shadow_rs/host/process.rs
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
//! An emulated Linux process.
use std::cell::{Cell, Ref, RefCell, RefMut};
use std::collections::BTreeMap;
use std::ffi::{c_char, c_void, CStr, CString};
use std::fmt::Write;
use std::num::TryFromIntError;
use std::ops::{Deref, DerefMut};
use std::os::fd::AsRawFd;
use std::path::{Path, PathBuf};
use std::sync::atomic::Ordering;
use std::sync::Arc;
#[cfg(feature = "perf_timers")]
use std::time::Duration;
use linux_api::errno::Errno;
use linux_api::fcntl::OFlag;
use linux_api::posix_types::Pid;
use linux_api::sched::{CloneFlags, SuidDump};
use linux_api::signal::{
defaultaction, siginfo_t, sigset_t, LinuxDefaultAction, SigActionFlags, Signal,
SignalFromI32Error,
};
use log::{debug, trace, warn};
use rustix::process::{WaitOptions, WaitStatus};
use shadow_shim_helper_rs::explicit_drop::{ExplicitDrop, ExplicitDropper};
use shadow_shim_helper_rs::rootedcell::rc::RootedRc;
use shadow_shim_helper_rs::rootedcell::refcell::RootedRefCell;
use shadow_shim_helper_rs::rootedcell::Root;
use shadow_shim_helper_rs::shim_shmem::ProcessShmem;
use shadow_shim_helper_rs::simulation_time::SimulationTime;
use shadow_shim_helper_rs::syscall_types::{ForeignPtr, ManagedPhysicalMemoryAddr};
use shadow_shim_helper_rs::HostId;
use shadow_shmem::allocator::ShMemBlock;
use super::descriptor::descriptor_table::{DescriptorHandle, DescriptorTable};
use super::descriptor::listener::StateEventSource;
use super::descriptor::{FileSignals, FileState};
use super::host::Host;
use super::memory_manager::{MemoryManager, ProcessMemoryRef, ProcessMemoryRefMut};
use super::syscall::formatter::StraceFmtMode;
use super::syscall::types::ForeignArrayPtr;
use super::thread::{Thread, ThreadId};
use super::timer::Timer;
use crate::core::configuration::{ProcessFinalState, RunningVal};
use crate::core::work::task::TaskRef;
use crate::core::worker::Worker;
use crate::cshadow;
use crate::host::context::ProcessContext;
use crate::host::descriptor::Descriptor;
use crate::host::managed_thread::ManagedThread;
use crate::host::syscall::formatter::FmtOptions;
use crate::utility::callback_queue::CallbackQueue;
#[cfg(feature = "perf_timers")]
use crate::utility::perf_timer::PerfTimer;
use crate::utility::{self, debug_assert_cloexec};
/// Virtual pid of a shadow process
#[derive(Debug, PartialEq, Eq, Hash, Copy, Clone, Ord, PartialOrd)]
pub struct ProcessId(u32);
impl ProcessId {
// The first Process to run after boot is the "init" process, and has pid=1.
// In Shadow simulations, this roughly corresponds to Shadow itself. e.g.
// processes spawned by Shadow itself have a parent pid of 1.
pub const INIT: Self = ProcessId(1);
/// Returns what the `ProcessId` would be of a `Process` whose thread
/// group leader has id `thread_group_leader_tid`.
pub fn from_thread_group_leader_tid(thread_group_leader_tid: ThreadId) -> Self {
ProcessId::try_from(libc::pid_t::from(thread_group_leader_tid)).unwrap()
}
}
impl std::fmt::Display for ProcessId {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
write!(f, "{}", self.0)
}
}
impl TryFrom<u32> for ProcessId {
type Error = TryFromIntError;
fn try_from(val: u32) -> Result<Self, Self::Error> {
// we don't actually want the value as a `pid_t`, we just want to make sure it can be
// converted successfully
let _ = libc::pid_t::try_from(val)?;
Ok(ProcessId(val))
}
}
impl TryFrom<libc::pid_t> for ProcessId {
type Error = TryFromIntError;
fn try_from(value: libc::pid_t) -> Result<Self, Self::Error> {
Ok(ProcessId(value.try_into()?))
}
}
impl From<ProcessId> for u32 {
fn from(val: ProcessId) -> Self {
val.0
}
}
impl From<ProcessId> for libc::pid_t {
fn from(val: ProcessId) -> Self {
val.0.try_into().unwrap()
}
}
impl From<ThreadId> for ProcessId {
fn from(value: ThreadId) -> Self {
ProcessId::try_from(libc::pid_t::from(value)).unwrap()
}
}
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub enum ExitStatus {
Normal(i32),
Signaled(Signal),
/// The process was killed by Shadow rather than exiting "naturally" as part
/// of the simulation. Currently this only happens when the process is still
/// running when the simulation stop_time is reached.
///
/// A signal delivered via `shutdown_signal` does not result in this status;
/// e.g. if the process is killed directly by the signal the ExitStatus will
/// be `Signaled`; if the process handles the signal and exits by calling
/// `exit`, the status will be `Normal`.
StoppedByShadow,
}
#[derive(Debug)]
struct StraceLogging {
file: RootedRefCell<std::fs::File>,
options: FmtOptions,
}
/// Parts of the process that are present in all states.
struct Common {
id: ProcessId,
host_id: HostId,
// Parent pid (aka `ppid`), as returned e.g. by `getppid`. This can change
// at runtime if the original parent exits and is reaped.
parent_pid: Cell<ProcessId>,
// Process group id (aka `pgid`), as returned e.g. by `getpgid`.
group_id: Cell<ProcessId>,
// Session id, as returned e.g. by `getsid`.
session_id: Cell<ProcessId>,
// Signal to send to parent on death.
exit_signal: Option<Signal>,
// unique id of the program that this process should run
name: CString,
// the name of the executable as provided in shadow's config, for logging purposes
plugin_name: CString,
// absolute path to the process's working directory.
// This must remain in sync with the actual working dir of the native process.
// See https://github.com/shadow/shadow/issues/2960
working_dir: CString,
}
impl Common {
fn id(&self) -> ProcessId {
self.id
}
fn physical_address(&self, vptr: ForeignPtr<()>) -> ManagedPhysicalMemoryAddr {
// We currently don't keep a true system-wide virtual <-> physical address
// mapping. Instead we simply assume that no shadow processes map the same
// underlying physical memory, and that therefore (pid, virtual address)
// uniquely defines a physical address.
//
// If we ever want to support futexes in memory shared between processes,
// we'll need to change this. The most foolproof way to do so is probably
// to change ManagedPhysicalMemoryAddr to be a bigger struct that identifies where
// the mapped region came from (e.g. what file), and the offset into that
// region. Such "fat" physical pointers might make memory management a
// little more cumbersome though, e.g. when using them as keys in the futex
// table.
//
// Alternatively we could hash the region+offset to a 64-bit value, but
// then we'd need to deal with potential collisions. On average we'd expect
// a collision after 2**32 physical addresses; i.e. they *probably*
// wouldn't happen in practice for realistic simulations.
// Linux uses the bottom 48-bits for user-space virtual addresses, giving
// us 16 bits for the pid.
const PADDR_BITS: i32 = 64;
const VADDR_BITS: i32 = 48;
const PID_BITS: i32 = 16;
assert_eq!(PADDR_BITS, PID_BITS + VADDR_BITS);
let high_part: u64 = u64::from(u32::from(self.id())) << VADDR_BITS;
assert_eq!(
ProcessId::try_from((high_part >> VADDR_BITS) as u32),
Ok(self.id())
);
let low_part = u64::from(vptr);
assert_eq!(low_part >> VADDR_BITS, 0);
ManagedPhysicalMemoryAddr::from(high_part | low_part)
}
fn name(&self) -> &str {
self.name.to_str().unwrap()
}
pub fn thread_group_leader_id(&self) -> ThreadId {
// tid of the thread group leader is equal to the pid.
ThreadId::from(self.id())
}
}
/// A process that is currently runnable.
pub struct RunnableProcess {
common: Common,
// Expected end state, if any. We'll report an error if this is present and
// doesn't match the actual exit status.
//
// This will be None e.g. for processes created via `fork` instead of
// spawned directly from Shadow's config file. In those cases it's the
// parent's responsibility to reap and interpret the exit status.
expected_final_state: Option<ProcessFinalState>,
// Shared memory allocation for shared state with shim.
shim_shared_mem_block: ShMemBlock<'static, ProcessShmem>,
// Shared with forked Processes
strace_logging: Option<Arc<StraceLogging>>,
// The shim's log file. This gets dup'd into the ManagedProcess
// where the shim can write to it directly. We persist it to handle the case
// where we need to recreatea a ManagedProcess and have it continue writing
// to the same file.
//
// Shared with forked Processes
shimlog_file: Arc<std::fs::File>,
// "dumpable" state, as manipulated via the prctl operations PR_SET_DUMPABLE
// and PR_GET_DUMPABLE.
dumpable: Cell<SuidDump>,
native_pid: Pid,
// timer that tracks the amount of CPU time we spend on plugin execution and processing
#[cfg(feature = "perf_timers")]
cpu_delay_timer: RefCell<PerfTimer>,
#[cfg(feature = "perf_timers")]
total_run_time: Cell<Duration>,
itimer_real: RefCell<Timer>,
// The `RootedRc` lets us hold a reference to a thread without holding a
// reference to the thread list. e.g. this lets us implement the `clone`
// syscall, which adds a thread to the list while we have a reference to the
// parent thread.
threads: RefCell<BTreeMap<ThreadId, RootedRc<RootedRefCell<Thread>>>>,
// References to `Self::memory_manager` cached on behalf of C code using legacy
// C memory access APIs.
// TODO: Remove these when we've migrated Shadow off of the APIs that need
// them (probably by migrating all the calling code to Rust).
//
// SAFETY: Must be before memory_manager for drop order.
unsafe_borrow_mut: RefCell<Option<UnsafeBorrowMut>>,
unsafe_borrows: RefCell<Vec<UnsafeBorrow>>,
// `clone(2)` documents that if `CLONE_THREAD` is set, then `CLONE_VM` must
// also be set. Hence all threads in a process always share the same virtual
// address space, and hence we have a `MemoryManager` at the `Process` level
// rather than the `Thread` level.
// SAFETY: Must come after `unsafe_borrows` and `unsafe_borrow_mut`.
// Boxed to avoid invalidating those if Self is moved.
memory_manager: Box<RefCell<MemoryManager>>,
// Listeners for child-events.
// e.g. these listeners are notified when a child of this process exits.
child_process_event_listeners: RefCell<StateEventSource>,
}
impl RunnableProcess {
/// Spawn a `ManagedThread` corresponding to the given `exec` syscall
/// parameters. Intended for use by the `exec` syscall handlers. Whether it
/// succeeds or fails, does *not* mutate `self`, though `self`'s strace and
/// shim log files will be passed into the new `ManagedThread`.
///
/// In case the native `exec` syscall fails, the corresponding error is returned.
pub fn spawn_mthread_for_exec(
&self,
host: &Host,
plugin_path: &CStr,
argv: Vec<CString>,
envv: Vec<CString>,
) -> Result<ManagedThread, Errno> {
ManagedThread::spawn(
plugin_path,
argv,
envv,
self.strace_logging
.as_ref()
.map(|s| s.file.borrow(host.root()))
.as_deref(),
&self.shimlog_file,
host.preload_paths(),
)
}
/// Call after a thread has exited. Removes the thread and does corresponding cleanup and notifications.
fn reap_thread(&self, host: &Host, threadrc: RootedRc<RootedRefCell<Thread>>) {
let threadrc = ExplicitDropper::new(threadrc, |t| {
t.explicit_drop_recursive(host.root(), host);
});
let thread = threadrc.borrow(host.root());
assert!(!thread.is_running());
// If the `clear_child_tid` attribute on the thread is set, and there are
// any other threads left alive in the process, perform a futex wake on
// that address. This mechanism is typically used in `pthread_join` etc.
// See `set_tid_address(2)`.
let clear_child_tid_pvp = thread.get_tid_address();
if !clear_child_tid_pvp.is_null() && self.threads.borrow().len() > 0 {
self.memory_manager
.borrow_mut()
.write(clear_child_tid_pvp, &0)
.unwrap();
// Wake the corresponding futex.
let futexes = host.futextable_borrow();
let addr = self
.common
.physical_address(clear_child_tid_pvp.cast::<()>());
if let Some(futex) = futexes.get(addr) {
futex.wake(1);
}
}
}
/// This cleans up memory references left over from legacy C code; usually
/// a syscall handler.
///
/// Writes the leftover mutable ref to memory (if any), and frees
/// all memory refs.
pub fn free_unsafe_borrows_flush(&self) -> Result<(), Errno> {
self.unsafe_borrows.borrow_mut().clear();
let unsafe_borrow_mut = self.unsafe_borrow_mut.borrow_mut().take();
if let Some(borrow) = unsafe_borrow_mut {
borrow.flush()
} else {
Ok(())
}
}
/// This cleans up memory references left over from legacy C code; usually
/// a syscall handler.
///
/// Frees all memory refs without writing back to memory.
pub fn free_unsafe_borrows_noflush(&self) {
self.unsafe_borrows.borrow_mut().clear();
let unsafe_borrow_mut = self.unsafe_borrow_mut.borrow_mut().take();
if let Some(borrow) = unsafe_borrow_mut {
borrow.noflush();
}
}
#[track_caller]
pub fn memory_borrow(&self) -> impl Deref<Target = MemoryManager> + '_ {
self.memory_manager.borrow()
}
#[track_caller]
pub fn memory_borrow_mut(&self) -> impl DerefMut<Target = MemoryManager> + '_ {
self.memory_manager.borrow_mut()
}
pub fn strace_logging_options(&self) -> Option<FmtOptions> {
self.strace_logging.as_ref().map(|x| x.options)
}
/// If strace logging is disabled, this function will do nothing and return `None`.
pub fn with_strace_file<T>(&self, f: impl FnOnce(&mut std::fs::File) -> T) -> Option<T> {
// TODO: get Host from caller. Would need t update syscall-logger.
Worker::with_active_host(|host| {
let strace_logging = self.strace_logging.as_ref()?;
let mut file = strace_logging.file.borrow_mut(host.root());
Some(f(&mut file))
})
.unwrap()
}
pub fn native_pid(&self) -> Pid {
self.native_pid
}
#[track_caller]
fn first_live_thread(&self, root: &Root) -> Option<Ref<RootedRc<RootedRefCell<Thread>>>> {
Ref::filter_map(self.threads.borrow(), |threads| {
threads.values().next().inspect(|thread| {
// There shouldn't be any non-running threads in the table.
assert!(thread.borrow(root).is_running());
})
})
.ok()
}
/// Returns a dynamically borrowed reference to the first live thread.
/// This is meant primarily for the MemoryManager.
#[track_caller]
pub fn first_live_thread_borrow(
&self,
root: &Root,
) -> Option<impl Deref<Target = RootedRc<RootedRefCell<Thread>>> + '_> {
self.first_live_thread(root)
}
#[track_caller]
fn thread(&self, virtual_tid: ThreadId) -> Option<Ref<RootedRc<RootedRefCell<Thread>>>> {
Ref::filter_map(self.threads.borrow(), |threads| threads.get(&virtual_tid)).ok()
}
#[track_caller]
pub fn thread_borrow(
&self,
virtual_tid: ThreadId,
) -> Option<impl Deref<Target = RootedRc<RootedRefCell<Thread>>> + '_> {
self.thread(virtual_tid)
}
// Disposes of `self`, returning the internal `Common` for reuse.
// Used internally when changing states.
fn into_common(self) -> Common {
// There shouldn't be any outstanding unsafe borrows when changing
// states, since that would indicate C code might still have a pointer
// to memory.
assert!(self.unsafe_borrow_mut.take().is_none());
assert!(self.unsafe_borrows.take().is_empty());
self.common
}
/// Starts the CPU delay timer.
/// Panics if the timer is already running.
#[cfg(feature = "perf_timers")]
pub fn start_cpu_delay_timer(&self) {
self.cpu_delay_timer.borrow_mut().start()
}
/// Stop the timer and return the most recent (not cumulative) duration.
/// Panics if the timer was not already running.
#[cfg(feature = "perf_timers")]
pub fn stop_cpu_delay_timer(&self, host: &Host) -> Duration {
let mut timer = self.cpu_delay_timer.borrow_mut();
timer.stop();
let total_elapsed = timer.elapsed();
let prev_total = self.total_run_time.replace(total_elapsed);
let delta = total_elapsed - prev_total;
host.cpu_borrow_mut().add_delay(delta);
delta
}
fn interrupt_with_signal(&self, host: &Host, signal: Signal) {
let threads = self.threads.borrow();
for thread in threads.values() {
let thread = thread.borrow(host.root());
{
let thread_shmem = thread.shmem();
let host_lock = host.shim_shmem_lock_borrow().unwrap();
let thread_shmem_protected = thread_shmem.protected.borrow(&host_lock.root);
let blocked_signals = thread_shmem_protected.blocked_signals;
if blocked_signals.has(signal) {
continue;
}
}
let Some(mut cond) = thread.syscall_condition_mut() else {
// Defensively handle this gracefully, but it probably shouldn't happen.
// The only thread in the process not blocked on a syscall should be
// the current-running thread (if any), but the caller should have
// delivered the signal synchronously instead of using this function
// in that case.
warn!("thread {:?} has no syscall_condition. How?", thread.id());
continue;
};
cond.wakeup_for_signal(host, signal);
break;
}
}
/// Send the signal described in `siginfo` to `process`. `current_thread`
/// should be set if there is one (e.g. if this is being called from a syscall
/// handler), and `None` otherwise (e.g. when called from a timer expiration event).
///
/// An event will be scheduled to deliver the signal unless `current_thread`
/// is set, and belongs to the process `self`, and doesn't have the signal
/// blocked. In that the signal will be processed synchronously when
/// returning from the current syscall.
pub fn signal(&self, host: &Host, current_thread: Option<&Thread>, siginfo_t: &siginfo_t) {
let signal = match siginfo_t.signal() {
Ok(s) => s,
Err(SignalFromI32Error(0)) => return,
Err(SignalFromI32Error(n)) => panic!("Bad signo {n}"),
};
// Scope for `process_shmem_protected`
{
let host_shmem = host.shim_shmem_lock_borrow().unwrap();
let mut process_shmem_protected = self
.shim_shared_mem_block
.protected
.borrow_mut(&host_shmem.root);
// SAFETY: We don't try to call any of the function pointers.
let action = unsafe { process_shmem_protected.signal_action(signal) };
match unsafe { action.handler() } {
linux_api::signal::SignalHandler::Handler(_) => (),
linux_api::signal::SignalHandler::Action(_) => (),
linux_api::signal::SignalHandler::SigIgn => return,
linux_api::signal::SignalHandler::SigDfl => {
if defaultaction(signal) == LinuxDefaultAction::IGN {
return;
}
}
}
if process_shmem_protected.pending_signals.has(signal) {
// Signal is already pending. From signal(7):In the case where a
// standard signal is already pending, the siginfo_t structure (see
// sigaction(2)) associated with that signal is not overwritten on
// arrival of subsequent instances of the same signal.
return;
}
process_shmem_protected.pending_signals.add(signal);
process_shmem_protected.set_pending_standard_siginfo(signal, siginfo_t);
}
if let Some(thread) = current_thread {
if thread.process_id() == self.common.id() {
let host_shmem = host.shim_shmem_lock_borrow().unwrap();
let threadmem = thread.shmem();
let threadprotmem = threadmem.protected.borrow(&host_shmem.root);
if !threadprotmem.blocked_signals.has(signal) {
// Target process is this process, and current thread hasn't blocked
// the signal. It will be delivered to this thread when it resumes.
return;
}
}
}
self.interrupt_with_signal(host, signal);
}
/// Adds a new thread to the process and schedules it to run.
/// Intended for use by `clone`.
pub fn add_thread(&self, host: &Host, thread: RootedRc<RootedRefCell<Thread>>) {
let pid = self.common.id();
let tid = thread.borrow(host.root()).id();
self.threads.borrow_mut().insert(tid, thread);
// Schedule thread to start. We're giving the caller's reference to thread
// to the TaskRef here, which is why we don't increment its ref count to
// create the TaskRef, but do decrement it on cleanup.
let task = TaskRef::new(move |host| {
host.resume(pid, tid);
});
host.schedule_task_with_delay(task, SimulationTime::ZERO);
}
/// Create a new `Process`, forked from `self`, with the thread `new_thread_group_leader`.
pub fn new_forked_process(
&self,
host: &Host,
flags: CloneFlags,
exit_signal: Option<Signal>,
new_thread_group_leader: RootedRc<RootedRefCell<Thread>>,
) -> RootedRc<RootedRefCell<Process>> {
let new_tgl_tid;
let native_pid;
{
let new_tgl = new_thread_group_leader.borrow(host.root());
new_tgl_tid = new_tgl.id();
native_pid = new_tgl.native_pid();
}
let pid = ProcessId::from_thread_group_leader_tid(new_tgl_tid);
assert_eq!(
pid,
new_thread_group_leader.borrow(host.root()).process_id()
);
let plugin_name = self.common.plugin_name.clone();
let name = make_name(host, plugin_name.to_str().unwrap(), pid);
let parent_pid = if flags.contains(CloneFlags::CLONE_PARENT) {
self.common.parent_pid.get()
} else {
self.common.id
};
// Process group is always inherited from the parent process.
let process_group_id = self.common.group_id.get();
// Session is always inherited from the parent process.
let session_id = self.common.session_id.get();
let common = Common {
id: pid,
host_id: host.id(),
name,
plugin_name,
working_dir: self.common.working_dir.clone(),
parent_pid: Cell::new(parent_pid),
group_id: Cell::new(process_group_id),
session_id: Cell::new(session_id),
exit_signal,
};
// The child will log to the same strace log file. Entries contain thread IDs,
// though it might be tricky to map those back to processes.
let strace_logging = self.strace_logging.as_ref().cloned();
// `fork(2)`:
// > The child does not inherit timers from its parent
// > (setitimer(2), alarm(2), timer_create(2)).
let itimer_real = RefCell::new(Timer::new(move |host| itimer_real_expiration(host, pid)));
let threads = RefCell::new(BTreeMap::from([(new_tgl_tid, new_thread_group_leader)]));
let shim_shared_mem = ProcessShmem::new(
&host.shim_shmem_lock_borrow().unwrap().root,
host.shim_shmem().serialize(),
host.id(),
strace_logging
.as_ref()
.map(|x| x.file.borrow(host.root()).as_raw_fd()),
);
let shim_shared_mem_block = shadow_shmem::allocator::shmalloc(shim_shared_mem);
let runnable_process = RunnableProcess {
common,
expected_final_state: None,
shim_shared_mem_block,
strace_logging,
dumpable: self.dumpable.clone(),
native_pid,
#[cfg(feature = "perf_timers")]
cpu_delay_timer: RefCell::new(PerfTimer::new()),
#[cfg(feature = "perf_timers")]
total_run_time: Cell::new(Duration::ZERO),
itimer_real,
threads,
unsafe_borrow_mut: RefCell::new(None),
unsafe_borrows: RefCell::new(Vec::new()),
memory_manager: Box::new(RefCell::new(unsafe { MemoryManager::new(native_pid) })),
child_process_event_listeners: Default::default(),
shimlog_file: self.shimlog_file.clone(),
};
let child_process = Process {
state: RefCell::new(Some(ProcessState::Runnable(runnable_process))),
};
RootedRc::new(host.root(), RootedRefCell::new(host.root(), child_process))
}
/// Shared memory for this process.
pub fn shmem(&self) -> impl Deref<Target = ShMemBlock<'static, ProcessShmem>> + '_ {
&self.shim_shared_mem_block
}
}
impl ExplicitDrop for RunnableProcess {
type ExplicitDropParam = Host;
type ExplicitDropResult = ();
fn explicit_drop(mut self, host: &Self::ExplicitDropParam) -> Self::ExplicitDropResult {
let threads = std::mem::take(self.threads.get_mut());
for thread in threads.into_values() {
thread.explicit_drop_recursive(host.root(), host);
}
}
}
/// A process that has exited.
pub struct ZombieProcess {
common: Common,
exit_status: ExitStatus,
}
impl ZombieProcess {
pub fn exit_status(&self) -> ExitStatus {
self.exit_status
}
/// Process that can reap this zombie process, if any.
pub fn reaper<'host>(
&self,
host: &'host Host,
) -> Option<impl Deref<Target = RootedRc<RootedRefCell<Process>>> + 'host> {
let parent_pid = self.common.parent_pid.get();
if parent_pid == ProcessId::INIT {
return None;
}
let parentrc = host.process_borrow(parent_pid)?;
// If the parent has *explicitly* ignored the exit signal, then it
// doesn't reap.
//
// `waitpid(2)`:
// > POSIX.1-2001 specifies that if the disposition of SIGCHLD is set to SIG_IGN or the SA_NOCLDWAIT flag is set for SIGCHLD (see
// > sigaction(2)), then children that terminate do not become zombies and a call to wait() or waitpid() will block until all
// > children have terminated, and then fail with errno set to ECHILD. (The original POSIX standard left the behavior of setting
// > SIGCHLD to SIG_IGN unspecified. Note that even though the default disposition of SIGCHLD is "ignore", explicitly setting the
// > disposition to SIG_IGN results in different treatment of zombie process children.)
//
// TODO: validate that this applies to whatever signal is configured as the exit
// signal, even if it's not SIGCHLD.
if let Some(exit_signal) = self.common.exit_signal {
let parent = parentrc.borrow(host.root());
let parent_shmem = parent.shmem();
let host_shmem_lock = host.shim_shmem_lock_borrow().unwrap();
let parent_shmem_protected = parent_shmem.protected.borrow(&host_shmem_lock.root);
// SAFETY: We don't dereference function pointers.
let action = unsafe { parent_shmem_protected.signal_action(exit_signal) };
if action.is_ignore() {
return None;
}
}
Some(parentrc)
}
fn notify_parent_of_exit(&self, host: &Host) {
let Some(exit_signal) = self.common.exit_signal else {
trace!("Not notifying parent of exit: no signal specified");
return;
};
let parent_pid = self.common.parent_pid.get();
if parent_pid == ProcessId::INIT {
trace!("Not notifying parent of exit: parent is 'init'");
return;
}
let Some(parent_rc) = host.process_borrow(parent_pid) else {
trace!("Not notifying parent of exit: parent {parent_pid:?} not found");
return;
};
let parent = parent_rc.borrow(host.root());
let siginfo = self.exit_siginfo(exit_signal);
let Some(parent_runnable) = parent.as_runnable() else {
trace!("Not notifying parent of exit: {parent_pid:?} not running");
debug_panic!("Non-running parent process shouldn't be possible.");
#[allow(unreachable_code)]
{
return;
}
};
parent_runnable.signal(host, None, &siginfo);
CallbackQueue::queue_and_run_with_legacy(|q| {
let mut parent_child_listeners =
parent_runnable.child_process_event_listeners.borrow_mut();
parent_child_listeners.notify_listeners(
FileState::CHILD_EVENT,
FileState::CHILD_EVENT,
FileSignals::empty(),
q,
);
});
}
/// Construct a siginfo containing information about how the process exited.
/// Used internally to send a signal to the parent process, and by the
/// `waitid` syscall handler.
///
/// `exit_signal` is the signal to set in the `siginfo_t`.
pub fn exit_siginfo(&self, exit_signal: Signal) -> siginfo_t {
match self.exit_status {
ExitStatus::Normal(exit_code) => siginfo_t::new_for_sigchld_exited(
exit_signal,
self.common.id.into(),
0,
exit_code,
0,
0,
),
ExitStatus::Signaled(fatal_signal) => {
// This ought to be `siginfo_t::new_for_sigchld_dumped` if
// the child dumped core, but that depends on various other
// system variables outside of our control. We always report
// that no core was dropped for determinism.
siginfo_t::new_for_sigchld_killed(
exit_signal,
self.common.id.into(),
0,
fatal_signal,
0,
0,
)
}
ExitStatus::StoppedByShadow => unreachable!(),
}
}
}
/// Inner implementation of a simulated process.
enum ProcessState {
Runnable(RunnableProcess),
Zombie(ZombieProcess),
}
impl ProcessState {
fn common(&self) -> &Common {
match self {
ProcessState::Runnable(r) => &r.common,
ProcessState::Zombie(z) => &z.common,
}
}
fn common_mut(&mut self) -> &mut Common {
match self {
ProcessState::Runnable(r) => &mut r.common,
ProcessState::Zombie(z) => &mut z.common,
}
}
fn as_runnable(&self) -> Option<&RunnableProcess> {
match self {
ProcessState::Runnable(r) => Some(r),
ProcessState::Zombie(_) => None,
}
}
fn as_runnable_mut(&mut self) -> Option<&mut RunnableProcess> {
match self {
ProcessState::Runnable(r) => Some(r),
ProcessState::Zombie(_) => None,
}
}
fn as_zombie(&self) -> Option<&ZombieProcess> {
match self {
ProcessState::Runnable(_) => None,
ProcessState::Zombie(z) => Some(z),
}
}
}
impl ExplicitDrop for ProcessState {
type ExplicitDropParam = Host;
type ExplicitDropResult = ();
fn explicit_drop(self, host: &Self::ExplicitDropParam) -> Self::ExplicitDropResult {
match self {
ProcessState::Runnable(r) => r.explicit_drop(host),
ProcessState::Zombie(_) => (),
}
}
}
/// A simulated process.
pub struct Process {
// Most of the implementation should be in [`ProcessState`].
// This wrapper allows us to change the state.
state: RefCell<Option<ProcessState>>,
}
fn itimer_real_expiration(host: &Host, pid: ProcessId) {
let Some(process) = host.process_borrow(pid) else {
debug!("Process {:?} no longer exists", pid);
return;
};
let process = process.borrow(host.root());
let Some(runnable) = process.as_runnable() else {
debug!("Process {:?} no longer running", &*process.name());
return;
};
let timer = runnable.itimer_real.borrow();
// The siginfo_t structure only has an i32. Presumably we want to just truncate in
// case of overflow.
let expiration_count = timer.expiration_count() as i32;
let siginfo_t = siginfo_t::new_for_timer(Signal::SIGALRM, 0, expiration_count);
process.signal(host, None, &siginfo_t);
}
impl Process {
fn common(&self) -> Ref<Common> {
Ref::map(self.state.borrow(), |state| {
state.as_ref().unwrap().common()
})
}
fn common_mut(&self) -> RefMut<Common> {
RefMut::map(self.state.borrow_mut(), |state| {
state.as_mut().unwrap().common_mut()
})
}
fn as_runnable(&self) -> Option<Ref<RunnableProcess>> {
Ref::filter_map(self.state.borrow(), |state| {
state.as_ref().unwrap().as_runnable()
})
.ok()
}
fn as_runnable_mut(&self) -> Option<RefMut<RunnableProcess>> {
RefMut::filter_map(self.state.borrow_mut(), |state| {
state.as_mut().unwrap().as_runnable_mut()
})
.ok()
}
/// Borrows a reference to the internal [`RunnableProcess`] if `self` is runnable.
pub fn borrow_as_runnable(&self) -> Option<impl Deref<Target = RunnableProcess> + '_> {
self.as_runnable()
}
fn as_zombie(&self) -> Option<Ref<ZombieProcess>> {
Ref::filter_map(self.state.borrow(), |state| {
state.as_ref().unwrap().as_zombie()
})
.ok()
}
/// Borrows a reference to the internal [`ZombieProcess`] if `self` is a zombie.
pub fn borrow_as_zombie(&self) -> Option<impl Deref<Target = ZombieProcess> + '_> {
self.as_zombie()
}
/// Spawn a new process. The process will be runnable via [`Self::resume`]
/// once it has been added to the `Host`'s process list.
pub fn spawn(
host: &Host,
plugin_name: CString,
plugin_path: &CStr,
argv: Vec<CString>,
envv: Vec<CString>,
pause_for_debugging: bool,
strace_logging_options: Option<FmtOptions>,
expected_final_state: ProcessFinalState,
) -> Result<RootedRc<RootedRefCell<Process>>, Errno> {
debug!("starting process '{:?}'", plugin_name);
let main_thread_id = host.get_new_thread_id();
let process_id = ProcessId::from(main_thread_id);
let desc_table = RootedRc::new(
host.root(),
RootedRefCell::new(host.root(), DescriptorTable::new()),
);
let itimer_real = RefCell::new(Timer::new(move |host| {
itimer_real_expiration(host, process_id)
}));
let name = make_name(host, plugin_name.to_str().unwrap(), process_id);
let mut file_basename = PathBuf::new();
file_basename.push(host.data_dir_path());
file_basename.push(format!(
"{exe_name}.{id}",
exe_name = plugin_name.to_str().unwrap(),
id = u32::from(process_id)
));
let strace_logging = strace_logging_options.map(|options| {
let file =
std::fs::File::create(Self::static_output_file_name(&file_basename, "strace"))
.unwrap();
debug_assert_cloexec(&file);
Arc::new(StraceLogging {
file: RootedRefCell::new(host.root(), file),
options,
})
});
let shim_shared_mem = ProcessShmem::new(
&host.shim_shmem_lock_borrow().unwrap().root,
host.shim_shmem().serialize(),
host.id(),
strace_logging
.as_ref()
.map(|x| x.file.borrow(host.root()).as_raw_fd()),
);
let shim_shared_mem_block = shadow_shmem::allocator::shmalloc(shim_shared_mem);
let working_dir = utility::pathbuf_to_nul_term_cstring(
std::fs::canonicalize(host.data_dir_path()).unwrap(),
);
#[cfg(feature = "perf_timers")]
let cpu_delay_timer = {
let mut t = PerfTimer::new();
t.stop();
RefCell::new(t)
};
// TODO: measure execution time of creating the main_thread with
// cpu_delay_timer? We previously did, but it's a little complex to do so,
// and it shouldn't matter much.
{
let mut descriptor_table = desc_table.borrow_mut(host.root());
Self::open_stdio_file_helper(
&mut descriptor_table,
libc::STDIN_FILENO.try_into().unwrap(),
"/dev/null".into(),
OFlag::O_RDONLY,
);
let name = Self::static_output_file_name(&file_basename, "stdout");
Self::open_stdio_file_helper(
&mut descriptor_table,
libc::STDOUT_FILENO.try_into().unwrap(),
name,
OFlag::O_WRONLY,
);
let name = Self::static_output_file_name(&file_basename, "stderr");
Self::open_stdio_file_helper(
&mut descriptor_table,
libc::STDERR_FILENO.try_into().unwrap(),
name,
OFlag::O_WRONLY,
);
}
let shimlog_file = Arc::new(
std::fs::File::create(Self::static_output_file_name(&file_basename, "shimlog"))
.unwrap(),
);
debug_assert_cloexec(&shimlog_file);
let mthread = ManagedThread::spawn(
plugin_path,
argv,
envv,
strace_logging
.as_ref()
.map(|s| s.file.borrow(host.root()))
.as_deref(),
&shimlog_file,
host.preload_paths(),
)?;
let native_pid = mthread.native_pid();
let main_thread =
Thread::wrap_mthread(host, mthread, desc_table, process_id, main_thread_id).unwrap();
debug!("process '{:?}' started", plugin_name);
if pause_for_debugging {
// will block until logger output has been flushed
// there is a race condition where other threads may log between the
// `eprintln` and `raise` below, but it should be rare
log::logger().flush();
// Use a single `eprintln` to ensure we hold the lock for the whole message.
// Defensively pre-construct a single string so that `eprintln` is
// more likely to use a single `write` call, to minimize the chance
// of more lines being written to stdout in the meantime, and in
// case of C code writing to `STDERR` directly without taking Rust's
// lock.
let msg = format!(
"\
\n** Pausing with SIGTSTP to enable debugger attachment to managed process\
\n** '{plugin_name:?}' (pid {native_pid:?}).\
\n** If running Shadow under Bash, resume Shadow by pressing Ctrl-Z to background\
\n** this task, and then typing \"fg\".\
\n** If running GDB, resume Shadow by typing \"signal SIGCONT\"."
);
eprintln!("{}", msg);
rustix::process::kill_process(rustix::process::getpid(), rustix::process::Signal::Tstp)
.unwrap();
}
let memory_manager = unsafe { MemoryManager::new(native_pid) };
let threads = RefCell::new(BTreeMap::from([(
main_thread_id,
RootedRc::new(host.root(), RootedRefCell::new(host.root(), main_thread)),
)]));
let common = Common {
id: process_id,
host_id: host.id(),
working_dir,
name,
plugin_name,
parent_pid: Cell::new(ProcessId::INIT),
group_id: Cell::new(ProcessId::INIT),
session_id: Cell::new(ProcessId::INIT),
// Exit signal is moot; since parent is INIT there will never
// be a valid target for it.
exit_signal: None,
};
Ok(RootedRc::new(
host.root(),
RootedRefCell::new(
host.root(),
Self {
state: RefCell::new(Some(ProcessState::Runnable(RunnableProcess {
common,
expected_final_state: Some(expected_final_state),
shim_shared_mem_block,
memory_manager: Box::new(RefCell::new(memory_manager)),
itimer_real,
strace_logging,
dumpable: Cell::new(SuidDump::SUID_DUMP_USER),
native_pid,
unsafe_borrow_mut: RefCell::new(None),
unsafe_borrows: RefCell::new(Vec::new()),
threads,
#[cfg(feature = "perf_timers")]
cpu_delay_timer,
#[cfg(feature = "perf_timers")]
total_run_time: Cell::new(Duration::ZERO),
child_process_event_listeners: Default::default(),
shimlog_file,
}))),
},
),
))
}
pub fn id(&self) -> ProcessId {
self.common().id
}
pub fn parent_id(&self) -> ProcessId {
self.common().parent_pid.get()
}
pub fn set_parent_id(&self, pid: ProcessId) {
self.common().parent_pid.set(pid)
}
pub fn group_id(&self) -> ProcessId {
self.common().group_id.get()
}
pub fn set_group_id(&self, id: ProcessId) {
self.common().group_id.set(id)
}
pub fn session_id(&self) -> ProcessId {
self.common().session_id.get()
}
pub fn set_session_id(&self, id: ProcessId) {
self.common().session_id.set(id)
}
pub fn host_id(&self) -> HostId {
self.common().host_id
}
/// Get process's "dumpable" state, as manipulated by the prctl operations `PR_SET_DUMPABLE` and
/// `PR_GET_DUMPABLE`.
pub fn dumpable(&self) -> SuidDump {
self.as_runnable().unwrap().dumpable.get()
}
/// Set process's "dumpable" state, as manipulated by the prctl operations `PR_SET_DUMPABLE` and
/// `PR_GET_DUMPABLE`.
pub fn set_dumpable(&self, val: SuidDump) {
assert!(val == SuidDump::SUID_DUMP_DISABLE || val == SuidDump::SUID_DUMP_USER);
self.as_runnable().unwrap().dumpable.set(val)
}
/// Deprecated wrapper for `RunnableProcess::start_cpu_delay_timer`
#[cfg(feature = "perf_timers")]
pub fn start_cpu_delay_timer(&self) {
self.as_runnable().unwrap().start_cpu_delay_timer()
}
/// Deprecated wrapper for `RunnableProcess::stop_cpu_delay_timer`
#[cfg(feature = "perf_timers")]
pub fn stop_cpu_delay_timer(&self, host: &Host) -> Duration {
self.as_runnable().unwrap().stop_cpu_delay_timer(host)
}
pub fn thread_group_leader_id(&self) -> ThreadId {
self.common().thread_group_leader_id()
}
/// Resume execution of `tid` (if it exists).
/// Should only be called from `Host::resume`.
pub fn resume(&self, host: &Host, tid: ThreadId) {
trace!("Continuing thread {} in process {}", tid, self.id());
let threadrc = {
let Some(runnable) = self.as_runnable() else {
debug!("Process {} is no longer running", &*self.name());
return;
};
let threads = runnable.threads.borrow();
let Some(thread) = threads.get(&tid) else {
debug!("Thread {} no longer exists", tid);
return;
};
// Clone the thread reference, so that we don't hold a dynamically
// borrowed reference to the thread list while running the thread.
thread.clone(host.root())
};
let threadrc = ExplicitDropper::new(threadrc, |t| {
t.explicit_drop_recursive(host.root(), host);
});
let thread = threadrc.borrow(host.root());
Worker::set_active_thread(&threadrc);
#[cfg(feature = "perf_timers")]
self.start_cpu_delay_timer();
Process::set_shared_time(host);
// Discard any unapplied latency.
// We currently only want this mechanism to force a yield if the thread itself
// never yields; we don't want unapplied latency to accumulate and force a yield
// under normal circumstances.
host.shim_shmem_lock_borrow_mut()
.unwrap()
.unapplied_cpu_latency = SimulationTime::ZERO;
let ctx = ProcessContext::new(host, self);
let res = thread.resume(&ctx);
#[cfg(feature = "perf_timers")]
{
let delay = self.stop_cpu_delay_timer(host);
debug!("process '{}' ran for {:?}", &*self.name(), delay);
}
#[cfg(not(feature = "perf_timers"))]
debug!("process '{}' done continuing", &*self.name());
match res {
crate::host::thread::ResumeResult::Blocked => {
debug!(
"thread {tid} in process '{}' still running, but blocked",
&*self.name()
);
}
crate::host::thread::ResumeResult::ExitedThread(return_code) => {
debug!(
"thread {tid} in process '{}' exited with code {return_code}",
&*self.name(),
);
let (threadrc, last_thread) = {
let runnable = self.as_runnable().unwrap();
let mut threads = runnable.threads.borrow_mut();
let threadrc = threads.remove(&tid).unwrap();
(threadrc, threads.is_empty())
};
self.as_runnable().unwrap().reap_thread(host, threadrc);
if last_thread {
self.handle_process_exit(host, false);
}
}
crate::host::thread::ResumeResult::ExitedProcess => {
debug!(
"Process {} exited while running thread {tid}",
&*self.name(),
);
self.handle_process_exit(host, false);
}
};
Worker::clear_active_thread();
}
/// Terminate the Process.
///
/// Should only be called from [`Host::free_all_applications`].
pub fn stop(&self, host: &Host) {
// Scope for `runnable`
{
let Some(runnable) = self.as_runnable() else {
debug!("process {} has already stopped", &*self.name());
return;
};
debug!("terminating process {}", &*self.name());
#[cfg(feature = "perf_timers")]
runnable.start_cpu_delay_timer();
if let Err(err) = rustix::process::kill_process(
runnable.native_pid().into(),
rustix::process::Signal::Kill,
) {
warn!("kill: {:?}", err);
}
#[cfg(feature = "perf_timers")]
{
let delay = runnable.stop_cpu_delay_timer(host);
debug!("process '{}' stopped in {:?}", &*self.name(), delay);
}
#[cfg(not(feature = "perf_timers"))]
debug!("process '{}' stopped", &*self.name());
}
// Mutates `self.state`, so we need to have dropped `runnable`.
self.handle_process_exit(host, true);
}
/// See `RunnableProcess::signal`.
///
/// No-op if the `self` is a `ZombieProcess`.
pub fn signal(&self, host: &Host, current_thread: Option<&Thread>, siginfo_t: &siginfo_t) {
// Using full-match here to force update if we add more states later.
match self.state.borrow().as_ref().unwrap() {
ProcessState::Runnable(r) => r.signal(host, current_thread, siginfo_t),
ProcessState::Zombie(_) => {
// Sending a signal to a zombie process is a no-op.
debug!("Process {} no longer running", &*self.name());
}
}
}
fn open_stdio_file_helper(
descriptor_table: &mut DescriptorTable,
fd: DescriptorHandle,
path: PathBuf,
access_mode: OFlag,
) {
let stdfile = unsafe { cshadow::regularfile_new() };
let cwd = rustix::process::getcwd(Vec::new()).unwrap();
let path = utility::pathbuf_to_nul_term_cstring(path);
// "Convert" to libc int, assuming here that the kernel's `OFlag` values
// are compatible with libc's values.
// XXX: We're assuming here that the kernel and libc flags are ABI
// compatible, which isn't guaranteed, but is mostly true in practice.
// TODO: We probably ought to change `regularfile_open` and friends to
// use a direct syscall instead of libc's wrappers, and explicitly take
// the kernel version of flags, mode, etc.
let access_mode = access_mode.bits();
let errorcode = unsafe {
cshadow::regularfile_open(
stdfile,
path.as_ptr(),
access_mode | libc::O_CREAT | libc::O_TRUNC,
libc::S_IRUSR | libc::S_IWUSR | libc::S_IRGRP | libc::S_IROTH,
cwd.as_ptr(),
)
};
if errorcode != 0 {
panic!(
"Opening {}: {:?}",
path.to_str().unwrap(),
linux_api::errno::Errno::try_from(-errorcode).unwrap()
);
}
let desc = unsafe {
Descriptor::from_legacy_file(
stdfile as *mut cshadow::LegacyFile,
linux_api::fcntl::OFlag::empty(),
)
};
let prev = descriptor_table.register_descriptor_with_fd(desc, fd);
assert!(prev.is_none());
trace!(
"Successfully opened fd {} at {}",
fd,
path.to_str().unwrap()
);
}
// Needed during early init, before `Self` is created.
fn static_output_file_name(file_basename: &Path, extension: &str) -> PathBuf {
let mut path = file_basename.to_owned().into_os_string();
path.push(".");
path.push(extension);
path.into()
}
pub fn name(&self) -> impl Deref<Target = str> + '_ {
Ref::map(self.common(), |c| c.name.to_str().unwrap())
}
pub fn plugin_name(&self) -> impl Deref<Target = str> + '_ {
Ref::map(self.common(), |c| c.plugin_name.to_str().unwrap())
}
/// Deprecated wrapper for `RunnableProcess::memory_borrow_mut`
#[track_caller]
pub fn memory_borrow_mut(&self) -> impl DerefMut<Target = MemoryManager> + '_ {
std_util::nested_ref::NestedRefMut::map(self.as_runnable().unwrap(), |runnable| {
runnable.memory_manager.borrow_mut()
})
}
/// Deprecated wrapper for `RunnableProcess::memory_borrow`
#[track_caller]
pub fn memory_borrow(&self) -> impl Deref<Target = MemoryManager> + '_ {
std_util::nested_ref::NestedRef::map(self.as_runnable().unwrap(), |runnable| {
runnable.memory_manager.borrow()
})
}
/// Deprecated wrapper for `RunnableProcess::strace_logging_options`
pub fn strace_logging_options(&self) -> Option<FmtOptions> {
self.as_runnable().unwrap().strace_logging_options()
}
/// Deprecated wrapper for `RunnableProcess::with_strace_file`
pub fn with_strace_file<T>(&self, f: impl FnOnce(&mut std::fs::File) -> T) -> Option<T> {
self.as_runnable().unwrap().with_strace_file(f)
}
/// Deprecated wrapper for `RunnableProcess::native_pid`
pub fn native_pid(&self) -> Pid {
self.as_runnable().unwrap().native_pid()
}
/// Deprecated wrapper for `RunnableProcess::realtime_timer_borrow`
#[track_caller]
pub fn realtime_timer_borrow(&self) -> impl Deref<Target = Timer> + '_ {
std_util::nested_ref::NestedRef::map(self.as_runnable().unwrap(), |runnable| {
runnable.itimer_real.borrow()
})
}
/// Deprecated wrapper for `RunnableProcess::realtime_timer_borrow_mut`
#[track_caller]
pub fn realtime_timer_borrow_mut(&self) -> impl DerefMut<Target = Timer> + '_ {
std_util::nested_ref::NestedRefMut::map(self.as_runnable().unwrap(), |runnable| {
runnable.itimer_real.borrow_mut()
})
}
/// Deprecated wrapper for `RunnableProcess::first_live_thread_borrow`
#[track_caller]
pub fn first_live_thread_borrow(
&self,
root: &Root,
) -> Option<impl Deref<Target = RootedRc<RootedRefCell<Thread>>> + '_> {
std_util::nested_ref::NestedRef::filter_map(self.as_runnable()?, |runnable| {
runnable.first_live_thread(root)
})
}
/// Deprecated wrapper for `RunnableProcess::thread_borrow`
pub fn thread_borrow(
&self,
virtual_tid: ThreadId,
) -> Option<impl Deref<Target = RootedRc<RootedRefCell<Thread>>> + '_> {
std_util::nested_ref::NestedRef::filter_map(self.as_runnable()?, |runnable| {
runnable.thread(virtual_tid)
})
}
/// Deprecated wrapper for [`RunnableProcess::free_unsafe_borrows_flush`].
pub fn free_unsafe_borrows_flush(&self) -> Result<(), Errno> {
self.as_runnable().unwrap().free_unsafe_borrows_flush()
}
/// Deprecated wrapper for [`RunnableProcess::free_unsafe_borrows_noflush`].
pub fn free_unsafe_borrows_noflush(&self) {
self.as_runnable().unwrap().free_unsafe_borrows_noflush()
}
pub fn physical_address(&self, vptr: ForeignPtr<()>) -> ManagedPhysicalMemoryAddr {
self.common().physical_address(vptr)
}
pub fn is_running(&self) -> bool {
self.as_runnable().is_some()
}
/// Transitions `self` from a `RunnableProcess` to a `ZombieProcess`.
fn handle_process_exit(&self, host: &Host, killed_by_shadow: bool) {
debug!(
"process '{}' has completed or is otherwise no longer running",
&*self.name()
);
// Take and dispose of all of the threads.
// TODO: consider doing this while the `self.state` mutable reference is held
// as with the other cleanup below. Right now this breaks some C code that expects
// to be able to lookup the thread's process name.
{
let runnable = self.as_runnable().unwrap();
let threads = std::mem::take(&mut *runnable.threads.borrow_mut());
for (_tid, threadrc) in threads.into_iter() {
threadrc.borrow(host.root()).handle_process_exit();
runnable.reap_thread(host, threadrc);
}
}
// Intentionally hold the borrow on self.state to ensure the state
// transition is "atomic".
let mut opt_state = self.state.borrow_mut();
let state = opt_state.take().unwrap();
let ProcessState::Runnable(runnable) = state else {
unreachable!("Tried to handle process exit of non-running process");
};
#[cfg(feature = "perf_timers")]
debug!(
"total runtime for process '{}' was {:?}",
runnable.common.name(),
runnable.total_run_time.get()
);
let wait_res: Option<WaitStatus> =
rustix::process::waitpid(Some(runnable.native_pid().into()), WaitOptions::empty())
.unwrap_or_else(|e| {
panic!("Error waiting for {:?}: {:?}", runnable.native_pid(), e)
});
let wait_status = wait_res.unwrap();
let exit_status = if killed_by_shadow {
if wait_status.terminating_signal()
!= Some(Signal::SIGKILL.as_i32().try_into().unwrap())
{
warn!("Unexpected waitstatus after killed by shadow: {wait_status:?}");
}
ExitStatus::StoppedByShadow
} else if let Some(code) = wait_status.exit_status() {
ExitStatus::Normal(code.try_into().unwrap())
} else if let Some(signal) = wait_status.terminating_signal() {
ExitStatus::Signaled(Signal::try_from(i32::try_from(signal).unwrap()).unwrap())
} else {
panic!(
"Unexpected status: {wait_status:?} for pid {:?}",
runnable.native_pid()
);
};
let (main_result_string, log_level) = {
let mut s = format!(
"process '{name}' exited with status {exit_status:?}",
name = runnable.common.name()
);
if let Some(expected_final_state) = runnable.expected_final_state {
let actual_final_state = match exit_status {
ExitStatus::Normal(i) => ProcessFinalState::Exited { exited: i },
ExitStatus::Signaled(s) => ProcessFinalState::Signaled {
// This conversion will fail on realtime signals, but that
// should currently be impossible since we don't support
// sending realtime signals.
signaled: s.try_into().unwrap(),
},
ExitStatus::StoppedByShadow => ProcessFinalState::Running(RunningVal::Running),
};
if expected_final_state == actual_final_state {
(s, log::Level::Debug)
} else {
Worker::increment_plugin_error_count();
write!(s, "; expected end state was {expected_final_state} but was {actual_final_state}").unwrap();
(s, log::Level::Error)
}
} else {
(s, log::Level::Debug)
}
};
log::log!(log_level, "{}", main_result_string);
let zombie = ZombieProcess {
common: runnable.into_common(),
exit_status,
};
zombie.notify_parent_of_exit(host);
*opt_state = Some(ProcessState::Zombie(zombie));
}
/// Deprecated wrapper for `RunnableProcess::add_thread`
pub fn add_thread(&self, host: &Host, thread: RootedRc<RootedRefCell<Thread>>) {
self.as_runnable().unwrap().add_thread(host, thread)
}
/// FIXME: still needed? Time is now updated more granularly in the Thread code
/// when xferring control to/from shim.
fn set_shared_time(host: &Host) {
let mut host_shmem = host.shim_shmem_lock_borrow_mut().unwrap();
host_shmem.max_runahead_time = Worker::max_event_runahead_time(host);
host.shim_shmem()
.sim_time
.store(Worker::current_time().unwrap(), Ordering::Relaxed);
}
/// Deprecated wrapper for `RunnableProcess::shmem`
pub fn shmem(&self) -> impl Deref<Target = ShMemBlock<'static, ProcessShmem>> + '_ {
Ref::map(self.as_runnable().unwrap(), |r| &r.shim_shared_mem_block)
}
/// Resource usage, as returned e.g. by the `getrusage` syscall.
pub fn rusage(&self) -> linux_api::resource::rusage {
warn_once_then_debug!(
"resource usage (rusage) tracking unimplemented; Returning bogus zeroed values"
);
// TODO: Actually track some of these.
// Assuming we want to support `RUSAGE_THREAD` in the `getrusage`
// syscall, we'll actually want to track at the thread level, and either
// increment at both thread and process level at the points where we do
// the tracking, or dynamically iterate over the threads here and sum
// the results.
linux_api::resource::rusage {
ru_utime: linux_api::time::kernel_old_timeval {
tv_sec: 0,
tv_usec: 0,
},
ru_stime: linux_api::time::kernel_old_timeval {
tv_sec: 0,
tv_usec: 0,
},
ru_maxrss: 0,
ru_ixrss: 0,
ru_idrss: 0,
ru_isrss: 0,
ru_minflt: 0,
ru_majflt: 0,
ru_nswap: 0,
ru_inblock: 0,
ru_oublock: 0,
ru_msgsnd: 0,
ru_msgrcv: 0,
ru_nsignals: 0,
ru_nvcsw: 0,
ru_nivcsw: 0,
}
}
/// Signal that will be sent to parent process on exit. Typically `Some(SIGCHLD)`.
pub fn exit_signal(&self) -> Option<Signal> {
self.common().exit_signal
}
pub fn current_working_dir(&self) -> impl Deref<Target = CString> + '_ {
Ref::map(self.common(), |common| &common.working_dir)
}
/// Set the process's working directory.
/// This must be kept in sync with the actual working dir of the native process.
/// See <https://github.com/shadow/shadow/issues/2960>
// TODO: This ought to be at the thread level, to support `CLONE_FS`.
pub fn set_current_working_dir(&self, path: CString) {
self.common_mut().working_dir = path;
}
/// Update `self` to complete an `exec` syscall from thread `tid`, replacing
/// the running managed process with `mthread`.
pub fn update_for_exec(&mut self, host: &Host, tid: ThreadId, mthread: ManagedThread) {
let Some(mut runnable) = self.as_runnable_mut() else {
// This could happen if another event runs before the "execve completion" event
// and kills the process. e.g. another thread in the process could run and
// execute the `exit_group` syscall.
log::debug!(
"Process {:?} exited before it could complete execve",
self.id()
);
mthread.kill_and_drop();
return;
};
let old_native_pid = std::mem::replace(&mut runnable.native_pid, mthread.native_pid());
// Kill the previous native process
rustix::process::kill_process(old_native_pid.into(), rustix::process::Signal::Kill)
.expect("Unable to send kill signal to managed process {old_native_pid:?}");
let wait_res = rustix::process::waitpid(Some(old_native_pid.into()), WaitOptions::empty())
.unwrap()
.unwrap();
assert_eq!(
wait_res.terminating_signal(),
Some(Signal::SIGKILL.as_i32().try_into().unwrap())
);
let execing_thread = runnable.threads.borrow_mut().remove(&tid).unwrap();
// Dispose of all threads other than the thread that's running `exec`.
for (_tid, thread) in runnable.threads.replace(BTreeMap::new()) {
// Notify the ManagedThread that the native process has exited.
thread.borrow(host.root()).mthread().handle_process_exit();
thread.explicit_drop_recursive(host.root(), host);
}
// Recreate the `MemoryManager`
{
// We can't safely replace the memory manager if there are outstanding
// unsafe references in C code. There shouldn't be any, though, since
// this is only called from the `execve` and `execveat` syscall handlers,
// which are in Rust.
let unsafe_borrow_mut = runnable.unsafe_borrow_mut.borrow();
let unsafe_borrows = runnable.unsafe_borrows.borrow();
assert!(unsafe_borrow_mut.is_none());
assert!(unsafe_borrows.is_empty());
// Replace the MM, while still holding the references to the unsafe borrows
// to ensure none exist.
runnable
.memory_manager
.replace(unsafe { MemoryManager::new(mthread.native_pid()) });
}
let new_tid = runnable.common.thread_group_leader_id();
log::trace!(
"updating for exec; pid:{pid}, tid:{tid:?}, new_tid:{new_tid:?}",
pid = runnable.common.id
);
execing_thread
.borrow_mut(host.root())
.update_for_exec(host, mthread, new_tid);
runnable
.threads
.borrow_mut()
.insert(new_tid, execing_thread);
// Exit signal is reset to SIGCHLD.
runnable.common.exit_signal = Some(Signal::SIGCHLD);
// Reset signal actions to default.
// `execve(2)`:
// POSIX.1 specifies that the dispositions of any signals that
// are ignored or set to the default are left unchanged. POSIX.1
// specifies one exception: if SIGCHLD is being ignored, then an
// implementation may leave the disposition unchanged or reset it
// to the default; Linux does the former.
let host_shmem_prot = host.shim_shmem_lock_borrow_mut().unwrap();
let mut shmem_prot = runnable
.shim_shared_mem_block
.protected
.borrow_mut(&host_shmem_prot.root);
for signal in Signal::standard_signals() {
let current_action = unsafe { shmem_prot.signal_action(signal) };
if !(current_action.is_default()
|| current_action.is_ignore()
|| signal == Signal::SIGCHLD && current_action.is_ignore())
{
unsafe {
*shmem_prot.signal_action_mut(signal) = linux_api::signal::sigaction::new_raw(
linux_api::signal::SignalHandler::SigDfl,
SigActionFlags::empty(),
sigset_t::EMPTY,
None,
)
};
}
}
}
}
impl Drop for Process {
fn drop(&mut self) {
// Should have been explicitly dropped.
debug_assert!(self.state.borrow().is_none());
}
}
impl ExplicitDrop for Process {
type ExplicitDropParam = Host;
type ExplicitDropResult = ();
fn explicit_drop(mut self, host: &Self::ExplicitDropParam) -> Self::ExplicitDropResult {
// Should normally only be dropped in the zombie state.
debug_assert!(self.as_zombie().is_some() || std::thread::panicking());
let state = self.state.get_mut().take().unwrap();
state.explicit_drop(host);
}
}
/// Tracks a memory reference made by a legacy C memory-read API.
struct UnsafeBorrow {
// Must come before `manager`, so that it's dropped first, since it's
// borrowed from it.
_memory: ProcessMemoryRef<'static, u8>,
_manager: Ref<'static, MemoryManager>,
}
impl UnsafeBorrow {
/// Creates a raw readable pointer, and saves an instance of `Self` into
/// `process` for later clean-up.
///
/// # Safety
///
/// The pointer is invalidated when one of the Process memory flush methods is called.
unsafe fn readable_ptr(
process: &Process,
ptr: ForeignArrayPtr<u8>,
) -> Result<*const c_void, Errno> {
let runnable = process.as_runnable().unwrap();
let manager = runnable.memory_manager.borrow();
// SAFETY: We ensure that the `memory` is dropped before the `manager`,
// and `Process` ensures that this whole object is dropped before
// `MemoryManager` can be moved, freed, etc.
let manager = unsafe {
std::mem::transmute::<Ref<'_, MemoryManager>, Ref<'static, MemoryManager>>(manager)
};
let memory = manager.memory_ref(ptr)?;
let memory = unsafe {
std::mem::transmute::<ProcessMemoryRef<'_, u8>, ProcessMemoryRef<'static, u8>>(memory)
};
let vptr = memory.as_ptr() as *mut c_void;
runnable.unsafe_borrows.borrow_mut().push(Self {
_manager: manager,
_memory: memory,
});
Ok(vptr)
}
/// Creates a raw readable string, and saves an instance of `Self` into
/// `process` for later clean-up.
///
/// # Safety
///
/// The pointer is invalidated when one of the Process memory flush methods is called.
unsafe fn readable_string(
process: &Process,
ptr: ForeignArrayPtr<c_char>,
) -> Result<(*const c_char, libc::size_t), Errno> {
let runnable = process.as_runnable().unwrap();
let manager = runnable.memory_manager.borrow();
// SAFETY: We ensure that the `memory` is dropped before the `manager`,
// and `Process` ensures that this whole object is dropped before
// `MemoryManager` can be moved, freed, etc.
let manager = unsafe {
std::mem::transmute::<Ref<'_, MemoryManager>, Ref<'static, MemoryManager>>(manager)
};
let ptr = ptr.cast_u8();
let memory = manager.memory_ref_prefix(ptr)?;
let memory = unsafe {
std::mem::transmute::<ProcessMemoryRef<'_, u8>, ProcessMemoryRef<'static, u8>>(memory)
};
if !memory.contains(&0) {
return Err(Errno::ENAMETOOLONG);
}
assert_eq!(std::mem::size_of::<c_char>(), std::mem::size_of::<u8>());
let ptr = memory.as_ptr() as *const c_char;
let len = memory.len();
runnable.unsafe_borrows.borrow_mut().push(Self {
_manager: manager,
_memory: memory,
});
Ok((ptr, len))
}
}
// Safety: Normally the Ref would make this non-Send, since it could end then
// end up trying to manipulate the source RefCell (which is !Sync) from multiple
// threads. We ensure that these objects never escape Process, which itself is
// non-Sync, ensuring this doesn't happen.
//
// This is admittedly hand-wavy and making some assumptions about the
// implementation of RefCell, but this whole type is temporary scaffolding to
// support legacy C code.
unsafe impl Send for UnsafeBorrow {}
/// Tracks a memory reference made by a legacy C memory-write API.
struct UnsafeBorrowMut {
// Must come before `manager`, so that it's dropped first, since it's
// borrowed from it.
memory: Option<ProcessMemoryRefMut<'static, u8>>,
_manager: RefMut<'static, MemoryManager>,
}
impl UnsafeBorrowMut {
/// Creates a raw writable pointer, and saves an instance of `Self` into
/// `process` for later clean-up. The initial contents of the pointer is unspecified.
///
/// # Safety
///
/// The pointer is invalidated when one of the Process memory flush methods is called.
unsafe fn writable_ptr(
process: &Process,
ptr: ForeignArrayPtr<u8>,
) -> Result<*mut c_void, Errno> {
let runnable = process.as_runnable().unwrap();
let manager = runnable.memory_manager.borrow_mut();
// SAFETY: We ensure that the `memory` is dropped before the `manager`,
// and `Process` ensures that this whole object is dropped before
// `MemoryManager` can be moved, freed, etc.
let mut manager = unsafe {
std::mem::transmute::<RefMut<'_, MemoryManager>, RefMut<'static, MemoryManager>>(
manager,
)
};
let memory = manager.memory_ref_mut_uninit(ptr)?;
let mut memory = unsafe {
std::mem::transmute::<ProcessMemoryRefMut<'_, u8>, ProcessMemoryRefMut<'static, u8>>(
memory,
)
};
let vptr = memory.as_mut_ptr() as *mut c_void;
let prev = runnable.unsafe_borrow_mut.borrow_mut().replace(Self {
_manager: manager,
memory: Some(memory),
});
assert!(prev.is_none());
Ok(vptr)
}
/// Creates a raw mutable pointer, and saves an instance of `Self` into
/// `process` for later clean-up.
///
/// # Safety
///
/// The pointer is invalidated when one of the Process memory flush methods is called.
unsafe fn mutable_ptr(
process: &Process,
ptr: ForeignArrayPtr<u8>,
) -> Result<*mut c_void, Errno> {
let runnable = process.as_runnable().unwrap();
let manager = runnable.memory_manager.borrow_mut();
// SAFETY: We ensure that the `memory` is dropped before the `manager`,
// and `Process` ensures that this whole object is dropped before
// `MemoryManager` can be moved, freed, etc.
let mut manager = unsafe {
std::mem::transmute::<RefMut<'_, MemoryManager>, RefMut<'static, MemoryManager>>(
manager,
)
};
let memory = manager.memory_ref_mut(ptr)?;
let mut memory = unsafe {
std::mem::transmute::<ProcessMemoryRefMut<'_, u8>, ProcessMemoryRefMut<'static, u8>>(
memory,
)
};
let vptr = memory.as_mut_ptr() as *mut c_void;
let prev = runnable.unsafe_borrow_mut.borrow_mut().replace(Self {
_manager: manager,
memory: Some(memory),
});
assert!(prev.is_none());
Ok(vptr)
}
/// Free this reference, writing back to process memory.
fn flush(mut self) -> Result<(), Errno> {
self.memory.take().unwrap().flush()
}
/// Free this reference without writing back to process memory.
fn noflush(mut self) {
self.memory.take().unwrap().noflush()
}
}
// Safety: Normally the RefMut would make this non-Send, since it could end then
// end up trying to manipulate the source RefCell (which is !Sync) from multiple
// threads. We ensure that these objects never escape Process, which itself is
// non-Sync, ensuring this doesn't happen.
//
// This is admittedly hand-wavy and making some assumptions about the implementation of
// RefCell, but this whole type is temporary scaffolding to support legacy C code.
unsafe impl Send for UnsafeBorrowMut {}
fn make_name(host: &Host, exe_name: &str, id: ProcessId) -> CString {
CString::new(format!(
"{host_name}.{exe_name}.{id}",
host_name = host.name(),
exe_name = exe_name,
id = u32::from(id)
))
.unwrap()
}
mod export {
use std::os::raw::c_void;
use libc::size_t;
use log::trace;
use shadow_shim_helper_rs::notnull::*;
use shadow_shim_helper_rs::shim_shmem::export::ShimShmemProcess;
use shadow_shim_helper_rs::syscall_types::UntypedForeignPtr;
use super::*;
use crate::utility::HostTreePointer;
/// Copy `n` bytes from `src` to `dst`. Returns 0 on success or -EFAULT if any of
/// the specified range couldn't be accessed. Always succeeds with n==0.
#[no_mangle]
pub extern "C-unwind" fn process_readPtr(
proc: *const Process,
dst: *mut c_void,
src: UntypedForeignPtr,
n: usize,
) -> i32 {
let proc = unsafe { proc.as_ref().unwrap() };
let src = ForeignArrayPtr::new(src.cast::<u8>(), n);
let dst = unsafe { std::slice::from_raw_parts_mut(notnull_mut_debug(dst) as *mut u8, n) };
match proc.memory_borrow().copy_from_ptr(dst, src) {
Ok(_) => 0,
Err(e) => {
trace!("Couldn't read {:?} into {:?}: {:?}", src, dst, e);
e.to_negated_i32()
}
}
}
/// Copy `n` bytes from `src` to `dst`. Returns 0 on success or -EFAULT if any of
/// the specified range couldn't be accessed. The write is flushed immediately.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_writePtr(
proc: *const Process,
dst: UntypedForeignPtr,
src: *const c_void,
n: usize,
) -> i32 {
let proc = unsafe { proc.as_ref().unwrap() };
let dst = ForeignArrayPtr::new(dst.cast::<u8>(), n);
let src = unsafe { std::slice::from_raw_parts(notnull_debug(src) as *const u8, n) };
match proc.memory_borrow_mut().copy_to_ptr(dst, src) {
Ok(_) => 0,
Err(e) => {
trace!("Couldn't write {:?} into {:?}: {:?}", src, dst, e);
e.to_negated_i32()
}
}
}
/// Make the data at plugin_src available in shadow's address space.
///
/// The returned pointer is invalidated when one of the process memory flush
/// methods is called; typically after a syscall has completed.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getReadablePtr(
proc: *const Process,
plugin_src: UntypedForeignPtr,
n: usize,
) -> *const c_void {
let proc = unsafe { proc.as_ref().unwrap() };
let plugin_src = ForeignArrayPtr::new(plugin_src.cast::<u8>(), n);
unsafe { UnsafeBorrow::readable_ptr(proc, plugin_src).unwrap_or(std::ptr::null()) }
}
/// Returns a writable pointer corresponding to the named region. The
/// initial contents of the returned memory are unspecified.
///
/// The returned pointer is invalidated when one of the process memory flush
/// methods is called; typically after a syscall has completed.
///
/// CAUTION: if the unspecified contents aren't overwritten, and the pointer
/// isn't explicitly freed via `process_freePtrsWithoutFlushing`, those
/// unspecified contents may be written back into process memory.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getWriteablePtr(
proc: *const Process,
plugin_src: UntypedForeignPtr,
n: usize,
) -> *mut c_void {
let proc = unsafe { proc.as_ref().unwrap() };
let plugin_src = ForeignArrayPtr::new(plugin_src.cast::<u8>(), n);
unsafe { UnsafeBorrowMut::writable_ptr(proc, plugin_src).unwrap_or(std::ptr::null_mut()) }
}
/// Returns a writeable pointer corresponding to the specified src. Use when
/// the data at the given address needs to be both read and written.
///
/// The returned pointer is invalidated when one of the process memory flush
/// methods is called; typically after a syscall has completed.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getMutablePtr(
proc: *const Process,
plugin_src: UntypedForeignPtr,
n: usize,
) -> *mut c_void {
let proc = unsafe { proc.as_ref().unwrap() };
let plugin_src = ForeignArrayPtr::new(plugin_src.cast::<u8>(), n);
unsafe { UnsafeBorrowMut::mutable_ptr(proc, plugin_src).unwrap_or(std::ptr::null_mut()) }
}
/// Reads up to `n` bytes into `str`.
///
/// Returns:
/// strlen(str) on success.
/// -ENAMETOOLONG if there was no NULL byte in the first `n` characters.
/// -EFAULT if the string extends beyond the accessible address space.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_readString(
proc: *const Process,
strbuf: *mut libc::c_char,
ptr: UntypedForeignPtr,
maxlen: libc::size_t,
) -> libc::ssize_t {
let proc = unsafe { proc.as_ref().unwrap() };
let memory_manager = proc.memory_borrow();
let buf =
unsafe { std::slice::from_raw_parts_mut(notnull_mut_debug(strbuf) as *mut u8, maxlen) };
let cstr = match memory_manager
.copy_str_from_ptr(buf, ForeignArrayPtr::new(ptr.cast::<u8>(), maxlen))
{
Ok(cstr) => cstr,
Err(e) => return e.to_negated_i32() as isize,
};
cstr.to_bytes().len().try_into().unwrap()
}
/// Reads up to `n` bytes into `str`.
///
/// Returns:
/// strlen(str) on success.
/// -ENAMETOOLONG if there was no NULL byte in the first `n` characters.
/// -EFAULT if the string extends beyond the accessible address space.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getReadableString(
proc: *const Process,
plugin_src: UntypedForeignPtr,
n: usize,
out_str: *mut *const c_char,
out_strlen: *mut size_t,
) -> i32 {
let proc = unsafe { proc.as_ref().unwrap() };
let ptr = ForeignArrayPtr::new(plugin_src.cast::<c_char>(), n);
match unsafe { UnsafeBorrow::readable_string(proc, ptr) } {
Ok((str, strlen)) => {
assert!(!out_str.is_null());
unsafe { out_str.write(str) };
if !out_strlen.is_null() {
unsafe { out_strlen.write(strlen) };
}
0
}
Err(e) => e.to_negated_i32(),
}
}
/// Returns the processID that was assigned to us in process_new
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getProcessID(proc: *const Process) -> libc::pid_t {
let proc = unsafe { proc.as_ref().unwrap() };
proc.id().into()
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getName(proc: *const Process) -> *const c_char {
let proc = unsafe { proc.as_ref().unwrap() };
proc.common().name.as_ptr()
}
/// Safety:
///
/// The returned pointer is invalidated when the host shmem lock is released, e.g. via
/// Host::unlock_shmem.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getSharedMem(
proc: *const Process,
) -> *const ShimShmemProcess {
let proc = unsafe { proc.as_ref().unwrap() };
std::ptr::from_ref(proc.as_runnable().unwrap().shim_shared_mem_block.deref())
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getWorkingDir(proc: *const Process) -> *const c_char {
let proc = unsafe { proc.as_ref().unwrap() };
proc.common().working_dir.as_ptr()
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_straceLoggingMode(
proc: *const Process,
) -> StraceFmtMode {
let proc = unsafe { proc.as_ref().unwrap() };
proc.strace_logging_options().into()
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getNativePid(proc: *const Process) -> libc::pid_t {
let proc = unsafe { proc.as_ref().unwrap() };
proc.native_pid().as_raw_nonzero().get()
}
/// Flushes and invalidates all previously returned readable/writable plugin
/// pointers, as if returning control to the plugin. This can be useful in
/// conjunction with `thread_nativeSyscall` operations that touch memory, or
/// to gracefully handle failed writes.
///
/// Returns 0 on success or a negative errno on failure.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_flushPtrs(proc: *const Process) -> i32 {
let proc = unsafe { proc.as_ref().unwrap() };
match proc.free_unsafe_borrows_flush() {
Ok(_) => 0,
Err(e) => e.to_negated_i32(),
}
}
/// Frees all readable/writable foreign pointers. Unlike process_flushPtrs, any
/// previously returned writable pointer is *not* written back. Useful
/// if an uninitialized writable pointer was obtained via `process_getWriteablePtr`,
/// and we end up not wanting to write anything after all (in particular, don't
/// write back whatever garbage data was in the uninialized bueffer).
#[no_mangle]
pub unsafe extern "C-unwind" fn process_freePtrsWithoutFlushing(proc: *const Process) {
let proc = unsafe { proc.as_ref().unwrap() };
proc.free_unsafe_borrows_noflush();
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getThread(
proc: *const Process,
tid: libc::pid_t,
) -> *const Thread {
let proc = unsafe { proc.as_ref().unwrap() };
Worker::with_active_host(|host| {
let tid = ThreadId::try_from(tid).unwrap();
let Some(thread) = proc.thread_borrow(tid) else {
return std::ptr::null();
};
let thread = thread.borrow(host.root());
&*thread
})
.unwrap()
}
/// Returns a pointer to an arbitrary live thread in the process.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_firstLiveThread(proc: *const Process) -> *const Thread {
let proc = unsafe { proc.as_ref().unwrap() };
Worker::with_active_host(|host| {
let Some(thread) = proc.first_live_thread_borrow(host.root()) else {
return std::ptr::null();
};
let thread = thread.borrow(host.root());
&*thread
})
.unwrap()
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_isRunning(proc: *const Process) -> bool {
let proc = unsafe { proc.as_ref().unwrap() };
proc.is_running()
}
// FIXME: still needed? Time is now updated more granularly in the Thread code
// when xferring control to/from shim.
#[no_mangle]
pub unsafe extern "C-unwind" fn process_setSharedTime() {
Worker::with_active_host(Process::set_shared_time).unwrap();
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_getPhysicalAddress(
proc: *const Process,
vptr: UntypedForeignPtr,
) -> ManagedPhysicalMemoryAddr {
let proc = unsafe { proc.as_ref().unwrap() };
proc.physical_address(vptr)
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_addChildEventListener(
host: *const Host,
process: *const Process,
listener: *mut cshadow::StatusListener,
) {
let host = unsafe { host.as_ref().unwrap() };
let process = unsafe { process.as_ref().unwrap() };
let listener = HostTreePointer::new_for_host(host.id(), listener);
process
.borrow_as_runnable()
.unwrap()
.child_process_event_listeners
.borrow_mut()
.add_legacy_listener(listener)
}
#[no_mangle]
pub unsafe extern "C-unwind" fn process_removeChildEventListener(
_host: *const Host,
process: *const Process,
listener: *mut cshadow::StatusListener,
) {
let process = unsafe { process.as_ref().unwrap() };
process
.borrow_as_runnable()
.unwrap()
.child_process_event_listeners
.borrow_mut()
.remove_legacy_listener(listener)
}
}