1 /* $OpenBSD: file.c,v 1.72 2024/11/21 13:20:27 claudio Exp $ */ 2 3 /* 4 * Copyright (c) 2015 Nicholas Marriott <nicm@openbsd.org> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF MIND, USE, DATA OR PROFITS, WHETHER 15 * IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING 16 * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 */ 18 19 #include <sys/types.h> 20 #include <sys/ioctl.h> 21 #include <sys/mman.h> 22 #include <sys/queue.h> 23 #include <sys/socket.h> 24 #include <sys/stat.h> 25 #include <sys/uio.h> 26 #include <sys/wait.h> 27 28 #include <err.h> 29 #include <errno.h> 30 #include <fcntl.h> 31 #include <getopt.h> 32 #include <imsg.h> 33 #include <libgen.h> 34 #include <limits.h> 35 #include <pwd.h> 36 #include <stdlib.h> 37 #include <string.h> 38 #include <time.h> 39 #include <unistd.h> 40 41 #include "file.h" 42 #include "magic.h" 43 #include "xmalloc.h" 44 45 struct input_msg { 46 int idx; 47 48 struct stat sb; 49 int error; 50 51 char link_path[PATH_MAX]; 52 int link_error; 53 int link_target; 54 }; 55 56 struct input_ack { 57 int idx; 58 }; 59 60 struct input_file { 61 struct magic *m; 62 struct input_msg *msg; 63 64 const char *path; 65 int fd; 66 67 void *base; 68 size_t size; 69 int mapped; 70 char *result; 71 }; 72 73 extern char *__progname; 74 75 __dead void usage(void); 76 77 static int prepare_message(struct input_msg *, int, const char *); 78 static void send_message(struct imsgbuf *, void *, size_t, int); 79 static int read_message(struct imsgbuf *, struct imsg *, pid_t); 80 81 static void read_link(struct input_msg *, const char *); 82 83 static __dead void child(int, pid_t, int, char **); 84 85 static void test_file(struct input_file *, size_t); 86 87 static int try_stat(struct input_file *); 88 static int try_empty(struct input_file *); 89 static int try_access(struct input_file *); 90 static int try_text(struct input_file *); 91 static int try_magic(struct input_file *); 92 static int try_unknown(struct input_file *); 93 94 static int bflag; 95 static int cflag; 96 static int iflag; 97 static int Lflag; 98 static int sflag; 99 static int Wflag; 100 101 static char *magicpath; 102 static FILE *magicfp; 103 104 static struct option longopts[] = { 105 { "brief", no_argument, NULL, 'b' }, 106 { "dereference", no_argument, NULL, 'L' }, 107 { "mime", no_argument, NULL, 'i' }, 108 { "mime-type", no_argument, NULL, 'i' }, 109 { NULL, 0, NULL, 0 } 110 }; 111 112 __dead void 113 usage(void) 114 { 115 fprintf(stderr, "usage: %s [-bchiLsW] file ...\n", __progname); 116 exit(1); 117 } 118 119 int 120 main(int argc, char **argv) 121 { 122 int opt, pair[2], fd, idx; 123 char *home; 124 struct passwd *pw; 125 struct imsgbuf ibuf; 126 struct imsg imsg; 127 struct input_msg msg; 128 struct input_ack ack; 129 pid_t pid, parent; 130 131 tzset(); 132 133 for (;;) { 134 opt = getopt_long(argc, argv, "bchiLsW", longopts, NULL); 135 if (opt == -1) 136 break; 137 switch (opt) { 138 case 'b': 139 bflag = 1; 140 break; 141 case 'c': 142 cflag = 1; 143 break; 144 case 'h': 145 Lflag = 0; 146 break; 147 case 'i': 148 iflag = 1; 149 break; 150 case 'L': 151 Lflag = 1; 152 break; 153 case 's': 154 sflag = 1; 155 break; 156 case 'W': 157 Wflag = 1; 158 break; 159 default: 160 usage(); 161 } 162 } 163 argc -= optind; 164 argv += optind; 165 if (cflag) { 166 if (argc != 0) 167 usage(); 168 } else if (argc == 0) 169 usage(); 170 171 if (pledge("stdio rpath getpw recvfd sendfd id proc", NULL) == -1) 172 err(1, "pledge"); 173 174 magicfp = NULL; 175 if (geteuid() != 0 && !issetugid()) { 176 home = getenv("HOME"); 177 if (home == NULL || *home == '\0') { 178 pw = getpwuid(getuid()); 179 if (pw != NULL) 180 home = pw->pw_dir; 181 else 182 home = NULL; 183 } 184 if (home != NULL) { 185 xasprintf(&magicpath, "%s/.magic", home); 186 magicfp = fopen(magicpath, "r"); 187 if (magicfp == NULL) 188 free(magicpath); 189 } 190 } 191 if (magicfp == NULL) { 192 magicpath = xstrdup("/etc/magic"); 193 magicfp = fopen(magicpath, "r"); 194 } 195 if (magicfp == NULL) 196 err(1, "%s", magicpath); 197 198 parent = getpid(); 199 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, pair) != 0) 200 err(1, "socketpair"); 201 switch (pid = fork()) { 202 case -1: 203 err(1, "fork"); 204 case 0: 205 close(pair[0]); 206 child(pair[1], parent, argc, argv); 207 } 208 close(pair[1]); 209 210 if (pledge("stdio rpath sendfd", NULL) == -1) 211 err(1, "pledge"); 212 213 fclose(magicfp); 214 magicfp = NULL; 215 216 if (cflag) 217 goto wait_for_child; 218 219 imsgbuf_init(&ibuf, pair[0]); 220 for (idx = 0; idx < argc; idx++) { 221 fd = prepare_message(&msg, idx, argv[idx]); 222 send_message(&ibuf, &msg, sizeof msg, fd); 223 224 if (read_message(&ibuf, &imsg, pid) == 0) 225 break; 226 if (imsg_get_data(&imsg, &ack, sizeof ack) == -1) 227 err(1, "bad message"); 228 if (ack.idx != idx) 229 errx(1, "index not expected"); 230 imsg_free(&imsg); 231 } 232 233 wait_for_child: 234 close(pair[0]); 235 while (wait(NULL) == -1 && errno != ECHILD) { 236 if (errno != EINTR) 237 err(1, "wait"); 238 } 239 _exit(0); /* let the child flush */ 240 } 241 242 static int 243 prepare_message(struct input_msg *msg, int idx, const char *path) 244 { 245 int fd, mode, error; 246 247 memset(msg, 0, sizeof *msg); 248 msg->idx = idx; 249 250 if (strcmp(path, "-") == 0) { 251 if (fstat(STDIN_FILENO, &msg->sb) == -1) { 252 msg->error = errno; 253 return (-1); 254 } 255 return (STDIN_FILENO); 256 } 257 258 if (Lflag) 259 error = stat(path, &msg->sb); 260 else 261 error = lstat(path, &msg->sb); 262 if (error == -1) { 263 msg->error = errno; 264 return (-1); 265 } 266 267 /* 268 * pledge(2) doesn't let us pass directory file descriptors around - 269 * but in fact we don't need them, so just don't open directories or 270 * symlinks (which could be to directories). 271 */ 272 mode = msg->sb.st_mode; 273 if (!S_ISDIR(mode) && !S_ISLNK(mode)) { 274 fd = open(path, O_RDONLY|O_NONBLOCK); 275 if (fd == -1 && (errno == ENFILE || errno == EMFILE)) 276 err(1, "open"); 277 } else 278 fd = -1; 279 if (S_ISLNK(mode)) 280 read_link(msg, path); 281 return (fd); 282 283 } 284 285 static void 286 send_message(struct imsgbuf *ibuf, void *msg, size_t msglen, int fd) 287 { 288 if (imsg_compose(ibuf, -1, -1, 0, fd, msg, msglen) != 1) 289 err(1, "imsg_compose"); 290 if (imsgbuf_flush(ibuf) != 0) 291 err(1, "imsgbuf_flush"); 292 } 293 294 static int 295 read_message(struct imsgbuf *ibuf, struct imsg *imsg, pid_t from) 296 { 297 int n; 298 299 while ((n = imsgbuf_read(ibuf)) == -1 && errno == EAGAIN) 300 /* nothing */ ; 301 if (n == -1) 302 err(1, "imsgbuf_read"); 303 if (n == 0) 304 return (0); 305 306 if ((n = imsg_get(ibuf, imsg)) == -1) 307 err(1, "imsg_get"); 308 if (n == 0) 309 return (0); 310 311 if ((pid_t)imsg->hdr.pid != from) 312 errx(1, "PIDs don't match"); 313 314 return (n); 315 316 } 317 318 static void 319 read_link(struct input_msg *msg, const char *path) 320 { 321 struct stat sb; 322 char lpath[PATH_MAX]; 323 char *copy, *root; 324 int used; 325 ssize_t size; 326 327 size = readlink(path, lpath, sizeof lpath - 1); 328 if (size == -1) { 329 msg->link_error = errno; 330 return; 331 } 332 lpath[size] = '\0'; 333 334 if (*lpath == '/') 335 strlcpy(msg->link_path, lpath, sizeof msg->link_path); 336 else { 337 copy = xstrdup(path); 338 339 root = dirname(copy); 340 if (*root == '\0' || strcmp(root, ".") == 0 || 341 strcmp (root, "/") == 0) 342 strlcpy(msg->link_path, lpath, sizeof msg->link_path); 343 else { 344 used = snprintf(msg->link_path, sizeof msg->link_path, 345 "%s/%s", root, lpath); 346 if (used < 0 || (size_t)used >= sizeof msg->link_path) { 347 msg->link_error = ENAMETOOLONG; 348 free(copy); 349 return; 350 } 351 } 352 353 free(copy); 354 } 355 356 if (!Lflag && stat(path, &sb) == -1) 357 msg->link_target = errno; 358 } 359 360 static __dead void 361 child(int fd, pid_t parent, int argc, char **argv) 362 { 363 struct passwd *pw; 364 struct magic *m; 365 struct imsgbuf ibuf; 366 struct imsg imsg; 367 struct input_msg msg; 368 struct input_ack ack; 369 struct input_file inf; 370 int i, idx; 371 size_t len, width = 0; 372 373 if (pledge("stdio getpw recvfd id", NULL) == -1) 374 err(1, "pledge"); 375 376 if (geteuid() == 0) { 377 pw = getpwnam(FILE_USER); 378 if (pw == NULL) 379 errx(1, "unknown user %s", FILE_USER); 380 if (setgroups(1, &pw->pw_gid) != 0) 381 err(1, "setgroups"); 382 if (setresgid(pw->pw_gid, pw->pw_gid, pw->pw_gid) != 0) 383 err(1, "setresgid"); 384 if (setresuid(pw->pw_uid, pw->pw_uid, pw->pw_uid) != 0) 385 err(1, "setresuid"); 386 } 387 388 if (pledge("stdio recvfd", NULL) == -1) 389 err(1, "pledge"); 390 391 m = magic_load(magicfp, magicpath, cflag || Wflag); 392 if (cflag) { 393 magic_dump(m); 394 exit(0); 395 } 396 397 for (i = 0; i < argc; i++) { 398 len = strlen(argv[i]) + 1; 399 if (len > width) 400 width = len; 401 } 402 403 imsgbuf_init(&ibuf, fd); 404 for (;;) { 405 if (read_message(&ibuf, &imsg, parent) == 0) 406 break; 407 if (imsg_get_data(&imsg, &msg, sizeof msg) == -1) 408 err(1, "bad message"); 409 410 idx = msg.idx; 411 if (idx < 0 || idx >= argc) 412 errx(1, "index out of range"); 413 414 memset(&inf, 0, sizeof inf); 415 inf.m = m; 416 inf.msg = &msg; 417 418 inf.path = argv[idx]; 419 inf.fd = imsg_get_fd(&imsg); 420 421 test_file(&inf, width); 422 423 if (inf.fd != -1) 424 close(inf.fd); 425 imsg_free(&imsg); 426 427 ack.idx = idx; 428 send_message(&ibuf, &ack, sizeof ack, -1); 429 } 430 exit(0); 431 } 432 433 static void * 434 fill_buffer(int fd, size_t size, size_t *used) 435 { 436 static void *buffer; 437 ssize_t got; 438 size_t left; 439 void *next; 440 441 if (buffer == NULL) 442 buffer = xmalloc(FILE_READ_SIZE); 443 444 next = buffer; 445 left = size; 446 while (left != 0) { 447 got = read(fd, next, left); 448 if (got == -1) { 449 if (errno == EINTR) 450 continue; 451 return (NULL); 452 } 453 if (got == 0) 454 break; 455 next = (char *)next + got; 456 left -= got; 457 } 458 *used = size - left; 459 return (buffer); 460 } 461 462 static int 463 load_file(struct input_file *inf) 464 { 465 size_t used; 466 467 if (inf->msg->sb.st_size == 0 && S_ISREG(inf->msg->sb.st_mode)) 468 return (0); /* empty file */ 469 if (inf->msg->sb.st_size == 0 || inf->msg->sb.st_size > FILE_READ_SIZE) 470 inf->size = FILE_READ_SIZE; 471 else 472 inf->size = inf->msg->sb.st_size; 473 474 if (!S_ISREG(inf->msg->sb.st_mode)) 475 goto try_read; 476 477 inf->base = mmap(NULL, inf->size, PROT_READ, MAP_PRIVATE, inf->fd, 0); 478 if (inf->base == MAP_FAILED) 479 goto try_read; 480 inf->mapped = 1; 481 return (0); 482 483 try_read: 484 inf->base = fill_buffer(inf->fd, inf->size, &used); 485 if (inf->base == NULL) { 486 xasprintf(&inf->result, "cannot read '%s' (%s)", inf->path, 487 strerror(errno)); 488 return (1); 489 } 490 inf->size = used; 491 return (0); 492 } 493 494 static int 495 try_stat(struct input_file *inf) 496 { 497 if (inf->msg->error != 0) { 498 xasprintf(&inf->result, "cannot stat '%s' (%s)", inf->path, 499 strerror(inf->msg->error)); 500 return (1); 501 } 502 if (sflag || strcmp(inf->path, "-") == 0) { 503 switch (inf->msg->sb.st_mode & S_IFMT) { 504 case S_IFIFO: 505 if (strcmp(inf->path, "-") != 0) 506 break; 507 case S_IFBLK: 508 case S_IFCHR: 509 case S_IFREG: 510 return (0); 511 } 512 } 513 514 if (iflag && (inf->msg->sb.st_mode & S_IFMT) != S_IFREG) { 515 xasprintf(&inf->result, "application/x-not-regular-file"); 516 return (1); 517 } 518 519 switch (inf->msg->sb.st_mode & S_IFMT) { 520 case S_IFDIR: 521 xasprintf(&inf->result, "directory"); 522 return (1); 523 case S_IFLNK: 524 if (inf->msg->link_error != 0) { 525 xasprintf(&inf->result, "unreadable symlink '%s' (%s)", 526 inf->path, strerror(inf->msg->link_error)); 527 return (1); 528 } 529 if (inf->msg->link_target == ELOOP) 530 xasprintf(&inf->result, "symbolic link in a loop"); 531 else if (inf->msg->link_target != 0) { 532 xasprintf(&inf->result, "broken symbolic link to '%s'", 533 inf->msg->link_path); 534 } else { 535 xasprintf(&inf->result, "symbolic link to '%s'", 536 inf->msg->link_path); 537 } 538 return (1); 539 case S_IFSOCK: 540 xasprintf(&inf->result, "socket"); 541 return (1); 542 case S_IFBLK: 543 xasprintf(&inf->result, "block special (%lu/%lu)", 544 (long)major(inf->msg->sb.st_rdev), 545 (long)minor(inf->msg->sb.st_rdev)); 546 return (1); 547 case S_IFCHR: 548 xasprintf(&inf->result, "character special (%lu/%lu)", 549 (long)major(inf->msg->sb.st_rdev), 550 (long)minor(inf->msg->sb.st_rdev)); 551 return (1); 552 case S_IFIFO: 553 xasprintf(&inf->result, "fifo (named pipe)"); 554 return (1); 555 } 556 return (0); 557 } 558 559 static int 560 try_empty(struct input_file *inf) 561 { 562 if (inf->size != 0) 563 return (0); 564 565 if (iflag) 566 xasprintf(&inf->result, "application/x-empty"); 567 else 568 xasprintf(&inf->result, "empty"); 569 return (1); 570 } 571 572 static int 573 try_access(struct input_file *inf) 574 { 575 char tmp[256] = ""; 576 577 if (inf->msg->sb.st_size == 0 && S_ISREG(inf->msg->sb.st_mode)) 578 return (0); /* empty file */ 579 if (inf->fd != -1) 580 return (0); 581 582 if (inf->msg->sb.st_mode & (S_IWUSR|S_IWGRP|S_IWOTH)) 583 strlcat(tmp, "writable, ", sizeof tmp); 584 if (inf->msg->sb.st_mode & (S_IXUSR|S_IXGRP|S_IXOTH)) 585 strlcat(tmp, "executable, ", sizeof tmp); 586 if (S_ISREG(inf->msg->sb.st_mode)) 587 strlcat(tmp, "regular file, ", sizeof tmp); 588 strlcat(tmp, "no read permission", sizeof tmp); 589 590 inf->result = xstrdup(tmp); 591 return (1); 592 } 593 594 static int 595 try_text(struct input_file *inf) 596 { 597 const char *type, *s; 598 int flags; 599 600 flags = MAGIC_TEST_TEXT; 601 if (iflag) 602 flags |= MAGIC_TEST_MIME; 603 604 type = text_get_type(inf->base, inf->size); 605 if (type == NULL) 606 return (0); 607 608 s = magic_test(inf->m, inf->base, inf->size, flags); 609 if (s != NULL) { 610 inf->result = xstrdup(s); 611 return (1); 612 } 613 614 s = text_try_words(inf->base, inf->size, flags); 615 if (s != NULL) { 616 if (iflag) 617 inf->result = xstrdup(s); 618 else 619 xasprintf(&inf->result, "%s %s text", type, s); 620 return (1); 621 } 622 623 if (iflag) 624 inf->result = xstrdup("text/plain"); 625 else 626 xasprintf(&inf->result, "%s text", type); 627 return (1); 628 } 629 630 static int 631 try_magic(struct input_file *inf) 632 { 633 const char *s; 634 int flags; 635 636 flags = 0; 637 if (iflag) 638 flags |= MAGIC_TEST_MIME; 639 640 s = magic_test(inf->m, inf->base, inf->size, flags); 641 if (s != NULL) { 642 inf->result = xstrdup(s); 643 return (1); 644 } 645 return (0); 646 } 647 648 static int 649 try_unknown(struct input_file *inf) 650 { 651 if (iflag) 652 xasprintf(&inf->result, "application/octet-stream"); 653 else 654 xasprintf(&inf->result, "data"); 655 return (1); 656 } 657 658 static void 659 test_file(struct input_file *inf, size_t width) 660 { 661 char *label; 662 int stop; 663 664 stop = 0; 665 if (!stop) 666 stop = try_stat(inf); 667 if (!stop) 668 stop = try_access(inf); 669 if (!stop) 670 stop = load_file(inf); 671 if (!stop) 672 stop = try_empty(inf); 673 if (!stop) 674 stop = try_magic(inf); 675 if (!stop) 676 stop = try_text(inf); 677 if (!stop) 678 stop = try_unknown(inf); 679 680 if (bflag) 681 printf("%s\n", inf->result); 682 else { 683 if (strcmp(inf->path, "-") == 0) 684 xasprintf(&label, "/dev/stdin:"); 685 else 686 xasprintf(&label, "%s:", inf->path); 687 printf("%-*s %s\n", (int)width, label, inf->result); 688 free(label); 689 } 690 free(inf->result); 691 692 if (inf->mapped && inf->base != NULL) 693 munmap(inf->base, inf->size); 694 } 695