sys/dev/raid/mpr/mpr.c
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 | /*- * Copyright (c) 2009 Yahoo! Inc. * Copyright (c) 2011-2015 LSI Corp. * Copyright (c) 2013-2016 Avago Technologies * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * Avago Technologies (LSI) MPT-Fusion Host Adapter FreeBSD * * $FreeBSD: head/sys/dev/mpr/mpr.c 330789 2018-03-12 05:02:22Z scottl $ */ /* Communications core for Avago Technologies (LSI) MPT3 */ /* TODO Move headers to mprvar */ #include <sys/types.h> #include <sys/param.h> #include <sys/systm.h> #include <sys/kernel.h> #include <sys/lock.h> #include <sys/module.h> #include <sys/bus.h> #include <sys/conf.h> #include <sys/bio.h> #include <sys/malloc.h> #include <sys/uio.h> #include <sys/sysctl.h> #include <sys/queue.h> #include <sys/kthread.h> #include <sys/taskqueue.h> #include <sys/endian.h> #include <sys/eventhandler.h> #include <sys/sbuf.h> #include <sys/caps.h> #include <sys/rman.h> #include <sys/proc.h> #include <bus/pci/pcivar.h> #include <bus/cam/cam.h> #include <bus/cam/cam_ccb.h> #include <bus/cam/scsi/scsi_all.h> #include <dev/raid/mpr/mpi/mpi2_type.h> #include <dev/raid/mpr/mpi/mpi2.h> #include <dev/raid/mpr/mpi/mpi2_ioc.h> #include <dev/raid/mpr/mpi/mpi2_sas.h> #include <dev/raid/mpr/mpi/mpi2_pci.h> #include <dev/raid/mpr/mpi/mpi2_cnfg.h> #include <dev/raid/mpr/mpi/mpi2_init.h> #include <dev/raid/mpr/mpi/mpi2_tool.h> #include <dev/raid/mpr/mpr_ioctl.h> #include <dev/raid/mpr/mprvar.h> #include <dev/raid/mpr/mpr_table.h> #include <dev/raid/mpr/mpr_sas.h> static int mpr_diag_reset(struct mpr_softc *sc, int sleep_flag); static int mpr_init_queues(struct mpr_softc *sc); static void mpr_resize_queues(struct mpr_softc *sc); static int mpr_message_unit_reset(struct mpr_softc *sc, int sleep_flag); static int mpr_transition_operational(struct mpr_softc *sc); static int mpr_iocfacts_allocate(struct mpr_softc *sc, uint8_t attaching); static void mpr_iocfacts_free(struct mpr_softc *sc); static void mpr_startup(void *arg); static int mpr_send_iocinit(struct mpr_softc *sc); static int mpr_alloc_queues(struct mpr_softc *sc); static int mpr_alloc_hw_queues(struct mpr_softc *sc); static int mpr_alloc_replies(struct mpr_softc *sc); static int mpr_alloc_requests(struct mpr_softc *sc); static int mpr_alloc_nvme_prp_pages(struct mpr_softc *sc); static int mpr_attach_log(struct mpr_softc *sc); static __inline void mpr_complete_command(struct mpr_softc *sc, struct mpr_command *cm); static void mpr_dispatch_event(struct mpr_softc *sc, uintptr_t data, MPI2_EVENT_NOTIFICATION_REPLY *reply); static void mpr_config_complete(struct mpr_softc *sc, struct mpr_command *cm); static void mpr_periodic(void *); static int mpr_reregister_events(struct mpr_softc *sc); static void mpr_enqueue_request(struct mpr_softc *sc, struct mpr_command *cm); static int mpr_get_iocfacts(struct mpr_softc *sc, MPI2_IOC_FACTS_REPLY *facts); static int mpr_wait_db_ack(struct mpr_softc *sc, int timeout, int sleep_flag); static int mpr_dump_reqs(SYSCTL_HANDLER_ARGS); SYSCTL_NODE(_hw, OID_AUTO, mpr, CTLFLAG_RD, 0, "MPR Driver Parameters"); MALLOC_DEFINE(M_MPR, "mpr", "mpr driver memory"); /* * Do a "Diagnostic Reset" aka a hard reset. This should get the chip out of * any state and back to its initialization state machine. */ static char mpt2_reset_magic[] = { 0x00, 0x0f, 0x04, 0x0b, 0x02, 0x07, 0x0d }; /* * Added this union to smoothly convert le64toh cm->cm_desc.Words. * Compiler only supports uint64_t to be passed as an argument. * Otherwise it will throw this error: * "aggregate value used where an integer was expected" */ typedef union _reply_descriptor { u64 word; struct { u32 low; u32 high; } u; } reply_descriptor, request_descriptor; /* Rate limit chain-fail messages to 1 per minute */ static struct timeval mpr_chainfail_interval = { 60, 0 }; /* * sleep_flag can be either CAN_SLEEP or NO_SLEEP. * If this function is called from process context, it can sleep * and there is no harm to sleep, in case if this fuction is called * from Interrupt handler, we can not sleep and need NO_SLEEP flag set. * based on sleep flags driver will call either msleep, pause or DELAY. * msleep and pause are of same variant, but pause is used when mpr_mtx * is not hold by driver. */ static int mpr_diag_reset(struct mpr_softc *sc,int sleep_flag) { uint32_t reg; int i, error, tries = 0; uint8_t first_wait_done = FALSE; mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); /* Clear any pending interrupts */ mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); mpr_dprint(sc, MPR_INIT, "sequence start, sleep_flag=%d\n", sleep_flag); /* Push the magic sequence */ error = ETIMEDOUT; while (tries++ < 20) { for (i = 0; i < sizeof(mpt2_reset_magic); i++) mpr_regwrite(sc, MPI2_WRITE_SEQUENCE_OFFSET, mpt2_reset_magic[i]); /* wait 100 msec */ if (lockowned(&sc->mpr_lock) && sleep_flag == CAN_SLEEP) lksleep(&sc->msleep_fake_chan, &sc->mpr_lock, 0, "mprdiag", hz < 10 ? 1 : hz / 10); else if (sleep_flag == CAN_SLEEP) tsleep(mpr_diag_reset, 0, "mprdiag", hz < 10 ? 1 : hz / 10); else DELAY(100 * 1000); reg = mpr_regread(sc, MPI2_HOST_DIAGNOSTIC_OFFSET); if (reg & MPI2_DIAG_DIAG_WRITE_ENABLE) { error = 0; break; } } if (error) { mpr_dprint(sc, MPR_INIT, "sequence failed, error=%d, exit\n", error); return (error); } /* Send the actual reset. XXX need to refresh the reg? */ reg |= MPI2_DIAG_RESET_ADAPTER; mpr_dprint(sc, MPR_INIT, "sequence success, sending reset, reg= 0x%x\n", reg); mpr_regwrite(sc, MPI2_HOST_DIAGNOSTIC_OFFSET, reg); /* Wait up to 300 seconds in 50ms intervals */ error = ETIMEDOUT; for (i = 0; i < 6000; i++) { /* * Wait 50 msec. If this is the first time through, wait 256 * msec to satisfy Diag Reset timing requirements. */ if (first_wait_done) { if (lockowned(&sc->mpr_lock) && sleep_flag == CAN_SLEEP) lksleep(&sc->msleep_fake_chan, &sc->mpr_lock, 0, "mprdiag", hz < 20 ? 1 : hz / 20); else if (sleep_flag == CAN_SLEEP) tsleep(mpr_diag_reset, 0, "mprdiag", hz < 20 ? 1: hz / 20); else DELAY(50 * 1000); } else { DELAY(256 * 1000); first_wait_done = TRUE; } /* * Check for the RESET_ADAPTER bit to be cleared first, then * wait for the RESET state to be cleared, which takes a little * longer. */ reg = mpr_regread(sc, MPI2_HOST_DIAGNOSTIC_OFFSET); if (reg & MPI2_DIAG_RESET_ADAPTER) { continue; } reg = mpr_regread(sc, MPI2_DOORBELL_OFFSET); if ((reg & MPI2_IOC_STATE_MASK) != MPI2_IOC_STATE_RESET) { error = 0; break; } } if (error) { mpr_dprint(sc, MPR_INIT, "reset failed, error= %d, exit\n", error); return (error); } mpr_regwrite(sc, MPI2_WRITE_SEQUENCE_OFFSET, 0x0); mpr_dprint(sc, MPR_INIT, "diag reset success, exit\n"); return (0); } static int mpr_message_unit_reset(struct mpr_softc *sc, int sleep_flag) { int error; MPR_FUNCTRACE(sc); mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); error = 0; mpr_regwrite(sc, MPI2_DOORBELL_OFFSET, MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET << MPI2_DOORBELL_FUNCTION_SHIFT); if (mpr_wait_db_ack(sc, 5, sleep_flag) != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Doorbell handshake failed\n"); error = ETIMEDOUT; } mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); return (error); } static int mpr_transition_ready(struct mpr_softc *sc) { uint32_t reg, state; int error, tries = 0; int sleep_flags; MPR_FUNCTRACE(sc); /* If we are in attach call, do not sleep */ sleep_flags = (sc->mpr_flags & MPR_FLAGS_ATTACH_DONE) ? CAN_SLEEP : NO_SLEEP; error = 0; mpr_dprint(sc, MPR_INIT, "%s entered, sleep_flags= %d\n", __func__, sleep_flags); while (tries++ < 1200) { reg = mpr_regread(sc, MPI2_DOORBELL_OFFSET); mpr_dprint(sc, MPR_INIT, " Doorbell= 0x%x\n", reg); /* * Ensure the IOC is ready to talk. If it's not, try * resetting it. */ if (reg & MPI2_DOORBELL_USED) { mpr_dprint(sc, MPR_INIT, " Not ready, sending diag " "reset\n"); mpr_diag_reset(sc, sleep_flags); DELAY(50000); continue; } /* Is the adapter owned by another peer? */ if ((reg & MPI2_DOORBELL_WHO_INIT_MASK) == (MPI2_WHOINIT_PCI_PEER << MPI2_DOORBELL_WHO_INIT_SHIFT)) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "IOC is under the " "control of another peer host, aborting " "initialization.\n"); error = ENXIO; break; } state = reg & MPI2_IOC_STATE_MASK; if (state == MPI2_IOC_STATE_READY) { /* Ready to go! */ error = 0; break; } else if (state == MPI2_IOC_STATE_FAULT) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "IOC in fault " "state 0x%x, resetting\n", state & MPI2_DOORBELL_FAULT_CODE_MASK); mpr_diag_reset(sc, sleep_flags); } else if (state == MPI2_IOC_STATE_OPERATIONAL) { /* Need to take ownership */ mpr_message_unit_reset(sc, sleep_flags); } else if (state == MPI2_IOC_STATE_RESET) { /* Wait a bit, IOC might be in transition */ mpr_dprint(sc, MPR_INIT|MPR_FAULT, "IOC in unexpected reset state\n"); } else { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "IOC in unknown state 0x%x\n", state); error = EINVAL; break; } /* Wait 50ms for things to settle down. */ DELAY(50000); } if (error) mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Cannot transition IOC to ready\n"); mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); return (error); } static int mpr_transition_operational(struct mpr_softc *sc) { uint32_t reg, state; int error; MPR_FUNCTRACE(sc); error = 0; reg = mpr_regread(sc, MPI2_DOORBELL_OFFSET); mpr_dprint(sc, MPR_INIT, "%s entered, Doorbell= 0x%x\n", __func__, reg); state = reg & MPI2_IOC_STATE_MASK; if (state != MPI2_IOC_STATE_READY) { mpr_dprint(sc, MPR_INIT, "IOC not ready\n"); if ((error = mpr_transition_ready(sc)) != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "failed to transition ready, exit\n"); return (error); } } error = mpr_send_iocinit(sc); mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); return (error); } static void mpr_resize_queues(struct mpr_softc *sc) { u_int reqcr, prireqcr, maxio, sges_per_frame, chain_seg_size; /* * Size the queues. Since the reply queues always need one free * entry, we'll deduct one reply message here. The LSI documents * suggest instead to add a count to the request queue, but I think * that it's better to deduct from reply queue. */ prireqcr = MAX(1, sc->max_prireqframes); prireqcr = MIN(prireqcr, sc->facts->HighPriorityCredit); reqcr = MAX(2, sc->max_reqframes); reqcr = MIN(reqcr, sc->facts->RequestCredit); sc->num_reqs = prireqcr + reqcr; sc->num_prireqs = prireqcr; sc->num_replies = MIN(sc->max_replyframes + sc->max_evtframes, sc->facts->MaxReplyDescriptorPostQueueDepth) - 1; /* Store the request frame size in bytes rather than as 32bit words */ sc->reqframesz = sc->facts->IOCRequestFrameSize * 4; /* * Gen3 and beyond uses the IOCMaxChainSegmentSize from IOC Facts to * get the size of a Chain Frame. Previous versions use the size as a * Request Frame for the Chain Frame size. If IOCMaxChainSegmentSize * is 0, use the default value. The IOCMaxChainSegmentSize is the * number of 16-byte elelements that can fit in a Chain Frame, which is * the size of an IEEE Simple SGE. */ if (sc->facts->MsgVersion >= MPI2_VERSION_02_05) { chain_seg_size = htole16(sc->facts->IOCMaxChainSegmentSize); if (chain_seg_size == 0) chain_seg_size = MPR_DEFAULT_CHAIN_SEG_SIZE; sc->chain_frame_size = chain_seg_size * MPR_MAX_CHAIN_ELEMENT_SIZE; } else { sc->chain_frame_size = sc->reqframesz; } /* * Max IO Size is Page Size * the following: * ((SGEs per frame - 1 for chain element) * Max Chain Depth) * + 1 for no chain needed in last frame * * If user suggests a Max IO size to use, use the smaller of the * user's value and the calculated value as long as the user's * value is larger than 0. The user's value is in pages. */ sges_per_frame = sc->chain_frame_size/sizeof(MPI2_IEEE_SGE_SIMPLE64)-1; maxio = (sges_per_frame * sc->facts->MaxChainDepth + 1) * PAGE_SIZE; /* * If I/O size limitation requested then use it and pass up to CAM. * If not, use MAXPHYS as an optimization hint, but report HW limit. */ if (sc->max_io_pages > 0) { maxio = min(maxio, sc->max_io_pages * PAGE_SIZE); sc->maxio = maxio; } else { sc->maxio = maxio; maxio = min(maxio, MAXPHYS); } sc->num_chains = (maxio / PAGE_SIZE + sges_per_frame - 2) / sges_per_frame * reqcr; if (sc->max_chains > 0 && sc->max_chains < sc->num_chains) sc->num_chains = sc->max_chains; /* * Figure out the number of MSIx-based queues. If the firmware or * user has done something crazy and not allowed enough credit for * the queues to be useful then don't enable multi-queue. */ if (sc->facts->MaxMSIxVectors < 2) sc->msi_msgs = 1; if (sc->msi_msgs > 1) { sc->msi_msgs = MIN(sc->msi_msgs, ncpus); sc->msi_msgs = MIN(sc->msi_msgs, sc->facts->MaxMSIxVectors); if (sc->num_reqs / sc->msi_msgs < 2) sc->msi_msgs = 1; } mpr_dprint(sc, MPR_INIT, "Sized queues to q=%d reqs=%d replies=%d\n", sc->msi_msgs, sc->num_reqs, sc->num_replies); } /* * This is called during attach and when re-initializing due to a Diag Reset. * IOC Facts is used to allocate many of the structures needed by the driver. * If called from attach, de-allocation is not required because the driver has * not allocated any structures yet, but if called from a Diag Reset, previously * allocated structures based on IOC Facts will need to be freed and re- * allocated bases on the latest IOC Facts. */ static int mpr_iocfacts_allocate(struct mpr_softc *sc, uint8_t attaching) { int error; Mpi2IOCFactsReply_t saved_facts; uint8_t saved_mode, reallocating; mpr_dprint(sc, MPR_INIT|MPR_TRACE, "%s entered\n", __func__); /* Save old IOC Facts and then only reallocate if Facts have changed */ if (!attaching) { bcopy(sc->facts, &saved_facts, sizeof(MPI2_IOC_FACTS_REPLY)); } /* * Get IOC Facts. In all cases throughout this function, panic if doing * a re-initialization and only return the error if attaching so the OS * can handle it. */ if ((error = mpr_get_iocfacts(sc, sc->facts)) != 0) { if (attaching) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Failed to get " "IOC Facts with error %d, exit\n", error); return (error); } else { panic("%s failed to get IOC Facts with error %d\n", __func__, error); } } MPR_DPRINT_PAGE(sc, MPR_XINFO, iocfacts, sc->facts); ksnprintf(sc->fw_version, sizeof(sc->fw_version), "%02d.%02d.%02d.%02d", sc->facts->FWVersion.Struct.Major, sc->facts->FWVersion.Struct.Minor, sc->facts->FWVersion.Struct.Unit, sc->facts->FWVersion.Struct.Dev); mpr_dprint(sc, MPR_INFO, "Firmware: %s, Driver: %s\n", sc->fw_version, MPR_DRIVER_VERSION); mpr_dprint(sc, MPR_INFO, "IOCCapabilities: %pb%i\n", "\20" "\3ScsiTaskFull" "\4DiagTrace" "\5SnapBuf" "\6ExtBuf" "\7EEDP" "\10BiDirTarg" "\11Multicast" "\14TransRetry" "\15IR" "\16EventReplay" "\17RaidAccel" "\20MSIXIndex" "\21HostDisc" "\22FastPath" "\23RDPQArray" "\24AtomicReqDesc" "\25PCIeSRIOV", sc->facts->IOCCapabilities); /* * If the chip doesn't support event replay then a hard reset will be * required to trigger a full discovery. Do the reset here then * retransition to Ready. A hard reset might have already been done, * but it doesn't hurt to do it again. Only do this if attaching, not * for a Diag Reset. */ if (attaching && ((sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_EVENT_REPLAY) == 0)) { mpr_dprint(sc, MPR_INIT, "No event replay, resetting\n"); mpr_diag_reset(sc, NO_SLEEP); if ((error = mpr_transition_ready(sc)) != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Failed to " "transition to ready with error %d, exit\n", error); return (error); } } /* * Set flag if IR Firmware is loaded. If the RAID Capability has * changed from the previous IOC Facts, log a warning, but only if * checking this after a Diag Reset and not during attach. */ saved_mode = sc->ir_firmware; if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_INTEGRATED_RAID) sc->ir_firmware = 1; if (!attaching) { if (sc->ir_firmware != saved_mode) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "new IR/IT mode " "in IOC Facts does not match previous mode\n"); } } /* Only deallocate and reallocate if relevant IOC Facts have changed */ reallocating = FALSE; sc->mpr_flags &= ~MPR_FLAGS_REALLOCATED; if ((!attaching) && ((saved_facts.MsgVersion != sc->facts->MsgVersion) || (saved_facts.HeaderVersion != sc->facts->HeaderVersion) || (saved_facts.MaxChainDepth != sc->facts->MaxChainDepth) || (saved_facts.RequestCredit != sc->facts->RequestCredit) || (saved_facts.ProductID != sc->facts->ProductID) || (saved_facts.IOCCapabilities != sc->facts->IOCCapabilities) || (saved_facts.IOCRequestFrameSize != sc->facts->IOCRequestFrameSize) || (saved_facts.IOCMaxChainSegmentSize != sc->facts->IOCMaxChainSegmentSize) || (saved_facts.MaxTargets != sc->facts->MaxTargets) || (saved_facts.MaxSasExpanders != sc->facts->MaxSasExpanders) || (saved_facts.MaxEnclosures != sc->facts->MaxEnclosures) || (saved_facts.HighPriorityCredit != sc->facts->HighPriorityCredit) || (saved_facts.MaxReplyDescriptorPostQueueDepth != sc->facts->MaxReplyDescriptorPostQueueDepth) || (saved_facts.ReplyFrameSize != sc->facts->ReplyFrameSize) || (saved_facts.MaxVolumes != sc->facts->MaxVolumes) || (saved_facts.MaxPersistentEntries != sc->facts->MaxPersistentEntries))) { reallocating = TRUE; /* Record that we reallocated everything */ sc->mpr_flags |= MPR_FLAGS_REALLOCATED; } /* * Some things should be done if attaching or re-allocating after a Diag * Reset, but are not needed after a Diag Reset if the FW has not * changed. */ if (attaching || reallocating) { /* * Check if controller supports FW diag buffers and set flag to * enable each type. */ if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_DIAG_TRACE_BUFFER) sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_TRACE]. enabled = TRUE; if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_SNAPSHOT_BUFFER) sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_SNAPSHOT]. enabled = TRUE; if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_EXTENDED_BUFFER) sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_EXTENDED]. enabled = TRUE; /* * Set flags for some supported items. */ if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_EEDP) sc->eedp_enabled = TRUE; if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_TLR) sc->control_TLR = TRUE; if (sc->facts->IOCCapabilities & MPI26_IOCFACTS_CAPABILITY_ATOMIC_REQ) sc->atomic_desc_capable = TRUE; mpr_resize_queues(sc); /* * Initialize all Tail Queues */ TAILQ_INIT(&sc->req_list); TAILQ_INIT(&sc->high_priority_req_list); TAILQ_INIT(&sc->chain_list); TAILQ_INIT(&sc->prp_page_list); TAILQ_INIT(&sc->tm_list); } /* * If doing a Diag Reset and the FW is significantly different * (reallocating will be set above in IOC Facts comparison), then all * buffers based on the IOC Facts will need to be freed before they are * reallocated. */ if (reallocating) { mpr_iocfacts_free(sc); mprsas_realloc_targets(sc, saved_facts.MaxTargets + saved_facts.MaxVolumes); } /* * Any deallocation has been completed. Now start reallocating * if needed. Will only need to reallocate if attaching or if the new * IOC Facts are different from the previous IOC Facts after a Diag * Reset. Targets have already been allocated above if needed. */ error = 0; while (attaching || reallocating) { if ((error = mpr_alloc_hw_queues(sc)) != 0) break; if ((error = mpr_alloc_replies(sc)) != 0) break; if ((error = mpr_alloc_requests(sc)) != 0) break; if ((error = mpr_alloc_queues(sc)) != 0) break; break; } if (error) { mpr_dprint(sc, MPR_INIT|MPR_ERROR, "Failed to alloc queues with error %d\n", error); mpr_free(sc); return (error); } /* Always initialize the queues */ bzero(sc->free_queue, sc->fqdepth * 4); mpr_init_queues(sc); /* * Always get the chip out of the reset state, but only panic if not * attaching. If attaching and there is an error, that is handled by * the OS. */ error = mpr_transition_operational(sc); if (error != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Failed to " "transition to operational with error %d\n", error); mpr_free(sc); return (error); } /* * Finish the queue initialization. * These are set here instead of in mpr_init_queues() because the * IOC resets these values during the state transition in * mpr_transition_operational(). The free index is set to 1 * because the corresponding index in the IOC is set to 0, and the * IOC treats the queues as full if both are set to the same value. * Hence the reason that the queue can't hold all of the possible * replies. */ sc->replypostindex = 0; mpr_regwrite(sc, MPI2_REPLY_FREE_HOST_INDEX_OFFSET, sc->replyfreeindex); mpr_regwrite(sc, MPI2_REPLY_POST_HOST_INDEX_OFFSET, 0); /* * Attach the subsystems so they can prepare their event masks. * XXX Should be dynamic so that IM/IR and user modules can attach */ error = 0; while (attaching) { mpr_dprint(sc, MPR_INIT, "Attaching subsystems\n"); if ((error = mpr_attach_log(sc)) != 0) break; if ((error = mpr_attach_sas(sc)) != 0) break; if ((error = mpr_attach_user(sc)) != 0) break; break; } if (error) { mpr_dprint(sc, MPR_INIT|MPR_ERROR, "Failed to attach all subsystems: error %d\n", error); mpr_free(sc); return (error); } /* * XXX If the number of MSI-X vectors changes during re-init, this * won't see it and adjust. */ if (attaching && (error = mpr_pci_setup_interrupts(sc)) != 0) { mpr_dprint(sc, MPR_INIT|MPR_ERROR, "Failed to setup interrupts\n"); mpr_free(sc); return (error); } return (error); } /* * This is called if memory is being free (during detach for example) and when * buffers need to be reallocated due to a Diag Reset. */ static void mpr_iocfacts_free(struct mpr_softc *sc) { struct mpr_command *cm; int i; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); if (sc->free_busaddr != 0) bus_dmamap_unload(sc->queues_dmat, sc->queues_map); if (sc->free_queue != NULL) bus_dmamem_free(sc->queues_dmat, sc->free_queue, sc->queues_map); if (sc->queues_dmat != NULL) bus_dma_tag_destroy(sc->queues_dmat); if (sc->chain_frames != NULL) { bus_dmamap_unload(sc->chain_dmat, sc->chain_map); bus_dmamem_free(sc->chain_dmat, sc->chain_frames, sc->chain_map); } if (sc->chain_dmat != NULL) bus_dma_tag_destroy(sc->chain_dmat); if (sc->sense_busaddr != 0) bus_dmamap_unload(sc->sense_dmat, sc->sense_map); if (sc->sense_frames != NULL) bus_dmamem_free(sc->sense_dmat, sc->sense_frames, sc->sense_map); if (sc->sense_dmat != NULL) bus_dma_tag_destroy(sc->sense_dmat); if (sc->prp_page_busaddr != 0) bus_dmamap_unload(sc->prp_page_dmat, sc->prp_page_map); if (sc->prp_pages != NULL) bus_dmamem_free(sc->prp_page_dmat, sc->prp_pages, sc->prp_page_map); if (sc->prp_page_dmat != NULL) bus_dma_tag_destroy(sc->prp_page_dmat); if (sc->reply_busaddr != 0) bus_dmamap_unload(sc->reply_dmat, sc->reply_map); if (sc->reply_frames != NULL) bus_dmamem_free(sc->reply_dmat, sc->reply_frames, sc->reply_map); if (sc->reply_dmat != NULL) bus_dma_tag_destroy(sc->reply_dmat); if (sc->req_busaddr != 0) bus_dmamap_unload(sc->req_dmat, sc->req_map); if (sc->req_frames != NULL) bus_dmamem_free(sc->req_dmat, sc->req_frames, sc->req_map); if (sc->req_dmat != NULL) bus_dma_tag_destroy(sc->req_dmat); if (sc->chains != NULL) kfree(sc->chains, M_MPR); if (sc->prps != NULL) kfree(sc->prps, M_MPR); if (sc->commands != NULL) { for (i = 1; i < sc->num_reqs; i++) { cm = &sc->commands[i]; bus_dmamap_destroy(sc->buffer_dmat, cm->cm_dmamap); } kfree(sc->commands, M_MPR); } if (sc->buffer_dmat != NULL) bus_dma_tag_destroy(sc->buffer_dmat); mpr_pci_free_interrupts(sc); kfree(sc->queues, M_MPR); sc->queues = NULL; } /* * The terms diag reset and hard reset are used interchangeably in the MPI * docs to mean resetting the controller chip. In this code diag reset * cleans everything up, and the hard reset function just sends the reset * sequence to the chip. This should probably be refactored so that every * subsystem gets a reset notification of some sort, and can clean up * appropriately. */ int mpr_reinit(struct mpr_softc *sc) { int error; struct mprsas_softc *sassc; sassc = sc->sassc; MPR_FUNCTRACE(sc); KKASSERT(lockowned(&sc->mpr_lock)); mpr_dprint(sc, MPR_INIT|MPR_INFO, "Reinitializing controller\n"); if (sc->mpr_flags & MPR_FLAGS_DIAGRESET) { mpr_dprint(sc, MPR_INIT, "Reset already in progress\n"); return 0; } /* * Make sure the completion callbacks can recognize they're getting * a NULL cm_reply due to a reset. */ sc->mpr_flags |= MPR_FLAGS_DIAGRESET; /* * Mask interrupts here. */ mpr_dprint(sc, MPR_INIT, "Masking interrupts and resetting\n"); mpr_mask_intr(sc); error = mpr_diag_reset(sc, CAN_SLEEP); if (error != 0) { panic("%s hard reset failed with error %d\n", __func__, error); } /* Restore the PCI state, including the MSI-X registers */ mpr_pci_restore(sc); /* Give the I/O subsystem special priority to get itself prepared */ mprsas_handle_reinit(sc); /* * Get IOC Facts and allocate all structures based on this information. * The attach function will also call mpr_iocfacts_allocate at startup. * If relevant values have changed in IOC Facts, this function will free * all of the memory based on IOC Facts and reallocate that memory. */ if ((error = mpr_iocfacts_allocate(sc, FALSE)) != 0) { panic("%s IOC Facts based allocation failed with error %d\n", __func__, error); } /* * Mapping structures will be re-allocated after getting IOC Page8, so * free these structures here. */ mpr_mapping_exit(sc); /* * The static page function currently read is IOC Page8. Others can be * added in future. It's possible that the values in IOC Page8 have * changed after a Diag Reset due to user modification, so always read * these. Interrupts are masked, so unmask them before getting config * pages. */ mpr_unmask_intr(sc); sc->mpr_flags &= ~MPR_FLAGS_DIAGRESET; mpr_base_static_config_pages(sc); /* * Some mapping info is based in IOC Page8 data, so re-initialize the * mapping tables. */ mpr_mapping_initialize(sc); /* * Restart will reload the event masks clobbered by the reset, and * then enable the port. */ mpr_reregister_events(sc); /* the end of discovery will release the simq, so we're done. */ mpr_dprint(sc, MPR_INIT|MPR_XINFO, "Finished sc %p post %u free %u\n", sc, sc->replypostindex, sc->replyfreeindex); mprsas_release_simq_reinit(sassc); mpr_dprint(sc, MPR_INIT, "%s exit error= %d\n", __func__, error); return 0; } /* Wait for the chip to ACK a word that we've put into its FIFO * Wait for <timeout> seconds. In single loop wait for busy loop * for 500 microseconds. * Total is [ 0.5 * (2000 * <timeout>) ] in milliseconds. * */ static int mpr_wait_db_ack(struct mpr_softc *sc, int timeout, int sleep_flag) { u32 cntdn, count; u32 int_status; u32 doorbell; count = 0; cntdn = (sleep_flag == CAN_SLEEP) ? 1000*timeout : 2000*timeout; do { int_status = mpr_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET); if (!(int_status & MPI2_HIS_SYS2IOC_DB_STATUS)) { mpr_dprint(sc, MPR_TRACE, "%s: successful count(%d), " "timeout(%d)\n", __func__, count, timeout); return 0; } else if (int_status & MPI2_HIS_IOC2SYS_DB_STATUS) { doorbell = mpr_regread(sc, MPI2_DOORBELL_OFFSET); if ((doorbell & MPI2_IOC_STATE_MASK) == MPI2_IOC_STATE_FAULT) { mpr_dprint(sc, MPR_FAULT, "fault_state(0x%04x)!\n", doorbell); return (EFAULT); } } else if (int_status == 0xFFFFFFFF) goto out; /* * If it can sleep, sleep for 1 millisecond, else busy loop for * 0.5 millisecond */ if (lockowned(&sc->mpr_lock) && sleep_flag == CAN_SLEEP) lksleep(&sc->msleep_fake_chan, &sc->mpr_lock, 0, "mprdba", hz < 1000 ? 1 : hz / 1000); else if (sleep_flag == CAN_SLEEP) tsleep(mpr_wait_db_ack, 0, "mprdba", hz < 1000 ? 1 : hz / 1000); else DELAY(500); count++; } while (--cntdn); out: mpr_dprint(sc, MPR_FAULT, "%s: failed due to timeout count(%d), " "int_status(%x)!\n", __func__, count, int_status); return (ETIMEDOUT); } /* Wait for the chip to signal that the next word in its FIFO can be fetched */ static int mpr_wait_db_int(struct mpr_softc *sc) { int retry; for (retry = 0; retry < MPR_DB_MAX_WAIT; retry++) { if ((mpr_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET) & MPI2_HIS_IOC2SYS_DB_STATUS) != 0) return (0); DELAY(2000); } return (ETIMEDOUT); } /* Step through the synchronous command state machine, i.e. "Doorbell mode" */ static int mpr_request_sync(struct mpr_softc *sc, void *req, MPI2_DEFAULT_REPLY *reply, int req_sz, int reply_sz, int timeout) { uint32_t *data32; uint16_t *data16; int i, count, ioc_sz, residual; int sleep_flags = CAN_SLEEP; /* Step 1 */ mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); /* Step 2 */ if (mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED) return (EBUSY); /* Step 3 * Announce that a message is coming through the doorbell. Messages * are pushed at 32bit words, so round up if needed. */ count = (req_sz + 3) / 4; mpr_regwrite(sc, MPI2_DOORBELL_OFFSET, (MPI2_FUNCTION_HANDSHAKE << MPI2_DOORBELL_FUNCTION_SHIFT) | (count << MPI2_DOORBELL_ADD_DWORDS_SHIFT)); /* Step 4 */ if (mpr_wait_db_int(sc) || (mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED) == 0) { mpr_dprint(sc, MPR_FAULT, "Doorbell failed to activate\n"); return (ENXIO); } mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); if (mpr_wait_db_ack(sc, 5, sleep_flags) != 0) { mpr_dprint(sc, MPR_FAULT, "Doorbell handshake failed\n"); return (ENXIO); } /* Step 5 */ /* Clock out the message data synchronously in 32-bit dwords*/ data32 = (uint32_t *)req; for (i = 0; i < count; i++) { mpr_regwrite(sc, MPI2_DOORBELL_OFFSET, htole32(data32[i])); if (mpr_wait_db_ack(sc, 5, sleep_flags) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout while writing doorbell\n"); return (ENXIO); } } /* Step 6 */ /* Clock in the reply in 16-bit words. The total length of the * message is always in the 4th byte, so clock out the first 2 words * manually, then loop the rest. */ data16 = (uint16_t *)reply; if (mpr_wait_db_int(sc) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout reading doorbell 0\n"); return (ENXIO); } data16[0] = mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_DATA_MASK; mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); if (mpr_wait_db_int(sc) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout reading doorbell 1\n"); return (ENXIO); } data16[1] = mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_DATA_MASK; mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); /* Number of 32bit words in the message */ ioc_sz = reply->MsgLength; /* * Figure out how many 16bit words to clock in without overrunning. * The precision loss with dividing reply_sz can safely be * ignored because the messages can only be multiples of 32bits. */ residual = 0; count = MIN((reply_sz / 4), ioc_sz) * 2; if (count < ioc_sz * 2) { residual = ioc_sz * 2 - count; mpr_dprint(sc, MPR_ERROR, "Driver error, throwing away %d " "residual message words\n", residual); } for (i = 2; i < count; i++) { if (mpr_wait_db_int(sc) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout reading doorbell %d\n", i); return (ENXIO); } data16[i] = mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_DATA_MASK; mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); } /* * Pull out residual words that won't fit into the provided buffer. * This keeps the chip from hanging due to a driver programming * error. */ while (residual--) { if (mpr_wait_db_int(sc) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout reading doorbell\n"); return (ENXIO); } (void)mpr_regread(sc, MPI2_DOORBELL_OFFSET); mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); } /* Step 7 */ if (mpr_wait_db_int(sc) != 0) { mpr_dprint(sc, MPR_FAULT, "Timeout waiting to exit doorbell\n"); return (ENXIO); } if (mpr_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED) mpr_dprint(sc, MPR_FAULT, "Warning, doorbell still active\n"); mpr_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0); return (0); } static void mpr_enqueue_request(struct mpr_softc *sc, struct mpr_command *cm) { request_descriptor rd; MPR_FUNCTRACE(sc); mpr_dprint(sc, MPR_TRACE, "SMID %u cm %p ccb %p\n", cm->cm_desc.Default.SMID, cm, cm->cm_ccb); if (sc->mpr_flags & MPR_FLAGS_ATTACH_DONE && !(sc->mpr_flags & MPR_FLAGS_SHUTDOWN)) KKASSERT(lockowned(&sc->mpr_lock)); if (++sc->io_cmds_active > sc->io_cmds_highwater) sc->io_cmds_highwater++; KASSERT(cm->cm_state == MPR_CM_STATE_BUSY, ("command not busy\n")); cm->cm_state = MPR_CM_STATE_INQUEUE; if (sc->atomic_desc_capable) { rd.u.low = cm->cm_desc.Words.Low; mpr_regwrite(sc, MPI26_ATOMIC_REQUEST_DESCRIPTOR_POST_OFFSET, rd.u.low); } else { rd.u.low = cm->cm_desc.Words.Low; rd.u.high = cm->cm_desc.Words.High; rd.word = htole64(rd.word); mpr_regwrite(sc, MPI2_REQUEST_DESCRIPTOR_POST_LOW_OFFSET, rd.u.low); mpr_regwrite(sc, MPI2_REQUEST_DESCRIPTOR_POST_HIGH_OFFSET, rd.u.high); } } /* * Just the FACTS, ma'am. */ static int mpr_get_iocfacts(struct mpr_softc *sc, MPI2_IOC_FACTS_REPLY *facts) { MPI2_DEFAULT_REPLY *reply; MPI2_IOC_FACTS_REQUEST request; int error, req_sz, reply_sz; MPR_FUNCTRACE(sc); mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); req_sz = sizeof(MPI2_IOC_FACTS_REQUEST); reply_sz = sizeof(MPI2_IOC_FACTS_REPLY); reply = (MPI2_DEFAULT_REPLY *)facts; bzero(&request, req_sz); request.Function = MPI2_FUNCTION_IOC_FACTS; error = mpr_request_sync(sc, &request, reply, req_sz, reply_sz, 5); mpr_dprint(sc, MPR_INIT, "%s exit, error= %d\n", __func__, error); return (error); } static int mpr_send_iocinit(struct mpr_softc *sc) { MPI2_IOC_INIT_REQUEST init; MPI2_DEFAULT_REPLY reply; int req_sz, reply_sz, error; struct timeval now; uint64_t time_in_msec; MPR_FUNCTRACE(sc); mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); /* Do a quick sanity check on proper initialization */ if ((sc->pqdepth == 0) || (sc->fqdepth == 0) || (sc->reqframesz == 0) || (sc->replyframesz == 0)) { mpr_dprint(sc, MPR_INIT|MPR_ERROR, "Driver not fully initialized for IOCInit\n"); return (EINVAL); } req_sz = sizeof(MPI2_IOC_INIT_REQUEST); reply_sz = sizeof(MPI2_IOC_INIT_REPLY); bzero(&init, req_sz); bzero(&reply, reply_sz); /* * Fill in the init block. Note that most addresses are * deliberately in the lower 32bits of memory. This is a micro- * optimzation for PCI/PCIX, though it's not clear if it helps PCIe. */ init.Function = MPI2_FUNCTION_IOC_INIT; init.WhoInit = MPI2_WHOINIT_HOST_DRIVER; init.MsgVersion = htole16(MPI2_VERSION); init.HeaderVersion = htole16(MPI2_HEADER_VERSION); init.SystemRequestFrameSize = htole16((uint16_t)(sc->reqframesz / 4)); init.ReplyDescriptorPostQueueDepth = htole16(sc->pqdepth); init.ReplyFreeQueueDepth = htole16(sc->fqdepth); init.SenseBufferAddressHigh = 0; init.SystemReplyAddressHigh = 0; init.SystemRequestFrameBaseAddress.High = 0; init.SystemRequestFrameBaseAddress.Low = htole32((uint32_t)sc->req_busaddr); init.ReplyDescriptorPostQueueAddress.High = 0; init.ReplyDescriptorPostQueueAddress.Low = htole32((uint32_t)sc->post_busaddr); init.ReplyFreeQueueAddress.High = 0; init.ReplyFreeQueueAddress.Low = htole32((uint32_t)sc->free_busaddr); getmicrotime(&now); time_in_msec = (now.tv_sec * 1000 + now.tv_usec/1000); init.TimeStamp.High = htole32((time_in_msec >> 32) & 0xFFFFFFFF); init.TimeStamp.Low = htole32(time_in_msec & 0xFFFFFFFF); init.HostPageSize = HOST_PAGE_SIZE_4K; error = mpr_request_sync(sc, &init, &reply, req_sz, reply_sz, 5); if ((reply.IOCStatus & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS) error = ENXIO; mpr_dprint(sc, MPR_INIT, "IOCInit status= 0x%x\n", reply.IOCStatus); mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); return (error); } void mpr_memaddr_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) { bus_addr_t *addr; addr = arg; *addr = segs[0].ds_addr; } void mpr_memaddr_wait_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) { struct mpr_busdma_context *ctx; int need_unload, need_free; ctx = (struct mpr_busdma_context *)arg; need_unload = 0; need_free = 0; mpr_lock(ctx->softc); ctx->error = error; ctx->completed = 1; if ((error == 0) && (ctx->abandoned == 0)) { *ctx->addr = segs[0].ds_addr; } else { if (nsegs != 0) need_unload = 1; if (ctx->abandoned != 0) need_free = 1; } if (need_free == 0) wakeup(ctx); mpr_unlock(ctx->softc); if (need_unload != 0) { bus_dmamap_unload(ctx->buffer_dmat, ctx->buffer_dmamap); *ctx->addr = 0; } if (need_free != 0) kfree(ctx, M_MPR); } static int mpr_alloc_queues(struct mpr_softc *sc) { struct mpr_queue *q; int nq, i; nq = sc->msi_msgs; mpr_dprint(sc, MPR_INIT|MPR_XINFO, "Allocating %d I/O queues\n", nq); sc->queues = kmalloc(sizeof(struct mpr_queue) * nq, M_MPR, M_NOWAIT|M_ZERO); if (sc->queues == NULL) return (ENOMEM); for (i = 0; i < nq; i++) { q = &sc->queues[i]; mpr_dprint(sc, MPR_INIT, "Configuring queue %d %p\n", i, q); q->sc = sc; q->qnum = i; } return (0); } static int mpr_alloc_hw_queues(struct mpr_softc *sc) { bus_addr_t queues_busaddr; uint8_t *queues; int qsize, fqsize, pqsize; /* * The reply free queue contains 4 byte entries in multiples of 16 and * aligned on a 16 byte boundary. There must always be an unused entry. * This queue supplies fresh reply frames for the firmware to use. * * The reply descriptor post queue contains 8 byte entries in * multiples of 16 and aligned on a 16 byte boundary. This queue * contains filled-in reply frames sent from the firmware to the host. * * These two queues are allocated together for simplicity. */ sc->fqdepth = roundup2(sc->num_replies + 1, 16); sc->pqdepth = roundup2(sc->num_replies + 1, 16); fqsize= sc->fqdepth * 4; pqsize = sc->pqdepth * 8; qsize = fqsize + pqsize; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 16, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR_32BIT,/* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ qsize, /* maxsize */ 1, /* nsegments */ qsize, /* maxsegsize */ 0, /* flags */ &sc->queues_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate queues DMA tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->queues_dmat, (void **)&queues, BUS_DMA_NOWAIT, &sc->queues_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate queues memory\n"); return (ENOMEM); } bzero(queues, qsize); bus_dmamap_load(sc->queues_dmat, sc->queues_map, queues, qsize, mpr_memaddr_cb, &queues_busaddr, 0); sc->free_queue = (uint32_t *)queues; sc->free_busaddr = queues_busaddr; sc->post_queue = (MPI2_REPLY_DESCRIPTORS_UNION *)(queues + fqsize); sc->post_busaddr = queues_busaddr + fqsize; mpr_dprint(sc, MPR_INIT, "free queue busaddr= %#016jx size= %d\n", (uintmax_t)sc->free_busaddr, fqsize); mpr_dprint(sc, MPR_INIT, "reply queue busaddr= %#016jx size= %d\n", (uintmax_t)sc->post_busaddr, pqsize); return (0); } static int mpr_alloc_replies(struct mpr_softc *sc) { int rsize, num_replies; /* Store the reply frame size in bytes rather than as 32bit words */ sc->replyframesz = sc->facts->ReplyFrameSize * 4; /* * sc->num_replies should be one less than sc->fqdepth. We need to * allocate space for sc->fqdepth replies, but only sc->num_replies * replies can be used at once. */ num_replies = max(sc->fqdepth, sc->num_replies); rsize = sc->replyframesz * num_replies; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 4, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR_32BIT,/* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ rsize, /* maxsize */ 1, /* nsegments */ rsize, /* maxsegsize */ 0, /* flags */ &sc->reply_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate replies DMA tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->reply_dmat, (void **)&sc->reply_frames, BUS_DMA_NOWAIT, &sc->reply_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate replies memory\n"); return (ENOMEM); } bzero(sc->reply_frames, rsize); bus_dmamap_load(sc->reply_dmat, sc->reply_map, sc->reply_frames, rsize, mpr_memaddr_cb, &sc->reply_busaddr, 0); mpr_dprint(sc, MPR_INIT, "reply frames busaddr= %#016jx size= %d\n", (uintmax_t)sc->reply_busaddr, rsize); return (0); } static void mpr_load_chains_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) { struct mpr_softc *sc = arg; struct mpr_chain *chain; bus_size_t bo; int i, o, s; if (error != 0) return; for (i = 0, o = 0, s = 0; s < nsegs; s++) { for (bo = 0; bo + sc->chain_frame_size <= segs[s].ds_len; bo += sc->chain_frame_size) { chain = &sc->chains[i++]; chain->chain =(MPI2_SGE_IO_UNION *)(sc->chain_frames+o); chain->chain_busaddr = segs[s].ds_addr + bo; o += sc->chain_frame_size; mpr_free_chain(sc, chain); } if (bo != segs[s].ds_len) o += segs[s].ds_len - bo; } sc->chain_free_lowwater = i; } static int mpr_alloc_requests(struct mpr_softc *sc) { struct mpr_command *cm; int i, rsize, nsegs; rsize = sc->reqframesz * sc->num_reqs; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 16, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR_32BIT,/* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ rsize, /* maxsize */ 1, /* nsegments */ rsize, /* maxsegsize */ 0, /* flags */ &sc->req_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate request DMA tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->req_dmat, (void **)&sc->req_frames, BUS_DMA_NOWAIT, &sc->req_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate request memory\n"); return (ENOMEM); } bzero(sc->req_frames, rsize); bus_dmamap_load(sc->req_dmat, sc->req_map, sc->req_frames, rsize, mpr_memaddr_cb, &sc->req_busaddr, 0); mpr_dprint(sc, MPR_INIT, "request frames busaddr= %#016jx size= %d\n", (uintmax_t)sc->req_busaddr, rsize); sc->chains = kmalloc(sizeof(struct mpr_chain) * sc->num_chains, M_MPR, M_NOWAIT | M_ZERO); if (!sc->chains) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate chain memory\n"); return (ENOMEM); } rsize = sc->chain_frame_size * sc->num_chains; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 16, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ rsize, /* maxsize */ howmany(rsize, PAGE_SIZE), /* nsegments */ rsize, /* maxsegsize */ 0, /* flags */ &sc->chain_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate chain DMA tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->chain_dmat, (void **)&sc->chain_frames, BUS_DMA_NOWAIT | BUS_DMA_ZERO, &sc->chain_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate chain memory\n"); return (ENOMEM); } if (bus_dmamap_load(sc->chain_dmat, sc->chain_map, sc->chain_frames, rsize, mpr_load_chains_cb, sc, BUS_DMA_NOWAIT)) { mpr_dprint(sc, MPR_ERROR, "Cannot load chain memory\n"); bus_dmamem_free(sc->chain_dmat, sc->chain_frames, sc->chain_map); return (ENOMEM); } rsize = MPR_SENSE_LEN * sc->num_reqs; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 1, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR_32BIT,/* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ rsize, /* maxsize */ 1, /* nsegments */ rsize, /* maxsegsize */ 0, /* flags */ &sc->sense_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate sense DMA tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->sense_dmat, (void **)&sc->sense_frames, BUS_DMA_NOWAIT, &sc->sense_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate sense memory\n"); return (ENOMEM); } bzero(sc->sense_frames, rsize); bus_dmamap_load(sc->sense_dmat, sc->sense_map, sc->sense_frames, rsize, mpr_memaddr_cb, &sc->sense_busaddr, 0); mpr_dprint(sc, MPR_INIT, "sense frames busaddr= %#016jx size= %d\n", (uintmax_t)sc->sense_busaddr, rsize); /* * Allocate NVMe PRP Pages for NVMe SGL support only if the FW supports * these devices. */ if ((sc->facts->MsgVersion >= MPI2_VERSION_02_06) && (sc->facts->ProtocolFlags & MPI2_IOCFACTS_PROTOCOL_NVME_DEVICES)) { if (mpr_alloc_nvme_prp_pages(sc) == ENOMEM) return (ENOMEM); } nsegs = (sc->maxio / PAGE_SIZE) + 1; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 1, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXSIZE_32BIT,/* maxsize */ nsegs, /* nsegments */ BUS_SPACE_MAXSIZE_32BIT,/* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ &sc->buffer_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate buffer DMA tag\n"); return (ENOMEM); } /* * SMID 0 cannot be used as a free command per the firmware spec. * Just drop that command instead of risking accounting bugs. */ sc->commands = kmalloc(sizeof(struct mpr_command) * sc->num_reqs, M_MPR, M_WAITOK | M_ZERO); if (!sc->commands) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate command memory\n"); return (ENOMEM); } for (i = 1; i < sc->num_reqs; i++) { cm = &sc->commands[i]; cm->cm_req = sc->req_frames + i * sc->reqframesz; cm->cm_req_busaddr = sc->req_busaddr + i * sc->reqframesz; cm->cm_sense = &sc->sense_frames[i]; cm->cm_sense_busaddr = sc->sense_busaddr + i * MPR_SENSE_LEN; cm->cm_desc.Default.SMID = i; cm->cm_sc = sc; cm->cm_state = MPR_CM_STATE_BUSY; TAILQ_INIT(&cm->cm_chain_list); TAILQ_INIT(&cm->cm_prp_page_list); callout_init_lk(&cm->cm_callout, &sc->mpr_lock); /* XXX Is a failure here a critical problem? */ if (bus_dmamap_create(sc->buffer_dmat, 0, &cm->cm_dmamap) == 0) { if (i <= sc->num_prireqs) mpr_free_high_priority_command(sc, cm); else mpr_free_command(sc, cm); } else { panic("failed to allocate command %d\n", i); sc->num_reqs = i; break; } } return (0); } /* * Allocate contiguous buffers for PCIe NVMe devices for building native PRPs, * which are scatter/gather lists for NVMe devices. * * This buffer must be contiguous due to the nature of how NVMe PRPs are built * and translated by FW. * * returns ENOMEM if memory could not be allocated, otherwise returns 0. */ static int mpr_alloc_nvme_prp_pages(struct mpr_softc *sc) { int PRPs_per_page, PRPs_required, pages_required; int rsize, i; struct mpr_prp_page *prp_page; /* * Assuming a MAX_IO_SIZE of 1MB and a PAGE_SIZE of 4k, the max number * of PRPs (NVMe's Scatter/Gather Element) needed per I/O is: * MAX_IO_SIZE / PAGE_SIZE = 256 * * 1 PRP entry in main frame for PRP list pointer still leaves 255 PRPs * required for the remainder of the 1MB I/O. 512 PRPs can fit into one * page (4096 / 8 = 512), so only one page is required for each I/O. * * Each of these buffers will need to be contiguous. For simplicity, * only one buffer is allocated here, which has all of the space * required for the NVMe Queue Depth. If there are problems allocating * this one buffer, this function will need to change to allocate * individual, contiguous NVME_QDEPTH buffers. * * The real calculation will use the real max io size. Above is just an * example. * */ PRPs_required = sc->maxio / PAGE_SIZE; PRPs_per_page = (PAGE_SIZE / PRP_ENTRY_SIZE) - 1; pages_required = (PRPs_required / PRPs_per_page) + 1; sc->prp_buffer_size = PAGE_SIZE * pages_required; rsize = sc->prp_buffer_size * NVME_QDEPTH; if (bus_dma_tag_create( sc->mpr_parent_dmat, /* parent */ 4, 0, /* algnmnt, boundary */ BUS_SPACE_MAXADDR_32BIT,/* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ rsize, /* maxsize */ 1, /* nsegments */ rsize, /* maxsegsize */ 0, /* flags */ &sc->prp_page_dmat)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate NVMe PRP DMA " "tag\n"); return (ENOMEM); } if (bus_dmamem_alloc(sc->prp_page_dmat, (void **)&sc->prp_pages, BUS_DMA_NOWAIT, &sc->prp_page_map)) { mpr_dprint(sc, MPR_ERROR, "Cannot allocate NVMe PRP memory\n"); return (ENOMEM); } bzero(sc->prp_pages, rsize); bus_dmamap_load(sc->prp_page_dmat, sc->prp_page_map, sc->prp_pages, rsize, mpr_memaddr_cb, &sc->prp_page_busaddr, 0); sc->prps = kmalloc(sizeof(struct mpr_prp_page) * NVME_QDEPTH, M_MPR, M_WAITOK | M_ZERO); for (i = 0; i < NVME_QDEPTH; i++) { prp_page = &sc->prps[i]; prp_page->prp_page = (uint64_t *)(sc->prp_pages + i * sc->prp_buffer_size); prp_page->prp_page_busaddr = (uint64_t)(sc->prp_page_busaddr + i * sc->prp_buffer_size); mpr_free_prp_page(sc, prp_page); sc->prp_pages_free_lowwater++; } return (0); } static int mpr_init_queues(struct mpr_softc *sc) { int i; memset((uint8_t *)sc->post_queue, 0xff, sc->pqdepth * 8); /* * According to the spec, we need to use one less reply than we * have space for on the queue. So sc->num_replies (the number we * use) should be less than sc->fqdepth (allocated size). */ if (sc->num_replies >= sc->fqdepth) return (EINVAL); /* * Initialize all of the free queue entries. */ for (i = 0; i < sc->fqdepth; i++) { sc->free_queue[i] = sc->reply_busaddr + (i * sc->replyframesz); } sc->replyfreeindex = sc->num_replies; return (0); } /* Get the driver parameter tunables. Lowest priority are the driver defaults. * Next are the global settings, if they exist. Highest are the per-unit * settings, if they exist. */ void mpr_get_tunables(struct mpr_softc *sc) { char tmpstr[80]; /* XXX default to some debugging for now */ sc->mpr_debug = MPR_INFO | MPR_FAULT; sc->msi_enable = 1; sc->max_chains = MPR_CHAIN_FRAMES; sc->max_io_pages = MPR_MAXIO_PAGES; sc->enable_ssu = MPR_SSU_ENABLE_SSD_DISABLE_HDD; sc->spinup_wait_time = DEFAULT_SPINUP_WAIT; sc->use_phynum = 1; sc->max_reqframes = MPR_REQ_FRAMES; sc->max_prireqframes = MPR_PRI_REQ_FRAMES; sc->max_replyframes = MPR_REPLY_FRAMES; sc->max_evtframes = MPR_EVT_REPLY_FRAMES; /* * Grab the global variables. */ TUNABLE_INT_FETCH("hw.mpr.debug_level", &sc->mpr_debug); TUNABLE_INT_FETCH("hw.mpr.msi_enable", &sc->msi_enable); TUNABLE_INT_FETCH("hw.mpr.max_chains", &sc->max_chains); TUNABLE_INT_FETCH("hw.mpr.max_io_pages", &sc->max_io_pages); TUNABLE_INT_FETCH("hw.mpr.enable_ssu", &sc->enable_ssu); TUNABLE_INT_FETCH("hw.mpr.spinup_wait_time", &sc->spinup_wait_time); TUNABLE_INT_FETCH("hw.mpr.use_phy_num", &sc->use_phynum); TUNABLE_INT_FETCH("hw.mpr.max_reqframes", &sc->max_reqframes); TUNABLE_INT_FETCH("hw.mpr.max_prireqframes", &sc->max_prireqframes); TUNABLE_INT_FETCH("hw.mpr.max_replyframes", &sc->max_replyframes); TUNABLE_INT_FETCH("hw.mpr.max_evtframes", &sc->max_evtframes); /* Grab the unit-instance variables */ ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.debug_level", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->mpr_debug); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.msi_enable", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->msi_enable); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_chains", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_chains); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_io_pages", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_io_pages); bzero(sc->exclude_ids, sizeof(sc->exclude_ids)); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.exclude_ids", device_get_unit(sc->mpr_dev)); TUNABLE_STR_FETCH(tmpstr, sc->exclude_ids, sizeof(sc->exclude_ids)); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.enable_ssu", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->enable_ssu); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.spinup_wait_time", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->spinup_wait_time); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.use_phy_num", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->use_phynum); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_reqframes", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_reqframes); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_prireqframes", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_prireqframes); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_replyframes", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_replyframes); ksnprintf(tmpstr, sizeof(tmpstr), "dev.mpr.%d.max_evtframes", device_get_unit(sc->mpr_dev)); TUNABLE_INT_FETCH(tmpstr, &sc->max_evtframes); } static void mpr_setup_sysctl(struct mpr_softc *sc) { struct sysctl_ctx_list *sysctl_ctx = NULL; struct sysctl_oid *sysctl_tree = NULL; char tmpstr[80], tmpstr2[80]; /* * Setup the sysctl variable so the user can change the debug level * on the fly. */ ksnprintf(tmpstr, sizeof(tmpstr), "MPR controller %d", device_get_unit(sc->mpr_dev)); ksnprintf(tmpstr2, sizeof(tmpstr2), "%d", device_get_unit(sc->mpr_dev)); sysctl_ctx = device_get_sysctl_ctx(sc->mpr_dev); if (sysctl_ctx != NULL) sysctl_tree = device_get_sysctl_tree(sc->mpr_dev); if (sysctl_tree == NULL) { sysctl_ctx_init(&sc->sysctl_ctx); sc->sysctl_tree = SYSCTL_ADD_NODE(&sc->sysctl_ctx, SYSCTL_STATIC_CHILDREN(_hw_mpr), OID_AUTO, tmpstr2, CTLFLAG_RD, 0, tmpstr); if (sc->sysctl_tree == NULL) return; sysctl_ctx = &sc->sysctl_ctx; sysctl_tree = sc->sysctl_tree; } SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "debug_level", CTLFLAG_RW, &sc->mpr_debug, 0, "mpr debug level"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_reqframes", CTLFLAG_RD, &sc->max_reqframes, 0, "Total number of allocated request frames"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_prireqframes", CTLFLAG_RD, &sc->max_prireqframes, 0, "Total number of allocated high priority request frames"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_replyframes", CTLFLAG_RD, &sc->max_replyframes, 0, "Total number of allocated reply frames"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_evtframes", CTLFLAG_RD, &sc->max_evtframes, 0, "Total number of event frames allocated"); SYSCTL_ADD_STRING(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "firmware_version", CTLFLAG_RW, sc->fw_version, strlen(sc->fw_version), "firmware version"); SYSCTL_ADD_STRING(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "driver_version", CTLFLAG_RW, MPR_DRIVER_VERSION, strlen(MPR_DRIVER_VERSION), "driver version"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "io_cmds_active", CTLFLAG_RD, &sc->io_cmds_active, 0, "number of currently active commands"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "io_cmds_highwater", CTLFLAG_RD, &sc->io_cmds_highwater, 0, "maximum active commands seen"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "chain_free", CTLFLAG_RD, &sc->chain_free, 0, "number of free chain elements"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "chain_free_lowwater", CTLFLAG_RD, &sc->chain_free_lowwater, 0,"lowest number of free chain elements"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_chains", CTLFLAG_RD, &sc->max_chains, 0,"maximum chain frames that will be allocated"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "max_io_pages", CTLFLAG_RD, &sc->max_io_pages, 0,"maximum pages to allow per I/O (if <1 use " "IOCFacts)"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "enable_ssu", CTLFLAG_RW, &sc->enable_ssu, 0, "enable SSU to SATA SSD/HDD at shutdown"); SYSCTL_ADD_UQUAD(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "chain_alloc_fail", CTLFLAG_RD, &sc->chain_alloc_fail, 0, "chain allocation failures"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "spinup_wait_time", CTLFLAG_RD, &sc->spinup_wait_time, DEFAULT_SPINUP_WAIT, "seconds to wait for " "spinup after SATA ID error"); SYSCTL_ADD_PROC(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "dump_reqs", CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_SKIP, sc, 0, mpr_dump_reqs, "I", "Dump Active Requests"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "use_phy_num", CTLFLAG_RD, &sc->use_phynum, 0, "Use the phy number for enumeration"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "prp_pages_free", CTLFLAG_RD, &sc->prp_pages_free, 0, "number of free PRP pages"); SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "prp_pages_free_lowwater", CTLFLAG_RD, &sc->prp_pages_free_lowwater, 0,"lowest number of free PRP pages"); SYSCTL_ADD_UQUAD(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree), OID_AUTO, "prp_page_alloc_fail", CTLFLAG_RD, &sc->prp_page_alloc_fail, 0, "PRP page allocation failures"); } struct mpr_dumpreq_hdr { uint32_t smid; uint32_t state; uint32_t numframes; uint32_t deschi; uint32_t desclo; }; static int mpr_dump_reqs(SYSCTL_HANDLER_ARGS) { struct mpr_softc *sc; struct mpr_chain *chain, *chain1; struct mpr_command *cm; struct mpr_dumpreq_hdr hdr; struct sbuf *sb; uint32_t smid, state; int i, numreqs, error = 0; sc = (struct mpr_softc *)arg1; if ((error = caps_priv_check_self(SYSCAP_NODRIVER)) != 0) return (error); state = MPR_CM_STATE_INQUEUE; smid = 1; numreqs = sc->num_reqs; if (req->newptr != NULL) return (EINVAL); if (smid == 0 || smid > sc->num_reqs) return (EINVAL); if (numreqs <= 0 || (numreqs + smid > sc->num_reqs)) numreqs = sc->num_reqs; sb = sbuf_new_for_sysctl(NULL, NULL, 4096, req); /* Best effort, no locking */ for (i = smid; i < numreqs; i++) { cm = &sc->commands[i]; if (cm->cm_state != state) continue; hdr.smid = i; hdr.state = cm->cm_state; hdr.numframes = 1; hdr.deschi = cm->cm_desc.Words.High; hdr.desclo = cm->cm_desc.Words.Low; TAILQ_FOREACH_MUTABLE(chain, &cm->cm_chain_list, chain_link, chain1) hdr.numframes++; sbuf_bcat(sb, &hdr, sizeof(hdr)); sbuf_bcat(sb, cm->cm_req, 128); TAILQ_FOREACH_MUTABLE(chain, &cm->cm_chain_list, chain_link, chain1) sbuf_bcat(sb, chain->chain, 128); } error = sbuf_finish(sb); sbuf_delete(sb); return (error); } int mpr_attach(struct mpr_softc *sc) { int error; MPR_FUNCTRACE(sc); mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); lockinit(&sc->mpr_lock, "MPR lock", 0, LK_CANRECURSE); callout_init_lk(&sc->periodic, &sc->mpr_lock); callout_init_lk(&sc->device_check_callout, &sc->mpr_lock); TAILQ_INIT(&sc->event_list); timevalclear(&sc->lastfail); if ((error = mpr_transition_ready(sc)) != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Failed to transition ready\n"); return (error); } sc->facts = kmalloc(sizeof(MPI2_IOC_FACTS_REPLY), M_MPR, M_ZERO|M_NOWAIT); if (!sc->facts) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "Cannot allocate memory, exit\n"); return (ENOMEM); } /* * Get IOC Facts and allocate all structures based on this information. * A Diag Reset will also call mpr_iocfacts_allocate and re-read the IOC * Facts. If relevant values have changed in IOC Facts, this function * will free all of the memory based on IOC Facts and reallocate that * memory. If this fails, any allocated memory should already be freed. */ if ((error = mpr_iocfacts_allocate(sc, TRUE)) != 0) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "IOC Facts allocation " "failed with error %d\n", error); return (error); } /* Start the periodic watchdog check on the IOC Doorbell */ mpr_periodic(sc); /* * The portenable will kick off discovery events that will drive the * rest of the initialization process. The CAM/SAS module will * hold up the boot sequence until discovery is complete. */ sc->mpr_ich.ich_func = mpr_startup; sc->mpr_ich.ich_arg = sc; if (config_intrhook_establish(&sc->mpr_ich) != 0) { mpr_dprint(sc, MPR_INIT|MPR_ERROR, "Cannot establish MPR config hook\n"); error = EINVAL; } /* * Allow IR to shutdown gracefully when shutdown occurs. */ sc->shutdown_eh = EVENTHANDLER_REGISTER(shutdown_final, mprsas_ir_shutdown, sc, SHUTDOWN_PRI_DEFAULT); if (sc->shutdown_eh == NULL) mpr_dprint(sc, MPR_INIT|MPR_ERROR, "shutdown event registration failed\n"); mpr_setup_sysctl(sc); sc->mpr_flags |= MPR_FLAGS_ATTACH_DONE; mpr_dprint(sc, MPR_INIT, "%s exit error= %d\n", __func__, error); return (error); } /* Run through any late-start handlers. */ static void mpr_startup(void *arg) { struct mpr_softc *sc; sc = (struct mpr_softc *)arg; mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); mpr_lock(sc); mpr_unmask_intr(sc); /* initialize device mapping tables */ mpr_base_static_config_pages(sc); mpr_mapping_initialize(sc); mprsas_startup(sc); mpr_unlock(sc); mpr_dprint(sc, MPR_INIT, "disestablish config intrhook\n"); config_intrhook_disestablish(&sc->mpr_ich); sc->mpr_ich.ich_arg = NULL; mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); } /* Periodic watchdog. Is called with the driver lock already held. */ static void mpr_periodic(void *arg) { struct mpr_softc *sc; uint32_t db; sc = (struct mpr_softc *)arg; if (sc->mpr_flags & MPR_FLAGS_SHUTDOWN) return; db = mpr_regread(sc, MPI2_DOORBELL_OFFSET); if ((db & MPI2_IOC_STATE_MASK) == MPI2_IOC_STATE_FAULT) { if ((db & MPI2_DOORBELL_FAULT_CODE_MASK) == IFAULT_IOP_OVER_TEMP_THRESHOLD_EXCEEDED) { panic("TEMPERATURE FAULT: STOPPING."); } mpr_dprint(sc, MPR_FAULT, "IOC Fault 0x%08x, Resetting\n", db); mpr_reinit(sc); } callout_reset(&sc->periodic, MPR_PERIODIC_DELAY * hz, mpr_periodic, sc); } static void mpr_log_evt_handler(struct mpr_softc *sc, uintptr_t data, MPI2_EVENT_NOTIFICATION_REPLY *event) { MPI2_EVENT_DATA_LOG_ENTRY_ADDED *entry; MPR_DPRINT_EVENT(sc, generic, event); switch (event->Event) { case MPI2_EVENT_LOG_DATA: mpr_dprint(sc, MPR_EVENT, "MPI2_EVENT_LOG_DATA:\n"); if (sc->mpr_debug & MPR_EVENT) hexdump(event->EventData, event->EventDataLength, NULL, 0); break; case MPI2_EVENT_LOG_ENTRY_ADDED: entry = (MPI2_EVENT_DATA_LOG_ENTRY_ADDED *)event->EventData; mpr_dprint(sc, MPR_EVENT, "MPI2_EVENT_LOG_ENTRY_ADDED event " "0x%x Sequence %d:\n", entry->LogEntryQualifier, entry->LogSequence); break; default: break; } return; } static int mpr_attach_log(struct mpr_softc *sc) { uint8_t events[16]; bzero(events, 16); setbit(events, MPI2_EVENT_LOG_DATA); setbit(events, MPI2_EVENT_LOG_ENTRY_ADDED); mpr_register_events(sc, events, mpr_log_evt_handler, NULL, &sc->mpr_log_eh); return (0); } static int mpr_detach_log(struct mpr_softc *sc) { if (sc->mpr_log_eh != NULL) mpr_deregister_events(sc, sc->mpr_log_eh); return (0); } /* * Free all of the driver resources and detach submodules. Should be called * without the lock held. */ int mpr_free(struct mpr_softc *sc) { int error; mpr_dprint(sc, MPR_INIT, "%s entered\n", __func__); /* Turn off the watchdog */ mpr_lock(sc); sc->mpr_flags |= MPR_FLAGS_SHUTDOWN; mpr_unlock(sc); /* Lock must not be held for this */ callout_drain(&sc->periodic); callout_drain(&sc->device_check_callout); if (((error = mpr_detach_log(sc)) != 0) || ((error = mpr_detach_sas(sc)) != 0)) { mpr_dprint(sc, MPR_INIT|MPR_FAULT, "failed to detach " "subsystems, error= %d, exit\n", error); return (error); } mpr_detach_user(sc); /* Put the IOC back in the READY state. */ mpr_lock(sc); if ((error = mpr_transition_ready(sc)) != 0) { mpr_unlock(sc); return (error); } mpr_unlock(sc); if (sc->facts != NULL) kfree(sc->facts, M_MPR); /* * Free all buffers that are based on IOC Facts. A Diag Reset may need * to free these buffers too. */ mpr_iocfacts_free(sc); if (sc->sysctl_tree != NULL) sysctl_ctx_free(&sc->sysctl_ctx); /* Deregister the shutdown function */ if (sc->shutdown_eh != NULL) EVENTHANDLER_DEREGISTER(shutdown_final, sc->shutdown_eh); lockuninit(&sc->mpr_lock); mpr_dprint(sc, MPR_INIT, "%s exit\n", __func__); return (0); } static __inline void mpr_complete_command(struct mpr_softc *sc, struct mpr_command *cm) { MPR_FUNCTRACE(sc); if (cm == NULL) { mpr_dprint(sc, MPR_ERROR, "Completing NULL command\n"); return; } if (cm->cm_flags & MPR_CM_FLAGS_POLLED) cm->cm_flags |= MPR_CM_FLAGS_COMPLETE; if (cm->cm_complete != NULL) { mpr_dprint(sc, MPR_TRACE, "%s cm %p calling cm_complete %p data %p reply %p\n", __func__, cm, cm->cm_complete, cm->cm_complete_data, cm->cm_reply); cm->cm_complete(sc, cm); } if (cm->cm_flags & MPR_CM_FLAGS_WAKEUP) { mpr_dprint(sc, MPR_TRACE, "waking up %p\n", cm); wakeup(cm); } if (sc->io_cmds_active != 0) { sc->io_cmds_active--; } else { mpr_dprint(sc, MPR_ERROR, "Warning: io_cmds_active is " "out of sync - resynching to 0\n"); } } static void mpr_sas_log_info(struct mpr_softc *sc , u32 log_info) { union loginfo_type { u32 loginfo; struct { u32 subcode:16; u32 code:8; u32 originator:4; u32 bus_type:4; } dw; }; union loginfo_type sas_loginfo; char *originator_str = NULL; sas_loginfo.loginfo = log_info; if (sas_loginfo.dw.bus_type != 3 /*SAS*/) return; /* each nexus loss loginfo */ if (log_info == 0x31170000) return; /* eat the loginfos associated with task aborts */ if ((log_info == 30050000) || (log_info == 0x31140000) || (log_info == 0x31130000)) return; switch (sas_loginfo.dw.originator) { case 0: originator_str = "IOP"; break; case 1: originator_str = "PL"; break; case 2: originator_str = "IR"; break; } mpr_dprint(sc, MPR_LOG, "log_info(0x%08x): originator(%s), " "code(0x%02x), sub_code(0x%04x)\n", log_info, originator_str, sas_loginfo.dw.code, sas_loginfo.dw.subcode); } static void mpr_display_reply_info(struct mpr_softc *sc, uint8_t *reply) { MPI2DefaultReply_t *mpi_reply; u16 sc_status; mpi_reply = (MPI2DefaultReply_t*)reply; sc_status = le16toh(mpi_reply->IOCStatus); if (sc_status & MPI2_IOCSTATUS_FLAG_LOG_INFO_AVAILABLE) mpr_sas_log_info(sc, le32toh(mpi_reply->IOCLogInfo)); } void mpr_intr(void *data) { struct mpr_softc *sc; uint32_t status; sc = (struct mpr_softc *)data; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); /* * Check interrupt status register to flush the bus. This is * needed for both INTx interrupts and driver-driven polling */ status = mpr_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET); if ((status & MPI2_HIS_REPLY_DESCRIPTOR_INTERRUPT) == 0) return; mpr_lock(sc); mpr_intr_locked(data); mpr_unlock(sc); return; } /* * In theory, MSI/MSIX interrupts shouldn't need to read any registers on the * chip. Hopefully this theory is correct. */ void mpr_intr_msi(void *data) { struct mpr_softc *sc; sc = (struct mpr_softc *)data; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); mpr_lock(sc); mpr_intr_locked(data); mpr_unlock(sc); return; } /* * The locking is overly broad and simplistic, but easy to deal with for now. */ void mpr_intr_locked(void *data) { MPI2_REPLY_DESCRIPTORS_UNION *desc; struct mpr_softc *sc; struct mpr_command *cm = NULL; uint8_t flags; u_int pq; MPI2_DIAG_RELEASE_REPLY *rel_rep; mpr_fw_diagnostic_buffer_t *pBuffer; sc = (struct mpr_softc *)data; pq = sc->replypostindex; mpr_dprint(sc, MPR_TRACE, "%s sc %p starting with replypostindex %u\n", __func__, sc, sc->replypostindex); for ( ;; ) { cm = NULL; desc = &sc->post_queue[sc->replypostindex]; flags = desc->Default.ReplyFlags & MPI2_RPY_DESCRIPT_FLAGS_TYPE_MASK; if ((flags == MPI2_RPY_DESCRIPT_FLAGS_UNUSED) || (le32toh(desc->Words.High) == 0xffffffff)) break; /* increment the replypostindex now, so that event handlers * and cm completion handlers which decide to do a diag * reset can zero it without it getting incremented again * afterwards, and we break out of this loop on the next * iteration since the reply post queue has been cleared to * 0xFF and all descriptors look unused (which they are). */ if (++sc->replypostindex >= sc->pqdepth) sc->replypostindex = 0; switch (flags) { case MPI2_RPY_DESCRIPT_FLAGS_SCSI_IO_SUCCESS: case MPI25_RPY_DESCRIPT_FLAGS_FAST_PATH_SCSI_IO_SUCCESS: case MPI26_RPY_DESCRIPT_FLAGS_PCIE_ENCAPSULATED_SUCCESS: cm = &sc->commands[le16toh(desc->SCSIIOSuccess.SMID)]; KASSERT(cm->cm_state == MPR_CM_STATE_INQUEUE, ("command not inqueue\n")); cm->cm_state = MPR_CM_STATE_BUSY; cm->cm_reply = NULL; break; case MPI2_RPY_DESCRIPT_FLAGS_ADDRESS_REPLY: { uint32_t baddr; uint8_t *reply; /* * Re-compose the reply address from the address * sent back from the chip. The ReplyFrameAddress * is the lower 32 bits of the physical address of * particular reply frame. Convert that address to * host format, and then use that to provide the * offset against the virtual address base * (sc->reply_frames). */ baddr = le32toh(desc->AddressReply.ReplyFrameAddress); reply = sc->reply_frames + (baddr - ((uint32_t)sc->reply_busaddr)); /* * Make sure the reply we got back is in a valid * range. If not, go ahead and panic here, since * we'll probably panic as soon as we deference the * reply pointer anyway. */ if ((reply < sc->reply_frames) || (reply > (sc->reply_frames + (sc->fqdepth * sc->replyframesz)))) { kprintf("%s: WARNING: reply %p out of range!\n", __func__, reply); kprintf("%s: reply_frames %p, fqdepth %d, " "frame size %d\n", __func__, sc->reply_frames, sc->fqdepth, sc->replyframesz); kprintf("%s: baddr %#x,\n", __func__, baddr); /* LSI-TODO. See Linux Code for Graceful exit */ panic("Reply address out of range"); } if (le16toh(desc->AddressReply.SMID) == 0) { if (((MPI2_DEFAULT_REPLY *)reply)->Function == MPI2_FUNCTION_DIAG_BUFFER_POST) { /* * If SMID is 0 for Diag Buffer Post, * this implies that the reply is due to * a release function with a status that * the buffer has been released. Set * the buffer flags accordingly. */ rel_rep = (MPI2_DIAG_RELEASE_REPLY *)reply; if ((le16toh(rel_rep->IOCStatus) & MPI2_IOCSTATUS_MASK) == MPI2_IOCSTATUS_DIAGNOSTIC_RELEASED) { pBuffer = &sc->fw_diag_buffer_list[ rel_rep->BufferType]; pBuffer->valid_data = TRUE; pBuffer->owned_by_firmware = FALSE; pBuffer->immediate = FALSE; } } else mpr_dispatch_event(sc, baddr, (MPI2_EVENT_NOTIFICATION_REPLY *) reply); } else { cm = &sc->commands[ le16toh(desc->AddressReply.SMID)]; KASSERT(cm->cm_state == MPR_CM_STATE_INQUEUE, ("command not inqueue\n")); cm->cm_state = MPR_CM_STATE_BUSY; cm->cm_reply = reply; cm->cm_reply_data = le32toh(desc->AddressReply. ReplyFrameAddress); } break; } case MPI2_RPY_DESCRIPT_FLAGS_TARGETASSIST_SUCCESS: case MPI2_RPY_DESCRIPT_FLAGS_TARGET_COMMAND_BUFFER: case MPI2_RPY_DESCRIPT_FLAGS_RAID_ACCELERATOR_SUCCESS: default: /* Unhandled */ mpr_dprint(sc, MPR_ERROR, "Unhandled reply 0x%x\n", desc->Default.ReplyFlags); cm = NULL; break; } if (cm != NULL) { // Print Error reply frame if (cm->cm_reply) mpr_display_reply_info(sc,cm->cm_reply); mpr_complete_command(sc, cm); } desc->Words.Low = 0xffffffff; desc->Words.High = 0xffffffff; } if (pq != sc->replypostindex) { mpr_dprint(sc, MPR_TRACE, "%s sc %p writing postindex %d\n", __func__, sc, sc->replypostindex); mpr_regwrite(sc, MPI2_REPLY_POST_HOST_INDEX_OFFSET, sc->replypostindex); } return; } static void mpr_dispatch_event(struct mpr_softc *sc, uintptr_t data, MPI2_EVENT_NOTIFICATION_REPLY *reply) { struct mpr_event_handle *eh; int event, handled = 0; event = le16toh(reply->Event); TAILQ_FOREACH(eh, &sc->event_list, eh_list) { if (isset(eh->mask, event)) { eh->callback(sc, data, reply); handled++; } } if (handled == 0) mpr_dprint(sc, MPR_EVENT, "Unhandled event 0x%x\n", le16toh(event)); /* * This is the only place that the event/reply should be freed. * Anything wanting to hold onto the event data should have * already copied it into their own storage. */ mpr_free_reply(sc, data); } static void mpr_reregister_events_complete(struct mpr_softc *sc, struct mpr_command *cm) { mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); if (cm->cm_reply) MPR_DPRINT_EVENT(sc, generic, (MPI2_EVENT_NOTIFICATION_REPLY *)cm->cm_reply); mpr_free_command(sc, cm); /* next, send a port enable */ mprsas_startup(sc); } /* * For both register_events and update_events, the caller supplies a bitmap * of events that it _wants_. These functions then turn that into a bitmask * suitable for the controller. */ int mpr_register_events(struct mpr_softc *sc, uint8_t *mask, mpr_evt_callback_t *cb, void *data, struct mpr_event_handle **handle) { struct mpr_event_handle *eh; int error = 0; eh = kmalloc(sizeof(struct mpr_event_handle), M_MPR, M_WAITOK|M_ZERO); if (!eh) { mpr_dprint(sc, MPR_EVENT|MPR_ERROR, "Cannot allocate event memory\n"); return (ENOMEM); } eh->callback = cb; eh->data = data; TAILQ_INSERT_TAIL(&sc->event_list, eh, eh_list); if (mask != NULL) error = mpr_update_events(sc, eh, mask); *handle = eh; return (error); } int mpr_update_events(struct mpr_softc *sc, struct mpr_event_handle *handle, uint8_t *mask) { MPI2_EVENT_NOTIFICATION_REQUEST *evtreq; MPI2_EVENT_NOTIFICATION_REPLY *reply = NULL; struct mpr_command *cm = NULL; struct mpr_event_handle *eh; int error, i; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); if ((mask != NULL) && (handle != NULL)) bcopy(mask, &handle->mask[0], 16); memset(sc->event_mask, 0xff, 16); TAILQ_FOREACH(eh, &sc->event_list, eh_list) { for (i = 0; i < 16; i++) sc->event_mask[i] &= ~eh->mask[i]; } if ((cm = mpr_alloc_command(sc)) == NULL) return (EBUSY); evtreq = (MPI2_EVENT_NOTIFICATION_REQUEST *)cm->cm_req; evtreq->Function = MPI2_FUNCTION_EVENT_NOTIFICATION; evtreq->MsgFlags = 0; evtreq->SASBroadcastPrimitiveMasks = 0; #ifdef MPR_DEBUG_ALL_EVENTS { u_char fullmask[16]; memset(fullmask, 0x00, 16); bcopy(fullmask, (uint8_t *)&evtreq->EventMasks, 16); } #else bcopy(sc->event_mask, (uint8_t *)&evtreq->EventMasks, 16); #endif cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; cm->cm_data = NULL; error = mpr_request_polled(sc, &cm); if (cm != NULL) reply = (MPI2_EVENT_NOTIFICATION_REPLY *)cm->cm_reply; if ((reply == NULL) || (reply->IOCStatus & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS) error = ENXIO; if (reply) MPR_DPRINT_EVENT(sc, generic, reply); mpr_dprint(sc, MPR_TRACE, "%s finished error %d\n", __func__, error); if (cm != NULL) mpr_free_command(sc, cm); return (error); } static int mpr_reregister_events(struct mpr_softc *sc) { MPI2_EVENT_NOTIFICATION_REQUEST *evtreq; struct mpr_command *cm; struct mpr_event_handle *eh; int error, i; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); /* first, reregister events */ memset(sc->event_mask, 0xff, 16); TAILQ_FOREACH(eh, &sc->event_list, eh_list) { for (i = 0; i < 16; i++) sc->event_mask[i] &= ~eh->mask[i]; } if ((cm = mpr_alloc_command(sc)) == NULL) return (EBUSY); evtreq = (MPI2_EVENT_NOTIFICATION_REQUEST *)cm->cm_req; evtreq->Function = MPI2_FUNCTION_EVENT_NOTIFICATION; evtreq->MsgFlags = 0; evtreq->SASBroadcastPrimitiveMasks = 0; #ifdef MPR_DEBUG_ALL_EVENTS { u_char fullmask[16]; memset(fullmask, 0x00, 16); bcopy(fullmask, (uint8_t *)&evtreq->EventMasks, 16); } #else bcopy(sc->event_mask, (uint8_t *)&evtreq->EventMasks, 16); #endif cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; cm->cm_data = NULL; cm->cm_complete = mpr_reregister_events_complete; error = mpr_map_command(sc, cm); mpr_dprint(sc, MPR_TRACE, "%s finished with error %d\n", __func__, error); return (error); } int mpr_deregister_events(struct mpr_softc *sc, struct mpr_event_handle *handle) { TAILQ_REMOVE(&sc->event_list, handle, eh_list); kfree(handle, M_MPR); return (mpr_update_events(sc, NULL, NULL)); } /** * mpr_build_nvme_prp - This function is called for NVMe end devices to build a * native SGL (NVMe PRP). The native SGL is built starting in the first PRP entry * of the NVMe message (PRP1). If the data buffer is small enough to be described * entirely using PRP1, then PRP2 is not used. If needed, PRP2 is used to * describe a larger data buffer. If the data buffer is too large to describe * using the two PRP entriess inside the NVMe message, then PRP1 describes the * first data memory segment, and PRP2 contains a pointer to a PRP list located * elsewhere in memory to describe the remaining data memory segments. The PRP * list will be contiguous. * The native SGL for NVMe devices is a Physical Region Page (PRP). A PRP * consists of a list of PRP entries to describe a number of noncontigous * physical memory segments as a single memory buffer, just as a SGL does. Note * however, that this function is only used by the IOCTL call, so the memory * given will be guaranteed to be contiguous. There is no need to translate * non-contiguous SGL into a PRP in this case. All PRPs will describe contiguous * space that is one page size each. * * Each NVMe message contains two PRP entries. The first (PRP1) either contains * a PRP list pointer or a PRP element, depending upon the command. PRP2 contains * the second PRP element if the memory being described fits within 2 PRP * entries, or a PRP list pointer if the PRP spans more than two entries. * * A PRP list pointer contains the address of a PRP list, structured as a linear * array of PRP entries. Each PRP entry in this list describes a segment of * physical memory. * * Each 64-bit PRP entry comprises an address and an offset field. The address * always points to the beginning of a PAGE_SIZE physical memory page, and the * offset describes where within that page the memory segment begins. Only the * first element in a PRP list may contain a non-zero offest, implying that all * memory segments following the first begin at the start of a PAGE_SIZE page. * * Each PRP element normally describes a chunck of PAGE_SIZE physical memory, * with exceptions for the first and last elements in the list. If the memory * being described by the list begins at a non-zero offset within the first page, * then the first PRP element will contain a non-zero offset indicating where the * region begins within the page. The last memory segment may end before the end * of the PAGE_SIZE segment, depending upon the overall size of the memory being * described by the PRP list. * * Since PRP entries lack any indication of size, the overall data buffer length * is used to determine where the end of the data memory buffer is located, and * how many PRP entries are required to describe it. * * Returns nothing. */ void mpr_build_nvme_prp(struct mpr_softc *sc, struct mpr_command *cm, Mpi26NVMeEncapsulatedRequest_t *nvme_encap_request, void *data, uint32_t data_in_sz, uint32_t data_out_sz) { int prp_size = PRP_ENTRY_SIZE; uint64_t *prp_entry, *prp1_entry, *prp2_entry; uint64_t *prp_entry_phys, *prp_page, *prp_page_phys; uint32_t offset, entry_len, page_mask_result, page_mask; bus_addr_t paddr; size_t length; struct mpr_prp_page *prp_page_info = NULL; /* * Not all commands require a data transfer. If no data, just return * without constructing any PRP. */ if (!data_in_sz && !data_out_sz) return; /* * Set pointers to PRP1 and PRP2, which are in the NVMe command. PRP1 is * located at a 24 byte offset from the start of the NVMe command. Then * set the current PRP entry pointer to PRP1. */ prp1_entry = (uint64_t *)(nvme_encap_request->NVMe_Command + NVME_CMD_PRP1_OFFSET); prp2_entry = (uint64_t *)(nvme_encap_request->NVMe_Command + NVME_CMD_PRP2_OFFSET); prp_entry = prp1_entry; /* * For the PRP entries, use the specially allocated buffer of * contiguous memory. PRP Page allocation failures should not happen * because there should be enough PRP page buffers to account for the * possible NVMe QDepth. */ prp_page_info = mpr_alloc_prp_page(sc); KASSERT(prp_page_info != NULL, ("%s: There are no PRP Pages left to be " "used for building a native NVMe SGL.\n", __func__)); prp_page = (uint64_t *)prp_page_info->prp_page; prp_page_phys = (uint64_t *)(uintptr_t)prp_page_info->prp_page_busaddr; /* * Insert the allocated PRP page into the command's PRP page list. This * will be freed when the command is freed. */ TAILQ_INSERT_TAIL(&cm->cm_prp_page_list, prp_page_info, prp_page_link); /* * Check if we are within 1 entry of a page boundary we don't want our * first entry to be a PRP List entry. */ page_mask = PAGE_SIZE - 1; page_mask_result = (uintptr_t)((uint8_t *)prp_page + prp_size) & page_mask; if (!page_mask_result) { /* Bump up to next page boundary. */ prp_page = (uint64_t *)((uint8_t *)prp_page + prp_size); prp_page_phys = (uint64_t *)((uint8_t *)prp_page_phys + prp_size); } /* * Set PRP physical pointer, which initially points to the current PRP * DMA memory page. */ prp_entry_phys = prp_page_phys; /* Get physical address and length of the data buffer. */ paddr = (bus_addr_t)(uintptr_t)data; if (data_in_sz) length = data_in_sz; else length = data_out_sz; /* Loop while the length is not zero. */ while (length) { /* * Check if we need to put a list pointer here if we are at page * boundary - prp_size (8 bytes). */ page_mask_result = (uintptr_t)((uint8_t *)prp_entry_phys + prp_size) & page_mask; if (!page_mask_result) { /* * This is the last entry in a PRP List, so we need to * put a PRP list pointer here. What this does is: * - bump the current memory pointer to the next * address, which will be the next full page. * - set the PRP Entry to point to that page. This is * now the PRP List pointer. * - bump the PRP Entry pointer the start of the next * page. Since all of this PRP memory is contiguous, * no need to get a new page - it's just the next * address. */ prp_entry_phys++; *prp_entry = htole64((uint64_t)(uintptr_t)prp_entry_phys); prp_entry++; } /* Need to handle if entry will be part of a page. */ offset = (uint32_t)paddr & page_mask; entry_len = PAGE_SIZE - offset; if (prp_entry == prp1_entry) { /* * Must fill in the first PRP pointer (PRP1) before * moving on. */ *prp1_entry = htole64((uint64_t)paddr); /* * Now point to the second PRP entry within the * command (PRP2). */ prp_entry = prp2_entry; } else if (prp_entry == prp2_entry) { /* * Should the PRP2 entry be a PRP List pointer or just a * regular PRP pointer? If there is more than one more * page of data, must use a PRP List pointer. */ if (length > PAGE_SIZE) { /* * PRP2 will contain a PRP List pointer because * more PRP's are needed with this command. The * list will start at the beginning of the * contiguous buffer. */ *prp2_entry = htole64( (uint64_t)(uintptr_t)prp_entry_phys); /* * The next PRP Entry will be the start of the * first PRP List. */ prp_entry = prp_page; } else { /* * After this, the PRP Entries are complete. * This command uses 2 PRP's and no PRP list. */ *prp2_entry = htole64((uint64_t)paddr); } } else { /* * Put entry in list and bump the addresses. * * After PRP1 and PRP2 are filled in, this will fill in * all remaining PRP entries in a PRP List, one per each * time through the loop. */ *prp_entry = htole64((uint64_t)paddr); prp_entry++; prp_entry_phys++; } /* * Bump the phys address of the command's data buffer by the * entry_len. */ paddr += entry_len; /* Decrement length accounting for last partial page. */ if (entry_len > length) length = 0; else length -= entry_len; } } /* * mpr_check_pcie_native_sgl - This function is called for PCIe end devices to * determine if the driver needs to build a native SGL. If so, that native SGL * is built in the contiguous buffers allocated especially for PCIe SGL * creation. If the driver will not build a native SGL, return TRUE and a * normal IEEE SGL will be built. Currently this routine supports NVMe devices * only. * * Returns FALSE (0) if native SGL was built, TRUE (1) if no SGL was built. */ static int mpr_check_pcie_native_sgl(struct mpr_softc *sc, struct mpr_command *cm, bus_dma_segment_t *segs, int segs_left) { uint32_t i, sge_dwords, length, offset, entry_len; uint32_t num_entries, buff_len = 0, sges_in_segment; uint32_t page_mask, page_mask_result, *curr_buff; uint32_t *ptr_sgl, *ptr_first_sgl, first_page_offset; uint32_t first_page_data_size, end_residual; uint64_t *msg_phys; bus_addr_t paddr; int build_native_sgl = 0, first_prp_entry; int prp_size = PRP_ENTRY_SIZE; Mpi25IeeeSgeChain64_t *main_chain_element = NULL; struct mpr_prp_page *prp_page_info = NULL; mpr_dprint(sc, MPR_TRACE, "%s\n", __func__); /* * Add up the sizes of each segment length to get the total transfer * size, which will be checked against the Maximum Data Transfer Size. * If the data transfer length exceeds the MDTS for this device, just * return 1 so a normal IEEE SGL will be built. F/W will break the I/O * up into multiple I/O's. [nvme_mdts = 0 means unlimited] */ for (i = 0; i < segs_left; i++) buff_len += htole32(segs[i].ds_len); if ((cm->cm_targ->MDTS > 0) && (buff_len > cm->cm_targ->MDTS)) return 1; /* Create page_mask (to get offset within page) */ page_mask = PAGE_SIZE - 1; /* * Check if the number of elements exceeds the max number that can be * put in the main message frame (H/W can only translate an SGL that * is contained entirely in the main message frame). */ sges_in_segment = (sc->reqframesz - offsetof(Mpi25SCSIIORequest_t, SGL)) / sizeof(MPI25_SGE_IO_UNION); if (segs_left > sges_in_segment) build_native_sgl = 1; else { /* * NVMe uses one PRP for each physical page (or part of physical * page). * if 4 pages or less then IEEE is OK * if > 5 pages then we need to build a native SGL * if > 4 and <= 5 pages, then check the physical address of * the first SG entry, then if this first size in the page * is >= the residual beyond 4 pages then use IEEE, * otherwise use native SGL */ if (buff_len > (PAGE_SIZE * 5)) build_native_sgl = 1; else if ((buff_len > (PAGE_SIZE * 4)) && (buff_len <= (PAGE_SIZE * 5)) ) { msg_phys = (uint64_t *)(uintptr_t)segs[0].ds_addr; first_page_offset = ((uint32_t)(uint64_t)(uintptr_t)msg_phys & page_mask); first_page_data_size = PAGE_SIZE - first_page_offset; end_residual = buff_len % PAGE_SIZE; /* * If offset into first page pushes the end of the data * beyond end of the 5th page, we need the extra PRP * list. */ if (first_page_data_size < end_residual) build_native_sgl = 1; /* * Check if first SG entry size is < residual beyond 4 * pages. */ if (htole32(segs[0].ds_len) < (buff_len - (PAGE_SIZE * 4))) build_native_sgl = 1; } } /* check if native SGL is needed */ if (!build_native_sgl) return 1; /* * Native SGL is needed. * Put a chain element in main message frame that points to the first * chain buffer. * * NOTE: The ChainOffset field must be 0 when using a chain pointer to * a native SGL. */ /* Set main message chain element pointer */ main_chain_element = (pMpi25IeeeSgeChain64_t)cm->cm_sge; /* * For NVMe the chain element needs to be the 2nd SGL entry in the main * message. */ main_chain_element = (Mpi25IeeeSgeChain64_t *) ((uint8_t *)main_chain_element + sizeof(MPI25_IEEE_SGE_CHAIN64)); /* * For the PRP entries, use the specially allocated buffer of * contiguous memory. PRP Page allocation failures should not happen * because there should be enough PRP page buffers to account for the * possible NVMe QDepth. */ prp_page_info = mpr_alloc_prp_page(sc); KASSERT(prp_page_info != NULL, ("%s: There are no PRP Pages left to be " "used for building a native NVMe SGL.\n", __func__)); curr_buff = (uint32_t *)prp_page_info->prp_page; msg_phys = (uint64_t *)(uintptr_t)prp_page_info->prp_page_busaddr; /* * Insert the allocated PRP page into the command's PRP page list. This * will be freed when the command is freed. */ TAILQ_INSERT_TAIL(&cm->cm_prp_page_list, prp_page_info, prp_page_link); /* * Check if we are within 1 entry of a page boundary we don't want our * first entry to be a PRP List entry. */ page_mask_result = (uintptr_t)((uint8_t *)curr_buff + prp_size) & page_mask; if (!page_mask_result) { /* Bump up to next page boundary. */ curr_buff = (uint32_t *)((uint8_t *)curr_buff + prp_size); msg_phys = (uint64_t *)((uint8_t *)msg_phys + prp_size); } /* Fill in the chain element and make it an NVMe segment type. */ main_chain_element->Address.High = htole32((uint32_t)((uint64_t)(uintptr_t)msg_phys >> 32)); main_chain_element->Address.Low = htole32((uint32_t)(uintptr_t)msg_phys); main_chain_element->NextChainOffset = 0; main_chain_element->Flags = MPI2_IEEE_SGE_FLAGS_CHAIN_ELEMENT | MPI2_IEEE_SGE_FLAGS_SYSTEM_ADDR | MPI26_IEEE_SGE_FLAGS_NSF_NVME_PRP; /* Set SGL pointer to start of contiguous PCIe buffer. */ ptr_sgl = curr_buff; sge_dwords = 2; num_entries = 0; /* * NVMe has a very convoluted PRP format. One PRP is required for each * page or partial page. We need to split up OS SG entries if they are * longer than one page or cross a page boundary. We also have to insert * a PRP list pointer entry as the last entry in each physical page of * the PRP list. * * NOTE: The first PRP "entry" is actually placed in the first SGL entry * in the main message in IEEE 64 format. The 2nd entry in the main * message is the chain element, and the rest of the PRP entries are * built in the contiguous PCIe buffer. */ first_prp_entry = 1; ptr_first_sgl = (uint32_t *)cm->cm_sge; for (i = 0; i < segs_left; i++) { /* Get physical address and length of this SG entry. */ paddr = segs[i].ds_addr; length = segs[i].ds_len; /* * Check whether a given SGE buffer lies on a non-PAGED * boundary if this is not the first page. If so, this is not * expected so have FW build the SGL. */ if ((i != 0) && (((uint32_t)paddr & page_mask) != 0)) { mpr_dprint(sc, MPR_ERROR, "Unaligned SGE while " "building NVMe PRPs, low address is 0x%x\n", (uint32_t)paddr); return 1; } /* Apart from last SGE, if any other SGE boundary is not page * aligned then it means that hole exists. Existence of hole * leads to data corruption. So fallback to IEEE SGEs. */ if (i != (segs_left - 1)) { if (((uint32_t)paddr + length) & page_mask) { mpr_dprint(sc, MPR_ERROR, "Unaligned SGE " "boundary while building NVMe PRPs, low " "address: 0x%x and length: %u\n", (uint32_t)paddr, length); return 1; } } /* Loop while the length is not zero. */ while (length) { /* * Check if we need to put a list pointer here if we are * at page boundary - prp_size. */ page_mask_result = (uintptr_t)((uint8_t *)ptr_sgl + prp_size) & page_mask; if (!page_mask_result) { /* * Need to put a PRP list pointer here. */ msg_phys = (uint64_t *)((uint8_t *)msg_phys + prp_size); *ptr_sgl = htole32((uintptr_t)msg_phys); *(ptr_sgl+1) = htole32((uint64_t)(uintptr_t) msg_phys >> 32); ptr_sgl += sge_dwords; num_entries++; } /* Need to handle if entry will be part of a page. */ offset = (uint32_t)paddr & page_mask; entry_len = PAGE_SIZE - offset; if (first_prp_entry) { /* * Put IEEE entry in first SGE in main message. * (Simple element, System addr, not end of * list.) */ *ptr_first_sgl = htole32((uint32_t)paddr); *(ptr_first_sgl + 1) = htole32((uint32_t)((uint64_t)paddr >> 32)); *(ptr_first_sgl + 2) = htole32(entry_len); *(ptr_first_sgl + 3) = 0; /* No longer the first PRP entry. */ first_prp_entry = 0; } else { /* Put entry in list. */ *ptr_sgl = htole32((uint32_t)paddr); *(ptr_sgl + 1) = htole32((uint32_t)((uint64_t)paddr >> 32)); /* Bump ptr_sgl, msg_phys, and num_entries. */ ptr_sgl += sge_dwords; msg_phys = (uint64_t *)((uint8_t *)msg_phys + prp_size); num_entries++; } /* Bump the phys address by the entry_len. */ paddr += entry_len; /* Decrement length accounting for last partial page. */ if (entry_len > length) length = 0; else length -= entry_len; } } /* Set chain element Length. */ main_chain_element->Length = htole32(num_entries * prp_size); /* Return 0, indicating we built a native SGL. */ return 0; } /* * Add a chain element as the next SGE for the specified command. * Reset cm_sge and cm_sgesize to indicate all the available space. Chains are * only required for IEEE commands. Therefore there is no code for commands * that have the MPR_CM_FLAGS_SGE_SIMPLE flag set (and those commands * shouldn't be requesting chains). */ static int mpr_add_chain(struct mpr_command *cm, int segsleft) { struct mpr_softc *sc = cm->cm_sc; MPI2_REQUEST_HEADER *req; MPI25_IEEE_SGE_CHAIN64 *ieee_sgc; struct mpr_chain *chain; int sgc_size, current_segs, rem_segs, segs_per_frame; uint8_t next_chain_offset = 0; /* * Fail if a command is requesting a chain for SIMPLE SGE's. For SAS3 * only IEEE commands should be requesting chains. Return some error * code other than 0. */ if (cm->cm_flags & MPR_CM_FLAGS_SGE_SIMPLE) { mpr_dprint(sc, MPR_ERROR, "A chain element cannot be added to " "an MPI SGL.\n"); return(ENOBUFS); } sgc_size = sizeof(MPI25_IEEE_SGE_CHAIN64); if (cm->cm_sglsize < sgc_size) panic("MPR: Need SGE Error Code\n"); chain = mpr_alloc_chain(cm->cm_sc); if (chain == NULL) return (ENOBUFS); /* * Note: a double-linked list is used to make it easier to walk for * debugging. */ TAILQ_INSERT_TAIL(&cm->cm_chain_list, chain, chain_link); /* * Need to know if the number of frames left is more than 1 or not. If * more than 1 frame is required, NextChainOffset will need to be set, * which will just be the last segment of the frame. */ rem_segs = 0; if (cm->cm_sglsize < (sgc_size * segsleft)) { /* * rem_segs is the number of segements remaining after the * segments that will go into the current frame. Since it is * known that at least one more frame is required, account for * the chain element. To know if more than one more frame is * required, just check if there will be a remainder after using * the current frame (with this chain) and the next frame. If * so the NextChainOffset must be the last element of the next * frame. */ current_segs = (cm->cm_sglsize / sgc_size) - 1; rem_segs = segsleft - current_segs; segs_per_frame = sc->chain_frame_size / sgc_size; if (rem_segs > segs_per_frame) { next_chain_offset = segs_per_frame - 1; } } ieee_sgc = &((MPI25_SGE_IO_UNION *)cm->cm_sge)->IeeeChain; ieee_sgc->Length = next_chain_offset ? htole32((uint32_t)sc->chain_frame_size) : htole32((uint32_t)rem_segs * (uint32_t)sgc_size); ieee_sgc->NextChainOffset = next_chain_offset; ieee_sgc->Flags = (MPI2_IEEE_SGE_FLAGS_CHAIN_ELEMENT | MPI2_IEEE_SGE_FLAGS_SYSTEM_ADDR); ieee_sgc->Address.Low = htole32(chain->chain_busaddr); ieee_sgc->Address.High = htole32(chain->chain_busaddr >> 32); cm->cm_sge = &((MPI25_SGE_IO_UNION *)chain->chain)->IeeeSimple; req = (MPI2_REQUEST_HEADER *)cm->cm_req; req->ChainOffset = (sc->chain_frame_size - sgc_size) >> 4; cm->cm_sglsize = sc->chain_frame_size; return (0); } /* * Add one scatter-gather element to the scatter-gather list for a command. * Maintain cm_sglsize and cm_sge as the remaining size and pointer to the * next SGE to fill in, respectively. In Gen3, the MPI SGL does not have a * chain, so don't consider any chain additions. */ int mpr_push_sge(struct mpr_command *cm, MPI2_SGE_SIMPLE64 *sge, size_t len, int segsleft) { uint32_t saved_buf_len, saved_address_low, saved_address_high; u32 sge_flags; /* * case 1: >=1 more segment, no room for anything (error) * case 2: 1 more segment and enough room for it */ if (cm->cm_sglsize < (segsleft * sizeof(MPI2_SGE_SIMPLE64))) { mpr_dprint(cm->cm_sc, MPR_ERROR, "%s: warning: Not enough room for MPI SGL in frame.\n", __func__); return(ENOBUFS); } KASSERT(segsleft == 1, ("segsleft cannot be more than 1 for an MPI SGL; segsleft = %d\n", segsleft)); /* * There is one more segment left to add for the MPI SGL and there is * enough room in the frame to add it. This is the normal case because * MPI SGL's don't have chains, otherwise something is wrong. * * If this is a bi-directional request, need to account for that * here. Save the pre-filled sge values. These will be used * either for the 2nd SGL or for a single direction SGL. If * cm_out_len is non-zero, this is a bi-directional request, so * fill in the OUT SGL first, then the IN SGL, otherwise just * fill in the IN SGL. Note that at this time, when filling in * 2 SGL's for a bi-directional request, they both use the same * DMA buffer (same cm command). */ saved_buf_len = sge->FlagsLength & 0x00FFFFFF; saved_address_low = sge->Address.Low; saved_address_high = sge->Address.High; if (cm->cm_out_len) { sge->FlagsLength = cm->cm_out_len | ((uint32_t)(MPI2_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_SGE_FLAGS_END_OF_BUFFER | MPI2_SGE_FLAGS_HOST_TO_IOC | MPI2_SGE_FLAGS_64_BIT_ADDRESSING) << MPI2_SGE_FLAGS_SHIFT); cm->cm_sglsize -= len; /* Endian Safe code */ sge_flags = sge->FlagsLength; sge->FlagsLength = htole32(sge_flags); sge->Address.High = htole32(sge->Address.High); sge->Address.Low = htole32(sge->Address.Low); bcopy(sge, cm->cm_sge, len); cm->cm_sge = (MPI2_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + len); } sge->FlagsLength = saved_buf_len | ((uint32_t)(MPI2_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_SGE_FLAGS_END_OF_BUFFER | MPI2_SGE_FLAGS_LAST_ELEMENT | MPI2_SGE_FLAGS_END_OF_LIST | MPI2_SGE_FLAGS_64_BIT_ADDRESSING) << MPI2_SGE_FLAGS_SHIFT); if (cm->cm_flags & MPR_CM_FLAGS_DATAIN) { sge->FlagsLength |= ((uint32_t)(MPI2_SGE_FLAGS_IOC_TO_HOST) << MPI2_SGE_FLAGS_SHIFT); } else { sge->FlagsLength |= ((uint32_t)(MPI2_SGE_FLAGS_HOST_TO_IOC) << MPI2_SGE_FLAGS_SHIFT); } sge->Address.Low = saved_address_low; sge->Address.High = saved_address_high; cm->cm_sglsize -= len; /* Endian Safe code */ sge_flags = sge->FlagsLength; sge->FlagsLength = htole32(sge_flags); sge->Address.High = htole32(sge->Address.High); sge->Address.Low = htole32(sge->Address.Low); bcopy(sge, cm->cm_sge, len); cm->cm_sge = (MPI2_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + len); return (0); } /* * Add one IEEE scatter-gather element (chain or simple) to the IEEE scatter- * gather list for a command. Maintain cm_sglsize and cm_sge as the * remaining size and pointer to the next SGE to fill in, respectively. */ int mpr_push_ieee_sge(struct mpr_command *cm, void *sgep, int segsleft) { MPI2_IEEE_SGE_SIMPLE64 *sge = sgep; int error, ieee_sge_size = sizeof(MPI25_SGE_IO_UNION); uint32_t saved_buf_len, saved_address_low, saved_address_high; uint32_t sge_length; /* * case 1: No room for chain or segment (error). * case 2: Two or more segments left but only room for chain. * case 3: Last segment and room for it, so set flags. */ /* * There should be room for at least one element, or there is a big * problem. */ if (cm->cm_sglsize < ieee_sge_size) panic("MPR: Need SGE Error Code\n"); if ((segsleft >= 2) && (cm->cm_sglsize < (ieee_sge_size * 2))) { if ((error = mpr_add_chain(cm, segsleft)) != 0) return (error); } if (segsleft == 1) { /* * If this is a bi-directional request, need to account for that * here. Save the pre-filled sge values. These will be used * either for the 2nd SGL or for a single direction SGL. If * cm_out_len is non-zero, this is a bi-directional request, so * fill in the OUT SGL first, then the IN SGL, otherwise just * fill in the IN SGL. Note that at this time, when filling in * 2 SGL's for a bi-directional request, they both use the same * DMA buffer (same cm command). */ saved_buf_len = sge->Length; saved_address_low = sge->Address.Low; saved_address_high = sge->Address.High; if (cm->cm_out_len) { sge->Length = cm->cm_out_len; sge->Flags = (MPI2_IEEE_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_IEEE_SGE_FLAGS_SYSTEM_ADDR); cm->cm_sglsize -= ieee_sge_size; /* Endian Safe code */ sge_length = sge->Length; sge->Length = htole32(sge_length); sge->Address.High = htole32(sge->Address.High); sge->Address.Low = htole32(sge->Address.Low); bcopy(sgep, cm->cm_sge, ieee_sge_size); cm->cm_sge = (MPI25_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + ieee_sge_size); } sge->Length = saved_buf_len; sge->Flags = (MPI2_IEEE_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_IEEE_SGE_FLAGS_SYSTEM_ADDR | MPI25_IEEE_SGE_FLAGS_END_OF_LIST); sge->Address.Low = saved_address_low; sge->Address.High = saved_address_high; } cm->cm_sglsize -= ieee_sge_size; /* Endian Safe code */ sge_length = sge->Length; sge->Length = htole32(sge_length); sge->Address.High = htole32(sge->Address.High); sge->Address.Low = htole32(sge->Address.Low); bcopy(sgep, cm->cm_sge, ieee_sge_size); cm->cm_sge = (MPI25_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + ieee_sge_size); return (0); } /* * Add one dma segment to the scatter-gather list for a command. */ int mpr_add_dmaseg(struct mpr_command *cm, vm_paddr_t pa, size_t len, u_int flags, int segsleft) { MPI2_SGE_SIMPLE64 sge; MPI2_IEEE_SGE_SIMPLE64 ieee_sge; if (!(cm->cm_flags & MPR_CM_FLAGS_SGE_SIMPLE)) { ieee_sge.Flags = (MPI2_IEEE_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_IEEE_SGE_FLAGS_SYSTEM_ADDR); ieee_sge.Length = len; mpr_from_u64(pa, &ieee_sge.Address); return (mpr_push_ieee_sge(cm, &ieee_sge, segsleft)); } else { /* * This driver always uses 64-bit address elements for * simplicity. */ flags |= MPI2_SGE_FLAGS_SIMPLE_ELEMENT | MPI2_SGE_FLAGS_64_BIT_ADDRESSING; /* Set Endian safe macro in mpr_push_sge */ sge.FlagsLength = len | (flags << MPI2_SGE_FLAGS_SHIFT); mpr_from_u64(pa, &sge.Address); return (mpr_push_sge(cm, &sge, sizeof sge, segsleft)); } } static void mpr_data_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) { struct mpr_softc *sc; struct mpr_command *cm; u_int i, dir, sflags; cm = (struct mpr_command *)arg; sc = cm->cm_sc; /* * In this case, just print out a warning and let the chip tell the * user they did the wrong thing. */ if ((cm->cm_max_segs != 0) && (nsegs > cm->cm_max_segs)) { mpr_dprint(sc, MPR_ERROR, "%s: warning: busdma returned %d " "segments, more than the %d allowed\n", __func__, nsegs, cm->cm_max_segs); } /* * Set up DMA direction flags. Bi-directional requests are also handled * here. In that case, both direction flags will be set. */ sflags = 0; if (cm->cm_flags & MPR_CM_FLAGS_SMP_PASS) { /* * We have to add a special case for SMP passthrough, there * is no easy way to generically handle it. The first * S/G element is used for the command (therefore the * direction bit needs to be set). The second one is used * for the reply. We'll leave it to the caller to make * sure we only have two buffers. */ /* * Even though the busdma man page says it doesn't make * sense to have both direction flags, it does in this case. * We have one s/g element being accessed in each direction. */ dir = BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD; /* * Set the direction flag on the first buffer in the SMP * passthrough request. We'll clear it for the second one. */ sflags |= MPI2_SGE_FLAGS_DIRECTION | MPI2_SGE_FLAGS_END_OF_BUFFER; } else if (cm->cm_flags & MPR_CM_FLAGS_DATAOUT) { sflags |= MPI2_SGE_FLAGS_HOST_TO_IOC; dir = BUS_DMASYNC_PREWRITE; } else dir = BUS_DMASYNC_PREREAD; /* Check if a native SG list is needed for an NVMe PCIe device. */ if (cm->cm_targ && cm->cm_targ->is_nvme && mpr_check_pcie_native_sgl(sc, cm, segs, nsegs) == 0) { /* A native SG list was built, skip to end. */ goto out; } for (i = 0; i < nsegs; i++) { if ((cm->cm_flags & MPR_CM_FLAGS_SMP_PASS) && (i != 0)) { sflags &= ~MPI2_SGE_FLAGS_DIRECTION; } error = mpr_add_dmaseg(cm, segs[i].ds_addr, segs[i].ds_len, sflags, nsegs - i); if (error != 0) { /* Resource shortage, roll back! */ if (ratecheck(&sc->lastfail, &mpr_chainfail_interval)) mpr_dprint(sc, MPR_INFO, "Out of chain frames, " "consider increasing hw.mpr.max_chains.\n"); cm->cm_flags |= MPR_CM_FLAGS_CHAIN_FAILED; mpr_complete_command(sc, cm); return; } } out: bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap, dir); mpr_enqueue_request(sc, cm); return; } static void mpr_data_cb2(void *arg, bus_dma_segment_t *segs, int nsegs, bus_size_t mapsize, int error) { mpr_data_cb(arg, segs, nsegs, error); } /* * This is the routine to enqueue commands ansynchronously. * Note that the only error path here is from bus_dmamap_load(), which can * return EINPROGRESS if it is waiting for resources. Other than this, it's * assumed that if you have a command in-hand, then you have enough credits * to use it. */ int mpr_map_command(struct mpr_softc *sc, struct mpr_command *cm) { int error = 0; if (cm->cm_flags & MPR_CM_FLAGS_USE_UIO) { error = bus_dmamap_load_uio(sc->buffer_dmat, cm->cm_dmamap, &cm->cm_uio, mpr_data_cb2, cm, 0); } else if (cm->cm_flags & MPR_CM_FLAGS_USE_CCB) { error = bus_dmamap_load_ccb(sc->buffer_dmat, cm->cm_dmamap, cm->cm_data, mpr_data_cb, cm, 0); } else if ((cm->cm_data != NULL) && (cm->cm_length != 0)) { error = bus_dmamap_load(sc->buffer_dmat, cm->cm_dmamap, cm->cm_data, cm->cm_length, mpr_data_cb, cm, 0); } else { /* Add a zero-length element as needed */ if (cm->cm_sge != NULL) mpr_add_dmaseg(cm, 0, 0, 0, 1); mpr_enqueue_request(sc, cm); } return (error); } /* * This is the routine to enqueue commands synchronously. An error of * EINPROGRESS from mpr_map_command() is ignored since the command will * be executed and enqueued automatically. Other errors come from msleep(). */ int mpr_wait_command(struct mpr_softc *sc, struct mpr_command **cmp, int timeout, int sleep_flag) { int error, rc; struct timeval cur_time, start_time; struct mpr_command *cm = *cmp; if (sc->mpr_flags & MPR_FLAGS_DIAGRESET) return EBUSY; cm->cm_complete = NULL; cm->cm_flags |= (MPR_CM_FLAGS_WAKEUP + MPR_CM_FLAGS_POLLED); error = mpr_map_command(sc, cm); if ((error != 0) && (error != EINPROGRESS)) return (error); // Check for context and wait for 50 mSec at a time until time has // expired or the command has finished. If msleep can't be used, need // to poll. getmicrouptime(&start_time); if (lockowned(&sc->mpr_lock) && sleep_flag == CAN_SLEEP) { error = lksleep(cm, &sc->mpr_lock, 0, "mprwait", timeout*hz); if (error == EWOULDBLOCK) { /* * Record the actual elapsed time in the case of a * timeout for the message below. */ getmicrouptime(&cur_time); timevalsub(&cur_time, &start_time); } } else { while ((cm->cm_flags & MPR_CM_FLAGS_COMPLETE) == 0) { mpr_intr_locked(sc); if (sleep_flag == CAN_SLEEP) tsleep(mpr_wait_command, 0, "mprwait", hz < 20 ? 1 : hz / 20); else DELAY(50000); getmicrouptime(&cur_time); timevalsub(&cur_time, &start_time); if (cur_time.tv_sec > timeout) { error = EWOULDBLOCK; break; } } } if (error == EWOULDBLOCK) { mpr_dprint(sc, MPR_FAULT, "Calling Reinit from %s, timeout=%d," " elapsed=%jd\n", __func__, timeout, (intmax_t)cur_time.tv_sec); rc = mpr_reinit(sc); mpr_dprint(sc, MPR_FAULT, "Reinit %s\n", (rc == 0) ? "success" : "failed"); if (sc->mpr_flags & MPR_FLAGS_REALLOCATED) { /* * Tell the caller that we freed the command in a * reinit. */ *cmp = NULL; } error = ETIMEDOUT; } return (error); } /* * This is the routine to enqueue a command synchonously and poll for * completion. Its use should be rare. */ int mpr_request_polled(struct mpr_softc *sc, struct mpr_command **cmp) { int error, rc; struct timeval cur_time, start_time; struct mpr_command *cm = *cmp; error = 0; cm->cm_flags |= MPR_CM_FLAGS_POLLED; cm->cm_complete = NULL; mpr_map_command(sc, cm); getmicrouptime(&start_time); while ((cm->cm_flags & MPR_CM_FLAGS_COMPLETE) == 0) { mpr_intr_locked(sc); if (lockowned(&sc->mpr_lock)) lksleep(&sc->msleep_fake_chan, &sc->mpr_lock, 0, "mprpoll", hz < 20 ? 1 : hz / 20); else tsleep(mpr_request_polled, 0, "mprpoll", hz < 20 ? 1 : hz / 20); /* * Check for real-time timeout and fail if more than 60 seconds. */ getmicrouptime(&cur_time); timevalsub(&cur_time, &start_time); if (cur_time.tv_sec > 60) { mpr_dprint(sc, MPR_FAULT, "polling failed\n"); error = ETIMEDOUT; break; } } if (error) { mpr_dprint(sc, MPR_FAULT, "Calling Reinit from %s\n", __func__); rc = mpr_reinit(sc); mpr_dprint(sc, MPR_FAULT, "Reinit %s\n", (rc == 0) ? "success" : "failed"); if (sc->mpr_flags & MPR_FLAGS_REALLOCATED) { /* * Tell the caller that we freed the command in a * reinit. */ *cmp = NULL; } } return (error); } /* * The MPT driver had a verbose interface for config pages. In this driver, * reduce it to much simpler terms, similar to the Linux driver. */ int mpr_read_config_page(struct mpr_softc *sc, struct mpr_config_params *params) { MPI2_CONFIG_REQUEST *req; struct mpr_command *cm; int error; if (sc->mpr_flags & MPR_FLAGS_BUSY) { return (EBUSY); } cm = mpr_alloc_command(sc); if (cm == NULL) { return (EBUSY); } req = (MPI2_CONFIG_REQUEST *)cm->cm_req; req->Function = MPI2_FUNCTION_CONFIG; req->Action = params->action; req->SGLFlags = 0; req->ChainOffset = 0; req->PageAddress = params->page_address; if (params->hdr.Struct.PageType == MPI2_CONFIG_PAGETYPE_EXTENDED) { MPI2_CONFIG_EXTENDED_PAGE_HEADER *hdr; hdr = ¶ms->hdr.Ext; req->ExtPageType = hdr->ExtPageType; req->ExtPageLength = hdr->ExtPageLength; req->Header.PageType = MPI2_CONFIG_PAGETYPE_EXTENDED; req->Header.PageLength = 0; /* Must be set to zero */ req->Header.PageNumber = hdr->PageNumber; req->Header.PageVersion = hdr->PageVersion; } else { MPI2_CONFIG_PAGE_HEADER *hdr; hdr = ¶ms->hdr.Struct; req->Header.PageType = hdr->PageType; req->Header.PageNumber = hdr->PageNumber; req->Header.PageLength = hdr->PageLength; req->Header.PageVersion = hdr->PageVersion; } cm->cm_data = params->buffer; cm->cm_length = params->length; if (cm->cm_data != NULL) { cm->cm_sge = &req->PageBufferSGE; cm->cm_sglsize = sizeof(MPI2_SGE_IO_UNION); cm->cm_flags = MPR_CM_FLAGS_SGE_SIMPLE | MPR_CM_FLAGS_DATAIN; } else cm->cm_sge = NULL; cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE; cm->cm_complete_data = params; if (params->callback != NULL) { cm->cm_complete = mpr_config_complete; return (mpr_map_command(sc, cm)); } else { error = mpr_wait_command(sc, &cm, 0, CAN_SLEEP); if (error) { mpr_dprint(sc, MPR_FAULT, "Error %d reading config page\n", error); if (cm != NULL) mpr_free_command(sc, cm); return (error); } mpr_config_complete(sc, cm); } return (0); } int mpr_write_config_page(struct mpr_softc *sc, struct mpr_config_params *params) { return (EINVAL); } static void mpr_config_complete(struct mpr_softc *sc, struct mpr_command *cm) { MPI2_CONFIG_REPLY *reply; struct mpr_config_params *params; MPR_FUNCTRACE(sc); params = cm->cm_complete_data; if (cm->cm_data != NULL) { bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap, BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->buffer_dmat, cm->cm_dmamap); } /* * XXX KDM need to do more error recovery? This results in the * device in question not getting probed. */ if ((cm->cm_flags & MPR_CM_FLAGS_ERROR_MASK) != 0) { params->status = MPI2_IOCSTATUS_BUSY; goto done; } reply = (MPI2_CONFIG_REPLY *)cm->cm_reply; if (reply == NULL) { params->status = MPI2_IOCSTATUS_BUSY; goto done; } params->status = reply->IOCStatus; if (params->hdr.Struct.PageType == MPI2_CONFIG_PAGETYPE_EXTENDED) { params->hdr.Ext.ExtPageType = reply->ExtPageType; params->hdr.Ext.ExtPageLength = reply->ExtPageLength; params->hdr.Ext.PageType = reply->Header.PageType; params->hdr.Ext.PageNumber = reply->Header.PageNumber; params->hdr.Ext.PageVersion = reply->Header.PageVersion; } else { params->hdr.Struct.PageType = reply->Header.PageType; params->hdr.Struct.PageNumber = reply->Header.PageNumber; params->hdr.Struct.PageLength = reply->Header.PageLength; params->hdr.Struct.PageVersion = reply->Header.PageVersion; } done: mpr_free_command(sc, cm); if (params->callback != NULL) params->callback(sc, params); return; } |