1 /*- 2 * Copyright (c) 2003-2007 Tim Kientzle 3 * Copyright (c) 2009 Andreas Henriksson <andreas@fatal.se> 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: src/lib/libarchive/archive_read_support_format_iso9660.c,v 1.30 2008/12/06 06:57:45 kientzle Exp $"); 29 30 #ifdef HAVE_ERRNO_H 31 #include <errno.h> 32 #endif 33 /* #include <stdint.h> */ /* See archive_platform.h */ 34 #include <stdio.h> 35 #ifdef HAVE_STDLIB_H 36 #include <stdlib.h> 37 #endif 38 #ifdef HAVE_STRING_H 39 #include <string.h> 40 #endif 41 #include <time.h> 42 43 #include "archive.h" 44 #include "archive_entry.h" 45 #include "archive_private.h" 46 #include "archive_read_private.h" 47 #include "archive_string.h" 48 49 /* 50 * An overview of ISO 9660 format: 51 * 52 * Each disk is laid out as follows: 53 * * 32k reserved for private use 54 * * Volume descriptor table. Each volume descriptor 55 * is 2k and specifies basic format information. 56 * The "Primary Volume Descriptor" (PVD) is defined by the 57 * standard and should always be present; other volume 58 * descriptors include various vendor-specific extensions. 59 * * Files and directories. Each file/dir is specified by 60 * an "extent" (starting sector and length in bytes). 61 * Dirs are just files with directory records packed one 62 * after another. The PVD contains a single dir entry 63 * specifying the location of the root directory. Everything 64 * else follows from there. 65 * 66 * This module works by first reading the volume descriptors, then 67 * building a list of directory entries, sorted by starting 68 * sector. At each step, I look for the earliest dir entry that 69 * hasn't yet been read, seek forward to that location and read 70 * that entry. If it's a dir, I slurp in the new dir entries and 71 * add them to the heap; if it's a regular file, I return the 72 * corresponding archive_entry and wait for the client to request 73 * the file body. This strategy allows us to read most compliant 74 * CDs with a single pass through the data, as required by libarchive. 75 */ 76 77 /* Structure of on-disk primary volume descriptor. */ 78 #define PVD_type_offset 0 79 #define PVD_type_size 1 80 #define PVD_id_offset (PVD_type_offset + PVD_type_size) 81 #define PVD_id_size 5 82 #define PVD_version_offset (PVD_id_offset + PVD_id_size) 83 #define PVD_version_size 1 84 #define PVD_reserved1_offset (PVD_version_offset + PVD_version_size) 85 #define PVD_reserved1_size 1 86 #define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size) 87 #define PVD_system_id_size 32 88 #define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size) 89 #define PVD_volume_id_size 32 90 #define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size) 91 #define PVD_reserved2_size 8 92 #define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size) 93 #define PVD_volume_space_size_size 8 94 #define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size) 95 #define PVD_reserved3_size 32 96 #define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size) 97 #define PVD_volume_set_size_size 4 98 #define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size) 99 #define PVD_volume_sequence_number_size 4 100 #define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size) 101 #define PVD_logical_block_size_size 4 102 #define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size) 103 #define PVD_path_table_size_size 8 104 #define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size) 105 #define PVD_type_1_path_table_size 4 106 #define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size) 107 #define PVD_opt_type_1_path_table_size 4 108 #define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size) 109 #define PVD_type_m_path_table_size 4 110 #define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size) 111 #define PVD_opt_type_m_path_table_size 4 112 #define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size) 113 #define PVD_root_directory_record_size 34 114 #define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size) 115 #define PVD_volume_set_id_size 128 116 #define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size) 117 #define PVD_publisher_id_size 128 118 #define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size) 119 #define PVD_preparer_id_size 128 120 #define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size) 121 #define PVD_application_id_size 128 122 #define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size) 123 #define PVD_copyright_file_id_size 37 124 #define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size) 125 #define PVD_abstract_file_id_size 37 126 #define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size) 127 #define PVD_bibliographic_file_id_size 37 128 #define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size) 129 #define PVD_creation_date_size 17 130 #define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size) 131 #define PVD_modification_date_size 17 132 #define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size) 133 #define PVD_expiration_date_size 17 134 #define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size) 135 #define PVD_effective_date_size 17 136 #define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size) 137 #define PVD_file_structure_version_size 1 138 #define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size) 139 #define PVD_reserved4_size 1 140 #define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size) 141 #define PVD_application_data_size 512 142 #define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size) 143 #define PVD_reserved5_size (2048 - PVD_reserved5_offset) 144 145 /* TODO: It would make future maintenance easier to just hardcode the 146 * above values. In particular, ECMA119 states the offsets as part of 147 * the standard. That would eliminate the need for the following check.*/ 148 #if PVD_reserved5_offset != 1395 149 #error PVD offset and size definitions are wrong. 150 #endif 151 152 153 /* Structure of optional on-disk supplementary volume descriptor. */ 154 #define SVD_type_offset 0 155 #define SVD_type_size 1 156 #define SVD_id_offset (SVD_type_offset + SVD_type_size) 157 #define SVD_id_size 5 158 #define SVD_version_offset (SVD_id_offset + SVD_id_size) 159 #define SVD_version_size 1 160 /* ... */ 161 #define SVD_volume_space_size_offset 80 162 #define SVD_volume_space_size_size 8 163 #define SVD_escape_sequences_offset (SVD_volume_space_size_offset + SVD_volume_space_size_size) 164 #define SVD_escape_sequences_size 32 165 /* ... */ 166 #define SVD_logical_block_size_offset 128 167 #define SVD_logical_block_size_size 4 168 /* ... */ 169 #define SVD_root_directory_record_offset 156 170 #define SVD_root_directory_record_size 34 171 /* ... */ 172 /* FIXME: validate correctness of last SVD entry offset. */ 173 174 /* Structure of an on-disk directory record. */ 175 /* Note: ISO9660 stores each multi-byte integer twice, once in 176 * each byte order. The sizes here are the size of just one 177 * of the two integers. (This is why the offset of a field isn't 178 * the same as the offset+size of the previous field.) */ 179 #define DR_length_offset 0 180 #define DR_length_size 1 181 #define DR_ext_attr_length_offset 1 182 #define DR_ext_attr_length_size 1 183 #define DR_extent_offset 2 184 #define DR_extent_size 4 185 #define DR_size_offset 10 186 #define DR_size_size 4 187 #define DR_date_offset 18 188 #define DR_date_size 7 189 #define DR_flags_offset 25 190 #define DR_flags_size 1 191 #define DR_file_unit_size_offset 26 192 #define DR_file_unit_size_size 1 193 #define DR_interleave_offset 27 194 #define DR_interleave_size 1 195 #define DR_volume_sequence_number_offset 28 196 #define DR_volume_sequence_number_size 2 197 #define DR_name_len_offset 32 198 #define DR_name_len_size 1 199 #define DR_name_offset 33 200 201 /* In-memory storage for a directory record. */ 202 struct file_info { 203 struct file_info *parent; 204 int refcount; 205 uint64_t offset; /* Offset on disk. */ 206 uint64_t size; /* File size in bytes. */ 207 uint64_t ce_offset; /* Offset of CE */ 208 uint64_t ce_size; /* Size of CE */ 209 time_t birthtime; /* File created time. */ 210 time_t mtime; /* File last modified time. */ 211 time_t atime; /* File last accessed time. */ 212 time_t ctime; /* File attribute change time. */ 213 uint64_t rdev; /* Device number */ 214 mode_t mode; 215 uid_t uid; 216 gid_t gid; 217 ino_t inode; 218 int nlinks; 219 struct archive_string name; /* Pathname */ 220 char name_continues; /* Non-zero if name continues */ 221 struct archive_string symlink; 222 char symlink_continues; /* Non-zero if link continues */ 223 }; 224 225 226 struct iso9660 { 227 int magic; 228 #define ISO9660_MAGIC 0x96609660 229 230 int option_ignore_joliet; 231 232 struct archive_string pathname; 233 char seenRockridge; /* Set true if RR extensions are used. */ 234 unsigned char suspOffset; 235 char seenJoliet; 236 237 uint64_t previous_offset; 238 uint64_t previous_size; 239 struct archive_string previous_pathname; 240 241 /* TODO: Make this a heap for fast inserts and deletions. */ 242 struct file_info **pending_files; 243 int pending_files_allocated; 244 int pending_files_used; 245 246 uint64_t current_position; 247 ssize_t logical_block_size; 248 uint64_t volume_size; /* Total size of volume in bytes. */ 249 250 off_t entry_sparse_offset; 251 int64_t entry_bytes_remaining; 252 }; 253 254 static void add_entry(struct iso9660 *iso9660, struct file_info *file); 255 static int archive_read_format_iso9660_bid(struct archive_read *); 256 static int archive_read_format_iso9660_options(struct archive_read *, 257 const char *, const char *); 258 static int archive_read_format_iso9660_cleanup(struct archive_read *); 259 static int archive_read_format_iso9660_read_data(struct archive_read *, 260 const void **, size_t *, off_t *); 261 static int archive_read_format_iso9660_read_data_skip(struct archive_read *); 262 static int archive_read_format_iso9660_read_header(struct archive_read *, 263 struct archive_entry *); 264 static const char *build_pathname(struct archive_string *, struct file_info *); 265 #if DEBUG 266 static void dump_isodirrec(FILE *, const unsigned char *isodirrec); 267 #endif 268 static time_t time_from_tm(struct tm *); 269 static time_t isodate17(const unsigned char *); 270 static time_t isodate7(const unsigned char *); 271 static int isJolietSVD(struct iso9660 *, const unsigned char *); 272 static int isPVD(struct iso9660 *, const unsigned char *); 273 static struct file_info *next_entry(struct iso9660 *); 274 static int next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, 275 struct file_info **pfile); 276 static struct file_info * 277 parse_file_info(struct iso9660 *iso9660, 278 struct file_info *parent, const unsigned char *isodirrec); 279 static void parse_rockridge(struct iso9660 *iso9660, 280 struct file_info *file, const unsigned char *start, 281 const unsigned char *end); 282 static void parse_rockridge_NM1(struct file_info *, 283 const unsigned char *, int); 284 static void parse_rockridge_SL1(struct file_info *, 285 const unsigned char *, int); 286 static void parse_rockridge_TF1(struct file_info *, 287 const unsigned char *, int); 288 static void release_file(struct iso9660 *, struct file_info *); 289 static unsigned toi(const void *p, int n); 290 291 int 292 archive_read_support_format_iso9660(struct archive *_a) 293 { 294 struct archive_read *a = (struct archive_read *)_a; 295 struct iso9660 *iso9660; 296 int r; 297 298 iso9660 = (struct iso9660 *)malloc(sizeof(*iso9660)); 299 if (iso9660 == NULL) { 300 archive_set_error(&a->archive, ENOMEM, "Can't allocate iso9660 data"); 301 return (ARCHIVE_FATAL); 302 } 303 memset(iso9660, 0, sizeof(*iso9660)); 304 iso9660->magic = ISO9660_MAGIC; 305 306 r = __archive_read_register_format(a, 307 iso9660, 308 "iso9660", 309 archive_read_format_iso9660_bid, 310 archive_read_format_iso9660_options, 311 archive_read_format_iso9660_read_header, 312 archive_read_format_iso9660_read_data, 313 archive_read_format_iso9660_read_data_skip, 314 archive_read_format_iso9660_cleanup); 315 316 if (r != ARCHIVE_OK) { 317 free(iso9660); 318 return (r); 319 } 320 return (ARCHIVE_OK); 321 } 322 323 324 static int 325 archive_read_format_iso9660_bid(struct archive_read *a) 326 { 327 struct iso9660 *iso9660; 328 ssize_t bytes_read, brsvd; 329 const void *h; 330 const unsigned char *p, *psvd; 331 int bid; 332 333 iso9660 = (struct iso9660 *)(a->format->data); 334 335 /* 336 * Skip the first 32k (reserved area) and get the first 337 * 8 sectors of the volume descriptor table. Of course, 338 * if the I/O layer gives us more, we'll take it. 339 */ 340 h = __archive_read_ahead(a, 32768 + 8*2048, &bytes_read); 341 if (h == NULL) 342 return (-1); 343 p = (const unsigned char *)h; 344 345 /* Skip the reserved area. */ 346 bytes_read -= 32768; 347 p += 32768; 348 349 /* Check each volume descriptor to locate possible SVD with Joliet. */ 350 for (brsvd = bytes_read, psvd = p; 351 !iso9660->option_ignore_joliet && brsvd > 2048; 352 brsvd -= 2048, psvd += 2048) { 353 bid = isJolietSVD(iso9660, psvd); 354 if (bid > 0) 355 return (bid); 356 if (*p == '\177') /* End-of-volume-descriptor marker. */ 357 break; 358 } 359 360 /* Check each volume descriptor to locate the PVD. */ 361 for (; bytes_read > 2048; bytes_read -= 2048, p += 2048) { 362 bid = isPVD(iso9660, p); 363 if (bid > 0) 364 return (bid); 365 if (*p == '\177') /* End-of-volume-descriptor marker. */ 366 break; 367 } 368 369 /* We didn't find a valid PVD; return a bid of zero. */ 370 return (0); 371 } 372 373 static int 374 archive_read_format_iso9660_options(struct archive_read *a, 375 const char *key, const char *val) 376 { 377 struct iso9660 *iso9660; 378 379 iso9660 = (struct iso9660 *)(a->format->data); 380 381 if (strcmp(key, "joliet") == 0) { 382 if (val == NULL || strcmp(val, "off") == 0 || 383 strcmp(val, "ignore") == 0 || 384 strcmp(val, "disable") == 0 || 385 strcmp(val, "0") == 0) 386 iso9660->option_ignore_joliet = 1; 387 else 388 iso9660->option_ignore_joliet = 0; 389 return (ARCHIVE_OK); 390 } 391 392 /* Note: The "warn" return is just to inform the options 393 * supervisor that we didn't handle it. It will generate 394 * a suitable error if noone used this option. */ 395 return (ARCHIVE_WARN); 396 } 397 398 static int 399 isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) 400 { 401 struct file_info *file; 402 const unsigned char *p; 403 404 /* Type 2 means it's a SVD. */ 405 if (h[SVD_type_offset] != 2) 406 return (0); 407 408 /* ID must be "CD001" */ 409 if (memcmp(h + SVD_id_offset, "CD001", 5) != 0) 410 return (0); 411 412 /* FIXME: do more validations according to joliet spec. */ 413 414 /* check if this SVD contains joliet extension! */ 415 p = h + SVD_escape_sequences_offset; 416 /* N.B. Joliet spec says p[1] == '\\', but.... */ 417 if (p[0] == '%' && p[1] == '/') { 418 int level = 0; 419 420 if (p[2] == '@') 421 level = 1; 422 else if (p[2] == 'C') 423 level = 2; 424 else if (p[2] == 'E') 425 level = 3; 426 else /* not joliet */ 427 return (0); 428 429 iso9660->seenJoliet = level; 430 431 } else /* not joliet */ 432 return (0); 433 434 iso9660->logical_block_size = toi(h + SVD_logical_block_size_offset, 2); 435 if (iso9660->logical_block_size <= 0) 436 return (0); 437 438 iso9660->volume_size = iso9660->logical_block_size 439 * (uint64_t)toi(h + SVD_volume_space_size_offset, 4); 440 441 #if DEBUG 442 fprintf(stderr, "Joliet UCS-2 level %d with " 443 "logical block size:%d, volume size:%d\n", 444 iso9660->seenJoliet, 445 iso9660->logical_block_size, iso9660->volume_size); 446 #endif 447 448 /* Store the root directory in the pending list. */ 449 file = parse_file_info(iso9660, NULL, h + SVD_root_directory_record_offset); 450 add_entry(iso9660, file); 451 452 return (48); 453 } 454 455 static int 456 isPVD(struct iso9660 *iso9660, const unsigned char *h) 457 { 458 struct file_info *file; 459 int i; 460 461 /* Type of the Primary Volume Descriptor must be 1. */ 462 if (h[PVD_type_offset] != 1) 463 return (0); 464 465 /* ID must be "CD001" */ 466 if (memcmp(h + PVD_id_offset, "CD001", 5) != 0) 467 return (0); 468 469 /* PVD version must be 1. */ 470 if (h[PVD_version_offset] != 1) 471 return (0); 472 473 /* Reserved field must be 0. */ 474 if (h[PVD_reserved1_offset] != 0) 475 return (0); 476 477 /* Reserved field must be 0. */ 478 for (i = 0; i < PVD_reserved2_size; ++i) 479 if (h[PVD_reserved2_offset + i] != 0) 480 return (0); 481 482 /* Reserved field must be 0. */ 483 for (i = 0; i < PVD_reserved3_size; ++i) 484 if (h[PVD_reserved3_offset + i] != 0) 485 return (0); 486 487 /* Logical block size must be > 0. */ 488 /* I've looked at Ecma 119 and can't find any stronger 489 * restriction on this field. */ 490 iso9660->logical_block_size = toi(h + PVD_logical_block_size_offset, 2); 491 if (iso9660->logical_block_size <= 0) 492 return (0); 493 494 iso9660->volume_size = iso9660->logical_block_size 495 * (uint64_t)toi(h + PVD_volume_space_size_offset, 4); 496 497 /* File structure version must be 1 for ISO9660/ECMA119. */ 498 if (h[PVD_file_structure_version_offset] != 1) 499 return (0); 500 501 502 /* Reserved field must be 0. */ 503 for (i = 0; i < PVD_reserved4_size; ++i) 504 if (h[PVD_reserved4_offset + i] != 0) 505 return (0); 506 507 /* Reserved field must be 0. */ 508 for (i = 0; i < PVD_reserved5_size; ++i) 509 if (h[PVD_reserved5_offset + i] != 0) 510 return (0); 511 512 /* XXX TODO: Check other values for sanity; reject more 513 * malformed PVDs. XXX */ 514 515 /* Store the root directory in the pending list. */ 516 file = parse_file_info(iso9660, NULL, h + PVD_root_directory_record_offset); 517 add_entry(iso9660, file); 518 return (48); 519 } 520 521 static int 522 archive_read_format_iso9660_read_header(struct archive_read *a, 523 struct archive_entry *entry) 524 { 525 struct iso9660 *iso9660; 526 struct file_info *file; 527 int r; 528 529 iso9660 = (struct iso9660 *)(a->format->data); 530 531 if (!a->archive.archive_format) { 532 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; 533 a->archive.archive_format_name = "ISO9660"; 534 } 535 536 /* Get the next entry that appears after the current offset. */ 537 r = next_entry_seek(a, iso9660, &file); 538 if (r != ARCHIVE_OK) { 539 release_file(iso9660, file); 540 return (r); 541 } 542 543 iso9660->entry_bytes_remaining = file->size; 544 iso9660->entry_sparse_offset = 0; /* Offset for sparse-file-aware clients. */ 545 546 if (file->offset + file->size > iso9660->volume_size) { 547 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 548 "File is beyond end-of-media: %s", file->name); 549 iso9660->entry_bytes_remaining = 0; 550 iso9660->entry_sparse_offset = 0; 551 release_file(iso9660, file); 552 return (ARCHIVE_WARN); 553 } 554 555 /* Set up the entry structure with information about this entry. */ 556 archive_entry_set_mode(entry, file->mode); 557 archive_entry_set_uid(entry, file->uid); 558 archive_entry_set_gid(entry, file->gid); 559 archive_entry_set_nlink(entry, file->nlinks); 560 archive_entry_set_ino(entry, file->inode); 561 archive_entry_set_birthtime(entry, file->birthtime, 0); 562 archive_entry_set_mtime(entry, file->mtime, 0); 563 archive_entry_set_ctime(entry, file->ctime, 0); 564 archive_entry_set_atime(entry, file->atime, 0); 565 /* N.B.: Rock Ridge supports 64-bit device numbers. */ 566 archive_entry_set_rdev(entry, (dev_t)file->rdev); 567 archive_entry_set_size(entry, iso9660->entry_bytes_remaining); 568 archive_string_empty(&iso9660->pathname); 569 archive_entry_set_pathname(entry, 570 build_pathname(&iso9660->pathname, file)); 571 if (file->symlink.s != NULL) 572 archive_entry_copy_symlink(entry, file->symlink.s); 573 574 /* If this entry points to the same data as the previous 575 * entry, convert this into a hardlink to that entry. 576 * But don't bother for zero-length files. */ 577 if (file->offset == iso9660->previous_offset 578 && file->size == iso9660->previous_size 579 && file->size > 0) { 580 archive_entry_set_hardlink(entry, 581 iso9660->previous_pathname.s); 582 iso9660->entry_bytes_remaining = 0; 583 iso9660->entry_sparse_offset = 0; 584 release_file(iso9660, file); 585 return (ARCHIVE_OK); 586 } 587 588 /* If the offset is before our current position, we can't 589 * seek backwards to extract it, so issue a warning. */ 590 if (file->offset < iso9660->current_position) { 591 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 592 "Ignoring out-of-order file @%x (%s) %jd < %jd", 593 file, 594 iso9660->pathname.s, 595 file->offset, iso9660->current_position); 596 iso9660->entry_bytes_remaining = 0; 597 iso9660->entry_sparse_offset = 0; 598 release_file(iso9660, file); 599 return (ARCHIVE_WARN); 600 } 601 602 iso9660->previous_size = file->size; 603 iso9660->previous_offset = file->offset; 604 archive_strcpy(&iso9660->previous_pathname, iso9660->pathname.s); 605 606 /* If this is a directory, read in all of the entries right now. */ 607 if (archive_entry_filetype(entry) == AE_IFDIR) { 608 while (iso9660->entry_bytes_remaining > 0) { 609 const void *block; 610 const unsigned char *p; 611 ssize_t step = iso9660->logical_block_size; 612 if (step > iso9660->entry_bytes_remaining) 613 step = iso9660->entry_bytes_remaining; 614 block = __archive_read_ahead(a, step, NULL); 615 if (block == NULL) { 616 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 617 "Failed to read full block when scanning ISO9660 directory list"); 618 release_file(iso9660, file); 619 return (ARCHIVE_FATAL); 620 } 621 __archive_read_consume(a, step); 622 iso9660->current_position += step; 623 iso9660->entry_bytes_remaining -= step; 624 for (p = (const unsigned char *)block; 625 *p != 0 && p < (const unsigned char *)block + step; 626 p += *p) { 627 struct file_info *child; 628 629 /* N.B.: these special directory identifiers 630 * are 8 bit "values" even on a 631 * Joliet CD with UCS-2 (16bit) encoding. 632 */ 633 634 /* Skip '.' entry. */ 635 if (*(p + DR_name_len_offset) == 1 636 && *(p + DR_name_offset) == '\0') 637 continue; 638 /* Skip '..' entry. */ 639 if (*(p + DR_name_len_offset) == 1 640 && *(p + DR_name_offset) == '\001') 641 continue; 642 child = parse_file_info(iso9660, file, p); 643 add_entry(iso9660, child); 644 if (iso9660->seenRockridge) { 645 a->archive.archive_format = 646 ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; 647 a->archive.archive_format_name = 648 "ISO9660 with Rockridge extensions"; 649 } 650 } 651 } 652 } 653 654 release_file(iso9660, file); 655 return (ARCHIVE_OK); 656 } 657 658 static int 659 archive_read_format_iso9660_read_data_skip(struct archive_read *a) 660 { 661 /* Because read_next_header always does an explicit skip 662 * to the next entry, we don't need to do anything here. */ 663 (void)a; /* UNUSED */ 664 return (ARCHIVE_OK); 665 } 666 667 static int 668 archive_read_format_iso9660_read_data(struct archive_read *a, 669 const void **buff, size_t *size, off_t *offset) 670 { 671 ssize_t bytes_read; 672 struct iso9660 *iso9660; 673 674 iso9660 = (struct iso9660 *)(a->format->data); 675 if (iso9660->entry_bytes_remaining <= 0) { 676 *buff = NULL; 677 *size = 0; 678 *offset = iso9660->entry_sparse_offset; 679 return (ARCHIVE_EOF); 680 } 681 682 *buff = __archive_read_ahead(a, 1, &bytes_read); 683 if (bytes_read == 0) 684 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 685 "Truncated input file"); 686 if (*buff == NULL) 687 return (ARCHIVE_FATAL); 688 if (bytes_read > iso9660->entry_bytes_remaining) 689 bytes_read = iso9660->entry_bytes_remaining; 690 *size = bytes_read; 691 *offset = iso9660->entry_sparse_offset; 692 iso9660->entry_sparse_offset += bytes_read; 693 iso9660->entry_bytes_remaining -= bytes_read; 694 iso9660->current_position += bytes_read; 695 __archive_read_consume(a, bytes_read); 696 return (ARCHIVE_OK); 697 } 698 699 static int 700 archive_read_format_iso9660_cleanup(struct archive_read *a) 701 { 702 struct iso9660 *iso9660; 703 struct file_info *file; 704 705 iso9660 = (struct iso9660 *)(a->format->data); 706 while ((file = next_entry(iso9660)) != NULL) 707 release_file(iso9660, file); 708 archive_string_free(&iso9660->pathname); 709 archive_string_free(&iso9660->previous_pathname); 710 if (iso9660->pending_files) 711 free(iso9660->pending_files); 712 free(iso9660); 713 (a->format->data) = NULL; 714 return (ARCHIVE_OK); 715 } 716 717 /* 718 * This routine parses a single ISO directory record, makes sense 719 * of any extensions, and stores the result in memory. 720 */ 721 static struct file_info * 722 parse_file_info(struct iso9660 *iso9660, struct file_info *parent, 723 const unsigned char *isodirrec) 724 { 725 struct file_info *file; 726 size_t name_len; 727 const unsigned char *rr_start, *rr_end; 728 const unsigned char *p; 729 int flags; 730 731 /* TODO: Sanity check that name_len doesn't exceed length, etc. */ 732 733 /* Create a new file entry and copy data from the ISO dir record. */ 734 file = (struct file_info *)malloc(sizeof(*file)); 735 if (file == NULL) 736 return (NULL); 737 memset(file, 0, sizeof(*file)); 738 file->parent = parent; 739 if (parent != NULL) 740 parent->refcount++; 741 file->offset = (uint64_t)toi(isodirrec + DR_extent_offset, DR_extent_size) 742 * iso9660->logical_block_size; 743 file->size = toi(isodirrec + DR_size_offset, DR_size_size); 744 file->mtime = isodate7(isodirrec + DR_date_offset); 745 file->ctime = file->atime = file->mtime; 746 747 name_len = (size_t)isodirrec[DR_name_len_offset]; 748 p = isodirrec + DR_name_offset; 749 /* Rockridge extensions (if any) follow name. Compute this 750 * before fidgeting the name_len below. */ 751 rr_start = p + name_len + (name_len & 1 ? 0 : 1) + iso9660->suspOffset; 752 rr_end = isodirrec + isodirrec[DR_length_offset]; 753 754 if (iso9660->seenJoliet) { 755 /* Joliet names are max 64 chars (128 bytes) according to spec, 756 * but genisoimage (and others?) will allow you to have more. 757 */ 758 wchar_t wbuff[64+1], *wp; 759 const unsigned char *c; 760 761 /* TODO: warn when name_len > 128 ? */ 762 763 /* convert BE UTF-16 to wchar_t */ 764 for (c = p, wp = wbuff; 765 c < (p + name_len) && 766 wp < (wbuff + sizeof(wbuff)/sizeof(*wbuff) - 1); 767 c += 2) { 768 *wp++ = (((255 & (int)c[0]) << 8) | (255 & (int)c[1])); 769 } 770 *wp = L'\0'; 771 772 #if 0 /* untested code, is it at all useful on Joliet? */ 773 /* trim trailing first version and dot from filename. 774 * 775 * Remember we where in UTF-16BE land! 776 * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both 777 * 16 bits big endian characters on Joliet. 778 * 779 * TODO: sanitize filename? 780 * Joliet allows any UCS-2 char except: 781 * *, /, :, ;, ? and \. 782 */ 783 /* Chop off trailing ';1' from files. */ 784 if (*(wp-2) == ';' && *(wp-1) == '1') { 785 wp-=2; 786 *wp = L'\0'; 787 } 788 789 /* Chop off trailing '.' from filenames. */ 790 if (*(wp-1) == '.') 791 *(--wp) = L'\0'; 792 #endif 793 794 /* store the result in the file name field. */ 795 archive_strappend_w_utf8(&file->name, wbuff); 796 } else { 797 /* Chop off trailing ';1' from files. */ 798 if (name_len > 2 && p[name_len - 2] == ';' && 799 p[name_len - 1] == '1') 800 name_len -= 2; 801 /* Chop off trailing '.' from filenames. */ 802 if (name_len > 1 && p[name_len - 1] == '.') 803 --name_len; 804 805 archive_strncpy(&file->name, (const char *)p, name_len); 806 } 807 808 flags = isodirrec[DR_flags_offset]; 809 if (flags & 0x02) 810 file->mode = AE_IFDIR | 0700; 811 else 812 file->mode = AE_IFREG | 0400; 813 814 /* Rockridge extensions overwrite information from above. */ 815 parse_rockridge(iso9660, file, rr_start, rr_end); 816 817 #if DEBUG 818 /* DEBUGGING: Warn about attributes I don't yet fully support. */ 819 if ((flags & ~0x02) != 0) { 820 fprintf(stderr, "\n ** Unrecognized flag: "); 821 dump_isodirrec(stderr, isodirrec); 822 fprintf(stderr, "\n"); 823 } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { 824 fprintf(stderr, "\n ** Unrecognized sequence number: "); 825 dump_isodirrec(stderr, isodirrec); 826 fprintf(stderr, "\n"); 827 } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { 828 fprintf(stderr, "\n ** Unexpected file unit size: "); 829 dump_isodirrec(stderr, isodirrec); 830 fprintf(stderr, "\n"); 831 } else if (*(isodirrec + DR_interleave_offset) != 0) { 832 fprintf(stderr, "\n ** Unexpected interleave: "); 833 dump_isodirrec(stderr, isodirrec); 834 fprintf(stderr, "\n"); 835 } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { 836 fprintf(stderr, "\n ** Unexpected extended attribute length: "); 837 dump_isodirrec(stderr, isodirrec); 838 fprintf(stderr, "\n"); 839 } 840 #endif 841 return (file); 842 } 843 844 static void 845 add_entry(struct iso9660 *iso9660, struct file_info *file) 846 { 847 /* Expand our pending files list as necessary. */ 848 if (iso9660->pending_files_used >= iso9660->pending_files_allocated) { 849 struct file_info **new_pending_files; 850 int new_size = iso9660->pending_files_allocated * 2; 851 852 if (iso9660->pending_files_allocated < 1024) 853 new_size = 1024; 854 /* Overflow might keep us from growing the list. */ 855 if (new_size <= iso9660->pending_files_allocated) 856 __archive_errx(1, "Out of memory"); 857 new_pending_files = (struct file_info **)malloc(new_size * sizeof(new_pending_files[0])); 858 if (new_pending_files == NULL) 859 __archive_errx(1, "Out of memory"); 860 memcpy(new_pending_files, iso9660->pending_files, 861 iso9660->pending_files_allocated * sizeof(new_pending_files[0])); 862 if (iso9660->pending_files != NULL) 863 free(iso9660->pending_files); 864 iso9660->pending_files = new_pending_files; 865 iso9660->pending_files_allocated = new_size; 866 } 867 868 iso9660->pending_files[iso9660->pending_files_used++] = file; 869 } 870 871 static void 872 parse_rockridge(struct iso9660 *iso9660, struct file_info *file, 873 const unsigned char *p, const unsigned char *end) 874 { 875 (void)iso9660; /* UNUSED */ 876 file->name_continues = 0; 877 file->symlink_continues = 0; 878 879 while (p + 4 < end /* Enough space for another entry. */ 880 && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ 881 && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ 882 && p[2] >= 4 /* Sanity-check length. */ 883 && p + p[2] <= end) { /* Sanity-check length. */ 884 const unsigned char *data = p + 4; 885 int data_length = p[2] - 4; 886 int version = p[3]; 887 888 /* 889 * Yes, each 'if' here does test p[0] again. 890 * Otherwise, the fall-through handling to catch 891 * unsupported extensions doesn't work. 892 */ 893 switch(p[0]) { 894 case 'C': 895 if (p[0] == 'C' && p[1] == 'E') { 896 if (version == 1 && data_length == 24) { 897 /* 898 * CE extension comprises: 899 * 8 byte sector containing extension 900 * 8 byte offset w/in above sector 901 * 8 byte length of continuation 902 */ 903 file->ce_offset = (uint64_t)toi(data, 4) 904 * iso9660->logical_block_size 905 + toi(data + 8, 4); 906 file->ce_size = toi(data + 16, 4); 907 /* If the result is rediculous, 908 * ignore it. */ 909 if (file->ce_offset + file->ce_size 910 > iso9660->volume_size) { 911 file->ce_offset = 0; 912 file->ce_size = 0; 913 } 914 } 915 break; 916 } 917 /* FALLTHROUGH */ 918 case 'N': 919 if (p[0] == 'N' && p[1] == 'M') { 920 if (version == 1) 921 parse_rockridge_NM1(file, 922 data, data_length); 923 break; 924 } 925 /* FALLTHROUGH */ 926 case 'P': 927 if (p[0] == 'P' && p[1] == 'D') { 928 /* 929 * PD extension is padding; 930 * contents are always ignored. 931 */ 932 break; 933 } 934 if (p[0] == 'P' && p[1] == 'N') { 935 if (version == 1 && data_length == 16) { 936 file->rdev = toi(data,4); 937 file->rdev <<= 32; 938 file->rdev |= toi(data + 8, 4); 939 } 940 break; 941 } 942 if (p[0] == 'P' && p[1] == 'X') { 943 /* 944 * PX extension comprises: 945 * 8 bytes for mode, 946 * 8 bytes for nlinks, 947 * 8 bytes for uid, 948 * 8 bytes for gid, 949 * 8 bytes for inode. 950 */ 951 if (version == 1) { 952 if (data_length >= 8) 953 file->mode 954 = toi(data, 4); 955 if (data_length >= 16) 956 file->nlinks 957 = toi(data + 8, 4); 958 if (data_length >= 24) 959 file->uid 960 = toi(data + 16, 4); 961 if (data_length >= 32) 962 file->gid 963 = toi(data + 24, 4); 964 if (data_length >= 40) 965 file->inode 966 = toi(data + 32, 4); 967 } 968 break; 969 } 970 /* FALLTHROUGH */ 971 case 'R': 972 if (p[0] == 'R' && p[1] == 'R' && version == 1) { 973 iso9660->seenRockridge = 1; 974 /* 975 * RR extension comprises: 976 * one byte flag value 977 */ 978 /* TODO: Handle RR extension. */ 979 break; 980 } 981 /* FALLTHROUGH */ 982 case 'S': 983 if (p[0] == 'S' && p[1] == 'L') { 984 if (version == 1) 985 parse_rockridge_SL1(file, 986 data, data_length); 987 break; 988 } 989 if (p[0] == 'S' && p[1] == 'P' 990 && version == 1 && data_length == 3 991 && data[0] == (unsigned char)'\xbe' 992 && data[1] == (unsigned char)'\xef') { 993 /* 994 * SP extension stores the suspOffset 995 * (Number of bytes to skip between 996 * filename and SUSP records.) 997 * It is mandatory by the SUSP standard 998 * (IEEE 1281). 999 * 1000 * It allows SUSP to coexist with 1001 * non-SUSP uses of the System 1002 * Use Area by placing non-SUSP data 1003 * before SUSP data. 1004 * 1005 * TODO: Add a check for 'SP' in 1006 * first directory entry, disable all SUSP 1007 * processing if not found. 1008 */ 1009 iso9660->suspOffset = data[2]; 1010 break; 1011 } 1012 if (p[0] == 'S' && p[1] == 'T' 1013 && data_length == 0 && version == 1) { 1014 /* 1015 * ST extension marks end of this 1016 * block of SUSP entries. 1017 * 1018 * It allows SUSP to coexist with 1019 * non-SUSP uses of the System 1020 * Use Area by placing non-SUSP data 1021 * after SUSP data. 1022 */ 1023 return; 1024 } 1025 case 'T': 1026 if (p[0] == 'T' && p[1] == 'F') { 1027 if (version == 1) 1028 parse_rockridge_TF1(file, 1029 data, data_length); 1030 break; 1031 } 1032 /* FALLTHROUGH */ 1033 default: 1034 /* The FALLTHROUGHs above leave us here for 1035 * any unsupported extension. */ 1036 #if DEBUG 1037 { 1038 const unsigned char *t; 1039 fprintf(stderr, "\nUnsupported RRIP extension for %s\n", file->name.s); 1040 fprintf(stderr, " %c%c(%d):", p[0], p[1], data_length); 1041 for (t = data; t < data + data_length && t < data + 16; t++) 1042 fprintf(stderr, " %02x", *t); 1043 fprintf(stderr, "\n"); 1044 } 1045 #endif 1046 break; 1047 } 1048 1049 1050 1051 p += p[2]; 1052 } 1053 } 1054 1055 static void 1056 parse_rockridge_NM1(struct file_info *file, 1057 const unsigned char *data, int data_length) 1058 { 1059 if (!file->name_continues) 1060 archive_string_empty(&file->name); 1061 file->name_continues = 0; 1062 if (data_length < 1) 1063 return; 1064 /* 1065 * NM version 1 extension comprises: 1066 * 1 byte flag, value is one of: 1067 * = 0: remainder is name 1068 * = 1: remainder is name, next NM entry continues name 1069 * = 2: "." 1070 * = 4: ".." 1071 * = 32: Implementation specific 1072 * All other values are reserved. 1073 */ 1074 switch(data[0]) { 1075 case 0: 1076 if (data_length < 2) 1077 return; 1078 archive_strncat(&file->name, (const char *)data + 1, data_length - 1); 1079 break; 1080 case 1: 1081 if (data_length < 2) 1082 return; 1083 archive_strncat(&file->name, (const char *)data + 1, data_length - 1); 1084 file->name_continues = 1; 1085 break; 1086 case 2: 1087 archive_strcat(&file->name, "."); 1088 break; 1089 case 4: 1090 archive_strcat(&file->name, ".."); 1091 break; 1092 default: 1093 return; 1094 } 1095 1096 } 1097 1098 static void 1099 parse_rockridge_TF1(struct file_info *file, const unsigned char *data, 1100 int data_length) 1101 { 1102 char flag; 1103 /* 1104 * TF extension comprises: 1105 * one byte flag 1106 * create time (optional) 1107 * modify time (optional) 1108 * access time (optional) 1109 * attribute time (optional) 1110 * Time format and presence of fields 1111 * is controlled by flag bits. 1112 */ 1113 if (data_length < 1) 1114 return; 1115 flag = data[0]; 1116 ++data; 1117 --data_length; 1118 if (flag & 0x80) { 1119 /* Use 17-byte time format. */ 1120 if ((flag & 1) && data_length >= 17) { 1121 /* Create time. */ 1122 file->birthtime = isodate17(data); 1123 data += 17; 1124 data_length -= 17; 1125 } 1126 if ((flag & 2) && data_length >= 17) { 1127 /* Modify time. */ 1128 file->mtime = isodate17(data); 1129 data += 17; 1130 data_length -= 17; 1131 } 1132 if ((flag & 4) && data_length >= 17) { 1133 /* Access time. */ 1134 file->atime = isodate17(data); 1135 data += 17; 1136 data_length -= 17; 1137 } 1138 if ((flag & 8) && data_length >= 17) { 1139 /* Attribute change time. */ 1140 file->ctime = isodate17(data); 1141 data += 17; 1142 data_length -= 17; 1143 } 1144 } else { 1145 /* Use 7-byte time format. */ 1146 if ((flag & 1) && data_length >= 7) { 1147 /* Create time. */ 1148 file->birthtime = isodate17(data); 1149 data += 7; 1150 data_length -= 7; 1151 } 1152 if ((flag & 2) && data_length >= 7) { 1153 /* Modify time. */ 1154 file->mtime = isodate7(data); 1155 data += 7; 1156 data_length -= 7; 1157 } 1158 if ((flag & 4) && data_length >= 7) { 1159 /* Access time. */ 1160 file->atime = isodate7(data); 1161 data += 7; 1162 data_length -= 7; 1163 } 1164 if ((flag & 8) && data_length >= 7) { 1165 /* Attribute change time. */ 1166 file->ctime = isodate7(data); 1167 data += 7; 1168 data_length -= 7; 1169 } 1170 } 1171 } 1172 1173 static void 1174 parse_rockridge_SL1(struct file_info *file, const unsigned char *data, 1175 int data_length) 1176 { 1177 int component_continues = 1; 1178 1179 if (!file->symlink_continues) 1180 archive_string_empty(&file->symlink); 1181 else 1182 archive_strcat(&file->symlink, "/"); 1183 file->symlink_continues = 0; 1184 1185 /* 1186 * Defined flag values: 1187 * 0: This is the last SL record for this symbolic link 1188 * 1: this symbolic link field continues in next SL entry 1189 * All other values are reserved. 1190 */ 1191 if (data_length < 1) 1192 return; 1193 switch(*data) { 1194 case 0: 1195 break; 1196 case 1: 1197 file->symlink_continues = 1; 1198 break; 1199 default: 1200 return; 1201 } 1202 ++data; /* Skip flag byte. */ 1203 --data_length; 1204 1205 /* 1206 * SL extension body stores "components". 1207 * Basically, this is a complicated way of storing 1208 * a POSIX path. It also interferes with using 1209 * symlinks for storing non-path data. <sigh> 1210 * 1211 * Each component is 2 bytes (flag and length) 1212 * possibly followed by name data. 1213 */ 1214 while (data_length >= 2) { 1215 unsigned char flag = *data++; 1216 unsigned char nlen = *data++; 1217 data_length -= 2; 1218 1219 if (!component_continues) 1220 archive_strcat(&file->symlink, "/"); 1221 component_continues = 0; 1222 1223 switch(flag) { 1224 case 0: /* Usual case, this is text. */ 1225 if (data_length < nlen) 1226 return; 1227 archive_strncat(&file->symlink, 1228 (const char *)data, nlen); 1229 break; 1230 case 0x01: /* Text continues in next component. */ 1231 if (data_length < nlen) 1232 return; 1233 archive_strncat(&file->symlink, 1234 (const char *)data, nlen); 1235 component_continues = 1; 1236 break; 1237 case 0x02: /* Current dir. */ 1238 archive_strcat(&file->symlink, "."); 1239 break; 1240 case 0x04: /* Parent dir. */ 1241 archive_strcat(&file->symlink, ".."); 1242 break; 1243 case 0x08: /* Root of filesystem. */ 1244 archive_string_empty(&file->symlink); 1245 archive_strcat(&file->symlink, "/"); 1246 break; 1247 case 0x10: /* Undefined (historically "volume root" */ 1248 archive_string_empty(&file->symlink); 1249 archive_strcat(&file->symlink, "ROOT"); 1250 break; 1251 case 0x20: /* Undefined (historically "hostname") */ 1252 archive_strcat(&file->symlink, "hostname"); 1253 break; 1254 default: 1255 /* TODO: issue a warning ? */ 1256 return; 1257 } 1258 data += nlen; 1259 data_length -= nlen; 1260 } 1261 } 1262 1263 1264 static void 1265 release_file(struct iso9660 *iso9660, struct file_info *file) 1266 { 1267 struct file_info *parent; 1268 1269 if (file == NULL) 1270 return; 1271 1272 if (file->refcount == 0) { 1273 parent = file->parent; 1274 archive_string_free(&file->name); 1275 archive_string_free(&file->symlink); 1276 free(file); 1277 if (parent != NULL) { 1278 parent->refcount--; 1279 release_file(iso9660, parent); 1280 } 1281 } 1282 } 1283 1284 static int 1285 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, 1286 struct file_info **pfile) 1287 { 1288 struct file_info *file; 1289 uint64_t offset; 1290 1291 *pfile = NULL; 1292 for (;;) { 1293 *pfile = file = next_entry(iso9660); 1294 if (file == NULL) 1295 return (ARCHIVE_EOF); 1296 1297 /* CE area precedes actual file data? Ignore it. */ 1298 if (file->ce_offset > file->offset) { 1299 /* fprintf(stderr, " *** Discarding CE data.\n"); */ 1300 file->ce_offset = 0; 1301 file->ce_size = 0; 1302 } 1303 1304 /* Don't waste time seeking for zero-length bodies. */ 1305 if (file->size == 0) { 1306 file->offset = iso9660->current_position; 1307 } 1308 1309 /* If CE exists, find and read it now. */ 1310 if (file->ce_offset > 0) 1311 offset = file->ce_offset; 1312 else 1313 offset = file->offset; 1314 1315 /* Seek forward to the start of the entry. */ 1316 if (iso9660->current_position < offset) { 1317 off_t step = offset - iso9660->current_position; 1318 off_t bytes_read; 1319 bytes_read = __archive_read_skip(a, step); 1320 if (bytes_read < 0) 1321 return (bytes_read); 1322 iso9660->current_position = offset; 1323 } 1324 1325 /* We found body of file; handle it now. */ 1326 if (offset == file->offset) 1327 return (ARCHIVE_OK); 1328 1329 /* Found CE? Process it and push the file back onto list. */ 1330 if (offset == file->ce_offset) { 1331 const void *p; 1332 ssize_t size = file->ce_size; 1333 const unsigned char *rr_start; 1334 1335 file->ce_offset = 0; 1336 file->ce_size = 0; 1337 p = __archive_read_ahead(a, size, NULL); 1338 if (p == NULL) 1339 return (ARCHIVE_FATAL); 1340 rr_start = (const unsigned char *)p; 1341 parse_rockridge(iso9660, file, rr_start, 1342 rr_start + size); 1343 __archive_read_consume(a, size); 1344 iso9660->current_position += size; 1345 add_entry(iso9660, file); 1346 } 1347 } 1348 } 1349 1350 static struct file_info * 1351 next_entry(struct iso9660 *iso9660) 1352 { 1353 int least_index; 1354 uint64_t least_end_offset; 1355 int i; 1356 struct file_info *r; 1357 1358 if (iso9660->pending_files_used < 1) 1359 return (NULL); 1360 1361 /* Assume the first file in the list is the earliest on disk. */ 1362 least_index = 0; 1363 least_end_offset = iso9660->pending_files[0]->offset 1364 + iso9660->pending_files[0]->size; 1365 1366 /* Now, try to find an earlier one. */ 1367 for (i = 0; i < iso9660->pending_files_used; i++) { 1368 /* Use the position of the file *end* as our comparison. */ 1369 uint64_t end_offset = iso9660->pending_files[i]->offset 1370 + iso9660->pending_files[i]->size; 1371 if (iso9660->pending_files[i]->ce_offset > 0 1372 && iso9660->pending_files[i]->ce_offset < iso9660->pending_files[i]->offset) 1373 end_offset = iso9660->pending_files[i]->ce_offset 1374 + iso9660->pending_files[i]->ce_size; 1375 if (least_end_offset > end_offset) { 1376 least_index = i; 1377 least_end_offset = end_offset; 1378 } 1379 } 1380 r = iso9660->pending_files[least_index]; 1381 iso9660->pending_files[least_index] 1382 = iso9660->pending_files[--iso9660->pending_files_used]; 1383 return (r); 1384 } 1385 1386 static unsigned int 1387 toi(const void *p, int n) 1388 { 1389 const unsigned char *v = (const unsigned char *)p; 1390 if (n > 1) 1391 return v[0] + 256 * toi(v + 1, n - 1); 1392 if (n == 1) 1393 return v[0]; 1394 return (0); 1395 } 1396 1397 static time_t 1398 isodate7(const unsigned char *v) 1399 { 1400 struct tm tm; 1401 int offset; 1402 memset(&tm, 0, sizeof(tm)); 1403 tm.tm_year = v[0]; 1404 tm.tm_mon = v[1] - 1; 1405 tm.tm_mday = v[2]; 1406 tm.tm_hour = v[3]; 1407 tm.tm_min = v[4]; 1408 tm.tm_sec = v[5]; 1409 /* v[6] is the signed timezone offset, in 1/4-hour increments. */ 1410 offset = ((const signed char *)v)[6]; 1411 if (offset > -48 && offset < 52) { 1412 tm.tm_hour -= offset / 4; 1413 tm.tm_min -= (offset % 4) * 15; 1414 } 1415 return (time_from_tm(&tm)); 1416 } 1417 1418 static time_t 1419 isodate17(const unsigned char *v) 1420 { 1421 struct tm tm; 1422 int offset; 1423 memset(&tm, 0, sizeof(tm)); 1424 tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 1425 + (v[2] - '0') * 10 + (v[3] - '0') 1426 - 1900; 1427 tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); 1428 tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); 1429 tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); 1430 tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); 1431 tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); 1432 /* v[16] is the signed timezone offset, in 1/4-hour increments. */ 1433 offset = ((const signed char *)v)[16]; 1434 if (offset > -48 && offset < 52) { 1435 tm.tm_hour -= offset / 4; 1436 tm.tm_min -= (offset % 4) * 15; 1437 } 1438 return (time_from_tm(&tm)); 1439 } 1440 1441 static time_t 1442 time_from_tm(struct tm *t) 1443 { 1444 #if HAVE_TIMEGM 1445 /* Use platform timegm() if available. */ 1446 return (timegm(t)); 1447 #else 1448 /* Else use direct calculation using POSIX assumptions. */ 1449 /* First, fix up tm_yday based on the year/month/day. */ 1450 mktime(t); 1451 /* Then we can compute timegm() from first principles. */ 1452 return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 1453 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 1454 + ((t->tm_year - 69) / 4) * 86400 - 1455 ((t->tm_year - 1) / 100) * 86400 1456 + ((t->tm_year + 299) / 400) * 86400); 1457 #endif 1458 } 1459 1460 static const char * 1461 build_pathname(struct archive_string *as, struct file_info *file) 1462 { 1463 if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { 1464 build_pathname(as, file->parent); 1465 archive_strcat(as, "/"); 1466 } 1467 if (archive_strlen(&file->name) == 0) 1468 archive_strcat(as, "."); 1469 else 1470 archive_string_concat(as, &file->name); 1471 return (as->s); 1472 } 1473 1474 #if DEBUG 1475 static void 1476 dump_isodirrec(FILE *out, const unsigned char *isodirrec) 1477 { 1478 fprintf(out, " l %d,", 1479 toi(isodirrec + DR_length_offset, DR_length_size)); 1480 fprintf(out, " a %d,", 1481 toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); 1482 fprintf(out, " ext 0x%x,", 1483 toi(isodirrec + DR_extent_offset, DR_extent_size)); 1484 fprintf(out, " s %d,", 1485 toi(isodirrec + DR_size_offset, DR_extent_size)); 1486 fprintf(out, " f 0x%02x,", 1487 toi(isodirrec + DR_flags_offset, DR_flags_size)); 1488 fprintf(out, " u %d,", 1489 toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); 1490 fprintf(out, " ilv %d,", 1491 toi(isodirrec + DR_interleave_offset, DR_interleave_size)); 1492 fprintf(out, " seq %d,", 1493 toi(isodirrec + DR_volume_sequence_number_offset, DR_volume_sequence_number_size)); 1494 fprintf(out, " nl %d:", 1495 toi(isodirrec + DR_name_len_offset, DR_name_len_size)); 1496 fprintf(out, " `%.*s'", 1497 toi(isodirrec + DR_name_len_offset, DR_name_len_size), isodirrec + DR_name_offset); 1498 } 1499 #endif 1500