Loading...
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 | /* * Copyright (c) 2015-2022 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * * This file contains Original Code and/or Modifications of Original Code * as defined in and that are subject to the Apple Public Source License * Version 2.0 (the 'License'). You may not use this file except in * compliance with the License. The rights granted to you under the License * may not be used to create, or enable the creation or redistribution of, * unlawful or unlicensed copies of an Apple operating system, or to * circumvent, violate, or enable the circumvention or violation of, any * terms of an Apple operating system software license agreement. * * Please obtain a copy of the License at * http://www.opensource.apple.com/apsl/ and read it before using this file. * * The Original Code and all software distributed under the License are * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. * Please see the License for the specific language governing rights and * limitations under the License. * * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ */ /* * The netif nexus domain has two domain providers: native and compat, with * the latter being the default provider of this domain. The compat provider * has special handlers for NXCFG_CMD_ATTACH and NXCFG_CMD_DETACH, etc. * * A netif nexus instance can be in a native or compat mode; in either case, * it is associated with two instances of a nexus_adapter structure, and allows * at most two channels opened to the nexus. Two two adapters correspond to * host and device ports, respectively. * * By itself, a netif nexus isn't associated with a network interface. The * association happens by attaching a network interface to the nexus instance. * A channel can only be successfully opened to a netif nexus after it has an * interface attached to it. * * During an attach, the interface is marked as Skywalk-capable, and its ifnet * structure refers to the attached netif nexus adapter via its if_na field. * The nexus also holds a reference to the interface on its na_ifp field. Note * that attaching to a netif_compat nexus does not alter the input/output data * path, nor does it remove any of the interface's hardware offload flags. It * merely associates the interface and netif nexus together. * * During a detach, the above references are dropped and the fields are cleared; * the interface is also marked as non-Skywalk-capable. This detach can happen * explicitly via a command down the nexus, or implicitly when the nexus goes * away (assuming there's no channel opened to it.) * * A userland channel can be opened to a netif nexus via the usual ch_open() * way, assuming the nexus provider is setup to allow access for the userland * process (either by binding the nexus port to PID, etc. or by creating the * nexus in the anonymous mode.) * * Alternatively, a kernel channel can also be opened to it by some kernel * subsystem, via ch_open_special(), e.g. by the flowswitch. Kernel channels * don't have any task mapping created, and the flag CHANF_KERNEL is used to * indicate that. * * Opening a channel to the host port of a native or compat netif causes the * ifnet output path to be redirected to nx_netif_host_transmit(). We also, * at present, disable any hardware offload features. * * Opening a channel to the device port of a compat netif causes the ifnet * input path to be redirected to nx_netif_compat_receive(). This is specific * to the compat variant, as the native variant's RX path already goes to * the native netif. * * During channel close, we restore the original I/O callbacks, as well as the * interface's offload flags. */ #include <skywalk/os_skywalk_private.h> #include <skywalk/nexus/netif/nx_netif.h> #include <skywalk/nexus/upipe/nx_user_pipe.h> #include <skywalk/nexus/flowswitch/nx_flowswitch.h> #include <sys/kdebug.h> #include <sys/sdt.h> #include <os/refcnt.h> #include <libkern/OSDebug.h> #define NX_NETIF_MAXRINGS NX_MAX_NUM_RING_PAIR #define NX_NETIF_MINSLOTS 2 /* XXX same as above */ #define NX_NETIF_MAXSLOTS NX_MAX_NUM_SLOT_PER_RING /* max # of slots */ #define NX_NETIF_TXRINGSIZE 512 /* default TX ring size */ #define NX_NETIF_RXRINGSIZE 1024 /* default RX ring size */ #define NX_NETIF_BUFSIZE (2 * 1024) /* default buffer size */ #define NX_NETIF_MINBUFSIZE (128) /* min buffer size */ #define NX_NETIF_MAXBUFSIZE (16 * 1024) /* max buffer size */ /* * TODO: adi@apple.com -- minimum buflets for now; we will need to * have a way to adjust this based on the underlying interface's * parameters, e.g. jumbo MTU, large segment offload, etc. */ #define NX_NETIF_UMD_SIZE _USER_PACKET_SIZE(BUFLETS_MIN) #define NX_NETIF_KMD_SIZE _KERN_PACKET_SIZE(BUFLETS_MIN) /* * minimum stack space required for IOSkywalkFamily and Driver execution. */ #if XNU_TARGET_OS_OSX #define NX_NETIF_MIN_DRIVER_STACK_SIZE (kernel_stack_size >> 1) #else /* !XNU_TARGET_OS_OSX */ #define NX_NETIF_MIN_DRIVER_STACK_SIZE (kernel_stack_size >> 2) #endif /* XNU_TARGET_OS_OSX */ static void nx_netif_dom_init(struct nxdom *); static void nx_netif_dom_terminate(struct nxdom *); static void nx_netif_dom_fini(struct nxdom *); static int nx_netif_prov_params_adjust( const struct kern_nexus_domain_provider *, const struct nxprov_params *, struct nxprov_adjusted_params *); static int nx_netif_dom_bind_port(struct kern_nexus *, nexus_port_t *, struct nxbind *, void *); static int nx_netif_dom_unbind_port(struct kern_nexus *, nexus_port_t); static int nx_netif_dom_connect(struct kern_nexus_domain_provider *, struct kern_nexus *, struct kern_channel *, struct chreq *, struct kern_channel *, struct nxbind *, struct proc *); static void nx_netif_dom_disconnect(struct kern_nexus_domain_provider *, struct kern_nexus *, struct kern_channel *); static void nx_netif_dom_defunct(struct kern_nexus_domain_provider *, struct kern_nexus *, struct kern_channel *, struct proc *); static void nx_netif_dom_defunct_finalize(struct kern_nexus_domain_provider *, struct kern_nexus *, struct kern_channel *, boolean_t); static void nx_netif_doorbell(struct ifnet *); static int nx_netif_na_txsync(struct __kern_channel_ring *, struct proc *, uint32_t); static int nx_netif_na_rxsync(struct __kern_channel_ring *, struct proc *, uint32_t); static void nx_netif_na_dtor(struct nexus_adapter *na); static int nx_netif_na_notify_tx(struct __kern_channel_ring *, struct proc *, uint32_t); static int nx_netif_na_notify_rx(struct __kern_channel_ring *, struct proc *, uint32_t); static int nx_netif_na_activate(struct nexus_adapter *, na_activate_mode_t); static int nx_netif_ctl(struct kern_nexus *, nxcfg_cmd_t, void *, struct proc *); static int nx_netif_ctl_attach(struct kern_nexus *, struct nx_spec_req *, struct proc *); static int nx_netif_ctl_detach(struct kern_nexus *, struct nx_spec_req *); static int nx_netif_attach(struct kern_nexus *, struct ifnet *); static void nx_netif_flags_init(struct nx_netif *); static void nx_netif_flags_fini(struct nx_netif *); static int nx_netif_na_channel_event_notify(struct nexus_adapter *, struct __kern_packet *, struct __kern_channel_event *, uint16_t); static void nx_netif_capabilities_fini(struct nx_netif *); static errno_t nx_netif_interface_advisory_notify(void *, const struct ifnet_interface_advisory *); struct nxdom nx_netif_dom_s = { .nxdom_prov_head = STAILQ_HEAD_INITIALIZER(nx_netif_dom_s.nxdom_prov_head), .nxdom_type = NEXUS_TYPE_NET_IF, .nxdom_md_type = NEXUS_META_TYPE_PACKET, .nxdom_md_subtype = NEXUS_META_SUBTYPE_RAW, .nxdom_name = "netif", .nxdom_ports = { .nb_def = 2, .nb_min = 2, .nb_max = NX_NETIF_MAXPORTS, }, .nxdom_tx_rings = { .nb_def = 1, .nb_min = 1, .nb_max = NX_NETIF_MAXRINGS, }, .nxdom_rx_rings = { .nb_def = 1, .nb_min = 1, .nb_max = NX_NETIF_MAXRINGS, }, .nxdom_tx_slots = { .nb_def = NX_NETIF_TXRINGSIZE, .nb_min = NX_NETIF_MINSLOTS, .nb_max = NX_NETIF_MAXSLOTS, }, .nxdom_rx_slots = { .nb_def = NX_NETIF_RXRINGSIZE, .nb_min = NX_NETIF_MINSLOTS, .nb_max = NX_NETIF_MAXSLOTS, }, .nxdom_buf_size = { .nb_def = NX_NETIF_BUFSIZE, .nb_min = NX_NETIF_MINBUFSIZE, .nb_max = NX_NETIF_MAXBUFSIZE, }, .nxdom_meta_size = { .nb_def = NX_NETIF_UMD_SIZE, .nb_min = NX_NETIF_UMD_SIZE, .nb_max = NX_METADATA_USR_MAX_SZ, }, .nxdom_stats_size = { .nb_def = 0, .nb_min = 0, .nb_max = NX_STATS_MAX_SZ, }, .nxdom_pipes = { .nb_def = 0, .nb_min = 0, .nb_max = NX_UPIPE_MAXPIPES, }, .nxdom_flowadv_max = { .nb_def = 0, .nb_min = 0, .nb_max = NX_FLOWADV_MAX, }, .nxdom_nexusadv_size = { .nb_def = 0, .nb_min = 0, .nb_max = NX_NEXUSADV_MAX_SZ, }, .nxdom_capabilities = { .nb_def = NXPCAP_USER_CHANNEL, .nb_min = 0, .nb_max = NXPCAP_USER_CHANNEL, }, .nxdom_qmap = { .nb_def = NEXUS_QMAP_TYPE_DEFAULT, .nb_min = NEXUS_QMAP_TYPE_DEFAULT, .nb_max = NEXUS_QMAP_TYPE_WMM, }, .nxdom_max_frags = { .nb_def = NX_PBUF_FRAGS_DEFAULT, .nb_min = NX_PBUF_FRAGS_MIN, .nb_max = NX_PBUF_FRAGS_MAX, }, .nxdom_init = nx_netif_dom_init, .nxdom_terminate = nx_netif_dom_terminate, .nxdom_fini = nx_netif_dom_fini, .nxdom_find_port = NULL, .nxdom_port_is_reserved = NULL, .nxdom_bind_port = nx_netif_dom_bind_port, .nxdom_unbind_port = nx_netif_dom_unbind_port, .nxdom_connect = nx_netif_dom_connect, .nxdom_disconnect = nx_netif_dom_disconnect, .nxdom_defunct = nx_netif_dom_defunct, .nxdom_defunct_finalize = nx_netif_dom_defunct_finalize, }; struct kern_nexus_domain_provider nx_netif_prov_s = { .nxdom_prov_name = NEXUS_PROVIDER_NET_IF, /* * Don't install this as the default domain provider, i.e. * NXDOMPROVF_DEFAULT flag not set; we want netif_compat * provider to be the one handling userland-issued requests * coming down thru nxprov_create() instead. */ .nxdom_prov_flags = 0, .nxdom_prov_cb = { .dp_cb_init = nx_netif_prov_init, .dp_cb_fini = nx_netif_prov_fini, .dp_cb_params = nx_netif_prov_params, .dp_cb_mem_new = nx_netif_prov_mem_new, .dp_cb_config = nx_netif_prov_config, .dp_cb_nx_ctor = nx_netif_prov_nx_ctor, .dp_cb_nx_dtor = nx_netif_prov_nx_dtor, .dp_cb_nx_mem_info = nx_netif_prov_nx_mem_info, .dp_cb_nx_mib_get = nx_netif_prov_nx_mib_get, .dp_cb_nx_stop = nx_netif_prov_nx_stop, }, }; struct nexus_ifnet_ops na_netif_ops = { .ni_finalize = na_netif_finalize, .ni_reap = nx_netif_reap, .ni_dequeue = nx_netif_native_tx_dequeue, .ni_get_len = nx_netif_native_tx_get_len }; #define NX_NETIF_DOORBELL_MAX_DEQUEUE 64 uint32_t nx_netif_doorbell_max_dequeue = NX_NETIF_DOORBELL_MAX_DEQUEUE; SYSCTL_EXTENSIBLE_NODE(_kern_skywalk, OID_AUTO, netif, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "Skywalk network interface"); #if (DEVELOPMENT || DEBUG) SYSCTL_STRING(_kern_skywalk_netif, OID_AUTO, sk_ll_prefix, CTLFLAG_RW | CTLFLAG_LOCKED, sk_ll_prefix, sizeof(sk_ll_prefix), "ifname prefix for enabling low latency support"); static uint32_t nx_netif_force_ifnet_start = 0; SYSCTL_UINT(_kern_skywalk_netif, OID_AUTO, force_ifnet_start, CTLFLAG_RW | CTLFLAG_LOCKED, &nx_netif_force_ifnet_start, 0, "always use ifnet starter thread"); SYSCTL_UINT(_kern_skywalk_netif, OID_AUTO, doorbell_max_dequeue, CTLFLAG_RW | CTLFLAG_LOCKED, &nx_netif_doorbell_max_dequeue, NX_NETIF_DOORBELL_MAX_DEQUEUE, "max packets to dequeue in doorbell context"); #endif /* !DEVELOPMENT && !DEBUG */ static ZONE_DEFINE(na_netif_zone, SKMEM_ZONE_PREFIX ".na.netif", sizeof(struct nexus_netif_adapter), ZC_ZFREE_CLEARMEM); static ZONE_DEFINE(nx_netif_zone, SKMEM_ZONE_PREFIX ".nx.netif", sizeof(struct nx_netif), ZC_ZFREE_CLEARMEM); #define SKMEM_TAG_NETIF_MIT "com.apple.skywalk.netif.mit" static kern_allocation_name_t skmem_tag_netif_mit; #define SKMEM_TAG_NETIF_FILTER "com.apple.skywalk.netif.filter" kern_allocation_name_t skmem_tag_netif_filter; #define SKMEM_TAG_NETIF_FLOW "com.apple.skywalk.netif.flow" kern_allocation_name_t skmem_tag_netif_flow; #define SKMEM_TAG_NETIF_AGENT_FLOW "com.apple.skywalk.netif.agent_flow" kern_allocation_name_t skmem_tag_netif_agent_flow; #define SKMEM_TAG_NETIF_LLINK "com.apple.skywalk.netif.llink" kern_allocation_name_t skmem_tag_netif_llink; #define SKMEM_TAG_NETIF_QSET "com.apple.skywalk.netif.qset" kern_allocation_name_t skmem_tag_netif_qset; #define SKMEM_TAG_NETIF_LLINK_INFO "com.apple.skywalk.netif.llink_info" kern_allocation_name_t skmem_tag_netif_llink_info; static void nx_netif_dom_init(struct nxdom *nxdom) { SK_LOCK_ASSERT_HELD(); ASSERT(!(nxdom->nxdom_flags & NEXUSDOMF_INITIALIZED)); _CASSERT(NEXUS_PORT_NET_IF_DEV == 0); _CASSERT(NEXUS_PORT_NET_IF_HOST == 1); _CASSERT(NEXUS_PORT_NET_IF_CLIENT == 2); _CASSERT(SK_NETIF_MIT_FORCE_OFF < SK_NETIF_MIT_FORCE_SIMPLE); _CASSERT(SK_NETIF_MIT_FORCE_SIMPLE < SK_NETIF_MIT_FORCE_ADVANCED); _CASSERT(SK_NETIF_MIT_FORCE_ADVANCED < SK_NETIF_MIT_AUTO); _CASSERT(SK_NETIF_MIT_AUTO == SK_NETIF_MIT_MAX); (void) nxdom_prov_add(nxdom, &nx_netif_prov_s); ASSERT(skmem_tag_netif_mit == NULL); skmem_tag_netif_mit = kern_allocation_name_allocate(SKMEM_TAG_NETIF_MIT, 0); ASSERT(skmem_tag_netif_mit != NULL); ASSERT(skmem_tag_netif_filter == NULL); skmem_tag_netif_filter = kern_allocation_name_allocate(SKMEM_TAG_NETIF_FILTER, 0); ASSERT(skmem_tag_netif_filter != NULL); ASSERT(skmem_tag_netif_flow == NULL); skmem_tag_netif_flow = kern_allocation_name_allocate(SKMEM_TAG_NETIF_FLOW, 0); ASSERT(skmem_tag_netif_flow != NULL); ASSERT(skmem_tag_netif_agent_flow == NULL); skmem_tag_netif_agent_flow = kern_allocation_name_allocate(SKMEM_TAG_NETIF_AGENT_FLOW, 0); ASSERT(skmem_tag_netif_agent_flow != NULL); ASSERT(skmem_tag_netif_llink == NULL); skmem_tag_netif_llink = kern_allocation_name_allocate(SKMEM_TAG_NETIF_LLINK, 0); ASSERT(skmem_tag_netif_llink != NULL); ASSERT(skmem_tag_netif_qset == NULL); skmem_tag_netif_qset = kern_allocation_name_allocate(SKMEM_TAG_NETIF_QSET, 0); ASSERT(skmem_tag_netif_qset != NULL); ASSERT(skmem_tag_netif_llink_info == NULL); skmem_tag_netif_llink_info = kern_allocation_name_allocate(SKMEM_TAG_NETIF_LLINK_INFO, 0); ASSERT(skmem_tag_netif_llink_info != NULL); nx_netif_compat_init(nxdom); ASSERT(nxdom_prov_default[nxdom->nxdom_type] != NULL && strcmp(nxdom_prov_default[nxdom->nxdom_type]->nxdom_prov_name, NEXUS_PROVIDER_NET_IF_COMPAT) == 0); netif_gso_init(); nx_netif_llink_module_init(); } static void nx_netif_dom_terminate(struct nxdom *nxdom) { struct kern_nexus_domain_provider *nxdom_prov, *tnxdp; SK_LOCK_ASSERT_HELD(); nx_netif_llink_module_fini(); netif_gso_fini(); nx_netif_compat_fini(); if (skmem_tag_netif_llink_info != NULL) { kern_allocation_name_release(skmem_tag_netif_llink_info); skmem_tag_netif_llink_info = NULL; } if (skmem_tag_netif_qset != NULL) { kern_allocation_name_release(skmem_tag_netif_qset); skmem_tag_netif_qset = NULL; } if (skmem_tag_netif_llink != NULL) { kern_allocation_name_release(skmem_tag_netif_llink); skmem_tag_netif_llink = NULL; } if (skmem_tag_netif_agent_flow != NULL) { kern_allocation_name_release(skmem_tag_netif_agent_flow); skmem_tag_netif_agent_flow = NULL; } if (skmem_tag_netif_flow != NULL) { kern_allocation_name_release(skmem_tag_netif_flow); skmem_tag_netif_flow = NULL; } if (skmem_tag_netif_filter != NULL) { kern_allocation_name_release(skmem_tag_netif_filter); skmem_tag_netif_filter = NULL; } if (skmem_tag_netif_mit != NULL) { kern_allocation_name_release(skmem_tag_netif_mit); skmem_tag_netif_mit = NULL; } STAILQ_FOREACH_SAFE(nxdom_prov, &nxdom->nxdom_prov_head, nxdom_prov_link, tnxdp) { (void) nxdom_prov_del(nxdom_prov); } } static void nx_netif_dom_fini(struct nxdom *nxdom) { #pragma unused(nxdom) } int nx_netif_prov_init(struct kern_nexus_domain_provider *nxdom_prov) { #pragma unused(nxdom_prov) SK_D("initializing %s", nxdom_prov->nxdom_prov_name); return 0; } static int nx_netif_na_notify_drop(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags) { #pragma unused(kring, p, flags) return ENXIO; } int nx_netif_prov_nx_stop(struct kern_nexus *nx) { uint32_t r; struct nexus_adapter *na = nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV); struct nexus_netif_adapter *nifna = (struct nexus_netif_adapter *)na; SK_LOCK_ASSERT_HELD(); ASSERT(nx != NULL); /* place all rings in drop mode */ na_kr_drop(na, TRUE); /* ensure global visibility */ membar_sync(); /* reset all TX notify callbacks */ for (r = 0; r < na_get_nrings(na, NR_TX); r++) { while (!atomic_test_set_ptr(&na->na_tx_rings[r].ckr_na_notify, ptrauth_nop_cast(void *, na->na_tx_rings[r].ckr_na_notify), ptrauth_nop_cast(void *, &nx_netif_na_notify_drop))) { ; } membar_sync(); if (nifna->nifna_tx_mit != NULL) { nx_netif_mit_cleanup(&nifna->nifna_tx_mit[r]); } } if (nifna->nifna_tx_mit != NULL) { skn_free_type_array(tx, struct nx_netif_mit, na_get_nrings(na, NR_TX), nifna->nifna_tx_mit); nifna->nifna_tx_mit = NULL; } /* reset all RX notify callbacks */ for (r = 0; r < na_get_nrings(na, NR_RX); r++) { while (!atomic_test_set_ptr(&na->na_rx_rings[r].ckr_na_notify, ptrauth_nop_cast(void *, na->na_rx_rings[r].ckr_na_notify), ptrauth_nop_cast(void *, &nx_netif_na_notify_drop))) { ; } membar_sync(); if (nifna->nifna_rx_mit != NULL) { nx_netif_mit_cleanup(&nifna->nifna_rx_mit[r]); } } if (nifna->nifna_rx_mit != NULL) { skn_free_type_array(rx, struct nx_netif_mit, na_get_nrings(na, NR_RX), nifna->nifna_rx_mit); nifna->nifna_rx_mit = NULL; } return 0; } static inline void nx_netif_compat_adjust_ring_size(struct nxprov_adjusted_params *adj, ifnet_t ifp) { if (IFNET_IS_CELLULAR(ifp) && (ifp->if_unit != 0)) { *(adj->adj_rx_slots) = sk_netif_compat_aux_cell_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_aux_cell_tx_ring_sz; } else if (IFNET_IS_WIFI(ifp)) { if (ifp->if_name[0] == 'a' && ifp->if_name[1] == 'p' && ifp->if_name[2] == '\0') { /* Wi-Fi Access Point */ *(adj->adj_rx_slots) = sk_netif_compat_wap_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_wap_tx_ring_sz; } else if (ifp->if_eflags & IFEF_AWDL) { /* AWDL */ *(adj->adj_rx_slots) = sk_netif_compat_awdl_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_awdl_tx_ring_sz; } else { /* Wi-Fi infrastructure */ *(adj->adj_rx_slots) = sk_netif_compat_wif_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_wif_tx_ring_sz; } } else if (IFNET_IS_ETHERNET(ifp)) { #if !XNU_TARGET_OS_OSX /* * On non-macOS platforms, treat all compat Ethernet * interfaces as USB Ethernet with reduced ring sizes. */ *(adj->adj_rx_slots) = sk_netif_compat_usb_eth_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_usb_eth_tx_ring_sz; #else /* XNU_TARGET_OS_OSX */ if (ifp->if_subfamily == IFNET_SUBFAMILY_USB) { *(adj->adj_rx_slots) = sk_netif_compat_usb_eth_rx_ring_sz; *(adj->adj_tx_slots) = sk_netif_compat_usb_eth_tx_ring_sz; } #endif /* XNU_TARGET_OS_OSX */ } } static int nx_netif_prov_params_adjust(const struct kern_nexus_domain_provider *nxdom_prov, const struct nxprov_params *nxp, struct nxprov_adjusted_params *adj) { /* * for netif compat adjust the following parameters for memory * optimization: * - change the size of buffer object to 128 bytes. * - don't allocate rx ring for host port and tx ring for dev port. * - for cellular interfaces other than pdp_ip0 reduce the ring size. * Assumption here is that pdp_ip0 is always used as the data * interface. * - reduce the ring size for AWDL interface. * - reduce the ring size for USB ethernet interface. */ if (strcmp(nxdom_prov->nxdom_prov_name, NEXUS_PROVIDER_NET_IF_COMPAT) == 0) { /* * Leave the parameters default if userspace access may be * needed. We can't use skywalk_direct_allowed() here because * the drivers have not attached yet. */ if (skywalk_netif_direct_enabled()) { goto done; } *(adj->adj_buf_size) = NETIF_COMPAT_BUF_SIZE; *(adj->adj_tx_rings) = 1; if (IF_INDEX_IN_RANGE(nxp->nxp_ifindex)) { ifnet_t ifp; ifnet_head_lock_shared(); ifp = ifindex2ifnet[nxp->nxp_ifindex]; ifnet_head_done(); VERIFY(ifp != NULL); nx_netif_compat_adjust_ring_size(adj, ifp); } if (adj->adj_buf_srp->srp_r_seg_size == 0) { adj->adj_buf_srp->srp_r_seg_size = skmem_usr_buf_seg_size; } } else { /* netif native */ if (nxp->nxp_flags & NXPF_NETIF_LLINK) { *(adj->adj_tx_slots) = NX_NETIF_MINSLOTS; *(adj->adj_rx_slots) = NX_NETIF_MINSLOTS; } /* * Add another extra ring for host port. Note that if the * nexus isn't configured to use the same pbufpool for all of * its ports, we'd end up allocating extra here. * Not a big deal since that case isn't the default. */ *(adj->adj_tx_rings) += 1; *(adj->adj_rx_rings) += 1; if ((*(adj->adj_buf_size) < PKT_MAX_PROTO_HEADER_SIZE)) { SK_ERR("buf size too small, min (%d)", PKT_MAX_PROTO_HEADER_SIZE); return EINVAL; } _CASSERT(sizeof(struct __kern_netif_intf_advisory) == NX_INTF_ADV_SIZE); *(adj->adj_nexusadv_size) = sizeof(struct netif_nexus_advisory); } done: /* enable magazines layer for metadata */ *(adj->adj_md_magazines) = TRUE; return 0; } int nx_netif_prov_params(struct kern_nexus_domain_provider *nxdom_prov, const uint32_t req, const struct nxprov_params *nxp0, struct nxprov_params *nxp, struct skmem_region_params srp[SKMEM_REGIONS]) { struct nxdom *nxdom = nxdom_prov->nxdom_prov_dom; return nxprov_params_adjust(nxdom_prov, req, nxp0, nxp, srp, nxdom, nxdom, nxdom, nx_netif_prov_params_adjust); } int nx_netif_prov_mem_new(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct nexus_adapter *na) { #pragma unused(nxdom_prov) int err = 0; boolean_t pp_truncated_buf = FALSE; boolean_t allow_direct; boolean_t kernel_only; SK_DF(SK_VERB_NETIF, "nx 0x%llx (\"%s\":\"%s\") na \"%s\" (0x%llx)", SK_KVA(nx), NX_DOM(nx)->nxdom_name, nxdom_prov->nxdom_prov_name, na->na_name, SK_KVA(na)); ASSERT(na->na_arena == NULL); if ((na->na_type == NA_NETIF_COMPAT_DEV) || (na->na_type == NA_NETIF_COMPAT_HOST)) { pp_truncated_buf = TRUE; } /* * We do this check to determine whether to create the extra * regions needed for userspace access. This is per interface. * NX_USER_CHANNEL_PROV() is systemwide so it can't be used. */ allow_direct = skywalk_netif_direct_allowed(na->na_name); /* * Both ports (host and dev) share the same packet buffer pool; * the first time a port gets opened will allocate the pp that * gets stored in the nexus, which will then be used by any * subsequent opens. */ kernel_only = !allow_direct || !NX_USER_CHANNEL_PROV(nx); na->na_arena = skmem_arena_create_for_nexus(na, NX_PROV(nx)->nxprov_region_params, &nx->nx_tx_pp, &nx->nx_rx_pp, pp_truncated_buf, kernel_only, &nx->nx_adv, &err); ASSERT(na->na_arena != NULL || err != 0); ASSERT(nx->nx_tx_pp == NULL || (nx->nx_tx_pp->pp_md_type == NX_DOM(nx)->nxdom_md_type && nx->nx_tx_pp->pp_md_subtype == NX_DOM(nx)->nxdom_md_subtype)); return err; } SK_NO_INLINE_ATTRIBUTE static int nx_netif_get_llink_info(struct sockopt *sopt, struct kern_nexus *nx) { struct nx_llink_info_req *nlir = NULL; struct nx_netif *nif; struct netif_llink *llink; uint16_t llink_cnt; size_t len, user_len; int err, i; nif = NX_NETIF_PRIVATE(nx); if (!NETIF_LLINK_ENABLED(nif)) { SK_ERR("llink mode not enabled"); return ENOTSUP; } lck_rw_lock_shared(&nif->nif_llink_lock); llink_cnt = nif->nif_llink_cnt; if (llink_cnt == 0) { SK_ERR("zero llink cnt"); err = ENXIO; goto done; } len = sizeof(*nlir) + (sizeof(struct nx_llink_info) * llink_cnt); /* preserve sopt_valsize because it gets overwritten by copyin */ user_len = sopt->sopt_valsize; if (user_len < len) { SK_ERR("buffer too small"); err = ENOBUFS; goto done; } nlir = sk_alloc_data(len, Z_WAITOK, skmem_tag_netif_llink_info); if (nlir == NULL) { SK_ERR("failed to allocate nlir"); err = ENOMEM; goto done; } err = sooptcopyin(sopt, nlir, sizeof(*nlir), sizeof(*nlir)); if (err != 0) { SK_ERR("copyin failed: %d", err); goto done; } if (nlir->nlir_version != NETIF_LLINK_INFO_VERSION) { SK_ERR("nlir version mismatch: %d != %d", nlir->nlir_version, NETIF_LLINK_INFO_VERSION); err = ENOTSUP; goto done; } nlir->nlir_llink_cnt = llink_cnt; i = 0; STAILQ_FOREACH(llink, &nif->nif_llink_list, nll_link) { struct nx_llink_info *nli; struct netif_qset *qset; uint16_t qset_cnt; int j; nli = &nlir->nlir_llink[i]; nli->nli_link_id = llink->nll_link_id; nli->nli_link_id_internal = llink->nll_link_id_internal; nli->nli_state = llink->nll_state; nli->nli_flags = llink->nll_flags; qset_cnt = llink->nll_qset_cnt; ASSERT(qset_cnt <= NETIF_LLINK_MAX_QSETS); nli->nli_qset_cnt = qset_cnt; j = 0; SLIST_FOREACH(qset, &llink->nll_qset_list, nqs_list) { struct nx_qset_info *nqi; nqi = &nli->nli_qset[j]; nqi->nqi_id = qset->nqs_id; nqi->nqi_flags = qset->nqs_flags; nqi->nqi_num_rx_queues = qset->nqs_num_rx_queues; nqi->nqi_num_tx_queues = qset->nqs_num_tx_queues; j++; } ASSERT(j == qset_cnt); i++; } ASSERT(i == llink_cnt); sopt->sopt_valsize = user_len; err = sooptcopyout(sopt, nlir, len); if (err != 0) { SK_ERR("sooptcopyout failed: %d", err); } done: lck_rw_unlock_shared(&nif->nif_llink_lock); if (nlir != NULL) { sk_free_data(nlir, len); } return err; } int nx_netif_prov_config(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct nx_cfg_req *ncr, int sopt_dir, struct proc *p, kauth_cred_t cred) { #pragma unused(nxdom_prov) struct sockopt sopt; int err = 0; SK_LOCK_ASSERT_HELD(); /* proceed only if the client possesses netif entitlement */ if ((err = skywalk_priv_check_cred(p, cred, PRIV_SKYWALK_REGISTER_NET_IF)) != 0) { goto done; } if (ncr->nc_req == USER_ADDR_NULL) { err = EINVAL; goto done; } /* to make life easier for handling copies */ bzero(&sopt, sizeof(sopt)); sopt.sopt_dir = sopt_dir; sopt.sopt_val = ncr->nc_req; sopt.sopt_valsize = ncr->nc_req_len; sopt.sopt_p = p; switch (ncr->nc_cmd) { case NXCFG_CMD_ATTACH: case NXCFG_CMD_DETACH: { struct nx_spec_req nsr; bzero(&nsr, sizeof(nsr)); err = sooptcopyin(&sopt, &nsr, sizeof(nsr), sizeof(nsr)); if (err != 0) { goto done; } /* * Null-terminate in case this has an interface name; * the union is already large enough for uuid_t. */ nsr.nsr_name[sizeof(nsr.nsr_name) - 1] = '\0'; if (p != kernproc) { nsr.nsr_flags &= NXSPECREQ_MASK; } err = nx_netif_ctl(nx, ncr->nc_cmd, &nsr, p); if (err != 0) { goto done; } /* XXX: adi@apple.com -- can this copyout fail? */ (void) sooptcopyout(&sopt, &nsr, sizeof(nsr)); break; } case NXCFG_CMD_FLOW_ADD: case NXCFG_CMD_FLOW_DEL: { _CASSERT(offsetof(struct nx_flow_req, _nfr_kernel_field_end) == offsetof(struct nx_flow_req, _nfr_common_field_end)); struct nx_flow_req nfr; bzero(&nfr, sizeof(nfr)); err = sooptcopyin(&sopt, &nfr, sizeof(nfr), sizeof(nfr)); if (err != 0) { goto done; } err = nx_netif_ctl(nx, ncr->nc_cmd, &nfr, p); if (err != 0) { goto done; } /* XXX: adi@apple.com -- can this copyout fail? */ (void) sooptcopyout(&sopt, &nfr, sizeof(nfr)); break; } case NXCFG_CMD_GET_LLINK_INFO: { err = nx_netif_get_llink_info(&sopt, nx); break; } default: err = EINVAL; goto done; } done: SK_DF(err ? SK_VERB_ERROR : SK_VERB_NETIF, "nexus 0x%llx (%s) cmd %d err %d", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name, ncr->nc_cmd, err); return err; } void nx_netif_prov_fini(struct kern_nexus_domain_provider *nxdom_prov) { #pragma unused(nxdom_prov) SK_D("destroying %s", nxdom_prov->nxdom_prov_name); } int nx_netif_prov_nx_ctor(struct kern_nexus *nx) { struct nx_netif *n; char name[64]; int error; SK_LOCK_ASSERT_HELD(); ASSERT(nx->nx_arg == NULL); SK_D("nexus 0x%llx (%s)", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name); nx->nx_arg = nx_netif_alloc(Z_WAITOK); n = NX_NETIF_PRIVATE(nx); if (NX_USER_CHANNEL_PROV(nx) && NX_PROV(nx)->nxprov_params->nxp_nexusadv_size != 0) { (void) snprintf(name, sizeof(name), "netif_%llu", nx->nx_id); error = nx_advisory_alloc(nx, name, &NX_PROV(nx)->nxprov_region_params[SKMEM_REGION_NEXUSADV], NEXUS_ADVISORY_TYPE_NETIF); if (error != 0) { nx_netif_free(n); return error; } } n->nif_nx = nx; SK_D("create new netif 0x%llx for nexus 0x%llx", SK_KVA(NX_NETIF_PRIVATE(nx)), SK_KVA(nx)); return 0; } void nx_netif_prov_nx_dtor(struct kern_nexus *nx) { struct nx_netif *n = NX_NETIF_PRIVATE(nx); SK_LOCK_ASSERT_HELD(); SK_D("nexus 0x%llx (%s) netif 0x%llx", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name, SK_KVA(n)); /* * XXX * detach should be done separately to be symmetrical with attach. */ nx_advisory_free(nx); if (nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV) != NULL) { /* we're called by nx_detach(), so this cannot fail */ int err = nx_netif_ctl_detach(nx, NULL); VERIFY(err == 0); } if (n->nif_dev_nxb != NULL) { nxb_free(n->nif_dev_nxb); n->nif_dev_nxb = NULL; } if (n->nif_host_nxb != NULL) { nxb_free(n->nif_host_nxb); n->nif_host_nxb = NULL; } SK_DF(SK_VERB_NETIF, "marking netif 0x%llx as free", SK_KVA(n)); nx_netif_free(n); nx->nx_arg = NULL; } int nx_netif_prov_nx_mem_info(struct kern_nexus *nx, struct kern_pbufpool **tpp, struct kern_pbufpool **rpp) { ASSERT(nx->nx_tx_pp != NULL); ASSERT(nx->nx_rx_pp != NULL); if (tpp != NULL) { *tpp = nx->nx_tx_pp; } if (rpp != NULL) { *rpp = nx->nx_rx_pp; } return 0; } static size_t __netif_mib_get_stats(struct kern_nexus *nx, void *out, size_t len) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); struct ifnet *ifp = nif->nif_ifp; struct sk_stats_net_if *sns = out; size_t actual_space = sizeof(struct sk_stats_net_if); if (out != NULL && actual_space <= len) { uuid_copy(sns->sns_nx_uuid, nx->nx_uuid); if (ifp != NULL) { (void) strlcpy(sns->sns_if_name, if_name(ifp), IFNAMSIZ); } sns->sns_nifs = nif->nif_stats; } return actual_space; } static size_t __netif_mib_get_llinks(struct kern_nexus *nx, void *out, size_t len) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); struct nx_llink_info *nli_list = out; size_t actual_space = 0; if (NETIF_LLINK_ENABLED(nif)) { lck_rw_lock_shared(&nif->nif_llink_lock); actual_space += nif->nif_llink_cnt * sizeof(struct nx_llink_info); if (out != NULL && actual_space <= len) { struct netif_llink *llink; int i = 0; STAILQ_FOREACH(llink, &nif->nif_llink_list, nll_link) { struct nx_llink_info *nli; struct netif_qset *qset; uint16_t qset_cnt; int j; nli = &nli_list[i]; uuid_copy(nli->nli_netif_uuid, nx->nx_uuid); nli->nli_link_id = llink->nll_link_id; nli->nli_link_id_internal = llink->nll_link_id_internal; nli->nli_state = llink->nll_state; nli->nli_flags = llink->nll_flags; qset_cnt = llink->nll_qset_cnt; ASSERT(qset_cnt <= NETIF_LLINK_MAX_QSETS); nli->nli_qset_cnt = qset_cnt; j = 0; SLIST_FOREACH(qset, &llink->nll_qset_list, nqs_list) { struct nx_qset_info *nqi; nqi = &nli->nli_qset[j]; nqi->nqi_id = qset->nqs_id; nqi->nqi_flags = qset->nqs_flags; nqi->nqi_num_rx_queues = qset->nqs_num_rx_queues; nqi->nqi_num_tx_queues = qset->nqs_num_tx_queues; j++; } ASSERT(j == qset_cnt); i++; } ASSERT(i == nif->nif_llink_cnt); } lck_rw_unlock_shared(&nif->nif_llink_lock); } return actual_space; } size_t nx_netif_prov_nx_mib_get(struct kern_nexus *nx, struct nexus_mib_filter *filter, void *out, size_t len, struct proc *p) { #pragma unused(p) size_t ret; if ((filter->nmf_bitmap & NXMIB_FILTER_NX_UUID) && (uuid_compare(filter->nmf_nx_uuid, nx->nx_uuid)) != 0) { return 0; } switch (filter->nmf_type) { case NXMIB_NETIF_STATS: ret = __netif_mib_get_stats(nx, out, len); break; case NXMIB_LLINK_LIST: ret = __netif_mib_get_llinks(nx, out, len); break; default: ret = 0; break; } return ret; } static int nx_netif_dom_bind_port(struct kern_nexus *nx, nexus_port_t *nx_port, struct nxbind *nxb, void *info) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); nexus_port_t first, last, port; int error; ASSERT(nx_port != NULL); ASSERT(nxb != NULL); port = *nx_port; /* * If port is: * != NEXUS_PORT_ANY: attempt to bind to the specified port * == NEXUS_PORT_ANY: find an available port, bind to it, and * return back the assigned port. */ first = NEXUS_PORT_NET_IF_CLIENT; last = NXDOM_MAX(NX_DOM(nx), ports); ASSERT(first <= last); NETIF_WLOCK(nif); if (__improbable(first == last)) { error = ENOMEM; } else if (port != NEXUS_PORT_ANY) { error = nx_port_bind_info(nx, port, nxb, info); SK_DF(SK_VERB_NETIF, "port %d, bind err %d", port, error); } else { error = nx_port_find(nx, first, last - 1, &port); ASSERT(error != 0 || (port >= first && port < last)); if (error == 0) { error = nx_port_bind_info(nx, port, nxb, info); SK_DF(SK_VERB_NETIF, "found port %d, bind err %d", port, error); } } NETIF_WUNLOCK(nif); ASSERT(*nx_port == NEXUS_PORT_ANY || *nx_port == port); if (error == 0) { *nx_port = port; } SK_DF(error ? SK_VERB_ERROR : SK_VERB_NETIF, "+++ netif 0x%llx nx_port %d, total %u active %u (err %d)", SK_KVA(nif), (int)*nx_port, NX_NETIF_MAXPORTS, nx->nx_active_ports, error); return error; } static int nx_netif_dom_unbind_port(struct kern_nexus *nx, nexus_port_t nx_port) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); int error = 0; ASSERT(nx_port != NEXUS_PORT_ANY); NETIF_WLOCK(nif); error = nx_port_unbind(nx, nx_port); NETIF_WUNLOCK(nif); return error; } static int nx_netif_dom_connect(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct kern_channel *ch, struct chreq *chr, struct kern_channel *ch0, struct nxbind *nxb, struct proc *p) { #pragma unused(nxdom_prov) int err = 0; SK_LOCK_ASSERT_HELD(); ASSERT(NX_DOM_PROV(nx) == nxdom_prov); ASSERT(nx->nx_prov->nxprov_params->nxp_type == nxdom_prov->nxdom_prov_dom->nxdom_type && nx->nx_prov->nxprov_params->nxp_type == NEXUS_TYPE_NET_IF); ASSERT(!(ch->ch_flags & CHANF_HOST)); switch (chr->cr_port) { case NEXUS_PORT_NET_IF_DEV: if (chr->cr_mode & CHMODE_HOST) { err = EINVAL; goto done; } break; case NEXUS_PORT_NET_IF_HOST: if (!(chr->cr_mode & CHMODE_HOST)) { if (ch->ch_flags & CHANF_KERNEL) { err = EINVAL; goto done; } chr->cr_mode |= CHMODE_HOST; } /* * This channel is exclusively opened to the host * rings; don't notify the external provider. */ atomic_bitset_32(&ch->ch_flags, CHANF_HOST | CHANF_EXT_SKIP); break; default: /* * This channel is shared between netif and user process; * don't notify the external provider. */ atomic_bitset_32(&ch->ch_flags, CHANF_EXT_SKIP); break; } chr->cr_ring_set = RING_SET_DEFAULT; chr->cr_real_endpoint = chr->cr_endpoint = CH_ENDPOINT_NET_IF; (void) snprintf(chr->cr_name, sizeof(chr->cr_name), "netif:%llu:%.*s", nx->nx_id, (int)nx->nx_prov->nxprov_params->nxp_namelen, nx->nx_prov->nxprov_params->nxp_name); if (ch->ch_flags & CHANF_KERNEL) { err = na_connect_spec(nx, ch, chr, p); } else { err = na_connect(nx, ch, chr, ch0, nxb, p); } if (err == 0) { /* * Mark the kernel slot descriptor region as busy; this * prevents it from being torn-down at channel defunct * time, as the (external) nexus owner may be calling * KPIs that require accessing the slots. */ skmem_arena_nexus_sd_set_noidle( skmem_arena_nexus(ch->ch_na->na_arena), 1); } done: return err; } static void nx_netif_dom_disconnect(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct kern_channel *ch) { #pragma unused(nxdom_prov) SK_LOCK_ASSERT_HELD(); SK_D("channel 0x%llx -!- nexus 0x%llx (%s:\"%s\":%u:%d)", SK_KVA(ch), SK_KVA(nx), nxdom_prov->nxdom_prov_name, ch->ch_na->na_name, ch->ch_info->cinfo_nx_port, (int)ch->ch_info->cinfo_ch_ring_id); /* * Release busy assertion held earlier in nx_netif_dom_connect(); * this allows for the final arena teardown to succeed. */ skmem_arena_nexus_sd_set_noidle( skmem_arena_nexus(ch->ch_na->na_arena), -1); if (ch->ch_flags & CHANF_KERNEL) { na_disconnect_spec(nx, ch); } else { na_disconnect(nx, ch); } } static void nx_netif_dom_defunct(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct kern_channel *ch, struct proc *p) { #pragma unused(nxdom_prov, nx) LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_OWNED); ASSERT(!(ch->ch_flags & CHANF_KERNEL)); ASSERT(ch->ch_na->na_type == NA_NETIF_DEV || ch->ch_na->na_type == NA_NETIF_HOST || ch->ch_na->na_type == NA_NETIF_COMPAT_DEV || ch->ch_na->na_type == NA_NETIF_COMPAT_HOST); na_ch_rings_defunct(ch, p); } static void nx_netif_dom_defunct_finalize(struct kern_nexus_domain_provider *nxdom_prov, struct kern_nexus *nx, struct kern_channel *ch, boolean_t locked) { #pragma unused(nxdom_prov) if (!locked) { SK_LOCK_ASSERT_NOTHELD(); SK_LOCK(); LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_NOTOWNED); } else { SK_LOCK_ASSERT_HELD(); LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_OWNED); } ASSERT(ch->ch_na->na_type == NA_NETIF_DEV || ch->ch_na->na_type == NA_NETIF_HOST || ch->ch_na->na_type == NA_NETIF_COMPAT_DEV || ch->ch_na->na_type == NA_NETIF_COMPAT_HOST); na_defunct(nx, ch, ch->ch_na, locked); SK_D("%s(%d): ch 0x%llx -/- nx 0x%llx (%s:\"%s\":%u:%d)", ch->ch_name, ch->ch_pid, SK_KVA(ch), SK_KVA(nx), nxdom_prov->nxdom_prov_name, ch->ch_na->na_name, ch->ch_info->cinfo_nx_port, (int)ch->ch_info->cinfo_ch_ring_id); if (!locked) { LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_NOTOWNED); SK_UNLOCK(); } else { LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_OWNED); SK_LOCK_ASSERT_HELD(); } } struct nexus_netif_adapter * na_netif_alloc(zalloc_flags_t how) { _CASSERT(offsetof(struct nexus_netif_adapter, nifna_up) == 0); return zalloc_flags(na_netif_zone, how | Z_ZERO); } void na_netif_free(struct nexus_adapter *na) { struct nexus_netif_adapter *nifna = (struct nexus_netif_adapter *)na; SK_LOCK_ASSERT_HELD(); SK_DF(SK_VERB_MEM, "nifna 0x%llx FREE", SK_KVA(nifna)); ASSERT(na->na_refcount == 0); ASSERT(nifna->nifna_tx_mit == NULL); ASSERT(nifna->nifna_rx_mit == NULL); bzero(nifna, sizeof(*nifna)); zfree(na_netif_zone, nifna); } /* Process NXCFG_CMD_ATTACH */ SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl_attach(struct kern_nexus *nx, struct nx_spec_req *nsr, struct proc *p) { struct nx_netif *n = NX_NETIF_PRIVATE(nx); struct ifnet *ifp = NULL; boolean_t compat; int err = 0; SK_LOCK_ASSERT_HELD(); ASSERT(NX_DOM(nx)->nxdom_type == NEXUS_TYPE_NET_IF); compat = (strcmp(NX_DOM_PROV(nx)->nxdom_prov_name, NEXUS_PROVIDER_NET_IF_COMPAT) == 0); uuid_clear(nsr->nsr_if_uuid); /* * The netif accepts either an interface name or a pointer to * an ifnet, but never a UUID. */ if (nsr->nsr_flags & NXSPECREQ_UUID) { err = EINVAL; goto done; } if (nsr->nsr_flags & NXSPECREQ_IFP) { if (p != kernproc || (ifp = nsr->nsr_ifp) == NULL) { err = EINVAL; goto done; } } else if ((ifp = ifunit_ref(nsr->nsr_name)) == NULL) { err = ENXIO; goto done; } if ((compat && SKYWALK_NATIVE(ifp)) || (!compat && !SKYWALK_NATIVE(ifp))) { /* native driver for netif; non-native for netif_compat */ err = ENODEV; } else if (ifp->if_na != NULL || !uuid_is_null(n->nif_uuid)) { err = EBUSY; } else { ASSERT(uuid_is_null(n->nif_uuid)); /* * Upon success, callee will hold its own ifnet iorefcnt * as well as a retain count on the nexus adapter. */ if (compat) { err = nx_netif_compat_attach(nx, ifp); } else { err = nx_netif_attach(nx, ifp); } if (err == 0) { /* return the adapter UUID */ uuid_generate_random(n->nif_uuid); uuid_copy(nsr->nsr_if_uuid, n->nif_uuid); #if (DEVELOPMENT || DEBUG) skoid_create(&n->nif_skoid, SKOID_SNODE(_kern_skywalk_netif), if_name(ifp), CTLFLAG_RW); #endif /* !DEVELOPMENT && !DEBUG */ } } done: /* drop I/O refcnt from ifunit_ref() */ if (ifp != NULL && !(nsr->nsr_flags & NXSPECREQ_IFP)) { ifnet_decr_iorefcnt(ifp); } #if SK_LOG uuid_string_t uuidstr, ifuuidstr; const char *nustr; if (nsr->nsr_flags & NXSPECREQ_UUID) { nustr = sk_uuid_unparse(nsr->nsr_uuid, uuidstr); } else if (nsr->nsr_flags & NXSPECREQ_IFP) { (void) snprintf((char *)uuidstr, sizeof(uuidstr), "0x%llx", SK_KVA(nsr->nsr_ifp)); nustr = uuidstr; } else { nustr = nsr->nsr_name; } SK_DF(err ? SK_VERB_ERROR : SK_VERB_NETIF, "nexus 0x%llx (%s) name/uuid \"%s\" if_uuid %s flags 0x%x err %d", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name, nustr, sk_uuid_unparse(nsr->nsr_if_uuid, ifuuidstr), nsr->nsr_flags, err); #endif /* SK_LOG */ return err; } /* process NXCFG_CMD_DETACH */ SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl_detach(struct kern_nexus *nx, struct nx_spec_req *nsr) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); int err = 0; SK_LOCK_ASSERT_HELD(); /* * nsr is NULL when we're called from the destructor, and it * implies that we'll detach whatever that is attached. */ if (nsr != NULL && uuid_is_null(nsr->nsr_if_uuid)) { err = EINVAL; } else if (nsr != NULL && uuid_compare(nsr->nsr_if_uuid, nif->nif_uuid) != 0) { err = ESRCH; } else if (nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV) == NULL) { /* nx_netif_ctl_attach() not yet done or already detached */ err = ENXIO; } else if (nx->nx_ch_count != 0) { /* * There's at least a channel opened; we can't * yank the interface from underneath the nexus * since our dlil input/output handler may be * running now. Bail out and come back here * again when the nexus detaches. */ err = EBUSY; } else { struct ifnet *ifp; boolean_t suspended = FALSE; ifp = nif->nif_ifp; if (ifp == NULL) { err = EALREADY; goto done; } /* * For regular kernel-attached interfaces, quiescing is handled by * the ifnet detach thread, which calls dlil_quiesce_and_detach_nexuses(). * For interfaces created by skywalk test cases, flowswitch/netif nexuses * are constructed on the fly and can also be torn down on the fly. * dlil_quiesce_and_detach_nexuses() won't help here because any nexus * can be detached while the interface is still attached. */ if (ifnet_datamov_suspend_if_needed(ifp)) { SK_UNLOCK(); suspended = TRUE; ifnet_datamov_drain(ifp); SK_LOCK(); } nx_netif_agent_fini(nif); nx_netif_capabilities_fini(nif); nx_netif_flow_fini(nif); nx_netif_filter_fini(nif); nx_netif_llink_fini(nif); nx_netif_flags_fini(nif); uuid_clear(nif->nif_uuid); /* nx_netif_{compat_}attach() held both references */ na_release_locked(nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV)); na_release_locked(nx_port_get_na(nx, NEXUS_PORT_NET_IF_HOST)); nx_port_free(nx, NEXUS_PORT_NET_IF_DEV); nx_port_free(nx, NEXUS_PORT_NET_IF_HOST); ifp->if_na_ops = NULL; ifp->if_na = NULL; nif->nif_ifp = NULL; nif->nif_netif_nxadv = NULL; SKYWALK_CLEAR_CAPABLE(ifp); if (suspended) { ifnet_datamov_resume(ifp); } #if (DEVELOPMENT || DEBUG) skoid_destroy(&nif->nif_skoid); #endif /* !DEVELOPMENT && !DEBUG */ } done: #if SK_LOG if (nsr != NULL) { uuid_string_t ifuuidstr; SK_DF(err ? SK_VERB_ERROR : SK_VERB_NETIF, "nexus 0x%llx (%s) if_uuid %s flags 0x%x err %d", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name, sk_uuid_unparse(nsr->nsr_if_uuid, ifuuidstr), nsr->nsr_flags, err); } else { SK_DF(err ? SK_VERB_ERROR : SK_VERB_NETIF, "nexus 0x%llx (%s) err %d", SK_KVA(nx), NX_DOM_PROV(nx)->nxdom_prov_name, err); } #endif /* SK_LOG */ return err; } /* * XXX * These checks are copied from fsw.c * There are no tests exercising this code. Do we still need this? */ SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl_flow_check(struct nx_netif *nif, nxcfg_cmd_t cmd, struct proc *p, struct nx_flow_req *req) { #pragma unused(nif) boolean_t need_check; int error; if (uuid_is_null(req->nfr_flow_uuid)) { return EINVAL; } req->nfr_flags &= NXFLOWREQF_MASK; req->nfr_flowadv_idx = FLOWADV_IDX_NONE; if (cmd == NXCFG_CMD_FLOW_DEL) { return 0; } need_check = FALSE; if (req->nfr_epid != -1 && proc_pid(p) != req->nfr_epid) { need_check = TRUE; } else if (!uuid_is_null(req->nfr_euuid)) { uuid_t uuid; /* get the UUID of the issuing process */ proc_getexecutableuuid(p, uuid, sizeof(uuid)); /* * If this is not issued by a process for its own * executable UUID and if the process does not have * the necessary privilege, reject the request. * The logic is similar to so_set_effective_uuid(). */ if (uuid_compare(req->nfr_euuid, uuid) != 0) { need_check = TRUE; } } if (need_check) { kauth_cred_t cred = kauth_cred_proc_ref(p); error = priv_check_cred(cred, PRIV_NET_PRIVILEGED_SOCKET_DELEGATE, 0); kauth_cred_unref(&cred); if (error != 0) { return error; } } return 0; } SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl_flow_add(struct nx_netif *nif, struct proc *p, struct nx_flow_req *req) { int err; ASSERT(p != PROC_NULL); err = nx_netif_ctl_flow_check(nif, NXCFG_CMD_FLOW_ADD, p, req); if (err != 0) { return err; } /* init kernel only fields */ nx_flow_req_internalize(req); req->nfr_context = NULL; req->nfr_flow_stats = NULL; req->nfr_port_reservation = NULL; req->nfr_pid = proc_pid(p); err = nx_netif_netagent_flow_add(nif, req); nx_flow_req_externalize(req); return err; } SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl_flow_del(struct nx_netif *nif, struct proc *p, struct nx_flow_req *req) { int err; err = nx_netif_ctl_flow_check(nif, NXCFG_CMD_FLOW_DEL, p, req); if (err != 0) { return err; } nx_flow_req_internalize(req); req->nfr_pid = proc_pid(p); err = nx_netif_netagent_flow_del(nif, req); nx_flow_req_externalize(req); return err; } SK_NO_INLINE_ATTRIBUTE static int nx_netif_ctl(struct kern_nexus *nx, nxcfg_cmd_t nc_cmd, void *data, struct proc *p) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); struct nx_spec_req *nsr = data; struct nx_flow_req *nfr = data; int error = 0; SK_LOCK_ASSERT_HELD(); switch (nc_cmd) { case NXCFG_CMD_ATTACH: error = nx_netif_ctl_attach(nx, nsr, p); break; case NXCFG_CMD_DETACH: error = nx_netif_ctl_detach(nx, nsr); break; case NXCFG_CMD_FLOW_ADD: error = nx_netif_ctl_flow_add(nif, p, nfr); break; case NXCFG_CMD_FLOW_DEL: error = nx_netif_ctl_flow_del(nif, p, nfr); break; default: SK_ERR("invalid cmd %u", nc_cmd); error = EINVAL; break; } return error; } static void nx_netif_llink_notify(struct kern_nexus *nx, struct netif_llink *llink, uint32_t flags) { #pragma unused(flags) struct netif_qset *qset; SLIST_FOREACH(qset, &llink->nll_qset_list, nqs_list) { (void) nx_tx_qset_notify(nx, qset->nqs_ctx); } } static void nx_netif_llink_notify_all(struct kern_nexus *nx, uint32_t flags) { struct nx_netif *nif; struct netif_llink *llink; nif = NX_NETIF_PRIVATE(nx); lck_rw_lock_shared(&nif->nif_llink_lock); STAILQ_FOREACH(llink, &nif->nif_llink_list, nll_link) { nx_netif_llink_notify(nx, llink, flags); } lck_rw_unlock_shared(&nif->nif_llink_lock); } /* * if_start() callback for native Skywalk interfaces, registered * at ifnet_allocate_extended() time, and invoked by the ifnet * starter thread. */ static void nx_netif_doorbell_internal(struct ifnet *ifp, uint32_t flags) { if (__improbable(ifp->if_na == NULL)) { return; } /* * Do this only if the nexus adapter is active, i.e. a channel * has been opened to it by the module above (flowswitch, etc.) */ struct nexus_adapter *hwna = &NA(ifp)->nifna_up; if (__probable(NA_IS_ACTIVE(hwna))) { struct kern_nexus *nx = hwna->na_nx; /* update our work timestamp */ hwna->na_work_ts = _net_uptime; if (NX_LLINK_PROV(nx)) { nx_netif_llink_notify_all(nx, flags); } else { struct __kern_channel_ring *kring; /* for doorbell purposes, use TX ring 0 */ kring = &hwna->na_tx_rings[0]; /* Issue a synchronous TX doorbell on the netif device ring */ kring->ckr_na_sync(kring, PROC_NULL, (NA_SYNCF_NETIF_DOORBELL | NA_SYNCF_NETIF_IFSTART)); } } else { struct netif_stats *nifs = &NX_NETIF_PRIVATE(hwna->na_nx)->nif_stats; STATS_INC(nifs, NETIF_STATS_DROP_NA_INACTIVE); } } static void nx_netif_doorbell(struct ifnet *ifp) { nx_netif_doorbell_internal(ifp, NETIF_XMIT_FLAG_HOST); } /* * TX sync callback, called from nx_netif_doorbell() where we'd expect to * perform synchronous TX doorbell to the driver, by invoking the driver's * doorbell callback directly in the same thread context. It is also called * when the layer above performs a TX sync operation, where we might need * to do an asynchronous doorbell instead, by simply calling ifnet_start(). */ static int nx_netif_na_txsync(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags) { #pragma unused(p) struct ifnet *ifp = KRNA(kring)->na_ifp; boolean_t sync_only; int ret = 0; ASSERT(ifp != NULL); SK_DF(SK_VERB_NETIF | SK_VERB_SYNC | SK_VERB_TX, "%s(%d) kr \"%s\" (0x%llx) krflags 0x%b ring %u flags 0%x", sk_proc_name_address(p), sk_proc_pid(p), kring->ckr_name, SK_KVA(kring), kring->ckr_flags, CKRF_BITS, kring->ckr_ring_id, flags); if (__improbable(!IF_FULLY_ATTACHED(ifp))) { SK_ERR("kr 0x%llx ifp %s (0x%llx), interface not attached", SK_KVA(kring), if_name(ifp), SK_KVA(ifp)); return ENXIO; } if (__improbable((ifp->if_start_flags & IFSF_FLOW_CONTROLLED) != 0)) { SK_DF(SK_VERB_SYNC | SK_VERB_TX, "kr 0x%llx ifp %s (0x%llx), " "flow control ON", SK_KVA(kring), if_name(ifp), SK_KVA(ifp)); return ENXIO; } /* update our work timestamp */ KRNA(kring)->na_work_ts = _net_uptime; sync_only = ((flags & NA_SYNCF_SYNC_ONLY) != 0) || !KR_KERNEL_ONLY(kring); /* regular sync (reclaim) */ if ((flags & NA_SYNCF_NETIF) != 0 || __improbable(sync_only)) { ret = nx_sync_tx(kring, (flags & NA_SYNCF_FORCE_RECLAIM) || kring->ckr_pending_intr != 0); kring->ckr_pending_intr = 0; /* direct user channels do not need to use the doorbell */ if (__improbable(sync_only)) { return ret; } } /* * Doorbell call. Here we do doorbell explicitly if the flag is * set or implicitly if we're opened directly by a user channel. * Synchronous vs. asynchronous depending on the context. */ if (__probable((flags & NA_SYNCF_NETIF_DOORBELL) != 0)) { if ((flags & NA_SYNCF_NETIF_IFSTART) != 0) { ASSERT(!(flags & NA_SYNCF_NETIF_IFSTART) || !(flags & NA_SYNCF_NETIF_ASYNC)); nx_tx_doorbell(kring, (flags & NA_SYNCF_NETIF_ASYNC)); } else { ifnet_start(ifp); } } return ret; } static int nx_netif_na_rxsync(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags) { #pragma unused(p) int ret; SK_DF(SK_VERB_NETIF | SK_VERB_SYNC | SK_VERB_RX, "%s(%d) kr \"%s\" (0x%llx) krflags 0x%b ring %u flags 0%x", sk_proc_name_address(p), sk_proc_pid(p), kring->ckr_name, SK_KVA(kring), kring->ckr_flags, CKRF_BITS, kring->ckr_ring_id, flags); ASSERT(kring->ckr_rhead <= kring->ckr_lim); /* update our work timestamp */ KRNA(kring)->na_work_ts = _net_uptime; ret = nx_sync_rx(kring, (flags & NA_SYNCF_FORCE_READ) || kring->ckr_pending_intr != 0); kring->ckr_pending_intr = 0; return ret; } static void nx_netif_na_dtor(struct nexus_adapter *na) { struct ifnet *ifp; struct nexus_netif_adapter *nifna = NIFNA(na); SK_LOCK_ASSERT_HELD(); ASSERT(na->na_type == NA_NETIF_DEV || na->na_type == NA_NETIF_HOST); SK_DF(SK_VERB_NETIF, "na \"%s\" (0x%llx)", na->na_name, SK_KVA(na)); /* * If the finalizer callback hasn't been called for whatever * reasons, pick up the embryonic ifnet stored in na_private. * Otherwise, release the I/O refcnt of a non-NULL na_ifp. */ if ((ifp = na->na_ifp) == NULL) { ifp = na->na_private; na->na_private = NULL; } else { ifnet_decr_iorefcnt(ifp); na->na_ifp = NULL; } if (nifna->nifna_netif != NULL) { nx_netif_release(nifna->nifna_netif); nifna->nifna_netif = NULL; } ASSERT(SKYWALK_NATIVE(ifp)); } /* * Dispatch rx/tx interrupts to the channel rings. * * The 'notify' routine depends on what the ring is attached to. * - for a channel file descriptor, do an event wakeup on the individual * waitqueue, plus one on the global one if needed (see na_notify) * - for a device port connected to a FlowSwitch, call the proper * forwarding routine; see nx_fsw_tx_hwna_notify() * or nx_fsw_rx_hwna_notify(). */ int nx_netif_common_intr(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags, uint32_t *work_done) { struct netif_stats *nifs = &NX_NETIF_PRIVATE(KRNA(kring)->na_nx)->nif_stats; int (*notify)(struct __kern_channel_ring *kring, struct proc *, uint32_t flags); int ret; KDBG((SK_KTRACE_NETIF_COMMON_INTR | DBG_FUNC_START), SK_KVA(kring)); SK_DF(SK_VERB_NETIF | SK_VERB_INTR | ((kring->ckr_tx == NR_RX) ? SK_VERB_RX : SK_VERB_TX), "na \"%s\" (0x%llx) kr \"%s\" (0x%llx) krflags 0x%b", KRNA(kring)->na_name, SK_KVA(KRNA(kring)), kring->ckr_name, SK_KVA(kring), kring->ckr_flags, CKRF_BITS); /* update our work timestamp */ KRNA(kring)->na_work_ts = _net_uptime; kring->ckr_pending_intr++; if (work_done != NULL) { *work_done = 1; /* do not fire again */ } /* * We can't be calling ckr_na_notify here since we could already be * intercepting it, else we'd end up recursively calling ourselves. * Use the original na_notify callback saved during na_activate, or in * the case when the module above us is the flowswitch, the notify * routine that it has installed in place of our original one. */ if (__probable(!KR_DROP(kring) && (notify = kring->ckr_netif_notify) != NULL)) { ret = notify(kring, p, flags); } else { /* * If the ring is in drop mode, pretend as if it's busy. * This allows the mitigation thread to pause for a while * before attempting again. */ ret = EBUSY; } if (__improbable(ret != 0)) { switch (kring->ckr_tx) { case NR_RX: if (ret == EBUSY) { STATS_INC(nifs, NETIF_STATS_RX_IRQ_BUSY); } else if (ret == EAGAIN) { STATS_INC(nifs, NETIF_STATS_RX_IRQ_AGAIN); } else { STATS_INC(nifs, NETIF_STATS_RX_IRQ_ERR); } break; case NR_TX: if (ret == EBUSY) { STATS_INC(nifs, NETIF_STATS_TX_IRQ_BUSY); } else if (ret == EAGAIN) { STATS_INC(nifs, NETIF_STATS_TX_IRQ_AGAIN); } else { STATS_INC(nifs, NETIF_STATS_TX_IRQ_ERR); } break; default: break; } } KDBG((SK_KTRACE_NETIF_COMMON_INTR | DBG_FUNC_END), SK_KVA(kring), ret); return ret; } static int nx_netif_na_notify_tx(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags) { return nx_netif_mit_tx_intr(kring, p, flags, NULL); } static int nx_netif_na_notify_rx(struct __kern_channel_ring *kring, struct proc *p, uint32_t flags) { int ret; /* * In the event the mitigation thread is disabled, protect * against recursion by detecting if we're already in the * context of an RX notify. IOSkywalkFamily may invoke the * notify callback as part of its RX sync callback. */ if (__probable(!sk_is_rx_notify_protected())) { sk_protect_t protect; uint32_t work_done; protect = sk_rx_notify_protect(); ret = nx_netif_mit_rx_intr(kring, p, flags, &work_done); sk_sync_unprotect(protect); } else { ret = EAGAIN; } return ret; } void nx_netif_mit_config(struct nexus_netif_adapter *nifna, boolean_t *tx_mit, boolean_t *tx_mit_simple, boolean_t *rx_mit, boolean_t *rx_mit_simple) { struct nx_netif *nif = nifna->nifna_netif; /* * TX mitigation is disabled by default, but can be * overridden via "sk_netif_tx_mit=N" boot-arg, where * N is one of SK_NETIF_MIT_FORCE_* values. */ *tx_mit = *tx_mit_simple = FALSE; switch (sk_netif_tx_mit) { case SK_NETIF_MIT_FORCE_SIMPLE: *tx_mit_simple = TRUE; OS_FALLTHROUGH; case SK_NETIF_MIT_FORCE_ADVANCED: *tx_mit = TRUE; break; case SK_NETIF_MIT_FORCE_OFF: case SK_NETIF_MIT_AUTO: ASSERT(*tx_mit == FALSE); break; default: VERIFY(0); /* NOTREACHED */ __builtin_unreachable(); } /* * RX mitigation is enabled by default only for BSD-style * virtual network interfaces, but can be overridden * via "sk_netif_rx_mit=N" boot-arg, where N is one of * SK_NETIF_MIT_FORCE_* values. */ *rx_mit = *rx_mit_simple = FALSE; switch (sk_netif_rx_mit) { case SK_NETIF_MIT_FORCE_OFF: ASSERT(*rx_mit == FALSE); break; case SK_NETIF_MIT_FORCE_SIMPLE: *rx_mit_simple = TRUE; OS_FALLTHROUGH; case SK_NETIF_MIT_FORCE_ADVANCED: *rx_mit = TRUE; break; case SK_NETIF_MIT_AUTO: *rx_mit_simple = TRUE; #if !XNU_TARGET_OS_OSX /* * On non-macOS platforms, enable RX mitigation * thread only for BSD-style virtual (and regular) * interfaces, since otherwise we may run out of * stack when subjected to IPsec processing, etc. */ *rx_mit = (NX_PROV(nifna->nifna_up.na_nx)->nxprov_flags & NXPROVF_VIRTUAL_DEVICE) && !NETIF_IS_LOW_LATENCY(nif); #else /* XNU_TARGET_OS_OSX */ /* * On macOS platform, enable RX mitigation on all but * low-latency interfaces, since we could potentially * have filter providers, etc. Ideally this should * be detected and dealt with dynamically. */ *rx_mit = !NETIF_IS_LOW_LATENCY(nif); #endif /* XNU_TARGET_OS_OSX */ break; default: VERIFY(0); /* NOTREACHED */ __builtin_unreachable(); } } static int nx_netif_na_activate(struct nexus_adapter *na, na_activate_mode_t mode) { struct nexus_netif_adapter *nifna = (struct nexus_netif_adapter *)na; boolean_t tx_mit, rx_mit, tx_mit_simple, rx_mit_simple; struct nx_netif *nif = nifna->nifna_netif; struct ifnet *ifp = na->na_ifp; int error = 0; uint32_t r; ASSERT(na->na_type == NA_NETIF_DEV); ASSERT(!(na->na_flags & NAF_HOST_ONLY)); SK_DF(SK_VERB_NETIF, "na \"%s\" (0x%llx) %s [%s]", na->na_name, SK_KVA(na), ifp->if_xname, na_activate_mode2str(mode)); switch (mode) { case NA_ACTIVATE_MODE_ON: ASSERT(SKYWALK_CAPABLE(ifp)); nx_netif_mit_config(nifna, &tx_mit, &tx_mit_simple, &rx_mit, &rx_mit_simple); /* * Init the mitigation support on all the dev TX rings. */ if (tx_mit) { nifna->nifna_tx_mit = skn_alloc_type_array(tx_on, struct nx_netif_mit, na_get_nrings(na, NR_TX), Z_WAITOK, skmem_tag_netif_mit); if (nifna->nifna_tx_mit == NULL) { SK_ERR("TX mitigation allocation failed"); error = ENOMEM; goto out; } } else { ASSERT(nifna->nifna_tx_mit == NULL); } /* * Init the mitigation support on all the dev RX rings. */ if (rx_mit) { nifna->nifna_rx_mit = skn_alloc_type_array(rx_on, struct nx_netif_mit, na_get_nrings(na, NR_RX), Z_WAITOK, skmem_tag_netif_mit); if (nifna->nifna_rx_mit == NULL) { SK_ERR("RX mitigation allocation failed"); if (nifna->nifna_tx_mit != NULL) { skn_free_type_array(rx_fail, struct nx_netif_mit, na_get_nrings(na, NR_TX), nifna->nifna_tx_mit); nifna->nifna_tx_mit = NULL; } error = ENOMEM; goto out; } } else { ASSERT(nifna->nifna_rx_mit == NULL); } /* intercept na_notify callback on the TX rings */ for (r = 0; r < na_get_nrings(na, NR_TX); r++) { na->na_tx_rings[r].ckr_netif_notify = na->na_tx_rings[r].ckr_na_notify; na->na_tx_rings[r].ckr_na_notify = nx_netif_na_notify_tx; if (nifna->nifna_tx_mit != NULL) { nx_netif_mit_init(nif, ifp, &nifna->nifna_tx_mit[r], &na->na_tx_rings[r], tx_mit_simple); } } /* intercept na_notify callback on the RX rings */ for (r = 0; r < na_get_nrings(na, NR_RX); r++) { na->na_rx_rings[r].ckr_netif_notify = na->na_rx_rings[r].ckr_na_notify; na->na_rx_rings[r].ckr_na_notify = nx_netif_na_notify_rx; if (nifna->nifna_rx_mit != NULL) { nx_netif_mit_init(nif, ifp, &nifna->nifna_rx_mit[r], &na->na_rx_rings[r], rx_mit_simple); } } nx_netif_filter_enable(nif); nx_netif_flow_enable(nif); atomic_bitset_32(&na->na_flags, NAF_ACTIVE); /* steer all start requests to netif; this must not fail */ lck_mtx_lock(&ifp->if_start_lock); error = ifnet_set_start_handler(ifp, nx_netif_doorbell); VERIFY(error == 0); lck_mtx_unlock(&ifp->if_start_lock); break; case NA_ACTIVATE_MODE_DEFUNCT: ASSERT(SKYWALK_CAPABLE(ifp)); break; case NA_ACTIVATE_MODE_OFF: /* * Note that here we cannot assert SKYWALK_CAPABLE() * as we're called in the destructor path. */ atomic_bitclear_32(&na->na_flags, NAF_ACTIVE); nx_netif_flow_disable(nif); nx_netif_filter_disable(nif); /* * Here we may block while holding sk_lock, but because * we've cleared NAF_ACTIVE above, kern_channel_tx_refill() * should immediately return. A better approach would be * to drop sk_lock and add a monitor for this routine. */ lck_mtx_lock(&ifp->if_start_lock); while (ifp->if_start_active != 0) { ++ifp->if_start_waiters; (void) msleep(&ifp->if_start_waiters, &ifp->if_start_lock, (PZERO - 1), na->na_name, NULL); } /* steer all start requests to default handler */ ifnet_reset_start_handler(ifp); lck_mtx_unlock(&ifp->if_start_lock); /* reset all TX notify callbacks */ for (r = 0; r < na_get_nrings(na, NR_TX); r++) { na->na_tx_rings[r].ckr_na_notify = na->na_tx_rings[r].ckr_netif_notify; na->na_tx_rings[r].ckr_netif_notify = NULL; if (nifna->nifna_tx_mit != NULL) { na->na_tx_rings[r].ckr_netif_mit_stats = NULL; nx_netif_mit_cleanup(&nifna->nifna_tx_mit[r]); } } if (nifna->nifna_tx_mit != NULL) { skn_free_type_array(tx_off, struct nx_netif_mit, na_get_nrings(na, NR_TX), nifna->nifna_tx_mit); nifna->nifna_tx_mit = NULL; } /* reset all RX notify callbacks */ for (r = 0; r < na_get_nrings(na, NR_RX); r++) { na->na_rx_rings[r].ckr_na_notify = na->na_rx_rings[r].ckr_netif_notify; na->na_rx_rings[r].ckr_netif_notify = NULL; if (nifna->nifna_rx_mit != NULL) { na->na_rx_rings[r].ckr_netif_mit_stats = NULL; nx_netif_mit_cleanup(&nifna->nifna_rx_mit[r]); } } if (nifna->nifna_rx_mit != NULL) { skn_free_type_array(rx_off, struct nx_netif_mit, na_get_nrings(na, NR_RX), nifna->nifna_rx_mit); nifna->nifna_rx_mit = NULL; } break; default: VERIFY(0); /* NOTREACHED */ __builtin_unreachable(); } out: return error; } SK_NO_INLINE_ATTRIBUTE static int nx_netif_attach(struct kern_nexus *nx, struct ifnet *ifp) __attribute__((optnone)) { struct nx_netif *nif = NX_NETIF_PRIVATE(nx); struct nxprov_params *nxp = NX_PROV(nx)->nxprov_params; struct nexus_netif_adapter *devnifna = NULL; struct nexus_netif_adapter *hostnifna = NULL; struct nexus_adapter *devna = NULL; struct nexus_adapter *hostna = NULL; boolean_t embryonic = FALSE; int retval = 0; uint32_t na_flags; SK_LOCK_ASSERT_HELD(); ASSERT(SKYWALK_NATIVE(ifp)); ASSERT(!SKYWALK_CAPABLE(ifp)); ASSERT(ifp->if_na == NULL); ASSERT(ifp->if_na_ops == NULL); devnifna = na_netif_alloc(Z_WAITOK); hostnifna = na_netif_alloc(Z_WAITOK); /* * We can be called for two different interface states: * * Fully attached: get an io ref count; upon success, this * holds a reference to the ifnet for the ifp pointer stored * in 'na_ifp' down below for both adapters. * * Embryonic: temporary hold the ifnet in na_private, which * upon a successful ifnet_attach(), will be moved over to * the 'na_ifp' with an io ref count held. * * The ifnet in 'na_ifp' will be released by na_release_locked(). */ if (!ifnet_is_attached(ifp, 1)) { if (!(ifp->if_refflags & IFRF_EMBRYONIC)) { ifp = NULL; retval = ENXIO; goto err; } embryonic = TRUE; } /* initialize the device netif adapter */ devnifna->nifna_netif = nif; nx_netif_retain(nif); devna = &devnifna->nifna_up; devna->na_type = NA_NETIF_DEV; devna->na_free = na_netif_free; (void) strncpy(devna->na_name, ifp->if_xname, sizeof(devna->na_name) - 1); devna->na_name[sizeof(devna->na_name) - 1] = '\0'; uuid_generate_random(devna->na_uuid); if (embryonic) { /* * We will move this over to na_ifp once * the interface is fully attached. */ devna->na_private = ifp; ASSERT(devna->na_ifp == NULL); } else { ASSERT(devna->na_private == NULL); /* use I/O refcnt from ifnet_is_attached() */ devna->na_ifp = ifp; } devna->na_activate = nx_netif_na_activate; devna->na_channel_event_notify = nx_netif_na_channel_event_notify; devna->na_txsync = nx_netif_na_txsync; devna->na_rxsync = nx_netif_na_rxsync; devna->na_dtor = nx_netif_na_dtor; devna->na_krings_create = nx_netif_dev_krings_create; devna->na_krings_delete = nx_netif_dev_krings_delete; devna->na_special = nx_netif_na_special; na_flags = NAF_NATIVE; if (NX_PROV(nx)->nxprov_flags & NXPROVF_VIRTUAL_DEVICE) { na_flags |= NAF_VIRTUAL_DEVICE; } if (NX_LLINK_PROV(nx)) { /* * while operating in logical link mode, we don't need to * create backing memory regions for the rings as they are * not used. */ na_flags |= NAF_MEM_NO_INIT; } atomic_bitset_32(&devna->na_flags, na_flags); *(nexus_stats_type_t *)(uintptr_t)&devna->na_stats_type = NEXUS_STATS_TYPE_INVALID; na_set_nrings(devna, NR_TX, nxp->nxp_tx_rings); na_set_nrings(devna, NR_RX, nxp->nxp_rx_rings); na_set_nslots(devna, NR_TX, nxp->nxp_tx_slots); na_set_nslots(devna, NR_RX, nxp->nxp_rx_slots); /* * Verify upper bounds; the parameters must have already been * validated by nxdom_prov_params() by the time we get here. */ ASSERT(na_get_nrings(devna, NR_TX) <= NX_DOM(nx)->nxdom_tx_rings.nb_max); ASSERT(na_get_nrings(devna, NR_RX) <= NX_DOM(nx)->nxdom_rx_rings.nb_max); ASSERT(na_get_nslots(devna, NR_TX) <= NX_DOM(nx)->nxdom_tx_slots.nb_max); ASSERT(na_get_nslots(devna, NR_RX) <= NX_DOM(nx)->nxdom_rx_slots.nb_max); na_attach_common(devna, nx, &nx_netif_prov_s); if ((retval = NX_DOM_PROV(nx)->nxdom_prov_mem_new(NX_DOM_PROV(nx), nx, devna)) != 0) { ASSERT(devna->na_arena == NULL); goto err; } ASSERT(devna->na_arena != NULL); *(uint32_t *)(uintptr_t)&devna->na_flowadv_max = nxp->nxp_flowadv_max; ASSERT(devna->na_flowadv_max == 0 || skmem_arena_nexus(devna->na_arena)->arn_flowadv_obj != NULL); /* setup packet copy routines */ if (skmem_arena_nexus(devna->na_arena)->arn_rx_pp->pp_max_frags > 1) { nif->nif_pkt_copy_from_mbuf = pkt_copy_multi_buflet_from_mbuf; nif->nif_pkt_copy_to_mbuf = pkt_copy_multi_buflet_to_mbuf; nif->nif_pkt_copy_from_pkt = pkt_copy_multi_buflet_from_pkt; } else { nif->nif_pkt_copy_from_mbuf = pkt_copy_from_mbuf; nif->nif_pkt_copy_to_mbuf = pkt_copy_to_mbuf; nif->nif_pkt_copy_from_pkt = pkt_copy_from_pkt; } /* initialize the host netif adapter */ hostnifna->nifna_netif = nif; nx_netif_retain(nif); hostna = &hostnifna->nifna_up; (void) snprintf(hostna->na_name, sizeof(hostna->na_name), "%s^", devna->na_name); uuid_generate_random(hostna->na_uuid); if (embryonic) { /* * We will move this over to na_ifp once * the interface is fully attached. */ hostna->na_private = ifp; ASSERT(hostna->na_ifp == NULL); } else { ASSERT(hostna->na_private == NULL); hostna->na_ifp = devna->na_ifp; ifnet_incr_iorefcnt(hostna->na_ifp); } hostna->na_type = NA_NETIF_HOST; hostna->na_free = na_netif_free; hostna->na_activate = nx_netif_host_na_activate; hostna->na_txsync = nx_netif_host_na_txsync; hostna->na_rxsync = nx_netif_host_na_rxsync; hostna->na_dtor = nx_netif_na_dtor; hostna->na_krings_create = nx_netif_host_krings_create; hostna->na_krings_delete = nx_netif_host_krings_delete; hostna->na_special = nx_netif_host_na_special; na_flags = NAF_HOST_ONLY | NAF_NATIVE; if (NX_LLINK_PROV(nx)) { /* * while operating in logical link mode, we don't need to * create backing memory regions for the rings as they are * not used. */ na_flags |= NAF_MEM_NO_INIT; } atomic_bitset_32(&hostna->na_flags, na_flags); *(nexus_stats_type_t *)(uintptr_t)&hostna->na_stats_type = NEXUS_STATS_TYPE_INVALID; na_set_nrings(hostna, NR_TX, 1); na_set_nrings(hostna, NR_RX, 1); na_set_nslots(hostna, NR_TX, nxp->nxp_tx_slots); na_set_nslots(hostna, NR_RX, nxp->nxp_rx_slots); na_attach_common(hostna, nx, &nx_netif_prov_s); if ((retval = NX_DOM_PROV(nx)->nxdom_prov_mem_new(NX_DOM_PROV(nx), nx, hostna)) != 0) { ASSERT(hostna->na_arena == NULL); goto err; } ASSERT(hostna->na_arena != NULL); *(uint32_t *)(uintptr_t)&hostna->na_flowadv_max = nxp->nxp_flowadv_max; ASSERT(hostna->na_flowadv_max == 0 || skmem_arena_nexus(hostna->na_arena)->arn_flowadv_obj != NULL); /* adjust the classq packet drop limit */ if (embryonic) { uint32_t drop_lim; struct kern_pbufpool_memory_info pp_info; retval = kern_pbufpool_get_memory_info(nx->nx_tx_pp, &pp_info); VERIFY(retval == 0); /* set the drop limit as 80% of size of packet pool */ drop_lim = (pp_info.kpm_packets * 4) / 5; VERIFY(drop_lim != 0); IFCQ_PKT_DROP_LIMIT(ifp->if_snd) = drop_lim; } /* these will be undone by destructor */ ifp->if_na_ops = &na_netif_ops; ifp->if_na = devnifna; na_retain_locked(devna); na_retain_locked(hostna); SKYWALK_SET_CAPABLE(ifp); NETIF_WLOCK(nif); nif->nif_ifp = ifp; nif->nif_netif_nxadv = nx->nx_adv.netif_nxv_adv; retval = nx_port_alloc(nx, NEXUS_PORT_NET_IF_DEV, NULL, &devna, kernproc); ASSERT(retval == 0); retval = nx_port_alloc(nx, NEXUS_PORT_NET_IF_HOST, NULL, &hostna, kernproc); ASSERT(retval == 0); NETIF_WUNLOCK(nif); #if SK_LOG uuid_string_t uuidstr; SK_DF(SK_VERB_NETIF, "devna: \"%s\"", devna->na_name); SK_DF(SK_VERB_NETIF, " UUID: %s", sk_uuid_unparse(devna->na_uuid, uuidstr)); SK_DF(SK_VERB_NETIF, " nx: 0x%llx (\"%s\":\"%s\")", SK_KVA(devna->na_nx), NX_DOM(devna->na_nx)->nxdom_name, NX_DOM_PROV(devna->na_nx)->nxdom_prov_name); SK_DF(SK_VERB_NETIF, " flags: 0x%b", devna->na_flags, NAF_BITS); SK_DF(SK_VERB_NETIF, " flowadv_max: %u", devna->na_flowadv_max); SK_DF(SK_VERB_NETIF, " rings: tx %u rx %u", na_get_nrings(devna, NR_TX), na_get_nrings(devna, NR_RX)); SK_DF(SK_VERB_NETIF, " slots: tx %u rx %u", na_get_nslots(devna, NR_TX), na_get_nslots(devna, NR_RX)); #if CONFIG_NEXUS_USER_PIPE SK_DF(SK_VERB_NETIF, " next_pipe: %u", devna->na_next_pipe); SK_DF(SK_VERB_NETIF, " max_pipes: %u", devna->na_max_pipes); #endif /* CONFIG_NEXUS_USER_PIPE */ SK_DF(SK_VERB_NETIF, " ifp: 0x%llx %s [ioref %u]", SK_KVA(ifp), ifp->if_xname, ifp->if_refio); SK_DF(SK_VERB_NETIF, "hostna: \"%s\"", hostna->na_name); SK_DF(SK_VERB_NETIF, " UUID: %s", sk_uuid_unparse(hostna->na_uuid, uuidstr)); SK_DF(SK_VERB_NETIF, " nx: 0x%llx (\"%s\":\"%s\")", SK_KVA(hostna->na_nx), NX_DOM(hostna->na_nx)->nxdom_name, NX_DOM_PROV(hostna->na_nx)->nxdom_prov_name); SK_DF(SK_VERB_NETIF, " flags: 0x%b", hostna->na_flags, NAF_BITS); SK_DF(SK_VERB_NETIF, " flowadv_max: %u", hostna->na_flowadv_max); SK_DF(SK_VERB_NETIF, " rings: tx %u rx %u", na_get_nrings(hostna, NR_TX), na_get_nrings(hostna, NR_RX)); SK_DF(SK_VERB_NETIF, " slots: tx %u rx %u", na_get_nslots(hostna, NR_TX), na_get_nslots(hostna, NR_RX)); #if CONFIG_NEXUS_USER_PIPE SK_DF(SK_VERB_NETIF, " next_pipe: %u", hostna->na_next_pipe); SK_DF(SK_VERB_NETIF, " max_pipes: %u", hostna->na_max_pipes); #endif /* CONFIG_NEXUS_USER_PIPE */ SK_DF(SK_VERB_NETIF, " ifp: 0x%llx %s [ioref %u]", SK_KVA(ifp), ifp->if_xname, ifp->if_refio); #endif /* SK_LOG */ err: if (retval != 0) { if (ifp != NULL) { if (!embryonic) { ifnet_decr_iorefcnt(ifp); } ifp = NULL; } if (devna != NULL) { if (devna->na_arena != NULL) { skmem_arena_release(devna->na_arena); devna->na_arena = NULL; } if (devna->na_ifp != NULL) { ifnet_decr_iorefcnt(devna->na_ifp); devna->na_ifp = NULL; } devna->na_private = NULL; } if (hostna != NULL) { if (hostna->na_arena != NULL) { skmem_arena_release(hostna->na_arena); hostna->na_arena = NULL; } if (hostna->na_ifp != NULL) { ifnet_decr_iorefcnt(hostna->na_ifp); hostna->na_ifp = NULL; } hostna->na_private = NULL; } if (devnifna != NULL) { if (devnifna->nifna_netif != NULL) { nx_netif_release(devnifna->nifna_netif); devnifna->nifna_netif = NULL; } na_netif_free((struct nexus_adapter *)devnifna); } if (hostnifna != NULL) { if (hostnifna->nifna_netif != NULL) { nx_netif_release(hostnifna->nifna_netif); hostnifna->nifna_netif = NULL; } na_netif_free((struct nexus_adapter *)hostnifna); } } return retval; } /* * Any per-netif state that can be discovered at attach time should be * initialized here. */ static void nx_netif_flags_init(struct nx_netif *nif) { ifnet_t ifp = nif->nif_ifp; struct kern_nexus *nx = nif->nif_nx; struct nexus_adapter *devna = nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV); switch (devna->na_type) { case NA_NETIF_DEV: if (strcmp(ifp->if_name, sk_ll_prefix) == 0) { nif->nif_flags |= NETIF_FLAG_LOW_LATENCY; if_set_xflags(ifp, IFXF_LOW_LATENCY); } break; case NA_NETIF_COMPAT_DEV: nif->nif_flags |= NETIF_FLAG_COMPAT; break; default: break; } } /* * This is also supposed to check for any inconsistent state at detach time. */ static void nx_netif_flags_fini(struct nx_netif *nif) { ifnet_t ifp = nif->nif_ifp; if (ifp != NULL) { if_clear_xflags(ifp, IFXF_LOW_LATENCY); } nif->nif_flags = 0; } static void nx_netif_capabilities_init(struct nx_netif *nif) { struct kern_nexus_capab_interface_advisory kncia; struct kern_nexus *nx = nif->nif_nx; nxprov_capab_config_fn_t capab_fn; uint32_t capab_len; int error; if ((NX_PROV(nx)->nxprov_netif_ext.nxnpi_version) == KERN_NEXUS_PROVIDER_VERSION_NETIF) { capab_fn = NX_PROV(nx)->nxprov_netif_ext.nxnpi_config_capab; ASSERT(capab_fn != NULL); } else { capab_fn = NX_PROV(nx)->nxprov_ext.nxpi_config_capab; } if (capab_fn == NULL) { return; } /* check/configure interface advisory notifications */ if ((nif->nif_ifp->if_eflags & IFEF_ADV_REPORT) != 0) { bzero(&kncia, sizeof(kncia)); kncia.kncia_version = KERN_NEXUS_CAPAB_INTERFACE_ADVISORY_VERSION_1; *__DECONST(kern_nexus_capab_interface_advisory_notify_fn_t *, &(kncia.kncia_notify)) = nx_netif_interface_advisory_notify; *__DECONST(void **, &(kncia.kncia_kern_context)) = nx; capab_len = sizeof(kncia); error = capab_fn(NX_PROV(nx), nx, KERN_NEXUS_CAPAB_INTERFACE_ADVISORY, &kncia, &capab_len); if (error == 0) { VERIFY(kncia.kncia_config != NULL); VERIFY(kncia.kncia_provider_context != NULL); nif->nif_intf_adv_config = kncia.kncia_config; nif->nif_intf_adv_prov_ctx = kncia.kncia_provider_context; } } } static void nx_netif_capabilities_fini(struct nx_netif *nif) { nif->nif_intf_adv_config = NULL; nif->nif_intf_adv_prov_ctx = NULL; } void na_netif_finalize(struct nexus_netif_adapter *nifna, struct ifnet *ifp) { struct nx_netif *nif = nifna->nifna_netif; struct kern_nexus *nx = nif->nif_nx; struct nexus_adapter *devna = nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV); struct nexus_adapter *hostna = nx_port_get_na(nx, NEXUS_PORT_NET_IF_HOST); ASSERT(devna != NULL); ASSERT(hostna != NULL); if (!ifnet_is_attached(ifp, 1)) { VERIFY(0); /* NOTREACHED */ __builtin_unreachable(); } ASSERT(devna->na_private == ifp); ASSERT(devna->na_ifp == NULL); /* use I/O refcnt held by ifnet_is_attached() above */ devna->na_ifp = devna->na_private; devna->na_private = NULL; ASSERT(hostna->na_private == ifp); ASSERT(hostna->na_ifp == NULL); hostna->na_ifp = hostna->na_private; hostna->na_private = NULL; ifnet_incr_iorefcnt(hostna->na_ifp); nx_netif_flags_init(nif); nx_netif_llink_init(nif); nx_netif_filter_init(nif); nx_netif_flow_init(nif); nx_netif_capabilities_init(nif); nx_netif_agent_init(nif); } void nx_netif_reap(struct nexus_netif_adapter *nifna, struct ifnet *ifp, uint32_t thres, boolean_t low) { #pragma unused(ifp) struct nx_netif *nif = nifna->nifna_netif; struct kern_nexus *nx = nif->nif_nx; struct nexus_adapter *devna = nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV); uint64_t now = _net_uptime; boolean_t purge; ASSERT(thres != 0); if (devna->na_work_ts == 0) { return; } /* * Purge if it's has been inactive for some time (twice the drain * threshold), and clear the work timestamp to temporarily skip this * adapter until it's active again. Purging cached objects can be * expensive since we'd need to allocate and construct them again, * so we do it only when necessary. */ if (low || (now - devna->na_work_ts) >= (thres << 1)) { devna->na_work_ts = 0; purge = TRUE; } else { purge = FALSE; } SK_DF(SK_VERB_NETIF, "%s: %s na %s", ifp->if_xname, (purge ? "purging" : "pruning"), devna->na_name); /* * Device and host adapters share the same packet buffer pool, * so just reap the arena belonging to the device instance. */ skmem_arena_reap(devna->na_arena, purge); /* * Reap any caches configured for classq. */ ifclassq_reap_caches(purge); } void nx_netif_copy_stats(struct nexus_netif_adapter *nifna, struct if_netif_stats *if_ns) { struct nx_netif_mit *mit; struct mit_cfg_tbl *mit_cfg; if ((mit = nifna->nifna_rx_mit) == NULL) { return; } if ((mit->mit_flags & NETIF_MITF_INITIALIZED) == 0) { return; } if_ns->ifn_rx_mit_interval = mit->mit_interval; if_ns->ifn_rx_mit_mode = mit->mit_mode; if_ns->ifn_rx_mit_packets_avg = mit->mit_packets_avg; if_ns->ifn_rx_mit_packets_min = mit->mit_packets_min; if_ns->ifn_rx_mit_packets_max = mit->mit_packets_max; if_ns->ifn_rx_mit_bytes_avg = mit->mit_bytes_avg; if_ns->ifn_rx_mit_bytes_min = mit->mit_bytes_min; if_ns->ifn_rx_mit_bytes_max = mit->mit_bytes_max; if_ns->ifn_rx_mit_cfg_idx = mit->mit_cfg_idx; VERIFY(if_ns->ifn_rx_mit_cfg_idx < mit->mit_cfg_idx_max); mit_cfg = &mit->mit_tbl[if_ns->ifn_rx_mit_cfg_idx]; if_ns->ifn_rx_mit_cfg_packets_lowat = mit_cfg->cfg_plowat; if_ns->ifn_rx_mit_cfg_packets_hiwat = mit_cfg->cfg_phiwat; if_ns->ifn_rx_mit_cfg_bytes_lowat = mit_cfg->cfg_blowat; if_ns->ifn_rx_mit_cfg_bytes_hiwat = mit_cfg->cfg_bhiwat; if_ns->ifn_rx_mit_cfg_interval = mit_cfg->cfg_ival; } int nx_netif_na_special(struct nexus_adapter *na, struct kern_channel *ch, struct chreq *chr, nxspec_cmd_t spec_cmd) { ASSERT(na->na_type == NA_NETIF_DEV || na->na_type == NA_NETIF_COMPAT_DEV); return nx_netif_na_special_common(na, ch, chr, spec_cmd); } int nx_netif_na_special_common(struct nexus_adapter *na, struct kern_channel *ch, struct chreq *chr, nxspec_cmd_t spec_cmd) { int error = 0; ASSERT(na->na_type == NA_NETIF_DEV || na->na_type == NA_NETIF_HOST || na->na_type == NA_NETIF_COMPAT_DEV || na->na_type == NA_NETIF_COMPAT_HOST); SK_LOCK_ASSERT_HELD(); switch (spec_cmd) { case NXSPEC_CMD_CONNECT: /* * netif adapter isn't created exclusively for kernel. * We mark (and clear) NAF_KERNEL_ONLY flag upon a succesful * na_special() connect and disconnect. */ if (NA_KERNEL_ONLY(na)) { error = EBUSY; goto done; } ASSERT(!(na->na_flags & NAF_SPEC_INIT)); atomic_bitset_32(&na->na_flags, NAF_KERNEL_ONLY); error = na_bind_channel(na, ch, chr); if (error != 0) { atomic_bitclear_32(&na->na_flags, NAF_KERNEL_ONLY); goto done; } atomic_bitset_32(&na->na_flags, NAF_SPEC_INIT); break; case NXSPEC_CMD_DISCONNECT: ASSERT(NA_KERNEL_ONLY(na)); ASSERT(na->na_channels > 0); ASSERT(na->na_flags & NAF_SPEC_INIT); na_unbind_channel(ch); atomic_bitclear_32(&na->na_flags, (NAF_SPEC_INIT | NAF_KERNEL_ONLY)); break; case NXSPEC_CMD_START: na_kr_drop(na, FALSE); break; case NXSPEC_CMD_STOP: na_kr_drop(na, TRUE); LCK_MTX_ASSERT(&ch->ch_lock, LCK_MTX_ASSERT_NOTOWNED); lck_mtx_lock(&ch->ch_lock); nxprov_advise_disconnect(na->na_nx, ch); lck_mtx_unlock(&ch->ch_lock); break; default: error = EINVAL; break; } done: SK_DF(error ? SK_VERB_ERROR : SK_VERB_NETIF, "ch 0x%llx from na \"%s\" (0x%llx) naflags %b nx 0x%llx " "spec_cmd %u (err %d)", SK_KVA(ch), na->na_name, SK_KVA(na), na->na_flags, NAF_BITS, SK_KVA(ch->ch_nexus), spec_cmd, error); return error; } /* * Get a skywalk netif adapter for the port. */ int nx_netif_na_find(struct kern_nexus *nx, struct kern_channel *ch, struct chreq *chr, struct nxbind *nxb, struct proc *p, struct nexus_adapter **nap, boolean_t create) { #pragma unused(ch) struct nx_netif *nif = NX_NETIF_PRIVATE(nx); boolean_t anon = NX_ANONYMOUS_PROV(nx); ch_endpoint_t ep = chr->cr_endpoint; nexus_port_t nx_port = chr->cr_port; struct nexus_adapter *na = NULL; struct ifnet *ifp; int err = 0; SK_LOCK_ASSERT_HELD(); *nap = NULL; /* default */ #if SK_LOG uuid_string_t uuidstr; SK_D("name \"%s\" spec_uuid \"%s\" port %d mode 0x%b pipe_id %u " "ring_id %d ring_set %u ep_type %u:%u create %u%s", chr->cr_name, sk_uuid_unparse(chr->cr_spec_uuid, uuidstr), (int)chr->cr_port, chr->cr_mode, CHMODE_BITS, chr->cr_pipe_id, (int)chr->cr_ring_id, chr->cr_ring_set, chr->cr_real_endpoint, chr->cr_endpoint, create, (ep != CH_ENDPOINT_NET_IF) ? " (skipped)" : ""); #endif /* SK_LOG */ if (!create || ep != CH_ENDPOINT_NET_IF) { err = ENODEV; goto done; } ASSERT(NX_DOM(nx)->nxdom_type == NEXUS_TYPE_NET_IF); if (nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV) == NULL) { err = ENXIO; goto done; } ifp = nif->nif_ifp; if (!(SKYWALK_CAPABLE(ifp))) { SK_ERR("interface %s is no longer usable", if_name(ifp)); err = ENOTSUP; goto done; } if (chr->cr_mode & CHMODE_LOW_LATENCY) { SK_ERR("low latency is not supported for netif channel"); err = ENOTSUP; goto done; } switch (nx_port) { case NEXUS_PORT_NET_IF_DEV: /* * We have to reject direct user open that's not explicitly * allowed because netif nexuses do not by default have * user memory regions. */ if (p != kernproc && (!skywalk_netif_direct_allowed(ifp->if_xname) || (kauth_cred_issuser(kauth_cred_get()) == 0 && (anon || nif->nif_dev_nxb == NULL || nxb == NULL || !nxb_is_equal(nif->nif_dev_nxb, nxb))))) { DTRACE_SKYWALK2(direct__not__allowed, struct ifnet *, ifp, struct chreq *, chr); err = ENOTSUP; goto done; } if (chr->cr_mode & CHMODE_EVENT_RING) { SK_ERR("event ring is not supported for netif dev port channel"); err = ENOTSUP; goto done; } na = nx_port_get_na(nx, NEXUS_PORT_NET_IF_DEV); break; case NEXUS_PORT_NET_IF_HOST: if (p != kernproc) { err = ENOTSUP; goto done; } if (chr->cr_mode & CHMODE_EVENT_RING) { SK_ERR("event ring is not supported for netif host port channel"); err = ENOTSUP; goto done; } na = nx_port_get_na(nx, NEXUS_PORT_NET_IF_HOST); break; default: ASSERT(!(chr->cr_mode & CHMODE_CONFIG)); NETIF_WLOCK(nif); err = nx_port_alloc(nx, nx_port, nxb, &na, p); if (err != 0) { NETIF_WUNLOCK(nif); goto done; } if (na == NULL) { if (chr->cr_mode & CHMODE_FILTER) { err = netif_filter_na_create(nx, chr, &na); } else { err = netif_vp_na_create(nx, chr, &na); } if (err != 0) { NETIF_WUNLOCK(nif); goto done; } err = nx_port_alloc(nx, nx_port, nxb, &na, p); if (err != 0) { NETIF_WUNLOCK(nif); goto done; } } NETIF_WUNLOCK(nif); break; } ASSERT(err == 0); ASSERT(na != NULL); #if CONFIG_NEXUS_USER_PIPE if (NA_OWNED_BY_ANY(na) || na->na_next_pipe > 0) { #else /* !CONFIG_NEXUS_USER_PIPE */ if (NA_OWNED_BY_ANY(na)) { #endif /* !CONFIG_NEXUS_USER_PIPE */ err = EBUSY; na = NULL; goto done; } *nap = na; na_retain_locked(na); done: ASSERT(err != 0 || na != NULL); if (err) { SK_ERR("na not found, err(%d)", err); } else { SK_DF(SK_VERB_NETIF, "found na 0x%llu", na); } return err; } /* na_krings_create callback for all netif device adapters */ int nx_netif_dev_krings_create(struct nexus_adapter *na, struct kern_channel *ch) { int ret; ASSERT(na->na_type == NA_NETIF_DEV || na->na_type == NA_NETIF_COMPAT_DEV); /* * Allocate context structures for native netif only, for * IOSkywalkFamily to store its object references. */ ret = na_rings_mem_setup(na, 0, (na->na_flags & NAF_NATIVE), ch); /* * We mark CKRF_DROP for kernel-only rings (kernel channel * opened by the flowswitch, etc.) to prevent packets from * going thru until after the client of the kernel channel * has fully plumbed things on its side. For userland-facing * rings (regular channel opened to netif), this is not * required, and so don't mark CKRF_DROP there. */ if (ret == 0 && NA_KERNEL_ONLY(na)) { na_kr_drop(na, TRUE); } return ret; } /* call with SK_LOCK held */ void nx_netif_dev_krings_delete(struct nexus_adapter *na, struct kern_channel *ch, boolean_t defunct) { ASSERT(na->na_type == NA_NETIF_DEV || na->na_type == NA_NETIF_COMPAT_DEV); /* see comments in nx_netif_dev_krings_create() */ if (NA_KERNEL_ONLY(na)) { na_kr_drop(na, TRUE); } na_rings_mem_teardown(na, ch, defunct); } struct nx_netif * nx_netif_alloc(zalloc_flags_t how) { struct nx_netif *n; SK_LOCK_ASSERT_HELD(); n = zalloc_flags(nx_netif_zone, how | Z_ZERO); if (n == NULL) { return NULL; } NETIF_RWINIT(n); os_ref_init(&n->nif_refcnt, NULL); SK_DF(SK_VERB_MEM, "netif 0x%llx", SK_KVA(n)); return n; } static void nx_netif_destroy(struct nx_netif *n) { ASSERT(n->nif_dev_nxb == NULL); ASSERT(n->nif_host_nxb == NULL); ASSERT(os_ref_get_count(&n->nif_refcnt) == 0); nx_netif_llink_config_free(n); SK_DF(SK_VERB_MEM, "netif 0x%llx", SK_KVA(n)); NETIF_RWDESTROY(n); zfree(nx_netif_zone, n); } void nx_netif_release(struct nx_netif *n) { SK_LOCK_ASSERT_HELD(); SK_DF(SK_VERB_MEM, "netif 0x%llx, refcnt %d", SK_KVA(n), os_ref_get_count(&n->nif_refcnt)); if (os_ref_release(&n->nif_refcnt) == 0) { nx_netif_destroy(n); } } void nx_netif_retain(struct nx_netif *n) { SK_LOCK_ASSERT_HELD(); /* retaining an object with a zero refcount is not allowed */ ASSERT(os_ref_get_count(&n->nif_refcnt) >= 1); os_ref_retain(&n->nif_refcnt); SK_DF(SK_VERB_MEM, "netif 0x%llx, refcnt %d", SK_KVA(n), os_ref_get_count(&n->nif_refcnt)); } void nx_netif_free(struct nx_netif *n) { nx_netif_release(n); } static int nx_netif_na_channel_event_notify(struct nexus_adapter *na, struct __kern_packet *kpkt, struct __kern_channel_event *ev, uint16_t ev_len) { int err; struct netif_flow *nf; struct nexus_adapter *netif_vpna; struct nx_netif *nif = NIFNA(na)->nifna_netif; struct netif_stats *nifs = &NIFNA(na)->nifna_netif->nif_stats; NETIF_RLOCK(nif); if (!NETIF_IS_LOW_LATENCY(nif)) { err = ENOTSUP; goto error; } if (__improbable(!NA_IS_ACTIVE(na))) { STATS_INC(nifs, NETIF_STATS_EV_DROP_NA_INACTIVE); err = ENXIO; goto error; } if (__improbable(NA_IS_DEFUNCT(na))) { STATS_INC(nifs, NETIF_STATS_EV_DROP_NA_DEFUNCT); err = ENXIO; goto error; } if (__improbable(nif->nif_vp_cnt == 0)) { STATS_INC(nifs, NETIF_STATS_EV_DROP_NO_VPNA); err = ENXIO; goto error; } /* The returned netif flow is refcounted. */ nf = nx_netif_flow_classify(nif, kpkt, NETIF_FLOW_OUTBOUND); if (nf == NULL) { SK_ERR("unclassified event (%d) dropped", ev->ev_type); STATS_INC(nifs, NETIF_STATS_EV_DROP_DEMUX_ERR); err = ENOENT; goto error; } netif_vpna = (struct nexus_adapter *)nf->nf_cb_arg; if (netif_vpna->na_channel_event_notify != NULL) { err = netif_vpna->na_channel_event_notify(netif_vpna, kpkt, ev, ev_len); } else { STATS_INC(nifs, NETIF_STATS_EV_DROP_EV_VPNA_NOTSUP); err = ENOTSUP; } nx_netif_flow_release(nif, nf); NETIF_RUNLOCK(nif); nf = NULL; return err; error: STATS_INC(nifs, NETIF_STATS_EV_DROP); NETIF_RUNLOCK(nif); return err; } static int nx_netif_interface_advisory_notify_common(struct kern_nexus *nx, const struct ifnet_interface_advisory *advisory) { struct kern_nexus *notify_nx; struct __kern_netif_intf_advisory *intf_adv; struct nx_netif *nif = NX_NETIF_PRIVATE(nx); if (nif->nif_fsw_nxadv != NULL) { ASSERT(nif->nif_fsw != NULL); intf_adv = &nif->nif_fsw_nxadv->_nxadv_intf_adv; notify_nx = nif->nif_fsw->fsw_nx; } else { intf_adv = &nif->nif_netif_nxadv->__kern_intf_adv; notify_nx = nif->nif_nx; } /* * copy the advisory report in shared memory */ intf_adv->cksum = os_cpu_copy_in_cksum(advisory, &intf_adv->adv, sizeof(*advisory), 0); STATS_INC(&nif->nif_stats, NETIF_STATS_IF_ADV_UPD_RECV); /* * notify user channels on advisory report availability */ nx_interface_advisory_notify(notify_nx); return 0; } int nx_netif_interface_advisory_report(struct nexus_adapter *devna, const struct ifnet_interface_advisory *advisory) { ASSERT(devna->na_type == NA_NETIF_DEV); if (__improbable(!NA_IS_ACTIVE(devna))) { return ENXIO; } if (__improbable(NA_IS_DEFUNCT(devna))) { return ENXIO; } return nx_netif_interface_advisory_notify_common(devna->na_nx, advisory); } static errno_t nx_netif_interface_advisory_notify(void *kern_ctx, const struct ifnet_interface_advisory *advisory) { if (__improbable(kern_ctx == NULL || advisory == NULL || advisory->version != IF_INTERFACE_ADVISORY_VERSION_CURRENT)) { return EINVAL; } if (__improbable((advisory->direction != IF_INTERFACE_ADVISORY_DIRECTION_TX) && (advisory->direction != IF_INTERFACE_ADVISORY_DIRECTION_RX))) { return EINVAL; } return nx_netif_interface_advisory_notify_common(kern_ctx, advisory); } void nx_netif_config_interface_advisory(struct kern_nexus *nx, bool enable) { struct kern_nexus *nx_netif; struct nx_netif *nif; if (NX_REJECT_ACT(nx) || (nx->nx_flags & NXF_CLOSED) != 0) { return; } if (NX_PROV(nx)->nxprov_params->nxp_type == NEXUS_TYPE_FLOW_SWITCH) { struct nx_flowswitch *fsw = NX_FSW_PRIVATE(nx); nx_netif = fsw->fsw_nifna->na_nx; } else { nx_netif = nx; } ASSERT(NX_PROV(nx_netif)->nxprov_params->nxp_type == NEXUS_TYPE_NET_IF); nif = NX_NETIF_PRIVATE(nx_netif); if (nif->nif_intf_adv_config != NULL) { nif->nif_intf_adv_config(nif->nif_intf_adv_prov_ctx, enable); } } /* * This function has no use anymore since we are now passing truncated packets * to filters. We keep this logic just in case we need to prevent certain * packets from being passed to filters. */ static boolean_t packet_is_filterable(struct nexus_netif_adapter *nifna, struct __kern_packet *pkt) { #pragma unused (nifna, pkt) return TRUE; } /* * This function is only meant for supporting the RX path because the TX path * will not send packets > MTU size due to the disabling of TSO when filters * are enabled. */ static void get_filterable_packets(struct nexus_netif_adapter *nifna, struct __kern_packet *pkt_chain, struct __kern_packet **fpkt_chain, struct __kern_packet **passthrough_chain) { struct nx_netif *nif = nifna->nifna_netif; struct netif_stats *nifs = &nif->nif_stats; struct __kern_packet *pkt = pkt_chain, *next, *fpkt; struct __kern_packet *fpkt_head = NULL, *passthrough_head = NULL; struct __kern_packet **fpkt_tailp = &fpkt_head; struct __kern_packet **passthrough_tailp = &passthrough_head; int fcnt = 0, pcnt = 0, dcnt = 0; while (pkt != NULL) { next = pkt->pkt_nextpkt; pkt->pkt_nextpkt = NULL; if (!packet_is_filterable(nifna, pkt)) { pcnt++; *passthrough_tailp = pkt; passthrough_tailp = &pkt->pkt_nextpkt; pkt = next; continue; } fpkt = nx_netif_pkt_to_filter_pkt(nifna, pkt, NETIF_CONVERT_RX); if (fpkt != NULL) { fcnt++; *fpkt_tailp = fpkt; fpkt_tailp = &fpkt->pkt_nextpkt; } else { dcnt++; } pkt = next; } *fpkt_chain = fpkt_head; *passthrough_chain = passthrough_head; /* * No need to increment drop stats because that's already * done in nx_netif_pkt_to_filter_pkt. */ STATS_ADD(nifs, NETIF_STATS_FILTER_RX_NOT_FILTERABLE, pcnt); DTRACE_SKYWALK6(filterable, struct nexus_netif_adapter *, nifna, int, fcnt, int, pcnt, int, dcnt, struct __kern_packet *, fpkt_head, struct __kern_packet *, passthrough_head); } /* * This is only used by ring-based notify functions for now. * When a qset-based notify becomes available, this function can be used * unmodified. */ void netif_receive(struct nexus_netif_adapter *nifna, struct __kern_packet *pkt_chain, struct nexus_pkt_stats *stats) { struct nx_netif *nif = nifna->nifna_netif; struct nexus_adapter *na = &nifna->nifna_up; struct netif_stats *nifs = &nif->nif_stats; int err, dropcnt, dropstat = -1; /* update our work timestamp */ na->na_work_ts = _net_uptime; if (nif->nif_filter_cnt > 0) { struct __kern_packet *fpkt_chain = NULL; struct __kern_packet *passthrough_chain = NULL; get_filterable_packets(nifna, pkt_chain, &fpkt_chain, &passthrough_chain); if (fpkt_chain != NULL) { (void) nx_netif_filter_inject(nifna, NULL, fpkt_chain, NETIF_FILTER_RX | NETIF_FILTER_SOURCE); } if (passthrough_chain != NULL) { pkt_chain = passthrough_chain; } else { return; } } else if (nx_netif_filter_default_drop != 0) { DTRACE_SKYWALK2(rx__default__drop, struct nx_netif *, nif, struct __kern_packet *, pkt_chain); dropstat = NETIF_STATS_FILTER_DROP_DEFAULT; goto drop; } if (nif->nif_flow_cnt > 0) { struct __kern_packet *remain = NULL; err = nx_netif_demux(nifna, pkt_chain, &remain, NETIF_FLOW_SOURCE); if (remain == NULL) { return; } pkt_chain = remain; } if (na->na_rx != NULL) { na->na_rx(na, pkt_chain, stats); } else { DTRACE_SKYWALK2(no__rx__cb, struct nx_netif *, nif, struct __kern_packet *, pkt_chain); dropstat = NETIF_STATS_DROP_NO_RX_CB; goto drop; } return; drop: dropcnt = 0; nx_netif_free_packet_chain(pkt_chain, &dropcnt); if (dropstat != -1) { STATS_ADD(nifs, dropstat, dropcnt); } STATS_ADD(nifs, NETIF_STATS_DROP, dropcnt); } static slot_idx_t netif_rate_limit(struct __kern_channel_ring *r, uint64_t rate, slot_idx_t begin, slot_idx_t end, boolean_t *rate_limited) { uint64_t elapsed; uint64_t now; struct __kern_packet *pkt; clock_sec_t sec; clock_usec_t usec; slot_idx_t i; if (__probable(rate == 0)) { return end; } /* init tbr if not so */ if (__improbable(r->ckr_tbr_token == CKR_TBR_TOKEN_INVALID)) { r->ckr_tbr_token = rate; r->ckr_tbr_depth = rate; r->ckr_tbr_last = mach_absolute_time(); } else { now = mach_absolute_time(); elapsed = now - r->ckr_tbr_last; absolutetime_to_microtime(elapsed, &sec, &usec); r->ckr_tbr_token += ((sec * USEC_PER_SEC + usec) * rate / USEC_PER_SEC); if (__improbable(r->ckr_tbr_token > r->ckr_tbr_depth)) { r->ckr_tbr_token = r->ckr_tbr_depth; } r->ckr_tbr_last = now; } *rate_limited = FALSE; for (i = begin; i != end; i = SLOT_NEXT(i, r->ckr_lim)) { pkt = KR_KSD(r, i)->sd_pkt; if (__improbable(pkt == NULL)) { continue; } if (__improbable(r->ckr_tbr_token <= 0)) { end = i; *rate_limited = TRUE; break; } r->ckr_tbr_token -= pkt->pkt_length * 8; } SK_DF(SK_VERB_FSW | SK_VERB_RX, "ckr %p %s rate limited at %d", r, r->ckr_name, i); return end; } SK_NO_INLINE_ATTRIBUTE static struct __kern_packet * consume_pkts(struct __kern_channel_ring *ring, slot_idx_t end) { struct __kern_packet *pkt_chain = NULL, **tailp = &pkt_chain; slot_idx_t idx = ring->ckr_rhead; while (idx != end) { struct __kern_slot_desc *ksd = KR_KSD(ring, idx); struct __kern_packet *pkt = ksd->sd_pkt; ASSERT(pkt->pkt_nextpkt == NULL); KR_SLOT_DETACH_METADATA(ring, ksd); *tailp = pkt; tailp = &pkt->pkt_nextpkt; idx = SLOT_NEXT(idx, ring->ckr_lim); } ring->ckr_rhead = end; ring->ckr_rtail = ring->ckr_ktail; return pkt_chain; } int netif_rx_notify_default(struct __kern_channel_ring *ring, struct proc *p, uint32_t flags) { struct nexus_adapter *hwna; struct nexus_netif_adapter *nifna; struct nx_netif *nif; struct __kern_packet *pkt_chain; struct nexus_pkt_stats stats; sk_protect_t protect; slot_idx_t ktail; int err = 0; KDBG((SK_KTRACE_NETIF_RX_NOTIFY_DEFAULT | DBG_FUNC_START), SK_KVA(ring)); ASSERT(ring->ckr_tx == NR_RX); ASSERT(!NA_KERNEL_ONLY(KRNA(ring)) || KR_KERNEL_ONLY(ring)); err = kr_enter(ring, ((flags & NA_NOTEF_CAN_SLEEP) != 0)); if (err != 0) { /* not a serious error, so no need to be chatty here */ SK_DF(SK_VERB_FSW, "hwna \"%s\" (0x%llx) kr \"%s\" (0x%llx) krflags 0x%b " "(%d)", KRNA(ring)->na_name, SK_KVA(KRNA(ring)), ring->ckr_name, SK_KVA(ring), ring->ckr_flags, CKRF_BITS, err); goto out; } if (__improbable(KR_DROP(ring))) { kr_exit(ring); err = ENODEV; goto out; } hwna = KRNA(ring); nifna = NIFNA(hwna); nif = nifna->nifna_netif; if (__improbable(hwna->na_ifp == NULL)) { kr_exit(ring); err = ENODEV; goto out; } protect = sk_sync_protect(); err = ring->ckr_na_sync(ring, p, 0); if (err != 0 && err != EAGAIN) { goto put_out; } /* read the tail pointer once */ ktail = ring->ckr_ktail; if (__improbable(ring->ckr_khead == ktail)) { SK_DF(SK_VERB_FSW | SK_VERB_NOTIFY | SK_VERB_RX, "how strange, interrupt with no packets on hwna " "\"%s\" (0x%llx)", KRNA(ring)->na_name, SK_KVA(KRNA(ring))); goto put_out; } ktail = netif_rate_limit(ring, nif->nif_input_rate, ring->ckr_rhead, ktail, &ring->ckr_rate_limited); pkt_chain = consume_pkts(ring, ktail); if (pkt_chain != NULL) { netif_receive(nifna, pkt_chain, &stats); if (ring->ckr_netif_mit_stats != NULL && stats.nps_pkts != 0 && stats.nps_bytes != 0) { ring->ckr_netif_mit_stats(ring, stats.nps_pkts, stats.nps_bytes); } } put_out: sk_sync_unprotect(protect); kr_exit(ring); out: KDBG((SK_KTRACE_NETIF_RX_NOTIFY_DEFAULT | DBG_FUNC_END), SK_KVA(ring), err); return err; } int netif_rx_notify_fast(struct __kern_channel_ring *ring, struct proc *p, uint32_t flags) { #pragma unused(p, flags) sk_protect_t protect; struct nexus_adapter *hwna; struct nexus_pkt_stats stats = {}; uint32_t i, count; int err = 0; KDBG((SK_KTRACE_NETIF_RX_NOTIFY_FAST | DBG_FUNC_START), SK_KVA(ring)); /* XXX * sk_sync_protect() is not needed for this case because * we are not using the dev ring. Unfortunately lots of * macros used by fsw still require this. */ protect = sk_sync_protect(); hwna = KRNA(ring); count = na_get_nslots(hwna, NR_RX); err = nx_rx_sync_packets(ring, ring->ckr_scratch, &count); if (__improbable(err != 0)) { SK_ERR("nx_rx_sync_packets failed: %d", err); DTRACE_SKYWALK2(rx__sync__packets__failed, struct __kern_channel_ring *, ring, int, err); goto out; } DTRACE_SKYWALK1(chain__count, uint32_t, count); for (i = 0; i < count; i++) { struct __kern_packet *pkt_chain; pkt_chain = SK_PTR_ADDR_KPKT(ring->ckr_scratch[i]); ASSERT(pkt_chain != NULL); netif_receive(NIFNA(KRNA(ring)), pkt_chain, &stats); if (ring->ckr_netif_mit_stats != NULL && stats.nps_pkts != 0 && stats.nps_bytes != 0) { ring->ckr_netif_mit_stats(ring, stats.nps_pkts, stats.nps_bytes); } } out: sk_sync_unprotect(protect); KDBG((SK_KTRACE_NETIF_RX_NOTIFY_FAST | DBG_FUNC_END), SK_KVA(ring), err); return err; } /* * Configure the NA to operate in a particular mode. */ static channel_ring_notify_t netif_hwna_get_notify(struct __kern_channel_ring *ring, netif_mode_t mode) { channel_ring_notify_t notify = NULL; boolean_t has_sync_pkts = (sk_rx_sync_packets != 0 && nx_has_rx_sync_packets(ring)); if (mode == NETIF_MODE_FSW) { notify = (has_sync_pkts ? netif_rx_notify_fast : netif_rx_notify_default); } else if (mode == NETIF_MODE_LLW) { notify = (has_sync_pkts ? netif_llw_rx_notify_fast : netif_llw_rx_notify_default); } return notify; } static uint32_t netif_mode_to_flag(netif_mode_t mode) { uint32_t flag = 0; if (mode == NETIF_MODE_FSW) { flag = NAF_MODE_FSW; } else if (mode == NETIF_MODE_LLW) { flag = NAF_MODE_LLW; } return flag; } static void netif_hwna_config_mode(struct nexus_adapter *hwna, netif_mode_t mode, void (*rx)(struct nexus_adapter *, struct __kern_packet *, struct nexus_pkt_stats *), boolean_t set) { uint32_t i; uint32_t flag; ASSERT(hwna->na_type == NA_NETIF_DEV || hwna->na_type == NA_NETIF_COMPAT_DEV); for (i = 0; i < na_get_nrings(hwna, NR_RX); i++) { struct __kern_channel_ring *kr = &NAKR(hwna, NR_RX)[i]; channel_ring_notify_t notify = netif_hwna_get_notify(kr, mode); if (set) { kr->ckr_save_notify = kr->ckr_netif_notify; kr->ckr_netif_notify = notify; } else { kr->ckr_netif_notify = kr->ckr_save_notify; kr->ckr_save_notify = NULL; } } if (set) { hwna->na_rx = rx; flag = netif_mode_to_flag(mode); atomic_bitset_32(&hwna->na_flags, flag); } else { hwna->na_rx = NULL; atomic_bitclear_32(&hwna->na_flags, (NAF_MODE_FSW | NAF_MODE_LLW)); } } void netif_hwna_set_mode(struct nexus_adapter *hwna, netif_mode_t mode, void (*rx)(struct nexus_adapter *, struct __kern_packet *, struct nexus_pkt_stats *)) { return netif_hwna_config_mode(hwna, mode, rx, TRUE); } void netif_hwna_clear_mode(struct nexus_adapter *hwna) { return netif_hwna_config_mode(hwna, NETIF_MODE_NONE, NULL, FALSE); } static void netif_inject_rx(struct nexus_adapter *na, struct __kern_packet *pkt_chain) { struct nexus_netif_adapter *nifna = NIFNA(na); struct nx_netif *nif = nifna->nifna_netif; struct netif_stats *nifs = &nif->nif_stats; struct __kern_channel_ring *r; struct nexus_pkt_stats stats; sk_protect_t protect; boolean_t ring_drop = FALSE; int err, dropcnt; if (!NA_OWNED_BY_FSW(na)) { DTRACE_SKYWALK1(fsw__disabled, struct nexus_adapter *, na); goto fail; } ASSERT(na->na_rx != NULL); /* * XXX * This function is called when a filter injects a packet back to the * regular RX path. We can assume the ring is 0 for now because RSS * is not supported. This needs to be revisited when we add support for * RSS. */ r = &na->na_rx_rings[0]; ASSERT(r->ckr_tx == NR_RX); err = kr_enter(r, TRUE); VERIFY(err == 0); if (__improbable(KR_DROP(r))) { kr_exit(r); DTRACE_SKYWALK2(ring__drop, struct nexus_adapter *, na, struct __kern_channel_ring *, r); ring_drop = TRUE; goto fail; } protect = sk_sync_protect(); na->na_rx(na, pkt_chain, &stats); if (r->ckr_netif_mit_stats != NULL && stats.nps_pkts != 0 && stats.nps_bytes != 0) { r->ckr_netif_mit_stats(r, stats.nps_pkts, stats.nps_bytes); } sk_sync_unprotect(protect); kr_exit(r); return; fail: dropcnt = 0; nx_netif_free_packet_chain(pkt_chain, &dropcnt); if (ring_drop) { STATS_ADD(nifs, NETIF_STATS_DROP_KRDROP_MODE, dropcnt); } STATS_ADD(nifs, NETIF_STATS_DROP, dropcnt); } /* * This is called when an inbound packet has traversed all filters. */ errno_t nx_netif_filter_rx_cb(struct nexus_netif_adapter *nifna, struct __kern_packet *fpkt_chain, uint32_t flags) { #pragma unused (flags) struct nx_netif *nif = nifna->nifna_netif; struct netif_stats *nifs = &nif->nif_stats; struct nexus_adapter *na = &nifna->nifna_up; struct __kern_packet *pkt_chain; int err; pkt_chain = nx_netif_filter_pkt_to_pkt_chain(nifna, fpkt_chain, NETIF_CONVERT_RX); if (pkt_chain == NULL) { return ENOMEM; } if (nif->nif_flow_cnt > 0) { struct __kern_packet *remain = NULL; err = nx_netif_demux(nifna, pkt_chain, &remain, NETIF_FLOW_INJECT); if (remain == NULL) { return err; } pkt_chain = remain; } if (na->na_rx != NULL) { netif_inject_rx(na, pkt_chain); } else { int dropcnt = 0; nx_netif_free_packet_chain(pkt_chain, &dropcnt); STATS_ADD(nifs, NETIF_STATS_FILTER_DROP_NO_RX_CB, dropcnt); STATS_ADD(nifs, NETIF_STATS_DROP, dropcnt); } return 0; } /* * This is called when an outbound packet has traversed all filters. */ errno_t nx_netif_filter_tx_cb(struct nexus_netif_adapter *nifna, struct __kern_packet *fpkt_chain, uint32_t flags) { #pragma unused (flags) struct nx_netif *nif = nifna->nifna_netif; struct nexus_adapter *na = &nifna->nifna_up; int err; if (NETIF_IS_COMPAT(nif)) { struct mbuf *m_chain; mbuf_svc_class_t sc; m_chain = nx_netif_filter_pkt_to_mbuf_chain(nifna, fpkt_chain, NETIF_CONVERT_TX); if (m_chain == NULL) { return ENOMEM; } /* * All packets in the chain have the same service class. * If the sc is missing or invalid, a valid value will be * returned. */ sc = mbuf_get_service_class(m_chain); err = nx_netif_filter_tx_processed_mbuf_enqueue(nifna, sc, m_chain); } else { struct __kern_packet *pkt_chain; kern_packet_svc_class_t sc; pkt_chain = nx_netif_filter_pkt_to_pkt_chain(nifna, fpkt_chain, NETIF_CONVERT_TX); if (pkt_chain == NULL) { return ENOMEM; } /* * All packets in the chain have the same service class. * If the sc is missing or invalid, a valid value will be * returned. */ sc = kern_packet_get_service_class(SK_PKT2PH(pkt_chain)); err = nx_netif_filter_tx_processed_pkt_enqueue(nifna, sc, pkt_chain); } /* Tell driver to resume dequeuing */ ifnet_start(na->na_ifp); return err; } void nx_netif_vp_region_params_adjust(struct nexus_adapter *na, struct skmem_region_params *srp) { #pragma unused(na, srp) return; } /* returns true, if starter thread is utilized */ static bool netif_use_starter_thread(struct ifnet *ifp, uint32_t flags) { #if (DEVELOPMENT || DEBUG) if (__improbable(nx_netif_force_ifnet_start != 0)) { ifnet_start(ifp); return true; } #endif /* !DEVELOPMENT && !DEBUG */ /* * use starter thread in following conditions: * - interface is not skywalk native * - interface attached to virtual driver (ipsec, utun) * - TBR is enabled * - delayed start mechanism is in use * - remaining stack space on the thread is not enough for driver * - caller is in rx workloop context * - caller is from the flowswitch path doing ARP resolving * - caller requires the use of starter thread (stack usage) */ if (!SKYWALK_NATIVE(ifp) || NA(ifp) == NULL || !NA_IS_ACTIVE(&NA(ifp)->nifna_up) || ((NA(ifp)->nifna_up.na_flags & NAF_VIRTUAL_DEVICE) != 0) || IFCQ_TBR_IS_ENABLED(ifp->if_snd) || (ifp->if_eflags & IFEF_ENQUEUE_MULTI) || sk_is_rx_notify_protected() || sk_is_async_transmit_protected() || (sk_is_sync_protected() && (flags & NETIF_XMIT_FLAG_HOST) != 0)) { DTRACE_SKYWALK2(use__starter__thread, struct ifnet *, ifp, uint32_t, flags); ifnet_start(ifp); return true; } lck_mtx_lock_spin(&ifp->if_start_lock); /* interface is flow controlled */ if (__improbable(ifp->if_start_flags & IFSF_FLOW_CONTROLLED)) { lck_mtx_unlock(&ifp->if_start_lock); return true; } /* if starter thread is active, utilize it */ if (ifp->if_start_active) { ifp->if_start_req++; lck_mtx_unlock(&ifp->if_start_lock); return true; } lck_mtx_unlock(&ifp->if_start_lock); /* Check remaining stack space */ if ((OSKernelStackRemaining() < NX_NETIF_MIN_DRIVER_STACK_SIZE)) { ifnet_start(ifp); return true; } return false; } void netif_transmit(struct ifnet *ifp, uint32_t flags) { if (netif_use_starter_thread(ifp, flags)) { return; } /* * If no longer attached, don't issue doorbell as ifp * is being destroyed; else hold an IO refcnt to * prevent the interface from being detached. */ if (!ifnet_datamov_begin(ifp)) { return; } nx_netif_doorbell_internal(ifp, flags); /* * Release the IO refcnt taken above. */ ifnet_datamov_end(ifp); } static struct ifclassq * netif_get_default_ifcq(struct nexus_adapter *hwna) { struct nx_netif *nif; struct ifclassq *ifcq; nif = NX_NETIF_PRIVATE(hwna->na_nx); if (NETIF_LLINK_ENABLED(nif)) { struct netif_qset *qset; /* * Use the default ifcq for now. * In the future this could be chosen by the caller. */ qset = nx_netif_get_default_qset_noref(nif); ASSERT(qset != NULL); ifcq = qset->nqs_ifcq; } else { ifcq = nif->nif_ifp->if_snd; } return ifcq; } static errno_t netif_deq_packets(struct nexus_adapter *hwna, struct ifclassq *ifcq, uint32_t pkt_limit, uint32_t byte_limit, struct __kern_packet **head, boolean_t *pkts_pending, kern_packet_svc_class_t sc) { classq_pkt_t pkt_head = CLASSQ_PKT_INITIALIZER(pkt_head); struct ifnet *ifp = hwna->na_ifp; errno_t rc; ASSERT(ifp != NULL); ASSERT(ifp->if_output_sched_model < IFNET_SCHED_MODEL_MAX); ASSERT((pkt_limit != 0) && (byte_limit != 0)); if (ifcq == NULL) { ifcq = netif_get_default_ifcq(hwna); } if (ifp->if_output_sched_model == IFNET_SCHED_MODEL_DRIVER_MANAGED) { rc = ifclassq_dequeue_sc(ifcq, (mbuf_svc_class_t)sc, pkt_limit, byte_limit, &pkt_head, NULL, NULL, NULL); } else { rc = ifclassq_dequeue(ifcq, pkt_limit, byte_limit, &pkt_head, NULL, NULL, NULL); } ASSERT((rc == 0) || (rc == EAGAIN)); ASSERT((pkt_head.cp_ptype == QP_PACKET) || (pkt_head.cp_kpkt == NULL)); if (IFCQ_LEN(ifcq) != 0) { *pkts_pending = TRUE; } else { *pkts_pending = FALSE; } *head = pkt_head.cp_kpkt; return rc; } #if SK_LOG /* Hoisted out of line to reduce kernel stack footprint */ SK_LOG_ATTRIBUTE static void netif_no_ring_space_log(const struct nexus_adapter *na, const kern_channel_ring_t ring) { SK_DF(SK_VERB_SYNC | SK_VERB_TX, "no ring space: na \"%s\" [%u] " "\"%s\"(kh %u kt %u kl %u | rh %u rt %u)" "\"%s\"(kh %u kt %u kl %u | rh %u rt %u)", na->na_name, ring->ckr_ring_id, ring->ckr_name, ring->ckr_khead, ring->ckr_ktail, ring->ckr_klease, ring->ckr_rhead, ring->ckr_rtail); } #endif /* SK_LOG */ /* * netif refill function for rings */ errno_t netif_ring_tx_refill(const kern_channel_ring_t ring, uint32_t pkt_limit, uint32_t byte_limit, boolean_t tx_doorbell_ctxt, boolean_t *pkts_pending, boolean_t canblock) { struct nexus_adapter *hwna; struct ifnet *ifp; struct __kern_packet *head = NULL; sk_protect_t protect; errno_t rc = 0; errno_t sync_err = 0; uint32_t npkts = 0, consumed = 0; uint32_t flags; slot_idx_t idx, ktail; int ring_space = 0; KDBG((SK_KTRACE_NETIF_RING_TX_REFILL | DBG_FUNC_START), SK_KVA(ring)); VERIFY(ring != NULL); hwna = KRNA(ring); ifp = hwna->na_ifp; ASSERT(hwna->na_type == NA_NETIF_DEV); ASSERT(ring->ckr_tx == NR_TX); *pkts_pending = FALSE; if (__improbable(pkt_limit == 0 || byte_limit == 0)) { SK_ERR("invalid limits plim %d, blim %d", pkt_limit, byte_limit); rc = EINVAL; goto out; } if (__improbable(!IF_FULLY_ATTACHED(ifp))) { SK_ERR("hwna 0x%llx ifp %s (0x%llx), interface not attached", SK_KVA(hwna), if_name(ifp), SK_KVA(ifp)); rc = ENXIO; goto out; } if (__improbable((ifp->if_start_flags & IFSF_FLOW_CONTROLLED) != 0)) { SK_DF(SK_VERB_SYNC | SK_VERB_TX, "hwna 0x%llx ifp %s (0x%llx), " "flow control ON", SK_KVA(hwna), if_name(ifp), SK_KVA(ifp)); rc = ENXIO; goto out; } /* * if the ring is busy, it means another dequeue is in * progress, so ignore this request and return success. */ if (kr_enter(ring, canblock) != 0) { rc = 0; goto out; } /* mark thread with sync-in-progress flag */ protect = sk_sync_protect(); if (__improbable(KR_DROP(ring) || !NA_IS_ACTIVE(ring->ckr_na))) { SK_ERR("hw-kr 0x%llx stopped", SK_KVA(ring)); rc = ENXIO; goto done; } idx = ring->ckr_rhead; ktail = ring->ckr_ktail; /* calculate available space on tx ring */ ring_space = ktail - idx; if (ring_space < 0) { ring_space += ring->ckr_num_slots; } if (ring_space == 0) { struct ifclassq *ifcq; /* no space in ring, driver should retry */ #if SK_LOG if (__improbable((sk_verbose & (SK_VERB_SYNC | SK_VERB_TX)) != 0)) { netif_no_ring_space_log(hwna, ring); } #endif /* SK_LOG */ ifcq = netif_get_default_ifcq(hwna); if (IFCQ_LEN(ifcq) != 0) { *pkts_pending = TRUE; } /* * We ran out of space in ring, most probably * because the driver is slow to drain its TX queue. * We want another doorbell to be generated as soon * as the TX notify completion happens; mark this * through ckr_pending_doorbell counter. Do this * regardless of whether there's any pending packet. */ ring->ckr_pending_doorbell++; rc = EAGAIN; goto sync_ring; } if ((uint32_t)ring_space < pkt_limit) { pkt_limit = ring_space; } if (tx_doorbell_ctxt && ((hwna->na_flags & NAF_VIRTUAL_DEVICE) == 0)) { pkt_limit = MIN(pkt_limit, nx_netif_doorbell_max_dequeue); } rc = netif_deq_packets(hwna, NULL, pkt_limit, byte_limit, &head, pkts_pending, ring->ckr_svc); /* * There's room in ring; if we haven't dequeued everything, * mark ckr_pending_doorbell for the next TX notify to issue * a TX door bell; otherwise, clear it. The next packet that * gets enqueued will trigger a door bell again. */ if (*pkts_pending) { ring->ckr_pending_doorbell++; } else if (ring->ckr_pending_doorbell != 0) { ring->ckr_pending_doorbell = 0; } if (rc != 0) { /* * This is expected sometimes as the IOSkywalkFamily * errs on the side of caution to perform an extra * dequeue when multiple doorbells are pending; * nothing to dequeue, do a sync if there are slots * to reclaim else just return. */ SK_DF(SK_VERB_SYNC | SK_VERB_TX, "nothing to dequeue, err %d", rc); if ((uint32_t)ring_space == ring->ckr_lim) { goto done; } else { goto sync_ring; } } /* move the dequeued packets to tx ring */ while (head != NULL && idx != ktail) { ASSERT(npkts <= pkt_limit); struct __kern_packet *pkt = head; KR_SLOT_ATTACH_METADATA(ring, KR_KSD(ring, idx), (struct __kern_quantum *)pkt); npkts++; if (__improbable(pkt->pkt_trace_id != 0)) { KDBG(SK_KTRACE_PKT_TX_AQM | DBG_FUNC_END, pkt->pkt_trace_id); KDBG(SK_KTRACE_PKT_TX_DRV | DBG_FUNC_START, pkt->pkt_trace_id); } idx = SLOT_NEXT(idx, ring->ckr_lim); head = pkt->pkt_nextpkt; pkt->pkt_nextpkt = NULL; } /* * We checked for ring space earlier so the ring should have enough * space for the entire chain. */ ASSERT(head == NULL); ring->ckr_rhead = idx; sync_ring: flags = NA_SYNCF_NETIF; if (ring->ckr_pending_doorbell != 0) { flags |= (NA_SYNCF_NETIF_DOORBELL | NA_SYNCF_NETIF_ASYNC); } ring->ckr_khead_pre = ring->ckr_khead; sync_err = ring->ckr_na_sync(ring, kernproc, flags); if (sync_err != 0 && sync_err != EAGAIN) { SK_ERR("unexpected sync err %d", sync_err); if (rc == 0) { rc = sync_err; } goto done; } /* * Verify that the driver has detached packets from the consumed slots. */ idx = ring->ckr_khead_pre; consumed = 0; while (idx != ring->ckr_khead) { struct __kern_slot_desc *ksd = KR_KSD(ring, idx); consumed++; VERIFY(!KSD_VALID_METADATA(ksd)); idx = SLOT_NEXT(idx, ring->ckr_lim); } ring->ckr_khead_pre = ring->ckr_khead; done: sk_sync_unprotect(protect); kr_exit(ring); out: KDBG((SK_KTRACE_NETIF_RING_TX_REFILL | DBG_FUNC_END), SK_KVA(ring), rc, 0, npkts); return rc; } void kern_netif_queue_rx_enqueue(kern_netif_queue_t queue, kern_packet_t ph_chain, uint32_t count, uint32_t flags) { #pragma unused (count) struct netif_queue *q = queue; struct netif_llink *llink = q->nq_qset->nqs_llink; struct __kern_packet *pkt_chain = SK_PTR_ADDR_KPKT(ph_chain); bool flush = ((flags & KERN_NETIF_QUEUE_RX_ENQUEUE_FLAG_FLUSH) != 0); struct pktq *pktq = &q->nq_pktq; struct netif_stats *nifs = &llink->nll_nif->nif_stats; struct nexus_pkt_stats stats; sk_protect_t protect; ASSERT((q->nq_flags & NETIF_QUEUE_IS_RX) != 0); if (llink->nll_state == NETIF_LLINK_STATE_DESTROYED) { int drop_cnt = 0; pp_free_packet_chain(pkt_chain, &drop_cnt); STATS_ADD(nifs, NETIF_STATS_LLINK_RX_DROP_BAD_STATE, drop_cnt); return; } KPKTQ_ENQUEUE_LIST(pktq, pkt_chain); if (flush) { pkt_chain = KPKTQ_FIRST(pktq); KPKTQ_INIT(pktq); protect = sk_sync_protect(); netif_receive(NA(llink->nll_nif->nif_ifp), pkt_chain, &stats); sk_sync_unprotect(protect); } } errno_t kern_netif_queue_tx_dequeue(kern_netif_queue_t queue, uint32_t pkt_limit, uint32_t byte_limit, boolean_t *pending, kern_packet_t *ph_chain) { struct netif_queue *q = queue; struct netif_llink *llink = q->nq_qset->nqs_llink; struct netif_stats *nifs = &llink->nll_nif->nif_stats; struct nexus_adapter *hwna; struct __kern_packet *pkt_chain = NULL; errno_t rc; ASSERT((q->nq_flags & NETIF_QUEUE_IS_RX) == 0); if (llink->nll_state == NETIF_LLINK_STATE_DESTROYED) { STATS_INC(nifs, NETIF_STATS_LLINK_AQM_DEQ_BAD_STATE); return ENXIO; } hwna = &NA(llink->nll_nif->nif_ifp)->nifna_up; if (((hwna->na_flags & NAF_VIRTUAL_DEVICE) == 0) && sk_is_tx_notify_protected()) { pkt_limit = MIN(pkt_limit, nx_netif_doorbell_max_dequeue); } rc = netif_deq_packets(hwna, q->nq_qset->nqs_ifcq, pkt_limit, byte_limit, &pkt_chain, pending, q->nq_svc); if (pkt_chain != NULL) { *ph_chain = SK_PKT2PH(pkt_chain); } return rc; } errno_t kern_nexus_netif_llink_add(struct kern_nexus *nx, struct kern_nexus_netif_llink_init *llink_init) { errno_t err; struct nx_netif *nif; struct netif_llink *llink; struct netif_stats *nifs; VERIFY(nx != NULL); VERIFY(llink_init != NULL); VERIFY((nx->nx_flags & NXF_ATTACHED) != 0); nif = NX_NETIF_PRIVATE(nx); nifs = &nif->nif_stats; err = nx_netif_validate_llink_config(llink_init, false); if (err != 0) { SK_ERR("Invalid llink init params"); STATS_INC(nifs, NETIF_STATS_LLINK_ADD_BAD_PARAMS); return err; } err = nx_netif_llink_add(nif, llink_init, &llink); return err; } errno_t kern_nexus_netif_llink_remove(struct kern_nexus *nx, kern_nexus_netif_llink_id_t llink_id) { struct nx_netif *nif; VERIFY(nx != NULL); VERIFY((nx->nx_flags & NXF_ATTACHED) != 0); nif = NX_NETIF_PRIVATE(nx); return nx_netif_llink_remove(nif, llink_id); } errno_t kern_netif_queue_get_service_class(kern_netif_queue_t queue, kern_packet_svc_class_t *svc) { *svc = queue->nq_svc; return 0; } |