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