tar.c (13860B)
1 /* See LICENSE file for copyright and license details. */ 2 #include <sys/stat.h> 3 #include <sys/time.h> 4 #include <sys/types.h> 5 #ifndef major 6 #include <sys/sysmacros.h> 7 #endif 8 9 #include <errno.h> 10 #include <fcntl.h> 11 #include <grp.h> 12 #include <libgen.h> 13 #include <pwd.h> 14 #include <stdio.h> 15 #include <stdlib.h> 16 #include <string.h> 17 #include <unistd.h> 18 19 #include "fs.h" 20 #include "util.h" 21 22 #define BLKSIZ 512 23 24 enum Type { 25 REG = '0', 26 AREG = '\0', 27 HARDLINK = '1', 28 SYMLINK = '2', 29 CHARDEV = '3', 30 BLOCKDEV = '4', 31 DIRECTORY = '5', 32 FIFO = '6', 33 RESERVED = '7' 34 }; 35 36 struct header { 37 char name[100]; 38 char mode[8]; 39 char uid[8]; 40 char gid[8]; 41 char size[12]; 42 char mtime[12]; 43 char chksum[8]; 44 char type; 45 char linkname[100]; 46 char magic[6]; 47 char version[2]; 48 char uname[32]; 49 char gname[32]; 50 char major[8]; 51 char minor[8]; 52 char prefix[155]; 53 }; 54 55 static struct dirtime { 56 char *name; 57 time_t mtime; 58 } *dirtimes; 59 60 static size_t dirtimeslen; 61 62 static int tarfd; 63 static ino_t tarinode; 64 static dev_t tardev; 65 66 static int mflag, vflag; 67 static int filtermode; 68 static const char *filtertool; 69 70 static const char *filtertools[] = { 71 ['J'] = "xz", 72 ['Z'] = "compress", 73 ['a'] = "lzma", 74 ['j'] = "bzip2", 75 ['z'] = "gzip", 76 }; 77 78 static void 79 pushdirtime(char *name, time_t mtime) 80 { 81 dirtimes = ereallocarray(dirtimes, dirtimeslen + 1, sizeof(*dirtimes)); 82 dirtimes[dirtimeslen].name = estrdup(name); 83 dirtimes[dirtimeslen].mtime = mtime; 84 dirtimeslen++; 85 } 86 87 static struct dirtime * 88 popdirtime(void) 89 { 90 if (dirtimeslen) { 91 dirtimeslen--; 92 return &dirtimes[dirtimeslen]; 93 } 94 return NULL; 95 } 96 97 static int 98 comp(int fd, const char *tool, const char *flags) 99 { 100 int fds[2]; 101 102 if (pipe(fds) < 0) 103 eprintf("pipe:"); 104 105 switch (fork()) { 106 case -1: 107 eprintf("fork:"); 108 case 0: 109 dup2(fd, 1); 110 dup2(fds[0], 0); 111 close(fds[0]); 112 close(fds[1]); 113 114 execlp(tool, tool, flags, NULL); 115 weprintf("execlp %s:", tool); 116 _exit(1); 117 } 118 close(fds[0]); 119 return fds[1]; 120 } 121 122 static int 123 decomp(int fd, const char *tool, const char *flags) 124 { 125 int fds[2]; 126 127 if (pipe(fds) < 0) 128 eprintf("pipe:"); 129 130 switch (fork()) { 131 case -1: 132 eprintf("fork:"); 133 case 0: 134 dup2(fd, 0); 135 dup2(fds[1], 1); 136 close(fds[0]); 137 close(fds[1]); 138 139 execlp(tool, tool, flags, NULL); 140 weprintf("execlp %s:", tool); 141 _exit(1); 142 } 143 close(fds[1]); 144 return fds[0]; 145 } 146 147 static ssize_t 148 eread(int fd, void *buf, size_t n) 149 { 150 ssize_t r; 151 152 again: 153 r = read(fd, buf, n); 154 if (r < 0) { 155 if (errno == EINTR) 156 goto again; 157 eprintf("read:"); 158 } 159 return r; 160 } 161 162 static ssize_t 163 ewrite(int fd, const void *buf, size_t n) 164 { 165 ssize_t r; 166 167 if ((r = write(fd, buf, n)) != n) 168 eprintf("write:"); 169 return r; 170 } 171 172 static void 173 putoctal(char *dst, unsigned num, int size) 174 { 175 if (snprintf(dst, size, "%.*o", size - 1, num) >= size) 176 eprintf("snprintf: input number too large\n"); 177 } 178 179 static int 180 archive(const char *path) 181 { 182 char b[BLKSIZ]; 183 const char *base, *p; 184 struct group *gr; 185 struct header *h; 186 struct passwd *pw; 187 struct stat st; 188 size_t chksum, i, nlen, plen; 189 ssize_t l, r; 190 int fd = -1; 191 192 if (lstat(path, &st) < 0) { 193 weprintf("lstat %s:", path); 194 return 0; 195 } else if (st.st_ino == tarinode && st.st_dev == tardev) { 196 weprintf("ignoring %s\n", path); 197 return 0; 198 } 199 200 pw = getpwuid(st.st_uid); 201 gr = getgrgid(st.st_gid); 202 203 h = (struct header *)b; 204 memset(b, 0, sizeof(b)); 205 206 plen = 0; 207 base = path; 208 if ((nlen = strlen(base)) >= sizeof(h->name)) { 209 /* 210 * Cover case where path name is too long (in which case we 211 * need to split it to prefix and name). 212 */ 213 if ((base = strrchr(path, '/')) == NULL) 214 goto too_long; 215 for (p = base++; p > path && *p == '/'; --p) 216 ; 217 218 nlen -= base - path; 219 plen = p - path + 1; 220 if (nlen >= sizeof(h->name) || plen >= sizeof(h->prefix)) 221 goto too_long; 222 } 223 224 memcpy(h->name, base, nlen); 225 memcpy(h->prefix, path, plen); 226 227 putoctal(h->mode, (unsigned)st.st_mode & 0777, sizeof(h->mode)); 228 putoctal(h->uid, (unsigned)st.st_uid, sizeof(h->uid)); 229 putoctal(h->gid, (unsigned)st.st_gid, sizeof(h->gid)); 230 putoctal(h->size, 0, sizeof(h->size)); 231 putoctal(h->mtime, (unsigned)st.st_mtime, sizeof(h->mtime)); 232 memcpy( h->magic, "ustar", sizeof(h->magic)); 233 memcpy( h->version, "00", sizeof(h->version)); 234 estrlcpy(h->uname, pw ? pw->pw_name : "", sizeof(h->uname)); 235 estrlcpy(h->gname, gr ? gr->gr_name : "", sizeof(h->gname)); 236 237 if (S_ISREG(st.st_mode)) { 238 h->type = REG; 239 putoctal(h->size, (unsigned)st.st_size, sizeof(h->size)); 240 fd = open(path, O_RDONLY); 241 if (fd < 0) 242 eprintf("open %s:", path); 243 } else if (S_ISDIR(st.st_mode)) { 244 h->type = DIRECTORY; 245 } else if (S_ISLNK(st.st_mode)) { 246 h->type = SYMLINK; 247 if ((r = readlink(path, h->linkname, sizeof(h->linkname) - 1)) < 0) 248 eprintf("readlink %s:", path); 249 h->linkname[r] = '\0'; 250 } else if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode)) { 251 h->type = S_ISCHR(st.st_mode) ? CHARDEV : BLOCKDEV; 252 putoctal(h->major, (unsigned)major(st.st_dev), sizeof(h->major)); 253 putoctal(h->minor, (unsigned)minor(st.st_dev), sizeof(h->minor)); 254 } else if (S_ISFIFO(st.st_mode)) { 255 h->type = FIFO; 256 } 257 258 memset(h->chksum, ' ', sizeof(h->chksum)); 259 for (i = 0, chksum = 0; i < sizeof(*h); i++) 260 chksum += (unsigned char)b[i]; 261 putoctal(h->chksum, chksum, sizeof(h->chksum)); 262 ewrite(tarfd, b, BLKSIZ); 263 264 if (fd != -1) { 265 while ((l = eread(fd, b, BLKSIZ)) > 0) { 266 if (l < BLKSIZ) 267 memset(b + l, 0, BLKSIZ - l); 268 ewrite(tarfd, b, BLKSIZ); 269 } 270 close(fd); 271 } 272 273 return 0; 274 275 too_long: 276 eprintf("filename too long: %s\n", path); 277 } 278 279 static int 280 unarchive(char *fname, ssize_t l, char b[BLKSIZ]) 281 { 282 char lname[101], *tmp, *p; 283 long mode, major, minor, type, mtime, uid, gid; 284 struct header *h = (struct header *)b; 285 int fd = -1; 286 struct timespec times[2]; 287 288 if (!mflag && ((mtime = strtol(h->mtime, &p, 8)) < 0 || *p != '\0')) 289 eprintf("strtol %s: invalid number\n", h->mtime); 290 if (remove(fname) < 0 && errno != ENOENT) 291 weprintf("remove %s:", fname); 292 293 tmp = estrdup(fname); 294 mkdirp(dirname(tmp), 0777, 0777); 295 free(tmp); 296 297 switch (h->type) { 298 case REG: 299 case AREG: 300 case RESERVED: 301 if ((mode = strtol(h->mode, &p, 8)) < 0 || *p != '\0') 302 eprintf("strtol %s: invalid number\n", h->mode); 303 fd = open(fname, O_WRONLY | O_TRUNC | O_CREAT, 0600); 304 if (fd < 0) 305 eprintf("open %s:", fname); 306 break; 307 case HARDLINK: 308 case SYMLINK: 309 snprintf(lname, sizeof(lname), "%.*s", (int)sizeof(h->linkname), 310 h->linkname); 311 if (((h->type == HARDLINK) ? link : symlink)(lname, fname) < 0) 312 eprintf("%s %s -> %s:", 313 (h->type == HARDLINK) ? "link" : "symlink", 314 fname, lname); 315 break; 316 case DIRECTORY: 317 if ((mode = strtol(h->mode, &p, 8)) < 0 || *p != '\0') 318 eprintf("strtol %s: invalid number\n", h->mode); 319 if (mkdir(fname, (mode_t)mode) < 0 && errno != EEXIST) 320 eprintf("mkdir %s:", fname); 321 pushdirtime(fname, mtime); 322 break; 323 case CHARDEV: 324 case BLOCKDEV: 325 if ((mode = strtol(h->mode, &p, 8)) < 0 || *p != '\0') 326 eprintf("strtol %s: invalid number\n", h->mode); 327 if ((major = strtol(h->major, &p, 8)) < 0 || *p != '\0') 328 eprintf("strtol %s: invalid number\n", h->major); 329 if ((minor = strtol(h->minor, &p, 8)) < 0 || *p != '\0') 330 eprintf("strtol %s: invalid number\n", h->minor); 331 type = (h->type == CHARDEV) ? S_IFCHR : S_IFBLK; 332 if (mknod(fname, type | mode, makedev(major, minor)) < 0) 333 eprintf("mknod %s:", fname); 334 break; 335 case FIFO: 336 if ((mode = strtol(h->mode, &p, 8)) < 0 || *p != '\0') 337 eprintf("strtol %s: invalid number\n", h->mode); 338 if (mknod(fname, S_IFIFO | mode, 0) < 0) 339 eprintf("mknod %s:", fname); 340 break; 341 default: 342 eprintf("unsupported tar-filetype %c\n", h->type); 343 } 344 345 if ((uid = strtol(h->uid, &p, 8)) < 0 || *p != '\0') 346 eprintf("strtol %s: invalid number\n", h->uid); 347 if ((gid = strtol(h->gid, &p, 8)) < 0 || *p != '\0') 348 eprintf("strtol %s: invalid number\n", h->gid); 349 350 if (fd != -1) { 351 for (; l > 0; l -= BLKSIZ) 352 if (eread(tarfd, b, BLKSIZ) > 0) 353 ewrite(fd, b, MIN(l, BLKSIZ)); 354 close(fd); 355 } 356 357 if (h->type == HARDLINK) 358 return 0; 359 360 times[0].tv_sec = times[1].tv_sec = mtime; 361 times[0].tv_nsec = times[1].tv_nsec = 0; 362 if (!mflag && utimensat(AT_FDCWD, fname, times, AT_SYMLINK_NOFOLLOW) < 0) 363 weprintf("utimensat %s:", fname); 364 if (h->type == SYMLINK) { 365 if (!getuid() && lchown(fname, uid, gid)) 366 weprintf("lchown %s:", fname); 367 } else { 368 if (!getuid() && chown(fname, uid, gid)) 369 weprintf("chown %s:", fname); 370 if (chmod(fname, mode) < 0) 371 eprintf("fchmod %s:", fname); 372 } 373 374 return 0; 375 } 376 377 static void 378 skipblk(ssize_t l) 379 { 380 char b[BLKSIZ]; 381 382 for (; l > 0; l -= BLKSIZ) 383 if (!eread(tarfd, b, BLKSIZ)) 384 break; 385 } 386 387 static int 388 print(char *fname, ssize_t l, char b[BLKSIZ]) 389 { 390 puts(fname); 391 skipblk(l); 392 return 0; 393 } 394 395 static void 396 c(int dirfd, const char *name, struct stat *st, void *data, struct recursor *r) 397 { 398 archive(r->path); 399 if (vflag) 400 puts(r->path); 401 402 if (S_ISDIR(st->st_mode)) 403 recurse(dirfd, name, NULL, r); 404 } 405 406 static void 407 sanitize(struct header *h) 408 { 409 size_t i, j; 410 struct { 411 char *f; 412 size_t l; 413 } fields[] = { 414 { h->mode, sizeof(h->mode) }, 415 { h->uid, sizeof(h->uid) }, 416 { h->gid, sizeof(h->gid) }, 417 { h->size, sizeof(h->size) }, 418 { h->mtime, sizeof(h->mtime) }, 419 { h->chksum, sizeof(h->chksum) }, 420 { h->major, sizeof(h->major) }, 421 { h->minor, sizeof(h->minor) } 422 }; 423 424 /* Numeric fields can be terminated with spaces instead of 425 * NULs as per the ustar specification. Patch all of them to 426 * use NULs so we can perform string operations on them. */ 427 for (i = 0; i < LEN(fields); i++){ 428 for (j = 0; j < fields[i].l && fields[i].f[j] == ' '; j++); 429 for (; j < fields[i].l; j++) 430 if (fields[i].f[j] == ' ') 431 fields[i].f[j] = '\0'; 432 } 433 } 434 435 static void 436 chktar(struct header *h) 437 { 438 char tmp[8], *err, *p = (char *)h; 439 const char *reason; 440 long s1, s2, i; 441 442 if (h->prefix[0] == '\0' && h->name[0] == '\0') { 443 reason = "empty filename"; 444 goto bad; 445 } 446 if (h->magic[0] && strncmp("ustar", h->magic, 5)) { 447 reason = "not ustar format"; 448 goto bad; 449 } 450 memcpy(tmp, h->chksum, sizeof(tmp)); 451 for (i = 0; i < sizeof(tmp) && tmp[i] == ' '; i++); 452 for (; i < sizeof(tmp); i++) 453 if (tmp[i] == ' ') 454 tmp[i] = '\0'; 455 s1 = strtol(tmp, &err, 8); 456 if (s1 < 0 || *err != '\0') { 457 reason = "invalid checksum"; 458 goto bad; 459 } 460 memset(h->chksum, ' ', sizeof(h->chksum)); 461 for (i = 0, s2 = 0; i < sizeof(*h); i++) 462 s2 += (unsigned char)p[i]; 463 if (s1 != s2) { 464 reason = "incorrect checksum"; 465 goto bad; 466 } 467 memcpy(h->chksum, tmp, sizeof(h->chksum)); 468 return; 469 bad: 470 eprintf("malformed tar archive: %s\n", reason); 471 } 472 473 static void 474 xt(int argc, char *argv[], int mode) 475 { 476 char b[BLKSIZ], fname[256 + 1], *p; 477 struct timespec times[2]; 478 struct header *h = (struct header *)b; 479 struct dirtime *dirtime; 480 long size; 481 int i, n; 482 int (*fn)(char *, ssize_t, char[BLKSIZ]) = (mode == 'x') ? unarchive : print; 483 484 while (eread(tarfd, b, BLKSIZ) > 0 && (h->name[0] || h->prefix[0])) { 485 chktar(h); 486 sanitize(h), n = 0; 487 488 /* small dance around non-null terminated fields */ 489 if (h->prefix[0]) 490 n = snprintf(fname, sizeof(fname), "%.*s/", 491 (int)sizeof(h->prefix), h->prefix); 492 snprintf(fname + n, sizeof(fname) - n, "%.*s", 493 (int)sizeof(h->name), h->name); 494 495 if ((size = strtol(h->size, &p, 8)) < 0 || *p != '\0') 496 eprintf("strtol %s: invalid number\n", h->size); 497 498 if (argc) { 499 /* only extract the given files */ 500 for (i = 0; i < argc; i++) 501 if (!strcmp(argv[i], fname)) 502 break; 503 if (i == argc) { 504 skipblk(size); 505 continue; 506 } 507 } 508 509 /* ignore global pax header craziness */ 510 if (h->type == 'g' || h->type == 'x') { 511 skipblk(size); 512 continue; 513 } 514 515 fn(fname, size, b); 516 if (vflag && mode != 't') 517 puts(fname); 518 } 519 520 if (mode == 'x' && !mflag) { 521 while ((dirtime = popdirtime())) { 522 times[0].tv_sec = times[1].tv_sec = dirtime->mtime; 523 times[0].tv_nsec = times[1].tv_nsec = 0; 524 if (utimensat(AT_FDCWD, dirtime->name, times, 0) < 0) 525 eprintf("utimensat %s:", fname); 526 free(dirtime->name); 527 } 528 free(dirtimes); 529 dirtimes = NULL; 530 } 531 } 532 533 static void 534 usage(void) 535 { 536 eprintf("usage: %s [-C dir] [-J | -Z | -a | -j | -z] -x [-m | -t] " 537 "[-f file] [file ...]\n" 538 " %s [-C dir] [-J | -Z | -a | -j | -z] [-h] -c path ... " 539 "[-f file]\n", argv0, argv0); 540 } 541 542 int 543 main(int argc, char *argv[]) 544 { 545 struct recursor r = { .fn = c, .follow = 'P', .flags = DIRFIRST }; 546 struct stat st; 547 char *file = NULL, *dir = ".", mode = '\0'; 548 int fd; 549 550 ARGBEGIN { 551 case 'x': 552 case 'c': 553 case 't': 554 mode = ARGC(); 555 break; 556 case 'C': 557 dir = EARGF(usage()); 558 break; 559 case 'f': 560 file = EARGF(usage()); 561 break; 562 case 'm': 563 mflag = 1; 564 break; 565 case 'J': 566 case 'Z': 567 case 'a': 568 case 'j': 569 case 'z': 570 filtermode = ARGC(); 571 filtertool = filtertools[filtermode]; 572 break; 573 case 'h': 574 r.follow = 'L'; 575 break; 576 case 'v': 577 vflag = 1; 578 break; 579 default: 580 usage(); 581 } ARGEND 582 583 if (!mode) 584 usage(); 585 if (mode == 'c') 586 if (!argc) 587 usage(); 588 589 switch (mode) { 590 case 'c': 591 tarfd = 1; 592 if (file && *file != '-') { 593 tarfd = open(file, O_WRONLY | O_TRUNC | O_CREAT, 0644); 594 if (tarfd < 0) 595 eprintf("open %s:", file); 596 if (lstat(file, &st) < 0) 597 eprintf("lstat %s:", file); 598 tarinode = st.st_ino; 599 tardev = st.st_dev; 600 } 601 602 if (filtertool) 603 tarfd = comp(tarfd, filtertool, "-cf"); 604 605 if (chdir(dir) < 0) 606 eprintf("chdir %s:", dir); 607 for (; *argv; argc--, argv++) 608 recurse(AT_FDCWD, *argv, NULL, &r); 609 break; 610 case 't': 611 case 'x': 612 tarfd = 0; 613 if (file && *file != '-') { 614 tarfd = open(file, O_RDONLY); 615 if (tarfd < 0) 616 eprintf("open %s:", file); 617 } 618 619 if (filtertool) { 620 fd = tarfd; 621 tarfd = decomp(tarfd, filtertool, "-cdf"); 622 close(fd); 623 } 624 625 if (chdir(dir) < 0) 626 eprintf("chdir %s:", dir); 627 xt(argc, argv, mode); 628 break; 629 } 630 631 return recurse_status; 632 }