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
use std::{
cmp,
collections::VecDeque,
convert::TryFrom,
fmt, io, mem,
net::{IpAddr, SocketAddr},
sync::Arc,
time::{Duration, Instant},
};
use bytes::{Bytes, BytesMut};
use frame::StreamMetaVec;
use rand::{rngs::StdRng, Rng, SeedableRng};
use thiserror::Error;
use tracing::{debug, error, trace, trace_span, warn};
use crate::{
cid_generator::ConnectionIdGenerator,
cid_queue::CidQueue,
coding::BufMutExt,
config::{ServerConfig, TransportConfig},
crypto::{self, KeyPair, Keys, PacketKey},
frame,
frame::{Close, Datagram, FrameStruct},
packet::{
FixedLengthConnectionIdParser, Header, InitialHeader, InitialPacket, LongType, Packet,
PacketNumber, PartialDecode, SpaceId,
},
range_set::ArrayRangeSet,
shared::{
ConnectionEvent, ConnectionEventInner, ConnectionId, DatagramConnectionEvent, EcnCodepoint,
EndpointEvent, EndpointEventInner,
},
token::ResetToken,
transport_parameters::TransportParameters,
Dir, EndpointConfig, Frame, Side, StreamId, Transmit, TransportError, TransportErrorCode,
VarInt, MAX_STREAM_COUNT, MIN_INITIAL_SIZE, TIMER_GRANULARITY,
};
mod ack_frequency;
use ack_frequency::AckFrequencyState;
mod assembler;
pub use assembler::Chunk;
mod cid_state;
use cid_state::CidState;
mod datagrams;
use datagrams::DatagramState;
pub use datagrams::{Datagrams, SendDatagramError};
mod mtud;
mod pacing;
mod packet_builder;
use packet_builder::PacketBuilder;
mod packet_crypto;
use packet_crypto::{PrevCrypto, ZeroRttCrypto};
mod paths;
pub use paths::RttEstimator;
use paths::{PathData, PathResponses};
mod send_buffer;
mod spaces;
#[cfg(fuzzing)]
pub use spaces::Retransmits;
#[cfg(not(fuzzing))]
use spaces::Retransmits;
use spaces::{PacketNumberFilter, PacketSpace, SendableFrames, SentPacket, ThinRetransmits};
mod stats;
pub use stats::{ConnectionStats, FrameStats, PathStats, UdpStats};
mod streams;
#[cfg(fuzzing)]
pub use streams::StreamsState;
#[cfg(not(fuzzing))]
use streams::StreamsState;
pub use streams::{
BytesSource, Chunks, ClosedStream, FinishError, ReadError, ReadableError, RecvStream,
SendStream, StreamEvent, Streams, WriteError, Written,
};
mod timer;
use crate::congestion::Controller;
use timer::{Timer, TimerTable};
/// Protocol state and logic for a single QUIC connection
///
/// Objects of this type receive [`ConnectionEvent`]s and emit [`EndpointEvent`]s and application
/// [`Event`]s to make progress. To handle timeouts, a `Connection` returns timer updates and
/// expects timeouts through various methods. A number of simple getter methods are exposed
/// to allow callers to inspect some of the connection state.
///
/// `Connection` has roughly 4 types of methods:
///
/// - A. Simple getters, taking `&self`
/// - B. Handlers for incoming events from the network or system, named `handle_*`.
/// - C. State machine mutators, for incoming commands from the application. For convenience we
/// refer to this as "performing I/O" below, however as per the design of this library none of the
/// functions actually perform system-level I/O. For example, [`read`](RecvStream::read) and
/// [`write`](SendStream::write), but also things like [`reset`](SendStream::reset).
/// - D. Polling functions for outgoing events or actions for the caller to
/// take, named `poll_*`.
///
/// The simplest way to use this API correctly is to call (B) and (C) whenever
/// appropriate, then after each of those calls, as soon as feasible call all
/// polling methods (D) and deal with their outputs appropriately, e.g. by
/// passing it to the application or by making a system-level I/O call. You
/// should call the polling functions in this order:
///
/// 1. [`poll_transmit`](Self::poll_transmit)
/// 2. [`poll_timeout`](Self::poll_timeout)
/// 3. [`poll_endpoint_events`](Self::poll_endpoint_events)
/// 4. [`poll`](Self::poll)
///
/// Currently the only actual dependency is from (2) to (1), however additional
/// dependencies may be added in future, so the above order is recommended.
///
/// (A) may be called whenever desired.
///
/// Care should be made to ensure that the input events represent monotonically
/// increasing time. Specifically, calling [`handle_timeout`](Self::handle_timeout)
/// with events of the same [`Instant`] may be interleaved in any order with a
/// call to [`handle_event`](Self::handle_event) at that same instant; however
/// events or timeouts with different instants must not be interleaved.
pub struct Connection {
endpoint_config: Arc<EndpointConfig>,
server_config: Option<Arc<ServerConfig>>,
config: Arc<TransportConfig>,
rng: StdRng,
crypto: Box<dyn crypto::Session>,
/// The CID we initially chose, for use during the handshake
handshake_cid: ConnectionId,
/// The CID the peer initially chose, for use during the handshake
rem_handshake_cid: ConnectionId,
/// The "real" local IP address which was was used to receive the initial packet.
/// This is only populated for the server case, and if known
local_ip: Option<IpAddr>,
path: PathData,
/// Whether MTU detection is supported in this environment
allow_mtud: bool,
prev_path: Option<(ConnectionId, PathData)>,
state: State,
side: Side,
/// Whether or not 0-RTT was enabled during the handshake. Does not imply acceptance.
zero_rtt_enabled: bool,
/// Set if 0-RTT is supported, then cleared when no longer needed.
zero_rtt_crypto: Option<ZeroRttCrypto>,
key_phase: bool,
/// How many packets are in the current key phase. Used only for `Data` space.
key_phase_size: u64,
/// Transport parameters set by the peer
peer_params: TransportParameters,
/// Source ConnectionId of the first packet received from the peer
orig_rem_cid: ConnectionId,
/// Destination ConnectionId sent by the client on the first Initial
initial_dst_cid: ConnectionId,
/// The value that the server included in the Source Connection ID field of a Retry packet, if
/// one was received
retry_src_cid: Option<ConnectionId>,
/// Total number of outgoing packets that have been deemed lost
lost_packets: u64,
events: VecDeque<Event>,
endpoint_events: VecDeque<EndpointEventInner>,
/// Whether the spin bit is in use for this connection
spin_enabled: bool,
/// Outgoing spin bit state
spin: bool,
/// Packet number spaces: initial, handshake, 1-RTT
spaces: [PacketSpace; 3],
/// Highest usable packet number space
highest_space: SpaceId,
/// 1-RTT keys used prior to a key update
prev_crypto: Option<PrevCrypto>,
/// 1-RTT keys to be used for the next key update
///
/// These are generated in advance to prevent timing attacks and/or DoS by third-party attackers
/// spoofing key updates.
next_crypto: Option<KeyPair<Box<dyn PacketKey>>>,
accepted_0rtt: bool,
/// Whether the idle timer should be reset the next time an ack-eliciting packet is transmitted.
permit_idle_reset: bool,
/// Negotiated idle timeout
idle_timeout: Option<VarInt>,
timers: TimerTable,
/// Number of packets received which could not be authenticated
authentication_failures: u64,
/// Why the connection was lost, if it has been
error: Option<ConnectionError>,
/// Sent in every outgoing Initial packet. Always empty for servers and after Initial keys are
/// discarded.
retry_token: Bytes,
/// Identifies Data-space packet numbers to skip. Not used in earlier spaces.
packet_number_filter: PacketNumberFilter,
//
// Queued non-retransmittable 1-RTT data
//
/// Responses to PATH_CHALLENGE frames
path_responses: PathResponses,
close: bool,
//
// ACK frequency
//
ack_frequency: AckFrequencyState,
//
// Loss Detection
//
/// The number of times a PTO has been sent without receiving an ack.
pto_count: u32,
//
// Congestion Control
//
/// Whether the most recently received packet had an ECN codepoint set
receiving_ecn: bool,
/// Number of packets authenticated
total_authed_packets: u64,
/// Whether the last `poll_transmit` call yielded no data because there was
/// no outgoing application data.
app_limited: bool,
streams: StreamsState,
/// Surplus remote CIDs for future use on new paths
rem_cids: CidQueue,
// Attributes of CIDs generated by local peer
local_cid_state: CidState,
/// State of the unreliable datagram extension
datagrams: DatagramState,
/// Connection level statistics
stats: ConnectionStats,
/// QUIC version used for the connection.
version: u32,
}
impl Connection {
pub(crate) fn new(
endpoint_config: Arc<EndpointConfig>,
server_config: Option<Arc<ServerConfig>>,
config: Arc<TransportConfig>,
init_cid: ConnectionId,
loc_cid: ConnectionId,
rem_cid: ConnectionId,
pref_addr_cid: Option<ConnectionId>,
remote: SocketAddr,
local_ip: Option<IpAddr>,
crypto: Box<dyn crypto::Session>,
cid_gen: &dyn ConnectionIdGenerator,
now: Instant,
version: u32,
allow_mtud: bool,
rng_seed: [u8; 32],
path_validated: bool,
) -> Self {
let side = if server_config.is_some() {
Side::Server
} else {
Side::Client
};
let initial_space = PacketSpace {
crypto: Some(crypto.initial_keys(&init_cid, side)),
..PacketSpace::new(now)
};
let state = State::Handshake(state::Handshake {
rem_cid_set: side.is_server(),
expected_token: Bytes::new(),
client_hello: None,
});
let mut rng = StdRng::from_seed(rng_seed);
let mut this = Self {
endpoint_config,
server_config,
crypto,
handshake_cid: loc_cid,
rem_handshake_cid: rem_cid,
local_cid_state: CidState::new(
cid_gen.cid_len(),
cid_gen.cid_lifetime(),
now,
if pref_addr_cid.is_some() { 2 } else { 1 },
),
path: PathData::new(remote, allow_mtud, None, now, path_validated, &config),
allow_mtud,
local_ip,
prev_path: None,
side,
state,
zero_rtt_enabled: false,
zero_rtt_crypto: None,
key_phase: false,
// A small initial key phase size ensures peers that don't handle key updates correctly
// fail sooner rather than later. It's okay for both peers to do this, as the first one
// to perform an update will reset the other's key phase size in `update_keys`, and a
// simultaneous key update by both is just like a regular key update with a really fast
// response. Inspired by quic-go's similar behavior of performing the first key update
// at the 100th short-header packet.
key_phase_size: rng.gen_range(10..1000),
peer_params: TransportParameters::default(),
orig_rem_cid: rem_cid,
initial_dst_cid: init_cid,
retry_src_cid: None,
lost_packets: 0,
events: VecDeque::new(),
endpoint_events: VecDeque::new(),
spin_enabled: config.allow_spin && rng.gen_ratio(7, 8),
spin: false,
spaces: [initial_space, PacketSpace::new(now), PacketSpace::new(now)],
highest_space: SpaceId::Initial,
prev_crypto: None,
next_crypto: None,
accepted_0rtt: false,
permit_idle_reset: true,
idle_timeout: config.max_idle_timeout,
timers: TimerTable::default(),
authentication_failures: 0,
error: None,
retry_token: Bytes::new(),
#[cfg(test)]
packet_number_filter: match config.deterministic_packet_numbers {
false => PacketNumberFilter::new(&mut rng),
true => PacketNumberFilter::disabled(),
},
#[cfg(not(test))]
packet_number_filter: PacketNumberFilter::new(&mut rng),
path_responses: PathResponses::default(),
close: false,
ack_frequency: AckFrequencyState::new(get_max_ack_delay(
&TransportParameters::default(),
)),
pto_count: 0,
app_limited: false,
receiving_ecn: false,
total_authed_packets: 0,
streams: StreamsState::new(
side,
config.max_concurrent_uni_streams,
config.max_concurrent_bidi_streams,
config.send_window,
config.receive_window,
config.stream_receive_window,
),
datagrams: DatagramState::default(),
config,
rem_cids: CidQueue::new(rem_cid),
rng,
stats: ConnectionStats::default(),
version,
};
if side.is_client() {
// Kick off the connection
this.write_crypto();
this.init_0rtt();
}
this
}
/// Returns the next time at which `handle_timeout` should be called
///
/// The value returned may change after:
/// - the application performed some I/O on the connection
/// - a call was made to `handle_event`
/// - a call to `poll_transmit` returned `Some`
/// - a call was made to `handle_timeout`
#[must_use]
pub fn poll_timeout(&mut self) -> Option<Instant> {
self.timers.next_timeout()
}
/// Returns application-facing events
///
/// Connections should be polled for events after:
/// - a call was made to `handle_event`
/// - a call was made to `handle_timeout`
#[must_use]
pub fn poll(&mut self) -> Option<Event> {
if let Some(x) = self.events.pop_front() {
return Some(x);
}
if let Some(event) = self.streams.poll() {
return Some(Event::Stream(event));
}
if let Some(err) = self.error.take() {
return Some(Event::ConnectionLost { reason: err });
}
None
}
/// Return endpoint-facing events
#[must_use]
pub fn poll_endpoint_events(&mut self) -> Option<EndpointEvent> {
self.endpoint_events.pop_front().map(EndpointEvent)
}
/// Provide control over streams
#[must_use]
pub fn streams(&mut self) -> Streams<'_> {
Streams {
state: &mut self.streams,
conn_state: &self.state,
}
}
/// Provide control over streams
#[must_use]
pub fn recv_stream(&mut self, id: StreamId) -> RecvStream<'_> {
assert!(id.dir() == Dir::Bi || id.initiator() != self.side);
RecvStream {
id,
state: &mut self.streams,
pending: &mut self.spaces[SpaceId::Data].pending,
}
}
/// Provide control over streams
#[must_use]
pub fn send_stream(&mut self, id: StreamId) -> SendStream<'_> {
assert!(id.dir() == Dir::Bi || id.initiator() == self.side);
SendStream {
id,
state: &mut self.streams,
pending: &mut self.spaces[SpaceId::Data].pending,
conn_state: &self.state,
}
}
/// Returns packets to transmit
///
/// Connections should be polled for transmit after:
/// - the application performed some I/O on the connection
/// - a call was made to `handle_event`
/// - a call was made to `handle_timeout`
///
/// `max_datagrams` specifies how many datagrams can be returned inside a
/// single Transmit using GSO. This must be at least 1.
#[must_use]
pub fn poll_transmit(
&mut self,
now: Instant,
max_datagrams: usize,
buf: &mut Vec<u8>,
) -> Option<Transmit> {
assert!(max_datagrams != 0);
let max_datagrams = match self.config.enable_segmentation_offload {
false => 1,
true => max_datagrams.min(MAX_TRANSMIT_SEGMENTS),
};
let mut num_datagrams = 0;
// Position in `buf` of the first byte of the current UDP datagram. When coalescing QUIC
// packets, this can be earlier than the start of the current QUIC packet.
let mut datagram_start = 0;
let mut segment_size = usize::from(self.path.current_mtu());
// Send PATH_CHALLENGE for a previous path if necessary
if let Some((prev_cid, ref mut prev_path)) = self.prev_path {
if prev_path.challenge_pending {
prev_path.challenge_pending = false;
let token = prev_path
.challenge
.expect("previous path challenge pending without token");
let destination = prev_path.remote;
debug_assert_eq!(
self.highest_space,
SpaceId::Data,
"PATH_CHALLENGE queued without 1-RTT keys"
);
buf.reserve(MIN_INITIAL_SIZE as usize);
let buf_capacity = buf.capacity();
// Use the previous CID to avoid linking the new path with the previous path. We
// don't bother accounting for possible retirement of that prev_cid because this is
// sent once, immediately after migration, when the CID is known to be valid. Even
// if a post-migration packet caused the CID to be retired, it's fair to pretend
// this is sent first.
let mut builder = PacketBuilder::new(
now,
SpaceId::Data,
prev_cid,
buf,
buf_capacity,
0,
false,
self,
)?;
trace!("validating previous path with PATH_CHALLENGE {:08x}", token);
buf.write(frame::Type::PATH_CHALLENGE);
buf.write(token);
self.stats.frame_tx.path_challenge += 1;
// An endpoint MUST expand datagrams that contain a PATH_CHALLENGE frame
// to at least the smallest allowed maximum datagram size of 1200 bytes,
// unless the anti-amplification limit for the path does not permit
// sending a datagram of this size
builder.pad_to(MIN_INITIAL_SIZE);
builder.finish(self, buf);
self.stats.udp_tx.on_sent(1, buf.len());
return Some(Transmit {
destination,
size: buf.len(),
ecn: None,
segment_size: None,
src_ip: self.local_ip,
});
}
}
// If we need to send a probe, make sure we have something to send.
for space in SpaceId::iter() {
let request_immediate_ack =
space == SpaceId::Data && self.peer_supports_ack_frequency();
self.spaces[space].maybe_queue_probe(request_immediate_ack, &self.streams);
}
// Check whether we need to send a close message
let close = match self.state {
State::Drained => {
self.app_limited = true;
return None;
}
State::Draining | State::Closed(_) => {
// self.close is only reset once the associated packet had been
// encoded successfully
if !self.close {
self.app_limited = true;
return None;
}
true
}
_ => false,
};
// Check whether we need to send an ACK_FREQUENCY frame
if let Some(config) = &self.config.ack_frequency_config {
self.spaces[SpaceId::Data].pending.ack_frequency = self
.ack_frequency
.should_send_ack_frequency(self.path.rtt.get(), config, &self.peer_params)
&& self.highest_space == SpaceId::Data
&& self.peer_supports_ack_frequency();
}
// Reserving capacity can provide more capacity than we asked for. However, we are not
// allowed to write more than `segment_size`. Therefore the maximum capacity is tracked
// separately.
let mut buf_capacity = 0;
let mut coalesce = true;
let mut builder_storage: Option<PacketBuilder> = None;
let mut sent_frames = None;
let mut pad_datagram = false;
let mut congestion_blocked = false;
// Iterate over all spaces and find data to send
let mut space_idx = 0;
let spaces = [SpaceId::Initial, SpaceId::Handshake, SpaceId::Data];
// This loop will potentially spend multiple iterations in the same `SpaceId`,
// so we cannot trivially rewrite it to take advantage of `SpaceId::iter()`.
while space_idx < spaces.len() {
let space_id = spaces[space_idx];
// Number of bytes available for frames if this is a 1-RTT packet. We're guaranteed to
// be able to send an individual frame at least this large in the next 1-RTT
// packet. This could be generalized to support every space, but it's only needed to
// handle large fixed-size frames, which only exist in 1-RTT (application datagrams). We
// don't account for coalesced packets potentially occupying space because frames can
// always spill into the next datagram.
let pn = self.packet_number_filter.peek(&self.spaces[SpaceId::Data]);
let frame_space_1rtt =
segment_size.saturating_sub(self.predict_1rtt_overhead(Some(pn)));
// Is there data or a close message to send in this space?
let can_send = self.space_can_send(space_id, frame_space_1rtt);
if can_send.is_empty() && (!close || self.spaces[space_id].crypto.is_none()) {
space_idx += 1;
continue;
}
let mut ack_eliciting = !self.spaces[space_id].pending.is_empty(&self.streams)
|| self.spaces[space_id].ping_pending
|| self.spaces[space_id].immediate_ack_pending;
if space_id == SpaceId::Data {
ack_eliciting |= self.can_send_1rtt(frame_space_1rtt);
}
// Can we append more data into the current buffer?
// It is not safe to assume that `buf.len()` is the end of the data,
// since the last packet might not have been finished.
let buf_end = if let Some(builder) = &builder_storage {
buf.len().max(builder.min_size) + builder.tag_len
} else {
buf.len()
};
if !coalesce || buf_capacity - buf_end < MIN_PACKET_SPACE {
// We need to send 1 more datagram and extend the buffer for that.
// Is 1 more datagram allowed?
if buf_capacity >= segment_size * max_datagrams {
// No more datagrams allowed
break;
}
// Anti-amplification is only based on `total_sent`, which gets
// updated at the end of this method. Therefore we pass the amount
// of bytes for datagrams that are already created, as well as 1 byte
// for starting another datagram. If there is any anti-amplification
// budget left, we always allow a full MTU to be sent
// (see https://github.com/quinn-rs/quinn/issues/1082)
if self
.path
.anti_amplification_blocked(segment_size as u64 * num_datagrams + 1)
{
trace!("blocked by anti-amplification");
break;
}
// Congestion control and pacing checks
// Tail loss probes must not be blocked by congestion, or a deadlock could arise
if ack_eliciting && self.spaces[space_id].loss_probes == 0 {
// Assume the current packet will get padded to fill the segment
let untracked_bytes = if let Some(builder) = &builder_storage {
buf_capacity - builder.partial_encode.start
} else {
0
} as u64;
debug_assert!(untracked_bytes <= segment_size as u64);
let bytes_to_send = segment_size as u64 + untracked_bytes;
if self.path.in_flight.bytes + bytes_to_send >= self.path.congestion.window() {
space_idx += 1;
congestion_blocked = true;
// We continue instead of breaking here in order to avoid
// blocking loss probes queued for higher spaces.
trace!("blocked by congestion control");
continue;
}
// Check whether the next datagram is blocked by pacing
let smoothed_rtt = self.path.rtt.get();
if let Some(delay) = self.path.pacing.delay(
smoothed_rtt,
bytes_to_send,
self.path.current_mtu(),
self.path.congestion.window(),
now,
) {
self.timers.set(Timer::Pacing, delay);
congestion_blocked = true;
// Loss probes should be subject to pacing, even though
// they are not congestion controlled.
trace!("blocked by pacing");
break;
}
}
// Finish current packet
if let Some(mut builder) = builder_storage.take() {
if pad_datagram {
builder.pad_to(MIN_INITIAL_SIZE);
}
if num_datagrams > 1 {
// If too many padding bytes would be required to continue the GSO batch
// after this packet, end the GSO batch here. Ensures that fixed-size frames
// with heterogeneous sizes (e.g. application datagrams) won't inadvertently
// waste large amounts of bandwidth. The exact threshold is a bit arbitrary
// and might benefit from further tuning, though there's no universally
// optimal value.
const MAX_PADDING: usize = 16;
let packet_len_unpadded = cmp::max(builder.min_size, buf.len())
- datagram_start
+ builder.tag_len;
if packet_len_unpadded + MAX_PADDING < segment_size {
trace!(
"GSO truncated by demand for {} padding bytes",
segment_size - packet_len_unpadded
);
builder_storage = Some(builder);
break;
}
// Pad the current packet to GSO segment size so it can be included in the
// GSO batch.
builder.pad_to(segment_size as u16);
}
builder.finish_and_track(now, self, sent_frames.take(), buf);
if num_datagrams == 1 {
// Set the segment size for this GSO batch to the size of the first UDP
// datagram in the batch. Larger data that cannot be fragmented
// (e.g. application datagrams) will be included in a future batch. When
// sending large enough volumes of data for GSO to be useful, we expect
// packet sizes to usually be consistent, e.g. populated by max-size STREAM
// frames or uniformly sized datagrams.
segment_size = buf.len();
// Clip the unused capacity out of the buffer so future packets don't
// overrun
buf_capacity = buf.len();
// Check whether the data we planned to send will fit in the reduced segment
// size. If not, bail out and leave it for the next GSO batch so we don't
// end up trying to send an empty packet. We can't easily compute the right
// segment size before the original call to `space_can_send`, because at
// that time we haven't determined whether we're going to coalesce with the
// first datagram or potentially pad it to `MIN_INITIAL_SIZE`.
if space_id == SpaceId::Data {
let frame_space_1rtt =
segment_size.saturating_sub(self.predict_1rtt_overhead(Some(pn)));
if self.space_can_send(space_id, frame_space_1rtt).is_empty() {
break;
}
}
}
}
// Allocate space for another datagram
buf_capacity += segment_size;
if buf.capacity() < buf_capacity {
// We reserve the maximum space for sending `max_datagrams` upfront
// to avoid any reallocations if more datagrams have to be appended later on.
// Benchmarks have shown shown a 5-10% throughput improvement
// compared to continuously resizing the datagram buffer.
// While this will lead to over-allocation for small transmits
// (e.g. purely containing ACKs), modern memory allocators
// (e.g. mimalloc and jemalloc) will pool certain allocation sizes
// and therefore this is still rather efficient.
buf.reserve(max_datagrams * segment_size);
}
num_datagrams += 1;
coalesce = true;
pad_datagram = false;
datagram_start = buf.len();
debug_assert_eq!(
datagram_start % segment_size,
0,
"datagrams in a GSO batch must be aligned to the segment size"
);
} else {
// We can append/coalesce the next packet into the current
// datagram.
// Finish current packet without adding extra padding
if let Some(builder) = builder_storage.take() {
builder.finish_and_track(now, self, sent_frames.take(), buf);
}
}
debug_assert!(buf_capacity - buf.len() >= MIN_PACKET_SPACE);
//
// From here on, we've determined that a packet will definitely be sent.
//
if self.spaces[SpaceId::Initial].crypto.is_some()
&& space_id == SpaceId::Handshake
&& self.side.is_client()
{
// A client stops both sending and processing Initial packets when it
// sends its first Handshake packet.
self.discard_space(now, SpaceId::Initial);
}
if let Some(ref mut prev) = self.prev_crypto {
prev.update_unacked = false;
}
debug_assert!(
builder_storage.is_none() && sent_frames.is_none(),
"Previous packet must have been finished"
);
// This should really be `builder.insert()`, but `Option::insert`
// is not stable yet. Since we `debug_assert!(builder.is_none())` it
// doesn't make any functional difference.
let builder = builder_storage.get_or_insert(PacketBuilder::new(
now,
space_id,
self.rem_cids.active(),
buf,
buf_capacity,
datagram_start,
ack_eliciting,
self,
)?);
coalesce = coalesce && !builder.short_header;
// https://tools.ietf.org/html/draft-ietf-quic-transport-34#section-14.1
pad_datagram |=
space_id == SpaceId::Initial && (self.side.is_client() || ack_eliciting);
if close {
trace!("sending CONNECTION_CLOSE");
// Encode ACKs before the ConnectionClose message, to give the receiver
// a better approximate on what data has been processed. This is
// especially important with ack delay, since the peer might not
// have gotten any other ACK for the data earlier on.
if !self.spaces[space_id].pending_acks.ranges().is_empty() {
Self::populate_acks(
now,
self.receiving_ecn,
&mut SentFrames::default(),
&mut self.spaces[space_id],
buf,
&mut self.stats,
);
}
// Since there only 64 ACK frames there will always be enough space
// to encode the ConnectionClose frame too. However we still have the
// check here to prevent crashes if something changes.
debug_assert!(
buf.len() + frame::ConnectionClose::SIZE_BOUND < builder.max_size,
"ACKs should leave space for ConnectionClose"
);
if buf.len() + frame::ConnectionClose::SIZE_BOUND < builder.max_size {
let max_frame_size = builder.max_size - buf.len();
match self.state {
State::Closed(state::Closed { ref reason }) => {
if space_id == SpaceId::Data || reason.is_transport_layer() {
reason.encode(buf, max_frame_size)
} else {
frame::ConnectionClose {
error_code: TransportErrorCode::APPLICATION_ERROR,
frame_type: None,
reason: Bytes::new(),
}
.encode(buf, max_frame_size)
}
}
State::Draining => frame::ConnectionClose {
error_code: TransportErrorCode::NO_ERROR,
frame_type: None,
reason: Bytes::new(),
}
.encode(buf, max_frame_size),
_ => unreachable!(
"tried to make a close packet when the connection wasn't closed"
),
}
}
if space_id == self.highest_space {
// Don't send another close packet
self.close = false;
// `CONNECTION_CLOSE` is the final packet
break;
} else {
// Send a close frame in every possible space for robustness, per RFC9000
// "Immediate Close during the Handshake". Don't bother trying to send anything
// else.
space_idx += 1;
continue;
}
}
// Send an off-path PATH_RESPONSE. Prioritized over on-path data to ensure that path
// validation can occur while the link is saturated.
if space_id == SpaceId::Data && num_datagrams == 1 {
if let Some((token, remote)) = self.path_responses.pop_off_path(&self.path.remote) {
// `unwrap` guaranteed to succeed because `builder_storage` was populated just
// above.
let mut builder = builder_storage.take().unwrap();
trace!("PATH_RESPONSE {:08x} (off-path)", token);
buf.write(frame::Type::PATH_RESPONSE);
buf.write(token);
self.stats.frame_tx.path_response += 1;
builder.pad_to(MIN_INITIAL_SIZE);
builder.finish_and_track(
now,
self,
Some(SentFrames {
non_retransmits: true,
..SentFrames::default()
}),
buf,
);
self.stats.udp_tx.on_sent(1, buf.len());
return Some(Transmit {
destination: remote,
size: buf.len(),
ecn: None,
segment_size: None,
src_ip: self.local_ip,
});
}
}
let sent =
self.populate_packet(now, space_id, buf, builder.max_size, builder.exact_number);
// ACK-only packets should only be sent when explicitly allowed. If we write them due to
// any other reason, there is a bug which leads to one component announcing write
// readiness while not writing any data. This degrades performance. The condition is
// only checked if the full MTU is available and when potentially large fixed-size
// frames aren't queued, so that lack of space in the datagram isn't the reason for just
// writing ACKs.
debug_assert!(
!(sent.is_ack_only(&self.streams)
&& !can_send.acks
&& can_send.other
&& (buf_capacity - builder.datagram_start) == self.path.current_mtu() as usize
&& self.datagrams.outgoing.is_empty()),
"SendableFrames was {can_send:?}, but only ACKs have been written"
);
pad_datagram |= sent.requires_padding;
if sent.largest_acked.is_some() {
self.spaces[space_id].pending_acks.acks_sent();
self.timers.stop(Timer::MaxAckDelay);
}
// Keep information about the packet around until it gets finalized
sent_frames = Some(sent);
// Don't increment space_idx.
// We stay in the current space and check if there is more data to send.
}
// Finish the last packet
if let Some(mut builder) = builder_storage {
if pad_datagram {
builder.pad_to(MIN_INITIAL_SIZE);
}
let last_packet_number = builder.exact_number;
builder.finish_and_track(now, self, sent_frames, buf);
self.path
.congestion
.on_sent(now, buf.len() as u64, last_packet_number);
}
self.app_limited = buf.is_empty() && !congestion_blocked;
// Send MTU probe if necessary
if buf.is_empty() && self.state.is_established() {
let space_id = SpaceId::Data;
let probe_size = match self
.path
.mtud
.poll_transmit(now, self.packet_number_filter.peek(&self.spaces[space_id]))
{
Some(next_probe_size) => next_probe_size,
None => return None,
};
let buf_capacity = probe_size as usize;
buf.reserve(buf_capacity);
let mut builder = PacketBuilder::new(
now,
space_id,
self.rem_cids.active(),
buf,
buf_capacity,
0,
true,
self,
)?;
// We implement MTU probes as ping packets padded up to the probe size
buf.write(frame::Type::PING);
self.stats.frame_tx.ping += 1;
// If supported by the peer, we want no delays to the probe's ACK
if self.peer_supports_ack_frequency() {
buf.write(frame::Type::IMMEDIATE_ACK);
self.stats.frame_tx.immediate_ack += 1;
}
builder.pad_to(probe_size);
let sent_frames = SentFrames {
non_retransmits: true,
..Default::default()
};
builder.finish_and_track(now, self, Some(sent_frames), buf);
self.stats.path.sent_plpmtud_probes += 1;
num_datagrams = 1;
trace!(?probe_size, "writing MTUD probe");
}
if buf.is_empty() {
return None;
}
trace!("sending {} bytes in {} datagrams", buf.len(), num_datagrams);
self.path.total_sent = self.path.total_sent.saturating_add(buf.len() as u64);
self.stats.udp_tx.on_sent(num_datagrams, buf.len());
Some(Transmit {
destination: self.path.remote,
size: buf.len(),
ecn: if self.path.sending_ecn {
Some(EcnCodepoint::Ect0)
} else {
None
},
segment_size: match num_datagrams {
1 => None,
_ => Some(segment_size),
},
src_ip: self.local_ip,
})
}
/// Indicate what types of frames are ready to send for the given space
fn space_can_send(&self, space_id: SpaceId, frame_space_1rtt: usize) -> SendableFrames {
if self.spaces[space_id].crypto.is_none()
&& (space_id != SpaceId::Data
|| self.zero_rtt_crypto.is_none()
|| self.side.is_server())
{
// No keys available for this space
return SendableFrames::empty();
}
let mut can_send = self.spaces[space_id].can_send(&self.streams);
if space_id == SpaceId::Data {
can_send.other |= self.can_send_1rtt(frame_space_1rtt);
}
can_send
}
/// Process `ConnectionEvent`s generated by the associated `Endpoint`
///
/// Will execute protocol logic upon receipt of a connection event, in turn preparing signals
/// (including application `Event`s, `EndpointEvent`s and outgoing datagrams) that should be
/// extracted through the relevant methods.
pub fn handle_event(&mut self, event: ConnectionEvent) {
use self::ConnectionEventInner::*;
match event.0 {
Datagram(DatagramConnectionEvent {
now,
remote,
ecn,
first_decode,
remaining,
}) => {
// If this packet could initiate a migration and we're a client or a server that
// forbids migration, drop the datagram. This could be relaxed to heuristically
// permit NAT-rebinding-like migration.
if remote != self.path.remote
&& self.server_config.as_ref().map_or(true, |x| !x.migration)
{
trace!("discarding packet from unrecognized peer {}", remote);
return;
}
let was_anti_amplification_blocked = self.path.anti_amplification_blocked(1);
self.stats.udp_rx.datagrams += 1;
self.stats.udp_rx.bytes += first_decode.len() as u64;
let data_len = first_decode.len();
self.handle_decode(now, remote, ecn, first_decode);
// The current `path` might have changed inside `handle_decode`,
// since the packet could have triggered a migration. Make sure
// the data received is accounted for the most recent path by accessing
// `path` after `handle_decode`.
self.path.total_recvd = self.path.total_recvd.saturating_add(data_len as u64);
if let Some(data) = remaining {
self.stats.udp_rx.bytes += data.len() as u64;
self.handle_coalesced(now, remote, ecn, data);
}
if was_anti_amplification_blocked {
// A prior attempt to set the loss detection timer may have failed due to
// anti-amplification, so ensure it's set now. Prevents a handshake deadlock if
// the server's first flight is lost.
self.set_loss_detection_timer(now);
}
}
NewIdentifiers(ids, now) => {
self.local_cid_state.new_cids(&ids, now);
ids.into_iter().rev().for_each(|frame| {
self.spaces[SpaceId::Data].pending.new_cids.push(frame);
});
// Update Timer::PushNewCid
if self
.timers
.get(Timer::PushNewCid)
.map_or(true, |x| x <= now)
{
self.reset_cid_retirement();
}
}
}
}
/// Process timer expirations
///
/// Executes protocol logic, potentially preparing signals (including application `Event`s,
/// `EndpointEvent`s and outgoing datagrams) that should be extracted through the relevant
/// methods.
///
/// It is most efficient to call this immediately after the system clock reaches the latest
/// `Instant` that was output by `poll_timeout`; however spurious extra calls will simply
/// no-op and therefore are safe.
pub fn handle_timeout(&mut self, now: Instant) {
for &timer in &Timer::VALUES {
if !self.timers.is_expired(timer, now) {
continue;
}
self.timers.stop(timer);
trace!(timer = ?timer, "timeout");
match timer {
Timer::Close => {
self.state = State::Drained;
self.endpoint_events.push_back(EndpointEventInner::Drained);
}
Timer::Idle => {
self.kill(ConnectionError::TimedOut);
}
Timer::KeepAlive => {
trace!("sending keep-alive");
self.ping();
}
Timer::LossDetection => {
self.on_loss_detection_timeout(now);
}
Timer::KeyDiscard => {
self.zero_rtt_crypto = None;
self.prev_crypto = None;
}
Timer::PathValidation => {
debug!("path validation failed");
if let Some((_, prev)) = self.prev_path.take() {
self.path = prev;
}
self.path.challenge = None;
self.path.challenge_pending = false;
}
Timer::Pacing => trace!("pacing timer expired"),
Timer::PushNewCid => {
// Update `retire_prior_to` field in NEW_CONNECTION_ID frame
let num_new_cid = self.local_cid_state.on_cid_timeout().into();
if !self.state.is_closed() {
trace!(
"push a new cid to peer RETIRE_PRIOR_TO field {}",
self.local_cid_state.retire_prior_to()
);
self.endpoint_events
.push_back(EndpointEventInner::NeedIdentifiers(now, num_new_cid));
}
}
Timer::MaxAckDelay => {
trace!("max ack delay reached");
// This timer is only armed in the Data space
self.spaces[SpaceId::Data]
.pending_acks
.on_max_ack_delay_timeout()
}
}
}
}
/// Close a connection immediately
///
/// This does not ensure delivery of outstanding data. It is the application's responsibility to
/// call this only when all important communications have been completed, e.g. by calling
/// [`SendStream::finish`] on outstanding streams and waiting for the corresponding
/// [`StreamEvent::Finished`] event.
///
/// If [`Streams::send_streams`] returns 0, all outstanding stream data has been
/// delivered. There may still be data from the peer that has not been received.
///
/// [`StreamEvent::Finished`]: crate::StreamEvent::Finished
pub fn close(&mut self, now: Instant, error_code: VarInt, reason: Bytes) {
self.close_inner(
now,
Close::Application(frame::ApplicationClose { error_code, reason }),
)
}
fn close_inner(&mut self, now: Instant, reason: Close) {
let was_closed = self.state.is_closed();
if !was_closed {
self.close_common();
self.set_close_timer(now);
self.close = true;
self.state = State::Closed(state::Closed { reason });
}
}
/// Control datagrams
pub fn datagrams(&mut self) -> Datagrams<'_> {
Datagrams { conn: self }
}
/// Returns connection statistics
pub fn stats(&self) -> ConnectionStats {
let mut stats = self.stats;
stats.path.rtt = self.path.rtt.get();
stats.path.cwnd = self.path.congestion.window();
stats.path.current_mtu = self.path.mtud.current_mtu();
stats
}
/// Ping the remote endpoint
///
/// Causes an ACK-eliciting packet to be transmitted.
pub fn ping(&mut self) {
self.spaces[self.highest_space].ping_pending = true;
}
#[doc(hidden)]
pub fn initiate_key_update(&mut self) {
self.update_keys(None, false);
}
/// Get a session reference
pub fn crypto_session(&self) -> &dyn crypto::Session {
&*self.crypto
}
/// Whether the connection is in the process of being established
///
/// If this returns `false`, the connection may be either established or closed, signaled by the
/// emission of a `Connected` or `ConnectionLost` message respectively.
pub fn is_handshaking(&self) -> bool {
self.state.is_handshake()
}
/// Whether the connection is closed
///
/// Closed connections cannot transport any further data. A connection becomes closed when
/// either peer application intentionally closes it, or when either transport layer detects an
/// error such as a time-out or certificate validation failure.
///
/// A `ConnectionLost` event is emitted with details when the connection becomes closed.
pub fn is_closed(&self) -> bool {
self.state.is_closed()
}
/// Whether there is no longer any need to keep the connection around
///
/// Closed connections become drained after a brief timeout to absorb any remaining in-flight
/// packets from the peer. All drained connections have been closed.
pub fn is_drained(&self) -> bool {
self.state.is_drained()
}
/// For clients, if the peer accepted the 0-RTT data packets
///
/// The value is meaningless until after the handshake completes.
pub fn accepted_0rtt(&self) -> bool {
self.accepted_0rtt
}
/// Whether 0-RTT is/was possible during the handshake
pub fn has_0rtt(&self) -> bool {
self.zero_rtt_enabled
}
/// Whether there are any pending retransmits
pub fn has_pending_retransmits(&self) -> bool {
!self.spaces[SpaceId::Data].pending.is_empty(&self.streams)
}
/// Look up whether we're the client or server of this Connection
pub fn side(&self) -> Side {
self.side
}
/// The latest socket address for this connection's peer
pub fn remote_address(&self) -> SocketAddr {
self.path.remote
}
/// The local IP address which was used when the peer established
/// the connection
///
/// This can be different from the address the endpoint is bound to, in case
/// the endpoint is bound to a wildcard address like `0.0.0.0` or `::`.
///
/// This will return `None` for clients, or when no `local_ip` was passed to
/// [`Endpoint::handle()`](crate::Endpoint::handle) for the datagrams establishing this
/// connection.
pub fn local_ip(&self) -> Option<IpAddr> {
self.local_ip
}
/// Current best estimate of this connection's latency (round-trip-time)
pub fn rtt(&self) -> Duration {
self.path.rtt.get()
}
/// Current state of this connection's congestion controller, for debugging purposes
pub fn congestion_state(&self) -> &dyn Controller {
self.path.congestion.as_ref()
}
/// Modify the number of remotely initiated streams that may be concurrently open
///
/// No streams may be opened by the peer unless fewer than `count` are already open. Large
/// `count`s increase both minimum and worst-case memory consumption.
pub fn set_max_concurrent_streams(&mut self, dir: Dir, count: VarInt) {
self.streams.set_max_concurrent(dir, count);
// If the limit was reduced, then a flow control update previously deemed insignificant may
// now be significant.
let pending = &mut self.spaces[SpaceId::Data].pending;
self.streams.queue_max_stream_id(pending);
}
/// Current number of remotely initiated streams that may be concurrently open
///
/// If the target for this limit is reduced using [`set_max_concurrent_streams`](Self::set_max_concurrent_streams),
/// it will not change immediately, even if fewer streams are open. Instead, it will
/// decrement by one for each time a remotely initiated stream of matching directionality is closed.
pub fn max_concurrent_streams(&self, dir: Dir) -> u64 {
self.streams.max_concurrent(dir)
}
/// See [`TransportConfig::receive_window()`]
pub fn set_receive_window(&mut self, receive_window: VarInt) {
if self.streams.set_receive_window(receive_window) {
self.spaces[SpaceId::Data].pending.max_data = true;
}
}
fn on_ack_received(
&mut self,
now: Instant,
space: SpaceId,
ack: frame::Ack,
) -> Result<(), TransportError> {
if ack.largest >= self.spaces[space].next_packet_number {
return Err(TransportError::PROTOCOL_VIOLATION("unsent packet acked"));
}
let new_largest = {
let space = &mut self.spaces[space];
if space
.largest_acked_packet
.map_or(true, |pn| ack.largest > pn)
{
space.largest_acked_packet = Some(ack.largest);
if let Some(info) = space.sent_packets.get(&ack.largest) {
// This should always succeed, but a misbehaving peer might ACK a packet we
// haven't sent. At worst, that will result in us spuriously reducing the
// congestion window.
space.largest_acked_packet_sent = info.time_sent;
}
true
} else {
false
}
};
// Avoid DoS from unreasonably huge ack ranges by filtering out just the new acks.
let mut newly_acked = ArrayRangeSet::new();
for range in ack.iter() {
self.packet_number_filter.check_ack(space, range.clone())?;
for (&pn, _) in self.spaces[space].sent_packets.range(range) {
newly_acked.insert_one(pn);
}
}
if newly_acked.is_empty() {
return Ok(());
}
let mut ack_eliciting_acked = false;
for packet in newly_acked.elts() {
if let Some(info) = self.spaces[space].take(packet) {
if let Some(acked) = info.largest_acked {
// Assume ACKs for all packets below the largest acknowledged in `packet` have
// been received. This can cause the peer to spuriously retransmit if some of
// our earlier ACKs were lost, but allows for simpler state tracking. See
// discussion at
// https://www.rfc-editor.org/rfc/rfc9000.html#name-limiting-ranges-by-tracking
self.spaces[space].pending_acks.subtract_below(acked);
}
ack_eliciting_acked |= info.ack_eliciting;
// Notify MTU discovery that a packet was acked, because it might be an MTU probe
let mtu_updated = self.path.mtud.on_acked(space, packet, info.size);
if mtu_updated {
self.path
.congestion
.on_mtu_update(self.path.mtud.current_mtu());
}
// Notify ack frequency that a packet was acked, because it might contain an ACK_FREQUENCY frame
self.ack_frequency.on_acked(packet);
self.on_packet_acked(now, packet, info);
}
}
self.path.congestion.on_end_acks(
now,
self.path.in_flight.bytes,
self.app_limited,
self.spaces[space].largest_acked_packet,
);
if new_largest && ack_eliciting_acked {
let ack_delay = if space != SpaceId::Data {
Duration::from_micros(0)
} else {
cmp::min(
self.ack_frequency.peer_max_ack_delay,
Duration::from_micros(ack.delay << self.peer_params.ack_delay_exponent.0),
)
};
let rtt = instant_saturating_sub(now, self.spaces[space].largest_acked_packet_sent);
self.path.rtt.update(ack_delay, rtt);
if self.path.first_packet_after_rtt_sample.is_none() {
self.path.first_packet_after_rtt_sample =
Some((space, self.spaces[space].next_packet_number));
}
}
// Must be called before crypto/pto_count are clobbered
self.detect_lost_packets(now, space, true);
if self.peer_completed_address_validation() {
self.pto_count = 0;
}
// Explicit congestion notification
if self.path.sending_ecn {
if let Some(ecn) = ack.ecn {
// We only examine ECN counters from ACKs that we are certain we received in transmit
// order, allowing us to compute an increase in ECN counts to compare against the number
// of newly acked packets that remains well-defined in the presence of arbitrary packet
// reordering.
if new_largest {
let sent = self.spaces[space].largest_acked_packet_sent;
self.process_ecn(now, space, newly_acked.len() as u64, ecn, sent);
}
} else {
// We always start out sending ECN, so any ack that doesn't acknowledge it disables it.
debug!("ECN not acknowledged by peer");
self.path.sending_ecn = false;
}
}
self.set_loss_detection_timer(now);
Ok(())
}
/// Process a new ECN block from an in-order ACK
fn process_ecn(
&mut self,
now: Instant,
space: SpaceId,
newly_acked: u64,
ecn: frame::EcnCounts,
largest_sent_time: Instant,
) {
match self.spaces[space].detect_ecn(newly_acked, ecn) {
Err(e) => {
debug!("halting ECN due to verification failure: {}", e);
self.path.sending_ecn = false;
// Wipe out the existing value because it might be garbage and could interfere with
// future attempts to use ECN on new paths.
self.spaces[space].ecn_feedback = frame::EcnCounts::ZERO;
}
Ok(false) => {}
Ok(true) => {
self.stats.path.congestion_events += 1;
self.path
.congestion
.on_congestion_event(now, largest_sent_time, false, 0);
}
}
}
// Not timing-aware, so it's safe to call this for inferred acks, such as arise from
// high-latency handshakes
fn on_packet_acked(&mut self, now: Instant, pn: u64, info: SentPacket) {
self.remove_in_flight(pn, &info);
if info.ack_eliciting && self.path.challenge.is_none() {
// Only pass ACKs to the congestion controller if we are not validating the current
// path, so as to ignore any ACKs from older paths still coming in.
self.path.congestion.on_ack(
now,
info.time_sent,
info.size.into(),
self.app_limited,
&self.path.rtt,
);
}
// Update state for confirmed delivery of frames
if let Some(retransmits) = info.retransmits.get() {
for (id, _) in retransmits.reset_stream.iter() {
self.streams.reset_acked(*id);
}
}
for frame in info.stream_frames {
self.streams.received_ack_of(frame);
}
}
fn set_key_discard_timer(&mut self, now: Instant, space: SpaceId) {
let start = if self.zero_rtt_crypto.is_some() {
now
} else {
self.prev_crypto
.as_ref()
.expect("no previous keys")
.end_packet
.as_ref()
.expect("update not acknowledged yet")
.1
};
self.timers
.set(Timer::KeyDiscard, start + self.pto(space) * 3);
}
fn on_loss_detection_timeout(&mut self, now: Instant) {
if let Some((_, pn_space)) = self.loss_time_and_space() {
// Time threshold loss Detection
self.detect_lost_packets(now, pn_space, false);
self.set_loss_detection_timer(now);
return;
}
let (_, space) = match self.pto_time_and_space(now) {
Some(x) => x,
None => {
error!("PTO expired while unset");
return;
}
};
trace!(
in_flight = self.path.in_flight.bytes,
count = self.pto_count,
?space,
"PTO fired"
);
let count = match self.path.in_flight.ack_eliciting {
// A PTO when we're not expecting any ACKs must be due to handshake anti-amplification
// deadlock preventions
0 => {
debug_assert!(!self.peer_completed_address_validation());
1
}
// Conventional loss probe
_ => 2,
};
self.spaces[space].loss_probes = self.spaces[space].loss_probes.saturating_add(count);
self.pto_count = self.pto_count.saturating_add(1);
self.set_loss_detection_timer(now);
}
fn detect_lost_packets(&mut self, now: Instant, pn_space: SpaceId, due_to_ack: bool) {
let mut lost_packets = Vec::<u64>::new();
let mut lost_mtu_probe = None;
let in_flight_mtu_probe = self.path.mtud.in_flight_mtu_probe();
let rtt = self.path.rtt.conservative();
let loss_delay = cmp::max(rtt.mul_f32(self.config.time_threshold), TIMER_GRANULARITY);
// Packets sent before this time are deemed lost.
let lost_send_time = now.checked_sub(loss_delay).unwrap();
let largest_acked_packet = self.spaces[pn_space].largest_acked_packet.unwrap();
let packet_threshold = self.config.packet_threshold as u64;
let mut size_of_lost_packets = 0u64;
// InPersistentCongestion: Determine if all packets in the time period before the newest
// lost packet, including the edges, are marked lost. PTO computation must always
// include max ACK delay, i.e. operate as if in Data space (see RFC9001 §7.6.1).
let congestion_period =
self.pto(SpaceId::Data) * self.config.persistent_congestion_threshold;
let mut persistent_congestion_start: Option<Instant> = None;
let mut prev_packet = None;
let mut in_persistent_congestion = false;
let space = &mut self.spaces[pn_space];
space.loss_time = None;
for (&packet, info) in space.sent_packets.range(0..largest_acked_packet) {
if prev_packet != Some(packet.wrapping_sub(1)) {
// An intervening packet was acknowledged
persistent_congestion_start = None;
}
if info.time_sent <= lost_send_time || largest_acked_packet >= packet + packet_threshold
{
if Some(packet) == in_flight_mtu_probe {
// Lost MTU probes are not included in `lost_packets`, because they should not
// trigger a congestion control response
lost_mtu_probe = in_flight_mtu_probe;
} else {
lost_packets.push(packet);
size_of_lost_packets += info.size as u64;
if info.ack_eliciting && due_to_ack {
match persistent_congestion_start {
// Two ACK-eliciting packets lost more than congestion_period apart, with no
// ACKed packets in between
Some(start) if info.time_sent - start > congestion_period => {
in_persistent_congestion = true;
}
// Persistent congestion must start after the first RTT sample
None if self
.path
.first_packet_after_rtt_sample
.map_or(false, |x| x < (pn_space, packet)) =>
{
persistent_congestion_start = Some(info.time_sent);
}
_ => {}
}
}
}
} else {
let next_loss_time = info.time_sent + loss_delay;
space.loss_time = Some(
space
.loss_time
.map_or(next_loss_time, |x| cmp::min(x, next_loss_time)),
);
persistent_congestion_start = None;
}
prev_packet = Some(packet);
}
// OnPacketsLost
if let Some(largest_lost) = lost_packets.last().cloned() {
let old_bytes_in_flight = self.path.in_flight.bytes;
let largest_lost_sent = self.spaces[pn_space].sent_packets[&largest_lost].time_sent;
self.lost_packets += lost_packets.len() as u64;
self.stats.path.lost_packets += lost_packets.len() as u64;
self.stats.path.lost_bytes += size_of_lost_packets;
trace!(
"packets lost: {:?}, bytes lost: {}",
lost_packets,
size_of_lost_packets
);
for &packet in &lost_packets {
let info = self.spaces[pn_space].take(packet).unwrap(); // safe: lost_packets is populated just above
self.remove_in_flight(packet, &info);
for frame in info.stream_frames {
self.streams.retransmit(frame);
}
self.spaces[pn_space].pending |= info.retransmits;
self.path.mtud.on_non_probe_lost(packet, info.size);
}
if self.path.mtud.black_hole_detected(now) {
self.stats.path.black_holes_detected += 1;
self.path
.congestion
.on_mtu_update(self.path.mtud.current_mtu());
if let Some(max_datagram_size) = self.datagrams().max_size() {
self.datagrams.drop_oversized(max_datagram_size);
}
}
// Don't apply congestion penalty for lost ack-only packets
let lost_ack_eliciting = old_bytes_in_flight != self.path.in_flight.bytes;
if lost_ack_eliciting {
self.stats.path.congestion_events += 1;
self.path.congestion.on_congestion_event(
now,
largest_lost_sent,
in_persistent_congestion,
size_of_lost_packets,
);
}
}
// Handle a lost MTU probe
if let Some(packet) = lost_mtu_probe {
let info = self.spaces[SpaceId::Data].take(packet).unwrap(); // safe: lost_mtu_probe is omitted from lost_packets, and therefore must not have been removed yet
self.remove_in_flight(packet, &info);
self.path.mtud.on_probe_lost();
self.stats.path.lost_plpmtud_probes += 1;
}
}
fn loss_time_and_space(&self) -> Option<(Instant, SpaceId)> {
SpaceId::iter()
.filter_map(|id| Some((self.spaces[id].loss_time?, id)))
.min_by_key(|&(time, _)| time)
}
fn pto_time_and_space(&self, now: Instant) -> Option<(Instant, SpaceId)> {
let backoff = 2u32.pow(self.pto_count.min(MAX_BACKOFF_EXPONENT));
let mut duration = self.path.rtt.pto_base() * backoff;
if self.path.in_flight.ack_eliciting == 0 {
debug_assert!(!self.peer_completed_address_validation());
let space = match self.highest_space {
SpaceId::Handshake => SpaceId::Handshake,
_ => SpaceId::Initial,
};
return Some((now + duration, space));
}
let mut result = None;
for space in SpaceId::iter() {
if self.spaces[space].in_flight == 0 {
continue;
}
if space == SpaceId::Data {
// Skip ApplicationData until handshake completes.
if self.is_handshaking() {
return result;
}
// Include max_ack_delay and backoff for ApplicationData.
duration += self.ack_frequency.max_ack_delay_for_pto() * backoff;
}
let last_ack_eliciting = match self.spaces[space].time_of_last_ack_eliciting_packet {
Some(time) => time,
None => continue,
};
let pto = last_ack_eliciting + duration;
if result.map_or(true, |(earliest_pto, _)| pto < earliest_pto) {
result = Some((pto, space));
}
}
result
}
#[allow(clippy::suspicious_operation_groupings)]
fn peer_completed_address_validation(&self) -> bool {
if self.side.is_server() || self.state.is_closed() {
return true;
}
// The server is guaranteed to have validated our address if any of our handshake or 1-RTT
// packets are acknowledged or we've seen HANDSHAKE_DONE and discarded handshake keys.
self.spaces[SpaceId::Handshake]
.largest_acked_packet
.is_some()
|| self.spaces[SpaceId::Data].largest_acked_packet.is_some()
|| (self.spaces[SpaceId::Data].crypto.is_some()
&& self.spaces[SpaceId::Handshake].crypto.is_none())
}
fn set_loss_detection_timer(&mut self, now: Instant) {
if self.state.is_closed() {
// No loss detection takes place on closed connections, and `close_common` already
// stopped time timer. Ensure we don't restart it inadvertently, e.g. in response to a
// reordered packet being handled by state-insensitive code.
return;
}
if let Some((loss_time, _)) = self.loss_time_and_space() {
// Time threshold loss detection.
self.timers.set(Timer::LossDetection, loss_time);
return;
}
if self.path.anti_amplification_blocked(1) {
// We wouldn't be able to send anything, so don't bother.
self.timers.stop(Timer::LossDetection);
return;
}
if self.path.in_flight.ack_eliciting == 0 && self.peer_completed_address_validation() {
// There is nothing to detect lost, so no timer is set. However, the client needs to arm
// the timer if the server might be blocked by the anti-amplification limit.
self.timers.stop(Timer::LossDetection);
return;
}
// Determine which PN space to arm PTO for.
// Calculate PTO duration
if let Some((timeout, _)) = self.pto_time_and_space(now) {
self.timers.set(Timer::LossDetection, timeout);
} else {
self.timers.stop(Timer::LossDetection);
}
}
/// Probe Timeout
fn pto(&self, space: SpaceId) -> Duration {
let max_ack_delay = match space {
SpaceId::Initial | SpaceId::Handshake => Duration::new(0, 0),
SpaceId::Data => self.ack_frequency.max_ack_delay_for_pto(),
};
self.path.rtt.pto_base() + max_ack_delay
}
fn on_packet_authenticated(
&mut self,
now: Instant,
space_id: SpaceId,
ecn: Option<EcnCodepoint>,
packet: Option<u64>,
spin: bool,
is_1rtt: bool,
) {
self.total_authed_packets += 1;
self.reset_keep_alive(now);
self.reset_idle_timeout(now, space_id);
self.permit_idle_reset = true;
self.receiving_ecn |= ecn.is_some();
if let Some(x) = ecn {
let space = &mut self.spaces[space_id];
space.ecn_counters += x;
if x.is_ce() {
space.pending_acks.set_immediate_ack_required();
}
}
let packet = match packet {
Some(x) => x,
None => return,
};
if self.side.is_server() {
if self.spaces[SpaceId::Initial].crypto.is_some() && space_id == SpaceId::Handshake {
// A server stops sending and processing Initial packets when it receives its first Handshake packet.
self.discard_space(now, SpaceId::Initial);
}
if self.zero_rtt_crypto.is_some() && is_1rtt {
// Discard 0-RTT keys soon after receiving a 1-RTT packet
self.set_key_discard_timer(now, space_id)
}
}
let space = &mut self.spaces[space_id];
space.pending_acks.insert_one(packet, now);
if packet >= space.rx_packet {
space.rx_packet = packet;
// Update outgoing spin bit, inverting iff we're the client
self.spin = self.side.is_client() ^ spin;
}
}
fn reset_idle_timeout(&mut self, now: Instant, space: SpaceId) {
let timeout = match self.idle_timeout {
None => return,
Some(x) => Duration::from_millis(x.0),
};
if self.state.is_closed() {
self.timers.stop(Timer::Idle);
return;
}
let dt = cmp::max(timeout, 3 * self.pto(space));
self.timers.set(Timer::Idle, now + dt);
}
fn reset_keep_alive(&mut self, now: Instant) {
let interval = match self.config.keep_alive_interval {
Some(x) if self.state.is_established() => x,
_ => return,
};
self.timers.set(Timer::KeepAlive, now + interval);
}
fn reset_cid_retirement(&mut self) {
if let Some(t) = self.local_cid_state.next_timeout() {
self.timers.set(Timer::PushNewCid, t);
}
}
/// Handle the already-decrypted first packet from the client
///
/// Decrypting the first packet in the `Endpoint` allows stateless packet handling to be more
/// efficient.
pub(crate) fn handle_first_packet(
&mut self,
now: Instant,
remote: SocketAddr,
ecn: Option<EcnCodepoint>,
packet_number: u64,
packet: InitialPacket,
remaining: Option<BytesMut>,
) -> Result<(), ConnectionError> {
let span = trace_span!("first recv");
let _guard = span.enter();
debug_assert!(self.side.is_server());
let len = packet.header_data.len() + packet.payload.len();
self.path.total_recvd = len as u64;
match self.state {
State::Handshake(ref mut state) => {
state.expected_token = packet.header.token.clone();
}
_ => unreachable!("first packet must be delivered in Handshake state"),
}
self.on_packet_authenticated(
now,
SpaceId::Initial,
ecn,
Some(packet_number),
false,
false,
);
self.process_decrypted_packet(now, remote, Some(packet_number), packet.into())?;
if let Some(data) = remaining {
self.handle_coalesced(now, remote, ecn, data);
}
Ok(())
}
fn init_0rtt(&mut self) {
let (header, packet) = match self.crypto.early_crypto() {
Some(x) => x,
None => return,
};
if self.side.is_client() {
match self.crypto.transport_parameters() {
Ok(params) => {
let params = params
.expect("crypto layer didn't supply transport parameters with ticket");
// Certain values must not be cached
let params = TransportParameters {
initial_src_cid: None,
original_dst_cid: None,
preferred_address: None,
retry_src_cid: None,
stateless_reset_token: None,
min_ack_delay: None,
ack_delay_exponent: TransportParameters::default().ack_delay_exponent,
max_ack_delay: TransportParameters::default().max_ack_delay,
..params
};
self.set_peer_params(params);
}
Err(e) => {
error!("session ticket has malformed transport parameters: {}", e);
return;
}
}
}
trace!("0-RTT enabled");
self.zero_rtt_enabled = true;
self.zero_rtt_crypto = Some(ZeroRttCrypto { header, packet });
}
fn read_crypto(
&mut self,
space: SpaceId,
crypto: &frame::Crypto,
payload_len: usize,
) -> Result<(), TransportError> {
let expected = if !self.state.is_handshake() {
SpaceId::Data
} else if self.highest_space == SpaceId::Initial {
SpaceId::Initial
} else {
// On the server, self.highest_space can be Data after receiving the client's first
// flight, but we expect Handshake CRYPTO until the handshake is complete.
SpaceId::Handshake
};
// We can't decrypt Handshake packets when highest_space is Initial, CRYPTO frames in 0-RTT
// packets are illegal, and we don't process 1-RTT packets until the handshake is
// complete. Therefore, we will never see CRYPTO data from a later-than-expected space.
debug_assert!(space <= expected, "received out-of-order CRYPTO data");
let end = crypto.offset + crypto.data.len() as u64;
if space < expected && end > self.spaces[space].crypto_stream.bytes_read() {
warn!(
"received new {:?} CRYPTO data when expecting {:?}",
space, expected
);
return Err(TransportError::PROTOCOL_VIOLATION(
"new data at unexpected encryption level",
));
}
let space = &mut self.spaces[space];
let max = end.saturating_sub(space.crypto_stream.bytes_read());
if max > self.config.crypto_buffer_size as u64 {
return Err(TransportError::CRYPTO_BUFFER_EXCEEDED(""));
}
space
.crypto_stream
.insert(crypto.offset, crypto.data.clone(), payload_len);
while let Some(chunk) = space.crypto_stream.read(usize::MAX, true) {
trace!("consumed {} CRYPTO bytes", chunk.bytes.len());
if self.crypto.read_handshake(&chunk.bytes)? {
self.events.push_back(Event::HandshakeDataReady);
}
}
Ok(())
}
fn write_crypto(&mut self) {
loop {
let space = self.highest_space;
let mut outgoing = Vec::new();
if let Some(crypto) = self.crypto.write_handshake(&mut outgoing) {
match space {
SpaceId::Initial => {
self.upgrade_crypto(SpaceId::Handshake, crypto);
}
SpaceId::Handshake => {
self.upgrade_crypto(SpaceId::Data, crypto);
}
_ => unreachable!("got updated secrets during 1-RTT"),
}
}
if outgoing.is_empty() {
if space == self.highest_space {
break;
} else {
// Keys updated, check for more data to send
continue;
}
}
let offset = self.spaces[space].crypto_offset;
let outgoing = Bytes::from(outgoing);
if let State::Handshake(ref mut state) = self.state {
if space == SpaceId::Initial && offset == 0 && self.side.is_client() {
state.client_hello = Some(outgoing.clone());
}
}
self.spaces[space].crypto_offset += outgoing.len() as u64;
trace!("wrote {} {:?} CRYPTO bytes", outgoing.len(), space);
self.spaces[space].pending.crypto.push_back(frame::Crypto {
offset,
data: outgoing,
});
}
}
/// Switch to stronger cryptography during handshake
fn upgrade_crypto(&mut self, space: SpaceId, crypto: Keys) {
debug_assert!(
self.spaces[space].crypto.is_none(),
"already reached packet space {space:?}"
);
trace!("{:?} keys ready", space);
if space == SpaceId::Data {
// Precompute the first key update
self.next_crypto = Some(
self.crypto
.next_1rtt_keys()
.expect("handshake should be complete"),
);
}
self.spaces[space].crypto = Some(crypto);
debug_assert!(space as usize > self.highest_space as usize);
self.highest_space = space;
if space == SpaceId::Data && self.side.is_client() {
// Discard 0-RTT keys because 1-RTT keys are available.
self.zero_rtt_crypto = None;
}
}
fn discard_space(&mut self, now: Instant, space_id: SpaceId) {
debug_assert!(space_id != SpaceId::Data);
trace!("discarding {:?} keys", space_id);
if space_id == SpaceId::Initial {
// No longer needed
self.retry_token = Bytes::new();
}
let space = &mut self.spaces[space_id];
space.crypto = None;
space.time_of_last_ack_eliciting_packet = None;
space.loss_time = None;
space.in_flight = 0;
let sent_packets = mem::take(&mut space.sent_packets);
for (pn, packet) in sent_packets.into_iter() {
self.remove_in_flight(pn, &packet);
}
self.set_loss_detection_timer(now)
}
fn handle_coalesced(
&mut self,
now: Instant,
remote: SocketAddr,
ecn: Option<EcnCodepoint>,
data: BytesMut,
) {
self.path.total_recvd = self.path.total_recvd.saturating_add(data.len() as u64);
let mut remaining = Some(data);
while let Some(data) = remaining {
match PartialDecode::new(
data,
&FixedLengthConnectionIdParser::new(self.local_cid_state.cid_len()),
&[self.version],
self.endpoint_config.grease_quic_bit,
) {
Ok((partial_decode, rest)) => {
remaining = rest;
self.handle_decode(now, remote, ecn, partial_decode);
}
Err(e) => {
trace!("malformed header: {}", e);
return;
}
}
}
}
fn handle_decode(
&mut self,
now: Instant,
remote: SocketAddr,
ecn: Option<EcnCodepoint>,
partial_decode: PartialDecode,
) {
if let Some(decoded) = packet_crypto::unprotect_header(
partial_decode,
&self.spaces,
self.zero_rtt_crypto.as_ref(),
self.peer_params.stateless_reset_token,
) {
self.handle_packet(now, remote, ecn, decoded.packet, decoded.stateless_reset);
}
}
fn handle_packet(
&mut self,
now: Instant,
remote: SocketAddr,
ecn: Option<EcnCodepoint>,
packet: Option<Packet>,
stateless_reset: bool,
) {
self.stats.udp_rx.ios += 1;
if let Some(ref packet) = packet {
trace!(
"got {:?} packet ({} bytes) from {} using id {}",
packet.header.space(),
packet.payload.len() + packet.header_data.len(),
remote,
packet.header.dst_cid(),
);
}
if self.is_handshaking() && remote != self.path.remote {
debug!("discarding packet with unexpected remote during handshake");
return;
}
let was_closed = self.state.is_closed();
let was_drained = self.state.is_drained();
let decrypted = match packet {
None => Err(None),
Some(mut packet) => self
.decrypt_packet(now, &mut packet)
.map(move |number| (packet, number)),
};
let result = match decrypted {
_ if stateless_reset => {
debug!("got stateless reset");
Err(ConnectionError::Reset)
}
Err(Some(e)) => {
warn!("illegal packet: {}", e);
Err(e.into())
}
Err(None) => {
debug!("failed to authenticate packet");
self.authentication_failures += 1;
let integrity_limit = self.spaces[self.highest_space]
.crypto
.as_ref()
.unwrap()
.packet
.local
.integrity_limit();
if self.authentication_failures > integrity_limit {
Err(TransportError::AEAD_LIMIT_REACHED("integrity limit violated").into())
} else {
return;
}
}
Ok((packet, number)) => {
let span = match number {
Some(pn) => trace_span!("recv", space = ?packet.header.space(), pn),
None => trace_span!("recv", space = ?packet.header.space()),
};
let _guard = span.enter();
let is_duplicate = |n| self.spaces[packet.header.space()].dedup.insert(n);
if number.map_or(false, is_duplicate) {
debug!("discarding possible duplicate packet");
return;
} else if self.state.is_handshake() && packet.header.is_short() {
// TODO: SHOULD buffer these to improve reordering tolerance.
trace!("dropping short packet during handshake");
return;
} else {
if let Header::Initial(InitialHeader { ref token, .. }) = packet.header {
if let State::Handshake(ref hs) = self.state {
if self.side.is_server() && token != &hs.expected_token {
// Clients must send the same retry token in every Initial. Initial
// packets can be spoofed, so we discard rather than killing the
// connection.
warn!("discarding Initial with invalid retry token");
return;
}
}
}
if !self.state.is_closed() {
let spin = match packet.header {
Header::Short { spin, .. } => spin,
_ => false,
};
self.on_packet_authenticated(
now,
packet.header.space(),
ecn,
number,
spin,
packet.header.is_1rtt(),
);
}
self.process_decrypted_packet(now, remote, number, packet)
}
}
};
// State transitions for error cases
if let Err(conn_err) = result {
self.error = Some(conn_err.clone());
self.state = match conn_err {
ConnectionError::ApplicationClosed(reason) => State::closed(reason),
ConnectionError::ConnectionClosed(reason) => State::closed(reason),
ConnectionError::Reset
| ConnectionError::TransportError(TransportError {
code: TransportErrorCode::AEAD_LIMIT_REACHED,
..
}) => State::Drained,
ConnectionError::TimedOut => {
unreachable!("timeouts aren't generated by packet processing");
}
ConnectionError::TransportError(err) => {
debug!("closing connection due to transport error: {}", err);
State::closed(err)
}
ConnectionError::VersionMismatch => State::Draining,
ConnectionError::LocallyClosed => {
unreachable!("LocallyClosed isn't generated by packet processing");
}
ConnectionError::CidsExhausted => {
unreachable!("CidsExhausted isn't generated by packet processing");
}
};
}
if !was_closed && self.state.is_closed() {
self.close_common();
if !self.state.is_drained() {
self.set_close_timer(now);
}
}
if !was_drained && self.state.is_drained() {
self.endpoint_events.push_back(EndpointEventInner::Drained);
// Close timer may have been started previously, e.g. if we sent a close and got a
// stateless reset in response
self.timers.stop(Timer::Close);
}
// Transmit CONNECTION_CLOSE if necessary
if let State::Closed(_) = self.state {
self.close = remote == self.path.remote;
}
}
fn process_decrypted_packet(
&mut self,
now: Instant,
remote: SocketAddr,
number: Option<u64>,
packet: Packet,
) -> Result<(), ConnectionError> {
let state = match self.state {
State::Established => {
match packet.header.space() {
SpaceId::Data => self.process_payload(now, remote, number.unwrap(), packet)?,
_ if packet.header.has_frames() => self.process_early_payload(now, packet)?,
_ => {
trace!("discarding unexpected pre-handshake packet");
}
}
return Ok(());
}
State::Closed(_) => {
for result in frame::Iter::new(packet.payload.freeze())? {
let frame = match result {
Ok(frame) => frame,
Err(err) => {
debug!("frame decoding error: {err:?}");
continue;
}
};
if let Frame::Padding = frame {
continue;
};
self.stats.frame_rx.record(&frame);
if let Frame::Close(_) = frame {
trace!("draining");
self.state = State::Draining;
break;
}
}
return Ok(());
}
State::Draining | State::Drained => return Ok(()),
State::Handshake(ref mut state) => state,
};
match packet.header {
Header::Retry {
src_cid: rem_cid, ..
} => {
if self.side.is_server() {
return Err(TransportError::PROTOCOL_VIOLATION("client sent Retry").into());
}
if self.total_authed_packets > 1
|| packet.payload.len() <= 16 // token + 16 byte tag
|| !self.crypto.is_valid_retry(
&self.rem_cids.active(),
&packet.header_data,
&packet.payload,
)
{
trace!("discarding invalid Retry");
// - After the client has received and processed an Initial or Retry
// packet from the server, it MUST discard any subsequent Retry
// packets that it receives.
// - A client MUST discard a Retry packet with a zero-length Retry Token
// field.
// - Clients MUST discard Retry packets that have a Retry Integrity Tag
// that cannot be validated
return Ok(());
}
trace!("retrying with CID {}", rem_cid);
let client_hello = state.client_hello.take().unwrap();
self.retry_src_cid = Some(rem_cid);
self.rem_cids.update_initial_cid(rem_cid);
self.rem_handshake_cid = rem_cid;
let space = &mut self.spaces[SpaceId::Initial];
if let Some(info) = space.take(0) {
self.on_packet_acked(now, 0, info);
};
self.discard_space(now, SpaceId::Initial); // Make sure we clean up after any retransmitted Initials
self.spaces[SpaceId::Initial] = PacketSpace {
crypto: Some(self.crypto.initial_keys(&rem_cid, self.side)),
next_packet_number: self.spaces[SpaceId::Initial].next_packet_number,
crypto_offset: client_hello.len() as u64,
..PacketSpace::new(now)
};
self.spaces[SpaceId::Initial]
.pending
.crypto
.push_back(frame::Crypto {
offset: 0,
data: client_hello,
});
// Retransmit all 0-RTT data
let zero_rtt = mem::take(&mut self.spaces[SpaceId::Data].sent_packets);
for (pn, info) in zero_rtt {
self.remove_in_flight(pn, &info);
self.spaces[SpaceId::Data].pending |= info.retransmits;
}
self.streams.retransmit_all_for_0rtt();
let token_len = packet.payload.len() - 16;
self.retry_token = packet.payload.freeze().split_to(token_len);
self.state = State::Handshake(state::Handshake {
expected_token: Bytes::new(),
rem_cid_set: false,
client_hello: None,
});
Ok(())
}
Header::Long {
ty: LongType::Handshake,
src_cid: rem_cid,
..
} => {
if rem_cid != self.rem_handshake_cid {
debug!(
"discarding packet with mismatched remote CID: {} != {}",
self.rem_handshake_cid, rem_cid
);
return Ok(());
}
self.path.validated = true;
self.process_early_payload(now, packet)?;
if self.state.is_closed() {
return Ok(());
}
if self.crypto.is_handshaking() {
trace!("handshake ongoing");
return Ok(());
}
if self.side.is_client() {
// Client-only because server params were set from the client's Initial
let params =
self.crypto
.transport_parameters()?
.ok_or_else(|| TransportError {
code: TransportErrorCode::crypto(0x6d),
frame: None,
reason: "transport parameters missing".into(),
})?;
if self.has_0rtt() {
if !self.crypto.early_data_accepted().unwrap() {
debug_assert!(self.side.is_client());
debug!("0-RTT rejected");
self.accepted_0rtt = false;
self.streams.zero_rtt_rejected();
// Discard already-queued frames
self.spaces[SpaceId::Data].pending = Retransmits::default();
// Discard 0-RTT packets
let sent_packets =
mem::take(&mut self.spaces[SpaceId::Data].sent_packets);
for (pn, packet) in sent_packets {
self.remove_in_flight(pn, &packet);
}
} else {
self.accepted_0rtt = true;
params.validate_resumption_from(&self.peer_params)?;
}
}
if let Some(token) = params.stateless_reset_token {
self.endpoint_events
.push_back(EndpointEventInner::ResetToken(self.path.remote, token));
}
self.handle_peer_params(params)?;
self.issue_first_cids(now);
} else {
// Server-only
self.spaces[SpaceId::Data].pending.handshake_done = true;
self.discard_space(now, SpaceId::Handshake);
}
self.events.push_back(Event::Connected);
self.state = State::Established;
trace!("established");
Ok(())
}
Header::Initial(InitialHeader {
src_cid: rem_cid, ..
}) => {
if !state.rem_cid_set {
trace!("switching remote CID to {}", rem_cid);
let mut state = state.clone();
self.rem_cids.update_initial_cid(rem_cid);
self.rem_handshake_cid = rem_cid;
self.orig_rem_cid = rem_cid;
state.rem_cid_set = true;
self.state = State::Handshake(state);
} else if rem_cid != self.rem_handshake_cid {
debug!(
"discarding packet with mismatched remote CID: {} != {}",
self.rem_handshake_cid, rem_cid
);
return Ok(());
}
let starting_space = self.highest_space;
self.process_early_payload(now, packet)?;
if self.side.is_server()
&& starting_space == SpaceId::Initial
&& self.highest_space != SpaceId::Initial
{
let params =
self.crypto
.transport_parameters()?
.ok_or_else(|| TransportError {
code: TransportErrorCode::crypto(0x6d),
frame: None,
reason: "transport parameters missing".into(),
})?;
self.handle_peer_params(params)?;
self.issue_first_cids(now);
self.init_0rtt();
}
Ok(())
}
Header::Long {
ty: LongType::ZeroRtt,
..
} => {
self.process_payload(now, remote, number.unwrap(), packet)?;
Ok(())
}
Header::VersionNegotiate { .. } => {
if self.total_authed_packets > 1 {
return Ok(());
}
let supported = packet
.payload
.chunks(4)
.any(|x| match <[u8; 4]>::try_from(x) {
Ok(version) => self.version == u32::from_be_bytes(version),
Err(_) => false,
});
if supported {
return Ok(());
}
debug!("remote doesn't support our version");
Err(ConnectionError::VersionMismatch)
}
Header::Short { .. } => unreachable!(
"short packets received during handshake are discarded in handle_packet"
),
}
}
/// Process an Initial or Handshake packet payload
fn process_early_payload(
&mut self,
now: Instant,
packet: Packet,
) -> Result<(), TransportError> {
debug_assert_ne!(packet.header.space(), SpaceId::Data);
let payload_len = packet.payload.len();
let mut ack_eliciting = false;
for result in frame::Iter::new(packet.payload.freeze())? {
let frame = result?;
let span = match frame {
Frame::Padding => continue,
_ => Some(trace_span!("frame", ty = %frame.ty())),
};
self.stats.frame_rx.record(&frame);
let _guard = span.as_ref().map(|x| x.enter());
ack_eliciting |= frame.is_ack_eliciting();
// Process frames
match frame {
Frame::Padding | Frame::Ping => {}
Frame::Crypto(frame) => {
self.read_crypto(packet.header.space(), &frame, payload_len)?;
}
Frame::Ack(ack) => {
self.on_ack_received(now, packet.header.space(), ack)?;
}
Frame::Close(reason) => {
self.error = Some(reason.into());
self.state = State::Draining;
return Ok(());
}
_ => {
let mut err =
TransportError::PROTOCOL_VIOLATION("illegal frame type in handshake");
err.frame = Some(frame.ty());
return Err(err);
}
}
}
if ack_eliciting {
// In the initial and handshake spaces, ACKs must be sent immediately
self.spaces[packet.header.space()]
.pending_acks
.set_immediate_ack_required();
}
self.write_crypto();
Ok(())
}
fn process_payload(
&mut self,
now: Instant,
remote: SocketAddr,
number: u64,
packet: Packet,
) -> Result<(), TransportError> {
let payload = packet.payload.freeze();
let mut is_probing_packet = true;
let mut close = None;
let payload_len = payload.len();
let mut ack_eliciting = false;
for result in frame::Iter::new(payload)? {
let frame = result?;
let span = match frame {
Frame::Padding => continue,
_ => Some(trace_span!("frame", ty = %frame.ty())),
};
self.stats.frame_rx.record(&frame);
// Crypto, Stream and Datagram frames are special cased in order no pollute
// the log with payload data
match &frame {
Frame::Crypto(f) => {
trace!(offset = f.offset, len = f.data.len(), "got crypto frame");
}
Frame::Stream(f) => {
trace!(id = %f.id, offset = f.offset, len = f.data.len(), fin = f.fin, "got stream frame");
}
Frame::Datagram(f) => {
trace!(len = f.data.len(), "got datagram frame");
}
f => {
trace!("got frame {:?}", f);
}
}
let _guard = span.as_ref().map(|x| x.enter());
if packet.header.is_0rtt() {
match frame {
Frame::Crypto(_) | Frame::Close(Close::Application(_)) => {
return Err(TransportError::PROTOCOL_VIOLATION(
"illegal frame type in 0-RTT",
));
}
_ => {}
}
}
ack_eliciting |= frame.is_ack_eliciting();
// Check whether this could be a probing packet
match frame {
Frame::Padding
| Frame::PathChallenge(_)
| Frame::PathResponse(_)
| Frame::NewConnectionId(_) => {}
_ => {
is_probing_packet = false;
}
}
match frame {
Frame::Crypto(frame) => {
self.read_crypto(SpaceId::Data, &frame, payload_len)?;
}
Frame::Stream(frame) => {
if self.streams.received(frame, payload_len)?.should_transmit() {
self.spaces[SpaceId::Data].pending.max_data = true;
}
}
Frame::Ack(ack) => {
self.on_ack_received(now, SpaceId::Data, ack)?;
}
Frame::Padding | Frame::Ping => {}
Frame::Close(reason) => {
close = Some(reason);
}
Frame::PathChallenge(token) => {
self.path_responses.push(number, token, remote);
if remote == self.path.remote {
// PATH_CHALLENGE on active path, possible off-path packet forwarding
// attack. Send a non-probing packet to recover the active path.
match self.peer_supports_ack_frequency() {
true => self.immediate_ack(),
false => self.ping(),
}
}
}
Frame::PathResponse(token) => {
if self.path.challenge == Some(token) && remote == self.path.remote {
trace!("new path validated");
self.timers.stop(Timer::PathValidation);
self.path.challenge = None;
self.path.validated = true;
if let Some((_, ref mut prev_path)) = self.prev_path {
prev_path.challenge = None;
prev_path.challenge_pending = false;
}
} else {
debug!(token, "ignoring invalid PATH_RESPONSE");
}
}
Frame::MaxData(bytes) => {
self.streams.received_max_data(bytes);
}
Frame::MaxStreamData { id, offset } => {
self.streams.received_max_stream_data(id, offset)?;
}
Frame::MaxStreams { dir, count } => {
self.streams.received_max_streams(dir, count)?;
}
Frame::ResetStream(frame) => {
if self.streams.received_reset(frame)?.should_transmit() {
self.spaces[SpaceId::Data].pending.max_data = true;
}
}
Frame::DataBlocked { offset } => {
debug!(offset, "peer claims to be blocked at connection level");
}
Frame::StreamDataBlocked { id, offset } => {
if id.initiator() == self.side && id.dir() == Dir::Uni {
debug!("got STREAM_DATA_BLOCKED on send-only {}", id);
return Err(TransportError::STREAM_STATE_ERROR(
"STREAM_DATA_BLOCKED on send-only stream",
));
}
debug!(
stream = %id,
offset, "peer claims to be blocked at stream level"
);
}
Frame::StreamsBlocked { dir, limit } => {
if limit > MAX_STREAM_COUNT {
return Err(TransportError::FRAME_ENCODING_ERROR(
"unrepresentable stream limit",
));
}
debug!(
"peer claims to be blocked opening more than {} {} streams",
limit, dir
);
}
Frame::StopSending(frame::StopSending { id, error_code }) => {
if id.initiator() != self.side {
if id.dir() == Dir::Uni {
debug!("got STOP_SENDING on recv-only {}", id);
return Err(TransportError::STREAM_STATE_ERROR(
"STOP_SENDING on recv-only stream",
));
}
} else if self.streams.is_local_unopened(id) {
return Err(TransportError::STREAM_STATE_ERROR(
"STOP_SENDING on unopened stream",
));
}
self.streams.received_stop_sending(id, error_code);
}
Frame::RetireConnectionId { sequence } => {
let allow_more_cids = self
.local_cid_state
.on_cid_retirement(sequence, self.peer_params.issue_cids_limit())?;
self.endpoint_events
.push_back(EndpointEventInner::RetireConnectionId(
now,
sequence,
allow_more_cids,
));
}
Frame::NewConnectionId(frame) => {
trace!(
sequence = frame.sequence,
id = %frame.id,
retire_prior_to = frame.retire_prior_to,
);
if self.rem_cids.active().is_empty() {
return Err(TransportError::PROTOCOL_VIOLATION(
"NEW_CONNECTION_ID when CIDs aren't in use",
));
}
if frame.retire_prior_to > frame.sequence {
return Err(TransportError::PROTOCOL_VIOLATION(
"NEW_CONNECTION_ID retiring unissued CIDs",
));
}
use crate::cid_queue::InsertError;
match self.rem_cids.insert(frame) {
Ok(None) => {}
Ok(Some((retired, reset_token))) => {
let pending_retired =
&mut self.spaces[SpaceId::Data].pending.retire_cids;
/// Ensure `pending_retired` cannot grow without bound. Limit is
/// somewhat arbitrary but very permissive.
const MAX_PENDING_RETIRED_CIDS: u64 = CidQueue::LEN as u64 * 10;
// We don't bother counting in-flight frames because those are bounded
// by congestion control.
if (pending_retired.len() as u64)
.saturating_add(retired.end.saturating_sub(retired.start))
> MAX_PENDING_RETIRED_CIDS
{
return Err(TransportError::CONNECTION_ID_LIMIT_ERROR(
"queued too many retired CIDs",
));
}
pending_retired.extend(retired);
self.set_reset_token(reset_token);
}
Err(InsertError::ExceedsLimit) => {
return Err(TransportError::CONNECTION_ID_LIMIT_ERROR(""));
}
Err(InsertError::Retired) => {
trace!("discarding already-retired");
// RETIRE_CONNECTION_ID might not have been previously sent if e.g. a
// range of connection IDs larger than the active connection ID limit
// was retired all at once via retire_prior_to.
self.spaces[SpaceId::Data]
.pending
.retire_cids
.push(frame.sequence);
continue;
}
};
if self.side.is_server() && self.rem_cids.active_seq() == 0 {
// We're a server still using the initial remote CID for the client, so
// let's switch immediately to enable clientside stateless resets.
self.update_rem_cid();
}
}
Frame::NewToken { token } => {
if self.side.is_server() {
return Err(TransportError::PROTOCOL_VIOLATION("client sent NEW_TOKEN"));
}
if token.is_empty() {
return Err(TransportError::FRAME_ENCODING_ERROR("empty token"));
}
trace!("got new token");
// TODO: Cache, or perhaps forward to user?
}
Frame::Datagram(datagram) => {
if self
.datagrams
.received(datagram, &self.config.datagram_receive_buffer_size)?
{
self.events.push_back(Event::DatagramReceived);
}
}
Frame::AckFrequency(ack_frequency) => {
// This frame can only be sent in the Data space
let space = &mut self.spaces[SpaceId::Data];
if !self
.ack_frequency
.ack_frequency_received(&ack_frequency, &mut space.pending_acks)?
{
// The AckFrequency frame is stale (we have already received a more recent one)
continue;
}
// Our `max_ack_delay` has been updated, so we may need to adjust its associated
// timeout
if let Some(timeout) = space
.pending_acks
.max_ack_delay_timeout(self.ack_frequency.max_ack_delay)
{
self.timers.set(Timer::MaxAckDelay, timeout);
}
}
Frame::ImmediateAck => {
// This frame can only be sent in the Data space
self.spaces[SpaceId::Data]
.pending_acks
.set_immediate_ack_required();
}
Frame::HandshakeDone => {
if self.side.is_server() {
return Err(TransportError::PROTOCOL_VIOLATION(
"client sent HANDSHAKE_DONE",
));
}
if self.spaces[SpaceId::Handshake].crypto.is_some() {
self.discard_space(now, SpaceId::Handshake);
}
}
}
}
let space = &mut self.spaces[SpaceId::Data];
if space
.pending_acks
.packet_received(now, number, ack_eliciting, &space.dedup)
{
self.timers
.set(Timer::MaxAckDelay, now + self.ack_frequency.max_ack_delay);
}
// Issue stream ID credit due to ACKs of outgoing finish/resets and incoming finish/resets
// on stopped streams. Incoming finishes/resets on open streams are not handled here as they
// are only freed, and hence only issue credit, once the application has been notified
// during a read on the stream.
let pending = &mut self.spaces[SpaceId::Data].pending;
self.streams.queue_max_stream_id(pending);
if let Some(reason) = close {
self.error = Some(reason.into());
self.state = State::Draining;
self.close = true;
}
if remote != self.path.remote
&& !is_probing_packet
&& number == self.spaces[SpaceId::Data].rx_packet
{
debug_assert!(
self.server_config
.as_ref()
.expect("packets from unknown remote should be dropped by clients")
.migration,
"migration-initiating packets should have been dropped immediately"
);
self.migrate(now, remote);
// Break linkability, if possible
self.update_rem_cid();
self.spin = false;
}
Ok(())
}
fn migrate(&mut self, now: Instant, remote: SocketAddr) {
trace!(%remote, "migration initiated");
// Reset rtt/congestion state for new path unless it looks like a NAT rebinding.
// Note that the congestion window will not grow until validation terminates. Helps mitigate
// amplification attacks performed by spoofing source addresses.
let mut new_path = if remote.is_ipv4() && remote.ip() == self.path.remote.ip() {
PathData::from_previous(remote, &self.path, now)
} else {
let peer_max_udp_payload_size =
u16::try_from(self.peer_params.max_udp_payload_size.into_inner())
.unwrap_or(u16::MAX);
PathData::new(
remote,
self.allow_mtud,
Some(peer_max_udp_payload_size),
now,
false,
&self.config,
)
};
new_path.challenge = Some(self.rng.gen());
new_path.challenge_pending = true;
let prev_pto = self.pto(SpaceId::Data);
let mut prev = mem::replace(&mut self.path, new_path);
// Don't clobber the original path if the previous one hasn't been validated yet
if prev.challenge.is_none() {
prev.challenge = Some(self.rng.gen());
prev.challenge_pending = true;
// We haven't updated the remote CID yet, this captures the remote CID we were using on
// the previous path.
self.prev_path = Some((self.rem_cids.active(), prev));
}
self.timers.set(
Timer::PathValidation,
now + 3 * cmp::max(self.pto(SpaceId::Data), prev_pto),
);
}
/// Handle a change in the local address, i.e. an active migration
pub fn local_address_changed(&mut self) {
self.update_rem_cid();
self.ping();
}
/// Switch to a previously unused remote connection ID, if possible
fn update_rem_cid(&mut self) {
let (reset_token, retired) = match self.rem_cids.next() {
Some(x) => x,
None => return,
};
// Retire the current remote CID and any CIDs we had to skip.
self.spaces[SpaceId::Data]
.pending
.retire_cids
.extend(retired);
self.set_reset_token(reset_token);
}
fn set_reset_token(&mut self, reset_token: ResetToken) {
self.endpoint_events
.push_back(EndpointEventInner::ResetToken(
self.path.remote,
reset_token,
));
self.peer_params.stateless_reset_token = Some(reset_token);
}
/// Issue an initial set of connection IDs to the peer upon connection
fn issue_first_cids(&mut self, now: Instant) {
if self.local_cid_state.cid_len() == 0 {
return;
}
// Subtract 1 to account for the CID we supplied while handshaking
let n = self.peer_params.issue_cids_limit() - 1;
self.endpoint_events
.push_back(EndpointEventInner::NeedIdentifiers(now, n));
}
fn populate_packet(
&mut self,
now: Instant,
space_id: SpaceId,
buf: &mut Vec<u8>,
max_size: usize,
pn: u64,
) -> SentFrames {
let mut sent = SentFrames::default();
let space = &mut self.spaces[space_id];
let is_0rtt = space_id == SpaceId::Data && space.crypto.is_none();
space.pending_acks.maybe_ack_non_eliciting();
// HANDSHAKE_DONE
if !is_0rtt && mem::replace(&mut space.pending.handshake_done, false) {
buf.write(frame::Type::HANDSHAKE_DONE);
sent.retransmits.get_or_create().handshake_done = true;
// This is just a u8 counter and the frame is typically just sent once
self.stats.frame_tx.handshake_done =
self.stats.frame_tx.handshake_done.saturating_add(1);
}
// PING
if mem::replace(&mut space.ping_pending, false) {
trace!("PING");
buf.write(frame::Type::PING);
sent.non_retransmits = true;
self.stats.frame_tx.ping += 1;
}
// IMMEDIATE_ACK
if mem::replace(&mut space.immediate_ack_pending, false) {
trace!("IMMEDIATE_ACK");
buf.write(frame::Type::IMMEDIATE_ACK);
sent.non_retransmits = true;
self.stats.frame_tx.immediate_ack += 1;
}
// ACK
if space.pending_acks.can_send() {
Self::populate_acks(
now,
self.receiving_ecn,
&mut sent,
space,
buf,
&mut self.stats,
);
}
// ACK_FREQUENCY
if mem::replace(&mut space.pending.ack_frequency, false) {
let sequence_number = self.ack_frequency.next_sequence_number();
// Safe to unwrap because this is always provided when ACK frequency is enabled
let config = self.config.ack_frequency_config.as_ref().unwrap();
// Ensure the delay is within bounds to avoid a PROTOCOL_VIOLATION error
let max_ack_delay = self.ack_frequency.candidate_max_ack_delay(
self.path.rtt.get(),
config,
&self.peer_params,
);
trace!(?max_ack_delay, "ACK_FREQUENCY");
frame::AckFrequency {
sequence: sequence_number,
ack_eliciting_threshold: config.ack_eliciting_threshold,
request_max_ack_delay: max_ack_delay.as_micros().try_into().unwrap_or(VarInt::MAX),
reordering_threshold: config.reordering_threshold,
}
.encode(buf);
sent.retransmits.get_or_create().ack_frequency = true;
self.ack_frequency.ack_frequency_sent(pn, max_ack_delay);
self.stats.frame_tx.ack_frequency += 1;
}
// PATH_CHALLENGE
if buf.len() + 9 < max_size && space_id == SpaceId::Data {
// Transmit challenges with every outgoing frame on an unvalidated path
if let Some(token) = self.path.challenge {
// But only send a packet solely for that purpose at most once
self.path.challenge_pending = false;
sent.non_retransmits = true;
sent.requires_padding = true;
trace!("PATH_CHALLENGE {:08x}", token);
buf.write(frame::Type::PATH_CHALLENGE);
buf.write(token);
self.stats.frame_tx.path_challenge += 1;
}
}
// PATH_RESPONSE
if buf.len() + 9 < max_size && space_id == SpaceId::Data {
if let Some(token) = self.path_responses.pop_on_path(&self.path.remote) {
sent.non_retransmits = true;
sent.requires_padding = true;
trace!("PATH_RESPONSE {:08x}", token);
buf.write(frame::Type::PATH_RESPONSE);
buf.write(token);
self.stats.frame_tx.path_response += 1;
}
}
// CRYPTO
while buf.len() + frame::Crypto::SIZE_BOUND < max_size && !is_0rtt {
let mut frame = match space.pending.crypto.pop_front() {
Some(x) => x,
None => break,
};
// Calculate the maximum amount of crypto data we can store in the buffer.
// Since the offset is known, we can reserve the exact size required to encode it.
// For length we reserve 2bytes which allows to encode up to 2^14,
// which is more than what fits into normally sized QUIC frames.
let max_crypto_data_size = max_size
- buf.len()
- 1 // Frame Type
- VarInt::size(unsafe { VarInt::from_u64_unchecked(frame.offset) })
- 2; // Maximum encoded length for frame size, given we send less than 2^14 bytes
let len = frame
.data
.len()
.min(2usize.pow(14) - 1)
.min(max_crypto_data_size);
let data = frame.data.split_to(len);
let truncated = frame::Crypto {
offset: frame.offset,
data,
};
trace!(
"CRYPTO: off {} len {}",
truncated.offset,
truncated.data.len()
);
truncated.encode(buf);
self.stats.frame_tx.crypto += 1;
sent.retransmits.get_or_create().crypto.push_back(truncated);
if !frame.data.is_empty() {
frame.offset += len as u64;
space.pending.crypto.push_front(frame);
}
}
if space_id == SpaceId::Data {
self.streams.write_control_frames(
buf,
&mut space.pending,
&mut sent.retransmits,
&mut self.stats.frame_tx,
max_size,
);
}
// NEW_CONNECTION_ID
while buf.len() + 44 < max_size {
let issued = match space.pending.new_cids.pop() {
Some(x) => x,
None => break,
};
trace!(
sequence = issued.sequence,
id = %issued.id,
"NEW_CONNECTION_ID"
);
frame::NewConnectionId {
sequence: issued.sequence,
retire_prior_to: self.local_cid_state.retire_prior_to(),
id: issued.id,
reset_token: issued.reset_token,
}
.encode(buf);
sent.retransmits.get_or_create().new_cids.push(issued);
self.stats.frame_tx.new_connection_id += 1;
}
// RETIRE_CONNECTION_ID
while buf.len() + frame::RETIRE_CONNECTION_ID_SIZE_BOUND < max_size {
let seq = match space.pending.retire_cids.pop() {
Some(x) => x,
None => break,
};
trace!(sequence = seq, "RETIRE_CONNECTION_ID");
buf.write(frame::Type::RETIRE_CONNECTION_ID);
buf.write_var(seq);
sent.retransmits.get_or_create().retire_cids.push(seq);
self.stats.frame_tx.retire_connection_id += 1;
}
// DATAGRAM
let mut sent_datagrams = false;
while buf.len() + Datagram::SIZE_BOUND < max_size && space_id == SpaceId::Data {
match self.datagrams.write(buf, max_size) {
true => {
sent_datagrams = true;
sent.non_retransmits = true;
self.stats.frame_tx.datagram += 1;
}
false => break,
}
}
if self.datagrams.send_blocked && sent_datagrams {
self.events.push_back(Event::DatagramsUnblocked);
self.datagrams.send_blocked = false;
}
// STREAM
if space_id == SpaceId::Data {
sent.stream_frames = self.streams.write_stream_frames(buf, max_size);
self.stats.frame_tx.stream += sent.stream_frames.len() as u64;
}
sent
}
/// Write pending ACKs into a buffer
///
/// This method assumes ACKs are pending, and should only be called if
/// `!PendingAcks::ranges().is_empty()` returns `true`.
fn populate_acks(
now: Instant,
receiving_ecn: bool,
sent: &mut SentFrames,
space: &mut PacketSpace,
buf: &mut Vec<u8>,
stats: &mut ConnectionStats,
) {
debug_assert!(!space.pending_acks.ranges().is_empty());
// 0-RTT packets must never carry acks (which would have to be of handshake packets)
debug_assert!(space.crypto.is_some(), "tried to send ACK in 0-RTT");
let ecn = if receiving_ecn {
Some(&space.ecn_counters)
} else {
None
};
sent.largest_acked = space.pending_acks.ranges().max();
let delay_micros = space.pending_acks.ack_delay(now).as_micros() as u64;
// TODO: This should come from `TransportConfig` if that gets configurable.
let ack_delay_exp = TransportParameters::default().ack_delay_exponent;
let delay = delay_micros >> ack_delay_exp.into_inner();
trace!(
"ACK {:?}, Delay = {}us",
space.pending_acks.ranges(),
delay_micros
);
frame::Ack::encode(delay as _, space.pending_acks.ranges(), ecn, buf);
stats.frame_tx.acks += 1;
}
fn close_common(&mut self) {
trace!("connection closed");
for &timer in &Timer::VALUES {
self.timers.stop(timer);
}
}
fn set_close_timer(&mut self, now: Instant) {
self.timers
.set(Timer::Close, now + 3 * self.pto(self.highest_space));
}
/// Handle transport parameters received from the peer
fn handle_peer_params(&mut self, params: TransportParameters) -> Result<(), TransportError> {
if Some(self.orig_rem_cid) != params.initial_src_cid
|| (self.side.is_client()
&& (Some(self.initial_dst_cid) != params.original_dst_cid
|| self.retry_src_cid != params.retry_src_cid))
{
return Err(TransportError::TRANSPORT_PARAMETER_ERROR(
"CID authentication failure",
));
}
self.set_peer_params(params);
Ok(())
}
fn set_peer_params(&mut self, params: TransportParameters) {
self.streams.set_params(¶ms);
self.idle_timeout = match (self.config.max_idle_timeout, params.max_idle_timeout) {
(None, VarInt(0)) => None,
(None, x) => Some(x),
(Some(x), VarInt(0)) => Some(x),
(Some(x), y) => Some(cmp::min(x, y)),
};
if let Some(ref info) = params.preferred_address {
self.rem_cids.insert(frame::NewConnectionId {
sequence: 1,
id: info.connection_id,
reset_token: info.stateless_reset_token,
retire_prior_to: 0,
}).expect("preferred address CID is the first received, and hence is guaranteed to be legal");
}
self.ack_frequency.peer_max_ack_delay = get_max_ack_delay(¶ms);
self.peer_params = params;
self.path.mtud.on_peer_max_udp_payload_size_received(
u16::try_from(self.peer_params.max_udp_payload_size.into_inner()).unwrap_or(u16::MAX),
);
}
fn decrypt_packet(
&mut self,
now: Instant,
packet: &mut Packet,
) -> Result<Option<u64>, Option<TransportError>> {
let result = packet_crypto::decrypt_packet_body(
packet,
&self.spaces,
self.zero_rtt_crypto.as_ref(),
self.key_phase,
self.prev_crypto.as_ref(),
self.next_crypto.as_ref(),
)?;
let result = match result {
Some(r) => r,
None => return Ok(None),
};
if result.outgoing_key_update_acked {
if let Some(prev) = self.prev_crypto.as_mut() {
prev.end_packet = Some((result.number, now));
self.set_key_discard_timer(now, packet.header.space());
}
}
if result.incoming_key_update {
trace!("key update authenticated");
self.update_keys(Some((result.number, now)), true);
self.set_key_discard_timer(now, packet.header.space());
}
Ok(Some(result.number))
}
fn update_keys(&mut self, end_packet: Option<(u64, Instant)>, remote: bool) {
trace!("executing key update");
// Generate keys for the key phase after the one we're switching to, store them in
// `next_crypto`, make the contents of `next_crypto` current, and move the current keys into
// `prev_crypto`.
let new = self
.crypto
.next_1rtt_keys()
.expect("only called for `Data` packets");
self.key_phase_size = new
.local
.confidentiality_limit()
.saturating_sub(KEY_UPDATE_MARGIN);
let old = mem::replace(
&mut self.spaces[SpaceId::Data]
.crypto
.as_mut()
.unwrap() // safe because update_keys() can only be triggered by short packets
.packet,
mem::replace(self.next_crypto.as_mut().unwrap(), new),
);
self.spaces[SpaceId::Data].sent_with_keys = 0;
self.prev_crypto = Some(PrevCrypto {
crypto: old,
end_packet,
update_unacked: remote,
});
self.key_phase = !self.key_phase;
}
fn peer_supports_ack_frequency(&self) -> bool {
self.peer_params.min_ack_delay.is_some()
}
/// Send an IMMEDIATE_ACK frame to the remote endpoint
///
/// According to the spec, this will result in an error if the remote endpoint does not support
/// the Acknowledgement Frequency extension
pub(crate) fn immediate_ack(&mut self) {
self.spaces[self.highest_space].immediate_ack_pending = true;
}
/// Decodes a packet, returning its decrypted payload, so it can be inspected in tests
#[cfg(test)]
pub(crate) fn decode_packet(&self, event: &ConnectionEvent) -> Option<Vec<u8>> {
let (first_decode, remaining) = match &event.0 {
ConnectionEventInner::Datagram(DatagramConnectionEvent {
first_decode,
remaining,
..
}) => (first_decode, remaining),
_ => return None,
};
if remaining.is_some() {
panic!("Packets should never be coalesced in tests");
}
let decrypted_header = packet_crypto::unprotect_header(
first_decode.clone(),
&self.spaces,
self.zero_rtt_crypto.as_ref(),
self.peer_params.stateless_reset_token,
)?;
let mut packet = decrypted_header.packet?;
packet_crypto::decrypt_packet_body(
&mut packet,
&self.spaces,
self.zero_rtt_crypto.as_ref(),
self.key_phase,
self.prev_crypto.as_ref(),
self.next_crypto.as_ref(),
)
.ok()?;
Some(packet.payload.to_vec())
}
/// The number of bytes of packets containing retransmittable frames that have not been
/// acknowledged or declared lost.
#[cfg(test)]
pub(crate) fn bytes_in_flight(&self) -> u64 {
self.path.in_flight.bytes
}
/// Number of bytes worth of non-ack-only packets that may be sent
#[cfg(test)]
pub(crate) fn congestion_window(&self) -> u64 {
self.path
.congestion
.window()
.saturating_sub(self.path.in_flight.bytes)
}
/// Whether no timers but keepalive, idle, rtt and pushnewcid are running
#[cfg(test)]
pub(crate) fn is_idle(&self) -> bool {
Timer::VALUES
.iter()
.filter(|&&t| t != Timer::KeepAlive && t != Timer::PushNewCid)
.filter_map(|&t| Some((t, self.timers.get(t)?)))
.min_by_key(|&(_, time)| time)
.map_or(true, |(timer, _)| timer == Timer::Idle)
}
/// Total number of outgoing packets that have been deemed lost
#[cfg(test)]
pub(crate) fn lost_packets(&self) -> u64 {
self.lost_packets
}
/// Whether explicit congestion notification is in use on outgoing packets.
#[cfg(test)]
pub(crate) fn using_ecn(&self) -> bool {
self.path.sending_ecn
}
/// The number of received bytes in the current path
#[cfg(test)]
pub(crate) fn total_recvd(&self) -> u64 {
self.path.total_recvd
}
#[cfg(test)]
pub(crate) fn active_local_cid_seq(&self) -> (u64, u64) {
self.local_cid_state.active_seq()
}
/// Instruct the peer to replace previously issued CIDs by sending a NEW_CONNECTION_ID frame
/// with updated `retire_prior_to` field set to `v`
#[cfg(test)]
pub(crate) fn rotate_local_cid(&mut self, v: u64, now: Instant) {
let n = self.local_cid_state.assign_retire_seq(v);
self.endpoint_events
.push_back(EndpointEventInner::NeedIdentifiers(now, n));
}
/// Check the current active remote CID sequence
#[cfg(test)]
pub(crate) fn active_rem_cid_seq(&self) -> u64 {
self.rem_cids.active_seq()
}
/// Returns the detected maximum udp payload size for the current path
#[cfg(test)]
pub(crate) fn path_mtu(&self) -> u16 {
self.path.current_mtu()
}
/// Whether we have 1-RTT data to send
///
/// See also `self.space(SpaceId::Data).can_send()`
fn can_send_1rtt(&self, max_size: usize) -> bool {
self.streams.can_send_stream_data()
|| self.path.challenge_pending
|| self
.prev_path
.as_ref()
.map_or(false, |(_, x)| x.challenge_pending)
|| !self.path_responses.is_empty()
|| self
.datagrams
.outgoing
.front()
.map_or(false, |x| x.size(true) <= max_size)
}
/// Update counters to account for a packet becoming acknowledged, lost, or abandoned
fn remove_in_flight(&mut self, pn: u64, packet: &SentPacket) {
// Visit known paths from newest to oldest to find the one `pn` was sent on
for path in [&mut self.path]
.into_iter()
.chain(self.prev_path.as_mut().map(|(_, data)| data))
{
if path.remove_in_flight(pn, packet) {
return;
}
}
}
/// Terminate the connection instantly, without sending a close packet
fn kill(&mut self, reason: ConnectionError) {
self.close_common();
self.error = Some(reason);
self.state = State::Drained;
self.endpoint_events.push_back(EndpointEventInner::Drained);
}
/// Storage size required for the largest packet known to be supported by the current path
///
/// Buffers passed to [`Connection::poll_transmit`] should be at least this large.
pub fn current_mtu(&self) -> u16 {
self.path.current_mtu()
}
/// Size of non-frame data for a 1-RTT packet
///
/// Quantifies space consumed by the QUIC header and AEAD tag. All other bytes in a packet are
/// frames. Changes if the length of the remote connection ID changes, which is expected to be
/// rare. If `pn` is specified, may additionally change unpredictably due to variations in
/// latency and packet loss.
fn predict_1rtt_overhead(&self, pn: Option<u64>) -> usize {
let pn_len = match pn {
Some(pn) => PacketNumber::new(
pn,
self.spaces[SpaceId::Data].largest_acked_packet.unwrap_or(0),
)
.len(),
// Upper bound
None => 4,
};
// 1 byte for flags
1 + self.rem_cids.active().len() + pn_len + self.tag_len_1rtt()
}
fn tag_len_1rtt(&self) -> usize {
let key = match self.spaces[SpaceId::Data].crypto.as_ref() {
Some(crypto) => Some(&*crypto.packet.local),
None => self.zero_rtt_crypto.as_ref().map(|x| &*x.packet),
};
// If neither Data nor 0-RTT keys are available, make a reasonable tag length guess. As of
// this writing, all QUIC cipher suites use 16-byte tags. We could return `None` instead,
// but that would needlessly prevent sending datagrams during 0-RTT.
key.map_or(16, |x| x.tag_len())
}
}
impl fmt::Debug for Connection {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
f.debug_struct("Connection")
.field("handshake_cid", &self.handshake_cid)
.finish()
}
}
/// Reasons why a connection might be lost
#[derive(Debug, Error, Clone, PartialEq, Eq)]
pub enum ConnectionError {
/// The peer doesn't implement any supported version
#[error("peer doesn't implement any supported version")]
VersionMismatch,
/// The peer violated the QUIC specification as understood by this implementation
#[error(transparent)]
TransportError(#[from] TransportError),
/// The peer's QUIC stack aborted the connection automatically
#[error("aborted by peer: {0}")]
ConnectionClosed(frame::ConnectionClose),
/// The peer closed the connection
#[error("closed by peer: {0}")]
ApplicationClosed(frame::ApplicationClose),
/// The peer is unable to continue processing this connection, usually due to having restarted
#[error("reset by peer")]
Reset,
/// Communication with the peer has lapsed for longer than the negotiated idle timeout
///
/// If neither side is sending keep-alives, a connection will time out after a long enough idle
/// period even if the peer is still reachable. See also [`TransportConfig::max_idle_timeout()`]
/// and [`TransportConfig::keep_alive_interval()`].
#[error("timed out")]
TimedOut,
/// The local application closed the connection
#[error("closed")]
LocallyClosed,
/// The connection could not be created because not enough of the CID space is available
///
/// Try using longer connection IDs.
#[error("CIDs exhausted")]
CidsExhausted,
}
impl From<Close> for ConnectionError {
fn from(x: Close) -> Self {
match x {
Close::Connection(reason) => Self::ConnectionClosed(reason),
Close::Application(reason) => Self::ApplicationClosed(reason),
}
}
}
// For compatibility with API consumers
impl From<ConnectionError> for io::Error {
fn from(x: ConnectionError) -> Self {
use self::ConnectionError::*;
let kind = match x {
TimedOut => io::ErrorKind::TimedOut,
Reset => io::ErrorKind::ConnectionReset,
ApplicationClosed(_) | ConnectionClosed(_) => io::ErrorKind::ConnectionAborted,
TransportError(_) | VersionMismatch | LocallyClosed | CidsExhausted => {
io::ErrorKind::Other
}
};
Self::new(kind, x)
}
}
#[allow(unreachable_pub)] // fuzzing only
#[derive(Clone)]
pub enum State {
Handshake(state::Handshake),
Established,
Closed(state::Closed),
Draining,
/// Waiting for application to call close so we can dispose of the resources
Drained,
}
impl State {
fn closed<R: Into<Close>>(reason: R) -> Self {
Self::Closed(state::Closed {
reason: reason.into(),
})
}
fn is_handshake(&self) -> bool {
matches!(*self, Self::Handshake(_))
}
fn is_established(&self) -> bool {
matches!(*self, Self::Established)
}
fn is_closed(&self) -> bool {
matches!(*self, Self::Closed(_) | Self::Draining | Self::Drained)
}
fn is_drained(&self) -> bool {
matches!(*self, Self::Drained)
}
}
mod state {
use super::*;
#[allow(unreachable_pub)] // fuzzing only
#[derive(Clone)]
pub struct Handshake {
/// Whether the remote CID has been set by the peer yet
///
/// Always set for servers
pub(super) rem_cid_set: bool,
/// Stateless retry token received in the first Initial by a server.
///
/// Must be present in every Initial. Always empty for clients.
pub(super) expected_token: Bytes,
/// First cryptographic message
///
/// Only set for clients
pub(super) client_hello: Option<Bytes>,
}
#[allow(unreachable_pub)] // fuzzing only
#[derive(Clone)]
pub struct Closed {
pub(super) reason: Close,
}
}
/// Events of interest to the application
#[derive(Debug)]
pub enum Event {
/// The connection's handshake data is ready
HandshakeDataReady,
/// The connection was successfully established
Connected,
/// The connection was lost
///
/// Emitted if the peer closes the connection or an error is encountered.
ConnectionLost {
/// Reason that the connection was closed
reason: ConnectionError,
},
/// Stream events
Stream(StreamEvent),
/// One or more application datagrams have been received
DatagramReceived,
/// One or more application datagrams have been sent after blocking
DatagramsUnblocked,
}
fn instant_saturating_sub(x: Instant, y: Instant) -> Duration {
if x > y {
x - y
} else {
Duration::new(0, 0)
}
}
fn get_max_ack_delay(params: &TransportParameters) -> Duration {
Duration::from_micros(params.max_ack_delay.0 * 1000)
}
// Prevents overflow and improves behavior in extreme circumstances
const MAX_BACKOFF_EXPONENT: u32 = 16;
// Minimal remaining size to allow packet coalescing
const MIN_PACKET_SPACE: usize = 40;
/// The maximum amount of datagrams that are sent in a single transmit
///
/// This can be lower than the maximum platform capabilities, to avoid excessive
/// memory allocations when calling `poll_transmit()`. Benchmarks have shown
/// that numbers around 10 are a good compromise.
const MAX_TRANSMIT_SEGMENTS: usize = 10;
/// Perform key updates this many packets before the AEAD confidentiality limit.
///
/// Chosen arbitrarily, intended to be large enough to prevent spurious connection loss.
const KEY_UPDATE_MARGIN: u64 = 10_000;
#[derive(Default)]
struct SentFrames {
retransmits: ThinRetransmits,
largest_acked: Option<u64>,
stream_frames: StreamMetaVec,
/// Whether the packet contains non-retransmittable frames (like datagrams)
non_retransmits: bool,
requires_padding: bool,
}
impl SentFrames {
/// Returns whether the packet contains only ACKs
fn is_ack_only(&self, streams: &StreamsState) -> bool {
self.largest_acked.is_some()
&& !self.non_retransmits
&& self.stream_frames.is_empty()
&& self.retransmits.is_empty(streams)
}
}