1 /*- 2 * Copyright (c) 2003-2007 Tim Kientzle 3 * Copyright (c) 2010-2011 Michihiro NAKAJIMA 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 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 18 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 25 */ 26 27 #include "archive_platform.h" 28 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_cpio.c 201163 2009-12-29 05:50:34Z kientzle $"); 29 30 #ifdef HAVE_ERRNO_H 31 #include <errno.h> 32 #endif 33 /* #include <stdint.h> */ /* See archive_platform.h */ 34 #ifdef HAVE_STDLIB_H 35 #include <stdlib.h> 36 #endif 37 #ifdef HAVE_STRING_H 38 #include <string.h> 39 #endif 40 41 #include "archive.h" 42 #include "archive_entry.h" 43 #include "archive_entry_locale.h" 44 #include "archive_private.h" 45 #include "archive_read_private.h" 46 47 #define bin_magic_offset 0 48 #define bin_magic_size 2 49 #define bin_dev_offset 2 50 #define bin_dev_size 2 51 #define bin_ino_offset 4 52 #define bin_ino_size 2 53 #define bin_mode_offset 6 54 #define bin_mode_size 2 55 #define bin_uid_offset 8 56 #define bin_uid_size 2 57 #define bin_gid_offset 10 58 #define bin_gid_size 2 59 #define bin_nlink_offset 12 60 #define bin_nlink_size 2 61 #define bin_rdev_offset 14 62 #define bin_rdev_size 2 63 #define bin_mtime_offset 16 64 #define bin_mtime_size 4 65 #define bin_namesize_offset 20 66 #define bin_namesize_size 2 67 #define bin_filesize_offset 22 68 #define bin_filesize_size 4 69 #define bin_header_size 26 70 71 #define odc_magic_offset 0 72 #define odc_magic_size 6 73 #define odc_dev_offset 6 74 #define odc_dev_size 6 75 #define odc_ino_offset 12 76 #define odc_ino_size 6 77 #define odc_mode_offset 18 78 #define odc_mode_size 6 79 #define odc_uid_offset 24 80 #define odc_uid_size 6 81 #define odc_gid_offset 30 82 #define odc_gid_size 6 83 #define odc_nlink_offset 36 84 #define odc_nlink_size 6 85 #define odc_rdev_offset 42 86 #define odc_rdev_size 6 87 #define odc_mtime_offset 48 88 #define odc_mtime_size 11 89 #define odc_namesize_offset 59 90 #define odc_namesize_size 6 91 #define odc_filesize_offset 65 92 #define odc_filesize_size 11 93 #define odc_header_size 76 94 95 #define newc_magic_offset 0 96 #define newc_magic_size 6 97 #define newc_ino_offset 6 98 #define newc_ino_size 8 99 #define newc_mode_offset 14 100 #define newc_mode_size 8 101 #define newc_uid_offset 22 102 #define newc_uid_size 8 103 #define newc_gid_offset 30 104 #define newc_gid_size 8 105 #define newc_nlink_offset 38 106 #define newc_nlink_size 8 107 #define newc_mtime_offset 46 108 #define newc_mtime_size 8 109 #define newc_filesize_offset 54 110 #define newc_filesize_size 8 111 #define newc_devmajor_offset 62 112 #define newc_devmajor_size 8 113 #define newc_devminor_offset 70 114 #define newc_devminor_size 8 115 #define newc_rdevmajor_offset 78 116 #define newc_rdevmajor_size 8 117 #define newc_rdevminor_offset 86 118 #define newc_rdevminor_size 8 119 #define newc_namesize_offset 94 120 #define newc_namesize_size 8 121 #define newc_checksum_offset 102 122 #define newc_checksum_size 8 123 #define newc_header_size 110 124 125 /* 126 * An afio large ASCII header, which they named itself. 127 * afio utility uses this header, if a file size is larger than 2G bytes 128 * or inode/uid/gid is bigger than 65535(0xFFFF) or mtime is bigger than 129 * 0x7fffffff, which we cannot record to odc header because of its limit. 130 * If not, uses odc header. 131 */ 132 #define afiol_magic_offset 0 133 #define afiol_magic_size 6 134 #define afiol_dev_offset 6 135 #define afiol_dev_size 8 /* hex */ 136 #define afiol_ino_offset 14 137 #define afiol_ino_size 16 /* hex */ 138 #define afiol_ino_m_offset 30 /* 'm' */ 139 #define afiol_mode_offset 31 140 #define afiol_mode_size 6 /* oct */ 141 #define afiol_uid_offset 37 142 #define afiol_uid_size 8 /* hex */ 143 #define afiol_gid_offset 45 144 #define afiol_gid_size 8 /* hex */ 145 #define afiol_nlink_offset 53 146 #define afiol_nlink_size 8 /* hex */ 147 #define afiol_rdev_offset 61 148 #define afiol_rdev_size 8 /* hex */ 149 #define afiol_mtime_offset 69 150 #define afiol_mtime_size 16 /* hex */ 151 #define afiol_mtime_n_offset 85 /* 'n' */ 152 #define afiol_namesize_offset 86 153 #define afiol_namesize_size 4 /* hex */ 154 #define afiol_flag_offset 90 155 #define afiol_flag_size 4 /* hex */ 156 #define afiol_xsize_offset 94 157 #define afiol_xsize_size 4 /* hex */ 158 #define afiol_xsize_s_offset 98 /* 's' */ 159 #define afiol_filesize_offset 99 160 #define afiol_filesize_size 16 /* hex */ 161 #define afiol_filesize_c_offset 115 /* ':' */ 162 #define afiol_header_size 116 163 164 165 struct links_entry { 166 struct links_entry *next; 167 struct links_entry *previous; 168 int links; 169 dev_t dev; 170 int64_t ino; 171 char *name; 172 }; 173 174 #define CPIO_MAGIC 0x13141516 175 struct cpio { 176 int magic; 177 int (*read_header)(struct archive_read *, struct cpio *, 178 struct archive_entry *, size_t *, size_t *); 179 struct links_entry *links_head; 180 int64_t entry_bytes_remaining; 181 int64_t entry_bytes_unconsumed; 182 int64_t entry_offset; 183 int64_t entry_padding; 184 185 struct archive_string_conv *opt_sconv; 186 struct archive_string_conv *sconv_default; 187 int init_default_conversion; 188 }; 189 190 static int64_t atol16(const char *, unsigned); 191 static int64_t atol8(const char *, unsigned); 192 static int archive_read_format_cpio_bid(struct archive_read *, int); 193 static int archive_read_format_cpio_options(struct archive_read *, 194 const char *, const char *); 195 static int archive_read_format_cpio_cleanup(struct archive_read *); 196 static int archive_read_format_cpio_read_data(struct archive_read *, 197 const void **, size_t *, int64_t *); 198 static int archive_read_format_cpio_read_header(struct archive_read *, 199 struct archive_entry *); 200 static int archive_read_format_cpio_skip(struct archive_read *); 201 static int be4(const unsigned char *); 202 static int find_odc_header(struct archive_read *); 203 static int find_newc_header(struct archive_read *); 204 static int header_bin_be(struct archive_read *, struct cpio *, 205 struct archive_entry *, size_t *, size_t *); 206 static int header_bin_le(struct archive_read *, struct cpio *, 207 struct archive_entry *, size_t *, size_t *); 208 static int header_newc(struct archive_read *, struct cpio *, 209 struct archive_entry *, size_t *, size_t *); 210 static int header_odc(struct archive_read *, struct cpio *, 211 struct archive_entry *, size_t *, size_t *); 212 static int header_afiol(struct archive_read *, struct cpio *, 213 struct archive_entry *, size_t *, size_t *); 214 static int is_octal(const char *, size_t); 215 static int is_hex(const char *, size_t); 216 static int le4(const unsigned char *); 217 static int record_hardlink(struct archive_read *a, 218 struct cpio *cpio, struct archive_entry *entry); 219 220 int 221 archive_read_support_format_cpio(struct archive *_a) 222 { 223 struct archive_read *a = (struct archive_read *)_a; 224 struct cpio *cpio; 225 int r; 226 227 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 228 ARCHIVE_STATE_NEW, "archive_read_support_format_cpio"); 229 230 cpio = (struct cpio *)calloc(1, sizeof(*cpio)); 231 if (cpio == NULL) { 232 archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); 233 return (ARCHIVE_FATAL); 234 } 235 cpio->magic = CPIO_MAGIC; 236 237 r = __archive_read_register_format(a, 238 cpio, 239 "cpio", 240 archive_read_format_cpio_bid, 241 archive_read_format_cpio_options, 242 archive_read_format_cpio_read_header, 243 archive_read_format_cpio_read_data, 244 archive_read_format_cpio_skip, 245 archive_read_format_cpio_cleanup); 246 247 if (r != ARCHIVE_OK) 248 free(cpio); 249 return (ARCHIVE_OK); 250 } 251 252 253 static int 254 archive_read_format_cpio_bid(struct archive_read *a, int best_bid) 255 { 256 const unsigned char *p; 257 struct cpio *cpio; 258 int bid; 259 260 (void)best_bid; /* UNUSED */ 261 262 cpio = (struct cpio *)(a->format->data); 263 264 if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) 265 return (-1); 266 267 bid = 0; 268 if (memcmp(p, "070707", 6) == 0) { 269 /* ASCII cpio archive (odc, POSIX.1) */ 270 cpio->read_header = header_odc; 271 bid += 48; 272 /* 273 * XXX TODO: More verification; Could check that only octal 274 * digits appear in appropriate header locations. XXX 275 */ 276 } else if (memcmp(p, "070727", 6) == 0) { 277 /* afio large ASCII cpio archive */ 278 cpio->read_header = header_odc; 279 bid += 48; 280 /* 281 * XXX TODO: More verification; Could check that almost hex 282 * digits appear in appropriate header locations. XXX 283 */ 284 } else if (memcmp(p, "070701", 6) == 0) { 285 /* ASCII cpio archive (SVR4 without CRC) */ 286 cpio->read_header = header_newc; 287 bid += 48; 288 /* 289 * XXX TODO: More verification; Could check that only hex 290 * digits appear in appropriate header locations. XXX 291 */ 292 } else if (memcmp(p, "070702", 6) == 0) { 293 /* ASCII cpio archive (SVR4 with CRC) */ 294 /* XXX TODO: Flag that we should check the CRC. XXX */ 295 cpio->read_header = header_newc; 296 bid += 48; 297 /* 298 * XXX TODO: More verification; Could check that only hex 299 * digits appear in appropriate header locations. XXX 300 */ 301 } else if (p[0] * 256 + p[1] == 070707) { 302 /* big-endian binary cpio archives */ 303 cpio->read_header = header_bin_be; 304 bid += 16; 305 /* Is more verification possible here? */ 306 } else if (p[0] + p[1] * 256 == 070707) { 307 /* little-endian binary cpio archives */ 308 cpio->read_header = header_bin_le; 309 bid += 16; 310 /* Is more verification possible here? */ 311 } else 312 return (ARCHIVE_WARN); 313 314 return (bid); 315 } 316 317 static int 318 archive_read_format_cpio_options(struct archive_read *a, 319 const char *key, const char *val) 320 { 321 struct cpio *cpio; 322 int ret = ARCHIVE_FAILED; 323 324 cpio = (struct cpio *)(a->format->data); 325 if (strcmp(key, "compat-2x") == 0) { 326 /* Handle filnames as libarchive 2.x */ 327 cpio->init_default_conversion = (val != NULL)?1:0; 328 ret = ARCHIVE_OK; 329 } else if (strcmp(key, "hdrcharset") == 0) { 330 if (val == NULL || val[0] == 0) 331 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 332 "cpio: hdrcharset option needs a character-set name"); 333 else { 334 cpio->opt_sconv = 335 archive_string_conversion_from_charset( 336 &a->archive, val, 0); 337 if (cpio->opt_sconv != NULL) 338 ret = ARCHIVE_OK; 339 else 340 ret = ARCHIVE_FATAL; 341 } 342 } else 343 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 344 "cpio: unknown keyword ``%s''", key); 345 346 return (ret); 347 } 348 349 static int 350 archive_read_format_cpio_read_header(struct archive_read *a, 351 struct archive_entry *entry) 352 { 353 struct cpio *cpio; 354 const void *h; 355 struct archive_string_conv *sconv; 356 size_t namelength; 357 size_t name_pad; 358 int r; 359 360 cpio = (struct cpio *)(a->format->data); 361 sconv = cpio->opt_sconv; 362 if (sconv == NULL) { 363 if (!cpio->init_default_conversion) { 364 cpio->sconv_default = 365 archive_string_default_conversion_for_read( 366 &(a->archive)); 367 cpio->init_default_conversion = 1; 368 } 369 sconv = cpio->sconv_default; 370 } 371 372 r = (cpio->read_header(a, cpio, entry, &namelength, &name_pad)); 373 374 if (r < ARCHIVE_WARN) 375 return (r); 376 377 /* Read name from buffer. */ 378 h = __archive_read_ahead(a, namelength + name_pad, NULL); 379 if (h == NULL) 380 return (ARCHIVE_FATAL); 381 if (archive_entry_copy_pathname_l(entry, 382 (const char *)h, namelength, sconv) != 0) { 383 if (errno == ENOMEM) { 384 archive_set_error(&a->archive, ENOMEM, 385 "Can't allocate memory for Pathname"); 386 return (ARCHIVE_FATAL); 387 } 388 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 389 "Pathname can't be converted from %s to current locale.", 390 archive_string_conversion_charset_name(sconv)); 391 r = ARCHIVE_WARN; 392 } 393 cpio->entry_offset = 0; 394 395 __archive_read_consume(a, namelength + name_pad); 396 397 /* If this is a symlink, read the link contents. */ 398 if (archive_entry_filetype(entry) == AE_IFLNK) { 399 h = __archive_read_ahead(a, cpio->entry_bytes_remaining, NULL); 400 if (h == NULL) 401 return (ARCHIVE_FATAL); 402 if (archive_entry_copy_symlink_l(entry, (const char *)h, 403 cpio->entry_bytes_remaining, sconv) != 0) { 404 if (errno == ENOMEM) { 405 archive_set_error(&a->archive, ENOMEM, 406 "Can't allocate memory for Linkname"); 407 return (ARCHIVE_FATAL); 408 } 409 archive_set_error(&a->archive, 410 ARCHIVE_ERRNO_FILE_FORMAT, 411 "Linkname can't be converted from %s to " 412 "current locale.", 413 archive_string_conversion_charset_name(sconv)); 414 r = ARCHIVE_WARN; 415 } 416 __archive_read_consume(a, cpio->entry_bytes_remaining); 417 cpio->entry_bytes_remaining = 0; 418 } 419 420 /* XXX TODO: If the full mode is 0160200, then this is a Solaris 421 * ACL description for the following entry. Read this body 422 * and parse it as a Solaris-style ACL, then read the next 423 * header. XXX */ 424 425 /* Compare name to "TRAILER!!!" to test for end-of-archive. */ 426 if (namelength == 11 && strcmp((const char *)h, "TRAILER!!!") == 0) { 427 /* TODO: Store file location of start of block. */ 428 archive_clear_error(&a->archive); 429 return (ARCHIVE_EOF); 430 } 431 432 /* Detect and record hardlinks to previously-extracted entries. */ 433 if (record_hardlink(a, cpio, entry) != ARCHIVE_OK) { 434 return (ARCHIVE_FATAL); 435 } 436 437 return (r); 438 } 439 440 static int 441 archive_read_format_cpio_read_data(struct archive_read *a, 442 const void **buff, size_t *size, int64_t *offset) 443 { 444 ssize_t bytes_read; 445 struct cpio *cpio; 446 447 cpio = (struct cpio *)(a->format->data); 448 449 if (cpio->entry_bytes_unconsumed) { 450 __archive_read_consume(a, cpio->entry_bytes_unconsumed); 451 cpio->entry_bytes_unconsumed = 0; 452 } 453 454 if (cpio->entry_bytes_remaining > 0) { 455 *buff = __archive_read_ahead(a, 1, &bytes_read); 456 if (bytes_read <= 0) 457 return (ARCHIVE_FATAL); 458 if (bytes_read > cpio->entry_bytes_remaining) 459 bytes_read = cpio->entry_bytes_remaining; 460 *size = bytes_read; 461 cpio->entry_bytes_unconsumed = bytes_read; 462 *offset = cpio->entry_offset; 463 cpio->entry_offset += bytes_read; 464 cpio->entry_bytes_remaining -= bytes_read; 465 return (ARCHIVE_OK); 466 } else { 467 if (cpio->entry_padding != 468 __archive_read_consume(a, cpio->entry_padding)) { 469 return (ARCHIVE_FATAL); 470 } 471 cpio->entry_padding = 0; 472 *buff = NULL; 473 *size = 0; 474 *offset = cpio->entry_offset; 475 return (ARCHIVE_EOF); 476 } 477 } 478 479 static int 480 archive_read_format_cpio_skip(struct archive_read *a) 481 { 482 struct cpio *cpio = (struct cpio *)(a->format->data); 483 int64_t to_skip = cpio->entry_bytes_remaining + cpio->entry_padding + 484 cpio->entry_bytes_unconsumed; 485 486 if (to_skip != __archive_read_consume(a, to_skip)) { 487 return (ARCHIVE_FATAL); 488 } 489 cpio->entry_bytes_remaining = 0; 490 cpio->entry_padding = 0; 491 cpio->entry_bytes_unconsumed = 0; 492 return (ARCHIVE_OK); 493 } 494 495 /* 496 * Skip forward to the next cpio newc header by searching for the 497 * 07070[12] string. This should be generalized and merged with 498 * find_odc_header below. 499 */ 500 static int 501 is_hex(const char *p, size_t len) 502 { 503 while (len-- > 0) { 504 if ((*p >= '0' && *p <= '9') 505 || (*p >= 'a' && *p <= 'f') 506 || (*p >= 'A' && *p <= 'F')) 507 ++p; 508 else 509 return (0); 510 } 511 return (1); 512 } 513 514 static int 515 find_newc_header(struct archive_read *a) 516 { 517 const void *h; 518 const char *p, *q; 519 size_t skip, skipped = 0; 520 ssize_t bytes; 521 522 for (;;) { 523 h = __archive_read_ahead(a, newc_header_size, &bytes); 524 if (h == NULL) 525 return (ARCHIVE_FATAL); 526 p = h; 527 q = p + bytes; 528 529 /* Try the typical case first, then go into the slow search.*/ 530 if (memcmp("07070", p, 5) == 0 531 && (p[5] == '1' || p[5] == '2') 532 && is_hex(p, newc_header_size)) 533 return (ARCHIVE_OK); 534 535 /* 536 * Scan ahead until we find something that looks 537 * like a newc header. 538 */ 539 while (p + newc_header_size <= q) { 540 switch (p[5]) { 541 case '1': 542 case '2': 543 if (memcmp("07070", p, 5) == 0 544 && is_hex(p, newc_header_size)) { 545 skip = p - (const char *)h; 546 __archive_read_consume(a, skip); 547 skipped += skip; 548 if (skipped > 0) { 549 archive_set_error(&a->archive, 550 0, 551 "Skipped %d bytes before " 552 "finding valid header", 553 (int)skipped); 554 return (ARCHIVE_WARN); 555 } 556 return (ARCHIVE_OK); 557 } 558 p += 2; 559 break; 560 case '0': 561 p++; 562 break; 563 default: 564 p += 6; 565 break; 566 } 567 } 568 skip = p - (const char *)h; 569 __archive_read_consume(a, skip); 570 skipped += skip; 571 } 572 } 573 574 static int 575 header_newc(struct archive_read *a, struct cpio *cpio, 576 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 577 { 578 const void *h; 579 const char *header; 580 int r; 581 582 r = find_newc_header(a); 583 if (r < ARCHIVE_WARN) 584 return (r); 585 586 /* Read fixed-size portion of header. */ 587 h = __archive_read_ahead(a, newc_header_size, NULL); 588 if (h == NULL) 589 return (ARCHIVE_FATAL); 590 591 /* Parse out hex fields. */ 592 header = (const char *)h; 593 594 if (memcmp(header + newc_magic_offset, "070701", 6) == 0) { 595 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_NOCRC; 596 a->archive.archive_format_name = "ASCII cpio (SVR4 with no CRC)"; 597 } else if (memcmp(header + newc_magic_offset, "070702", 6) == 0) { 598 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_CRC; 599 a->archive.archive_format_name = "ASCII cpio (SVR4 with CRC)"; 600 } else { 601 /* TODO: Abort here? */ 602 } 603 604 archive_entry_set_devmajor(entry, atol16(header + newc_devmajor_offset, newc_devmajor_size)); 605 archive_entry_set_devminor(entry, atol16(header + newc_devminor_offset, newc_devminor_size)); 606 archive_entry_set_ino(entry, atol16(header + newc_ino_offset, newc_ino_size)); 607 archive_entry_set_mode(entry, atol16(header + newc_mode_offset, newc_mode_size)); 608 archive_entry_set_uid(entry, atol16(header + newc_uid_offset, newc_uid_size)); 609 archive_entry_set_gid(entry, atol16(header + newc_gid_offset, newc_gid_size)); 610 archive_entry_set_nlink(entry, atol16(header + newc_nlink_offset, newc_nlink_size)); 611 archive_entry_set_rdevmajor(entry, atol16(header + newc_rdevmajor_offset, newc_rdevmajor_size)); 612 archive_entry_set_rdevminor(entry, atol16(header + newc_rdevminor_offset, newc_rdevminor_size)); 613 archive_entry_set_mtime(entry, atol16(header + newc_mtime_offset, newc_mtime_size), 0); 614 *namelength = atol16(header + newc_namesize_offset, newc_namesize_size); 615 /* Pad name to 2 more than a multiple of 4. */ 616 *name_pad = (2 - *namelength) & 3; 617 618 /* 619 * Note: entry_bytes_remaining is at least 64 bits and 620 * therefore guaranteed to be big enough for a 33-bit file 621 * size. 622 */ 623 cpio->entry_bytes_remaining = 624 atol16(header + newc_filesize_offset, newc_filesize_size); 625 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 626 /* Pad file contents to a multiple of 4. */ 627 cpio->entry_padding = 3 & -cpio->entry_bytes_remaining; 628 __archive_read_consume(a, newc_header_size); 629 return (r); 630 } 631 632 /* 633 * Skip forward to the next cpio odc header by searching for the 634 * 070707 string. This is a hand-optimized search that could 635 * probably be easily generalized to handle all character-based 636 * cpio variants. 637 */ 638 static int 639 is_octal(const char *p, size_t len) 640 { 641 while (len-- > 0) { 642 if (*p < '0' || *p > '7') 643 return (0); 644 ++p; 645 } 646 return (1); 647 } 648 649 static int 650 is_afio_large(const char *h, size_t len) 651 { 652 if (len < afiol_header_size) 653 return (0); 654 if (h[afiol_ino_m_offset] != 'm' 655 || h[afiol_mtime_n_offset] != 'n' 656 || h[afiol_xsize_s_offset] != 's' 657 || h[afiol_filesize_c_offset] != ':') 658 return (0); 659 if (!is_hex(h + afiol_dev_offset, afiol_ino_m_offset - afiol_dev_offset)) 660 return (0); 661 if (!is_hex(h + afiol_mode_offset, afiol_mtime_n_offset - afiol_mode_offset)) 662 return (0); 663 if (!is_hex(h + afiol_namesize_offset, afiol_xsize_s_offset - afiol_namesize_offset)) 664 return (0); 665 if (!is_hex(h + afiol_filesize_offset, afiol_filesize_size)) 666 return (0); 667 return (1); 668 } 669 670 static int 671 find_odc_header(struct archive_read *a) 672 { 673 const void *h; 674 const char *p, *q; 675 size_t skip, skipped = 0; 676 ssize_t bytes; 677 678 for (;;) { 679 h = __archive_read_ahead(a, odc_header_size, &bytes); 680 if (h == NULL) 681 return (ARCHIVE_FATAL); 682 p = h; 683 q = p + bytes; 684 685 /* Try the typical case first, then go into the slow search.*/ 686 if (memcmp("070707", p, 6) == 0 && is_octal(p, odc_header_size)) 687 return (ARCHIVE_OK); 688 if (memcmp("070727", p, 6) == 0 && is_afio_large(p, bytes)) { 689 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 690 return (ARCHIVE_OK); 691 } 692 693 /* 694 * Scan ahead until we find something that looks 695 * like an odc header. 696 */ 697 while (p + odc_header_size <= q) { 698 switch (p[5]) { 699 case '7': 700 if ((memcmp("070707", p, 6) == 0 701 && is_octal(p, odc_header_size)) 702 || (memcmp("070727", p, 6) == 0 703 && is_afio_large(p, q - p))) { 704 skip = p - (const char *)h; 705 __archive_read_consume(a, skip); 706 skipped += skip; 707 if (p[4] == '2') 708 a->archive.archive_format = 709 ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 710 if (skipped > 0) { 711 archive_set_error(&a->archive, 712 0, 713 "Skipped %d bytes before " 714 "finding valid header", 715 (int)skipped); 716 return (ARCHIVE_WARN); 717 } 718 return (ARCHIVE_OK); 719 } 720 p += 2; 721 break; 722 case '0': 723 p++; 724 break; 725 default: 726 p += 6; 727 break; 728 } 729 } 730 skip = p - (const char *)h; 731 __archive_read_consume(a, skip); 732 skipped += skip; 733 } 734 } 735 736 static int 737 header_odc(struct archive_read *a, struct cpio *cpio, 738 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 739 { 740 const void *h; 741 int r; 742 const char *header; 743 744 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_POSIX; 745 a->archive.archive_format_name = "POSIX octet-oriented cpio"; 746 747 /* Find the start of the next header. */ 748 r = find_odc_header(a); 749 if (r < ARCHIVE_WARN) 750 return (r); 751 752 if (a->archive.archive_format == ARCHIVE_FORMAT_CPIO_AFIO_LARGE) { 753 int r2 = (header_afiol(a, cpio, entry, namelength, name_pad)); 754 if (r2 == ARCHIVE_OK) 755 return (r); 756 else 757 return (r2); 758 } 759 760 /* Read fixed-size portion of header. */ 761 h = __archive_read_ahead(a, odc_header_size, NULL); 762 if (h == NULL) 763 return (ARCHIVE_FATAL); 764 765 /* Parse out octal fields. */ 766 header = (const char *)h; 767 768 archive_entry_set_dev(entry, atol8(header + odc_dev_offset, odc_dev_size)); 769 archive_entry_set_ino(entry, atol8(header + odc_ino_offset, odc_ino_size)); 770 archive_entry_set_mode(entry, atol8(header + odc_mode_offset, odc_mode_size)); 771 archive_entry_set_uid(entry, atol8(header + odc_uid_offset, odc_uid_size)); 772 archive_entry_set_gid(entry, atol8(header + odc_gid_offset, odc_gid_size)); 773 archive_entry_set_nlink(entry, atol8(header + odc_nlink_offset, odc_nlink_size)); 774 archive_entry_set_rdev(entry, atol8(header + odc_rdev_offset, odc_rdev_size)); 775 archive_entry_set_mtime(entry, atol8(header + odc_mtime_offset, odc_mtime_size), 0); 776 *namelength = atol8(header + odc_namesize_offset, odc_namesize_size); 777 *name_pad = 0; /* No padding of filename. */ 778 779 /* 780 * Note: entry_bytes_remaining is at least 64 bits and 781 * therefore guaranteed to be big enough for a 33-bit file 782 * size. 783 */ 784 cpio->entry_bytes_remaining = 785 atol8(header + odc_filesize_offset, odc_filesize_size); 786 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 787 cpio->entry_padding = 0; 788 __archive_read_consume(a, odc_header_size); 789 return (r); 790 } 791 792 /* 793 * NOTE: if a filename suffix is ".z", it is the file gziped by afio. 794 * it would be nice that we can show uncompressed file size and we can 795 * uncompressed file contents automatically, unfortunately we have nothing 796 * to get a uncompressed file size while reading each header. it means 797 * we also cannot uncompressed file contens under the our framework. 798 */ 799 static int 800 header_afiol(struct archive_read *a, struct cpio *cpio, 801 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 802 { 803 const void *h; 804 const char *header; 805 806 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 807 a->archive.archive_format_name = "afio large ASCII"; 808 809 /* Read fixed-size portion of header. */ 810 h = __archive_read_ahead(a, afiol_header_size, NULL); 811 if (h == NULL) 812 return (ARCHIVE_FATAL); 813 814 /* Parse out octal fields. */ 815 header = (const char *)h; 816 817 archive_entry_set_dev(entry, atol16(header + afiol_dev_offset, afiol_dev_size)); 818 archive_entry_set_ino(entry, atol16(header + afiol_ino_offset, afiol_ino_size)); 819 archive_entry_set_mode(entry, atol8(header + afiol_mode_offset, afiol_mode_size)); 820 archive_entry_set_uid(entry, atol16(header + afiol_uid_offset, afiol_uid_size)); 821 archive_entry_set_gid(entry, atol16(header + afiol_gid_offset, afiol_gid_size)); 822 archive_entry_set_nlink(entry, atol16(header + afiol_nlink_offset, afiol_nlink_size)); 823 archive_entry_set_rdev(entry, atol16(header + afiol_rdev_offset, afiol_rdev_size)); 824 archive_entry_set_mtime(entry, atol16(header + afiol_mtime_offset, afiol_mtime_size), 0); 825 *namelength = atol16(header + afiol_namesize_offset, afiol_namesize_size); 826 *name_pad = 0; /* No padding of filename. */ 827 828 cpio->entry_bytes_remaining = 829 atol16(header + afiol_filesize_offset, afiol_filesize_size); 830 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 831 cpio->entry_padding = 0; 832 __archive_read_consume(a, afiol_header_size); 833 return (ARCHIVE_OK); 834 } 835 836 837 static int 838 header_bin_le(struct archive_read *a, struct cpio *cpio, 839 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 840 { 841 const void *h; 842 const unsigned char *header; 843 844 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_LE; 845 a->archive.archive_format_name = "cpio (little-endian binary)"; 846 847 /* Read fixed-size portion of header. */ 848 h = __archive_read_ahead(a, bin_header_size, NULL); 849 if (h == NULL) 850 return (ARCHIVE_FATAL); 851 852 /* Parse out binary fields. */ 853 header = (const unsigned char *)h; 854 855 archive_entry_set_dev(entry, header[bin_dev_offset] + header[bin_dev_offset + 1] * 256); 856 archive_entry_set_ino(entry, header[bin_ino_offset] + header[bin_ino_offset + 1] * 256); 857 archive_entry_set_mode(entry, header[bin_mode_offset] + header[bin_mode_offset + 1] * 256); 858 archive_entry_set_uid(entry, header[bin_uid_offset] + header[bin_uid_offset + 1] * 256); 859 archive_entry_set_gid(entry, header[bin_gid_offset] + header[bin_gid_offset + 1] * 256); 860 archive_entry_set_nlink(entry, header[bin_nlink_offset] + header[bin_nlink_offset + 1] * 256); 861 archive_entry_set_rdev(entry, header[bin_rdev_offset] + header[bin_rdev_offset + 1] * 256); 862 archive_entry_set_mtime(entry, le4(header + bin_mtime_offset), 0); 863 *namelength = header[bin_namesize_offset] + header[bin_namesize_offset + 1] * 256; 864 *name_pad = *namelength & 1; /* Pad to even. */ 865 866 cpio->entry_bytes_remaining = le4(header + bin_filesize_offset); 867 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 868 cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ 869 __archive_read_consume(a, bin_header_size); 870 return (ARCHIVE_OK); 871 } 872 873 static int 874 header_bin_be(struct archive_read *a, struct cpio *cpio, 875 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 876 { 877 const void *h; 878 const unsigned char *header; 879 880 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_BE; 881 a->archive.archive_format_name = "cpio (big-endian binary)"; 882 883 /* Read fixed-size portion of header. */ 884 h = __archive_read_ahead(a, bin_header_size, NULL); 885 if (h == NULL) 886 return (ARCHIVE_FATAL); 887 888 /* Parse out binary fields. */ 889 header = (const unsigned char *)h; 890 891 archive_entry_set_dev(entry, header[bin_dev_offset] * 256 + header[bin_dev_offset + 1]); 892 archive_entry_set_ino(entry, header[bin_ino_offset] * 256 + header[bin_ino_offset + 1]); 893 archive_entry_set_mode(entry, header[bin_mode_offset] * 256 + header[bin_mode_offset + 1]); 894 archive_entry_set_uid(entry, header[bin_uid_offset] * 256 + header[bin_uid_offset + 1]); 895 archive_entry_set_gid(entry, header[bin_gid_offset] * 256 + header[bin_gid_offset + 1]); 896 archive_entry_set_nlink(entry, header[bin_nlink_offset] * 256 + header[bin_nlink_offset + 1]); 897 archive_entry_set_rdev(entry, header[bin_rdev_offset] * 256 + header[bin_rdev_offset + 1]); 898 archive_entry_set_mtime(entry, be4(header + bin_mtime_offset), 0); 899 *namelength = header[bin_namesize_offset] * 256 + header[bin_namesize_offset + 1]; 900 *name_pad = *namelength & 1; /* Pad to even. */ 901 902 cpio->entry_bytes_remaining = be4(header + bin_filesize_offset); 903 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 904 cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ 905 __archive_read_consume(a, bin_header_size); 906 return (ARCHIVE_OK); 907 } 908 909 static int 910 archive_read_format_cpio_cleanup(struct archive_read *a) 911 { 912 struct cpio *cpio; 913 914 cpio = (struct cpio *)(a->format->data); 915 /* Free inode->name map */ 916 while (cpio->links_head != NULL) { 917 struct links_entry *lp = cpio->links_head->next; 918 919 if (cpio->links_head->name) 920 free(cpio->links_head->name); 921 free(cpio->links_head); 922 cpio->links_head = lp; 923 } 924 free(cpio); 925 (a->format->data) = NULL; 926 return (ARCHIVE_OK); 927 } 928 929 static int 930 le4(const unsigned char *p) 931 { 932 return ((p[0]<<16) + (p[1]<<24) + (p[2]<<0) + (p[3]<<8)); 933 } 934 935 936 static int 937 be4(const unsigned char *p) 938 { 939 return ((p[0]<<24) + (p[1]<<16) + (p[2]<<8) + (p[3])); 940 } 941 942 /* 943 * Note that this implementation does not (and should not!) obey 944 * locale settings; you cannot simply substitute strtol here, since 945 * it does obey locale. 946 */ 947 static int64_t 948 atol8(const char *p, unsigned char_cnt) 949 { 950 int64_t l; 951 int digit; 952 953 l = 0; 954 while (char_cnt-- > 0) { 955 if (*p >= '0' && *p <= '7') 956 digit = *p - '0'; 957 else 958 return (l); 959 p++; 960 l <<= 3; 961 l |= digit; 962 } 963 return (l); 964 } 965 966 static int64_t 967 atol16(const char *p, unsigned char_cnt) 968 { 969 int64_t l; 970 int digit; 971 972 l = 0; 973 while (char_cnt-- > 0) { 974 if (*p >= 'a' && *p <= 'f') 975 digit = *p - 'a' + 10; 976 else if (*p >= 'A' && *p <= 'F') 977 digit = *p - 'A' + 10; 978 else if (*p >= '0' && *p <= '9') 979 digit = *p - '0'; 980 else 981 return (l); 982 p++; 983 l <<= 4; 984 l |= digit; 985 } 986 return (l); 987 } 988 989 static int 990 record_hardlink(struct archive_read *a, 991 struct cpio *cpio, struct archive_entry *entry) 992 { 993 struct links_entry *le; 994 dev_t dev; 995 int64_t ino; 996 997 if (archive_entry_nlink(entry) <= 1) 998 return (ARCHIVE_OK); 999 1000 dev = archive_entry_dev(entry); 1001 ino = archive_entry_ino64(entry); 1002 1003 /* 1004 * First look in the list of multiply-linked files. If we've 1005 * already dumped it, convert this entry to a hard link entry. 1006 */ 1007 for (le = cpio->links_head; le; le = le->next) { 1008 if (le->dev == dev && le->ino == ino) { 1009 archive_entry_copy_hardlink(entry, le->name); 1010 1011 if (--le->links <= 0) { 1012 if (le->previous != NULL) 1013 le->previous->next = le->next; 1014 if (le->next != NULL) 1015 le->next->previous = le->previous; 1016 if (cpio->links_head == le) 1017 cpio->links_head = le->next; 1018 free(le->name); 1019 free(le); 1020 } 1021 1022 return (ARCHIVE_OK); 1023 } 1024 } 1025 1026 le = (struct links_entry *)malloc(sizeof(struct links_entry)); 1027 if (le == NULL) { 1028 archive_set_error(&a->archive, 1029 ENOMEM, "Out of memory adding file to list"); 1030 return (ARCHIVE_FATAL); 1031 } 1032 if (cpio->links_head != NULL) 1033 cpio->links_head->previous = le; 1034 le->next = cpio->links_head; 1035 le->previous = NULL; 1036 cpio->links_head = le; 1037 le->dev = dev; 1038 le->ino = ino; 1039 le->links = archive_entry_nlink(entry) - 1; 1040 le->name = strdup(archive_entry_pathname(entry)); 1041 if (le->name == NULL) { 1042 archive_set_error(&a->archive, 1043 ENOMEM, "Out of memory adding file to list"); 1044 return (ARCHIVE_FATAL); 1045 } 1046 1047 return (ARCHIVE_OK); 1048 } 1049