1 /*- 2 * Copyright (c) 2003-2007 Tim Kientzle 3 * Copyright (c) 2010-2012 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 int64_t 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 int64_t 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 NULL, 246 archive_read_format_cpio_cleanup, 247 NULL, 248 NULL); 249 250 if (r != ARCHIVE_OK) 251 free(cpio); 252 return (ARCHIVE_OK); 253 } 254 255 256 static int 257 archive_read_format_cpio_bid(struct archive_read *a, int best_bid) 258 { 259 const unsigned char *p; 260 struct cpio *cpio; 261 int bid; 262 263 (void)best_bid; /* UNUSED */ 264 265 cpio = (struct cpio *)(a->format->data); 266 267 if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) 268 return (-1); 269 270 bid = 0; 271 if (memcmp(p, "070707", 6) == 0) { 272 /* ASCII cpio archive (odc, POSIX.1) */ 273 cpio->read_header = header_odc; 274 bid += 48; 275 /* 276 * XXX TODO: More verification; Could check that only octal 277 * digits appear in appropriate header locations. XXX 278 */ 279 } else if (memcmp(p, "070727", 6) == 0) { 280 /* afio large ASCII cpio archive */ 281 cpio->read_header = header_odc; 282 bid += 48; 283 /* 284 * XXX TODO: More verification; Could check that almost hex 285 * digits appear in appropriate header locations. XXX 286 */ 287 } else if (memcmp(p, "070701", 6) == 0) { 288 /* ASCII cpio archive (SVR4 without CRC) */ 289 cpio->read_header = header_newc; 290 bid += 48; 291 /* 292 * XXX TODO: More verification; Could check that only hex 293 * digits appear in appropriate header locations. XXX 294 */ 295 } else if (memcmp(p, "070702", 6) == 0) { 296 /* ASCII cpio archive (SVR4 with CRC) */ 297 /* XXX TODO: Flag that we should check the CRC. XXX */ 298 cpio->read_header = header_newc; 299 bid += 48; 300 /* 301 * XXX TODO: More verification; Could check that only hex 302 * digits appear in appropriate header locations. XXX 303 */ 304 } else if (p[0] * 256 + p[1] == 070707) { 305 /* big-endian binary cpio archives */ 306 cpio->read_header = header_bin_be; 307 bid += 16; 308 /* Is more verification possible here? */ 309 } else if (p[0] + p[1] * 256 == 070707) { 310 /* little-endian binary cpio archives */ 311 cpio->read_header = header_bin_le; 312 bid += 16; 313 /* Is more verification possible here? */ 314 } else 315 return (ARCHIVE_WARN); 316 317 return (bid); 318 } 319 320 static int 321 archive_read_format_cpio_options(struct archive_read *a, 322 const char *key, const char *val) 323 { 324 struct cpio *cpio; 325 int ret = ARCHIVE_FAILED; 326 327 cpio = (struct cpio *)(a->format->data); 328 if (strcmp(key, "compat-2x") == 0) { 329 /* Handle filnames as libarchive 2.x */ 330 cpio->init_default_conversion = (val != NULL)?1:0; 331 return (ARCHIVE_OK); 332 } else if (strcmp(key, "hdrcharset") == 0) { 333 if (val == NULL || val[0] == 0) 334 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 335 "cpio: hdrcharset option needs a character-set name"); 336 else { 337 cpio->opt_sconv = 338 archive_string_conversion_from_charset( 339 &a->archive, val, 0); 340 if (cpio->opt_sconv != NULL) 341 ret = ARCHIVE_OK; 342 else 343 ret = ARCHIVE_FATAL; 344 } 345 return (ret); 346 } 347 348 /* Note: The "warn" return is just to inform the options 349 * supervisor that we didn't handle it. It will generate 350 * a suitable error if no one used this option. */ 351 return (ARCHIVE_WARN); 352 } 353 354 static int 355 archive_read_format_cpio_read_header(struct archive_read *a, 356 struct archive_entry *entry) 357 { 358 struct cpio *cpio; 359 const void *h; 360 struct archive_string_conv *sconv; 361 size_t namelength; 362 size_t name_pad; 363 int r; 364 365 cpio = (struct cpio *)(a->format->data); 366 sconv = cpio->opt_sconv; 367 if (sconv == NULL) { 368 if (!cpio->init_default_conversion) { 369 cpio->sconv_default = 370 archive_string_default_conversion_for_read( 371 &(a->archive)); 372 cpio->init_default_conversion = 1; 373 } 374 sconv = cpio->sconv_default; 375 } 376 377 r = (cpio->read_header(a, cpio, entry, &namelength, &name_pad)); 378 379 if (r < ARCHIVE_WARN) 380 return (r); 381 382 /* Read name from buffer. */ 383 h = __archive_read_ahead(a, namelength + name_pad, NULL); 384 if (h == NULL) 385 return (ARCHIVE_FATAL); 386 if (archive_entry_copy_pathname_l(entry, 387 (const char *)h, namelength, sconv) != 0) { 388 if (errno == ENOMEM) { 389 archive_set_error(&a->archive, ENOMEM, 390 "Can't allocate memory for Pathname"); 391 return (ARCHIVE_FATAL); 392 } 393 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 394 "Pathname can't be converted from %s to current locale.", 395 archive_string_conversion_charset_name(sconv)); 396 r = ARCHIVE_WARN; 397 } 398 cpio->entry_offset = 0; 399 400 __archive_read_consume(a, namelength + name_pad); 401 402 /* If this is a symlink, read the link contents. */ 403 if (archive_entry_filetype(entry) == AE_IFLNK) { 404 if (cpio->entry_bytes_remaining > 1024 * 1024) { 405 archive_set_error(&a->archive, ENOMEM, 406 "Rejecting malformed cpio archive: symlink contents exceed 1 megabyte"); 407 return (ARCHIVE_FATAL); 408 } 409 h = __archive_read_ahead(a, 410 (size_t)cpio->entry_bytes_remaining, NULL); 411 if (h == NULL) 412 return (ARCHIVE_FATAL); 413 if (archive_entry_copy_symlink_l(entry, (const char *)h, 414 (size_t)cpio->entry_bytes_remaining, sconv) != 0) { 415 if (errno == ENOMEM) { 416 archive_set_error(&a->archive, ENOMEM, 417 "Can't allocate memory for Linkname"); 418 return (ARCHIVE_FATAL); 419 } 420 archive_set_error(&a->archive, 421 ARCHIVE_ERRNO_FILE_FORMAT, 422 "Linkname can't be converted from %s to " 423 "current locale.", 424 archive_string_conversion_charset_name(sconv)); 425 r = ARCHIVE_WARN; 426 } 427 __archive_read_consume(a, cpio->entry_bytes_remaining); 428 cpio->entry_bytes_remaining = 0; 429 } 430 431 /* XXX TODO: If the full mode is 0160200, then this is a Solaris 432 * ACL description for the following entry. Read this body 433 * and parse it as a Solaris-style ACL, then read the next 434 * header. XXX */ 435 436 /* Compare name to "TRAILER!!!" to test for end-of-archive. */ 437 if (namelength == 11 && strcmp((const char *)h, "TRAILER!!!") == 0) { 438 /* TODO: Store file location of start of block. */ 439 archive_clear_error(&a->archive); 440 return (ARCHIVE_EOF); 441 } 442 443 /* Detect and record hardlinks to previously-extracted entries. */ 444 if (record_hardlink(a, cpio, entry) != ARCHIVE_OK) { 445 return (ARCHIVE_FATAL); 446 } 447 448 return (r); 449 } 450 451 static int 452 archive_read_format_cpio_read_data(struct archive_read *a, 453 const void **buff, size_t *size, int64_t *offset) 454 { 455 ssize_t bytes_read; 456 struct cpio *cpio; 457 458 cpio = (struct cpio *)(a->format->data); 459 460 if (cpio->entry_bytes_unconsumed) { 461 __archive_read_consume(a, cpio->entry_bytes_unconsumed); 462 cpio->entry_bytes_unconsumed = 0; 463 } 464 465 if (cpio->entry_bytes_remaining > 0) { 466 *buff = __archive_read_ahead(a, 1, &bytes_read); 467 if (bytes_read <= 0) 468 return (ARCHIVE_FATAL); 469 if (bytes_read > cpio->entry_bytes_remaining) 470 bytes_read = (ssize_t)cpio->entry_bytes_remaining; 471 *size = bytes_read; 472 cpio->entry_bytes_unconsumed = bytes_read; 473 *offset = cpio->entry_offset; 474 cpio->entry_offset += bytes_read; 475 cpio->entry_bytes_remaining -= bytes_read; 476 return (ARCHIVE_OK); 477 } else { 478 if (cpio->entry_padding != 479 __archive_read_consume(a, cpio->entry_padding)) { 480 return (ARCHIVE_FATAL); 481 } 482 cpio->entry_padding = 0; 483 *buff = NULL; 484 *size = 0; 485 *offset = cpio->entry_offset; 486 return (ARCHIVE_EOF); 487 } 488 } 489 490 static int 491 archive_read_format_cpio_skip(struct archive_read *a) 492 { 493 struct cpio *cpio = (struct cpio *)(a->format->data); 494 int64_t to_skip = cpio->entry_bytes_remaining + cpio->entry_padding + 495 cpio->entry_bytes_unconsumed; 496 497 if (to_skip != __archive_read_consume(a, to_skip)) { 498 return (ARCHIVE_FATAL); 499 } 500 cpio->entry_bytes_remaining = 0; 501 cpio->entry_padding = 0; 502 cpio->entry_bytes_unconsumed = 0; 503 return (ARCHIVE_OK); 504 } 505 506 /* 507 * Skip forward to the next cpio newc header by searching for the 508 * 07070[12] string. This should be generalized and merged with 509 * find_odc_header below. 510 */ 511 static int 512 is_hex(const char *p, size_t len) 513 { 514 while (len-- > 0) { 515 if ((*p >= '0' && *p <= '9') 516 || (*p >= 'a' && *p <= 'f') 517 || (*p >= 'A' && *p <= 'F')) 518 ++p; 519 else 520 return (0); 521 } 522 return (1); 523 } 524 525 static int 526 find_newc_header(struct archive_read *a) 527 { 528 const void *h; 529 const char *p, *q; 530 size_t skip, skipped = 0; 531 ssize_t bytes; 532 533 for (;;) { 534 h = __archive_read_ahead(a, newc_header_size, &bytes); 535 if (h == NULL) 536 return (ARCHIVE_FATAL); 537 p = h; 538 q = p + bytes; 539 540 /* Try the typical case first, then go into the slow search.*/ 541 if (memcmp("07070", p, 5) == 0 542 && (p[5] == '1' || p[5] == '2') 543 && is_hex(p, newc_header_size)) 544 return (ARCHIVE_OK); 545 546 /* 547 * Scan ahead until we find something that looks 548 * like a newc header. 549 */ 550 while (p + newc_header_size <= q) { 551 switch (p[5]) { 552 case '1': 553 case '2': 554 if (memcmp("07070", p, 5) == 0 555 && is_hex(p, newc_header_size)) { 556 skip = p - (const char *)h; 557 __archive_read_consume(a, skip); 558 skipped += skip; 559 if (skipped > 0) { 560 archive_set_error(&a->archive, 561 0, 562 "Skipped %d bytes before " 563 "finding valid header", 564 (int)skipped); 565 return (ARCHIVE_WARN); 566 } 567 return (ARCHIVE_OK); 568 } 569 p += 2; 570 break; 571 case '0': 572 p++; 573 break; 574 default: 575 p += 6; 576 break; 577 } 578 } 579 skip = p - (const char *)h; 580 __archive_read_consume(a, skip); 581 skipped += skip; 582 } 583 } 584 585 static int 586 header_newc(struct archive_read *a, struct cpio *cpio, 587 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 588 { 589 const void *h; 590 const char *header; 591 int r; 592 593 r = find_newc_header(a); 594 if (r < ARCHIVE_WARN) 595 return (r); 596 597 /* Read fixed-size portion of header. */ 598 h = __archive_read_ahead(a, newc_header_size, NULL); 599 if (h == NULL) 600 return (ARCHIVE_FATAL); 601 602 /* Parse out hex fields. */ 603 header = (const char *)h; 604 605 if (memcmp(header + newc_magic_offset, "070701", 6) == 0) { 606 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_NOCRC; 607 a->archive.archive_format_name = "ASCII cpio (SVR4 with no CRC)"; 608 } else if (memcmp(header + newc_magic_offset, "070702", 6) == 0) { 609 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_CRC; 610 a->archive.archive_format_name = "ASCII cpio (SVR4 with CRC)"; 611 } else { 612 /* TODO: Abort here? */ 613 } 614 615 archive_entry_set_devmajor(entry, 616 (dev_t)atol16(header + newc_devmajor_offset, newc_devmajor_size)); 617 archive_entry_set_devminor(entry, 618 (dev_t)atol16(header + newc_devminor_offset, newc_devminor_size)); 619 archive_entry_set_ino(entry, atol16(header + newc_ino_offset, newc_ino_size)); 620 archive_entry_set_mode(entry, 621 (mode_t)atol16(header + newc_mode_offset, newc_mode_size)); 622 archive_entry_set_uid(entry, atol16(header + newc_uid_offset, newc_uid_size)); 623 archive_entry_set_gid(entry, atol16(header + newc_gid_offset, newc_gid_size)); 624 archive_entry_set_nlink(entry, 625 (unsigned int)atol16(header + newc_nlink_offset, newc_nlink_size)); 626 archive_entry_set_rdevmajor(entry, 627 (dev_t)atol16(header + newc_rdevmajor_offset, newc_rdevmajor_size)); 628 archive_entry_set_rdevminor(entry, 629 (dev_t)atol16(header + newc_rdevminor_offset, newc_rdevminor_size)); 630 archive_entry_set_mtime(entry, atol16(header + newc_mtime_offset, newc_mtime_size), 0); 631 *namelength = (size_t)atol16(header + newc_namesize_offset, newc_namesize_size); 632 /* Pad name to 2 more than a multiple of 4. */ 633 *name_pad = (2 - *namelength) & 3; 634 635 /* 636 * Note: entry_bytes_remaining is at least 64 bits and 637 * therefore guaranteed to be big enough for a 33-bit file 638 * size. 639 */ 640 cpio->entry_bytes_remaining = 641 atol16(header + newc_filesize_offset, newc_filesize_size); 642 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 643 /* Pad file contents to a multiple of 4. */ 644 cpio->entry_padding = 3 & -cpio->entry_bytes_remaining; 645 __archive_read_consume(a, newc_header_size); 646 return (r); 647 } 648 649 /* 650 * Skip forward to the next cpio odc header by searching for the 651 * 070707 string. This is a hand-optimized search that could 652 * probably be easily generalized to handle all character-based 653 * cpio variants. 654 */ 655 static int 656 is_octal(const char *p, size_t len) 657 { 658 while (len-- > 0) { 659 if (*p < '0' || *p > '7') 660 return (0); 661 ++p; 662 } 663 return (1); 664 } 665 666 static int 667 is_afio_large(const char *h, size_t len) 668 { 669 if (len < afiol_header_size) 670 return (0); 671 if (h[afiol_ino_m_offset] != 'm' 672 || h[afiol_mtime_n_offset] != 'n' 673 || h[afiol_xsize_s_offset] != 's' 674 || h[afiol_filesize_c_offset] != ':') 675 return (0); 676 if (!is_hex(h + afiol_dev_offset, afiol_ino_m_offset - afiol_dev_offset)) 677 return (0); 678 if (!is_hex(h + afiol_mode_offset, afiol_mtime_n_offset - afiol_mode_offset)) 679 return (0); 680 if (!is_hex(h + afiol_namesize_offset, afiol_xsize_s_offset - afiol_namesize_offset)) 681 return (0); 682 if (!is_hex(h + afiol_filesize_offset, afiol_filesize_size)) 683 return (0); 684 return (1); 685 } 686 687 static int 688 find_odc_header(struct archive_read *a) 689 { 690 const void *h; 691 const char *p, *q; 692 size_t skip, skipped = 0; 693 ssize_t bytes; 694 695 for (;;) { 696 h = __archive_read_ahead(a, odc_header_size, &bytes); 697 if (h == NULL) 698 return (ARCHIVE_FATAL); 699 p = h; 700 q = p + bytes; 701 702 /* Try the typical case first, then go into the slow search.*/ 703 if (memcmp("070707", p, 6) == 0 && is_octal(p, odc_header_size)) 704 return (ARCHIVE_OK); 705 if (memcmp("070727", p, 6) == 0 && is_afio_large(p, bytes)) { 706 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 707 return (ARCHIVE_OK); 708 } 709 710 /* 711 * Scan ahead until we find something that looks 712 * like an odc header. 713 */ 714 while (p + odc_header_size <= q) { 715 switch (p[5]) { 716 case '7': 717 if ((memcmp("070707", p, 6) == 0 718 && is_octal(p, odc_header_size)) 719 || (memcmp("070727", p, 6) == 0 720 && is_afio_large(p, q - p))) { 721 skip = p - (const char *)h; 722 __archive_read_consume(a, skip); 723 skipped += skip; 724 if (p[4] == '2') 725 a->archive.archive_format = 726 ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 727 if (skipped > 0) { 728 archive_set_error(&a->archive, 729 0, 730 "Skipped %d bytes before " 731 "finding valid header", 732 (int)skipped); 733 return (ARCHIVE_WARN); 734 } 735 return (ARCHIVE_OK); 736 } 737 p += 2; 738 break; 739 case '0': 740 p++; 741 break; 742 default: 743 p += 6; 744 break; 745 } 746 } 747 skip = p - (const char *)h; 748 __archive_read_consume(a, skip); 749 skipped += skip; 750 } 751 } 752 753 static int 754 header_odc(struct archive_read *a, struct cpio *cpio, 755 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 756 { 757 const void *h; 758 int r; 759 const char *header; 760 761 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_POSIX; 762 a->archive.archive_format_name = "POSIX octet-oriented cpio"; 763 764 /* Find the start of the next header. */ 765 r = find_odc_header(a); 766 if (r < ARCHIVE_WARN) 767 return (r); 768 769 if (a->archive.archive_format == ARCHIVE_FORMAT_CPIO_AFIO_LARGE) { 770 int r2 = (header_afiol(a, cpio, entry, namelength, name_pad)); 771 if (r2 == ARCHIVE_OK) 772 return (r); 773 else 774 return (r2); 775 } 776 777 /* Read fixed-size portion of header. */ 778 h = __archive_read_ahead(a, odc_header_size, NULL); 779 if (h == NULL) 780 return (ARCHIVE_FATAL); 781 782 /* Parse out octal fields. */ 783 header = (const char *)h; 784 785 archive_entry_set_dev(entry, 786 (dev_t)atol8(header + odc_dev_offset, odc_dev_size)); 787 archive_entry_set_ino(entry, atol8(header + odc_ino_offset, odc_ino_size)); 788 archive_entry_set_mode(entry, 789 (mode_t)atol8(header + odc_mode_offset, odc_mode_size)); 790 archive_entry_set_uid(entry, atol8(header + odc_uid_offset, odc_uid_size)); 791 archive_entry_set_gid(entry, atol8(header + odc_gid_offset, odc_gid_size)); 792 archive_entry_set_nlink(entry, 793 (unsigned int)atol8(header + odc_nlink_offset, odc_nlink_size)); 794 archive_entry_set_rdev(entry, 795 (dev_t)atol8(header + odc_rdev_offset, odc_rdev_size)); 796 archive_entry_set_mtime(entry, atol8(header + odc_mtime_offset, odc_mtime_size), 0); 797 *namelength = (size_t)atol8(header + odc_namesize_offset, odc_namesize_size); 798 *name_pad = 0; /* No padding of filename. */ 799 800 /* 801 * Note: entry_bytes_remaining is at least 64 bits and 802 * therefore guaranteed to be big enough for a 33-bit file 803 * size. 804 */ 805 cpio->entry_bytes_remaining = 806 atol8(header + odc_filesize_offset, odc_filesize_size); 807 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 808 cpio->entry_padding = 0; 809 __archive_read_consume(a, odc_header_size); 810 return (r); 811 } 812 813 /* 814 * NOTE: if a filename suffix is ".z", it is the file gziped by afio. 815 * it would be nice that we can show uncompressed file size and we can 816 * uncompressed file contents automatically, unfortunately we have nothing 817 * to get a uncompressed file size while reading each header. it means 818 * we also cannot uncompressed file contens under the our framework. 819 */ 820 static int 821 header_afiol(struct archive_read *a, struct cpio *cpio, 822 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 823 { 824 const void *h; 825 const char *header; 826 827 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; 828 a->archive.archive_format_name = "afio large ASCII"; 829 830 /* Read fixed-size portion of header. */ 831 h = __archive_read_ahead(a, afiol_header_size, NULL); 832 if (h == NULL) 833 return (ARCHIVE_FATAL); 834 835 /* Parse out octal fields. */ 836 header = (const char *)h; 837 838 archive_entry_set_dev(entry, 839 (dev_t)atol16(header + afiol_dev_offset, afiol_dev_size)); 840 archive_entry_set_ino(entry, atol16(header + afiol_ino_offset, afiol_ino_size)); 841 archive_entry_set_mode(entry, 842 (mode_t)atol8(header + afiol_mode_offset, afiol_mode_size)); 843 archive_entry_set_uid(entry, atol16(header + afiol_uid_offset, afiol_uid_size)); 844 archive_entry_set_gid(entry, atol16(header + afiol_gid_offset, afiol_gid_size)); 845 archive_entry_set_nlink(entry, 846 (unsigned int)atol16(header + afiol_nlink_offset, afiol_nlink_size)); 847 archive_entry_set_rdev(entry, 848 (dev_t)atol16(header + afiol_rdev_offset, afiol_rdev_size)); 849 archive_entry_set_mtime(entry, atol16(header + afiol_mtime_offset, afiol_mtime_size), 0); 850 *namelength = (size_t)atol16(header + afiol_namesize_offset, afiol_namesize_size); 851 *name_pad = 0; /* No padding of filename. */ 852 853 cpio->entry_bytes_remaining = 854 atol16(header + afiol_filesize_offset, afiol_filesize_size); 855 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 856 cpio->entry_padding = 0; 857 __archive_read_consume(a, afiol_header_size); 858 return (ARCHIVE_OK); 859 } 860 861 862 static int 863 header_bin_le(struct archive_read *a, struct cpio *cpio, 864 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 865 { 866 const void *h; 867 const unsigned char *header; 868 869 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_LE; 870 a->archive.archive_format_name = "cpio (little-endian binary)"; 871 872 /* Read fixed-size portion of header. */ 873 h = __archive_read_ahead(a, bin_header_size, NULL); 874 if (h == NULL) { 875 archive_set_error(&a->archive, 0, 876 "End of file trying to read next cpio header"); 877 return (ARCHIVE_FATAL); 878 } 879 880 /* Parse out binary fields. */ 881 header = (const unsigned char *)h; 882 883 archive_entry_set_dev(entry, header[bin_dev_offset] + header[bin_dev_offset + 1] * 256); 884 archive_entry_set_ino(entry, header[bin_ino_offset] + header[bin_ino_offset + 1] * 256); 885 archive_entry_set_mode(entry, header[bin_mode_offset] + header[bin_mode_offset + 1] * 256); 886 archive_entry_set_uid(entry, header[bin_uid_offset] + header[bin_uid_offset + 1] * 256); 887 archive_entry_set_gid(entry, header[bin_gid_offset] + header[bin_gid_offset + 1] * 256); 888 archive_entry_set_nlink(entry, header[bin_nlink_offset] + header[bin_nlink_offset + 1] * 256); 889 archive_entry_set_rdev(entry, header[bin_rdev_offset] + header[bin_rdev_offset + 1] * 256); 890 archive_entry_set_mtime(entry, le4(header + bin_mtime_offset), 0); 891 *namelength = header[bin_namesize_offset] + header[bin_namesize_offset + 1] * 256; 892 *name_pad = *namelength & 1; /* Pad to even. */ 893 894 cpio->entry_bytes_remaining = le4(header + bin_filesize_offset); 895 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 896 cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ 897 __archive_read_consume(a, bin_header_size); 898 return (ARCHIVE_OK); 899 } 900 901 static int 902 header_bin_be(struct archive_read *a, struct cpio *cpio, 903 struct archive_entry *entry, size_t *namelength, size_t *name_pad) 904 { 905 const void *h; 906 const unsigned char *header; 907 908 a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_BE; 909 a->archive.archive_format_name = "cpio (big-endian binary)"; 910 911 /* Read fixed-size portion of header. */ 912 h = __archive_read_ahead(a, bin_header_size, NULL); 913 if (h == NULL) { 914 archive_set_error(&a->archive, 0, 915 "End of file trying to read next cpio header"); 916 return (ARCHIVE_FATAL); 917 } 918 919 /* Parse out binary fields. */ 920 header = (const unsigned char *)h; 921 922 archive_entry_set_dev(entry, header[bin_dev_offset] * 256 + header[bin_dev_offset + 1]); 923 archive_entry_set_ino(entry, header[bin_ino_offset] * 256 + header[bin_ino_offset + 1]); 924 archive_entry_set_mode(entry, header[bin_mode_offset] * 256 + header[bin_mode_offset + 1]); 925 archive_entry_set_uid(entry, header[bin_uid_offset] * 256 + header[bin_uid_offset + 1]); 926 archive_entry_set_gid(entry, header[bin_gid_offset] * 256 + header[bin_gid_offset + 1]); 927 archive_entry_set_nlink(entry, header[bin_nlink_offset] * 256 + header[bin_nlink_offset + 1]); 928 archive_entry_set_rdev(entry, header[bin_rdev_offset] * 256 + header[bin_rdev_offset + 1]); 929 archive_entry_set_mtime(entry, be4(header + bin_mtime_offset), 0); 930 *namelength = header[bin_namesize_offset] * 256 + header[bin_namesize_offset + 1]; 931 *name_pad = *namelength & 1; /* Pad to even. */ 932 933 cpio->entry_bytes_remaining = be4(header + bin_filesize_offset); 934 archive_entry_set_size(entry, cpio->entry_bytes_remaining); 935 cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ 936 __archive_read_consume(a, bin_header_size); 937 return (ARCHIVE_OK); 938 } 939 940 static int 941 archive_read_format_cpio_cleanup(struct archive_read *a) 942 { 943 struct cpio *cpio; 944 945 cpio = (struct cpio *)(a->format->data); 946 /* Free inode->name map */ 947 while (cpio->links_head != NULL) { 948 struct links_entry *lp = cpio->links_head->next; 949 950 if (cpio->links_head->name) 951 free(cpio->links_head->name); 952 free(cpio->links_head); 953 cpio->links_head = lp; 954 } 955 free(cpio); 956 (a->format->data) = NULL; 957 return (ARCHIVE_OK); 958 } 959 960 static int64_t 961 le4(const unsigned char *p) 962 { 963 return ((p[0] << 16) + (((int64_t)p[1]) << 24) + (p[2] << 0) + (p[3] << 8)); 964 } 965 966 967 static int64_t 968 be4(const unsigned char *p) 969 { 970 return ((((int64_t)p[0]) << 24) + (p[1] << 16) + (p[2] << 8) + (p[3])); 971 } 972 973 /* 974 * Note that this implementation does not (and should not!) obey 975 * locale settings; you cannot simply substitute strtol here, since 976 * it does obey locale. 977 */ 978 static int64_t 979 atol8(const char *p, unsigned char_cnt) 980 { 981 int64_t l; 982 int digit; 983 984 l = 0; 985 while (char_cnt-- > 0) { 986 if (*p >= '0' && *p <= '7') 987 digit = *p - '0'; 988 else 989 return (l); 990 p++; 991 l <<= 3; 992 l |= digit; 993 } 994 return (l); 995 } 996 997 static int64_t 998 atol16(const char *p, unsigned char_cnt) 999 { 1000 int64_t l; 1001 int digit; 1002 1003 l = 0; 1004 while (char_cnt-- > 0) { 1005 if (*p >= 'a' && *p <= 'f') 1006 digit = *p - 'a' + 10; 1007 else if (*p >= 'A' && *p <= 'F') 1008 digit = *p - 'A' + 10; 1009 else if (*p >= '0' && *p <= '9') 1010 digit = *p - '0'; 1011 else 1012 return (l); 1013 p++; 1014 l <<= 4; 1015 l |= digit; 1016 } 1017 return (l); 1018 } 1019 1020 static int 1021 record_hardlink(struct archive_read *a, 1022 struct cpio *cpio, struct archive_entry *entry) 1023 { 1024 struct links_entry *le; 1025 dev_t dev; 1026 int64_t ino; 1027 1028 if (archive_entry_nlink(entry) <= 1) 1029 return (ARCHIVE_OK); 1030 1031 dev = archive_entry_dev(entry); 1032 ino = archive_entry_ino64(entry); 1033 1034 /* 1035 * First look in the list of multiply-linked files. If we've 1036 * already dumped it, convert this entry to a hard link entry. 1037 */ 1038 for (le = cpio->links_head; le; le = le->next) { 1039 if (le->dev == dev && le->ino == ino) { 1040 archive_entry_copy_hardlink(entry, le->name); 1041 1042 if (--le->links <= 0) { 1043 if (le->previous != NULL) 1044 le->previous->next = le->next; 1045 if (le->next != NULL) 1046 le->next->previous = le->previous; 1047 if (cpio->links_head == le) 1048 cpio->links_head = le->next; 1049 free(le->name); 1050 free(le); 1051 } 1052 1053 return (ARCHIVE_OK); 1054 } 1055 } 1056 1057 le = (struct links_entry *)malloc(sizeof(struct links_entry)); 1058 if (le == NULL) { 1059 archive_set_error(&a->archive, 1060 ENOMEM, "Out of memory adding file to list"); 1061 return (ARCHIVE_FATAL); 1062 } 1063 if (cpio->links_head != NULL) 1064 cpio->links_head->previous = le; 1065 le->next = cpio->links_head; 1066 le->previous = NULL; 1067 cpio->links_head = le; 1068 le->dev = dev; 1069 le->ino = ino; 1070 le->links = archive_entry_nlink(entry) - 1; 1071 le->name = strdup(archive_entry_pathname(entry)); 1072 if (le->name == NULL) { 1073 archive_set_error(&a->archive, 1074 ENOMEM, "Out of memory adding file to list"); 1075 return (ARCHIVE_FATAL); 1076 } 1077 1078 return (ARCHIVE_OK); 1079 } 1080