1 /*- 2 * Copyright (c) 2003-2007 Tim Kientzle 3 * Copyright (c) 2009 Andreas Henriksson <andreas@fatal.se> 4 * Copyright (c) 2009-2011 Michihiro NAKAJIMA 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 #include "archive_platform.h" 29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_iso9660.c 201246 2009-12-30 05:30:35Z kientzle $"); 30 31 #ifdef HAVE_ERRNO_H 32 #include <errno.h> 33 #endif 34 /* #include <stdint.h> */ /* See archive_platform.h */ 35 #include <stdio.h> 36 #ifdef HAVE_STDLIB_H 37 #include <stdlib.h> 38 #endif 39 #ifdef HAVE_STRING_H 40 #include <string.h> 41 #endif 42 #include <time.h> 43 #ifdef HAVE_ZLIB_H 44 #include <zlib.h> 45 #endif 46 47 #include "archive.h" 48 #include "archive_endian.h" 49 #include "archive_entry.h" 50 #include "archive_entry_locale.h" 51 #include "archive_private.h" 52 #include "archive_read_private.h" 53 #include "archive_string.h" 54 55 /* 56 * An overview of ISO 9660 format: 57 * 58 * Each disk is laid out as follows: 59 * * 32k reserved for private use 60 * * Volume descriptor table. Each volume descriptor 61 * is 2k and specifies basic format information. 62 * The "Primary Volume Descriptor" (PVD) is defined by the 63 * standard and should always be present; other volume 64 * descriptors include various vendor-specific extensions. 65 * * Files and directories. Each file/dir is specified by 66 * an "extent" (starting sector and length in bytes). 67 * Dirs are just files with directory records packed one 68 * after another. The PVD contains a single dir entry 69 * specifying the location of the root directory. Everything 70 * else follows from there. 71 * 72 * This module works by first reading the volume descriptors, then 73 * building a list of directory entries, sorted by starting 74 * sector. At each step, I look for the earliest dir entry that 75 * hasn't yet been read, seek forward to that location and read 76 * that entry. If it's a dir, I slurp in the new dir entries and 77 * add them to the heap; if it's a regular file, I return the 78 * corresponding archive_entry and wait for the client to request 79 * the file body. This strategy allows us to read most compliant 80 * CDs with a single pass through the data, as required by libarchive. 81 */ 82 #define LOGICAL_BLOCK_SIZE 2048 83 #define SYSTEM_AREA_BLOCK 16 84 85 /* Structure of on-disk primary volume descriptor. */ 86 #define PVD_type_offset 0 87 #define PVD_type_size 1 88 #define PVD_id_offset (PVD_type_offset + PVD_type_size) 89 #define PVD_id_size 5 90 #define PVD_version_offset (PVD_id_offset + PVD_id_size) 91 #define PVD_version_size 1 92 #define PVD_reserved1_offset (PVD_version_offset + PVD_version_size) 93 #define PVD_reserved1_size 1 94 #define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size) 95 #define PVD_system_id_size 32 96 #define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size) 97 #define PVD_volume_id_size 32 98 #define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size) 99 #define PVD_reserved2_size 8 100 #define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size) 101 #define PVD_volume_space_size_size 8 102 #define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size) 103 #define PVD_reserved3_size 32 104 #define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size) 105 #define PVD_volume_set_size_size 4 106 #define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size) 107 #define PVD_volume_sequence_number_size 4 108 #define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size) 109 #define PVD_logical_block_size_size 4 110 #define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size) 111 #define PVD_path_table_size_size 8 112 #define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size) 113 #define PVD_type_1_path_table_size 4 114 #define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size) 115 #define PVD_opt_type_1_path_table_size 4 116 #define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size) 117 #define PVD_type_m_path_table_size 4 118 #define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size) 119 #define PVD_opt_type_m_path_table_size 4 120 #define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size) 121 #define PVD_root_directory_record_size 34 122 #define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size) 123 #define PVD_volume_set_id_size 128 124 #define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size) 125 #define PVD_publisher_id_size 128 126 #define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size) 127 #define PVD_preparer_id_size 128 128 #define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size) 129 #define PVD_application_id_size 128 130 #define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size) 131 #define PVD_copyright_file_id_size 37 132 #define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size) 133 #define PVD_abstract_file_id_size 37 134 #define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size) 135 #define PVD_bibliographic_file_id_size 37 136 #define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size) 137 #define PVD_creation_date_size 17 138 #define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size) 139 #define PVD_modification_date_size 17 140 #define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size) 141 #define PVD_expiration_date_size 17 142 #define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size) 143 #define PVD_effective_date_size 17 144 #define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size) 145 #define PVD_file_structure_version_size 1 146 #define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size) 147 #define PVD_reserved4_size 1 148 #define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size) 149 #define PVD_application_data_size 512 150 #define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size) 151 #define PVD_reserved5_size (2048 - PVD_reserved5_offset) 152 153 /* TODO: It would make future maintenance easier to just hardcode the 154 * above values. In particular, ECMA119 states the offsets as part of 155 * the standard. That would eliminate the need for the following check.*/ 156 #if PVD_reserved5_offset != 1395 157 #error PVD offset and size definitions are wrong. 158 #endif 159 160 161 /* Structure of optional on-disk supplementary volume descriptor. */ 162 #define SVD_type_offset 0 163 #define SVD_type_size 1 164 #define SVD_id_offset (SVD_type_offset + SVD_type_size) 165 #define SVD_id_size 5 166 #define SVD_version_offset (SVD_id_offset + SVD_id_size) 167 #define SVD_version_size 1 168 /* ... */ 169 #define SVD_reserved1_offset 72 170 #define SVD_reserved1_size 8 171 #define SVD_volume_space_size_offset 80 172 #define SVD_volume_space_size_size 8 173 #define SVD_escape_sequences_offset (SVD_volume_space_size_offset + SVD_volume_space_size_size) 174 #define SVD_escape_sequences_size 32 175 /* ... */ 176 #define SVD_logical_block_size_offset 128 177 #define SVD_logical_block_size_size 4 178 #define SVD_type_L_path_table_offset 140 179 #define SVD_type_M_path_table_offset 148 180 /* ... */ 181 #define SVD_root_directory_record_offset 156 182 #define SVD_root_directory_record_size 34 183 #define SVD_file_structure_version_offset 881 184 #define SVD_reserved2_offset 882 185 #define SVD_reserved2_size 1 186 #define SVD_reserved3_offset 1395 187 #define SVD_reserved3_size 653 188 /* ... */ 189 /* FIXME: validate correctness of last SVD entry offset. */ 190 191 /* Structure of an on-disk directory record. */ 192 /* Note: ISO9660 stores each multi-byte integer twice, once in 193 * each byte order. The sizes here are the size of just one 194 * of the two integers. (This is why the offset of a field isn't 195 * the same as the offset+size of the previous field.) */ 196 #define DR_length_offset 0 197 #define DR_length_size 1 198 #define DR_ext_attr_length_offset 1 199 #define DR_ext_attr_length_size 1 200 #define DR_extent_offset 2 201 #define DR_extent_size 4 202 #define DR_size_offset 10 203 #define DR_size_size 4 204 #define DR_date_offset 18 205 #define DR_date_size 7 206 #define DR_flags_offset 25 207 #define DR_flags_size 1 208 #define DR_file_unit_size_offset 26 209 #define DR_file_unit_size_size 1 210 #define DR_interleave_offset 27 211 #define DR_interleave_size 1 212 #define DR_volume_sequence_number_offset 28 213 #define DR_volume_sequence_number_size 2 214 #define DR_name_len_offset 32 215 #define DR_name_len_size 1 216 #define DR_name_offset 33 217 218 #ifdef HAVE_ZLIB_H 219 static const unsigned char zisofs_magic[8] = { 220 0x37, 0xE4, 0x53, 0x96, 0xC9, 0xDB, 0xD6, 0x07 221 }; 222 223 struct zisofs { 224 /* Set 1 if this file compressed by paged zlib */ 225 int pz; 226 int pz_log2_bs; /* Log2 of block size */ 227 uint64_t pz_uncompressed_size; 228 229 int initialized; 230 unsigned char *uncompressed_buffer; 231 size_t uncompressed_buffer_size; 232 233 uint32_t pz_offset; 234 unsigned char header[16]; 235 size_t header_avail; 236 int header_passed; 237 unsigned char *block_pointers; 238 size_t block_pointers_alloc; 239 size_t block_pointers_size; 240 size_t block_pointers_avail; 241 size_t block_off; 242 uint32_t block_avail; 243 244 z_stream stream; 245 int stream_valid; 246 }; 247 #else 248 struct zisofs { 249 /* Set 1 if this file compressed by paged zlib */ 250 int pz; 251 }; 252 #endif 253 254 struct content { 255 uint64_t offset;/* Offset on disk. */ 256 uint64_t size; /* File size in bytes. */ 257 struct content *next; 258 }; 259 260 /* In-memory storage for a directory record. */ 261 struct file_info { 262 struct file_info *use_next; 263 struct file_info *parent; 264 struct file_info *next; 265 struct file_info *re_next; 266 int subdirs; 267 uint64_t key; /* Heap Key. */ 268 uint64_t offset; /* Offset on disk. */ 269 uint64_t size; /* File size in bytes. */ 270 uint32_t ce_offset; /* Offset of CE. */ 271 uint32_t ce_size; /* Size of CE. */ 272 char rr_moved; /* Flag to rr_moved. */ 273 char rr_moved_has_re_only; 274 char re; /* Having RRIP "RE" extension. */ 275 char re_descendant; 276 uint64_t cl_offset; /* Having RRIP "CL" extension. */ 277 int birthtime_is_set; 278 time_t birthtime; /* File created time. */ 279 time_t mtime; /* File last modified time. */ 280 time_t atime; /* File last accessed time. */ 281 time_t ctime; /* File attribute change time. */ 282 uint64_t rdev; /* Device number. */ 283 mode_t mode; 284 uid_t uid; 285 gid_t gid; 286 int64_t number; 287 int nlinks; 288 struct archive_string name; /* Pathname */ 289 unsigned char *utf16be_name; 290 size_t utf16be_bytes; 291 char name_continues; /* Non-zero if name continues */ 292 struct archive_string symlink; 293 char symlink_continues; /* Non-zero if link continues */ 294 /* Set 1 if this file compressed by paged zlib(zisofs) */ 295 int pz; 296 int pz_log2_bs; /* Log2 of block size */ 297 uint64_t pz_uncompressed_size; 298 /* Set 1 if this file is multi extent. */ 299 int multi_extent; 300 struct { 301 struct content *first; 302 struct content **last; 303 } contents; 304 struct { 305 struct file_info *first; 306 struct file_info **last; 307 } rede_files; 308 }; 309 310 struct heap_queue { 311 struct file_info **files; 312 int allocated; 313 int used; 314 }; 315 316 struct iso9660 { 317 int magic; 318 #define ISO9660_MAGIC 0x96609660 319 320 int opt_support_joliet; 321 int opt_support_rockridge; 322 323 struct archive_string pathname; 324 char seenRockridge; /* Set true if RR extensions are used. */ 325 char seenSUSP; /* Set true if SUSP is beging used. */ 326 char seenJoliet; 327 328 unsigned char suspOffset; 329 struct file_info *rr_moved; 330 struct read_ce_queue { 331 struct read_ce_req { 332 uint64_t offset;/* Offset of CE on disk. */ 333 struct file_info *file; 334 } *reqs; 335 int cnt; 336 int allocated; 337 } read_ce_req; 338 339 int64_t previous_number; 340 struct archive_string previous_pathname; 341 342 struct file_info *use_files; 343 struct heap_queue pending_files; 344 struct { 345 struct file_info *first; 346 struct file_info **last; 347 } cache_files; 348 struct { 349 struct file_info *first; 350 struct file_info **last; 351 } re_files; 352 353 uint64_t current_position; 354 ssize_t logical_block_size; 355 uint64_t volume_size; /* Total size of volume in bytes. */ 356 int32_t volume_block;/* Total size of volume in logical blocks. */ 357 358 struct vd { 359 int location; /* Location of Extent. */ 360 uint32_t size; 361 } primary, joliet; 362 363 int64_t entry_sparse_offset; 364 int64_t entry_bytes_remaining; 365 size_t entry_bytes_unconsumed; 366 struct zisofs entry_zisofs; 367 struct content *entry_content; 368 struct archive_string_conv *sconv_utf16be; 369 /* 370 * Buffers for a full pathname in UTF-16BE in Joliet extensions. 371 */ 372 #define UTF16_NAME_MAX 1024 373 unsigned char *utf16be_path; 374 size_t utf16be_path_len; 375 unsigned char *utf16be_previous_path; 376 size_t utf16be_previous_path_len; 377 }; 378 379 static int archive_read_format_iso9660_bid(struct archive_read *, int); 380 static int archive_read_format_iso9660_options(struct archive_read *, 381 const char *, const char *); 382 static int archive_read_format_iso9660_cleanup(struct archive_read *); 383 static int archive_read_format_iso9660_read_data(struct archive_read *, 384 const void **, size_t *, int64_t *); 385 static int archive_read_format_iso9660_read_data_skip(struct archive_read *); 386 static int archive_read_format_iso9660_read_header(struct archive_read *, 387 struct archive_entry *); 388 static const char *build_pathname(struct archive_string *, struct file_info *); 389 static int build_pathname_utf16be(unsigned char *, size_t, size_t *, 390 struct file_info *); 391 #if DEBUG 392 static void dump_isodirrec(FILE *, const unsigned char *isodirrec); 393 #endif 394 static time_t time_from_tm(struct tm *); 395 static time_t isodate17(const unsigned char *); 396 static time_t isodate7(const unsigned char *); 397 static int isBootRecord(struct iso9660 *, const unsigned char *); 398 static int isVolumePartition(struct iso9660 *, const unsigned char *); 399 static int isVDSetTerminator(struct iso9660 *, const unsigned char *); 400 static int isJolietSVD(struct iso9660 *, const unsigned char *); 401 static int isSVD(struct iso9660 *, const unsigned char *); 402 static int isEVD(struct iso9660 *, const unsigned char *); 403 static int isPVD(struct iso9660 *, const unsigned char *); 404 static int next_cache_entry(struct archive_read *, struct iso9660 *, 405 struct file_info **); 406 static int next_entry_seek(struct archive_read *, struct iso9660 *, 407 struct file_info **); 408 static struct file_info * 409 parse_file_info(struct archive_read *a, 410 struct file_info *parent, const unsigned char *isodirrec); 411 static int parse_rockridge(struct archive_read *a, 412 struct file_info *file, const unsigned char *start, 413 const unsigned char *end); 414 static int register_CE(struct archive_read *a, int32_t location, 415 struct file_info *file); 416 static int read_CE(struct archive_read *a, struct iso9660 *iso9660); 417 static void parse_rockridge_NM1(struct file_info *, 418 const unsigned char *, int); 419 static void parse_rockridge_SL1(struct file_info *, 420 const unsigned char *, int); 421 static void parse_rockridge_TF1(struct file_info *, 422 const unsigned char *, int); 423 static void parse_rockridge_ZF1(struct file_info *, 424 const unsigned char *, int); 425 static void register_file(struct iso9660 *, struct file_info *); 426 static void release_files(struct iso9660 *); 427 static unsigned toi(const void *p, int n); 428 static inline void re_add_entry(struct iso9660 *, struct file_info *); 429 static inline struct file_info * re_get_entry(struct iso9660 *); 430 static inline int rede_add_entry(struct file_info *); 431 static inline struct file_info * rede_get_entry(struct file_info *); 432 static inline void cache_add_entry(struct iso9660 *iso9660, 433 struct file_info *file); 434 static inline struct file_info *cache_get_entry(struct iso9660 *iso9660); 435 static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, 436 struct file_info *file, uint64_t key); 437 static struct file_info *heap_get_entry(struct heap_queue *heap); 438 439 #define add_entry(arch, iso9660, file) \ 440 heap_add_entry(arch, &((iso9660)->pending_files), file, file->offset) 441 #define next_entry(iso9660) \ 442 heap_get_entry(&((iso9660)->pending_files)) 443 444 int 445 archive_read_support_format_iso9660(struct archive *_a) 446 { 447 struct archive_read *a = (struct archive_read *)_a; 448 struct iso9660 *iso9660; 449 int r; 450 451 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 452 ARCHIVE_STATE_NEW, "archive_read_support_format_iso9660"); 453 454 iso9660 = (struct iso9660 *)calloc(1, sizeof(*iso9660)); 455 if (iso9660 == NULL) { 456 archive_set_error(&a->archive, ENOMEM, 457 "Can't allocate iso9660 data"); 458 return (ARCHIVE_FATAL); 459 } 460 iso9660->magic = ISO9660_MAGIC; 461 iso9660->cache_files.first = NULL; 462 iso9660->cache_files.last = &(iso9660->cache_files.first); 463 iso9660->re_files.first = NULL; 464 iso9660->re_files.last = &(iso9660->re_files.first); 465 /* Enable to support Joliet extensions by default. */ 466 iso9660->opt_support_joliet = 1; 467 /* Enable to support Rock Ridge extensions by default. */ 468 iso9660->opt_support_rockridge = 1; 469 470 r = __archive_read_register_format(a, 471 iso9660, 472 "iso9660", 473 archive_read_format_iso9660_bid, 474 archive_read_format_iso9660_options, 475 archive_read_format_iso9660_read_header, 476 archive_read_format_iso9660_read_data, 477 archive_read_format_iso9660_read_data_skip, 478 archive_read_format_iso9660_cleanup); 479 480 if (r != ARCHIVE_OK) { 481 free(iso9660); 482 return (r); 483 } 484 return (ARCHIVE_OK); 485 } 486 487 488 static int 489 archive_read_format_iso9660_bid(struct archive_read *a, int best_bid) 490 { 491 struct iso9660 *iso9660; 492 ssize_t bytes_read; 493 const unsigned char *p; 494 int seenTerminator; 495 496 /* If there's already a better bid than we can ever 497 make, don't bother testing. */ 498 if (best_bid > 48) 499 return (-1); 500 501 iso9660 = (struct iso9660 *)(a->format->data); 502 503 /* 504 * Skip the first 32k (reserved area) and get the first 505 * 8 sectors of the volume descriptor table. Of course, 506 * if the I/O layer gives us more, we'll take it. 507 */ 508 #define RESERVED_AREA (SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE) 509 p = __archive_read_ahead(a, 510 RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE, 511 &bytes_read); 512 if (p == NULL) 513 return (-1); 514 515 /* Skip the reserved area. */ 516 bytes_read -= RESERVED_AREA; 517 p += RESERVED_AREA; 518 519 /* Check each volume descriptor. */ 520 seenTerminator = 0; 521 for (; bytes_read > LOGICAL_BLOCK_SIZE; 522 bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) { 523 /* Do not handle undefined Volume Descriptor Type. */ 524 if (p[0] >= 4 && p[0] <= 254) 525 return (0); 526 /* Standard Identifier must be "CD001" */ 527 if (memcmp(p + 1, "CD001", 5) != 0) 528 return (0); 529 if (isPVD(iso9660, p)) 530 continue; 531 if (!iso9660->joliet.location) { 532 if (isJolietSVD(iso9660, p)) 533 continue; 534 } 535 if (isBootRecord(iso9660, p)) 536 continue; 537 if (isEVD(iso9660, p)) 538 continue; 539 if (isSVD(iso9660, p)) 540 continue; 541 if (isVolumePartition(iso9660, p)) 542 continue; 543 if (isVDSetTerminator(iso9660, p)) { 544 seenTerminator = 1; 545 break; 546 } 547 return (0); 548 } 549 /* 550 * ISO 9660 format must have Primary Volume Descriptor and 551 * Volume Descriptor Set Terminator. 552 */ 553 if (seenTerminator && iso9660->primary.location > 16) 554 return (48); 555 556 /* We didn't find a valid PVD; return a bid of zero. */ 557 return (0); 558 } 559 560 static int 561 archive_read_format_iso9660_options(struct archive_read *a, 562 const char *key, const char *val) 563 { 564 struct iso9660 *iso9660; 565 566 iso9660 = (struct iso9660 *)(a->format->data); 567 568 if (strcmp(key, "joliet") == 0) { 569 if (val == NULL || strcmp(val, "off") == 0 || 570 strcmp(val, "ignore") == 0 || 571 strcmp(val, "disable") == 0 || 572 strcmp(val, "0") == 0) 573 iso9660->opt_support_joliet = 0; 574 else 575 iso9660->opt_support_joliet = 1; 576 return (ARCHIVE_OK); 577 } 578 if (strcmp(key, "rockridge") == 0 || 579 strcmp(key, "Rockridge") == 0) { 580 iso9660->opt_support_rockridge = val != NULL; 581 return (ARCHIVE_OK); 582 } 583 584 /* Note: The "warn" return is just to inform the options 585 * supervisor that we didn't handle it. It will generate 586 * a suitable error if no one used this option. */ 587 return (ARCHIVE_WARN); 588 } 589 590 static int 591 isBootRecord(struct iso9660 *iso9660, const unsigned char *h) 592 { 593 (void)iso9660; /* UNUSED */ 594 595 /* Type of the Volume Descriptor Boot Record must be 0. */ 596 if (h[0] != 0) 597 return (0); 598 599 /* Volume Descriptor Version must be 1. */ 600 if (h[6] != 1) 601 return (0); 602 603 return (1); 604 } 605 606 static int 607 isVolumePartition(struct iso9660 *iso9660, const unsigned char *h) 608 { 609 int32_t location; 610 611 /* Type of the Volume Partition Descriptor must be 3. */ 612 if (h[0] != 3) 613 return (0); 614 615 /* Volume Descriptor Version must be 1. */ 616 if (h[6] != 1) 617 return (0); 618 /* Unused Field */ 619 if (h[7] != 0) 620 return (0); 621 622 location = archive_le32dec(h + 72); 623 if (location <= SYSTEM_AREA_BLOCK || 624 location >= iso9660->volume_block) 625 return (0); 626 if ((uint32_t)location != archive_be32dec(h + 76)) 627 return (0); 628 629 return (1); 630 } 631 632 static int 633 isVDSetTerminator(struct iso9660 *iso9660, const unsigned char *h) 634 { 635 int i; 636 637 (void)iso9660; /* UNUSED */ 638 639 /* Type of the Volume Descriptor Set Terminator must be 255. */ 640 if (h[0] != 255) 641 return (0); 642 643 /* Volume Descriptor Version must be 1. */ 644 if (h[6] != 1) 645 return (0); 646 647 /* Reserved field must be 0. */ 648 for (i = 7; i < 2048; ++i) 649 if (h[i] != 0) 650 return (0); 651 652 return (1); 653 } 654 655 static int 656 isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) 657 { 658 const unsigned char *p; 659 ssize_t logical_block_size; 660 int32_t volume_block; 661 662 /* Check if current sector is a kind of Supplementary Volume 663 * Descriptor. */ 664 if (!isSVD(iso9660, h)) 665 return (0); 666 667 /* FIXME: do more validations according to joliet spec. */ 668 669 /* check if this SVD contains joliet extension! */ 670 p = h + SVD_escape_sequences_offset; 671 /* N.B. Joliet spec says p[1] == '\\', but.... */ 672 if (p[0] == '%' && p[1] == '/') { 673 int level = 0; 674 675 if (p[2] == '@') 676 level = 1; 677 else if (p[2] == 'C') 678 level = 2; 679 else if (p[2] == 'E') 680 level = 3; 681 else /* not joliet */ 682 return (0); 683 684 iso9660->seenJoliet = level; 685 686 } else /* not joliet */ 687 return (0); 688 689 logical_block_size = 690 archive_le16dec(h + SVD_logical_block_size_offset); 691 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 692 693 iso9660->logical_block_size = logical_block_size; 694 iso9660->volume_block = volume_block; 695 iso9660->volume_size = logical_block_size * (uint64_t)volume_block; 696 /* Read Root Directory Record in Volume Descriptor. */ 697 p = h + SVD_root_directory_record_offset; 698 iso9660->joliet.location = archive_le32dec(p + DR_extent_offset); 699 iso9660->joliet.size = archive_le32dec(p + DR_size_offset); 700 701 return (48); 702 } 703 704 static int 705 isSVD(struct iso9660 *iso9660, const unsigned char *h) 706 { 707 const unsigned char *p; 708 ssize_t logical_block_size; 709 int32_t volume_block; 710 int32_t location; 711 int i; 712 713 (void)iso9660; /* UNUSED */ 714 715 /* Type 2 means it's a SVD. */ 716 if (h[SVD_type_offset] != 2) 717 return (0); 718 719 /* Reserved field must be 0. */ 720 for (i = 0; i < SVD_reserved1_size; ++i) 721 if (h[SVD_reserved1_offset + i] != 0) 722 return (0); 723 for (i = 0; i < SVD_reserved2_size; ++i) 724 if (h[SVD_reserved2_offset + i] != 0) 725 return (0); 726 for (i = 0; i < SVD_reserved3_size; ++i) 727 if (h[SVD_reserved3_offset + i] != 0) 728 return (0); 729 730 /* File structure version must be 1 for ISO9660/ECMA119. */ 731 if (h[SVD_file_structure_version_offset] != 1) 732 return (0); 733 734 logical_block_size = 735 archive_le16dec(h + SVD_logical_block_size_offset); 736 if (logical_block_size <= 0) 737 return (0); 738 739 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 740 if (volume_block <= SYSTEM_AREA_BLOCK+4) 741 return (0); 742 743 /* Location of Occurrence of Type L Path Table must be 744 * available location, 745 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 746 location = archive_le32dec(h+SVD_type_L_path_table_offset); 747 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 748 return (0); 749 750 /* The Type M Path Table must be at a valid location (WinISO 751 * and probably other programs omit this, so we allow zero) 752 * 753 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 754 location = archive_be32dec(h+SVD_type_M_path_table_offset); 755 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 756 || location >= volume_block) 757 return (0); 758 759 /* Read Root Directory Record in Volume Descriptor. */ 760 p = h + SVD_root_directory_record_offset; 761 if (p[DR_length_offset] != 34) 762 return (0); 763 764 return (48); 765 } 766 767 static int 768 isEVD(struct iso9660 *iso9660, const unsigned char *h) 769 { 770 const unsigned char *p; 771 ssize_t logical_block_size; 772 int32_t volume_block; 773 int32_t location; 774 int i; 775 776 (void)iso9660; /* UNUSED */ 777 778 /* Type of the Enhanced Volume Descriptor must be 2. */ 779 if (h[PVD_type_offset] != 2) 780 return (0); 781 782 /* EVD version must be 2. */ 783 if (h[PVD_version_offset] != 2) 784 return (0); 785 786 /* Reserved field must be 0. */ 787 if (h[PVD_reserved1_offset] != 0) 788 return (0); 789 790 /* Reserved field must be 0. */ 791 for (i = 0; i < PVD_reserved2_size; ++i) 792 if (h[PVD_reserved2_offset + i] != 0) 793 return (0); 794 795 /* Reserved field must be 0. */ 796 for (i = 0; i < PVD_reserved3_size; ++i) 797 if (h[PVD_reserved3_offset + i] != 0) 798 return (0); 799 800 /* Logical block size must be > 0. */ 801 /* I've looked at Ecma 119 and can't find any stronger 802 * restriction on this field. */ 803 logical_block_size = 804 archive_le16dec(h + PVD_logical_block_size_offset); 805 if (logical_block_size <= 0) 806 return (0); 807 808 volume_block = 809 archive_le32dec(h + PVD_volume_space_size_offset); 810 if (volume_block <= SYSTEM_AREA_BLOCK+4) 811 return (0); 812 813 /* File structure version must be 2 for ISO9660:1999. */ 814 if (h[PVD_file_structure_version_offset] != 2) 815 return (0); 816 817 /* Location of Occurrence of Type L Path Table must be 818 * available location, 819 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 820 location = archive_le32dec(h+PVD_type_1_path_table_offset); 821 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 822 return (0); 823 824 /* Location of Occurrence of Type M Path Table must be 825 * available location, 826 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 827 location = archive_be32dec(h+PVD_type_m_path_table_offset); 828 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 829 || location >= volume_block) 830 return (0); 831 832 /* Reserved field must be 0. */ 833 for (i = 0; i < PVD_reserved4_size; ++i) 834 if (h[PVD_reserved4_offset + i] != 0) 835 return (0); 836 837 /* Reserved field must be 0. */ 838 for (i = 0; i < PVD_reserved5_size; ++i) 839 if (h[PVD_reserved5_offset + i] != 0) 840 return (0); 841 842 /* Read Root Directory Record in Volume Descriptor. */ 843 p = h + PVD_root_directory_record_offset; 844 if (p[DR_length_offset] != 34) 845 return (0); 846 847 return (48); 848 } 849 850 static int 851 isPVD(struct iso9660 *iso9660, const unsigned char *h) 852 { 853 const unsigned char *p; 854 ssize_t logical_block_size; 855 int32_t volume_block; 856 int32_t location; 857 int i; 858 859 /* Type of the Primary Volume Descriptor must be 1. */ 860 if (h[PVD_type_offset] != 1) 861 return (0); 862 863 /* PVD version must be 1. */ 864 if (h[PVD_version_offset] != 1) 865 return (0); 866 867 /* Reserved field must be 0. */ 868 if (h[PVD_reserved1_offset] != 0) 869 return (0); 870 871 /* Reserved field must be 0. */ 872 for (i = 0; i < PVD_reserved2_size; ++i) 873 if (h[PVD_reserved2_offset + i] != 0) 874 return (0); 875 876 /* Reserved field must be 0. */ 877 for (i = 0; i < PVD_reserved3_size; ++i) 878 if (h[PVD_reserved3_offset + i] != 0) 879 return (0); 880 881 /* Logical block size must be > 0. */ 882 /* I've looked at Ecma 119 and can't find any stronger 883 * restriction on this field. */ 884 logical_block_size = 885 archive_le16dec(h + PVD_logical_block_size_offset); 886 if (logical_block_size <= 0) 887 return (0); 888 889 volume_block = archive_le32dec(h + PVD_volume_space_size_offset); 890 if (volume_block <= SYSTEM_AREA_BLOCK+4) 891 return (0); 892 893 /* File structure version must be 1 for ISO9660/ECMA119. */ 894 if (h[PVD_file_structure_version_offset] != 1) 895 return (0); 896 897 /* Location of Occurrence of Type L Path Table must be 898 * available location, 899 * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 900 location = archive_le32dec(h+PVD_type_1_path_table_offset); 901 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 902 return (0); 903 904 /* The Type M Path Table must also be at a valid location 905 * (although ECMA 119 requires a Type M Path Table, WinISO and 906 * probably other programs omit it, so we permit a zero here) 907 * 908 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 909 location = archive_be32dec(h+PVD_type_m_path_table_offset); 910 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 911 || location >= volume_block) 912 return (0); 913 914 /* Reserved field must be 0. */ 915 /* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */ 916 for (i = 0; i < PVD_reserved4_size; ++i) 917 if (h[PVD_reserved4_offset + i] != 0 918 && h[PVD_reserved4_offset + i] != 0x20) 919 return (0); 920 921 /* Reserved field must be 0. */ 922 for (i = 0; i < PVD_reserved5_size; ++i) 923 if (h[PVD_reserved5_offset + i] != 0) 924 return (0); 925 926 /* XXX TODO: Check other values for sanity; reject more 927 * malformed PVDs. XXX */ 928 929 /* Read Root Directory Record in Volume Descriptor. */ 930 p = h + PVD_root_directory_record_offset; 931 if (p[DR_length_offset] != 34) 932 return (0); 933 934 if (!iso9660->primary.location) { 935 iso9660->logical_block_size = logical_block_size; 936 iso9660->volume_block = volume_block; 937 iso9660->volume_size = logical_block_size * (uint64_t)volume_block; 938 iso9660->primary.location = archive_le32dec(p + DR_extent_offset); 939 iso9660->primary.size = archive_le32dec(p + DR_size_offset); 940 } 941 942 return (48); 943 } 944 945 static int 946 read_children(struct archive_read *a, struct file_info *parent) 947 { 948 struct iso9660 *iso9660; 949 const unsigned char *b, *p; 950 struct file_info *multi; 951 size_t step, skip_size; 952 953 iso9660 = (struct iso9660 *)(a->format->data); 954 if (iso9660->current_position > parent->offset) { 955 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 956 "Ignoring out-of-order directory (%s) %jd > %jd", 957 parent->name.s, 958 (intmax_t)iso9660->current_position, 959 (intmax_t)parent->offset); 960 return (ARCHIVE_WARN); 961 } 962 if (parent->offset + parent->size > iso9660->volume_size) { 963 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 964 "Directory is beyond end-of-media: %s", 965 parent->name.s); 966 return (ARCHIVE_WARN); 967 } 968 if (iso9660->current_position < parent->offset) { 969 int64_t skipsize; 970 971 skipsize = parent->offset - iso9660->current_position; 972 skipsize = __archive_read_consume(a, skipsize); 973 if (skipsize < 0) 974 return ((int)skipsize); 975 iso9660->current_position = parent->offset; 976 } 977 978 step = ((parent->size + iso9660->logical_block_size -1) / 979 iso9660->logical_block_size) * iso9660->logical_block_size; 980 b = __archive_read_ahead(a, step, NULL); 981 if (b == NULL) { 982 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 983 "Failed to read full block when scanning " 984 "ISO9660 directory list"); 985 return (ARCHIVE_FATAL); 986 } 987 iso9660->current_position += step; 988 multi = NULL; 989 skip_size = step; 990 while (step) { 991 p = b; 992 b += iso9660->logical_block_size; 993 step -= iso9660->logical_block_size; 994 for (; *p != 0 && p < b && p + *p <= b; p += *p) { 995 struct file_info *child; 996 997 /* N.B.: these special directory identifiers 998 * are 8 bit "values" even on a 999 * Joliet CD with UCS-2 (16bit) encoding. 1000 */ 1001 1002 /* Skip '.' entry. */ 1003 if (*(p + DR_name_len_offset) == 1 1004 && *(p + DR_name_offset) == '\0') 1005 continue; 1006 /* Skip '..' entry. */ 1007 if (*(p + DR_name_len_offset) == 1 1008 && *(p + DR_name_offset) == '\001') 1009 continue; 1010 child = parse_file_info(a, parent, p); 1011 if (child == NULL) { 1012 __archive_read_consume(a, skip_size); 1013 return (ARCHIVE_FATAL); 1014 } 1015 if (child->cl_offset == 0 && 1016 (child->multi_extent || multi != NULL)) { 1017 struct content *con; 1018 1019 if (multi == NULL) { 1020 multi = child; 1021 multi->contents.first = NULL; 1022 multi->contents.last = 1023 &(multi->contents.first); 1024 } 1025 con = malloc(sizeof(struct content)); 1026 if (con == NULL) { 1027 archive_set_error( 1028 &a->archive, ENOMEM, 1029 "No memory for multi extent"); 1030 __archive_read_consume(a, skip_size); 1031 return (ARCHIVE_FATAL); 1032 } 1033 con->offset = child->offset; 1034 con->size = child->size; 1035 con->next = NULL; 1036 *multi->contents.last = con; 1037 multi->contents.last = &(con->next); 1038 if (multi == child) { 1039 if (add_entry(a, iso9660, child) 1040 != ARCHIVE_OK) 1041 return (ARCHIVE_FATAL); 1042 } else { 1043 multi->size += child->size; 1044 if (!child->multi_extent) 1045 multi = NULL; 1046 } 1047 } else 1048 if (add_entry(a, iso9660, child) != ARCHIVE_OK) 1049 return (ARCHIVE_FATAL); 1050 } 1051 } 1052 1053 __archive_read_consume(a, skip_size); 1054 1055 /* Read data which recorded by RRIP "CE" extension. */ 1056 if (read_CE(a, iso9660) != ARCHIVE_OK) 1057 return (ARCHIVE_FATAL); 1058 1059 return (ARCHIVE_OK); 1060 } 1061 1062 static int 1063 archive_read_format_iso9660_read_header(struct archive_read *a, 1064 struct archive_entry *entry) 1065 { 1066 struct iso9660 *iso9660; 1067 struct file_info *file; 1068 int r, rd_r = ARCHIVE_OK; 1069 1070 iso9660 = (struct iso9660 *)(a->format->data); 1071 1072 if (!a->archive.archive_format) { 1073 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; 1074 a->archive.archive_format_name = "ISO9660"; 1075 } 1076 1077 if (iso9660->current_position == 0) { 1078 int64_t skipsize; 1079 struct vd *vd; 1080 const void *block; 1081 char seenJoliet; 1082 1083 vd = &(iso9660->primary); 1084 if (!iso9660->opt_support_joliet) 1085 iso9660->seenJoliet = 0; 1086 if (iso9660->seenJoliet && 1087 vd->location > iso9660->joliet.location) 1088 /* This condition is unlikely; by way of caution. */ 1089 vd = &(iso9660->joliet); 1090 1091 skipsize = LOGICAL_BLOCK_SIZE * vd->location; 1092 skipsize = __archive_read_consume(a, skipsize); 1093 if (skipsize < 0) 1094 return ((int)skipsize); 1095 iso9660->current_position = skipsize; 1096 1097 block = __archive_read_ahead(a, vd->size, NULL); 1098 if (block == NULL) { 1099 archive_set_error(&a->archive, 1100 ARCHIVE_ERRNO_MISC, 1101 "Failed to read full block when scanning " 1102 "ISO9660 directory list"); 1103 return (ARCHIVE_FATAL); 1104 } 1105 1106 /* 1107 * While reading Root Directory, flag seenJoliet 1108 * must be zero to avoid converting special name 1109 * 0x00(Current Directory) and next byte to UCS2. 1110 */ 1111 seenJoliet = iso9660->seenJoliet;/* Save flag. */ 1112 iso9660->seenJoliet = 0; 1113 file = parse_file_info(a, NULL, block); 1114 if (file == NULL) 1115 return (ARCHIVE_FATAL); 1116 iso9660->seenJoliet = seenJoliet; 1117 if (vd == &(iso9660->primary) && iso9660->seenRockridge 1118 && iso9660->seenJoliet) 1119 /* 1120 * If iso image has RockRidge and Joliet, 1121 * we use RockRidge Extensions. 1122 */ 1123 iso9660->seenJoliet = 0; 1124 if (vd == &(iso9660->primary) && !iso9660->seenRockridge 1125 && iso9660->seenJoliet) { 1126 /* Switch reading data from primary to joliet. */ 1127 vd = &(iso9660->joliet); 1128 skipsize = LOGICAL_BLOCK_SIZE * vd->location; 1129 skipsize -= iso9660->current_position; 1130 skipsize = __archive_read_consume(a, skipsize); 1131 if (skipsize < 0) 1132 return ((int)skipsize); 1133 iso9660->current_position += skipsize; 1134 1135 block = __archive_read_ahead(a, vd->size, NULL); 1136 if (block == NULL) { 1137 archive_set_error(&a->archive, 1138 ARCHIVE_ERRNO_MISC, 1139 "Failed to read full block when scanning " 1140 "ISO9660 directory list"); 1141 return (ARCHIVE_FATAL); 1142 } 1143 iso9660->seenJoliet = 0; 1144 file = parse_file_info(a, NULL, block); 1145 if (file == NULL) 1146 return (ARCHIVE_FATAL); 1147 iso9660->seenJoliet = seenJoliet; 1148 } 1149 /* Store the root directory in the pending list. */ 1150 if (add_entry(a, iso9660, file) != ARCHIVE_OK) 1151 return (ARCHIVE_FATAL); 1152 if (iso9660->seenRockridge) { 1153 a->archive.archive_format = 1154 ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; 1155 a->archive.archive_format_name = 1156 "ISO9660 with Rockridge extensions"; 1157 } 1158 } 1159 1160 file = NULL;/* Eliminate a warning. */ 1161 /* Get the next entry that appears after the current offset. */ 1162 r = next_entry_seek(a, iso9660, &file); 1163 if (r != ARCHIVE_OK) 1164 return (r); 1165 1166 if (iso9660->seenJoliet) { 1167 /* 1168 * Convert UTF-16BE of a filename to local locale MBS 1169 * and store the result into a filename field. 1170 */ 1171 if (iso9660->sconv_utf16be == NULL) { 1172 iso9660->sconv_utf16be = 1173 archive_string_conversion_from_charset( 1174 &(a->archive), "UTF-16BE", 1); 1175 if (iso9660->sconv_utf16be == NULL) 1176 /* Coundn't allocate memory */ 1177 return (ARCHIVE_FATAL); 1178 } 1179 if (iso9660->utf16be_path == NULL) { 1180 iso9660->utf16be_path = malloc(UTF16_NAME_MAX); 1181 if (iso9660->utf16be_path == NULL) { 1182 archive_set_error(&a->archive, ENOMEM, 1183 "No memory"); 1184 return (ARCHIVE_FATAL); 1185 } 1186 } 1187 if (iso9660->utf16be_previous_path == NULL) { 1188 iso9660->utf16be_previous_path = malloc(UTF16_NAME_MAX); 1189 if (iso9660->utf16be_previous_path == NULL) { 1190 archive_set_error(&a->archive, ENOMEM, 1191 "No memory"); 1192 return (ARCHIVE_FATAL); 1193 } 1194 } 1195 1196 iso9660->utf16be_path_len = 0; 1197 if (build_pathname_utf16be(iso9660->utf16be_path, 1198 UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) { 1199 archive_set_error(&a->archive, 1200 ARCHIVE_ERRNO_FILE_FORMAT, 1201 "Pathname is too long"); 1202 } 1203 1204 r = archive_entry_copy_pathname_l(entry, 1205 (const char *)iso9660->utf16be_path, 1206 iso9660->utf16be_path_len, 1207 iso9660->sconv_utf16be); 1208 if (r != 0) { 1209 if (errno == ENOMEM) { 1210 archive_set_error(&a->archive, ENOMEM, 1211 "No memory for Pathname"); 1212 return (ARCHIVE_FATAL); 1213 } 1214 archive_set_error(&a->archive, 1215 ARCHIVE_ERRNO_FILE_FORMAT, 1216 "Pathname cannot be converted " 1217 "from %s to current locale.", 1218 archive_string_conversion_charset_name( 1219 iso9660->sconv_utf16be)); 1220 1221 rd_r = ARCHIVE_WARN; 1222 } 1223 } else { 1224 archive_string_empty(&iso9660->pathname); 1225 archive_entry_set_pathname(entry, 1226 build_pathname(&iso9660->pathname, file)); 1227 } 1228 1229 iso9660->entry_bytes_remaining = file->size; 1230 iso9660->entry_sparse_offset = 0; /* Offset for sparse-file-aware clients. */ 1231 1232 if (file->offset + file->size > iso9660->volume_size) { 1233 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1234 "File is beyond end-of-media: %s", 1235 archive_entry_pathname(entry)); 1236 iso9660->entry_bytes_remaining = 0; 1237 iso9660->entry_sparse_offset = 0; 1238 return (ARCHIVE_WARN); 1239 } 1240 1241 /* Set up the entry structure with information about this entry. */ 1242 archive_entry_set_mode(entry, file->mode); 1243 archive_entry_set_uid(entry, file->uid); 1244 archive_entry_set_gid(entry, file->gid); 1245 archive_entry_set_nlink(entry, file->nlinks); 1246 if (file->birthtime_is_set) 1247 archive_entry_set_birthtime(entry, file->birthtime, 0); 1248 else 1249 archive_entry_unset_birthtime(entry); 1250 archive_entry_set_mtime(entry, file->mtime, 0); 1251 archive_entry_set_ctime(entry, file->ctime, 0); 1252 archive_entry_set_atime(entry, file->atime, 0); 1253 /* N.B.: Rock Ridge supports 64-bit device numbers. */ 1254 archive_entry_set_rdev(entry, (dev_t)file->rdev); 1255 archive_entry_set_size(entry, iso9660->entry_bytes_remaining); 1256 if (file->symlink.s != NULL) 1257 archive_entry_copy_symlink(entry, file->symlink.s); 1258 1259 /* Note: If the input isn't seekable, we can't rewind to 1260 * return the same body again, so if the next entry refers to 1261 * the same data, we have to return it as a hardlink to the 1262 * original entry. */ 1263 if (file->number != -1 && 1264 file->number == iso9660->previous_number) { 1265 if (iso9660->seenJoliet) { 1266 r = archive_entry_copy_hardlink_l(entry, 1267 (const char *)iso9660->utf16be_previous_path, 1268 iso9660->utf16be_previous_path_len, 1269 iso9660->sconv_utf16be); 1270 if (r != 0) { 1271 if (errno == ENOMEM) { 1272 archive_set_error(&a->archive, ENOMEM, 1273 "No memory for Linkname"); 1274 return (ARCHIVE_FATAL); 1275 } 1276 archive_set_error(&a->archive, 1277 ARCHIVE_ERRNO_FILE_FORMAT, 1278 "Linkname cannot be converted " 1279 "from %s to current locale.", 1280 archive_string_conversion_charset_name( 1281 iso9660->sconv_utf16be)); 1282 rd_r = ARCHIVE_WARN; 1283 } 1284 } else 1285 archive_entry_set_hardlink(entry, 1286 iso9660->previous_pathname.s); 1287 archive_entry_unset_size(entry); 1288 iso9660->entry_bytes_remaining = 0; 1289 iso9660->entry_sparse_offset = 0; 1290 return (rd_r); 1291 } 1292 1293 /* Except for the hardlink case above, if the offset of the 1294 * next entry is before our current position, we can't seek 1295 * backwards to extract it, so issue a warning. Note that 1296 * this can only happen if this entry was added to the heap 1297 * after we passed this offset, that is, only if the directory 1298 * mentioning this entry is later than the body of the entry. 1299 * Such layouts are very unusual; most ISO9660 writers lay out 1300 * and record all directory information first, then store 1301 * all file bodies. */ 1302 /* TODO: Someday, libarchive's I/O core will support optional 1303 * seeking. When that day comes, this code should attempt to 1304 * seek and only return the error if the seek fails. That 1305 * will give us support for whacky ISO images that require 1306 * seeking while retaining the ability to read almost all ISO 1307 * images in a streaming fashion. */ 1308 if ((file->mode & AE_IFMT) != AE_IFDIR && 1309 file->offset < iso9660->current_position) { 1310 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1311 "Ignoring out-of-order file @%jx (%s) %jd < %jd", 1312 (intmax_t)file->number, 1313 iso9660->pathname.s, 1314 (intmax_t)file->offset, 1315 (intmax_t)iso9660->current_position); 1316 iso9660->entry_bytes_remaining = 0; 1317 iso9660->entry_sparse_offset = 0; 1318 return (ARCHIVE_WARN); 1319 } 1320 1321 /* Initialize zisofs variables. */ 1322 iso9660->entry_zisofs.pz = file->pz; 1323 if (file->pz) { 1324 #ifdef HAVE_ZLIB_H 1325 struct zisofs *zisofs; 1326 1327 zisofs = &iso9660->entry_zisofs; 1328 zisofs->initialized = 0; 1329 zisofs->pz_log2_bs = file->pz_log2_bs; 1330 zisofs->pz_uncompressed_size = file->pz_uncompressed_size; 1331 zisofs->pz_offset = 0; 1332 zisofs->header_avail = 0; 1333 zisofs->header_passed = 0; 1334 zisofs->block_pointers_avail = 0; 1335 #endif 1336 archive_entry_set_size(entry, file->pz_uncompressed_size); 1337 } 1338 1339 iso9660->previous_number = file->number; 1340 if (iso9660->seenJoliet) { 1341 memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path, 1342 iso9660->utf16be_path_len); 1343 iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len; 1344 } else 1345 archive_strcpy( 1346 &iso9660->previous_pathname, iso9660->pathname.s); 1347 1348 /* Reset entry_bytes_remaining if the file is multi extent. */ 1349 iso9660->entry_content = file->contents.first; 1350 if (iso9660->entry_content != NULL) 1351 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1352 1353 if (archive_entry_filetype(entry) == AE_IFDIR) { 1354 /* Overwrite nlinks by proper link number which is 1355 * calculated from number of sub directories. */ 1356 archive_entry_set_nlink(entry, 2 + file->subdirs); 1357 /* Directory data has been read completely. */ 1358 iso9660->entry_bytes_remaining = 0; 1359 iso9660->entry_sparse_offset = 0; 1360 } 1361 1362 if (rd_r != ARCHIVE_OK) 1363 return (rd_r); 1364 return (ARCHIVE_OK); 1365 } 1366 1367 static int 1368 archive_read_format_iso9660_read_data_skip(struct archive_read *a) 1369 { 1370 /* Because read_next_header always does an explicit skip 1371 * to the next entry, we don't need to do anything here. */ 1372 (void)a; /* UNUSED */ 1373 return (ARCHIVE_OK); 1374 } 1375 1376 #ifdef HAVE_ZLIB_H 1377 1378 static int 1379 zisofs_read_data(struct archive_read *a, 1380 const void **buff, size_t *size, int64_t *offset) 1381 { 1382 struct iso9660 *iso9660; 1383 struct zisofs *zisofs; 1384 const unsigned char *p; 1385 size_t avail; 1386 ssize_t bytes_read; 1387 size_t uncompressed_size; 1388 int r; 1389 1390 iso9660 = (struct iso9660 *)(a->format->data); 1391 zisofs = &iso9660->entry_zisofs; 1392 1393 p = __archive_read_ahead(a, 1, &bytes_read); 1394 if (bytes_read <= 0) { 1395 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1396 "Truncated zisofs file body"); 1397 return (ARCHIVE_FATAL); 1398 } 1399 if (bytes_read > iso9660->entry_bytes_remaining) 1400 bytes_read = iso9660->entry_bytes_remaining; 1401 avail = bytes_read; 1402 uncompressed_size = 0; 1403 1404 if (!zisofs->initialized) { 1405 size_t ceil, xsize; 1406 1407 /* Allocate block pointers buffer. */ 1408 ceil = (zisofs->pz_uncompressed_size + 1409 (1LL << zisofs->pz_log2_bs) - 1) 1410 >> zisofs->pz_log2_bs; 1411 xsize = (ceil + 1) * 4; 1412 if (zisofs->block_pointers_alloc < xsize) { 1413 size_t alloc; 1414 1415 if (zisofs->block_pointers != NULL) 1416 free(zisofs->block_pointers); 1417 alloc = ((xsize >> 10) + 1) << 10; 1418 zisofs->block_pointers = malloc(alloc); 1419 if (zisofs->block_pointers == NULL) { 1420 archive_set_error(&a->archive, ENOMEM, 1421 "No memory for zisofs decompression"); 1422 return (ARCHIVE_FATAL); 1423 } 1424 zisofs->block_pointers_alloc = alloc; 1425 } 1426 zisofs->block_pointers_size = xsize; 1427 1428 /* Allocate uncompressed data buffer. */ 1429 xsize = 1UL << zisofs->pz_log2_bs; 1430 if (zisofs->uncompressed_buffer_size < xsize) { 1431 if (zisofs->uncompressed_buffer != NULL) 1432 free(zisofs->uncompressed_buffer); 1433 zisofs->uncompressed_buffer = malloc(xsize); 1434 if (zisofs->uncompressed_buffer == NULL) { 1435 archive_set_error(&a->archive, ENOMEM, 1436 "No memory for zisofs decompression"); 1437 return (ARCHIVE_FATAL); 1438 } 1439 } 1440 zisofs->uncompressed_buffer_size = xsize; 1441 1442 /* 1443 * Read the file header, and check the magic code of zisofs. 1444 */ 1445 if (zisofs->header_avail < sizeof(zisofs->header)) { 1446 xsize = sizeof(zisofs->header) - zisofs->header_avail; 1447 if (avail < xsize) 1448 xsize = avail; 1449 memcpy(zisofs->header + zisofs->header_avail, p, xsize); 1450 zisofs->header_avail += xsize; 1451 avail -= xsize; 1452 p += xsize; 1453 } 1454 if (!zisofs->header_passed && 1455 zisofs->header_avail == sizeof(zisofs->header)) { 1456 int err = 0; 1457 1458 if (memcmp(zisofs->header, zisofs_magic, 1459 sizeof(zisofs_magic)) != 0) 1460 err = 1; 1461 if (archive_le32dec(zisofs->header + 8) 1462 != zisofs->pz_uncompressed_size) 1463 err = 1; 1464 if (zisofs->header[12] != 4) 1465 err = 1; 1466 if (zisofs->header[13] != zisofs->pz_log2_bs) 1467 err = 1; 1468 if (err) { 1469 archive_set_error(&a->archive, 1470 ARCHIVE_ERRNO_FILE_FORMAT, 1471 "Illegal zisofs file body"); 1472 return (ARCHIVE_FATAL); 1473 } 1474 zisofs->header_passed = 1; 1475 } 1476 /* 1477 * Read block pointers. 1478 */ 1479 if (zisofs->header_passed && 1480 zisofs->block_pointers_avail < zisofs->block_pointers_size) { 1481 xsize = zisofs->block_pointers_size 1482 - zisofs->block_pointers_avail; 1483 if (avail < xsize) 1484 xsize = avail; 1485 memcpy(zisofs->block_pointers 1486 + zisofs->block_pointers_avail, p, xsize); 1487 zisofs->block_pointers_avail += xsize; 1488 avail -= xsize; 1489 p += xsize; 1490 if (zisofs->block_pointers_avail 1491 == zisofs->block_pointers_size) { 1492 /* We've got all block pointers and initialize 1493 * related variables. */ 1494 zisofs->block_off = 0; 1495 zisofs->block_avail = 0; 1496 /* Complete a initialization */ 1497 zisofs->initialized = 1; 1498 } 1499 } 1500 1501 if (!zisofs->initialized) 1502 goto next_data; /* We need more data. */ 1503 } 1504 1505 /* 1506 * Get block offsets from block pointers. 1507 */ 1508 if (zisofs->block_avail == 0) { 1509 uint32_t bst, bed; 1510 1511 if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { 1512 /* There isn't a pair of offsets. */ 1513 archive_set_error(&a->archive, 1514 ARCHIVE_ERRNO_FILE_FORMAT, 1515 "Illegal zisofs block pointers"); 1516 return (ARCHIVE_FATAL); 1517 } 1518 bst = archive_le32dec( 1519 zisofs->block_pointers + zisofs->block_off); 1520 if (bst != zisofs->pz_offset + (bytes_read - avail)) { 1521 /* TODO: Should we seek offset of current file 1522 * by bst ? */ 1523 archive_set_error(&a->archive, 1524 ARCHIVE_ERRNO_FILE_FORMAT, 1525 "Illegal zisofs block pointers(cannot seek)"); 1526 return (ARCHIVE_FATAL); 1527 } 1528 bed = archive_le32dec( 1529 zisofs->block_pointers + zisofs->block_off + 4); 1530 if (bed < bst) { 1531 archive_set_error(&a->archive, 1532 ARCHIVE_ERRNO_FILE_FORMAT, 1533 "Illegal zisofs block pointers"); 1534 return (ARCHIVE_FATAL); 1535 } 1536 zisofs->block_avail = bed - bst; 1537 zisofs->block_off += 4; 1538 1539 /* Initialize compression library for new block. */ 1540 if (zisofs->stream_valid) 1541 r = inflateReset(&zisofs->stream); 1542 else 1543 r = inflateInit(&zisofs->stream); 1544 if (r != Z_OK) { 1545 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1546 "Can't initialize zisofs decompression."); 1547 return (ARCHIVE_FATAL); 1548 } 1549 zisofs->stream_valid = 1; 1550 zisofs->stream.total_in = 0; 1551 zisofs->stream.total_out = 0; 1552 } 1553 1554 /* 1555 * Make uncompressed data. 1556 */ 1557 if (zisofs->block_avail == 0) { 1558 memset(zisofs->uncompressed_buffer, 0, 1559 zisofs->uncompressed_buffer_size); 1560 uncompressed_size = zisofs->uncompressed_buffer_size; 1561 } else { 1562 zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; 1563 if (avail > zisofs->block_avail) 1564 zisofs->stream.avail_in = zisofs->block_avail; 1565 else 1566 zisofs->stream.avail_in = avail; 1567 zisofs->stream.next_out = zisofs->uncompressed_buffer; 1568 zisofs->stream.avail_out = zisofs->uncompressed_buffer_size; 1569 1570 r = inflate(&zisofs->stream, 0); 1571 switch (r) { 1572 case Z_OK: /* Decompressor made some progress.*/ 1573 case Z_STREAM_END: /* Found end of stream. */ 1574 break; 1575 default: 1576 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1577 "zisofs decompression failed (%d)", r); 1578 return (ARCHIVE_FATAL); 1579 } 1580 uncompressed_size = 1581 zisofs->uncompressed_buffer_size - zisofs->stream.avail_out; 1582 avail -= zisofs->stream.next_in - p; 1583 zisofs->block_avail -= zisofs->stream.next_in - p; 1584 } 1585 next_data: 1586 bytes_read -= avail; 1587 *buff = zisofs->uncompressed_buffer; 1588 *size = uncompressed_size; 1589 *offset = iso9660->entry_sparse_offset; 1590 iso9660->entry_sparse_offset += uncompressed_size; 1591 iso9660->entry_bytes_remaining -= bytes_read; 1592 iso9660->current_position += bytes_read; 1593 zisofs->pz_offset += bytes_read; 1594 iso9660->entry_bytes_unconsumed += bytes_read; 1595 1596 return (ARCHIVE_OK); 1597 } 1598 1599 #else /* HAVE_ZLIB_H */ 1600 1601 static int 1602 zisofs_read_data(struct archive_read *a, 1603 const void **buff, size_t *size, int64_t *offset) 1604 { 1605 1606 (void)buff;/* UNUSED */ 1607 (void)size;/* UNUSED */ 1608 (void)offset;/* UNUSED */ 1609 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1610 "zisofs is not supported on this platform."); 1611 return (ARCHIVE_FAILED); 1612 } 1613 1614 #endif /* HAVE_ZLIB_H */ 1615 1616 static int 1617 archive_read_format_iso9660_read_data(struct archive_read *a, 1618 const void **buff, size_t *size, int64_t *offset) 1619 { 1620 ssize_t bytes_read; 1621 struct iso9660 *iso9660; 1622 1623 iso9660 = (struct iso9660 *)(a->format->data); 1624 1625 if (iso9660->entry_bytes_unconsumed) { 1626 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 1627 iso9660->entry_bytes_unconsumed = 0; 1628 } 1629 1630 if (iso9660->entry_bytes_remaining <= 0) { 1631 if (iso9660->entry_content != NULL) 1632 iso9660->entry_content = iso9660->entry_content->next; 1633 if (iso9660->entry_content == NULL) { 1634 *buff = NULL; 1635 *size = 0; 1636 *offset = iso9660->entry_sparse_offset; 1637 return (ARCHIVE_EOF); 1638 } 1639 /* Seek forward to the start of the entry. */ 1640 if (iso9660->current_position < iso9660->entry_content->offset) { 1641 int64_t step; 1642 1643 step = iso9660->entry_content->offset - 1644 iso9660->current_position; 1645 step = __archive_read_consume(a, step); 1646 if (step < 0) 1647 return ((int)step); 1648 iso9660->current_position = 1649 iso9660->entry_content->offset; 1650 } 1651 if (iso9660->entry_content->offset < iso9660->current_position) { 1652 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1653 "Ignoring out-of-order file (%s) %jd < %jd", 1654 iso9660->pathname.s, 1655 (intmax_t)iso9660->entry_content->offset, 1656 (intmax_t)iso9660->current_position); 1657 *buff = NULL; 1658 *size = 0; 1659 *offset = iso9660->entry_sparse_offset; 1660 return (ARCHIVE_WARN); 1661 } 1662 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1663 } 1664 if (iso9660->entry_zisofs.pz) 1665 return (zisofs_read_data(a, buff, size, offset)); 1666 1667 *buff = __archive_read_ahead(a, 1, &bytes_read); 1668 if (bytes_read == 0) 1669 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1670 "Truncated input file"); 1671 if (*buff == NULL) 1672 return (ARCHIVE_FATAL); 1673 if (bytes_read > iso9660->entry_bytes_remaining) 1674 bytes_read = iso9660->entry_bytes_remaining; 1675 *size = bytes_read; 1676 *offset = iso9660->entry_sparse_offset; 1677 iso9660->entry_sparse_offset += bytes_read; 1678 iso9660->entry_bytes_remaining -= bytes_read; 1679 iso9660->entry_bytes_unconsumed = bytes_read; 1680 iso9660->current_position += bytes_read; 1681 return (ARCHIVE_OK); 1682 } 1683 1684 static int 1685 archive_read_format_iso9660_cleanup(struct archive_read *a) 1686 { 1687 struct iso9660 *iso9660; 1688 int r = ARCHIVE_OK; 1689 1690 iso9660 = (struct iso9660 *)(a->format->data); 1691 release_files(iso9660); 1692 free(iso9660->read_ce_req.reqs); 1693 archive_string_free(&iso9660->pathname); 1694 archive_string_free(&iso9660->previous_pathname); 1695 if (iso9660->pending_files.files) 1696 free(iso9660->pending_files.files); 1697 #ifdef HAVE_ZLIB_H 1698 free(iso9660->entry_zisofs.uncompressed_buffer); 1699 free(iso9660->entry_zisofs.block_pointers); 1700 if (iso9660->entry_zisofs.stream_valid) { 1701 if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) { 1702 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1703 "Failed to clean up zlib decompressor"); 1704 r = ARCHIVE_FATAL; 1705 } 1706 } 1707 #endif 1708 free(iso9660->utf16be_path); 1709 free(iso9660->utf16be_previous_path); 1710 free(iso9660); 1711 (a->format->data) = NULL; 1712 return (r); 1713 } 1714 1715 /* 1716 * This routine parses a single ISO directory record, makes sense 1717 * of any extensions, and stores the result in memory. 1718 */ 1719 static struct file_info * 1720 parse_file_info(struct archive_read *a, struct file_info *parent, 1721 const unsigned char *isodirrec) 1722 { 1723 struct iso9660 *iso9660; 1724 struct file_info *file; 1725 size_t name_len; 1726 const unsigned char *rr_start, *rr_end; 1727 const unsigned char *p; 1728 size_t dr_len; 1729 uint64_t fsize; 1730 int32_t location; 1731 int flags; 1732 1733 iso9660 = (struct iso9660 *)(a->format->data); 1734 1735 dr_len = (size_t)isodirrec[DR_length_offset]; 1736 name_len = (size_t)isodirrec[DR_name_len_offset]; 1737 location = archive_le32dec(isodirrec + DR_extent_offset); 1738 fsize = toi(isodirrec + DR_size_offset, DR_size_size); 1739 /* Sanity check that dr_len needs at least 34. */ 1740 if (dr_len < 34) { 1741 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1742 "Invalid length of directory record"); 1743 return (NULL); 1744 } 1745 /* Sanity check that name_len doesn't exceed dr_len. */ 1746 if (dr_len - 33 < name_len || name_len == 0) { 1747 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1748 "Invalid length of file identifier"); 1749 return (NULL); 1750 } 1751 /* Sanity check that location doesn't exceed volume block. 1752 * Don't check lower limit of location; it's possibility 1753 * the location has negative value when file type is symbolic 1754 * link or file size is zero. As far as I know latest mkisofs 1755 * do that. 1756 */ 1757 if (location > 0 && 1758 (location + ((fsize + iso9660->logical_block_size -1) 1759 / iso9660->logical_block_size)) > iso9660->volume_block) { 1760 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1761 "Invalid location of extent of file"); 1762 return (NULL); 1763 } 1764 /* Sanity check that location doesn't have a negative value 1765 * when the file is not empty. it's too large. */ 1766 if (fsize != 0 && location < 0) { 1767 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1768 "Invalid location of extent of file"); 1769 return (NULL); 1770 } 1771 1772 /* Create a new file entry and copy data from the ISO dir record. */ 1773 file = (struct file_info *)calloc(1, sizeof(*file)); 1774 if (file == NULL) { 1775 archive_set_error(&a->archive, ENOMEM, 1776 "No memory for file entry"); 1777 return (NULL); 1778 } 1779 file->parent = parent; 1780 file->offset = iso9660->logical_block_size * (uint64_t)location; 1781 file->size = fsize; 1782 file->mtime = isodate7(isodirrec + DR_date_offset); 1783 file->ctime = file->atime = file->mtime; 1784 file->rede_files.first = NULL; 1785 file->rede_files.last = &(file->rede_files.first); 1786 1787 p = isodirrec + DR_name_offset; 1788 /* Rockridge extensions (if any) follow name. Compute this 1789 * before fidgeting the name_len below. */ 1790 rr_start = p + name_len + (name_len & 1 ? 0 : 1); 1791 rr_end = isodirrec + dr_len; 1792 1793 if (iso9660->seenJoliet) { 1794 /* Joliet names are max 64 chars (128 bytes) according to spec, 1795 * but genisoimage/mkisofs allows recording longer Joliet 1796 * names which are 103 UCS2 characters(206 bytes) by their 1797 * option '-joliet-long'. 1798 */ 1799 if (name_len > 206) 1800 name_len = 206; 1801 name_len &= ~1; 1802 1803 /* trim trailing first version and dot from filename. 1804 * 1805 * Remember we were in UTF-16BE land! 1806 * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both 1807 * 16 bits big endian characters on Joliet. 1808 * 1809 * TODO: sanitize filename? 1810 * Joliet allows any UCS-2 char except: 1811 * *, /, :, ;, ? and \. 1812 */ 1813 /* Chop off trailing ';1' from files. */ 1814 if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';' 1815 && p[name_len-2] == 0 && p[name_len-1] == '1') 1816 name_len -= 4; 1817 #if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */ 1818 /* Chop off trailing '.' from filenames. */ 1819 if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.') 1820 name_len -= 2; 1821 #endif 1822 if ((file->utf16be_name = malloc(name_len)) == NULL) { 1823 archive_set_error(&a->archive, ENOMEM, 1824 "No memory for file name"); 1825 return (NULL); 1826 } 1827 memcpy(file->utf16be_name, p, name_len); 1828 file->utf16be_bytes = name_len; 1829 } else { 1830 /* Chop off trailing ';1' from files. */ 1831 if (name_len > 2 && p[name_len - 2] == ';' && 1832 p[name_len - 1] == '1') 1833 name_len -= 2; 1834 /* Chop off trailing '.' from filenames. */ 1835 if (name_len > 1 && p[name_len - 1] == '.') 1836 --name_len; 1837 1838 archive_strncpy(&file->name, (const char *)p, name_len); 1839 } 1840 1841 flags = isodirrec[DR_flags_offset]; 1842 if (flags & 0x02) 1843 file->mode = AE_IFDIR | 0700; 1844 else 1845 file->mode = AE_IFREG | 0400; 1846 if (flags & 0x80) 1847 file->multi_extent = 1; 1848 else 1849 file->multi_extent = 0; 1850 /* 1851 * Use a location for the file number, which is treated as an inode 1852 * number to find out hardlink target. If Rockridge extensions is 1853 * being used, the file number will be overwritten by FILE SERIAL 1854 * NUMBER of RRIP "PX" extension. 1855 * Note: Old mkisofs did not record that FILE SERIAL NUMBER 1856 * in ISO images. 1857 * Note2: xorriso set 0 to the location of a symlink file. 1858 */ 1859 if (file->size == 0 && location >= 0) { 1860 /* If file->size is zero, its location points wrong place, 1861 * and so we should not use it for the file number. 1862 * When the location has negative value, it can be used 1863 * for the file number. 1864 */ 1865 file->number = -1; 1866 /* Do not appear before any directory entries. */ 1867 file->offset = -1; 1868 } else 1869 file->number = (int64_t)(uint32_t)location; 1870 1871 /* Rockridge extensions overwrite information from above. */ 1872 if (iso9660->opt_support_rockridge) { 1873 if (parent == NULL && rr_end - rr_start >= 7) { 1874 p = rr_start; 1875 if (p[0] == 'S' && p[1] == 'P' 1876 && p[2] == 7 && p[3] == 1 1877 && p[4] == 0xBE && p[5] == 0xEF) { 1878 /* 1879 * SP extension stores the suspOffset 1880 * (Number of bytes to skip between 1881 * filename and SUSP records.) 1882 * It is mandatory by the SUSP standard 1883 * (IEEE 1281). 1884 * 1885 * It allows SUSP to coexist with 1886 * non-SUSP uses of the System 1887 * Use Area by placing non-SUSP data 1888 * before SUSP data. 1889 * 1890 * SP extension must be in the root 1891 * directory entry, disable all SUSP 1892 * processing if not found. 1893 */ 1894 iso9660->suspOffset = p[6]; 1895 iso9660->seenSUSP = 1; 1896 rr_start += 7; 1897 } 1898 } 1899 if (iso9660->seenSUSP) { 1900 int r; 1901 1902 file->name_continues = 0; 1903 file->symlink_continues = 0; 1904 rr_start += iso9660->suspOffset; 1905 r = parse_rockridge(a, file, rr_start, rr_end); 1906 if (r != ARCHIVE_OK) { 1907 free(file); 1908 return (NULL); 1909 } 1910 } else 1911 /* If there isn't SUSP, disable parsing 1912 * rock ridge extensions. */ 1913 iso9660->opt_support_rockridge = 0; 1914 } 1915 1916 file->nlinks = 1;/* Reset nlink. we'll calculate it later. */ 1917 /* Tell file's parent how many children that parent has. */ 1918 if (parent != NULL && (flags & 0x02)) 1919 parent->subdirs++; 1920 1921 if (iso9660->seenRockridge) { 1922 if (parent != NULL && parent->parent == NULL && 1923 (flags & 0x02) && iso9660->rr_moved == NULL && 1924 (strcmp(file->name.s, "rr_moved") == 0 || 1925 strcmp(file->name.s, ".rr_moved") == 0)) { 1926 iso9660->rr_moved = file; 1927 file->rr_moved = 1; 1928 file->rr_moved_has_re_only = 1; 1929 file->re = 0; 1930 parent->subdirs--; 1931 } else if (file->re) { 1932 /* 1933 * Sanity check: file's parent is rr_moved. 1934 */ 1935 if (parent == NULL || parent->rr_moved == 0) { 1936 archive_set_error(&a->archive, 1937 ARCHIVE_ERRNO_MISC, 1938 "Invalid Rockridge RE"); 1939 return (NULL); 1940 } 1941 /* 1942 * Sanity check: file does not have "CL" extension. 1943 */ 1944 if (file->cl_offset) { 1945 archive_set_error(&a->archive, 1946 ARCHIVE_ERRNO_MISC, 1947 "Invalid Rockridge RE and CL"); 1948 return (NULL); 1949 } 1950 /* 1951 * Sanity check: The file type must be a directory. 1952 */ 1953 if ((flags & 0x02) == 0) { 1954 archive_set_error(&a->archive, 1955 ARCHIVE_ERRNO_MISC, 1956 "Invalid Rockridge RE"); 1957 return (NULL); 1958 } 1959 } else if (parent != NULL && parent->rr_moved) 1960 file->rr_moved_has_re_only = 0; 1961 else if (parent != NULL && (flags & 0x02) && 1962 (parent->re || parent->re_descendant)) 1963 file->re_descendant = 1; 1964 if (file->cl_offset) { 1965 struct file_info *r; 1966 1967 if (parent == NULL || parent->parent == NULL) { 1968 archive_set_error(&a->archive, 1969 ARCHIVE_ERRNO_MISC, 1970 "Invalid Rockridge CL"); 1971 return (NULL); 1972 } 1973 /* 1974 * Sanity check: The file type must be a regular file. 1975 */ 1976 if ((flags & 0x02) != 0) { 1977 archive_set_error(&a->archive, 1978 ARCHIVE_ERRNO_MISC, 1979 "Invalid Rockridge CL"); 1980 return (NULL); 1981 } 1982 parent->subdirs++; 1983 /* Overwrite an offset and a number of this "CL" entry 1984 * to appear before other dirs. "+1" to those is to 1985 * make sure to appear after "RE" entry which this 1986 * "CL" entry should be connected with. */ 1987 file->offset = file->number = file->cl_offset + 1; 1988 1989 /* 1990 * Sanity check: cl_offset does not point at its 1991 * the parents or itself. 1992 */ 1993 for (r = parent; r; r = r->parent) { 1994 if (r->offset == file->cl_offset) { 1995 archive_set_error(&a->archive, 1996 ARCHIVE_ERRNO_MISC, 1997 "Invalid Rockridge CL"); 1998 return (NULL); 1999 } 2000 } 2001 if (file->cl_offset == file->offset || 2002 parent->rr_moved) { 2003 archive_set_error(&a->archive, 2004 ARCHIVE_ERRNO_MISC, 2005 "Invalid Rockridge CL"); 2006 return (NULL); 2007 } 2008 } 2009 } 2010 2011 #if DEBUG 2012 /* DEBUGGING: Warn about attributes I don't yet fully support. */ 2013 if ((flags & ~0x02) != 0) { 2014 fprintf(stderr, "\n ** Unrecognized flag: "); 2015 dump_isodirrec(stderr, isodirrec); 2016 fprintf(stderr, "\n"); 2017 } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { 2018 fprintf(stderr, "\n ** Unrecognized sequence number: "); 2019 dump_isodirrec(stderr, isodirrec); 2020 fprintf(stderr, "\n"); 2021 } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { 2022 fprintf(stderr, "\n ** Unexpected file unit size: "); 2023 dump_isodirrec(stderr, isodirrec); 2024 fprintf(stderr, "\n"); 2025 } else if (*(isodirrec + DR_interleave_offset) != 0) { 2026 fprintf(stderr, "\n ** Unexpected interleave: "); 2027 dump_isodirrec(stderr, isodirrec); 2028 fprintf(stderr, "\n"); 2029 } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { 2030 fprintf(stderr, "\n ** Unexpected extended attribute length: "); 2031 dump_isodirrec(stderr, isodirrec); 2032 fprintf(stderr, "\n"); 2033 } 2034 #endif 2035 register_file(iso9660, file); 2036 return (file); 2037 } 2038 2039 static int 2040 parse_rockridge(struct archive_read *a, struct file_info *file, 2041 const unsigned char *p, const unsigned char *end) 2042 { 2043 struct iso9660 *iso9660; 2044 2045 iso9660 = (struct iso9660 *)(a->format->data); 2046 2047 while (p + 4 <= end /* Enough space for another entry. */ 2048 && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ 2049 && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ 2050 && p[2] >= 4 /* Sanity-check length. */ 2051 && p + p[2] <= end) { /* Sanity-check length. */ 2052 const unsigned char *data = p + 4; 2053 int data_length = p[2] - 4; 2054 int version = p[3]; 2055 2056 /* 2057 * Yes, each 'if' here does test p[0] again. 2058 * Otherwise, the fall-through handling to catch 2059 * unsupported extensions doesn't work. 2060 */ 2061 switch(p[0]) { 2062 case 'C': 2063 if (p[0] == 'C' && p[1] == 'E') { 2064 if (version == 1 && data_length == 24) { 2065 /* 2066 * CE extension comprises: 2067 * 8 byte sector containing extension 2068 * 8 byte offset w/in above sector 2069 * 8 byte length of continuation 2070 */ 2071 int32_t location = 2072 archive_le32dec(data); 2073 file->ce_offset = 2074 archive_le32dec(data+8); 2075 file->ce_size = 2076 archive_le32dec(data+16); 2077 if (register_CE(a, location, file) 2078 != ARCHIVE_OK) 2079 return (ARCHIVE_FATAL); 2080 } 2081 break; 2082 } 2083 if (p[0] == 'C' && p[1] == 'L') { 2084 if (version == 1 && data_length == 8) { 2085 file->cl_offset = (uint64_t) 2086 iso9660->logical_block_size * 2087 (uint64_t)archive_le32dec(data); 2088 iso9660->seenRockridge = 1; 2089 } 2090 break; 2091 } 2092 /* FALLTHROUGH */ 2093 case 'N': 2094 if (p[0] == 'N' && p[1] == 'M') { 2095 if (version == 1) { 2096 parse_rockridge_NM1(file, 2097 data, data_length); 2098 iso9660->seenRockridge = 1; 2099 } 2100 break; 2101 } 2102 /* FALLTHROUGH */ 2103 case 'P': 2104 if (p[0] == 'P' && p[1] == 'D') { 2105 /* 2106 * PD extension is padding; 2107 * contents are always ignored. 2108 */ 2109 break; 2110 } 2111 if (p[0] == 'P' && p[1] == 'L') { 2112 /* 2113 * PL extension won't appear; 2114 * contents are always ignored. 2115 */ 2116 break; 2117 } 2118 if (p[0] == 'P' && p[1] == 'N') { 2119 if (version == 1 && data_length == 16) { 2120 file->rdev = toi(data,4); 2121 file->rdev <<= 32; 2122 file->rdev |= toi(data + 8, 4); 2123 iso9660->seenRockridge = 1; 2124 } 2125 break; 2126 } 2127 if (p[0] == 'P' && p[1] == 'X') { 2128 /* 2129 * PX extension comprises: 2130 * 8 bytes for mode, 2131 * 8 bytes for nlinks, 2132 * 8 bytes for uid, 2133 * 8 bytes for gid, 2134 * 8 bytes for inode. 2135 */ 2136 if (version == 1) { 2137 if (data_length >= 8) 2138 file->mode 2139 = toi(data, 4); 2140 if (data_length >= 16) 2141 file->nlinks 2142 = toi(data + 8, 4); 2143 if (data_length >= 24) 2144 file->uid 2145 = toi(data + 16, 4); 2146 if (data_length >= 32) 2147 file->gid 2148 = toi(data + 24, 4); 2149 if (data_length >= 40) 2150 file->number 2151 = toi(data + 32, 4); 2152 iso9660->seenRockridge = 1; 2153 } 2154 break; 2155 } 2156 /* FALLTHROUGH */ 2157 case 'R': 2158 if (p[0] == 'R' && p[1] == 'E' && version == 1) { 2159 file->re = 1; 2160 iso9660->seenRockridge = 1; 2161 break; 2162 } 2163 if (p[0] == 'R' && p[1] == 'R' && version == 1) { 2164 /* 2165 * RR extension comprises: 2166 * one byte flag value 2167 * This extension is obsolete, 2168 * so contents are always ignored. 2169 */ 2170 break; 2171 } 2172 /* FALLTHROUGH */ 2173 case 'S': 2174 if (p[0] == 'S' && p[1] == 'L') { 2175 if (version == 1) { 2176 parse_rockridge_SL1(file, 2177 data, data_length); 2178 iso9660->seenRockridge = 1; 2179 } 2180 break; 2181 } 2182 if (p[0] == 'S' && p[1] == 'T' 2183 && data_length == 0 && version == 1) { 2184 /* 2185 * ST extension marks end of this 2186 * block of SUSP entries. 2187 * 2188 * It allows SUSP to coexist with 2189 * non-SUSP uses of the System 2190 * Use Area by placing non-SUSP data 2191 * after SUSP data. 2192 */ 2193 iso9660->seenSUSP = 0; 2194 iso9660->seenRockridge = 0; 2195 return (ARCHIVE_OK); 2196 } 2197 case 'T': 2198 if (p[0] == 'T' && p[1] == 'F') { 2199 if (version == 1) { 2200 parse_rockridge_TF1(file, 2201 data, data_length); 2202 iso9660->seenRockridge = 1; 2203 } 2204 break; 2205 } 2206 /* FALLTHROUGH */ 2207 case 'Z': 2208 if (p[0] == 'Z' && p[1] == 'F') { 2209 if (version == 1) 2210 parse_rockridge_ZF1(file, 2211 data, data_length); 2212 break; 2213 } 2214 /* FALLTHROUGH */ 2215 default: 2216 /* The FALLTHROUGHs above leave us here for 2217 * any unsupported extension. */ 2218 break; 2219 } 2220 2221 2222 2223 p += p[2]; 2224 } 2225 return (ARCHIVE_OK); 2226 } 2227 2228 static int 2229 register_CE(struct archive_read *a, int32_t location, 2230 struct file_info *file) 2231 { 2232 struct iso9660 *iso9660; 2233 struct read_ce_queue *heap; 2234 struct read_ce_req *p; 2235 uint64_t offset, parent_offset; 2236 int hole, parent; 2237 2238 iso9660 = (struct iso9660 *)(a->format->data); 2239 offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size; 2240 if (((file->mode & AE_IFMT) == AE_IFREG && 2241 offset >= file->offset) || 2242 offset < iso9660->current_position || 2243 (((uint64_t)file->ce_offset) + file->ce_size) 2244 > iso9660->logical_block_size || 2245 offset + file->ce_offset + file->ce_size 2246 > iso9660->volume_size) { 2247 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2248 "Invalid parameter in SUSP \"CE\" extension"); 2249 return (ARCHIVE_FATAL); 2250 } 2251 2252 /* Expand our CE list as necessary. */ 2253 heap = &(iso9660->read_ce_req); 2254 if (heap->cnt >= heap->allocated) { 2255 int new_size; 2256 2257 if (heap->allocated < 16) 2258 new_size = 16; 2259 else 2260 new_size = heap->allocated * 2; 2261 /* Overflow might keep us from growing the list. */ 2262 if (new_size <= heap->allocated) { 2263 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2264 return (ARCHIVE_FATAL); 2265 } 2266 p = malloc(new_size * sizeof(p[0])); 2267 if (p == NULL) { 2268 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2269 return (ARCHIVE_FATAL); 2270 } 2271 if (heap->reqs != NULL) { 2272 memcpy(p, heap->reqs, heap->cnt * sizeof(*p)); 2273 free(heap->reqs); 2274 } 2275 heap->reqs = p; 2276 heap->allocated = new_size; 2277 } 2278 2279 /* 2280 * Start with hole at end, walk it up tree to find insertion point. 2281 */ 2282 hole = heap->cnt++; 2283 while (hole > 0) { 2284 parent = (hole - 1)/2; 2285 parent_offset = heap->reqs[parent].offset; 2286 if (offset >= parent_offset) { 2287 heap->reqs[hole].offset = offset; 2288 heap->reqs[hole].file = file; 2289 return (ARCHIVE_OK); 2290 } 2291 /* Move parent into hole <==> move hole up tree. */ 2292 heap->reqs[hole] = heap->reqs[parent]; 2293 hole = parent; 2294 } 2295 heap->reqs[0].offset = offset; 2296 heap->reqs[0].file = file; 2297 return (ARCHIVE_OK); 2298 } 2299 2300 static void 2301 next_CE(struct read_ce_queue *heap) 2302 { 2303 uint64_t a_offset, b_offset, c_offset; 2304 int a, b, c; 2305 struct read_ce_req tmp; 2306 2307 if (heap->cnt < 1) 2308 return; 2309 2310 /* 2311 * Move the last item in the heap to the root of the tree 2312 */ 2313 heap->reqs[0] = heap->reqs[--(heap->cnt)]; 2314 2315 /* 2316 * Rebalance the heap. 2317 */ 2318 a = 0; /* Starting element and its offset */ 2319 a_offset = heap->reqs[a].offset; 2320 for (;;) { 2321 b = a + a + 1; /* First child */ 2322 if (b >= heap->cnt) 2323 return; 2324 b_offset = heap->reqs[b].offset; 2325 c = b + 1; /* Use second child if it is smaller. */ 2326 if (c < heap->cnt) { 2327 c_offset = heap->reqs[c].offset; 2328 if (c_offset < b_offset) { 2329 b = c; 2330 b_offset = c_offset; 2331 } 2332 } 2333 if (a_offset <= b_offset) 2334 return; 2335 tmp = heap->reqs[a]; 2336 heap->reqs[a] = heap->reqs[b]; 2337 heap->reqs[b] = tmp; 2338 a = b; 2339 } 2340 } 2341 2342 2343 static int 2344 read_CE(struct archive_read *a, struct iso9660 *iso9660) 2345 { 2346 struct read_ce_queue *heap; 2347 const unsigned char *b, *p, *end; 2348 struct file_info *file; 2349 size_t step; 2350 int r; 2351 2352 /* Read data which RRIP "CE" extension points. */ 2353 heap = &(iso9660->read_ce_req); 2354 step = iso9660->logical_block_size; 2355 while (heap->cnt && 2356 heap->reqs[0].offset == iso9660->current_position) { 2357 b = __archive_read_ahead(a, step, NULL); 2358 if (b == NULL) { 2359 archive_set_error(&a->archive, 2360 ARCHIVE_ERRNO_MISC, 2361 "Failed to read full block when scanning " 2362 "ISO9660 directory list"); 2363 return (ARCHIVE_FATAL); 2364 } 2365 do { 2366 file = heap->reqs[0].file; 2367 if (file->ce_offset + file->ce_size > step) { 2368 archive_set_error(&a->archive, 2369 ARCHIVE_ERRNO_FILE_FORMAT, 2370 "Malformed CE information"); 2371 return (ARCHIVE_FATAL); 2372 } 2373 p = b + file->ce_offset; 2374 end = p + file->ce_size; 2375 next_CE(heap); 2376 r = parse_rockridge(a, file, p, end); 2377 if (r != ARCHIVE_OK) 2378 return (ARCHIVE_FATAL); 2379 } while (heap->cnt && 2380 heap->reqs[0].offset == iso9660->current_position); 2381 /* NOTE: Do not move this consume's code to fron of 2382 * do-while loop. Registration of nested CE extension 2383 * might cause error because of current position. */ 2384 __archive_read_consume(a, step); 2385 iso9660->current_position += step; 2386 } 2387 return (ARCHIVE_OK); 2388 } 2389 2390 static void 2391 parse_rockridge_NM1(struct file_info *file, 2392 const unsigned char *data, int data_length) 2393 { 2394 if (!file->name_continues) 2395 archive_string_empty(&file->name); 2396 file->name_continues = 0; 2397 if (data_length < 1) 2398 return; 2399 /* 2400 * NM version 1 extension comprises: 2401 * 1 byte flag, value is one of: 2402 * = 0: remainder is name 2403 * = 1: remainder is name, next NM entry continues name 2404 * = 2: "." 2405 * = 4: ".." 2406 * = 32: Implementation specific 2407 * All other values are reserved. 2408 */ 2409 switch(data[0]) { 2410 case 0: 2411 if (data_length < 2) 2412 return; 2413 archive_strncat(&file->name, 2414 (const char *)data + 1, data_length - 1); 2415 break; 2416 case 1: 2417 if (data_length < 2) 2418 return; 2419 archive_strncat(&file->name, 2420 (const char *)data + 1, data_length - 1); 2421 file->name_continues = 1; 2422 break; 2423 case 2: 2424 archive_strcat(&file->name, "."); 2425 break; 2426 case 4: 2427 archive_strcat(&file->name, ".."); 2428 break; 2429 default: 2430 return; 2431 } 2432 2433 } 2434 2435 static void 2436 parse_rockridge_TF1(struct file_info *file, const unsigned char *data, 2437 int data_length) 2438 { 2439 char flag; 2440 /* 2441 * TF extension comprises: 2442 * one byte flag 2443 * create time (optional) 2444 * modify time (optional) 2445 * access time (optional) 2446 * attribute time (optional) 2447 * Time format and presence of fields 2448 * is controlled by flag bits. 2449 */ 2450 if (data_length < 1) 2451 return; 2452 flag = data[0]; 2453 ++data; 2454 --data_length; 2455 if (flag & 0x80) { 2456 /* Use 17-byte time format. */ 2457 if ((flag & 1) && data_length >= 17) { 2458 /* Create time. */ 2459 file->birthtime_is_set = 1; 2460 file->birthtime = isodate17(data); 2461 data += 17; 2462 data_length -= 17; 2463 } 2464 if ((flag & 2) && data_length >= 17) { 2465 /* Modify time. */ 2466 file->mtime = isodate17(data); 2467 data += 17; 2468 data_length -= 17; 2469 } 2470 if ((flag & 4) && data_length >= 17) { 2471 /* Access time. */ 2472 file->atime = isodate17(data); 2473 data += 17; 2474 data_length -= 17; 2475 } 2476 if ((flag & 8) && data_length >= 17) { 2477 /* Attribute change time. */ 2478 file->ctime = isodate17(data); 2479 } 2480 } else { 2481 /* Use 7-byte time format. */ 2482 if ((flag & 1) && data_length >= 7) { 2483 /* Create time. */ 2484 file->birthtime_is_set = 1; 2485 file->birthtime = isodate7(data); 2486 data += 7; 2487 data_length -= 7; 2488 } 2489 if ((flag & 2) && data_length >= 7) { 2490 /* Modify time. */ 2491 file->mtime = isodate7(data); 2492 data += 7; 2493 data_length -= 7; 2494 } 2495 if ((flag & 4) && data_length >= 7) { 2496 /* Access time. */ 2497 file->atime = isodate7(data); 2498 data += 7; 2499 data_length -= 7; 2500 } 2501 if ((flag & 8) && data_length >= 7) { 2502 /* Attribute change time. */ 2503 file->ctime = isodate7(data); 2504 } 2505 } 2506 } 2507 2508 static void 2509 parse_rockridge_SL1(struct file_info *file, const unsigned char *data, 2510 int data_length) 2511 { 2512 const char *separator = ""; 2513 2514 if (!file->symlink_continues || file->symlink.length < 1) 2515 archive_string_empty(&file->symlink); 2516 else if (!file->symlink_continues && 2517 file->symlink.s[file->symlink.length - 1] != '/') 2518 separator = "/"; 2519 file->symlink_continues = 0; 2520 2521 /* 2522 * Defined flag values: 2523 * 0: This is the last SL record for this symbolic link 2524 * 1: this symbolic link field continues in next SL entry 2525 * All other values are reserved. 2526 */ 2527 if (data_length < 1) 2528 return; 2529 switch(*data) { 2530 case 0: 2531 break; 2532 case 1: 2533 file->symlink_continues = 1; 2534 break; 2535 default: 2536 return; 2537 } 2538 ++data; /* Skip flag byte. */ 2539 --data_length; 2540 2541 /* 2542 * SL extension body stores "components". 2543 * Basically, this is a complicated way of storing 2544 * a POSIX path. It also interferes with using 2545 * symlinks for storing non-path data. <sigh> 2546 * 2547 * Each component is 2 bytes (flag and length) 2548 * possibly followed by name data. 2549 */ 2550 while (data_length >= 2) { 2551 unsigned char flag = *data++; 2552 unsigned char nlen = *data++; 2553 data_length -= 2; 2554 2555 archive_strcat(&file->symlink, separator); 2556 separator = "/"; 2557 2558 switch(flag) { 2559 case 0: /* Usual case, this is text. */ 2560 if (data_length < nlen) 2561 return; 2562 archive_strncat(&file->symlink, 2563 (const char *)data, nlen); 2564 break; 2565 case 0x01: /* Text continues in next component. */ 2566 if (data_length < nlen) 2567 return; 2568 archive_strncat(&file->symlink, 2569 (const char *)data, nlen); 2570 separator = ""; 2571 break; 2572 case 0x02: /* Current dir. */ 2573 archive_strcat(&file->symlink, "."); 2574 break; 2575 case 0x04: /* Parent dir. */ 2576 archive_strcat(&file->symlink, ".."); 2577 break; 2578 case 0x08: /* Root of filesystem. */ 2579 archive_strcat(&file->symlink, "/"); 2580 separator = ""; 2581 break; 2582 case 0x10: /* Undefined (historically "volume root" */ 2583 archive_string_empty(&file->symlink); 2584 archive_strcat(&file->symlink, "ROOT"); 2585 break; 2586 case 0x20: /* Undefined (historically "hostname") */ 2587 archive_strcat(&file->symlink, "hostname"); 2588 break; 2589 default: 2590 /* TODO: issue a warning ? */ 2591 return; 2592 } 2593 data += nlen; 2594 data_length -= nlen; 2595 } 2596 } 2597 2598 static void 2599 parse_rockridge_ZF1(struct file_info *file, const unsigned char *data, 2600 int data_length) 2601 { 2602 2603 if (data[0] == 0x70 && data[1] == 0x7a && data_length == 12) { 2604 /* paged zlib */ 2605 file->pz = 1; 2606 file->pz_log2_bs = data[3]; 2607 file->pz_uncompressed_size = archive_le32dec(&data[4]); 2608 } 2609 } 2610 2611 static void 2612 register_file(struct iso9660 *iso9660, struct file_info *file) 2613 { 2614 2615 file->use_next = iso9660->use_files; 2616 iso9660->use_files = file; 2617 } 2618 2619 static void 2620 release_files(struct iso9660 *iso9660) 2621 { 2622 struct content *con, *connext; 2623 struct file_info *file; 2624 2625 file = iso9660->use_files; 2626 while (file != NULL) { 2627 struct file_info *next = file->use_next; 2628 2629 archive_string_free(&file->name); 2630 archive_string_free(&file->symlink); 2631 free(file->utf16be_name); 2632 con = file->contents.first; 2633 while (con != NULL) { 2634 connext = con->next; 2635 free(con); 2636 con = connext; 2637 } 2638 free(file); 2639 file = next; 2640 } 2641 } 2642 2643 static int 2644 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, 2645 struct file_info **pfile) 2646 { 2647 struct file_info *file; 2648 int r; 2649 2650 r = next_cache_entry(a, iso9660, pfile); 2651 if (r != ARCHIVE_OK) 2652 return (r); 2653 file = *pfile; 2654 2655 /* Don't waste time seeking for zero-length bodies. */ 2656 if (file->size == 0) 2657 file->offset = iso9660->current_position; 2658 2659 /* flush any remaining bytes from the last round to ensure 2660 * we're positioned */ 2661 if (iso9660->entry_bytes_unconsumed) { 2662 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 2663 iso9660->entry_bytes_unconsumed = 0; 2664 } 2665 2666 /* Seek forward to the start of the entry. */ 2667 if (iso9660->current_position < file->offset) { 2668 int64_t step; 2669 2670 step = file->offset - iso9660->current_position; 2671 step = __archive_read_consume(a, step); 2672 if (step < 0) 2673 return ((int)step); 2674 iso9660->current_position = file->offset; 2675 } 2676 2677 /* We found body of file; handle it now. */ 2678 return (ARCHIVE_OK); 2679 } 2680 2681 static int 2682 next_cache_entry(struct archive_read *a, struct iso9660 *iso9660, 2683 struct file_info **pfile) 2684 { 2685 struct file_info *file; 2686 struct { 2687 struct file_info *first; 2688 struct file_info **last; 2689 } empty_files; 2690 int64_t number; 2691 int count; 2692 2693 file = cache_get_entry(iso9660); 2694 if (file != NULL) { 2695 *pfile = file; 2696 return (ARCHIVE_OK); 2697 } 2698 2699 for (;;) { 2700 struct file_info *re, *d; 2701 2702 *pfile = file = next_entry(iso9660); 2703 if (file == NULL) { 2704 /* 2705 * If directory entries all which are descendant of 2706 * rr_moved are stil remaning, expose their. 2707 */ 2708 if (iso9660->re_files.first != NULL && 2709 iso9660->rr_moved != NULL && 2710 iso9660->rr_moved->rr_moved_has_re_only) 2711 /* Expose "rr_moved" entry. */ 2712 cache_add_entry(iso9660, iso9660->rr_moved); 2713 while ((re = re_get_entry(iso9660)) != NULL) { 2714 /* Expose its descendant dirs. */ 2715 while ((d = rede_get_entry(re)) != NULL) 2716 cache_add_entry(iso9660, d); 2717 } 2718 if (iso9660->cache_files.first != NULL) 2719 return (next_cache_entry(a, iso9660, pfile)); 2720 return (ARCHIVE_EOF); 2721 } 2722 2723 if (file->cl_offset) { 2724 struct file_info *first_re = NULL; 2725 int nexted_re = 0; 2726 2727 /* 2728 * Find "RE" dir for the current file, which 2729 * has "CL" flag. 2730 */ 2731 while ((re = re_get_entry(iso9660)) 2732 != first_re) { 2733 if (first_re == NULL) 2734 first_re = re; 2735 if (re->offset == file->cl_offset) { 2736 re->parent->subdirs--; 2737 re->parent = file->parent; 2738 re->re = 0; 2739 if (re->parent->re_descendant) { 2740 nexted_re = 1; 2741 re->re_descendant = 1; 2742 if (rede_add_entry(re) < 0) 2743 goto fatal_rr; 2744 /* Move a list of descendants 2745 * to a new ancestor. */ 2746 while ((d = rede_get_entry( 2747 re)) != NULL) 2748 if (rede_add_entry(d) 2749 < 0) 2750 goto fatal_rr; 2751 break; 2752 } 2753 /* Replace the current file 2754 * with "RE" dir */ 2755 *pfile = file = re; 2756 /* Expose its descendant */ 2757 while ((d = rede_get_entry( 2758 file)) != NULL) 2759 cache_add_entry( 2760 iso9660, d); 2761 break; 2762 } else 2763 re_add_entry(iso9660, re); 2764 } 2765 if (nexted_re) { 2766 /* 2767 * Do not expose this at this time 2768 * because we have not gotten its full-path 2769 * name yet. 2770 */ 2771 continue; 2772 } 2773 } else if ((file->mode & AE_IFMT) == AE_IFDIR) { 2774 int r; 2775 2776 /* Read file entries in this dir. */ 2777 r = read_children(a, file); 2778 if (r != ARCHIVE_OK) 2779 return (r); 2780 2781 /* 2782 * Handle a special dir of Rockridge extensions, 2783 * "rr_moved". 2784 */ 2785 if (file->rr_moved) { 2786 /* 2787 * If this has only the subdirectories which 2788 * have "RE" flags, do not expose at this time. 2789 */ 2790 if (file->rr_moved_has_re_only) 2791 continue; 2792 /* Otherwise expose "rr_moved" entry. */ 2793 } else if (file->re) { 2794 /* 2795 * Do not expose this at this time 2796 * because we have not gotten its full-path 2797 * name yet. 2798 */ 2799 re_add_entry(iso9660, file); 2800 continue; 2801 } else if (file->re_descendant) { 2802 /* 2803 * If the top level "RE" entry of this entry 2804 * is not exposed, we, accordingly, should not 2805 * expose this entry at this time because 2806 * we cannot make its proper full-path name. 2807 */ 2808 if (rede_add_entry(file) == 0) 2809 continue; 2810 /* Otherwise we can expose this entry because 2811 * it seems its top level "RE" has already been 2812 * exposed. */ 2813 } 2814 } 2815 break; 2816 } 2817 2818 if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1) 2819 return (ARCHIVE_OK); 2820 2821 count = 0; 2822 number = file->number; 2823 iso9660->cache_files.first = NULL; 2824 iso9660->cache_files.last = &(iso9660->cache_files.first); 2825 empty_files.first = NULL; 2826 empty_files.last = &empty_files.first; 2827 /* Collect files which has the same file serial number. 2828 * Peek pending_files so that file which number is different 2829 * is not put bak. */ 2830 while (iso9660->pending_files.used > 0 && 2831 (iso9660->pending_files.files[0]->number == -1 || 2832 iso9660->pending_files.files[0]->number == number)) { 2833 if (file->number == -1) { 2834 /* This file has the same offset 2835 * but it's wrong offset which empty files 2836 * and symlink files have. 2837 * NOTE: This wrong offse was recorded by 2838 * old mkisofs utility. If ISO images is 2839 * created by latest mkisofs, this does not 2840 * happen. 2841 */ 2842 file->next = NULL; 2843 *empty_files.last = file; 2844 empty_files.last = &(file->next); 2845 } else { 2846 count++; 2847 cache_add_entry(iso9660, file); 2848 } 2849 file = next_entry(iso9660); 2850 } 2851 2852 if (count == 0) { 2853 *pfile = file; 2854 return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2855 } 2856 if (file->number == -1) { 2857 file->next = NULL; 2858 *empty_files.last = file; 2859 empty_files.last = &(file->next); 2860 } else { 2861 count++; 2862 cache_add_entry(iso9660, file); 2863 } 2864 2865 if (count > 1) { 2866 /* The count is the same as number of hardlink, 2867 * so much so that each nlinks of files in cache_file 2868 * is overwritten by value of the count. 2869 */ 2870 for (file = iso9660->cache_files.first; 2871 file != NULL; file = file->next) 2872 file->nlinks = count; 2873 } 2874 /* If there are empty files, that files are added 2875 * to the tail of the cache_files. */ 2876 if (empty_files.first != NULL) { 2877 *iso9660->cache_files.last = empty_files.first; 2878 iso9660->cache_files.last = empty_files.last; 2879 } 2880 *pfile = cache_get_entry(iso9660); 2881 return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2882 2883 fatal_rr: 2884 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2885 "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of" 2886 "Rockridge extensions"); 2887 return (ARCHIVE_FATAL); 2888 } 2889 2890 static inline void 2891 re_add_entry(struct iso9660 *iso9660, struct file_info *file) 2892 { 2893 file->re_next = NULL; 2894 *iso9660->re_files.last = file; 2895 iso9660->re_files.last = &(file->re_next); 2896 } 2897 2898 static inline struct file_info * 2899 re_get_entry(struct iso9660 *iso9660) 2900 { 2901 struct file_info *file; 2902 2903 if ((file = iso9660->re_files.first) != NULL) { 2904 iso9660->re_files.first = file->re_next; 2905 if (iso9660->re_files.first == NULL) 2906 iso9660->re_files.last = 2907 &(iso9660->re_files.first); 2908 } 2909 return (file); 2910 } 2911 2912 static inline int 2913 rede_add_entry(struct file_info *file) 2914 { 2915 struct file_info *re; 2916 2917 /* 2918 * Find "RE" entry. 2919 */ 2920 re = file->parent; 2921 while (re != NULL && !re->re) 2922 re = re->parent; 2923 if (re == NULL) 2924 return (-1); 2925 2926 file->re_next = NULL; 2927 *re->rede_files.last = file; 2928 re->rede_files.last = &(file->re_next); 2929 return (0); 2930 } 2931 2932 static inline struct file_info * 2933 rede_get_entry(struct file_info *re) 2934 { 2935 struct file_info *file; 2936 2937 if ((file = re->rede_files.first) != NULL) { 2938 re->rede_files.first = file->re_next; 2939 if (re->rede_files.first == NULL) 2940 re->rede_files.last = 2941 &(re->rede_files.first); 2942 } 2943 return (file); 2944 } 2945 2946 static inline void 2947 cache_add_entry(struct iso9660 *iso9660, struct file_info *file) 2948 { 2949 file->next = NULL; 2950 *iso9660->cache_files.last = file; 2951 iso9660->cache_files.last = &(file->next); 2952 } 2953 2954 static inline struct file_info * 2955 cache_get_entry(struct iso9660 *iso9660) 2956 { 2957 struct file_info *file; 2958 2959 if ((file = iso9660->cache_files.first) != NULL) { 2960 iso9660->cache_files.first = file->next; 2961 if (iso9660->cache_files.first == NULL) 2962 iso9660->cache_files.last = 2963 &(iso9660->cache_files.first); 2964 } 2965 return (file); 2966 } 2967 2968 static int 2969 heap_add_entry(struct archive_read *a, struct heap_queue *heap, 2970 struct file_info *file, uint64_t key) 2971 { 2972 uint64_t file_key, parent_key; 2973 int hole, parent; 2974 2975 /* Expand our pending files list as necessary. */ 2976 if (heap->used >= heap->allocated) { 2977 struct file_info **new_pending_files; 2978 int new_size = heap->allocated * 2; 2979 2980 if (heap->allocated < 1024) 2981 new_size = 1024; 2982 /* Overflow might keep us from growing the list. */ 2983 if (new_size <= heap->allocated) { 2984 archive_set_error(&a->archive, 2985 ENOMEM, "Out of memory"); 2986 return (ARCHIVE_FATAL); 2987 } 2988 new_pending_files = (struct file_info **) 2989 malloc(new_size * sizeof(new_pending_files[0])); 2990 if (new_pending_files == NULL) { 2991 archive_set_error(&a->archive, 2992 ENOMEM, "Out of memory"); 2993 return (ARCHIVE_FATAL); 2994 } 2995 memcpy(new_pending_files, heap->files, 2996 heap->allocated * sizeof(new_pending_files[0])); 2997 if (heap->files != NULL) 2998 free(heap->files); 2999 heap->files = new_pending_files; 3000 heap->allocated = new_size; 3001 } 3002 3003 file_key = file->key = key; 3004 3005 /* 3006 * Start with hole at end, walk it up tree to find insertion point. 3007 */ 3008 hole = heap->used++; 3009 while (hole > 0) { 3010 parent = (hole - 1)/2; 3011 parent_key = heap->files[parent]->key; 3012 if (file_key >= parent_key) { 3013 heap->files[hole] = file; 3014 return (ARCHIVE_OK); 3015 } 3016 /* Move parent into hole <==> move hole up tree. */ 3017 heap->files[hole] = heap->files[parent]; 3018 hole = parent; 3019 } 3020 heap->files[0] = file; 3021 3022 return (ARCHIVE_OK); 3023 } 3024 3025 static struct file_info * 3026 heap_get_entry(struct heap_queue *heap) 3027 { 3028 uint64_t a_key, b_key, c_key; 3029 int a, b, c; 3030 struct file_info *r, *tmp; 3031 3032 if (heap->used < 1) 3033 return (NULL); 3034 3035 /* 3036 * The first file in the list is the earliest; we'll return this. 3037 */ 3038 r = heap->files[0]; 3039 3040 /* 3041 * Move the last item in the heap to the root of the tree 3042 */ 3043 heap->files[0] = heap->files[--(heap->used)]; 3044 3045 /* 3046 * Rebalance the heap. 3047 */ 3048 a = 0; /* Starting element and its heap key */ 3049 a_key = heap->files[a]->key; 3050 for (;;) { 3051 b = a + a + 1; /* First child */ 3052 if (b >= heap->used) 3053 return (r); 3054 b_key = heap->files[b]->key; 3055 c = b + 1; /* Use second child if it is smaller. */ 3056 if (c < heap->used) { 3057 c_key = heap->files[c]->key; 3058 if (c_key < b_key) { 3059 b = c; 3060 b_key = c_key; 3061 } 3062 } 3063 if (a_key <= b_key) 3064 return (r); 3065 tmp = heap->files[a]; 3066 heap->files[a] = heap->files[b]; 3067 heap->files[b] = tmp; 3068 a = b; 3069 } 3070 } 3071 3072 static unsigned int 3073 toi(const void *p, int n) 3074 { 3075 const unsigned char *v = (const unsigned char *)p; 3076 if (n > 1) 3077 return v[0] + 256 * toi(v + 1, n - 1); 3078 if (n == 1) 3079 return v[0]; 3080 return (0); 3081 } 3082 3083 static time_t 3084 isodate7(const unsigned char *v) 3085 { 3086 struct tm tm; 3087 int offset; 3088 memset(&tm, 0, sizeof(tm)); 3089 tm.tm_year = v[0]; 3090 tm.tm_mon = v[1] - 1; 3091 tm.tm_mday = v[2]; 3092 tm.tm_hour = v[3]; 3093 tm.tm_min = v[4]; 3094 tm.tm_sec = v[5]; 3095 /* v[6] is the signed timezone offset, in 1/4-hour increments. */ 3096 offset = ((const signed char *)v)[6]; 3097 if (offset > -48 && offset < 52) { 3098 tm.tm_hour -= offset / 4; 3099 tm.tm_min -= (offset % 4) * 15; 3100 } 3101 return (time_from_tm(&tm)); 3102 } 3103 3104 static time_t 3105 isodate17(const unsigned char *v) 3106 { 3107 struct tm tm; 3108 int offset; 3109 memset(&tm, 0, sizeof(tm)); 3110 tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 3111 + (v[2] - '0') * 10 + (v[3] - '0') 3112 - 1900; 3113 tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); 3114 tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); 3115 tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); 3116 tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); 3117 tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); 3118 /* v[16] is the signed timezone offset, in 1/4-hour increments. */ 3119 offset = ((const signed char *)v)[16]; 3120 if (offset > -48 && offset < 52) { 3121 tm.tm_hour -= offset / 4; 3122 tm.tm_min -= (offset % 4) * 15; 3123 } 3124 return (time_from_tm(&tm)); 3125 } 3126 3127 static time_t 3128 time_from_tm(struct tm *t) 3129 { 3130 #if HAVE_TIMEGM 3131 /* Use platform timegm() if available. */ 3132 return (timegm(t)); 3133 #elif HAVE__MKGMTIME64 3134 return (_mkgmtime64(t)); 3135 #else 3136 /* Else use direct calculation using POSIX assumptions. */ 3137 /* First, fix up tm_yday based on the year/month/day. */ 3138 mktime(t); 3139 /* Then we can compute timegm() from first principles. */ 3140 return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 3141 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 3142 + ((t->tm_year - 69) / 4) * 86400 - 3143 ((t->tm_year - 1) / 100) * 86400 3144 + ((t->tm_year + 299) / 400) * 86400); 3145 #endif 3146 } 3147 3148 static const char * 3149 build_pathname(struct archive_string *as, struct file_info *file) 3150 { 3151 if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { 3152 build_pathname(as, file->parent); 3153 archive_strcat(as, "/"); 3154 } 3155 if (archive_strlen(&file->name) == 0) 3156 archive_strcat(as, "."); 3157 else 3158 archive_string_concat(as, &file->name); 3159 return (as->s); 3160 } 3161 3162 static int 3163 build_pathname_utf16be(unsigned char *p, size_t max, size_t *len, 3164 struct file_info *file) 3165 { 3166 if (file->parent != NULL && file->parent->utf16be_bytes > 0) { 3167 if (build_pathname_utf16be(p, max, len, file->parent) != 0) 3168 return (-1); 3169 p[*len] = 0; 3170 p[*len + 1] = '/'; 3171 *len += 2; 3172 } 3173 if (file->utf16be_bytes == 0) { 3174 if (*len + 2 > max) 3175 return (-1);/* Path is too long! */ 3176 p[*len] = 0; 3177 p[*len + 1] = '.'; 3178 *len += 2; 3179 } else { 3180 if (*len + file->utf16be_bytes > max) 3181 return (-1);/* Path is too long! */ 3182 memcpy(p + *len, file->utf16be_name, file->utf16be_bytes); 3183 *len += file->utf16be_bytes; 3184 } 3185 return (0); 3186 } 3187 3188 #if DEBUG 3189 static void 3190 dump_isodirrec(FILE *out, const unsigned char *isodirrec) 3191 { 3192 fprintf(out, " l %d,", 3193 toi(isodirrec + DR_length_offset, DR_length_size)); 3194 fprintf(out, " a %d,", 3195 toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); 3196 fprintf(out, " ext 0x%x,", 3197 toi(isodirrec + DR_extent_offset, DR_extent_size)); 3198 fprintf(out, " s %d,", 3199 toi(isodirrec + DR_size_offset, DR_extent_size)); 3200 fprintf(out, " f 0x%x,", 3201 toi(isodirrec + DR_flags_offset, DR_flags_size)); 3202 fprintf(out, " u %d,", 3203 toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); 3204 fprintf(out, " ilv %d,", 3205 toi(isodirrec + DR_interleave_offset, DR_interleave_size)); 3206 fprintf(out, " seq %d,", 3207 toi(isodirrec + DR_volume_sequence_number_offset, DR_volume_sequence_number_size)); 3208 fprintf(out, " nl %d:", 3209 toi(isodirrec + DR_name_len_offset, DR_name_len_size)); 3210 fprintf(out, " `%.*s'", 3211 toi(isodirrec + DR_name_len_offset, DR_name_len_size), isodirrec + DR_name_offset); 3212 } 3213 #endif 3214