1 /* 2 * BSD syscalls 3 * 4 * Copyright (c) 2003-2008 Fabrice Bellard 5 * Copyright (c) 2013-2014 Stacey D. Son 6 * 7 * This program is free software; you can redistribute it and/or modify 8 * it under the terms of the GNU General Public License as published by 9 * the Free Software Foundation; either version 2 of the License, or 10 * (at your option) any later version. 11 * 12 * This program is distributed in the hope that it will be useful, 13 * but WITHOUT ANY WARRANTY; without even the implied warranty of 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 15 * GNU General Public License for more details. 16 * 17 * You should have received a copy of the GNU General Public License 18 * along with this program; if not, see <http://www.gnu.org/licenses/>. 19 */ 20 21 /* 22 * We need the FreeBSD "legacy" definitions. Rust needs the FreeBSD 11 system 23 * calls since it doesn't use libc at all, so we have to emulate that despite 24 * FreeBSD 11 being EOL'd. 25 */ 26 #define _WANT_FREEBSD11_STAT 27 #define _WANT_FREEBSD11_STATFS 28 #define _WANT_FREEBSD11_DIRENT 29 #define _WANT_KERNEL_ERRNO 30 #define _WANT_SEMUN 31 #include "qemu/osdep.h" 32 #include "qemu/cutils.h" 33 #include "qemu/path.h" 34 #include <sys/syscall.h> 35 #include <sys/param.h> 36 #include <sys/mount.h> 37 #include <sys/sysctl.h> 38 #include <utime.h> 39 40 #include "qemu.h" 41 #include "signal-common.h" 42 #include "user/syscall-trace.h" 43 44 #include "bsd-file.h" 45 #include "bsd-proc.h" 46 47 /* I/O */ 48 safe_syscall3(int, open, const char *, path, int, flags, mode_t, mode); 49 safe_syscall4(int, openat, int, fd, const char *, path, int, flags, mode_t, 50 mode); 51 52 safe_syscall3(ssize_t, read, int, fd, void *, buf, size_t, nbytes); 53 safe_syscall4(ssize_t, pread, int, fd, void *, buf, size_t, nbytes, off_t, 54 offset); 55 safe_syscall3(ssize_t, readv, int, fd, const struct iovec *, iov, int, iovcnt); 56 safe_syscall4(ssize_t, preadv, int, fd, const struct iovec *, iov, int, iovcnt, 57 off_t, offset); 58 59 safe_syscall3(ssize_t, write, int, fd, void *, buf, size_t, nbytes); 60 safe_syscall4(ssize_t, pwrite, int, fd, void *, buf, size_t, nbytes, off_t, 61 offset); 62 safe_syscall3(ssize_t, writev, int, fd, const struct iovec *, iov, int, iovcnt); 63 safe_syscall4(ssize_t, pwritev, int, fd, const struct iovec *, iov, int, iovcnt, 64 off_t, offset); 65 66 void target_set_brk(abi_ulong new_brk) 67 { 68 } 69 70 /* 71 * errno conversion. 72 */ 73 abi_long get_errno(abi_long ret) 74 { 75 if (ret == -1) { 76 return -host_to_target_errno(errno); 77 } else { 78 return ret; 79 } 80 } 81 82 int host_to_target_errno(int err) 83 { 84 /* 85 * All the BSDs have the property that the error numbers are uniform across 86 * all architectures for a given BSD, though they may vary between different 87 * BSDs. 88 */ 89 return err; 90 } 91 92 bool is_error(abi_long ret) 93 { 94 return (abi_ulong)ret >= (abi_ulong)(-4096); 95 } 96 97 /* 98 * Unlocks a iovec. Unlike unlock_iovec, it assumes the tvec array itself is 99 * already locked from target_addr. It will be unlocked as well as all the iovec 100 * elements. 101 */ 102 static void helper_unlock_iovec(struct target_iovec *target_vec, 103 abi_ulong target_addr, struct iovec *vec, 104 int count, int copy) 105 { 106 for (int i = 0; i < count; i++) { 107 abi_ulong base = tswapal(target_vec[i].iov_base); 108 109 if (vec[i].iov_base) { 110 unlock_user(vec[i].iov_base, base, copy ? vec[i].iov_len : 0); 111 } 112 } 113 unlock_user(target_vec, target_addr, 0); 114 } 115 116 struct iovec *lock_iovec(int type, abi_ulong target_addr, 117 int count, int copy) 118 { 119 struct target_iovec *target_vec; 120 struct iovec *vec; 121 abi_ulong total_len, max_len; 122 int i; 123 int err = 0; 124 125 if (count == 0) { 126 errno = 0; 127 return NULL; 128 } 129 if (count < 0 || count > IOV_MAX) { 130 errno = EINVAL; 131 return NULL; 132 } 133 134 vec = g_try_new0(struct iovec, count); 135 if (vec == NULL) { 136 errno = ENOMEM; 137 return NULL; 138 } 139 140 target_vec = lock_user(VERIFY_READ, target_addr, 141 count * sizeof(struct target_iovec), 1); 142 if (target_vec == NULL) { 143 err = EFAULT; 144 goto fail2; 145 } 146 147 max_len = 0x7fffffff & MIN(TARGET_PAGE_MASK, PAGE_MASK); 148 total_len = 0; 149 150 for (i = 0; i < count; i++) { 151 abi_ulong base = tswapal(target_vec[i].iov_base); 152 abi_long len = tswapal(target_vec[i].iov_len); 153 154 if (len < 0) { 155 err = EINVAL; 156 goto fail; 157 } else if (len == 0) { 158 /* Zero length pointer is ignored. */ 159 vec[i].iov_base = 0; 160 } else { 161 vec[i].iov_base = lock_user(type, base, len, copy); 162 /* 163 * If the first buffer pointer is bad, this is a fault. But 164 * subsequent bad buffers will result in a partial write; this is 165 * realized by filling the vector with null pointers and zero 166 * lengths. 167 */ 168 if (!vec[i].iov_base) { 169 if (i == 0) { 170 err = EFAULT; 171 goto fail; 172 } else { 173 /* 174 * Fail all the subsequent addresses, they are already 175 * zero'd. 176 */ 177 goto out; 178 } 179 } 180 if (len > max_len - total_len) { 181 len = max_len - total_len; 182 } 183 } 184 vec[i].iov_len = len; 185 total_len += len; 186 } 187 out: 188 unlock_user(target_vec, target_addr, 0); 189 return vec; 190 191 fail: 192 helper_unlock_iovec(target_vec, target_addr, vec, i, copy); 193 fail2: 194 g_free(vec); 195 errno = err; 196 return NULL; 197 } 198 199 void unlock_iovec(struct iovec *vec, abi_ulong target_addr, 200 int count, int copy) 201 { 202 struct target_iovec *target_vec; 203 204 target_vec = lock_user(VERIFY_READ, target_addr, 205 count * sizeof(struct target_iovec), 1); 206 if (target_vec) { 207 helper_unlock_iovec(target_vec, target_addr, vec, count, copy); 208 } 209 210 g_free(vec); 211 } 212 213 /* 214 * All errnos that freebsd_syscall() returns must be -TARGET_<errcode>. 215 */ 216 static abi_long freebsd_syscall(void *cpu_env, int num, abi_long arg1, 217 abi_long arg2, abi_long arg3, abi_long arg4, 218 abi_long arg5, abi_long arg6, abi_long arg7, 219 abi_long arg8) 220 { 221 abi_long ret; 222 223 switch (num) { 224 /* 225 * process system calls 226 */ 227 case TARGET_FREEBSD_NR_exit: /* exit(2) */ 228 ret = do_bsd_exit(cpu_env, arg1); 229 break; 230 231 /* 232 * File system calls. 233 */ 234 case TARGET_FREEBSD_NR_read: /* read(2) */ 235 ret = do_bsd_read(arg1, arg2, arg3); 236 break; 237 238 case TARGET_FREEBSD_NR_pread: /* pread(2) */ 239 ret = do_bsd_pread(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6); 240 break; 241 242 case TARGET_FREEBSD_NR_readv: /* readv(2) */ 243 ret = do_bsd_readv(arg1, arg2, arg3); 244 break; 245 246 case TARGET_FREEBSD_NR_preadv: /* preadv(2) */ 247 ret = do_bsd_preadv(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6); 248 249 case TARGET_FREEBSD_NR_write: /* write(2) */ 250 ret = do_bsd_write(arg1, arg2, arg3); 251 break; 252 253 case TARGET_FREEBSD_NR_pwrite: /* pwrite(2) */ 254 ret = do_bsd_pwrite(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6); 255 break; 256 257 case TARGET_FREEBSD_NR_writev: /* writev(2) */ 258 ret = do_bsd_writev(arg1, arg2, arg3); 259 break; 260 261 case TARGET_FREEBSD_NR_pwritev: /* pwritev(2) */ 262 ret = do_bsd_pwritev(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6); 263 break; 264 265 case TARGET_FREEBSD_NR_open: /* open(2) */ 266 ret = do_bsd_open(arg1, arg2, arg3); 267 break; 268 269 case TARGET_FREEBSD_NR_openat: /* openat(2) */ 270 ret = do_bsd_openat(arg1, arg2, arg3, arg4); 271 break; 272 273 case TARGET_FREEBSD_NR_close: /* close(2) */ 274 ret = do_bsd_close(arg1); 275 break; 276 277 case TARGET_FREEBSD_NR_fdatasync: /* fdatasync(2) */ 278 ret = do_bsd_fdatasync(arg1); 279 break; 280 281 case TARGET_FREEBSD_NR_fsync: /* fsync(2) */ 282 ret = do_bsd_fsync(arg1); 283 break; 284 285 case TARGET_FREEBSD_NR_freebsd12_closefrom: /* closefrom(2) */ 286 ret = do_bsd_closefrom(arg1); 287 break; 288 289 case TARGET_FREEBSD_NR_revoke: /* revoke(2) */ 290 ret = do_bsd_revoke(arg1); 291 break; 292 293 case TARGET_FREEBSD_NR_access: /* access(2) */ 294 ret = do_bsd_access(arg1, arg2); 295 break; 296 297 case TARGET_FREEBSD_NR_eaccess: /* eaccess(2) */ 298 ret = do_bsd_eaccess(arg1, arg2); 299 break; 300 301 case TARGET_FREEBSD_NR_faccessat: /* faccessat(2) */ 302 ret = do_bsd_faccessat(arg1, arg2, arg3, arg4); 303 break; 304 305 case TARGET_FREEBSD_NR_chdir: /* chdir(2) */ 306 ret = do_bsd_chdir(arg1); 307 break; 308 309 case TARGET_FREEBSD_NR_fchdir: /* fchdir(2) */ 310 ret = do_bsd_fchdir(arg1); 311 break; 312 313 case TARGET_FREEBSD_NR_rename: /* rename(2) */ 314 ret = do_bsd_rename(arg1, arg2); 315 break; 316 317 case TARGET_FREEBSD_NR_renameat: /* renameat(2) */ 318 ret = do_bsd_renameat(arg1, arg2, arg3, arg4); 319 break; 320 321 case TARGET_FREEBSD_NR_link: /* link(2) */ 322 ret = do_bsd_link(arg1, arg2); 323 break; 324 325 case TARGET_FREEBSD_NR_linkat: /* linkat(2) */ 326 ret = do_bsd_linkat(arg1, arg2, arg3, arg4, arg5); 327 break; 328 329 case TARGET_FREEBSD_NR_unlink: /* unlink(2) */ 330 ret = do_bsd_unlink(arg1); 331 break; 332 333 case TARGET_FREEBSD_NR_unlinkat: /* unlinkat(2) */ 334 ret = do_bsd_unlinkat(arg1, arg2, arg3); 335 break; 336 337 case TARGET_FREEBSD_NR_mkdir: /* mkdir(2) */ 338 ret = do_bsd_mkdir(arg1, arg2); 339 break; 340 341 case TARGET_FREEBSD_NR_mkdirat: /* mkdirat(2) */ 342 ret = do_bsd_mkdirat(arg1, arg2, arg3); 343 break; 344 345 case TARGET_FREEBSD_NR_rmdir: /* rmdir(2) (XXX no rmdirat()?) */ 346 ret = do_bsd_rmdir(arg1); 347 break; 348 349 case TARGET_FREEBSD_NR___getcwd: /* undocumented __getcwd() */ 350 ret = do_bsd___getcwd(arg1, arg2); 351 break; 352 353 case TARGET_FREEBSD_NR_dup: /* dup(2) */ 354 ret = do_bsd_dup(arg1); 355 break; 356 357 case TARGET_FREEBSD_NR_dup2: /* dup2(2) */ 358 ret = do_bsd_dup2(arg1, arg2); 359 break; 360 361 case TARGET_FREEBSD_NR_truncate: /* truncate(2) */ 362 ret = do_bsd_truncate(cpu_env, arg1, arg2, arg3, arg4); 363 break; 364 365 case TARGET_FREEBSD_NR_ftruncate: /* ftruncate(2) */ 366 ret = do_bsd_ftruncate(cpu_env, arg1, arg2, arg3, arg4); 367 break; 368 369 case TARGET_FREEBSD_NR_acct: /* acct(2) */ 370 ret = do_bsd_acct(arg1); 371 break; 372 373 case TARGET_FREEBSD_NR_sync: /* sync(2) */ 374 ret = do_bsd_sync(); 375 break; 376 377 case TARGET_FREEBSD_NR_mount: /* mount(2) */ 378 ret = do_bsd_mount(arg1, arg2, arg3, arg4); 379 break; 380 381 case TARGET_FREEBSD_NR_unmount: /* unmount(2) */ 382 ret = do_bsd_unmount(arg1, arg2); 383 break; 384 385 case TARGET_FREEBSD_NR_nmount: /* nmount(2) */ 386 ret = do_bsd_nmount(arg1, arg2, arg3); 387 break; 388 389 case TARGET_FREEBSD_NR_symlink: /* symlink(2) */ 390 ret = do_bsd_symlink(arg1, arg2); 391 break; 392 393 case TARGET_FREEBSD_NR_symlinkat: /* symlinkat(2) */ 394 ret = do_bsd_symlinkat(arg1, arg2, arg3); 395 break; 396 397 case TARGET_FREEBSD_NR_readlink: /* readlink(2) */ 398 ret = do_bsd_readlink(cpu_env, arg1, arg2, arg3); 399 break; 400 401 case TARGET_FREEBSD_NR_readlinkat: /* readlinkat(2) */ 402 ret = do_bsd_readlinkat(arg1, arg2, arg3, arg4); 403 break; 404 405 default: 406 qemu_log_mask(LOG_UNIMP, "Unsupported syscall: %d\n", num); 407 ret = -TARGET_ENOSYS; 408 break; 409 } 410 411 return ret; 412 } 413 414 /* 415 * do_freebsd_syscall() should always have a single exit point at the end so 416 * that actions, such as logging of syscall results, can be performed. This 417 * as a wrapper around freebsd_syscall() so that actually happens. Since 418 * that is a singleton, modern compilers will inline it anyway... 419 */ 420 abi_long do_freebsd_syscall(void *cpu_env, int num, abi_long arg1, 421 abi_long arg2, abi_long arg3, abi_long arg4, 422 abi_long arg5, abi_long arg6, abi_long arg7, 423 abi_long arg8) 424 { 425 CPUState *cpu = env_cpu(cpu_env); 426 int ret; 427 428 trace_guest_user_syscall(cpu, num, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8); 429 if (do_strace) { 430 print_freebsd_syscall(num, arg1, arg2, arg3, arg4, arg5, arg6); 431 } 432 433 ret = freebsd_syscall(cpu_env, num, arg1, arg2, arg3, arg4, arg5, arg6, 434 arg7, arg8); 435 if (do_strace) { 436 print_freebsd_syscall_ret(num, ret); 437 } 438 trace_guest_user_syscall_ret(cpu, num, ret); 439 440 return ret; 441 } 442 443 void syscall_init(void) 444 { 445 } 446