1 /*- 2 * Copyright (c) 2007 Kai Wang 3 * Copyright (c) 2007 Tim Kientzle 4 * All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer 11 * in this position and unchanged. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 #include "archive_platform.h" 29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_ar.c 201101 2009-12-28 03:06:27Z kientzle $"); 30 31 #ifdef HAVE_SYS_STAT_H 32 #include <sys/stat.h> 33 #endif 34 #ifdef HAVE_ERRNO_H 35 #include <errno.h> 36 #endif 37 #ifdef HAVE_STDLIB_H 38 #include <stdlib.h> 39 #endif 40 #ifdef HAVE_STRING_H 41 #include <string.h> 42 #endif 43 #ifdef HAVE_LIMITS_H 44 #include <limits.h> 45 #endif 46 47 #include "archive.h" 48 #include "archive_entry.h" 49 #include "archive_private.h" 50 #include "archive_read_private.h" 51 52 struct ar { 53 int64_t entry_bytes_remaining; 54 /* unconsumed is purely to track data we've gotten from readahead, 55 * but haven't yet marked as consumed. Must be paired with 56 * entry_bytes_remaining usage/modification. 57 */ 58 size_t entry_bytes_unconsumed; 59 int64_t entry_offset; 60 int64_t entry_padding; 61 char *strtab; 62 size_t strtab_size; 63 char read_global_header; 64 }; 65 66 /* 67 * Define structure of the "ar" header. 68 */ 69 #define AR_name_offset 0 70 #define AR_name_size 16 71 #define AR_date_offset 16 72 #define AR_date_size 12 73 #define AR_uid_offset 28 74 #define AR_uid_size 6 75 #define AR_gid_offset 34 76 #define AR_gid_size 6 77 #define AR_mode_offset 40 78 #define AR_mode_size 8 79 #define AR_size_offset 48 80 #define AR_size_size 10 81 #define AR_fmag_offset 58 82 #define AR_fmag_size 2 83 84 static int archive_read_format_ar_bid(struct archive_read *a, int); 85 static int archive_read_format_ar_cleanup(struct archive_read *a); 86 static int archive_read_format_ar_read_data(struct archive_read *a, 87 const void **buff, size_t *size, int64_t *offset); 88 static int archive_read_format_ar_skip(struct archive_read *a); 89 static int archive_read_format_ar_read_header(struct archive_read *a, 90 struct archive_entry *e); 91 static uint64_t ar_atol8(const char *p, unsigned char_cnt); 92 static uint64_t ar_atol10(const char *p, unsigned char_cnt); 93 static int ar_parse_gnu_filename_table(struct archive_read *a); 94 static int ar_parse_common_header(struct ar *ar, struct archive_entry *, 95 const char *h); 96 97 int 98 archive_read_support_format_ar(struct archive *_a) 99 { 100 struct archive_read *a = (struct archive_read *)_a; 101 struct ar *ar; 102 int r; 103 104 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 105 ARCHIVE_STATE_NEW, "archive_read_support_format_ar"); 106 107 ar = (struct ar *)calloc(1, sizeof(*ar)); 108 if (ar == NULL) { 109 archive_set_error(&a->archive, ENOMEM, 110 "Can't allocate ar data"); 111 return (ARCHIVE_FATAL); 112 } 113 ar->strtab = NULL; 114 115 r = __archive_read_register_format(a, 116 ar, 117 "ar", 118 archive_read_format_ar_bid, 119 NULL, 120 archive_read_format_ar_read_header, 121 archive_read_format_ar_read_data, 122 archive_read_format_ar_skip, 123 NULL, 124 archive_read_format_ar_cleanup, 125 NULL, 126 NULL); 127 128 if (r != ARCHIVE_OK) { 129 free(ar); 130 return (r); 131 } 132 return (ARCHIVE_OK); 133 } 134 135 static int 136 archive_read_format_ar_cleanup(struct archive_read *a) 137 { 138 struct ar *ar; 139 140 ar = (struct ar *)(a->format->data); 141 free(ar->strtab); 142 free(ar); 143 (a->format->data) = NULL; 144 return (ARCHIVE_OK); 145 } 146 147 static int 148 archive_read_format_ar_bid(struct archive_read *a, int best_bid) 149 { 150 const void *h; 151 152 (void)best_bid; /* UNUSED */ 153 154 /* 155 * Verify the 8-byte file signature. 156 * TODO: Do we need to check more than this? 157 */ 158 if ((h = __archive_read_ahead(a, 8, NULL)) == NULL) 159 return (-1); 160 if (memcmp(h, "!<arch>\n", 8) == 0) { 161 return (64); 162 } 163 return (-1); 164 } 165 166 static int 167 _ar_read_header(struct archive_read *a, struct archive_entry *entry, 168 struct ar *ar, const char *h, size_t *unconsumed) 169 { 170 char filename[AR_name_size + 1]; 171 uint64_t number; /* Used to hold parsed numbers before validation. */ 172 size_t bsd_name_length, entry_size; 173 char *p, *st; 174 const void *b; 175 int r; 176 177 /* Verify the magic signature on the file header. */ 178 if (strncmp(h + AR_fmag_offset, "`\n", 2) != 0) { 179 archive_set_error(&a->archive, EINVAL, 180 "Incorrect file header signature"); 181 return (ARCHIVE_FATAL); 182 } 183 184 /* Copy filename into work buffer. */ 185 strncpy(filename, h + AR_name_offset, AR_name_size); 186 filename[AR_name_size] = '\0'; 187 188 /* 189 * Guess the format variant based on the filename. 190 */ 191 if (a->archive.archive_format == ARCHIVE_FORMAT_AR) { 192 /* We don't already know the variant, so let's guess. */ 193 /* 194 * Biggest clue is presence of '/': GNU starts special 195 * filenames with '/', appends '/' as terminator to 196 * non-special names, so anything with '/' should be 197 * GNU except for BSD long filenames. 198 */ 199 if (strncmp(filename, "#1/", 3) == 0) 200 a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; 201 else if (strchr(filename, '/') != NULL) 202 a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU; 203 else if (strncmp(filename, "__.SYMDEF", 9) == 0) 204 a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; 205 /* 206 * XXX Do GNU/SVR4 'ar' programs ever omit trailing '/' 207 * if name exactly fills 16-byte field? If so, we 208 * can't assume entries without '/' are BSD. XXX 209 */ 210 } 211 212 /* Update format name from the code. */ 213 if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) 214 a->archive.archive_format_name = "ar (GNU/SVR4)"; 215 else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) 216 a->archive.archive_format_name = "ar (BSD)"; 217 else 218 a->archive.archive_format_name = "ar"; 219 220 /* 221 * Remove trailing spaces from the filename. GNU and BSD 222 * variants both pad filename area out with spaces. 223 * This will only be wrong if GNU/SVR4 'ar' implementations 224 * omit trailing '/' for 16-char filenames and we have 225 * a 16-char filename that ends in ' '. 226 */ 227 p = filename + AR_name_size - 1; 228 while (p >= filename && *p == ' ') { 229 *p = '\0'; 230 p--; 231 } 232 233 /* 234 * Remove trailing slash unless first character is '/'. 235 * (BSD entries never end in '/', so this will only trim 236 * GNU-format entries. GNU special entries start with '/' 237 * and are not terminated in '/', so we don't trim anything 238 * that starts with '/'.) 239 */ 240 if (filename[0] != '/' && p > filename && *p == '/') { 241 *p = '\0'; 242 } 243 244 if (p < filename) { 245 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 246 "Found entry with empty filename"); 247 return (ARCHIVE_FATAL); 248 } 249 250 /* 251 * '//' is the GNU filename table. 252 * Later entries can refer to names in this table. 253 */ 254 if (strcmp(filename, "//") == 0) { 255 /* This must come before any call to _read_ahead. */ 256 ar_parse_common_header(ar, entry, h); 257 archive_entry_copy_pathname(entry, filename); 258 archive_entry_set_filetype(entry, AE_IFREG); 259 /* Get the size of the filename table. */ 260 number = ar_atol10(h + AR_size_offset, AR_size_size); 261 if (number > SIZE_MAX || number > 1024 * 1024 * 1024) { 262 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 263 "Filename table too large"); 264 return (ARCHIVE_FATAL); 265 } 266 entry_size = (size_t)number; 267 if (entry_size == 0) { 268 archive_set_error(&a->archive, EINVAL, 269 "Invalid string table"); 270 return (ARCHIVE_FATAL); 271 } 272 if (ar->strtab != NULL) { 273 archive_set_error(&a->archive, EINVAL, 274 "More than one string tables exist"); 275 return (ARCHIVE_FATAL); 276 } 277 278 /* Read the filename table into memory. */ 279 st = malloc(entry_size); 280 if (st == NULL) { 281 archive_set_error(&a->archive, ENOMEM, 282 "Can't allocate filename table buffer"); 283 return (ARCHIVE_FATAL); 284 } 285 ar->strtab = st; 286 ar->strtab_size = entry_size; 287 288 if (*unconsumed) { 289 __archive_read_consume(a, *unconsumed); 290 *unconsumed = 0; 291 } 292 293 if ((b = __archive_read_ahead(a, entry_size, NULL)) == NULL) 294 return (ARCHIVE_FATAL); 295 memcpy(st, b, entry_size); 296 __archive_read_consume(a, entry_size); 297 /* All contents are consumed. */ 298 ar->entry_bytes_remaining = 0; 299 archive_entry_set_size(entry, ar->entry_bytes_remaining); 300 301 /* Parse the filename table. */ 302 return (ar_parse_gnu_filename_table(a)); 303 } 304 305 /* 306 * GNU variant handles long filenames by storing /<number> 307 * to indicate a name stored in the filename table. 308 * XXX TODO: Verify that it's all digits... Don't be fooled 309 * by "/9xyz" XXX 310 */ 311 if (filename[0] == '/' && filename[1] >= '0' && filename[1] <= '9') { 312 number = ar_atol10(h + AR_name_offset + 1, AR_name_size - 1); 313 /* 314 * If we can't look up the real name, warn and return 315 * the entry with the wrong name. 316 */ 317 if (ar->strtab == NULL || number >= ar->strtab_size) { 318 archive_set_error(&a->archive, EINVAL, 319 "Can't find long filename for GNU/SVR4 archive entry"); 320 archive_entry_copy_pathname(entry, filename); 321 /* Parse the time, owner, mode, size fields. */ 322 ar_parse_common_header(ar, entry, h); 323 return (ARCHIVE_FATAL); 324 } 325 326 archive_entry_copy_pathname(entry, &ar->strtab[(size_t)number]); 327 /* Parse the time, owner, mode, size fields. */ 328 return (ar_parse_common_header(ar, entry, h)); 329 } 330 331 /* 332 * BSD handles long filenames by storing "#1/" followed by the 333 * length of filename as a decimal number, then prepends the 334 * the filename to the file contents. 335 */ 336 if (strncmp(filename, "#1/", 3) == 0) { 337 /* Parse the time, owner, mode, size fields. */ 338 /* This must occur before _read_ahead is called again. */ 339 ar_parse_common_header(ar, entry, h); 340 341 /* Parse the size of the name, adjust the file size. */ 342 number = ar_atol10(h + AR_name_offset + 3, AR_name_size - 3); 343 /* Sanity check the filename length: 344 * = Must be <= SIZE_MAX - 1 345 * = Must be <= 1MB 346 * = Cannot be bigger than the entire entry 347 */ 348 if (number > SIZE_MAX - 1 349 || number > 1024 * 1024 350 || (int64_t)number > ar->entry_bytes_remaining) { 351 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 352 "Bad input file size"); 353 return (ARCHIVE_FATAL); 354 } 355 bsd_name_length = (size_t)number; 356 ar->entry_bytes_remaining -= bsd_name_length; 357 /* Adjust file size reported to client. */ 358 archive_entry_set_size(entry, ar->entry_bytes_remaining); 359 360 if (*unconsumed) { 361 __archive_read_consume(a, *unconsumed); 362 *unconsumed = 0; 363 } 364 365 /* Read the long name into memory. */ 366 if ((b = __archive_read_ahead(a, bsd_name_length, NULL)) == NULL) { 367 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 368 "Truncated input file"); 369 return (ARCHIVE_FATAL); 370 } 371 /* Store it in the entry. */ 372 p = (char *)malloc(bsd_name_length + 1); 373 if (p == NULL) { 374 archive_set_error(&a->archive, ENOMEM, 375 "Can't allocate fname buffer"); 376 return (ARCHIVE_FATAL); 377 } 378 strncpy(p, b, bsd_name_length); 379 p[bsd_name_length] = '\0'; 380 381 __archive_read_consume(a, bsd_name_length); 382 383 archive_entry_copy_pathname(entry, p); 384 free(p); 385 return (ARCHIVE_OK); 386 } 387 388 /* 389 * "/" is the SVR4/GNU archive symbol table. 390 * "/SYM64/" is the SVR4/GNU 64-bit variant archive symbol table. 391 */ 392 if (strcmp(filename, "/") == 0 || strcmp(filename, "/SYM64/") == 0) { 393 archive_entry_copy_pathname(entry, filename); 394 /* Parse the time, owner, mode, size fields. */ 395 r = ar_parse_common_header(ar, entry, h); 396 /* Force the file type to a regular file. */ 397 archive_entry_set_filetype(entry, AE_IFREG); 398 return (r); 399 } 400 401 /* 402 * "__.SYMDEF" is a BSD archive symbol table. 403 */ 404 if (strcmp(filename, "__.SYMDEF") == 0) { 405 archive_entry_copy_pathname(entry, filename); 406 /* Parse the time, owner, mode, size fields. */ 407 return (ar_parse_common_header(ar, entry, h)); 408 } 409 410 /* 411 * Otherwise, this is a standard entry. The filename 412 * has already been trimmed as much as possible, based 413 * on our current knowledge of the format. 414 */ 415 archive_entry_copy_pathname(entry, filename); 416 return (ar_parse_common_header(ar, entry, h)); 417 } 418 419 static int 420 archive_read_format_ar_read_header(struct archive_read *a, 421 struct archive_entry *entry) 422 { 423 struct ar *ar = (struct ar*)(a->format->data); 424 size_t unconsumed; 425 const void *header_data; 426 int ret; 427 428 if (!ar->read_global_header) { 429 /* 430 * We are now at the beginning of the archive, 431 * so we need first consume the ar global header. 432 */ 433 __archive_read_consume(a, 8); 434 ar->read_global_header = 1; 435 /* Set a default format code for now. */ 436 a->archive.archive_format = ARCHIVE_FORMAT_AR; 437 } 438 439 /* Read the header for the next file entry. */ 440 if ((header_data = __archive_read_ahead(a, 60, NULL)) == NULL) 441 /* Broken header. */ 442 return (ARCHIVE_EOF); 443 444 unconsumed = 60; 445 446 ret = _ar_read_header(a, entry, ar, (const char *)header_data, &unconsumed); 447 448 if (unconsumed) 449 __archive_read_consume(a, unconsumed); 450 451 return ret; 452 } 453 454 455 static int 456 ar_parse_common_header(struct ar *ar, struct archive_entry *entry, 457 const char *h) 458 { 459 uint64_t n; 460 461 /* Copy remaining header */ 462 archive_entry_set_filetype(entry, AE_IFREG); 463 archive_entry_set_mtime(entry, 464 (time_t)ar_atol10(h + AR_date_offset, AR_date_size), 0L); 465 archive_entry_set_uid(entry, 466 (uid_t)ar_atol10(h + AR_uid_offset, AR_uid_size)); 467 archive_entry_set_gid(entry, 468 (gid_t)ar_atol10(h + AR_gid_offset, AR_gid_size)); 469 archive_entry_set_mode(entry, 470 (mode_t)ar_atol8(h + AR_mode_offset, AR_mode_size)); 471 n = ar_atol10(h + AR_size_offset, AR_size_size); 472 473 ar->entry_offset = 0; 474 ar->entry_padding = n % 2; 475 archive_entry_set_size(entry, n); 476 ar->entry_bytes_remaining = n; 477 return (ARCHIVE_OK); 478 } 479 480 static int 481 archive_read_format_ar_read_data(struct archive_read *a, 482 const void **buff, size_t *size, int64_t *offset) 483 { 484 ssize_t bytes_read; 485 struct ar *ar; 486 487 ar = (struct ar *)(a->format->data); 488 489 if (ar->entry_bytes_unconsumed) { 490 __archive_read_consume(a, ar->entry_bytes_unconsumed); 491 ar->entry_bytes_unconsumed = 0; 492 } 493 494 if (ar->entry_bytes_remaining > 0) { 495 *buff = __archive_read_ahead(a, 1, &bytes_read); 496 if (bytes_read == 0) { 497 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 498 "Truncated ar archive"); 499 return (ARCHIVE_FATAL); 500 } 501 if (bytes_read < 0) 502 return (ARCHIVE_FATAL); 503 if (bytes_read > ar->entry_bytes_remaining) 504 bytes_read = (ssize_t)ar->entry_bytes_remaining; 505 *size = bytes_read; 506 ar->entry_bytes_unconsumed = bytes_read; 507 *offset = ar->entry_offset; 508 ar->entry_offset += bytes_read; 509 ar->entry_bytes_remaining -= bytes_read; 510 return (ARCHIVE_OK); 511 } else { 512 int64_t skipped = __archive_read_consume(a, ar->entry_padding); 513 if (skipped >= 0) { 514 ar->entry_padding -= skipped; 515 } 516 if (ar->entry_padding) { 517 if (skipped >= 0) { 518 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 519 "Truncated ar archive- failed consuming padding"); 520 } 521 return (ARCHIVE_FATAL); 522 } 523 *buff = NULL; 524 *size = 0; 525 *offset = ar->entry_offset; 526 return (ARCHIVE_EOF); 527 } 528 } 529 530 static int 531 archive_read_format_ar_skip(struct archive_read *a) 532 { 533 int64_t bytes_skipped; 534 struct ar* ar; 535 536 ar = (struct ar *)(a->format->data); 537 538 bytes_skipped = __archive_read_consume(a, 539 ar->entry_bytes_remaining + ar->entry_padding 540 + ar->entry_bytes_unconsumed); 541 if (bytes_skipped < 0) 542 return (ARCHIVE_FATAL); 543 544 ar->entry_bytes_remaining = 0; 545 ar->entry_bytes_unconsumed = 0; 546 ar->entry_padding = 0; 547 548 return (ARCHIVE_OK); 549 } 550 551 static int 552 ar_parse_gnu_filename_table(struct archive_read *a) 553 { 554 struct ar *ar; 555 char *p; 556 size_t size; 557 558 ar = (struct ar*)(a->format->data); 559 size = ar->strtab_size; 560 561 for (p = ar->strtab; p < ar->strtab + size - 1; ++p) { 562 if (*p == '/') { 563 *p++ = '\0'; 564 if (*p != '\n') 565 goto bad_string_table; 566 *p = '\0'; 567 } 568 } 569 /* 570 * GNU ar always pads the table to an even size. 571 * The pad character is either '\n' or '`'. 572 */ 573 if (p != ar->strtab + size && *p != '\n' && *p != '`') 574 goto bad_string_table; 575 576 /* Enforce zero termination. */ 577 ar->strtab[size - 1] = '\0'; 578 579 return (ARCHIVE_OK); 580 581 bad_string_table: 582 archive_set_error(&a->archive, EINVAL, 583 "Invalid string table"); 584 free(ar->strtab); 585 ar->strtab = NULL; 586 return (ARCHIVE_FATAL); 587 } 588 589 static uint64_t 590 ar_atol8(const char *p, unsigned char_cnt) 591 { 592 uint64_t l, limit, last_digit_limit; 593 unsigned int digit, base; 594 595 base = 8; 596 limit = UINT64_MAX / base; 597 last_digit_limit = UINT64_MAX % base; 598 599 while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) 600 p++; 601 602 l = 0; 603 digit = *p - '0'; 604 while (*p >= '0' && digit < base && char_cnt-- > 0) { 605 if (l>limit || (l == limit && digit > last_digit_limit)) { 606 l = UINT64_MAX; /* Truncate on overflow. */ 607 break; 608 } 609 l = (l * base) + digit; 610 digit = *++p - '0'; 611 } 612 return (l); 613 } 614 615 static uint64_t 616 ar_atol10(const char *p, unsigned char_cnt) 617 { 618 uint64_t l, limit, last_digit_limit; 619 unsigned int base, digit; 620 621 base = 10; 622 limit = UINT64_MAX / base; 623 last_digit_limit = UINT64_MAX % base; 624 625 while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) 626 p++; 627 l = 0; 628 digit = *p - '0'; 629 while (*p >= '0' && digit < base && char_cnt-- > 0) { 630 if (l > limit || (l == limit && digit > last_digit_limit)) { 631 l = UINT64_MAX; /* Truncate on overflow. */ 632 break; 633 } 634 l = (l * base) + digit; 635 digit = *++p - '0'; 636 } 637 return (l); 638 } 639