass intended to be subclassed by individual handlers
in order to store handler unique context between in-place alter API calls.
The handler is responsible for creating the object. This can be done
as early as during check_if_supported_inplace_alter().
The SQL layer is responsible for destroying the object.
@see Alter_inplace_info
*/
class inplace_alter_handler_ctx {
public:
inplace_alter_handler_ctx() = default;
virtual void set_shared_data(const inplace_alter_handler_ctx *ctx
[[maybe_unused]]) {}
virtual ~inplace_alter_handler_ctx() = default;
};
/**
Class describing changes to be done by ALTER TABLE.
Instance of this class is passed to storage engine in order
to determine if this ALTER TABLE can be done using in-place
algorithm. It is also used for executing the ALTER TABLE
using in-place algorithm.
*/
class Alter_inplace_info {
public:
/**
Bits to show in detail what operations the storage engine is
to execute.
All these operations are supported as in-place operations by the
SQL layer. This means that operations that by their nature must
be performed by copying the table to a temporary table, will not
have their own flags here (e.g. ALTER TABLE FORCE, ALTER TABLE
ENGINE).
We generally try to specify handler flags only if there are real
changes. But in cases when it is cumbersome to determine if some
attribute has really changed we might choose to set flag
pessimistically, for example, relying on parser output only.
*/
typedef ulonglong HA_ALTER_FLAGS;
// Add non-unique, non-primary index
static const HA_ALTER_FLAGS ADD_INDEX = 1ULL << 0;
// Drop non-unique, non-primary index
static const HA_ALTER_FLAGS DROP_INDEX = 1ULL << 1;
// Add unique, non-primary index
static const HA_ALTER_FLAGS ADD_UNIQUE_INDEX = 1ULL << 2;
// Drop unique, non-primary index
static const HA_ALTER_FLAGS DROP_UNIQUE_INDEX = 1ULL << 3;
// Add primary index
static const HA_ALTER_FLAGS ADD_PK_INDEX = 1ULL << 4;
// Drop primary index
static const HA_ALTER_FLAGS DROP_PK_INDEX = 1ULL << 5;
// Add column
// Virtual generated column
static const HA_ALTER_FLAGS ADD_VIRTUAL_COLUMN = 1ULL << 6;
// Stored base (non-generated) column
static const HA_ALTER_FLAGS ADD_STORED_BASE_COLUMN = 1ULL << 7;
// Stored generated column
static const HA_ALTER_FLAGS ADD_STORED_GENERATED_COLUMN = 1ULL << 8;
// Add generic column (convenience constant).
static const HA_ALTER_FLAGS ADD_COLUMN =
ADD_VIRTUAL_COLUMN | ADD_STORED_BASE_COLUMN | ADD_STORED_GENERATED_COLUMN;
// Drop column
static const HA_ALTER_FLAGS DROP_VIRTUAL_COLUMN = 1ULL << 9;
static const HA_ALTER_FLAGS DROP_STORED_COLUMN = 1ULL << 10;
static const HA_ALTER_FLAGS DROP_COLUMN =
DROP_VIRTUAL_COLUMN | DROP_STORED_COLUMN;
// Rename column
static const HA_ALTER_FLAGS ALTER_COLUMN_NAME = 1ULL << 11;
// Change column datatype
static const HA_ALTER_FLAGS ALTER_VIRTUAL_COLUMN_TYPE = 1ULL << 12;
static const HA_ALTER_FLAGS ALTER_STORED_COLUMN_TYPE = 1ULL << 13;
/**
Change column datatype in such way that new type has compatible
packed representation with old type, so it is theoretically
possible to perform change by only updating data dictionary
without changing table rows.
*/
static const HA_ALTER_FLAGS ALTER_COLUMN_EQUAL_PACK_LENGTH = 1ULL << 14;
/// A virtual column has changed its position
static const HA_ALTER_FLAGS ALTER_VIRTUAL_COLUMN_ORDER = 1ULL << 15;
/// A stored column has changed its position (disregarding virtual columns)
static const HA_ALTER_FLAGS ALTER_STORED_COLUMN_ORDER = 1ULL << 16;
// Change column from NOT NULL to NULL
static const HA_ALTER_FLAGS ALTER_COLUMN_NULLABLE = 1ULL << 17;
// Change column from NULL to NOT NULL
static const HA_ALTER_FLAGS ALTER_COLUMN_NOT_NULLABLE = 1ULL << 18;
// Set or remove default column value
static const HA_ALTER_FLAGS ALTER_COLUMN_DEFAULT = 1ULL << 19;
// Change column generation expression
static const HA_ALTER_FLAGS ALTER_VIRTUAL_GCOL_EXPR = 1ULL << 20;
static const HA_ALTER_FLAGS ALTER_STORED_GCOL_EXPR = 1ULL << 21;
// Add foreign key
static const HA_ALTER_FLAGS ADD_FOREIGN_KEY = 1ULL << 22;
// Drop foreign key
static const HA_ALTER_FLAGS DROP_FOREIGN_KEY = 1ULL << 23;
// table_options changed, see HA_CREATE_INFO::used_fields for details.
static const HA_ALTER_FLAGS CHANGE_CREATE_OPTION = 1ULL << 24;
// Table is renamed
static const HA_ALTER_FLAGS ALTER_RENAME = 1ULL << 25;
// Change the storage type of column
static const HA_ALTER_FLAGS ALTER_COLUMN_STORAGE_TYPE = 1ULL << 26;
// Change the column format of column
static const HA_ALTER_FLAGS ALTER_COLUMN_COLUMN_FORMAT = 1ULL << 27;
// Add partition
static const HA_ALTER_FLAGS ADD_PARTITION = 1ULL << 28;
// Drop partition
static const HA_ALTER_FLAGS DROP_PARTITION = 1ULL << 29;
// Changing partition options
static const HA_ALTER_FLAGS ALTER_PARTITION = 1ULL << 30;
// Coalesce partition
static const HA_ALTER_FLAGS COALESCE_PARTITION = 1ULL << 31;
// Reorganize partition ... into
static const HA_ALTER_FLAGS REORGANIZE_PARTITION = 1ULL << 32;
// Reorganize partition
static const HA_ALTER_FLAGS ALTER_TABLE_REORG = 1ULL << 33;
// Remove partitioning
static const HA_ALTER_FLAGS ALTER_REMOVE_PARTITIONING = 1ULL << 34;
// Partition operation with ALL keyword
static const HA_ALTER_FLAGS ALTER_ALL_PARTITION = 1ULL << 35;
/**
Rename index. Note that we set this flag only if there are no other
changes to the index being renamed. Also for simplicity we don't
detect renaming of indexes which is done by dropping index and then
re-creating index with identical definition under different name.
*/
static const HA_ALTER_FLAGS RENAME_INDEX = 1ULL << 36;
/**
Recreate the table for ALTER TABLE FORCE, ALTER TABLE ENGINE
and OPTIMIZE TABLE operations.
*/
static const HA_ALTER_FLAGS RECREATE_TABLE = 1ULL << 37;
// Add spatial index
static const HA_ALTER_FLAGS ADD_SPATIAL_INDEX = 1ULL << 38;
// Alter index comment
static const HA_ALTER_FLAGS ALTER_INDEX_COMMENT = 1ULL << 39;
// New/changed virtual generated column require validation
static const HA_ALTER_FLAGS VALIDATE_VIRTUAL_COLUMN = 1ULL << 40;
/**
Change index option in a way which is likely not to require index
recreation. For example, change COMMENT or KEY::is_algorithm_explicit
flag (without change of index algorithm itself).
*/
static const HA_ALTER_FLAGS CHANGE_INDEX_OPTION = 1LL << 41;
// Rebuild partition
static const HA_ALTER_FLAGS ALTER_REBUILD_PARTITION = 1ULL << 42;
/**
Change in index length such that it does not require index rebuild.
For example, change in index length due to column expansion like
varchar(X) changed to varchar(X + N).
*/
static const HA_ALTER_FLAGS ALTER_COLUMN_INDEX_LENGTH = 1ULL << 43;
/**
Change to one of columns on which virtual generated column depends,
so its values require re-evaluation.
*/
static const HA_ALTER_FLAGS VIRTUAL_GCOL_REEVAL = 1ULL << 44;
/**
Change to one of columns on which stored generated column depends,
so its values require re-evaluation.
*/
static const HA_ALTER_FLAGS STORED_GCOL_REEVAL = 1ULL << 45;
// Add check constraint.
static const HA_ALTER_FLAGS ADD_CHECK_CONSTRAINT = 1ULL << 46;
// Drop check constraint.
static const HA_ALTER_FLAGS DROP_CHECK_CONSTRAINT = 1ULL << 47;
// Suspend check constraint.
static const HA_ALTER_FLAGS SUSPEND_CHECK_CONSTRAINT = 1ULL << 48;
// Alter column visibility.
static const HA_ALTER_FLAGS ALTER_COLUMN_VISIBILITY = 1ULL << 49;
/**
Create options (like MAX_ROWS) for the new version of table.
@note The referenced instance of HA_CREATE_INFO object was already
used to create new .FRM file for table being altered. So it
has been processed by mysql_prepare_create_table() already.
For example, this means that it has HA_OPTION_PACK_RECORD
flag in HA_CREATE_INFO::table_options member correctly set.
*/
HA_CREATE_INFO *create_info;
/**
Alter options, fields and keys for the new version of table.
@note The referenced instance of Alter_info object was already
used to create new .FRM file for table being altered. So it
has been processed by mysql_prepare_create_table() already.
In particular, this means that in Create_field objects for
fields which were present in some form in the old version
of table, Create_field::field member points to corresponding
Field instance for old version of table.
*/
Alter_info *alter_info;
/**
Indicates whether operation should fail if table is non-empty.
Storage engines should not suggest/allow execution of such operations
using INSTANT algorithm since check whether table is empty done from
SQL-layer is not "instant". Also SEs might choose different algorithm for
ALTER TABLE execution knowing that it will be allowed to proceed only if
table is empty.
Unlike for Alter_table_ctx::error_if_not_empty, we use bool for this flag
and not bitmap, since SEs are really interested in the fact that ALTER
will fail if table is not empty and not in exact reason behind this fact,
and because we want to avoid extra dependency between Alter_table_ctx and
Alter_inplace_info.
*/
bool error_if_not_empty;
/**
Array of KEYs for new version of table - including KEYs to be added.
@note Currently this array is produced as result of
mysql_prepare_create_table() call.
This means that it follows different convention for
KEY_PART_INFO::fieldnr values than objects in TABLE::key_info
array.
@todo This is mainly due to the fact that we need to keep compatibility
with removed handler::add_index() call. We plan to switch to
TABLE::key_info numbering later.
KEYs are sorted - see sort_keys().
*/
KEY *key_info_buffer;
/** Size of key_info_buffer array. */
uint key_count;
/** Size of index_drop_buffer array. */
uint index_drop_count;
/**
Array of pointers to KEYs to be dropped belonging to the TABLE instance
for the old version of the table.
*/
KEY **index_drop_buffer;
/** Size of index_add_buffer array. */
uint index_add_count;
/**
Array of indexes into key_info_buffer for KEYs to be added,
sorted in increasing order.
*/
uint *index_add_buffer;
/** Size of index_rename_buffer array. */
uint index_rename_count;
/** Size of index_rename_buffer array. */
uint index_altered_visibility_count;
/**
Array of KEY_PAIR objects describing indexes being renamed.
For each index renamed it contains object with KEY_PAIR::old_key
pointing to KEY object belonging to the TABLE instance for old
version of table representing old version of index and with
KEY_PAIR::new_key pointing to KEY object for new version of
index in key_info_buffer member.
*/
KEY_PAIR *index_rename_buffer;
KEY_PAIR *index_altered_visibility_buffer;
/** Number of virtual columns to be added. */
uint virtual_column_add_count;
/** number of virtual columns to be dropped. */
uint virtual_column_drop_count;
/**
Context information to allow handlers to keep context between in-place
alter API calls.
@see inplace_alter_handler_ctx for information about object lifecycle.
*/
inplace_alter_handler_ctx *handler_ctx;
/**
If the table uses several handlers, like ha_partition uses one handler
per partition, this contains a Null terminated array of ctx pointers
that should all be committed together.
Or NULL if only handler_ctx should be committed.
Set to NULL if the low level handler::commit_inplace_alter_table uses it,
to signal to the main handler that everything was committed as atomically.
@see inplace_alter_handler_ctx for information about object lifecycle.
*/
inplace_alter_handler_ctx **group_commit_ctx;
/**
Flags describing in detail which operations the storage engine is to
execute.
*/
HA_ALTER_FLAGS handler_flags;
/**
Partition_info taking into account the partition changes to be performed.
Contains all partitions which are present in the old version of the table
with partitions to be dropped or changed marked as such + all partitions
to be added in the new version of table marked as such.
*/
partition_info *modified_part_info;
/** true for online operation (LOCK=NONE) */
bool online;
/**
Can be set by handler along with handler_ctx. The difference is that
this flag can be used to store SE-specific in-place ALTER context in cases
when constructing full-blown inplace_alter_handler_ctx descendant is
inconvenient.
*/
uint handler_trivial_ctx;
/**
Can be set by handler to describe why a given operation cannot be done
in-place (HA_ALTER_INPLACE_NOT_SUPPORTED) or why it cannot be done
online (HA_ALTER_INPLACE_NO_LOCK or HA_ALTER_INPLACE_NO_LOCK_AFTER_PREPARE)
If set, it will be used with ER_ALTER_OPERATION_NOT_SUPPORTED_REASON if
results from handler::check_if_supported_inplace_alter() doesn't match
requirements set by user. If not set, the more generic
ER_ALTER_OPERATION_NOT_SUPPORTED will be used.
Please set to a properly localized string, for example using
my_get_err_msg(), so that the error message as a whole is localized.
*/
const char *unsupported_reason;
Alter_inplace_info(HA_CREATE_INFO *create_info_arg,
Alter_info *alter_info_arg, bool error_if_not_empty_arg,
KEY *key_info_arg, uint key_count_arg,
partition_info *modified_part_info_arg)
: create_info(create_info_arg),
alter_info(alter_info_arg),
error_if_not_empty(error_if_not_empty_arg),
key_info_buffer(key_info_arg),
key_count(key_count_arg),
index_drop_count(0),
index_drop_buffer(nullptr),
index_add_count(0),
index_add_buffer(nullptr),
index_rename_count(0),
index_altered_visibility_count(0),
index_rename_buffer(nullptr),
virtual_column_add_count(0),
virtual_column_drop_count(0),
handler_ctx(nullptr),
group_commit_ctx(nullptr),
handler_flags(0),
modified_part_info(modified_part_info_arg),
online(false),
handler_trivial_ctx(0),
unsupported_reason(nullptr) {}
~Alter_inplace_info() { destroy(handler_ctx); }
/**
Used after check_if_supported_inplace_alter() to report
error if the result does not match the LOCK/ALGORITHM
requirements set by the user.
@param not_supported Part of statement that was not supported.
@param try_instead Suggestion as to what the user should
replace not_supported with.
*/
void report_unsupported_error(const char *not_supported,
const char *try_instead);
/** Add old and new version of key to array of indexes to be renamed. */
void add_renamed_key(KEY *old_key, KEY *new_key) {
KEY_PAIR *key_pair = index_rename_buffer + index_rename_count++;
key_pair->old_key = old_key;
key_pair->new_key = new_key;
DBUG_PRINT("info",
("index renamed: '%s' to '%s'", old_key->name, new_key->name));
}
void add_altered_index_visibility(KEY *old_key, KEY *new_key) {
KEY_PAIR *key_pair =
index_altered_visibility_buffer + index_altered_visibility_count++;
key_pair->old_key = old_key;
key_pair->new_key = new_key;
DBUG_PRINT("info", ("index had visibility altered: %i to %i",
old_key->is_visible, new_key->is_visible));
}
/**
Add old and new version of modified key to arrays of indexes to
be dropped and added (correspondingly).
*/
void add_modified_key(KEY *old_key, KEY *new_key) {
index_drop_buffer[index_drop_count++] = old_key;
index_add_buffer[index_add_count++] = (uint)(new_key - key_info_buffer);
DBUG_PRINT("info", ("index changed: '%s'", old_key->name));
}
/** Drop key to array of indexes to be dropped. */
void add_dropped_key(KEY *old_key) {
index_drop_buffer[index_drop_count++] = old_key;
DBUG_PRINT("info", ("index dropped: '%s'", old_key->name));
}
/** Add key to array of indexes to be added. */
void add_added_key(KEY *new_key) {
index_add_buffer[index_add_count++] = (uint)(new_key - key_info_buffer);
DBUG_PRINT("info", ("index added: '%s'", new_key->name));
}
};
struct HA_CHECK_OPT {
uint flags{0}; /* isam layer flags (e.g. for myisamchk) */
uint sql_flags{0}; /* sql layer flags - for something myisamchk cannot do */
KEY_CACHE *key_cache; /* new key cache when changing key cache */
};
/*
This is a buffer area that the handler can use to store rows.
'end_of_used_area' should be kept updated after calls to
read-functions so that other parts of the code can use the
remaining area (until next read calls is issued).
*/
struct HANDLER_BUFFER {
uchar *buffer; /* Buffer one can start using */
uchar *buffer_end; /* End of buffer */
uchar *end_of_used_area; /* End of area that was used by handler */
};
typedef void *range_seq_t;
struct RANGE_SEQ_IF {
/*
Initialize the traversal of range sequence
SYNOPSIS
init()
init_params The seq_init_param parameter
n_ranges The number of ranges obtained
flags A combination of HA_MRR_SINGLE_POINT, HA_MRR_FIXED_KEY
RETURN
An opaque value to be used as RANGE_SEQ_IF::next() parameter
*/
range_seq_t (*init)(void *init_params, uint n_ranges, uint flags);
/*
Get the next range in the range sequence
SYNOPSIS
next()
seq The value returned by RANGE_SEQ_IF::init()
range OUT Information about the next range
RETURN
0 - Ok, the range structure filled with info about the next range
1 - No more ranges
*/
uint (*next)(range_seq_t seq, KEY_MULTI_RANGE *range);
/*
Check whether range_info orders to skip the next record
SYNOPSIS
skip_record()
seq The value returned by RANGE_SEQ_IF::init()
range_info Information about the next range
(Ignored if MRR_NO_ASSOCIATION is set)
rowid Rowid of the record to be checked (ignored if set to 0)
RETURN
1 - Record with this range_info and/or this rowid shall be filtered
out from the stream of records returned by ha_multi_range_read_next()
0 - The record shall be left in the stream
*/
bool (*skip_record)(range_seq_t seq, char *range_info, uchar *rowid);
};
/**
Used to store optimizer cost estimates.
The class consists of PODs only: default operator=, copy constructor
and destructor are used.
*/
class Cost_estimate {
private:
double io_cost; ///< cost of I/O operations
double cpu_cost; ///< cost of CPU operations
double import_cost; ///< cost of remote operations
double mem_cost; ///< memory used (bytes)
public:
Cost_estimate() : io_cost(0), cpu_cost(0), import_cost(0), mem_cost(0) {}
/// Returns sum of time-consuming costs, i.e., not counting memory cost
double total_cost() const { return io_cost + cpu_cost + import_cost; }
double get_io_cost() const { return io_cost; }
double get_cpu_cost() const { return cpu_cost; }
double get_import_cost() const { return import_cost; }
double get_mem_cost() const { return mem_cost; }
/**
Whether or not all costs in the object are zero
@return true if all costs are zero, false otherwise
*/
bool is_zero() const {
return !(io_cost || cpu_cost || import_cost || mem_cost);
}
/**
Whether or not the total cost is the maximal double
@return true if total cost is the maximal double, false otherwise
*/
bool is_max_cost() const { return io_cost == DBL_MAX; }
/// Reset all costs to zero
void reset() { io_cost = cpu_cost = import_cost = mem_cost = 0; }
/// Set current cost to the maximal double
void set_max_cost() {
reset();
io_cost = DBL_MAX;
}
/// Multiply io, cpu and import costs by parameter
void multiply(double m) {
assert(!is_max_cost());
io_cost *= m;
cpu_cost *= m;
import_cost *= m;
/* Don't multiply mem_cost */
}
Cost_estimate &operator+=(const Cost_estimate &other) {
assert(!is_max_cost() && !other.is_max_cost());
io_cost += other.io_cost;
cpu_cost += other.cpu_cost;
import_cost += other.import_cost;
mem_cost += other.mem_cost;
return *this;
}
Cost_estimate operator+(const Cost_estimate &other) {
Cost_estimate result = *this;
result += other;
return result;
}
Cost_estimate operator-(const Cost_estimate &other) {
Cost_estimate result;
assert(!other.is_max_cost());
result.io_cost = io_cost - other.io_cost;
result.cpu_cost = cpu_cost - other.cpu_cost;
result.import_cost = import_cost - other.import_cost;
result.mem_cost = mem_cost - other.mem_cost;
return result;
}
bool operator>(const Cost_estimate &other) const {
return total_cost() > other.total_cost() ? true : false;
}
bool operator<(const Cost_estimate &other) const {
return other > *this ? true : false;
}
/// Add to IO cost
void add_io(double add_io_cost) {
assert(!is_max_cost());
io_cost += add_io_cost;
}
/// Add to CPU cost
void add_cpu(double add_cpu_cost) {
assert(!is_max_cost());
cpu_cost += add_cpu_cost;
}
/// Add to import cost
void add_import(double add_import_cost) {
assert(!is_max_cost());
import_cost += add_import_cost;
}
/// Add to memory cost
void add_mem(double add_mem_cost) {
assert(!is_max_cost());
mem_cost += add_mem_cost;
}
};
void get_sweep_read_cost(TABLE *table, ha_rows nrows, bool interrupted,
Cost_estimate *cost);
/*
The below two are not used (and not handled) in this milestone of this WL
entry because there seems to be no use for them at this stage of
implementation.
*/
#define HA_MRR_SINGLE_POINT 1
#define HA_MRR_FIXED_KEY 2
/*
Indicates that RANGE_SEQ_IF::next(&range) doesn't need to fill in the
'range' parameter.
*/
#define HA_MRR_NO_ASSOCIATION 4
/*
The MRR user will provide ranges in key order, and MRR implementation
must return rows in key order.
Passing this flag to multi_read_range_init() may cause the
default MRR handler to be used even if HA_MRR_USE_DEFAULT_IMPL
was not specified.
(If the native MRR impl. can not provide SORTED result)
*/
#define HA_MRR_SORTED 8
/* MRR implementation doesn't have to retrieve full records */
#define HA_MRR_INDEX_ONLY 16
/*
The passed memory buffer is of maximum possible size, the caller can't
assume larger buffer.
*/
#define HA_MRR_LIMITS 32
/*
Flag set <=> default MRR implementation is used
(The choice is made by **_info[_const]() function which may set this
flag. SQL layer remembers the flag value and then passes it to
multi_read_range_init().
*/
#define HA_MRR_USE_DEFAULT_IMPL 64
/*
Used only as parameter to multi_range_read_info():
Flag set <=> the caller guarantees that the bounds of the scanned ranges
will not have NULL values.
*/
#define HA_MRR_NO_NULL_ENDPOINTS 128
/*
Set by the MRR implementation to signal that it will natively
produced sorted result if multi_range_read_init() is called with
the HA_MRR_SORTED flag - Else multi_range_read_init(HA_MRR_SORTED)
will revert to use the default MRR implementation.
*/
#define HA_MRR_SUPPORT_SORTED 256
class ha_statistics {
public:
ulonglong data_file_length; /* Length off data file */
ulonglong max_data_file_length; /* Length off data file */
ulonglong index_file_length;
ulonglong max_index_file_length;
ulonglong delete_length; /* Free bytes */
ulonglong auto_increment_value;
/*
The number of records in the table.
0 - means the table has exactly 0 rows
other - if (table_flags() & HA_STATS_RECORDS_IS_EXACT)
the value is the exact number of records in the table
else
it is an estimate
*/
ha_rows records;
ha_rows deleted; /* Deleted records */
ulong mean_rec_length; /* physical reclength */
/* TODO: create_time should be retrieved from the new DD. Remove this. */
time_t create_time; /* When table was created */
ulong check_time;
ulong update_time;
uint block_size; /* index block size */
/*
number of buffer bytes that native mrr implementation needs,
*/
uint mrr_length_per_rec;
/**
Estimate for how much of the table that is available in a memory
buffer. Valid range is [0..1]. If it has the special value
IN_MEMORY_ESTIMATE_UNKNOWN (defined in structs.h), it means that
the storage engine has not supplied any value for it.
*/
double table_in_mem_estimate;
ha_statistics()
: data_file_length(0),
max_data_file_length(0),
index_file_length(0),
delete_length(0),
auto_increment_value(0),
records(0),
deleted(0),
mean_rec_length(0),
create_time(0),
check_time(0),
update_time(0),
block_size(0),
table_in_mem_estimate(IN_MEMORY_ESTIMATE_UNKNOWN) {}
};
/**
Calculates length of key.
Given a key index and a map of key parts return length of buffer used by key
parts.
@param table Table containing the key
@param key Key index
@param keypart_map which key parts that is used
@return Length of used key parts.
*/
uint calculate_key_len(TABLE *table, uint key, key_part_map keypart_map);
/*
bitmap with first N+1 bits set
(keypart_map for a key prefix of [0..N] keyparts)
*/
#define make_keypart_map(N) (((key_part_map)2 << (N)) - 1)
/*
bitmap with first N bits set
(keypart_map for a key prefix of [0..N-1] keyparts)
*/
#define make_prev_keypart_map(N) (((key_part_map)1 << (N)) - 1)
/** Base class to be used by handlers different shares */
class Handler_share {
public:
Handler_share() = default;
virtual ~Handler_share() = default;
};
/**
Wrapper for struct ft_hints.
*/
class Ft_hints {
private:
struct ft_hints hints;
public:
explicit Ft_hints(uint ft_flags) {
hints.flags = ft_flags;
hints.op_type = FT_OP_UNDEFINED;
hints.op_value = 0.0;
hints.limit = HA_POS_ERROR;
}
/**
Set comparison operation type and and value for master MATCH function.
@param type comparison operation type
@param value comparison operation value
*/
void set_hint_op(enum ft_operation type, double value) {
hints.op_type = type;
hints.op_value = value;
}
/**
Set Ft_hints flag.
@param ft_flag Ft_hints flag
*/
void set_hint_flag(uint ft_flag) { hints.flags |= ft_flag; }
/**
Set Ft_hints limit.
@param ft_limit limit
*/
void set_hint_limit(ha_rows ft_limit) { hints.limit = ft_limit; }
/**
Get Ft_hints limit.
@return Ft_hints limit
*/
ha_rows get_limit() const { return hints.limit; }
/**
Get Ft_hints operation value.
@return operation value
*/
double get_op_value() const { return hints.op_value; }
/**
Get Ft_hints operation type.
@return operation type
*/
enum ft_operation get_op_type() const { return hints.op_type; }
/**
Get Ft_hints flags.
@return Ft_hints flags
*/
uint get_flags() const { return hints.flags; }
/**
Get ft_hints struct.
@return pointer to ft_hints struct
*/
struct ft_hints *get_hints() { return &hints; }
};
/**
The handler class is the interface for dynamically loadable
storage engines. Do not add ifdefs and take care when adding or
changing virtual functions to avoid vtable confusion
Functions in this class accept and return table columns data. Two data
representation formats are used:
1. TableRecordFormat - Used to pass [partial] table records to/from
storage engine
2. KeyTupleFormat - used to pass index search tuples (aka "keys") to
storage engine. See opt_range.cc for description of this format.
TableRecordFormat
=================
[Warning: this description is work in progress and may be incomplete]
The table record is stored in a fixed-size buffer:
record: null_bytes, column1_data, column2_data, ...
The offsets of the parts of the buffer are also fixed: every column has
an offset to its column{i}_data, and if it is nullable it also has its own
bit in null_bytes.
The record buffer only includes data about columns that are marked in the
relevant column set (table->read_set and/or table->write_set, depending on
the situation).
<not-sure>It could be that it is required that null bits of non-present
columns are set to 1</not-sure>
VARIOUS EXCEPTIONS AND SPECIAL CASES
If the table has no nullable columns, then null_bytes is still
present, its length is one byte <not-sure> which must be set to 0xFF
at all times. </not-sure>
If the table has columns of type BIT, then certain bits from those columns
may be stored in null_bytes as well. Grep around for Field_bit for
details.
For blob columns (see Field_blob), the record buffer stores length of the
data, following by memory pointer to the blob data. The pointer is owned
by the storage engine and is valid until the next operation.
If a blob column has NULL value, then its length and blob data pointer
must be set to 0.
Overview of main modules of the handler API
===========================================
The overview below was copied from the storage/partition/ha_partition.h when
support for non-native partitioning was removed.
-------------------------------------------------------------------------
MODULE create/delete handler object
-------------------------------------------------------------------------
Object create/delete method. Normally called when a table object
exists.
-------------------------------------------------------------------------
MODULE meta data changes
-------------------------------------------------------------------------
Meta data routines to CREATE, DROP, RENAME table are often used at
ALTER TABLE (update_create_info used from ALTER TABLE and SHOW ..).
Methods:
delete_table()
rename_table()
create()
update_create_info()
-------------------------------------------------------------------------
MODULE open/close object
-------------------------------------------------------------------------
Open and close handler object to ensure all underlying files and
objects allocated and deallocated for query handling is handled
properly.
A handler object is opened as part of its initialisation and before
being used for normal queries (not before meta-data changes always.
If the object was opened it will also be closed before being deleted.
Methods:
open()
close()
-------------------------------------------------------------------------
MODULE start/end statement
-------------------------------------------------------------------------
This module contains methods that are used to understand start/end of
statements, transaction boundaries, and aid for proper concurrency
control.
Methods:
store_lock()
external_lock()
start_stmt()
lock_count()
unlock_row()
was_semi_consistent_read()
try_semi_consistent_read()
-------------------------------------------------------------------------
MODULE change record
-------------------------------------------------------------------------
This part of the handler interface is used to change the records
after INSERT, DELETE, UPDATE, REPLACE method calls but also other
special meta-data operations as ALTER TABLE, LOAD DATA, TRUNCATE.
These methods are used for insert (write_row), update (update_row)
and delete (delete_row). All methods to change data always work on
one row at a time. update_row and delete_row also contains the old
row.
delete_all_rows will delete all rows in the table in one call as a
special optimization for DELETE from table;
Bulk inserts are supported if all underlying handlers support it.
start_bulk_insert and end_bulk_insert is called before and after a
number of calls to write_row.
Methods:
write_row()
update_row()
delete_row()
delete_all_rows()
start_bulk_insert()
end_bulk_insert()
-------------------------------------------------------------------------
MODULE full table scan
-------------------------------------------------------------------------
This module is used for the most basic access method for any table
handler. This is to fetch all data through a full table scan. No
indexes are needed to implement this part.
It contains one method to start the scan (rnd_init) that can also be
called multiple times (typical in a nested loop join). Then proceeding
to the next record (rnd_next) and closing the scan (rnd_end).
To remember a record for later access there is a method (position)
and there is a method used to retrieve the record based on the stored
position.
The position can be a file position, a primary key, a ROWID dependent
on the handler below.
All functions that retrieve records and are callable through the
handler interface must indicate whether a record is present after the call
or not. Record found is indicated by returning 0 and setting table status
to "has row". Record not found is indicated by returning a non-zero value
and setting table status to "no row".
@see TABLE::set_found_row() and TABLE::set_no_row().
By enforcing these rules in the handler interface, storage handler functions
need not set any status in struct TABLE. These notes also apply to module
index scan, documented below.
Methods:
rnd_init()
rnd_end()
rnd_next()
rnd_pos()
rnd_pos_by_record()
position()
-------------------------------------------------------------------------
MODULE index scan
-------------------------------------------------------------------------
This part of the handler interface is used to perform access through
indexes. The interface is defined as a scan interface but the handler
can also use key lookup if the index is a unique index or a primary
key index.
Index scans are mostly useful for SELECT queries but are an important
part also of UPDATE, DELETE, REPLACE and CREATE TABLE table AS SELECT
and so forth.
Naturally an index is needed for an index scan and indexes can either
be ordered, hash based. Some ordered indexes can return data in order
but not necessarily all of them.
There are many flags that define the behavior of indexes in the
various handlers. These methods are found in the optimizer module.
index_read is called to start a scan of an index. The find_flag defines
the semantics of the scan. These flags are defined in
include/my_base.h
index_read_idx is the same but also initializes index before calling doing
the same thing as index_read. Thus it is similar to index_init followed
by index_read. This is also how we implement it.
index_read/index_read_idx does also return the first row. Thus for
key lookups, the index_read will be the only call to the handler in
the index scan.
index_init initializes an index before using it and index_end does
any end processing needed.
Methods:
index_read_map()
index_init()
index_end()
index_read_idx_map()
index_next()
index_prev()
index_first()
index_last()
index_next_same()
index_read_last_map()
read_range_first()
read_range_next()
-------------------------------------------------------------------------
MODULE information calls
-------------------------------------------------------------------------
This calls are used to inform the handler of specifics of the ongoing
scans and other actions. Most of these are used for optimisation
purposes.
Methods:
info()
get_dynamic_partition_info
extra()
extra_opt()
reset()
-------------------------------------------------------------------------
MODULE optimizer support
-------------------------------------------------------------------------
NOTE:
One important part of the public handler interface that is not depicted in
the methods is the attribute records which is defined in the base class.
This is looked upon directly and is set by calling info(HA_STATUS_INFO) ?
Methods:
min_rows_for_estimate()
get_biggest_used_partition()
scan_time()
read_time()
records_in_range()
estimate_rows_upper_bound()
records()
-------------------------------------------------------------------------
MODULE print messages
-------------------------------------------------------------------------
This module contains various methods that returns text messages for
table types, index type and error messages.
Methods:
table_type()
get_row_type()
print_error()
get_error_message()
-------------------------------------------------------------------------
MODULE handler characteristics
-------------------------------------------------------------------------
This module contains a number of methods defining limitations and
characteristics of the handler (see also documentation regarding the
individual flags).
Methods:
table_flags()
index_flags()
min_of_the_max_uint()
max_supported_record_length()
max_supported_keys()
max_supported_key_parts()
max_supported_key_length()
max_supported_key_part_length()
low_byte_first()
extra_rec_buf_length()
min_record_length(uint options)
primary_key_is_clustered()
ha_key_alg get_default_index_algorithm()
is_index_algorithm_supported()
-------------------------------------------------------------------------
MODULE compare records
-------------------------------------------------------------------------
cmp_ref checks if two references are the same. For most handlers this is
a simple memcmp of the reference. However some handlers use primary key
as reference and this can be the same even if memcmp says they are
different. This is due to character sets and end spaces and so forth.
Methods:
cmp_ref()
-------------------------------------------------------------------------
MODULE auto increment
-------------------------------------------------------------------------
This module is used to handle the support of auto increments.
This variable in the handler is used as part of the handler interface
It is maintained by the parent handler object and should not be
touched by child handler objects (see handler.cc for its use).
Methods:
get_auto_increment()
release_auto_increment()
-------------------------------------------------------------------------
MODULE initialize handler for HANDLER call
-------------------------------------------------------------------------
This method is a special InnoDB method called before a HANDLER query.
Methods:
init_table_handle_for_HANDLER()
-------------------------------------------------------------------------
MODULE fulltext index
-------------------------------------------------------------------------
Fulltext index support.
Methods:
ft_init_ext_with_hints()
ft_init()
ft_init_ext()
ft_read()
-------------------------------------------------------------------------
MODULE in-place ALTER TABLE
-------------------------------------------------------------------------
Methods for in-place ALTER TABLE support (implemented by InnoDB and NDB).
Methods:
check_if_supported_inplace_alter()
prepare_inplace_alter_table()
inplace_alter_table()
commit_inplace_alter_table()
notify_table_changed()
-------------------------------------------------------------------------
MODULE tablespace support
-------------------------------------------------------------------------
Methods:
discard_or_import_tablespace()
-------------------------------------------------------------------------
MODULE administrative DDL
-------------------------------------------------------------------------
Methods:
optimize()
analyze()
check()
repair()
check_and_repair()
auto_repair()
is_crashed()
check_for_upgrade()
checksum()
assign_to_keycache()
-------------------------------------------------------------------------
MODULE enable/disable indexes
-------------------------------------------------------------------------
Enable/Disable Indexes are only supported by HEAP and MyISAM.
Methods:
disable_indexes()
enable_indexes()
indexes_are_disabled()
-------------------------------------------------------------------------
MODULE append_create_info
-------------------------------------------------------------------------
Only used by MyISAM MERGE tables.
Methods:
append_create_info()
-------------------------------------------------------------------------
MODULE partitioning specific handler API
-------------------------------------------------------------------------
Methods:
get_partition_handler()
*/
class handler {
friend class Partition_handler;
public:
typedef ulonglong Table_flags;
protected:
TABLE_SHARE *table_share; /* The table definition */
TABLE *table; /* The current open table */
Table_flags cached_table_flags{0}; /* Set on init() and open() */
ha_rows estimation_rows_to_insert;
public:
handlerton *ht; /* storage engine of this handler */
/** Pointer to current row */
uchar *ref;
/** Pointer to duplicate row */
uchar *dup_ref;
ha_statistics stats;
/* MultiRangeRead-related members: */
range_seq_t mrr_iter; /* Iterator to traverse the range sequence */
RANGE_SEQ_IF mrr_funcs; /* Range sequence traversal functions */
HANDLER_BUFFER *multi_range_buffer; /* MRR buffer info */
uint ranges_in_seq; /* Total number of ranges in the traversed sequence */
/* true <=> source MRR ranges and the output are ordered */
bool mrr_is_output_sorted;
/* true <=> we're currently traversing a range in mrr_cur_range. */
bool mrr_have_range;
/* Current range (the one we're now returning rows from) */
KEY_MULTI_RANGE mrr_cur_range;
/*
The direction of the current range or index scan. This is used by
the ICP implementation to determine if it has reached the end
of the current range.
*/
enum enum_range_scan_direction { RANGE_SCAN_ASC, RANGE_SCAN_DESC };
private:
Record_buffer *m_record_buffer = nullptr; ///< Buffer for multi-row reads.
/*
Storage space for the end range value. Should only be accessed using
the end_range pointer. The content is invalid when end_range is NULL.
*/
key_range save_end_range;
enum_range_scan_direction range_scan_direction;
int key_compare_result_on_equal;
/**
Pointer to the handler of the table in the primary storage engine,
if this handler represents a table in a secondary storage engine.
*/
handler *m_primary_handler{nullptr};
protected:
KEY_PART_INFO *range_key_part;
bool eq_range;
/*
true <=> the engine guarantees that returned records are within the range
being scanned.
*/
bool in_range_check_pushed_down;
public:
/**
End value for a range scan. If this is NULL the range scan has no
end value. Should also be NULL when there is no ongoing range scan.
Used by the read_range() functions and also evaluated by pushed
index conditions.
*/
key_range *end_range;
/**
Flag which tells if #end_range contains a virtual generated column.
The content is invalid when #end_range is @c nullptr.
*/
bool m_virt_gcol_in_end_range = false;
uint errkey; /* Last dup key */
uint key_used_on_scan;
uint active_index;
/** Length of ref (1-8 or the clustered key length) */
uint ref_length;
FT_INFO *ft_handler;
enum { NONE = 0, INDEX, RND, SAMPLING } inited;
bool implicit_emptied; /* Can be !=0 only if HEAP */
const Item *pushed_cond;
Item *pushed_idx_cond;
uint pushed_idx_cond_keyno; /* The index which the above condition is for */
/**
next_insert_id is the next value which should be inserted into the
auto_increment column: in a inserting-multi-row statement (like INSERT
SELECT), for the first row where the autoinc value is not specified by the
statement, get_auto_increment() called and asked to generate a value,
next_insert_id is set to the next value, then for all other rows
next_insert_id is used (and increased each time) without calling
get_auto_increment().
*/
ulonglong next_insert_id;
/**
insert id for the current row (*autogenerated*; if not
autogenerated, it's 0).
At first successful insertion, this variable is stored into
THD::first_successful_insert_id_in_cur_stmt.
*/
ulonglong insert_id_for_cur_row;
/**
Interval returned by get_auto_increment() and being consumed by the
inserter.
*/
Discrete_interval auto_inc_interval_for_cur_row;
/**
Number of reserved auto-increment intervals. Serves as a heuristic
when we have no estimation of how many records the statement will insert:
the more intervals we have reserved, the bigger the next one. Reset in
handler::ha_release_auto_increment().
*/
uint auto_inc_intervals_count;
/**
Instrumented table associated with this handler.
*/
PSI_table *m_psi;
std::mt19937 m_random_number_engine;
double m_sampling_percentage;
private:
/** Internal state of the batch instrumentation. */
enum batch_mode_t {
/** Batch mode not used. */
PSI_BATCH_MODE_NONE,
/** Batch mode used, before first table io. */
PSI_BATCH_MODE_STARTING,
/** Batch mode used, after first table io. */
PSI_BATCH_MODE_STARTED
};
/**
Batch mode state.
@sa start_psi_batch_mode.
@sa end_psi_batch_mode.
*/
batch_mode_t m_psi_batch_mode;
/**
The number of rows in the batch.
@sa start_psi_batch_mode.
@sa end_psi_batch_mode.
*/
ulonglong m_psi_numrows;
/**
The current event in a batch.
@sa start_psi_batch_mode.
@sa end_psi_batch_mode.
*/
PSI_table_locker *m_psi_locker;
/**
Storage for the event in a batch.
@sa start_psi_batch_mode.
@sa end_psi_batch_mode.
*/
PSI_table_locker_state m_psi_locker_state;
public:
void unbind_psi();
void rebind_psi();
/**
Put the handler in 'batch' mode when collecting
table io instrumented events.
When operating in batch mode:
- a single start event is generated in the performance schema.
- all table io performed between @c start_psi_batch_mode
and @c end_psi_batch_mode is not instrumented:
the number of rows affected is counted instead in @c m_psi_numrows.
- a single end event is generated in the performance schema
when the batch mode ends with @c end_psi_batch_mode.
*/
void start_psi_batch_mode();
/** End a batch started with @c start_psi_batch_mode. */
void end_psi_batch_mode();
/**
If a PSI batch was started, turn if off.
@returns true if it was started.
*/
bool end_psi_batch_mode_if_started() {
bool rc = m_psi_batch_mode;
if (rc) end_psi_batch_mode();
return rc;
}
private:
/**
The lock type set by when calling::ha_external_lock(). This is
propagated down to the storage engine. The reason for also storing
it here, is that when doing MRR we need to create/clone a second handler
object. This cloned handler object needs to know about the lock_type used.
*/
int m_lock_type;
/**
Pointer where to store/retrieve the Handler_share pointer.
For non partitioned handlers this is &TABLE_SHARE::ha_share.
*/
Handler_share **ha_share;
/**
Some non-virtual ha_* functions, responsible for reading rows,
like ha_rnd_pos(), must ensure that virtual generated columns are
calculated before they return. For that, they should set this
member to true at their start, and check it before they return: if
the member is still true, it means they should calculate; if it's
false, it means the calculation has been done by some called
lower-level function and does not need to be re-done (which is why
we need this status flag: to avoid redundant calculations, for
performance).
Note that when updating generated fields, the NULL row status in
the underlying TABLE objects matter, so be sure to reset them if needed!
*/
bool m_update_generated_read_fields;
/* Filter row ids to weed out duplicates when multi-valued index is used */
Unique_on_insert *m_unique;
public:
handler(handlerton *ht_arg, TABLE_SHARE *share_arg)
: table_share(share_arg),
table(nullptr),
estimation_rows_to_insert(0),
ht(ht_arg),
ref(nullptr),
range_scan_direction(RANGE_SCAN_ASC),
in_range_check_pushed_down(false),
end_range(nullptr),
key_used_on_scan(MAX_KEY),
active_index(MAX_KEY),
ref_length(sizeof(my_off_t)),
ft_handler(nullptr),
inited(NONE),
implicit_emptied(false),
pushed_cond(nullptr),
pushed_idx_cond(nullptr),
pushed_idx_cond_keyno(MAX_KEY),
next_insert_id(0),
insert_id_for_cur_row(0),
auto_inc_intervals_count(0),
m_psi(nullptr),
m_psi_batch_mode(PSI_BATCH_MODE_NONE),
m_psi_numrows(0),
m_psi_locker(nullptr),
m_lock_type(F_UNLCK),
ha_share(nullptr),
m_update_generated_read_fields(false),
m_unique(nullptr) {
DBUG_PRINT("info", ("handler created F_UNLCK %d F_RDLCK %d F_WRLCK %d",
F_UNLCK, F_RDLCK, F_WRLCK));
}
virtual ~handler(void) {
assert(m_psi == nullptr);
assert(m_psi_batch_mode == PSI_BATCH_MODE_NONE);
assert(m_psi_locker == nullptr);
assert(m_lock_type == F_UNLCK);
assert(inited == NONE);
}
/**
Return extra handler specific text for EXPLAIN.
*/
virtual std::string explain_extra() const { return ""; }
/*
@todo reorganize functions, make proper public/protected/private qualifiers
*/
virtual handler *clone(const char *name, MEM_ROOT *mem_root);
/** This is called after create to allow us to set up cached variables */
void init() { cached_table_flags = table_flags(); }
/* ha_ methods: public wrappers for private virtual API */
/**
Set a record buffer that the storage engine can use for multi-row reads.
The buffer has to be provided prior to the first read from an index or a
table.
@param buffer the buffer to use for multi-row reads
*/
void ha_set_record_buffer(Record_buffer *buffer) { m_record_buffer = buffer; }
/**
Get the record buffer that was set with ha_set_record_buffer().
@return the buffer to use for multi-row reads, or nullptr if there is none
*/
Record_buffer *ha_get_record_buffer() const { return m_record_buffer; }
/**
Does this handler want to get a Record_buffer for multi-row reads
via the ha_set_record_buffer() function? And if so, what is the
maximum number of records to allocate space for in the buffer?
Storage engines that support using a Record_buffer should override
handler::is_record_buffer_wanted().
@param[out] max_rows gets set to the maximum number of records to
allocate space for in the buffer if the function
returns true
@retval true if the handler would like a Record_buffer
@retval false if the handler does not want a Record_buffer
*/
bool ha_is_record_buffer_wanted(ha_rows *const max_rows) const {
return is_record_buffer_wanted(max_rows);
}
int ha_open(TABLE *table, const char *name, int mode, int test_if_locked,
const dd::Table *table_def);
int ha_close(void);
int ha_index_init(uint idx, bool sorted);
int ha_index_end();
int ha_rnd_init(bool scan);
int ha_rnd_end();
int ha_rnd_next(uchar *buf);
// See the comment on m_update_generated_read_fields.
int ha_rnd_pos(uchar *buf, uchar *pos);
int ha_index_read_map(uchar *buf, const uchar *key, key_part_map keypart_map,
enum ha_rkey_function find_flag);
int ha_index_read_last_map(uchar *buf, const uchar *key,
key_part_map keypart_map);
int ha_index_read_idx_map(uchar *buf, uint index, const uchar *key,
key_part_map keypart_map,
enum ha_rkey_function find_flag);
int ha_index_next(uchar *buf);
int ha_index_prev(uchar *buf);
int ha_index_first(uchar *buf);
int ha_index_last(uchar *buf);
int ha_index_next_same(uchar *buf, const uchar *key, uint keylen);
int ha_reset();
/* this is necessary in many places, e.g. in HANDLER command */
int ha_index_or_rnd_end() {
return inited == INDEX ? ha_index_end() : inited == RND ? ha_rnd_end() : 0;
}
/**
The cached_table_flags is set at ha_open and ha_external_lock
*/
Table_flags ha_table_flags() const { return cached_table_flags; }
/**
These functions represent the public interface to *users* of the
handler class, hence they are *not* virtual. For the inheritance
interface, see the (private) functions write_row(), update_row(),
and delete_row() below.
*/
int ha_external_lock(THD *thd, int lock_type);
int ha_write_row(uchar *buf);
/**
Update the current row.
@param old_data the old contents of the row
@param new_data the new contents of the row
@return error status (zero on success, HA_ERR_* error code on error)
*/
int ha_update_row(const uchar *old_data, uchar *new_data);
int ha_delete_row(const uchar *buf);
void ha_release_auto_increment();
int ha_check_for_upgrade(HA_CHECK_OPT *check_opt);
/** to be actually called to get 'check()' functionality*/
int ha_check(THD *thd, HA_CHECK_OPT *check_opt);
int ha_repair(THD *thd, HA_CHECK_OPT *check_opt);
void ha_start_bulk_insert(ha_rows rows);
int ha_end_bulk_insert();
int ha_bulk_update_row(const uchar *old_data, uchar *new_data,
uint *dup_key_found);
int ha_delete_all_rows();
int ha_truncate(dd::Table *table_def);
int ha_optimize(THD *thd, HA_CHECK_OPT *check_opt);
int ha_analyze(THD *thd, HA_CHECK_OPT *check_opt);
bool ha_check_and_repair(THD *thd);
int ha_disable_indexes(uint mode);
int ha_enable_indexes(uint mode);
int ha_discard_or_import_tablespace(bool discard, dd::Table *table_def);
int ha_rename_table(const char *from, const char *to,
const dd::Table *from_table_def, dd::Table *to_table_def);
int ha_delete_table(const char *name, const dd::Table *table_def);
void ha_drop_table(const char *name);
int ha_create(const char *name, TABLE *form, HA_CREATE_INFO *info,
dd::Table *table_def);
int ha_load_table(const TABLE &table);
int ha_unload_table(const char *db_name, const char *table_name,
bool error_if_not_loaded);
/**
Initializes a parallel scan. It creates a parallel_scan_ctx that has to
be used across all parallel_scan methods. Also, gets the number of
threads that would be spawned for parallel scan.
@param[out] scan_ctx The parallel scan context.
@param[out] num_threads Number of threads used for the scan.
@param[in] use_reserved_threads true if reserved threads are to be used
if we exhaust the max cap of number of
parallel read threads that can be
spawned at a time
@return error code
@retval 0 on success
*/
virtual int parallel_scan_init(void *&scan_ctx [[maybe_unused]],
size_t *num_threads [[maybe_unused]],
bool use_reserved_threads [[maybe_unused]]) {
return 0;
}
/**
This callback is called by each parallel load thread at the beginning of
the parallel load for the adapter scan.
@param cookie The cookie for this thread
@param ncols Number of columns in each row
@param row_len The size of a row in bytes
@param col_offsets An array of size ncols, where each element represents
the offset of a column in the row data. The memory of
this array belongs to the caller and will be free-ed
after the pload_end_cbk call.
@param null_byte_offsets An array of size ncols, where each element
represents the offset of a column in the row data. The
memory of this array belongs to the caller and will be
free-ed after the pload_end_cbk call.
@param null_bitmasks An array of size ncols, where each element
represents the bitmask required to get the null bit. The
memory of this array belongs to the caller and will be
free-ed after the pload_end_cbk call.
*/
using Load_init_cbk = std::function<bool(
void *cookie, ulong ncols, ulong row_len, const ulong *col_offsets,
const ulong *null_byte_offsets, const ulong *null_bitmasks)>;
/**
This callback is called by each parallel load thread when processing
of rows is required for the adapter scan.
@param[in] cookie The cookie for this thread
@param[in] nrows The nrows that are available
@param[in] rowdata The mysql-in-memory row data buffer. This is a
memory buffer for nrows records. The length of each record is fixed and
communicated via Load_init_cbk
@param[in] partition_id Partition id if it's a partitioned table, else
std::numeric_limits<uint64_t>::max()
@returns true if there is an error, false otherwise.
*/
using Load_cbk = std::function<bool(void *cookie, uint nrows, void *rowdata,
uint64_t partition_id)>;
/**
This callback is called by each parallel load thread when processing
of rows has ended for the adapter scan.
@param[in] cookie The cookie for this thread
*/
using Load_end_cbk = std::function<void(void *cookie)>;
/**
Run the parallel read of data.
@param[in] scan_ctx Scan context of the parallel read.
@param[in,out] thread_ctxs Caller thread contexts.
@param[in] init_fn Callback called by each parallel load
thread at the beginning of the parallel load.
@param[in] load_fn Callback called by each parallel load
thread when processing of rows is required.
@param[in] end_fn Callback called by each parallel load
thread when processing of rows has ended.
@return error code
@retval 0 on success
*/
virtual int parallel_scan(void *scan_ctx [[maybe_unused]],
void **thread_ctxs [[maybe_unused]],
Load_init_cbk init_fn [[maybe_unused]],
Load_cbk load_fn [[maybe_unused]],
Load_end_cbk end_fn [[maybe_unused]]) {
return 0;
}
/**
End of the parallel scan.
@param[in] scan_ctx A scan context created by parallel_scan_init.
*/
virtual void parallel_scan_end(void *scan_ctx [[maybe_unused]]) { return; }
/**
Submit a dd::Table object representing a core DD table having
hardcoded data to be filled in by the DDSE. This function can be
used for retrieving the hard coded SE private data for the
mysql.dd_properties table, before creating or opening it, or for
retrieving the hard coded SE private data for a core table,
before creating or opening them.
@param dd_table [in,out] A dd::Table object representing
a core DD table.
@param reset Reset counters.
@retval true An error occurred.
@retval false Success - no errors.
*/
bool ha_get_se_private_data(dd::Table *dd_table, bool reset);
void adjust_next_insert_id_after_explicit_value(ulonglong nr);
int update_auto_increment();
virtual void print_error(int error, myf errflag);
virtual bool get_error_message(int error, String *buf);
uint get_dup_key(int error);
/**
Retrieves the names of the table and the key for which there was a
duplicate entry in the case of HA_ERR_FOREIGN_DUPLICATE_KEY.
If any of the table or key name is not available this method will return
false and will not change any of child_table_name or child_key_name.
@param [out] child_table_name Table name
@param [in] child_table_name_len Table name buffer size
@param [out] child_key_name Key name
@param [in] child_key_name_len Key name buffer size
@retval true table and key names were available
and were written into the corresponding
out parameters.
@retval false table and key names were not available,
the out parameters were not touched.
*/
virtual bool get_foreign_dup_key(char *child_table_name,
uint child_table_name_len,
char *child_key_name,
uint child_key_name_len);
/**
Change the internal TABLE_SHARE pointer.
@param table_arg TABLE object
@param share New share to use
@note Is used in error handling in ha_delete_table.
*/
virtual void change_table_ptr(TABLE *table_arg, TABLE_SHARE *share) {
table = table_arg;
table_share = share;
}
const TABLE_SHARE *get_table_share() const { return table_share; }
/* Estimates calculation */
/**
@deprecated This function is deprecated and will be removed in a future
version. Use table_scan_cost() instead.
*/
virtual double scan_time() {
return ulonglong2double(stats.data_file_length) / IO_SIZE + 2;
}
/**
The cost of reading a set of ranges from the table using an index
to access it.
@deprecated This function is deprecated and will be removed in a future
version. Use read_cost() instead.
@param index The index number.
@param ranges The number of ranges to be read.
@param rows Total number of rows to be read.
This method can be used to calculate the total cost of scanning a table
using an index by calling it using read_time(index, 1, table_size).
*/
virtual double read_time(uint index [[maybe_unused]], uint ranges,
ha_rows rows) {
return rows2double(ranges + rows);
}
/**
@deprecated This function is deprecated and will be removed in a future
version. Use index_scan_cost() instead.
*/
virtual double index_only_read_time(uint keynr, double records);
/**
Cost estimate for doing a complete table scan.
@note For this version it is recommended that storage engines continue
to override scan_time() instead of this function.
@returns the estimated cost
*/
virtual Cost_estimate table_scan_cost();
/**
Cost estimate for reading a number of ranges from an index.
The cost estimate will only include the cost of reading data that
is contained in the index. If the records need to be read, use
read_cost() instead.
@note The ranges parameter is currently ignored and is not taken
into account in the cost estimate.
@note For this version it is recommended that storage engines continue
to override index_only_read_time() instead of this function.
@param index the index number
@param ranges the number of ranges to be read
@param rows total number of rows to be read
@returns the estimated cost
*/
virtual Cost_estimate index_scan_cost(uint index, double ranges, double rows);
/**
Cost estimate for reading a set of ranges from the table using an index
to access it.
@note For this version it is recommended that storage engines continue
to override read_time() instead of this function.
@param index the index number
@param ranges the number of ranges to be read
@param rows total number of rows to be read
@returns the estimated cost
*/
virtual Cost_estimate read_cost(uint index, double ranges, double rows);
/**
Cost estimate for doing a number of non-sequentially accesses
against the storage engine. Such accesses can be either number
of rows to read, or number of disk pages to access.
Each handler implementation is free to interpret that as best
suited, depending on what is the dominating cost for that
storage engine.
This method is mainly provided as a temporary workaround for
bug#33317872, where we fix problems caused by calling
Cost_model::page_read_cost() directly from the optimizer.
That should be avoided, as it introduced assumption about all
storage engines being disk-page based, and having a 'page' cost.
Furthermore, this page cost was even compared against read_cost(),
which was computed with an entirely different algorithm, and thus
could not be compared.
The default implementation still call Cost_model::page_read_cost(),
thus behaving just as before. However, handler implementation may
override it to call handler::read_cost() instead(), which probably
will be more correct. (If a page_read_cost should be included
in the cost estimate, that should preferable be done inside
each read_cost() implementation)
Longer term we should consider to remove all page_read_cost()
usage from the optimizer itself, making this method obsolete.
@param index the index number
@param reads the number of accesses being made
@returns the estimated cost
*/
virtual double page_read_cost(uint index, double reads);
/**
Provide an upper cost-limit of doing a specified number of
seek-and-read key lookups. This need to be comparable and
calculated with the same 'metric' as page_read_cost.
@param reads the number of rows read in the 'worst' case.
@returns the estimated cost
*/
virtual double worst_seek_times(double reads);
/**
Return an estimate on the amount of memory the storage engine will
use for caching data in memory. If this is unknown or the storage
engine does not cache data in memory -1 is returned.
*/
virtual longlong get_memory_buffer_size() const { return -1; }
/**
Return an estimate of how much of the table that is currently stored
in main memory.
This estimate should be the fraction of the table that currently
is available in a main memory buffer. The estimate should be in the
range from 0.0 (nothing in memory) to 1.0 (entire table in memory).
@return The fraction of the table in main memory buffer
*/
double table_in_memory_estimate() const;
/**
Return an estimate of how much of the index that is currently stored
in main memory.
This estimate should be the fraction of the index that currently
is available in a main memory buffer. The estimate should be in the
range from 0.0 (nothing in memory) to 1.0 (entire index in memory).
@param keyno the index to get an estimate for
@return The fraction of the index in main memory buffer
*/
double index_in_memory_estimate(uint keyno) const;
/**
Initialize sampling.
@param[out] scan_ctx A scan context created by this method that has to be
used in sample_next
@param[in] sampling_percentage percentage of records that need to be
sampled
@param[in] sampling_seed random seed that the random generator will
use
@param[in] sampling_method sampling method to be used; currently only
SYSTEM sampling is supported
@param[in] tablesample true if the sampling is for tablesample
@return 0 for success, else one of the HA_xxx values in case of error.
*/
int ha_sample_init(void *&scan_ctx, double sampling_percentage,
int sampling_seed, enum_sampling_method sampling_method,
const bool tablesample);
/**
Get the next record for sampling.
@param[in] scan_ctx Scan context of the sampling
@param[in] buf buffer to place the read record
@return 0 for success, else one of the HA_xxx values in case of error.
*/
int ha_sample_next(void *scan_ctx, uchar *buf);
/**
End sampling.
@param[in] scan_ctx Scan context of the sampling
@return 0 for success, else one of the HA_xxx values in case of error.
*/
int ha_sample_end(void *scan_ctx);
private:
int check_collation_compatibility();
/**
Make a guesstimate for how much of a table or index is in a memory
buffer in the case where the storage engine has not provided any
estimate for this.
@param table_index_size size of the table or index
@return The fraction of the table or index in main memory buffer
*/
double estimate_in_memory_buffer(ulonglong table_index_size) const;
public:
virtual ha_rows multi_range_read_info_const(
uint keyno, RANGE_SEQ_IF *seq, void *seq_init_param, uint n_ranges,
uint *bufsz, uint *flags, bool *force_default_mrr, Cost_estimate *cost);
virtual ha_rows multi_range_read_info(uint keyno, uint n_ranges, uint keys,
uint *bufsz, uint *flags,
Cost_estimate *cost);
virtual int multi_range_read_init(RANGE_SEQ_IF *seq, void *seq_init_param,
uint n_ranges, uint mode,
HANDLER_BUFFER *buf);
int ha_multi_range_read_next(char **range_info);
int ha_read_range_first(const key_range *start_key, const key_range *end_key,
bool eq_range, bool sorted);
int ha_read_range_next();
bool has_transactions() {
return (ha_table_flags() & HA_NO_TRANSACTIONS) == 0;
}
virtual uint extra_rec_buf_length() const { return 0; }
/**
@brief Determine whether an error can be ignored or not.
@details This method is used to analyze the error to see whether the
error is ignorable or not. Such errors will be reported as warnings
instead of errors for IGNORE statements. This means that the statement
will not abort, but instead continue to the next row.
HA_ERR_FOUND_DUP_UNIQUE is a special case in MyISAM that means the
same thing as HA_ERR_FOUND_DUP_KEY, but can in some cases lead to
a slightly different error message.
@param error error code received from the handler interface (HA_ERR_...)
@return whether the error is ignorable or not
@retval true the error is ignorable
@retval false the error is not ignorable
*/
virtual bool is_ignorable_error(int error);
/**
@brief Determine whether an error is fatal or not.
@details This method is used to analyze the error to see whether the
error is fatal or not. A fatal error is an error that will not be
possible to handle with SP handlers and will not be subject to
retry attempts on the slave.
@param error error code received from the handler interface (HA_ERR_...)
@return whether the error is fatal or not
@retval true the error is fatal
@retval false the error is not fatal
*/
virtual bool is_fatal_error(int error);
protected:
virtual int multi_range_read_next(char **range_info);
/**
Number of rows in table. If HA_COUNT_ROWS_INSTANT is set, count is
available instantly. Else do a table scan.
@param num_rows [out] num_rows number of rows in table.
@retval 0 for OK, one of the HA_xxx values in case of error.
*/
virtual int records(ha_rows *num_rows);
/**
Number of rows in table counted using the secondary index chosen by
optimizer. See comments in optimize_aggregated_query() .
@param num_rows [out] Number of rows in table.
@param index Index chosen by optimizer for counting.
@retval 0 for OK, one of the HA_xxx values in case of error.
*/
virtual int records_from_index(ha_rows *num_rows, uint index);
private:
/**
Function will handle the error code from call to records() and
records_from_index().
@param error return code from records() and records_from_index().
@param num_rows Check if it contains HA_POS_ERROR in case error < 0.
@retval 0 for OK, one of the HA_xxx values in case of error.
*/
int handle_records_error(int error, ha_rows *num_rows);
public:
/**
Wrapper function to call records() in storage engine.
@param num_rows [out] Number of rows in table.
@retval 0 for OK, one of the HA_xxx values in case of error.
*/
int ha_records(ha_rows *num_rows) {
return handle_records_error(records(num_rows), num_rows);
}
/**
Wrapper function to call records_from_index() in storage engine.
@param num_rows [out] Number of rows in table.
@param index Index chosen by optimizer for counting.
@retval 0 for OK, one of the HA_xxx values in case of error.
*/
int ha_records(ha_rows *num_rows, uint index) {
return handle_records_error(records_from_index(num_rows, index), num_rows);
}
/**
Return upper bound of current number of records in the table
(max. of how many records one will retrieve when doing a full table scan)
If upper bound is not known, HA_POS_ERROR should be returned as a max
possible upper bound.
*/
virtual ha_rows estimate_rows_upper_bound() {
return stats.records + EXTRA_RECORDS;
}
/**
Get real row type for the table created based on one specified by user,
CREATE TABLE options and SE capabilities.
*/
virtual enum row_type get_real_row_type(
const HA_CREATE_INFO *create_info) const {
return (create_info->table_options & HA_OPTION_COMPRESS_RECORD)
? ROW_TYPE_COMPRESSED
: ((create_info->table_options & HA_OPTION_PACK_RECORD)
? ROW_TYPE_DYNAMIC
: ROW_TYPE_FIXED);
}
/**
Get default key algorithm for SE. It is used when user has not provided
algorithm explicitly or when algorithm specified is not supported by SE.
*/
virtual enum ha_key_alg get_default_index_algorithm() const {
return HA_KEY_ALG_SE_SPECIFIC;
}
/**
Check if SE supports specific key algorithm.
@note This method is never used for FULLTEXT or SPATIAL keys.
We rely on handler::ha_table_flags() to check if such keys
are supported.
*/
virtual bool is_index_algorithm_supported(enum ha_key_alg key_alg) const {
return key_alg == HA_KEY_ALG_SE_SPECIFIC;
}
/**
Signal that the table->read_set and table->write_set table maps changed
The handler is allowed to set additional bits in the above map in this
call. Normally the handler should ignore all calls until we have done
a ha_rnd_init() or ha_index_init(), write_row(), update_row or delete_row()
as there may be several calls to this routine.
*/
virtual void column_bitmaps_signal();
uint get_index(void) const { return active_index; }
/**
@retval false Bulk update used by handler
@retval true Bulk update not used, normal operation used
*/
virtual bool start_bulk_update() { return true; }
/**
@retval false Bulk delete used by handler
@retval true Bulk delete not used, normal operation used
*/
virtual bool start_bulk_delete() { return true; }
/**
After this call all outstanding updates must be performed. The number
of duplicate key errors are reported in the duplicate key parameter.
It is allowed to continue to the batched update after this call, the
handler has to wait until end_bulk_update with changing state.
@param dup_key_found Number of duplicate keys found
@retval 0 Success
@retval >0 Error code
*/
virtual int exec_bulk_update(uint *dup_key_found [[maybe_unused]]) {
assert(false);
return HA_ERR_WRONG_COMMAND;
}
/**
Perform any needed clean-up, no outstanding updates are there at the
moment.
*/
virtual void end_bulk_update() { return; }
/**
Execute all outstanding deletes and close down the bulk delete.
@retval 0 Success
@retval >0 Error code
*/
virtual int end_bulk_delete() {
assert(false);
return HA_ERR_WRONG_COMMAND;
}
protected:
/**
@brief
Positions an index cursor to the index specified in the handle
('active_index'). Fetches the row if available. If the key value is null,
begin at the first key of the index.
@returns 0 if success (found a record); non-zero if no record.
*/
virtual int index_read_map(uchar *buf, const uchar *key,
key_part_map keypart_map,
enum ha_rkey_function find_flag) {
uint key_len = calculate_key_len(table, active_index, keypart_map);
return index_read(buf, key, key_len, find_flag);
}
/**
Positions an index cursor to the index specified in argument. Fetches
the row if available. If the key value is null, begin at the first key of
the index.
@sa index_read_map()
*/
virtual int index_read_idx_map(uchar *buf, uint index, const uchar *key,
key_part_map keypart_map,
enum ha_rkey_function find_flag);
/*
These methods are used to jump to next or previous entry in the index
scan. There are also methods to jump to first and last entry.
*/
/// @see index_read_map().
virtual int index_next(uchar *) { return HA_ERR_WRONG_COMMAND; }
/// @see index_read_map().
virtual int index_prev(uchar *) { return HA_ERR_WRONG_COMMAND; }
/// @see index_read_map().
virtual int index_first(uchar *) { return HA_ERR_WRONG_COMMAND; }
/// @see index_read_map().
virtual int index_last(uchar *) { return HA_ERR_WRONG_COMMAND; }
/// @see index_read_map().
virtual int index_next_same(uchar *buf, const uchar *key, uint keylen);
/**
The following functions works like index_read, but it find the last
row with the current key value or prefix.
@see index_read_map().
*/
virtual int index_read_last_map(uchar *buf, const uchar *key,
key_part_map keypart_map) {
uint key_len = calculate_key_len(table, active_index, keypart_map);
return index_read_last(buf, key, key_len);
}
virtual int read_range_first(const key_range *start_key,
const key_range *end_key, bool eq_range_arg,
bool sorted);
virtual int read_range_next();
public:
/**
Set the end position for a range scan. This is used for checking
for when to end the range scan and by the ICP code to determine
that the next record is within the current range.
@param range The end value for the range scan
@param direction Direction of the range scan
*/
void set_end_range(const key_range *range,
enum_range_scan_direction direction);
int compare_key(key_range *range);
int compare_key_icp(const key_range *range) const;
int compare_key_in_buffer(const uchar *buf) const;
virtual int ft_init() { return HA_ERR_WRONG_COMMAND; }
virtual FT_INFO *ft_init_ext(uint flags, uint inx, String *key);
virtual FT_INFO *ft_init_ext_with_hints(uint inx, String *key,
Ft_hints *hints) {
return ft_init_ext(hints->get_flags(), inx, key);
}
int ha_ft_read(uchar *buf);
int ha_read_first_row(uchar *buf, uint primary_key);
protected:
/// @see index_read_map().
virtual int rnd_next(uchar *buf) = 0;
/// @see index_read_map().
virtual int rnd_pos(uchar *buf, uchar *pos) = 0;
virtual int ft_read(uchar *) { return HA_ERR_WRONG_COMMAND; }
public:
/**
This function only works for handlers having
HA_PRIMARY_KEY_REQUIRED_FOR_POSITION set.
It will return the row with the PK given in the record argument.
*/
virtual int rnd_pos_by_record(uchar *record) {
int error;
assert(table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION);
error = ha_rnd_init(false);
if (error != 0) return error;
position(record);
error = ha_rnd_pos(record, ref);
ha_rnd_end();
return error;
}
/**
Find number of records in a range.
Given a starting key, and an ending key estimate the number of rows that
will exist between the two. max_key may be empty which in case determine
if start_key matches any rows. Used by optimizer to calculate cost of
using a particular index.
@param inx Index number
@param min_key Start of range
@param max_key End of range
@return Number of rows in range.
*/
virtual ha_rows records_in_range(uint inx [[maybe_unused]],
key_range *min_key [[maybe_unused]],
key_range *max_key [[maybe_unused]]) {
return (ha_rows)10;
}
/*
If HA_PRIMARY_KEY_REQUIRED_FOR_POSITION is set, then it sets ref
(reference to the row, aka position, with the primary key given in
the record).
Otherwise it set ref to the current row.
*/
virtual void position(const uchar *record) = 0;
/**
General method to gather info from handler
::info() is used to return information to the optimizer.
SHOW also makes use of this data Another note, if your handler
doesn't proved exact record count, you will probably want to
have the following in your code:
if (records < 2)
records = 2;
The reason is that the server will optimize for cases of only a single
record. If in a table scan you don't know the number of records
it will probably be better to set records to two so you can return
as many records as you need.
Along with records a few more variables you may wish to set are:
records
deleted
data_file_length
index_file_length
delete_length
check_time
Take a look at the public variables in handler.h for more information.
See also my_base.h for a full description.
@param flag Specifies what info is requested
*/
virtual int info(uint flag) = 0;
virtual uint32 calculate_key_hash_value(Field **field_array
[[maybe_unused]]) {
assert(0);
return 0;
}
/**
Request storage engine to do an extra operation: enable,disable or run some
functionality.
@param operation the operation to perform
@returns
0 on success
error otherwise
*/
int ha_extra(enum ha_extra_function operation);
private:
/**
Storage engine specific implementation of ha_extra()
@param operation the operation to perform
@returns
0 on success
error otherwise
*/
virtual int extra(enum ha_extra_function operation [[maybe_unused]]) {
return 0;
}
public:
virtual int extra_opt(enum ha_extra_function operation,
ulong cache_size [[maybe_unused]]) {
return extra(operation);
}
/**
Get the handlerton of the storage engine if the SE is capable of
pushing down some of the AccessPath functionality.
(Join, Filter conditions, ... possiby more)
Call the handlerton::push_to_engine() method for performing the
actual pushdown of (parts of) the AccessPath functionality
@returns handlerton* of the SE if it may be capable of
off loading part of the query by calling
handlerton::push_to_engine()
Else, 'nullptr' is returned.
*/
virtual const handlerton *hton_supporting_engine_pushdown() {
return nullptr;
}
/**
Start read (before write) removal on the current table.
@see HA_READ_BEFORE_WRITE_REMOVAL
*/
virtual bool start_read_removal(void) {
assert(0);
return false;
}
/**
End read (before write) removal and return the number of rows
really written
@see HA_READ_BEFORE_WRITE_REMOVAL
*/
virtual ha_rows end_read_removal(void) {
assert(0);
return (ha_rows)0;
}
/**
Normally, when running UPDATE or DELETE queries, we need to wait for other
transactions to release their locks on a given row before we can read it and
potentially update it. However, in READ UNCOMMITTED and READ COMMITTED, we
can ignore these locks if we don't intend to modify the row (e.g., because
it failed a WHERE). This is signaled through enabling “semi-consistent
read”, by calling try_semi_consistent_read(true) (and then setting it back
to false after finishing the query).
If semi-consistent read is enabled, and we are in READ UNCOMMITTED or READ
COMMITTED, the storage engine is permitted to return rows that are locked
and thus un-updatable. If the optimizer doesn't want the row, e.g., because
it got filtered out, it can call unlock_row() as usual. However, if it
intends to update the row, it needs to call was_semi_consistent_read()
before doing so. If was_semi_consistent_read() returns false, the row was
never locked to begin with and can be updated as usual. However, if it
returns 1, it was read optimistically, must be discarded (ie., do not try to
update the row) and must be re-read with locking enabled. The next read call
after was_semi_consistent_read() will automatically re-read the same row,
this time with locking enabled.
Thus, typical use in an UPDATE scenario would look like this:
file->try_semi_consistent_read(true);
file->ha_rnd_init(true);
while (file->ha_rnd_next(table->record[0]) == 0) {
if (row is filtered...) {
file->unlock_row();
continue;
}
if (file->was_semi_consistent_read()) {
// Discard the row; next ha_rnd_next() will read it again with
// locking.
continue;
}
// Process row here.
}
file->ha_rnd_end();
file->try_semi_consistent_read(false);
If the transaction isolation level is REPEATABLE READ or SERIALIZABLE,
enabling this flag has no effect.
*/
virtual bool was_semi_consistent_read() { return false; }
/**
Tell the engine whether it should avoid unnecessary lock waits.
If yes, in an UPDATE or DELETE, if the row under the cursor was locked
by another transaction, the engine may try an optimistic read of
the last committed row value under the cursor.
*/
virtual void try_semi_consistent_read(bool) {}
/**
Unlock last accessed row.
Record currently processed was not in the result set of the statement
and is thus unlocked. Used for UPDATE and DELETE queries.
*/
virtual void unlock_row() {}
/**
Start a statement when table is locked
This method is called instead of external lock when the table is locked
before the statement is executed.
@param thd Thread object.
@param lock_type Type of external lock.
@retval >0 Error code.
@retval 0 Success.
*/
virtual int start_stmt(THD *thd [[maybe_unused]],
thr_lock_type lock_type [[maybe_unused]]) {
return 0;
}
virtual void get_auto_increment(ulonglong offset, ulonglong increment,
ulonglong nb_desired_values,
ulonglong *first_value,
ulonglong *nb_reserved_values);
void set_next_insert_id(ulonglong id) {
DBUG_PRINT("info", ("auto_increment: next value %lu", (ulong)id));
next_insert_id = id;
}
void restore_auto_increment(ulonglong prev_insert_id) {
/*
Insertion of a row failed, re-use the lastly generated auto_increment
id, for the next row. This is achieved by resetting next_insert_id to
what it was before the failed insertion (that old value is provided by
the caller). If that value was 0, it was the first row of the INSERT;
then if insert_id_for_cur_row contains 0 it means no id was generated
for this first row, so no id was generated since the INSERT started, so
we should set next_insert_id to 0; if insert_id_for_cur_row is not 0, it
is the generated id of the first and failed row, so we use it.
*/
next_insert_id =
(prev_insert_id > 0) ? prev_insert_id : insert_id_for_cur_row;
}
/**
Update create info as part of ALTER TABLE.
Forward this handler call to the storage engine foreach
partition handler. The data_file_name for each partition may
need to be reset if the tablespace was moved. Use a dummy
HA_CREATE_INFO structure and transfer necessary data.
@param create_info Create info from ALTER TABLE.
*/
virtual void update_create_info(HA_CREATE_INFO *create_info
[[maybe_unused]]) {}
virtual int assign_to_keycache(THD *, HA_CHECK_OPT *) {
return HA_ADMIN_NOT_IMPLEMENTED;
}
virtual int preload_keys(THD *, HA_CHECK_OPT *) {
return HA_ADMIN_NOT_IMPLEMENTED;
}
/* end of the list of admin commands */
/**
Check if indexes are disabled.
@retval 0 Indexes are enabled.
@retval != 0 Indexes are disabled.
*/
virtual int indexes_are_disabled(void) { return 0; }
virtual void append_create_info(String *packet [[maybe_unused]]) {}
virtual void init_table_handle_for_HANDLER() {
return;
} /* prepare InnoDB for HANDLER */
/** The following can be called without an open handler */
virtual const char *table_type() const = 0;
virtual ulong index_flags(uint idx, uint part, bool all_parts) const = 0;
uint max_record_length() const {
return std::min(HA_MAX_REC_LENGTH, max_supported_record_length());
}
uint max_keys() const {
return std::min<uint>(MAX_KEY, max_supported_keys());
}
uint max_key_parts() const {
return std::min(MAX_REF_PARTS, max_supported_key_parts());
}
uint max_key_length() const {
return std::min(MAX_KEY_LENGTH, max_supported_key_length());
}
uint max_key_part_length(HA_CREATE_INFO *create_info) const {
return std::min(MAX_KEY_LENGTH, max_supported_key_part_length(create_info));
}
virtual uint max_supported_record_length() const { return HA_MAX_REC_LENGTH; }
virtual uint max_supported_keys() const { return 0; }
virtual uint max_supported_key_parts() const { return MAX_REF_PARTS; }
virtual uint max_supported_key_length() const { return MAX_KEY_LENGTH; }
virtual uint max_supported_key_part_length(HA_CREATE_INFO *create_info
[[maybe_unused]]) const {
return 255;
}
virtual uint min_record_length(uint options [[maybe_unused]]) const {
return 1;
}
virtual bool low_byte_first() const { return true; }
virtual ha_checksum checksum() const { return 0; }
/**
Check if the table is crashed.
@retval true Crashed
@retval false Not crashed
*/
virtual bool is_crashed() const { return false; }
/**
Check if the table can be automatically repaired.
@retval true Can be auto repaired
@retval false Cannot be auto repaired
*/
virtual bool auto_repair() const { return false; }
/**
Get number of lock objects returned in store_lock.
Returns the number of store locks needed in call to store lock.
We return number of partitions we will lock multiplied with number of
locks needed by each partition. Assists the above functions in allocating
sufficient space for lock structures.
@returns Number of locks returned in call to store_lock.
@note lock_count() can return > 1 if the table is MERGE or partitioned.
*/
virtual uint lock_count(void) const { return 1; }
/**
Is not invoked for non-transactional temporary tables.
@note store_lock() can return more than one lock if the table is MERGE
or partitioned.
@note that one can NOT rely on table->in_use in store_lock(). It may
refer to a different thread if called from mysql_lock_abort_for_thread().
@note If the table is MERGE, store_lock() can return less locks
than lock_count() claimed. This can happen when the MERGE children
are not attached when this is called from another thread.
The idea with handler::store_lock() is the following:
The statement decided which locks we should need for the table
for updates/deletes/inserts we get WRITE locks, for SELECT... we get
read locks.
Before adding the lock into the table lock handler (see thr_lock.c)
mysqld calls store lock with the requested locks. Store lock can now
modify a write lock to a read lock (or some other lock), ignore the
lock (if we don't want to use MySQL table locks at all) or add locks
for many tables (like we do when we are using a MERGE handler).
In some exceptional cases MySQL may send a request for a TL_IGNORE;
This means that we are requesting the same lock as last time and this
should also be ignored.
Called from lock.cc by get_lock_data().
*/
virtual THR_LOCK_DATA **store_lock(THD *thd, THR_LOCK_DATA **to,
enum thr_lock_type lock_type) = 0;
/**
Check if the primary key is clustered or not.
@retval true Primary key (if there is one) is a clustered
key covering all fields
@retval false otherwise
*/
virtual bool primary_key_is_clustered() const { return false; }
/**
Compare two positions.
@param ref1 First position.
@param ref2 Second position.
@retval <0 ref1 < ref2.
@retval 0 Equal.
@retval >0 ref1 > ref2.
*/
virtual int cmp_ref(const uchar *ref1, const uchar *ref2) const {
return memcmp(ref1, ref2, ref_length);
}
/*
Condition pushdown to storage engines
*/
/**
Push condition down to the table handler.
@param cond Condition to be pushed. The condition tree
must not be modified by the caller.
@return
The 'remainder' condition that caller must use to filter out records.
NULL means the handler will not return rows that do not match the
passed condition.
@note
handler->ha_reset() call discard any pushed conditions.
Calls to rnd_init/rnd_end, index_init/index_end etc do not affect the
pushed conditions.
*/
virtual const Item *cond_push(const Item *cond) {
assert(pushed_cond == nullptr);
return cond;
}
/**
Push down an index condition to the handler.
The server will use this method to push down a condition it wants
the handler to evaluate when retrieving records using a specified
index. The pushed index condition will only refer to fields from
this handler that is contained in the index (but it may also refer
to fields in other handlers). Before the handler evaluates the
condition it must read the content of the index entry into the
record buffer.
The handler is free to decide if and how much of the condition it
will take responsibility for evaluating. Based on this evaluation
it should return the part of the condition it will not evaluate.
If it decides to evaluate the entire condition it should return
NULL. If it decides not to evaluate any part of the condition it
should return a pointer to the same condition as given as argument.
@param keyno the index number to evaluate the condition on
@param idx_cond the condition to be evaluated by the handler
@return The part of the pushed condition that the handler decides
not to evaluate
*/
virtual Item *idx_cond_push(uint keyno [[maybe_unused]], Item *idx_cond) {
return idx_cond;
}
/** Reset information about pushed index conditions */
virtual void cancel_pushed_idx_cond() {
pushed_idx_cond = nullptr;
pushed_idx_cond_keyno = MAX_KEY;
in_range_check_pushed_down = false;
}
/**
Reports number of tables included in pushed join which this
handler instance is part of. ==0 -> Not pushed
*/
virtual uint number_of_pushed_joins() const { return 0; }
/**
If this handler instance is part of a pushed join sequence
returned TABLE instance being root of the pushed query?
*/
virtual const TABLE *member_of_pushed_join() const { return nullptr; }
/**
If this handler instance is a child in a pushed join sequence
returned TABLE instance being my parent?
*/
virtual const TABLE *parent_of_pushed_join() const { return nullptr; }
/// @returns a map of the tables involved in this pushed join, or 0 if not
/// part of a pushed join.
virtual table_map tables_in_pushed_join() const { return 0; }
int ha_index_read_pushed(uchar *buf, const uchar *key,
key_part_map keypart_map);
int ha_index_next_pushed(uchar *buf);
protected:
virtual int index_read_pushed(uchar *, const uchar *, key_part_map) {
return HA_ERR_WRONG_COMMAND;
}
virtual int index_next_pushed(uchar *) { return HA_ERR_WRONG_COMMAND; }
public:
/**
Part of old, deprecated in-place ALTER API.
*/
virtual bool check_if_incompatible_data(HA_CREATE_INFO *create_info
[[maybe_unused]],
uint table_changes [[maybe_unused]]) {
return COMPATIBLE_DATA_NO;
}
/* On-line/in-place/instant ALTER TABLE interface. */
/*
Here is an outline of on-line/in-place ALTER TABLE execution through
this interface.
Phase 1 : Initialization
========================
During this phase we determine which algorithm should be used
for execution of ALTER TABLE and what level concurrency it will
require.
*) This phase starts by opening the table and preparing description
of the new version of the table.
*) Then we check if it is impossible even in theory to carry out
this ALTER TABLE using the in-place/instant algorithm. For example,
because we need to change storage engine or the user has explicitly
requested usage of the "copy" algorithm.
*) If in-place/instant ALTER TABLE is theoretically possible, we continue
by compiling differences between old and new versions of the table
in the form of HA_ALTER_FLAGS bitmap. We also build a few
auxiliary structures describing requested changes and store
all these data in the Alter_inplace_info object.
*) Then the handler::check_if_supported_inplace_alter() method is called
in order to find if the storage engine can carry out changes requested
by this ALTER TABLE using the in-place or instant algorithm.
To determine this, the engine can rely on data in HA_ALTER_FLAGS/
Alter_inplace_info passed to it as well as on its own checks.
If the in-place algorithm can be used for this ALTER TABLE, the level
of required concurrency for its execution is also returned.
If any errors occur during the handler call, ALTER TABLE is aborted
and no further handler functions are called.
Note that in cases when there is difference between in-place and
instant algorithm and user explicitly asked for usage of in-place
algorithm storage engine MUST return one of values corresponding
to in-place algorithm and not HA_ALTER_INPLACE_INSTANT from this
method.
*) Locking requirements of the in-place algorithm are compared to any
concurrency requirements specified by user. If there is a conflict
between them, we either switch to the copy algorithm or emit an error.
Phase 2 : Execution
===================
In this phase the operations are executed.
*) As the first step, we acquire a lock corresponding to the concurrency
level which was returned by handler::check_if_supported_inplace_alter()
and requested by the user. This lock is held for most of the
duration of in-place ALTER (if HA_ALTER_INPLACE_SHARED_LOCK_AFTER_PREPARE
or HA_ALTER_INPLACE_NO_LOCK_AFTER_PREPARE were returned we acquire an
exclusive lock for duration of the next step only).
For HA_ALTER_INPLACE_INSTANT we keep shared upgradable metadata lock
which was acquired at table open time.
*) After that we call handler::ha_prepare_inplace_alter_table() to give the
storage engine a chance to update its internal structures with a higher
lock level than the one that will be used for the main step of algorithm.
After that we downgrade the lock if it is necessary.
This step should be no-op for instant algorithm.
*) After that, the main step of this phase and algorithm is executed.
We call the handler::ha_inplace_alter_table() method, which carries out
the changes requested by ALTER TABLE but does not makes them visible to
other connections yet.
This step should be no-op for instant algorithm as well.
*) We ensure that no other connection uses the table by upgrading our
lock on it to exclusive.
*) a) If the previous step succeeds,
handler::ha_commit_inplace_alter_table() is called to allow the storage
engine to do any final updates to its structures, to make all earlier
changes durable and visible to other connections.
For instant algorithm this is the step during which SE changes are done.
Engines that support atomic DDL only prepare for the commit during this
step but do not finalize it. Real commit happens later when the whole
statement is committed. Also in some situations statement might be rolled
back after call to commit_inplace_alter_table() for such storage engines.
In the latter special case SE might require call to
handlerton::dict_cache_reset() in order to invalidate its internal table
definition cache after rollback.
b) If we have failed to upgrade lock or any errors have occurred during
the handler functions calls (including commit), we call
handler::ha_commit_inplace_alter_table() to rollback all changes which
were done during previous steps.
All the above calls to SE are provided with dd::Table objects describing old
and new version of table being altered. Engines which support atomic DDL are
allowed to adjust object corresponding to the new version. During phase 3
these changes are saved to the data-dictionary.
Phase 3 : Final
===============
In this phase we:
a) For engines which don't support atomic DDL:
*) Update the SQL-layer data-dictionary by replacing description of old
version of the table with its new version. This change is immediately
committed.
*) Inform the storage engine about this change by calling the
handler::ha_notify_table_changed() method.
*) Process the RENAME clause by calling handler::ha_rename_table() and
updating the data-dictionary accordingly. Again this change is
immediately committed.
*) Destroy the Alter_inplace_info and handler_ctx objects.
b) For engines which support atomic DDL:
*) Update the SQL-layer data-dictionary by replacing description of old
version of the table with its new version.
*) Process the RENAME clause by calling handler::ha_rename_table() and
updating the data-dictionary accordingly.
*) Commit the statement/transaction.
*) Finalize atomic DDL operation by calling handlerton::post_ddl() hook
for the storage engine.
*) Additionally inform the storage engine about completion of ALTER TABLE
for the table by calling the handler::ha_notify_table_changed()
method.
*) Destroy the Alter_inplace_info and handler_ctx objects.
*/
/**
Check if a storage engine supports a particular alter table in-place
@param altered_table TABLE object for new version of table.
@param ha_alter_info Structure describing changes to be done
by ALTER TABLE and holding data used
during in-place alter.
@retval HA_ALTER_ERROR Unexpected error.
@retval HA_ALTER_INPLACE_NOT_SUPPORTED Not supported, must use copy.
@retval HA_ALTER_INPLACE_EXCLUSIVE_LOCK Supported, but requires X lock.
@retval HA_ALTER_INPLACE_SHARED_LOCK_AFTER_PREPARE
Supported, but requires SNW lock
during main phase. Prepare phase
requires X lock.
@retval HA_ALTER_INPLACE_SHARED_LOCK Supported, but requires SNW lock.
@retval HA_ALTER_INPLACE_NO_LOCK_AFTER_PREPARE
Supported, concurrent
reads/writes allowed. However, prepare phase requires X lock.
@retval HA_ALTER_INPLACE_NO_LOCK Supported, concurrent
reads/writes allowed.
@retval HA_ALTER_INPLACE_INSTANT Instant algorithm is supported.
Prepare and main phases are
no-op. Changes happen during
commit phase and it should be
"instant". We keep SU lock,
allowing concurrent reads and
writes during no-op phases and
upgrade it to X lock before
commit phase.
@note The default implementation uses the old in-place ALTER API
to determine if the storage engine supports in-place ALTER or not.
@note In cases when there is difference between in-place and instant
algorithm and explicit ALGORITHM=INPLACE clause was provided SE MUST
return one of values corresponding to in-place algorithm and not
HA_ALTER_INPLACE_INSTANT from this method.
@note Called without holding thr_lock.c lock.
*/
virtual enum_alter_inplace_result check_if_supported_inplace_alter(
TABLE *altered_table, Alter_inplace_info *ha_alter_info);
/**
Public functions wrapping the actual handler call.
@see prepare_inplace_alter_table()
*/
bool ha_prepare_inplace_alter_table(TABLE *altered_table,
Alter_inplace_info *ha_alter_info,
const dd::Table *old_table_def,
dd::Table *new_table_def);
/**
Public function wrapping the actual handler call.
@see inplace_alter_table()
*/
bool ha_inplace_alter_table(TABLE *altered_table,
Alter_inplace_info *ha_alter_info,
const dd::Table *old_table_def,
dd::Table *new_table_def) {
return inplace_alter_table(altered_table, ha_alter_info, old_table_def,
new_table_def);
}
/**
Public function wrapping the actual handler call.
Allows us to enforce asserts regardless of handler implementation.
@see commit_inplace_alter_table()
*/
bool ha_commit_inplace_alter_table(TABLE *altered_table,
Alter_inplace_info *ha_alter_info,
bool commit,
const dd::Table *old_table_def,
dd::Table *new_table_def);
/**
Public function wrapping the actual handler call.
@see notify_table_changed()
*/
void ha_notify_table_changed(Alter_inplace_info *ha_alter_info) {
notify_table_changed(ha_alter_info);
}
protected:
/**
Allows the storage engine to update internal structures with concurrent
writes blocked. If check_if_supported_inplace_alter() returns
HA_ALTER_INPLACE_NO_LOCK_AFTER_PREPARE or
HA_ALTER_INPLACE_SHARED_AFTER_PREPARE, this function is called with
exclusive lock otherwise the same level of locking as for
inplace_alter_table() will be used.
@note Should be no-op for instant algorithm.
@note Storage engines are responsible for reporting any errors by
calling my_error()/print_error()
@note If this function reports error, commit_inplace_alter_table()
will be called with commit= false.
@note For partitioning, failing to prepare one partition, means that
commit_inplace_alter_table() will be called to roll back changes for
all partitions. This means that commit_inplace_alter_table() might be
called without prepare_inplace_alter_table() having been called first
for a given partition.
@param altered_table TABLE object for new version of table.
@param ha_alter_info Structure describing changes to be done
by ALTER TABLE and holding data used
during in-place alter.
@param old_table_def dd::Table object describing old version of
the table.
@param new_table_def dd::Table object for the new version of the
table. Can be adjusted by this call if SE
supports atomic DDL. These changes to the
table definition will be persisted in the
data-dictionary at statement commit time.
@retval true Error
@retval false Success
*/
virtual bool prepare_inplace_alter_table(
TABLE *altered_table [[maybe_unused]],
Alter_inplace_info *ha_alter_info [[maybe_unused]],
const dd::Table *old_table_def [[maybe_unused]],
dd::Table *new_table_def [[maybe_unused]]) {
return false;
}
/**
Alter the table structure in-place with operations specified using
HA_ALTER_FLAGS and Alter_inplace_info. The level of concurrency allowed
during this operation depends on the return value from
check_if_supported_inplace_alter().
@note Should be no-op for instant algorithm.
@note Storage engines are responsible for reporting any errors by
calling my_error()/print_error()
@note If this function reports error, commit_inplace_alter_table()
will be called with commit= false.
@param altered_table TABLE object for new version of table.
@param ha_alter_info Structure describing changes to be done
by ALTER TABLE and holding data used
during in-place alter.
@param old_table_def dd::Table object describing old version of
the table.
@param new_table_def dd::Table object for the new version of the
table. Can be adjusted by this call if SE
supports atomic DDL. These changes to the
table definition will be persisted in the
data-dictionary at statement commit time.
@retval true Error
@retval false Success
*/
virtual bool inplace_alter_table(TABLE *altered_table [[maybe_unused]],
Alter_inplace_info *ha_alter_info
[[maybe_unused]],
const dd::Table *old_table_def
[[maybe_unused]],
dd::Table *new_table_def [[maybe_unused]]) {
return false;
}
/**
Commit or rollback the changes made during prepare_inplace_alter_table()
and inplace_alter_table() inside the storage engine.
Note that in case of rollback the allowed level of concurrency during
this operation will be the same as for inplace_alter_table() and thus
might be higher than during prepare_inplace_alter_table(). (For example,
concurrent writes were blocked during prepare, but might not be during
rollback).
@note This is the place where SE changes happen for instant algorithm.
@note For storage engines supporting atomic DDL this method should only
prepare for the commit but do not finalize it. Real commit should happen
later when the whole statement is committed. Also in some situations
statement might be rolled back after call to commit_inplace_alter_table()
for such storage engines. In the latter special case SE might require call
to handlerton::dict_cache_reset() in order to invalidate its internal
table definition cache after rollback.
@note Storage engines are responsible for reporting any errors by
calling my_error()/print_error()
@note If this function with commit= true reports error, it will be called
again with commit= false.
@note In case of partitioning, this function might be called for rollback
without prepare_inplace_alter_table() having been called first.
Also partitioned tables sets ha_alter_info->group_commit_ctx to a NULL
terminated array of the partitions handlers and if all of them are
committed as one, then group_commit_ctx should be set to NULL to indicate
to the partitioning handler that all partitions handlers are committed.
@see prepare_inplace_alter_table().
@param altered_table TABLE object for new version of table.
@param ha_alter_info Structure describing changes to be done
by ALTER TABLE and holding data used
during in-place alter.
@param commit True => Commit, False => Rollback.
@param old_table_def dd::Table object describing old version of
the table.
@param new_table_def dd::Table object for the new version of the
table. Can be adjusted by this call if SE
supports atomic DDL. These changes to the
table definition will be persisted in the
data-dictionary at statement commit time.
@retval true Error
@retval false Success
*/
virtual bool commit_inplace_alter_table(TABLE *altered_table [[maybe_unused]],
Alter_inplace_info *ha_alter_info
[[maybe_unused]],
bool commit [[maybe_unused]],
const dd::Table *old_table_def
[[maybe_unused]],
dd::Table *new_table_def
[[maybe_unused]]) {
/* Nothing to commit/rollback, mark all handlers committed! */
ha_alter_info->group_commit_ctx = nullptr;
return false;
}
/**
Notify the storage engine that the table definition has been updated.
@param ha_alter_info Structure describing changes done by
ALTER TABLE and holding data used
during in-place alter.
@note No errors are allowed during notify_table_changed().
@note For storage engines supporting atomic DDL this method is invoked
after the whole ALTER TABLE is completed and committed.
Particularly this means that for ALTER TABLE statements with RENAME
clause TABLE/handler object used for invoking this method will be
associated with new table name. If storage engine needs to know
the old schema and table name in this method for some reason it
has to use ha_alter_info object to figure it out.
*/
virtual void notify_table_changed(Alter_inplace_info *ha_alter_info
[[maybe_unused]]) {}
public:
/* End of On-line/in-place ALTER TABLE interface. */
/**
use_hidden_primary_key() is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key
*/
virtual void use_hidden_primary_key();
protected:
/* Service methods for use by storage engines. */
void ha_statistic_increment(ulonglong System_status_var::*offset) const;
THD *ha_thd() const;
/**
Acquire the instrumented table information from a table share.
@param share a table share
@return an instrumented table share, or NULL.
*/
PSI_table_share *ha_table_share_psi(const TABLE_SHARE *share) const;
/**
Default rename_table() and delete_table() rename/delete files with a
given name and extensions from handlerton::file_extensions.
These methods can be overridden, but their default implementation
provide useful functionality.
@param [in] from Path for the old table name.
@param [in] to Path for the new table name.
@param [in] from_table_def Old version of definition for table
being renamed (i.e. prior to rename).
@param [in,out] to_table_def New version of definition for table
being renamed. Storage engines which
support atomic DDL (i.e. having
HTON_SUPPORTS_ATOMIC_DDL flag set)
are allowed to adjust this object.
@retval >0 Error.
@retval 0 Success.
*/
virtual int rename_table(const char *from, const char *to,
const dd::Table *from_table_def,
dd::Table *to_table_def);
/**
Delete a table.
Used to delete a table. By the time delete_table() has been called all
opened references to this table will have been closed (and your globally
shared references released. The variable name will just be the name of
the table. You will need to remove any files you have created at this
point. Called for base as well as temporary tables.
@param name Full path of table name.
@param table_def dd::Table describing table being deleted
(can be NULL for temporary tables created
by optimizer).
@return Zero on success, nonzero otherwise.
*/
virtual int delete_table(const char *name, const dd::Table *table_def);
private:
/* Private helpers */
void mark_trx_read_write();
/*
Low-level primitives for storage engines. These should be
overridden by the storage engine class. To call these methods, use
the corresponding 'ha_*' method above.
*/
virtual int open(const char *name, int mode, uint test_if_locked,
const dd::Table *table_def) = 0;
virtual int close(void) = 0;
virtual int index_init(uint idx, bool sorted [[maybe_unused]]) {
active_index = idx;
return 0;
}
virtual int index_end() {
active_index = MAX_KEY;
return 0;
}
/**
rnd_init() can be called two times without rnd_end() in between
(it only makes sense if scan=1).
then the second call should prepare for the new table scan (e.g
if rnd_init allocates the cursor, second call should position it
to the start of the table, no need to deallocate and allocate it again
*/
virtual int rnd_init(bool scan) = 0;
virtual int rnd_end() { return 0; }
/**
Write a row.
write_row() inserts a row. buf is a byte array of data, normally
record[0].
You can use the field information to extract the data from the native byte
array type.
Example of this would be:
for (Field **field=table->field ; *field ; field++)
{
...
}
@param buf Buffer to write from.
@return Operation status.
@retval 0 Success.
@retval != 0 Error code.
*/
virtual int write_row(uchar *buf [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
/**
Update a single row.
Note: If HA_ERR_FOUND_DUPP_KEY is returned, the handler must read
all columns of the row so MySQL can create an error message. If
the columns required for the error message are not read, the error
message will contain garbage.
*/
virtual int update_row(const uchar *old_data [[maybe_unused]],
uchar *new_data [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
virtual int delete_row(const uchar *buf [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
/**
Reset state of file to after 'open'.
This function is called after every statement for all tables used
by that statement.
*/
virtual int reset() { return 0; }
virtual Table_flags table_flags(void) const = 0;
/**
Is not invoked for non-transactional temporary tables.
Tells the storage engine that we intend to read or write data
from the table. This call is prefixed with a call to handler::store_lock()
and is invoked only for those handler instances that stored the lock.
Calls to @c rnd_init / @c index_init are prefixed with this call. When table
IO is complete, we call @code external_lock(F_UNLCK) @endcode.
A storage engine writer should expect that each call to
@code ::external_lock(F_[RD|WR]LOCK @endcode is followed by a call to
@code ::external_lock(F_UNLCK) @endcode. If it is not, it is a bug in MySQL.
The name and signature originate from the first implementation
in MyISAM, which would call @c fcntl to set/clear an advisory
lock on the data file in this method.
Originally this method was used to set locks on file level to enable
several MySQL Servers to work on the same data. For transactional
engines it has been "abused" to also mean start and end of statements
to enable proper rollback of statements and transactions. When LOCK
TABLES has been issued the start_stmt method takes over the role of
indicating start of statement but in this case there is no end of
statement indicator(?).
Called from lock.cc by lock_external() and unlock_external(). Also called
from sql_table.cc by copy_data_between_tables().
@param thd the current thread
@param lock_type F_RDLCK, F_WRLCK, F_UNLCK
@return non-0 in case of failure, 0 in case of success.
When lock_type is F_UNLCK, the return value is ignored.
*/
virtual int external_lock(THD *thd [[maybe_unused]],
int lock_type [[maybe_unused]]) {
return 0;
}
virtual void release_auto_increment() { return; }
/** admin commands - called from mysql_admin_table */
virtual int check_for_upgrade(HA_CHECK_OPT *) { return 0; }
virtual int check(THD *, HA_CHECK_OPT *) { return HA_ADMIN_NOT_IMPLEMENTED; }
/**
In this method check_opt can be modified
to specify CHECK option to use to call check()
upon the table.
*/
virtual int repair(THD *, HA_CHECK_OPT *) {
assert(!(ha_table_flags() & HA_CAN_REPAIR));
return HA_ADMIN_NOT_IMPLEMENTED;
}
virtual void start_bulk_insert(ha_rows) {}
virtual int end_bulk_insert() { return 0; }
/**
Does this handler want to get a Record_buffer for multi-row reads
via the ha_set_record_buffer() function? And if so, what is the
maximum number of records to allocate space for in the buffer?
Storage engines that support using a Record_buffer should override
this function and return true for scans that could benefit from a
buffer.
@param[out] max_rows gets set to the maximum number of records to
allocate space for in the buffer if the function
returns true
@retval true if the handler would like a Record_buffer
@retval false if the handler does not want a Record_buffer
*/
virtual bool is_record_buffer_wanted(ha_rows *const max_rows) const {
*max_rows = 0;
return false;
}
// Set se_private_id and se_private_data during upgrade
virtual bool upgrade_table(THD *thd [[maybe_unused]],
const char *dbname [[maybe_unused]],
const char *table_name [[maybe_unused]],
dd::Table *dd_table [[maybe_unused]]) {
return false;
}
/** Initialize sampling.
@param[out] scan_ctx A scan context created by this method that has to be
used in sample_next
@param[in] sampling_percentage percentage of records that need to be sampled
@param[in] sampling_seed random seed
@param[in] sampling_method sampling method to be used; currently only
SYSTEM sampling is supported
@param[in] tablesample true if the sampling is for tablesample
@return 0 for success, else failure. */
virtual int sample_init(void *&scan_ctx, double sampling_percentage,
int sampling_seed,
enum_sampling_method sampling_method,
const bool tablesample);
/** Get the next record for sampling.
@param[in] scan_ctx Scan context of the sampling
@param[in] buf buffer to place the read record
@return 0 for success, else failure. */
virtual int sample_next(void *scan_ctx, uchar *buf);
/** End sampling.
@param[in] scan_ctx Scan context of the sampling
@return 0 for success, else failure. */
virtual int sample_end(void *scan_ctx);
/**
* Loads a table into its defined secondary storage engine.
*
* @param table Table opened in primary storage engine. Its read_set tells
* which columns to load.
*
* @return 0 if success, error code otherwise.
*/
virtual int load_table(const TABLE &table [[maybe_unused]]) {
/* purecov: begin inspected */
assert(false);
return HA_ERR_WRONG_COMMAND;
/* purecov: end */
}
/**
* Unloads a table from its defined secondary storage engine.
*
* @param db_name Database name.
* @param table_name Table name.
* @param error_if_not_loaded If true, then errors will be reported by this
* function. If false, no errors will be reported
* (silently fail). This case of false is useful
* during DROP TABLE where a failure to unload
* should not prevent dropping the whole table.
* @return 0 if success, error code otherwise.
*/
virtual int unload_table(const char *db_name [[maybe_unused]],
const char *table_name [[maybe_unused]],
bool error_if_not_loaded [[maybe_unused]]) {
/* purecov: begin inspected */
assert(false);
return HA_ERR_WRONG_COMMAND;
/* purecov: end */
}
protected:
virtual int index_read(uchar *buf [[maybe_unused]],
const uchar *key [[maybe_unused]],
uint key_len [[maybe_unused]],
enum ha_rkey_function find_flag [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
virtual int index_read_last(uchar *buf [[maybe_unused]],
const uchar *key [[maybe_unused]],
uint key_len [[maybe_unused]]) {
set_my_errno(HA_ERR_WRONG_COMMAND);
return HA_ERR_WRONG_COMMAND;
}
public:
/**
This method is similar to update_row, however the handler doesn't need
to execute the updates at this point in time. The handler can be certain
that another call to bulk_update_row will occur OR a call to
exec_bulk_update before the set of updates in this query is concluded.
Note: If HA_ERR_FOUND_DUPP_KEY is returned, the handler must read
all columns of the row so MySQL can create an error message. If
the columns required for the error message are not read, the error
message will contain garbage.
@param old_data Old record
@param new_data New record
@param dup_key_found Number of duplicate keys found
*/
virtual int bulk_update_row(const uchar *old_data [[maybe_unused]],
uchar *new_data [[maybe_unused]],
uint *dup_key_found [[maybe_unused]]) {
assert(false);
return HA_ERR_WRONG_COMMAND;
}
/**
Delete all rows in a table.
This is called both for cases of truncate and for cases where the
optimizer realizes that all rows will be removed as a result of an
SQL statement.
If the handler don't support this, then this function will
return HA_ERR_WRONG_COMMAND and MySQL will delete the rows one
by one.
*/
virtual int delete_all_rows() {
set_my_errno(HA_ERR_WRONG_COMMAND);
return HA_ERR_WRONG_COMMAND;
}
/**
Quickly remove all rows from a table.
@param[in,out] table_def dd::Table object for table being truncated.
@remark This method is responsible for implementing MySQL's TRUNCATE
TABLE statement, which is a DDL operation. As such, a engine
can bypass certain integrity checks and in some cases avoid
fine-grained locking (e.g. row locks) which would normally be
required for a DELETE statement.
@remark Typically, truncate is not used if it can result in integrity
violation. For example, truncate is not used when a foreign
key references the table, but it might be used if foreign key
checks are disabled.
@remark Engine is responsible for resetting the auto-increment counter.
@remark The table is locked in exclusive mode. All open TABLE/handler
instances except the one which is used for truncate() call
are closed.
@note It is assumed that transactional storage engines implementing
this method can revert its effects if transaction is rolled
back (e.g. because we failed to write statement to the binary
log).
@note Changes to dd::Table object done by this method will be saved
to data-dictionary only if storage engine supports atomic DDL
(i.e. has HTON_SUPPORTS_ATOMIC_DDL flag set).
*/
virtual int truncate(dd::Table *table_def [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
virtual int optimize(THD *, HA_CHECK_OPT *) {
return HA_ADMIN_NOT_IMPLEMENTED;
}
virtual int analyze(THD *, HA_CHECK_OPT *) {
return HA_ADMIN_NOT_IMPLEMENTED;
}
/**
@brief Check and repair the table if necessary.
@param thd Thread object
@retval true Error/Not supported
@retval false Success
@note Called if open_table_from_share fails and is_crashed().
*/
virtual bool check_and_repair(THD *thd [[maybe_unused]]) { return true; }
/**
Disable indexes for a while.
@param mode Mode.
@retval 0 Success.
@retval != 0 Error.
*/
virtual int disable_indexes(uint mode [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
/**
Enable indexes again.
@param mode Mode.
@retval 0 Success.
@retval != 0 Error.
*/
virtual int enable_indexes(uint mode [[maybe_unused]]) {
return HA_ERR_WRONG_COMMAND;
}
/**
Discard or import tablespace.
@param [in] discard Indicates whether this is discard operation.
@param [in,out] table_def dd::Table object describing the table
in which tablespace needs to be discarded
or imported. This object can be adjusted by
storage engine if it supports atomic DDL
(i.e. has HTON_SUPPORTS_ATOMIC_DDL flag set).
These changes will be persisted in the
data-dictionary.
@retval 0 Success.
@retval != 0 Error.
*/
virtual int discard_or_import_tablespace(bool discard [[maybe_unused]],
dd::Table *table_def
[[maybe_unused]]) {
set_my_errno(HA_ERR_WRONG_COMMAND);
return HA_ERR_WRONG_COMMAND;
}
virtual void drop_table(const char *name);
/**
Create table (implementation).
@param [in] name Table name.
@param [in] form TABLE object describing the table to be
created.
@param [in] info HA_CREATE_INFO describing table.
@param [in,out] table_def dd::Table object describing the table
to be created. This object can be
adjusted by storage engine if it
supports atomic DDL (i.e. has
HTON_SUPPORTS_ATOMIC_DDL flag set).
These changes will be persisted in the
data-dictionary. Can be NULL for
temporary tables created by optimizer.
@retval 0 Success.
@retval non-0 Error.
*/
virtual int create(const char *name, TABLE *form, HA_CREATE_INFO *info,
dd::Table *table_def) = 0;
virtual bool get_se_private_data(dd::Table *dd_table [[maybe_unused]],
bool reset [[maybe_unused]]) {
return false;
}
/**
Adjust definition of table to be created by adding implicit columns
and indexes necessary for the storage engine.
@param [in] create_info HA_CREATE_INFO describing the table.
@param [in] create_list List of columns in the table.
@param [in] key_info Array of KEY objects describing table
indexes.
@param [in] key_count Number of indexes in the table.
@param [in,out] table_obj dd::Table object describing the table
to be created. Implicit columns and
indexes are to be added to this object.
Adjusted table description will be
saved into the data-dictionary.
@retval 0 Success.
@retval non-0 Error.
*/
virtual int get_extra_columns_and_keys(
const HA_CREATE_INFO *create_info [[maybe_unused]],
const List<Create_field> *create_list [[maybe_unused]],
const KEY *key_info [[maybe_unused]], uint key_count [[maybe_unused]],
dd::Table *table_obj [[maybe_unused]]) {
return 0;
}
virtual bool set_ha_share_ref(Handler_share **arg_ha_share) {
ha_share = arg_ha_share;
return false;
}
void set_ha_table(TABLE *table_arg) { table = table_arg; }
int get_lock_type() const { return m_lock_type; }
/**
Callback function that will be called by my_prepare_gcolumn_template
once the table has been opened.
*/
typedef void (*my_gcolumn_template_callback_t)(const TABLE *, void *);
static bool my_prepare_gcolumn_template(THD *thd, const char *db_name,
const char *table_name,
my_gcolumn_template_callback_t myc,
void *ib_table);
static bool my_eval_gcolumn_expr_with_open(THD *thd, const char *db_name,
const char *table_name,
const MY_BITMAP *const fields,
uchar *record,
const char **mv_data_ptr,
ulong *mv_length);
/**
Callback for computing generated column values.
Storage engines that need to have virtual column values for a row
can use this function to get the values computed. The storage
engine must have filled in the values for the base columns that
the virtual columns depend on.
@param thd thread handle
@param table table object
@param fields bitmap of field index of evaluated generated
column
@param[in,out] record buff of base columns generated column depends.
After calling this function, it will be
used to return the value of the generated
columns.
@param[out] mv_data_ptr When given (not null) and the field
needs to be calculated is a typed array field, it
will contain pointer to field's calculated value.
@param[out] mv_length Length of the data above
@retval true in case of error
@retval false on success
*/
static bool my_eval_gcolumn_expr(THD *thd, TABLE *table,
const MY_BITMAP *const fields, uchar *record,
const char **mv_data_ptr, ulong *mv_length);
/* This must be implemented if the handlerton's partition_flags() is set. */
virtual Partition_handler *get_partition_handler() { return nullptr; }
/**
Set se_private_id and se_private_data during upgrade
@param thd Pointer of THD
@param dbname Database name
@param table_name Table name
@param dd_table dd::Table for the table
@param table_arg TABLE object for the table.
@return Operation status
@retval false Success
@retval true Error
*/
bool ha_upgrade_table(THD *thd, const char *dbname, const char *table_name,
dd::Table *dd_table, TABLE *table_arg);
/**
Store a pointer to the handler of the primary table that
corresponds to the secondary table in this handler.
*/
void ha_set_primary_handler(handler *primary_handler);
/**
Get a pointer to a handler for the table in the primary storage
engine, if this handler is for a table in a secondary storage
engine.
*/
handler *ha_get_primary_handler() const { return m_primary_handler; }
/**
Return max limits for a single set of multi-valued keys
@param[out] num_keys number of keys to store
@param[out] keys_length total length of keys, bytes
*/
void ha_mv_key_capacity(uint *num_keys, size_t *keys_length) const {
return mv_key_capacity(num_keys, keys_length);
}
/**
Propagates the secondary storage engine offload failure reason for a query
to the external engine when the offloaded query fails in the secondary
storage engine.
*/
virtual void set_external_table_offload_error(const char * /*reason*/) {}
/**
Identifies and throws the propagated external engine query offload or exec
failure reason given by the external engine handler.
*/
virtual void external_table_offload_error() const {}
private:
/**
Engine-specific function for ha_can_store_mv_keys().
Dummy function. SE's overloaded method is used instead.
*/
/* purecov: begin inspected */
virtual void mv_key_capacity(uint *num_keys, size_t *keys_length) const {
*num_keys = 0;
*keys_length = 0;
}
/* purecov: end */
/**
Filter duplicate records when multi-valued index is used for retrieval
@returns
true duplicate, such row id was already seen
false row id is seen for the first time
*/
bool filter_dup_records();
protected:
Handler_share *get_ha_share_ptr();
void set_ha_share_ptr(Handler_share *arg_ha_share);
void lock_shared_ha_data();
void unlock_shared_ha_data();
friend class DsMrr_impl;
};
/* Temporary Table handle for opening uncached table */
class Temp_table_handle {
public:
Temp_table_handle() : table(nullptr) {}
/** Open the table handler
@param[in] thd Thread object
@param[in] db_name Database name
@param[in] table_name Table name
@return table object or nullptr */
TABLE *open(THD *thd, const char *db_name, const char *table_name);
~Temp_table_handle();
private:
TABLE *table;
};
/**
Function identifies any old data type present in table.
This function was handler::check_old_types().
Function is not part of SE API. It is now converted to
auxiliary standalone function.
@param[in] table TABLE object
@param[in] check_temporal_upgrade Check if temporal upgrade is needed
@retval 0 ON SUCCESS
@retval error code ON FAILURE
*/
int check_table_for_old_types(const TABLE *table, bool check_temporal_upgrade);
/*
A Disk-Sweep MRR interface implementation
This implementation makes range (and, in the future, 'ref') scans to read
table rows in disk sweeps.
Currently it is used by MyISAM and InnoDB. Potentially it can be used with
any table handler that has non-clustered indexes and on-disk rows.
*/
class DsMrr_impl {
public:
DsMrr_impl(handler *owner) : h(owner), table(nullptr), h2(nullptr) {}
~DsMrr_impl() {
/*
If ha_reset() has not been called then the h2 dialog might still
exist. This must be closed and deleted (this is the case for
internally created temporary tables).
*/
if (h2) reset();
assert(h2 == nullptr);
}
private:
/*
The "owner" handler object (the one that calls dsmrr_XXX functions.
It is used to retrieve full table rows by calling rnd_pos().
*/
handler *const h;
TABLE *table; /* Always equal to h->table */
/* Secondary handler object. It is used for scanning the index */
handler *h2;
/* Buffer to store rowids, or (rowid, range_id) pairs */
uchar *rowids_buf;
uchar *rowids_buf_cur; /* Current position when reading/writing */
uchar *rowids_buf_last; /* When reading: end of used buffer space */
uchar *rowids_buf_end; /* End of the buffer */
bool dsmrr_eof; /* true <=> We have reached EOF when reading index tuples */
/* true <=> need range association, buffer holds {rowid, range_id} pairs */
bool is_mrr_assoc;
bool use_default_impl; /* true <=> shortcut all calls to default MRR impl */
public:
/**
Initialize the DsMrr_impl object.
This object is used for both doing default MRR scans and DS-MRR scans.
This function just initializes the object. To do a DS-MRR scan,
this must also be initialized by calling dsmrr_init().
@param table_arg pointer to the TABLE that owns the handler
*/
void init(TABLE *table_arg) {
assert(table_arg != nullptr);
table = table_arg;
}
int dsmrr_init(RANGE_SEQ_IF *seq_funcs, void *seq_init_param, uint n_ranges,
uint mode, HANDLER_BUFFER *buf);
void dsmrr_close();
/**
Resets the DS-MRR object to the state it had after being initialized.
If there is an open scan then it will be closed.
This function should be called by handler::ha_reset() which is called
when a statement is completed in order to make the handler object ready
for re-use by a different statement.
*/
void reset();
int dsmrr_fill_buffer();
int dsmrr_next(char **range_info);
ha_rows dsmrr_info(uint keyno, uint n_ranges, uint keys, uint *bufsz,
uint *flags, Cost_estimate *cost);
ha_rows dsmrr_info_const(uint keyno, RANGE_SEQ_IF *seq, void *seq_init_param,
uint n_ranges, uint *bufsz, uint *flags,
Cost_estimate *cost);
private:
bool choose_mrr_impl(uint keyno, ha_rows rows, uint *flags, uint *bufsz,
Cost_estimate *cost);
bool get_disk_sweep_mrr_cost(uint keynr, ha_rows rows, uint flags,
uint *buffer_size, Cost_estimate *cost);
};
/* lookups */
handlerton *ha_default_handlerton(THD *thd);
handlerton *ha_default_temp_handlerton(THD *thd);
/**
Resolve handlerton plugin by name, without checking for "DEFAULT" or
HTON_NOT_USER_SELECTABLE.
@param thd Thread context.
@param name Plugin name.
@return plugin or NULL if not found.
*/
plugin_ref ha_resolve_by_name_raw(THD *thd, const LEX_CSTRING &name);
plugin_ref ha_resolve_by_name(THD *thd, const LEX_CSTRING *name,
bool is_temp_table);
plugin_ref ha_lock_engine(THD *thd, const handlerton *hton);
handlerton *ha_resolve_by_legacy_type(THD *thd, enum legacy_db_type db_type);
handler *get_new_handler(TABLE_SHARE *share, bool partitioned, MEM_ROOT *alloc,
handlerton *db_type);
handlerton *ha_checktype(THD *thd, enum legacy_db_type database_type,
bool no_substitute, bool report_error);
bool ha_secondary_engine_supports_ddl(
THD *thd, const LEX_CSTRING &secondary_engine) noexcept;
/**
Get default handlerton, if handler supplied is null.
@param thd Thread context.
@param hton The handlerton passed.
@returns pointer to handlerton.
*/
inline handlerton *get_default_handlerton(THD *thd, handlerton *hton) {
if (!hton) {
hton = ha_checktype(thd, DB_TYPE_UNKNOWN, false, false);
assert(hton);
}
return hton;
}
static inline enum legacy_db_type ha_legacy_type(const handlerton *db_type) {
return (db_type == nullptr) ? DB_TYPE_UNKNOWN : db_type->db_type;
}
const char *ha_resolve_storage_engine_name(const handlerton *db_type);
static inline bool ha_check_storage_engine_flag(const handlerton *db_type,
uint32 flag) {
return db_type == nullptr ? false : (db_type->flags & flag);
}
/**
Predicate to determine if a storage engine, represented by a handlerton*, is
enabled.
@note "Enabled" in this context refers only the state of the handlerton
object, and does not consider the disabled_storage_engines system variable.
This leads to the very counter-intuitive and confusing situation that it is
possible for a storage engine to be enabled, but at the same time also be
disabled.
*/
inline bool ha_storage_engine_is_enabled(const handlerton *db_type) {
return (db_type && db_type->create) ? (db_type->state == SHOW_OPTION_YES)
: false;
}
/* basic stuff */
int ha_init_errors(void);
int ha_init(void);
void ha_end();
int ha_initialize_handlerton(st_plugin_int *plugin);
int ha_finalize_handlerton(st_plugin_int *plugin);
TYPELIB *ha_known_exts();
int ha_panic(enum ha_panic_function flag);
void ha_close_connection(THD *thd);
void ha_kill_connection(THD *thd);
/** Invoke handlerton::pre_dd_shutdown() on every storage engine plugin. */
void ha_pre_dd_shutdown(void);
/**
Flush the log(s) of storage engine(s).
@param binlog_group_flush true if we got invoked by binlog group
commit during flush stage, false in other cases.
@retval false Succeed
@retval true Error
*/
bool ha_flush_logs(bool binlog_group_flush = false);
void ha_drop_database(char *path);
int ha_create_table(THD *thd, const char *path, const char *db,
const char *table_name, HA_CREATE_INFO *create_info,
bool update_create_info, bool is_temp_table,
dd::Table *table_def);
int ha_delete_table(THD *thd, handlerton *db_type, const char *path,
const char *db, const char *alias,
const dd::Table *table_def, bool generate_warning);
bool ha_check_reserved_db_name(const char *name);
/* statistics and info */
bool ha_show_status(THD *thd, handlerton *db_type, enum ha_stat_type stat);
typedef bool Log_func(THD *, TABLE *, bool, const uchar *, const uchar *);
int binlog_log_row(TABLE *table, const uchar *before_record,
const uchar *after_record, Log_func *log_func);
/* discovery */
int ha_create_table_from_engine(THD *thd, const char *db, const char *name);
bool ha_check_if_table_exists(THD *thd, const char *db, const char *name,
bool *exists);
int ha_find_files(THD *thd, const char *db, const char *path, const char *wild,
bool dir, List<LEX_STRING> *files);
int ha_table_exists_in_engine(THD *thd, const char *db, const char *name);
bool ha_check_if_supported_system_table(handlerton *hton, const char *db,
const char *table_name);
bool ha_rm_tmp_tables(THD *thd, List<LEX_STRING> *files);
bool default_rm_tmp_tables(handlerton *hton, THD *thd, List<LEX_STRING> *files);
/* key cache */
int ha_init_key_cache(std::string_view name, KEY_CACHE *key_cache);
int ha_resize_key_cache(KEY_CACHE *key_cache);
int ha_change_key_cache(KEY_CACHE *old_key_cache, KEY_CACHE *new_key_cache);
/* transactions: interface to handlerton functions */
int ha_start_consistent_snapshot(THD *thd);
int ha_commit_trans(THD *thd, bool all, bool ignore_global_read_lock = false);
int ha_commit_attachable(THD *thd);
int ha_rollback_trans(THD *thd, bool all);
/**
Stage of the recovery process where information is collected from the
storage engines (SE), merged with information from the transaction
coordinator (TC) and transactions states are determined and enforced.
Implemented heuristics is as follows:
1. The `commit_list` parameter contains the set of internally coordinated
transactions that the TC ensures were marked as committed.
2. The `xa_state_list` parameter contains the list of externally
coordinated transactions and their states, as recorded by the TC.
3. For each SE:
a. Collect list of transactions found in `PREPARED_IN_TC` state in the
SE and merge it with the information collected from the TC, in
`xa_state_list`.
b. Retrieve the list of transactions found in prepared state in the
SE.
c. For each internally coordinated transactions found in prepared
state:
1. If the transaction is found in `commit_list`, commit it.
2. If the transaction is NOT found in `commit_list` but
`tc_heuristic_recover = TC_HEURISTIC_RECOVER_COMMIT`, commit it.
3. Otherwise, roll it back.
d. For each externally coordinated transactions found in prepared
state:
1. If the transaction isn't found in `xa_state_list`, roll it back.
2. If the transaction is found in `xa_state_list` in `COMMITTED`
state, commit it.
3. If the transaction is found in `xa_state_list` in `ROLLEDBACK`
state, roll it back.
4. If the transaction is found in `xa_state_list` in `PREPARED`
state, ensure that the transaction state in the SE is
`PREPARED_IN_TC`.
@param commit_list Set of XIDs of internally coordinated transactions
found as been committed in the transaction coordinator
state.
@param xa_state_list Map between XIDs and states of externally
coordinated transactions as found in the internal
transaction coordinator state.
@return 0 if recovery was successful, non-zero otherwise.
*/
int ha_recover(Xid_commit_list *commit_list = nullptr,
Xa_state_list *xa_state_list = nullptr);
/**
Perform SE-specific cleanup after recovery of transactions.
@note SE supporting atomic DDL can use this method to perform
post-DDL actions for DDL statements which were committed
or rolled back during recovery stage.
*/
void ha_post_recover();
/*
transactions: interface to low-level handlerton functions. These are
intended to be used by the transaction coordinators to
commit/prepare/rollback transactions in the engines.
*/
int ha_commit_low(THD *thd, bool all, bool run_after_commit = true);
/**
Prepares the underlying transaction of the THD session object parameter
in the storage engines that participate in the transaction.
In case of failure, an error will be emitted by the function in the case
of internally coordinated transactions. In the case of externally
coordinated transactions (XA), the error treatment must follow the
XA/Open specification and is handled by the `Sql_cmd_xa_prepare` class.
@param thd The THD session object holding the transaction to be prepared.
@param all Whether or not the prepare regards a full transaction or the
statement being executed..
@return 0 if the transaction was successfully prepared, non-zero
otherwise.
*/
int ha_prepare_low(THD *thd, bool all);
int ha_rollback_low(THD *thd, bool all);
/* transactions: these functions never call handlerton functions directly */
int ha_enable_transaction(THD *thd, bool on);
/* savepoints */
int ha_rollback_to_savepoint(THD *thd, SAVEPOINT *sv);
bool ha_rollback_to_savepoint_can_release_mdl(THD *thd);
int ha_savepoint(THD *thd, SAVEPOINT *sv);
int ha_release_savepoint(THD *thd, SAVEPOINT *sv);
/* these are called by storage engines */
void trans_register_ha(THD *thd, bool all, handlerton *ht,
const ulonglong *trxid);
int ha_reset_logs(THD *thd);
/**
Inform storage engine(s) that a binary log file will be purged and any
references to it should be removed.
The function is called for all purged files, regardless if it is an explicit
PURGE BINARY LOGS statement, or an automatic purge performed by the server.
@note Since function is called with the LOCK_index mutex held the work
performed in this callback should be kept at minimum. One way to defer work is
to schedule work and use the `ha_binlog_index_purge_wait` callback to wait for
completion.
@param thd Thread handle of session purging file. The nullptr value indicates
that purge is done at server startup.
@param file Name of file being purged.
@return Always 0, return value are ignored by caller.
*/
int ha_binlog_index_purge_file(THD *thd, const char *file);
/**
Request the storage engine to complete any operations that were initiated
by `ha_binlog_index_purge_file` and which need to complete
before PURGE BINARY LOGS completes.
The function is called only from PURGE BINARY LOGS. Each PURGE BINARY LOGS
statement will result in 0, 1 or more calls to `ha_binlog_index_purge_file`,
followed by exactly 1 call to `ha_binlog_index_purge_wait`.
@note This function is called without LOCK_index mutex held and thus any
waiting performed will only affect the current session.
@param thd Thread handle of session.
*/
void ha_binlog_index_purge_wait(THD *thd);
void ha_reset_slave(THD *thd);
void ha_binlog_log_query(THD *thd, handlerton *db_type,
enum_binlog_command binlog_command, const char *query,
size_t query_length, const char *db,
const char *table_name);
void ha_acl_notify(THD *thd, class Acl_change_notification *);
void ha_binlog_wait(THD *thd);
/* It is required by basic binlog features on both MySQL server and libmysqld */
int ha_binlog_end(THD *thd);
const char *get_canonical_filename(handler *file, const char *path,
char *tmp_path);
const char *table_case_name(const HA_CREATE_INFO *info, const char *name);
void print_keydup_error(TABLE *table, KEY *key, const char *msg, myf errflag,
const char *org_table_name);
void print_keydup_error(TABLE *table, KEY *key, myf errflag,
const char *org_table_name);
inline void print_keydup_error(TABLE *table, KEY *key, const char *msg,
myf errflag) {
print_keydup_error(table, key, msg, errflag, nullptr);
}
inline void print_keydup_error(TABLE *table, KEY *key, myf errflag) {
print_keydup_error(table, key, errflag, nullptr);
}
bool ha_is_storage_engine_disabled(handlerton *se_engine);
bool ha_notify_exclusive_mdl(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type,
bool *victimized);
bool ha_notify_table_ddl(THD *thd, const MDL_key *mdl_key,
ha_notification_type notification_type,
ha_ddl_type ddl_type, const char *old_db_name,
const char *old_table_name, const char *new_db_name,
const char *new_table_name);
std::pair<int, bool> commit_owned_gtids(THD *thd, bool all);
bool set_tx_isolation(THD *thd, enum_tx_isolation tx_isolation, bool one_shot);
bool is_index_access_error(int error);
/*
This class is used by INFORMATION_SCHEMA.FILES to read SE specific
tablespace dynamic metadata. Some member like m_type and id, is not
really dynamic, but as this information is not stored in data dictionary
in a generic format and still is SE specific Some member like m_type and
id, is not really dynamic, but as this information is not stored in data
dictionary in a generic format and still needs SE specific decision, we
are requesting the same from SE.
*/
class ha_tablespace_statistics {
public:
ha_tablespace_statistics()
: m_id(0),
m_logfile_group_number(~0ULL),
m_free_extents(0),
m_total_extents(0),
m_extent_size(0),
m_initial_size(0),
m_maximum_size(0),
m_autoextend_size(0),
m_version(~0ULL),
m_data_free(0) {}
ulonglong m_id;
dd::String_type m_type;
dd::String_type m_logfile_group_name; // NDB only
ulonglong m_logfile_group_number; // NDB only
ulonglong m_free_extents;
ulonglong m_total_extents;
ulonglong m_extent_size;
ulonglong m_initial_size;
ulonglong m_maximum_size;
ulonglong m_autoextend_size;
ulonglong m_version; // NDB only
dd::String_type m_row_format; // NDB only
ulonglong m_data_free; // InnoDB
dd::String_type m_status;
dd::String_type m_extra; // NDB only
};
#endif /* HANDLER_INCLUDED */
|