1 /*- 2 * Copyright (c) 2003-2007 Tim Kientzle 3 * Copyright (c) 2009 Andreas Henriksson <andreas@fatal.se> 4 * Copyright (c) 2009-2012 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("$FreeBSD$"); 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 being 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 /* Null buffer used in bidder to improve its performance. */ 378 unsigned char null[2048]; 379 }; 380 381 static int archive_read_format_iso9660_bid(struct archive_read *, int); 382 static int archive_read_format_iso9660_options(struct archive_read *, 383 const char *, const char *); 384 static int archive_read_format_iso9660_cleanup(struct archive_read *); 385 static int archive_read_format_iso9660_read_data(struct archive_read *, 386 const void **, size_t *, int64_t *); 387 static int archive_read_format_iso9660_read_data_skip(struct archive_read *); 388 static int archive_read_format_iso9660_read_header(struct archive_read *, 389 struct archive_entry *); 390 static const char *build_pathname(struct archive_string *, struct file_info *, int); 391 static int build_pathname_utf16be(unsigned char *, size_t, size_t *, 392 struct file_info *); 393 #if DEBUG 394 static void dump_isodirrec(FILE *, const unsigned char *isodirrec); 395 #endif 396 static time_t time_from_tm(struct tm *); 397 static time_t isodate17(const unsigned char *); 398 static time_t isodate7(const unsigned char *); 399 static int isBootRecord(struct iso9660 *, const unsigned char *); 400 static int isVolumePartition(struct iso9660 *, const unsigned char *); 401 static int isVDSetTerminator(struct iso9660 *, const unsigned char *); 402 static int isJolietSVD(struct iso9660 *, const unsigned char *); 403 static int isSVD(struct iso9660 *, const unsigned char *); 404 static int isEVD(struct iso9660 *, const unsigned char *); 405 static int isPVD(struct iso9660 *, const unsigned char *); 406 static int next_cache_entry(struct archive_read *, struct iso9660 *, 407 struct file_info **); 408 static int next_entry_seek(struct archive_read *, struct iso9660 *, 409 struct file_info **); 410 static struct file_info * 411 parse_file_info(struct archive_read *a, 412 struct file_info *parent, const unsigned char *isodirrec); 413 static int parse_rockridge(struct archive_read *a, 414 struct file_info *file, const unsigned char *start, 415 const unsigned char *end); 416 static int register_CE(struct archive_read *a, int32_t location, 417 struct file_info *file); 418 static int read_CE(struct archive_read *a, struct iso9660 *iso9660); 419 static void parse_rockridge_NM1(struct file_info *, 420 const unsigned char *, int); 421 static void parse_rockridge_SL1(struct file_info *, 422 const unsigned char *, int); 423 static void parse_rockridge_TF1(struct file_info *, 424 const unsigned char *, int); 425 static void parse_rockridge_ZF1(struct file_info *, 426 const unsigned char *, int); 427 static void register_file(struct iso9660 *, struct file_info *); 428 static void release_files(struct iso9660 *); 429 static unsigned toi(const void *p, int n); 430 static inline void re_add_entry(struct iso9660 *, struct file_info *); 431 static inline struct file_info * re_get_entry(struct iso9660 *); 432 static inline int rede_add_entry(struct file_info *); 433 static inline struct file_info * rede_get_entry(struct file_info *); 434 static inline void cache_add_entry(struct iso9660 *iso9660, 435 struct file_info *file); 436 static inline struct file_info *cache_get_entry(struct iso9660 *iso9660); 437 static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, 438 struct file_info *file, uint64_t key); 439 static struct file_info *heap_get_entry(struct heap_queue *heap); 440 441 #define add_entry(arch, iso9660, file) \ 442 heap_add_entry(arch, &((iso9660)->pending_files), file, file->offset) 443 #define next_entry(iso9660) \ 444 heap_get_entry(&((iso9660)->pending_files)) 445 446 int 447 archive_read_support_format_iso9660(struct archive *_a) 448 { 449 struct archive_read *a = (struct archive_read *)_a; 450 struct iso9660 *iso9660; 451 int r; 452 453 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 454 ARCHIVE_STATE_NEW, "archive_read_support_format_iso9660"); 455 456 iso9660 = (struct iso9660 *)calloc(1, sizeof(*iso9660)); 457 if (iso9660 == NULL) { 458 archive_set_error(&a->archive, ENOMEM, 459 "Can't allocate iso9660 data"); 460 return (ARCHIVE_FATAL); 461 } 462 iso9660->magic = ISO9660_MAGIC; 463 iso9660->cache_files.first = NULL; 464 iso9660->cache_files.last = &(iso9660->cache_files.first); 465 iso9660->re_files.first = NULL; 466 iso9660->re_files.last = &(iso9660->re_files.first); 467 /* Enable to support Joliet extensions by default. */ 468 iso9660->opt_support_joliet = 1; 469 /* Enable to support Rock Ridge extensions by default. */ 470 iso9660->opt_support_rockridge = 1; 471 472 r = __archive_read_register_format(a, 473 iso9660, 474 "iso9660", 475 archive_read_format_iso9660_bid, 476 archive_read_format_iso9660_options, 477 archive_read_format_iso9660_read_header, 478 archive_read_format_iso9660_read_data, 479 archive_read_format_iso9660_read_data_skip, 480 NULL, 481 archive_read_format_iso9660_cleanup, 482 NULL, 483 NULL); 484 485 if (r != ARCHIVE_OK) { 486 free(iso9660); 487 return (r); 488 } 489 return (ARCHIVE_OK); 490 } 491 492 493 static int 494 archive_read_format_iso9660_bid(struct archive_read *a, int best_bid) 495 { 496 struct iso9660 *iso9660; 497 ssize_t bytes_read; 498 const unsigned char *p; 499 int seenTerminator; 500 501 /* If there's already a better bid than we can ever 502 make, don't bother testing. */ 503 if (best_bid > 48) 504 return (-1); 505 506 iso9660 = (struct iso9660 *)(a->format->data); 507 508 /* 509 * Skip the first 32k (reserved area) and get the first 510 * 8 sectors of the volume descriptor table. Of course, 511 * if the I/O layer gives us more, we'll take it. 512 */ 513 #define RESERVED_AREA (SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE) 514 p = __archive_read_ahead(a, 515 RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE, 516 &bytes_read); 517 if (p == NULL) 518 return (-1); 519 520 /* Skip the reserved area. */ 521 bytes_read -= RESERVED_AREA; 522 p += RESERVED_AREA; 523 524 /* Check each volume descriptor. */ 525 seenTerminator = 0; 526 for (; bytes_read > LOGICAL_BLOCK_SIZE; 527 bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) { 528 /* Do not handle undefined Volume Descriptor Type. */ 529 if (p[0] >= 4 && p[0] <= 254) 530 return (0); 531 /* Standard Identifier must be "CD001" */ 532 if (memcmp(p + 1, "CD001", 5) != 0) 533 return (0); 534 if (isPVD(iso9660, p)) 535 continue; 536 if (!iso9660->joliet.location) { 537 if (isJolietSVD(iso9660, p)) 538 continue; 539 } 540 if (isBootRecord(iso9660, p)) 541 continue; 542 if (isEVD(iso9660, p)) 543 continue; 544 if (isSVD(iso9660, p)) 545 continue; 546 if (isVolumePartition(iso9660, p)) 547 continue; 548 if (isVDSetTerminator(iso9660, p)) { 549 seenTerminator = 1; 550 break; 551 } 552 return (0); 553 } 554 /* 555 * ISO 9660 format must have Primary Volume Descriptor and 556 * Volume Descriptor Set Terminator. 557 */ 558 if (seenTerminator && iso9660->primary.location > 16) 559 return (48); 560 561 /* We didn't find a valid PVD; return a bid of zero. */ 562 return (0); 563 } 564 565 static int 566 archive_read_format_iso9660_options(struct archive_read *a, 567 const char *key, const char *val) 568 { 569 struct iso9660 *iso9660; 570 571 iso9660 = (struct iso9660 *)(a->format->data); 572 573 if (strcmp(key, "joliet") == 0) { 574 if (val == NULL || strcmp(val, "off") == 0 || 575 strcmp(val, "ignore") == 0 || 576 strcmp(val, "disable") == 0 || 577 strcmp(val, "0") == 0) 578 iso9660->opt_support_joliet = 0; 579 else 580 iso9660->opt_support_joliet = 1; 581 return (ARCHIVE_OK); 582 } 583 if (strcmp(key, "rockridge") == 0 || 584 strcmp(key, "Rockridge") == 0) { 585 iso9660->opt_support_rockridge = val != NULL; 586 return (ARCHIVE_OK); 587 } 588 589 /* Note: The "warn" return is just to inform the options 590 * supervisor that we didn't handle it. It will generate 591 * a suitable error if no one used this option. */ 592 return (ARCHIVE_WARN); 593 } 594 595 static int 596 isNull(struct iso9660 *iso9660, const unsigned char *h, unsigned offset, 597 unsigned bytes) 598 { 599 600 while (bytes >= sizeof(iso9660->null)) { 601 if (!memcmp(iso9660->null, h + offset, sizeof(iso9660->null))) 602 return (0); 603 offset += sizeof(iso9660->null); 604 bytes -= sizeof(iso9660->null); 605 } 606 if (bytes) 607 return memcmp(iso9660->null, h + offset, bytes) == 0; 608 else 609 return (1); 610 } 611 612 static int 613 isBootRecord(struct iso9660 *iso9660, const unsigned char *h) 614 { 615 (void)iso9660; /* UNUSED */ 616 617 /* Type of the Volume Descriptor Boot Record must be 0. */ 618 if (h[0] != 0) 619 return (0); 620 621 /* Volume Descriptor Version must be 1. */ 622 if (h[6] != 1) 623 return (0); 624 625 return (1); 626 } 627 628 static int 629 isVolumePartition(struct iso9660 *iso9660, const unsigned char *h) 630 { 631 int32_t location; 632 633 /* Type of the Volume Partition Descriptor must be 3. */ 634 if (h[0] != 3) 635 return (0); 636 637 /* Volume Descriptor Version must be 1. */ 638 if (h[6] != 1) 639 return (0); 640 /* Unused Field */ 641 if (h[7] != 0) 642 return (0); 643 644 location = archive_le32dec(h + 72); 645 if (location <= SYSTEM_AREA_BLOCK || 646 location >= iso9660->volume_block) 647 return (0); 648 if ((uint32_t)location != archive_be32dec(h + 76)) 649 return (0); 650 651 return (1); 652 } 653 654 static int 655 isVDSetTerminator(struct iso9660 *iso9660, const unsigned char *h) 656 { 657 (void)iso9660; /* UNUSED */ 658 659 /* Type of the Volume Descriptor Set Terminator must be 255. */ 660 if (h[0] != 255) 661 return (0); 662 663 /* Volume Descriptor Version must be 1. */ 664 if (h[6] != 1) 665 return (0); 666 667 /* Reserved field must be 0. */ 668 if (!isNull(iso9660, h, 7, 2048-7)) 669 return (0); 670 671 return (1); 672 } 673 674 static int 675 isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) 676 { 677 const unsigned char *p; 678 ssize_t logical_block_size; 679 int32_t volume_block; 680 681 /* Check if current sector is a kind of Supplementary Volume 682 * Descriptor. */ 683 if (!isSVD(iso9660, h)) 684 return (0); 685 686 /* FIXME: do more validations according to joliet spec. */ 687 688 /* check if this SVD contains joliet extension! */ 689 p = h + SVD_escape_sequences_offset; 690 /* N.B. Joliet spec says p[1] == '\\', but.... */ 691 if (p[0] == '%' && p[1] == '/') { 692 int level = 0; 693 694 if (p[2] == '@') 695 level = 1; 696 else if (p[2] == 'C') 697 level = 2; 698 else if (p[2] == 'E') 699 level = 3; 700 else /* not joliet */ 701 return (0); 702 703 iso9660->seenJoliet = level; 704 705 } else /* not joliet */ 706 return (0); 707 708 logical_block_size = 709 archive_le16dec(h + SVD_logical_block_size_offset); 710 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 711 712 iso9660->logical_block_size = logical_block_size; 713 iso9660->volume_block = volume_block; 714 iso9660->volume_size = logical_block_size * (uint64_t)volume_block; 715 /* Read Root Directory Record in Volume Descriptor. */ 716 p = h + SVD_root_directory_record_offset; 717 iso9660->joliet.location = archive_le32dec(p + DR_extent_offset); 718 iso9660->joliet.size = archive_le32dec(p + DR_size_offset); 719 720 return (48); 721 } 722 723 static int 724 isSVD(struct iso9660 *iso9660, const unsigned char *h) 725 { 726 const unsigned char *p; 727 ssize_t logical_block_size; 728 int32_t volume_block; 729 int32_t location; 730 731 (void)iso9660; /* UNUSED */ 732 733 /* Type 2 means it's a SVD. */ 734 if (h[SVD_type_offset] != 2) 735 return (0); 736 737 /* Reserved field must be 0. */ 738 if (!isNull(iso9660, h, SVD_reserved1_offset, SVD_reserved1_size)) 739 return (0); 740 if (!isNull(iso9660, h, SVD_reserved2_offset, SVD_reserved2_size)) 741 return (0); 742 if (!isNull(iso9660, h, SVD_reserved3_offset, SVD_reserved3_size)) 743 return (0); 744 745 /* File structure version must be 1 for ISO9660/ECMA119. */ 746 if (h[SVD_file_structure_version_offset] != 1) 747 return (0); 748 749 logical_block_size = 750 archive_le16dec(h + SVD_logical_block_size_offset); 751 if (logical_block_size <= 0) 752 return (0); 753 754 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 755 if (volume_block <= SYSTEM_AREA_BLOCK+4) 756 return (0); 757 758 /* Location of Occurrence of Type L Path Table must be 759 * available location, 760 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 761 location = archive_le32dec(h+SVD_type_L_path_table_offset); 762 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 763 return (0); 764 765 /* The Type M Path Table must be at a valid location (WinISO 766 * and probably other programs omit this, so we allow zero) 767 * 768 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 769 location = archive_be32dec(h+SVD_type_M_path_table_offset); 770 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 771 || location >= volume_block) 772 return (0); 773 774 /* Read Root Directory Record in Volume Descriptor. */ 775 p = h + SVD_root_directory_record_offset; 776 if (p[DR_length_offset] != 34) 777 return (0); 778 779 return (48); 780 } 781 782 static int 783 isEVD(struct iso9660 *iso9660, const unsigned char *h) 784 { 785 const unsigned char *p; 786 ssize_t logical_block_size; 787 int32_t volume_block; 788 int32_t location; 789 790 (void)iso9660; /* UNUSED */ 791 792 /* Type of the Enhanced Volume Descriptor must be 2. */ 793 if (h[PVD_type_offset] != 2) 794 return (0); 795 796 /* EVD version must be 2. */ 797 if (h[PVD_version_offset] != 2) 798 return (0); 799 800 /* Reserved field must be 0. */ 801 if (h[PVD_reserved1_offset] != 0) 802 return (0); 803 804 /* Reserved field must be 0. */ 805 if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) 806 return (0); 807 808 /* Reserved field must be 0. */ 809 if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) 810 return (0); 811 812 /* Logical block size must be > 0. */ 813 /* I've looked at Ecma 119 and can't find any stronger 814 * restriction on this field. */ 815 logical_block_size = 816 archive_le16dec(h + PVD_logical_block_size_offset); 817 if (logical_block_size <= 0) 818 return (0); 819 820 volume_block = 821 archive_le32dec(h + PVD_volume_space_size_offset); 822 if (volume_block <= SYSTEM_AREA_BLOCK+4) 823 return (0); 824 825 /* File structure version must be 2 for ISO9660:1999. */ 826 if (h[PVD_file_structure_version_offset] != 2) 827 return (0); 828 829 /* Location of Occurrence of Type L Path Table must be 830 * available location, 831 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 832 location = archive_le32dec(h+PVD_type_1_path_table_offset); 833 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 834 return (0); 835 836 /* Location of Occurrence of Type M Path Table must be 837 * available location, 838 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 839 location = archive_be32dec(h+PVD_type_m_path_table_offset); 840 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 841 || location >= volume_block) 842 return (0); 843 844 /* Reserved field must be 0. */ 845 if (!isNull(iso9660, h, PVD_reserved4_offset, PVD_reserved4_size)) 846 return (0); 847 848 /* Reserved field must be 0. */ 849 if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) 850 return (0); 851 852 /* Read Root Directory Record in Volume Descriptor. */ 853 p = h + PVD_root_directory_record_offset; 854 if (p[DR_length_offset] != 34) 855 return (0); 856 857 return (48); 858 } 859 860 static int 861 isPVD(struct iso9660 *iso9660, const unsigned char *h) 862 { 863 const unsigned char *p; 864 ssize_t logical_block_size; 865 int32_t volume_block; 866 int32_t location; 867 int i; 868 869 /* Type of the Primary Volume Descriptor must be 1. */ 870 if (h[PVD_type_offset] != 1) 871 return (0); 872 873 /* PVD version must be 1. */ 874 if (h[PVD_version_offset] != 1) 875 return (0); 876 877 /* Reserved field must be 0. */ 878 if (h[PVD_reserved1_offset] != 0) 879 return (0); 880 881 /* Reserved field must be 0. */ 882 if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) 883 return (0); 884 885 /* Reserved field must be 0. */ 886 if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) 887 return (0); 888 889 /* Logical block size must be > 0. */ 890 /* I've looked at Ecma 119 and can't find any stronger 891 * restriction on this field. */ 892 logical_block_size = 893 archive_le16dec(h + PVD_logical_block_size_offset); 894 if (logical_block_size <= 0) 895 return (0); 896 897 volume_block = archive_le32dec(h + PVD_volume_space_size_offset); 898 if (volume_block <= SYSTEM_AREA_BLOCK+4) 899 return (0); 900 901 /* File structure version must be 1 for ISO9660/ECMA119. */ 902 if (h[PVD_file_structure_version_offset] != 1) 903 return (0); 904 905 /* Location of Occurrence of Type L Path Table must be 906 * available location, 907 * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 908 location = archive_le32dec(h+PVD_type_1_path_table_offset); 909 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 910 return (0); 911 912 /* The Type M Path Table must also be at a valid location 913 * (although ECMA 119 requires a Type M Path Table, WinISO and 914 * probably other programs omit it, so we permit a zero here) 915 * 916 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 917 location = archive_be32dec(h+PVD_type_m_path_table_offset); 918 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 919 || location >= volume_block) 920 return (0); 921 922 /* Reserved field must be 0. */ 923 /* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */ 924 for (i = 0; i < PVD_reserved4_size; ++i) 925 if (h[PVD_reserved4_offset + i] != 0 926 && h[PVD_reserved4_offset + i] != 0x20) 927 return (0); 928 929 /* Reserved field must be 0. */ 930 if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) 931 return (0); 932 933 /* XXX TODO: Check other values for sanity; reject more 934 * malformed PVDs. XXX */ 935 936 /* Read Root Directory Record in Volume Descriptor. */ 937 p = h + PVD_root_directory_record_offset; 938 if (p[DR_length_offset] != 34) 939 return (0); 940 941 if (!iso9660->primary.location) { 942 iso9660->logical_block_size = logical_block_size; 943 iso9660->volume_block = volume_block; 944 iso9660->volume_size = 945 logical_block_size * (uint64_t)volume_block; 946 iso9660->primary.location = 947 archive_le32dec(p + DR_extent_offset); 948 iso9660->primary.size = archive_le32dec(p + DR_size_offset); 949 } 950 951 return (48); 952 } 953 954 static int 955 read_children(struct archive_read *a, struct file_info *parent) 956 { 957 struct iso9660 *iso9660; 958 const unsigned char *b, *p; 959 struct file_info *multi; 960 size_t step, skip_size; 961 962 iso9660 = (struct iso9660 *)(a->format->data); 963 /* flush any remaining bytes from the last round to ensure 964 * we're positioned */ 965 if (iso9660->entry_bytes_unconsumed) { 966 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 967 iso9660->entry_bytes_unconsumed = 0; 968 } 969 if (iso9660->current_position > parent->offset) { 970 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 971 "Ignoring out-of-order directory (%s) %jd > %jd", 972 parent->name.s, 973 (intmax_t)iso9660->current_position, 974 (intmax_t)parent->offset); 975 return (ARCHIVE_WARN); 976 } 977 if (parent->offset + parent->size > iso9660->volume_size) { 978 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 979 "Directory is beyond end-of-media: %s", 980 parent->name.s); 981 return (ARCHIVE_WARN); 982 } 983 if (iso9660->current_position < parent->offset) { 984 int64_t skipsize; 985 986 skipsize = parent->offset - iso9660->current_position; 987 skipsize = __archive_read_consume(a, skipsize); 988 if (skipsize < 0) 989 return ((int)skipsize); 990 iso9660->current_position = parent->offset; 991 } 992 993 step = (size_t)(((parent->size + iso9660->logical_block_size -1) / 994 iso9660->logical_block_size) * iso9660->logical_block_size); 995 b = __archive_read_ahead(a, step, NULL); 996 if (b == NULL) { 997 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 998 "Failed to read full block when scanning " 999 "ISO9660 directory list"); 1000 return (ARCHIVE_FATAL); 1001 } 1002 iso9660->current_position += step; 1003 multi = NULL; 1004 skip_size = step; 1005 while (step) { 1006 p = b; 1007 b += iso9660->logical_block_size; 1008 step -= iso9660->logical_block_size; 1009 for (; *p != 0 && p < b && p + *p <= b; p += *p) { 1010 struct file_info *child; 1011 1012 /* N.B.: these special directory identifiers 1013 * are 8 bit "values" even on a 1014 * Joliet CD with UCS-2 (16bit) encoding. 1015 */ 1016 1017 /* Skip '.' entry. */ 1018 if (*(p + DR_name_len_offset) == 1 1019 && *(p + DR_name_offset) == '\0') 1020 continue; 1021 /* Skip '..' entry. */ 1022 if (*(p + DR_name_len_offset) == 1 1023 && *(p + DR_name_offset) == '\001') 1024 continue; 1025 child = parse_file_info(a, parent, p); 1026 if (child == NULL) { 1027 __archive_read_consume(a, skip_size); 1028 return (ARCHIVE_FATAL); 1029 } 1030 if (child->cl_offset == 0 && 1031 (child->multi_extent || multi != NULL)) { 1032 struct content *con; 1033 1034 if (multi == NULL) { 1035 multi = child; 1036 multi->contents.first = NULL; 1037 multi->contents.last = 1038 &(multi->contents.first); 1039 } 1040 con = malloc(sizeof(struct content)); 1041 if (con == NULL) { 1042 archive_set_error( 1043 &a->archive, ENOMEM, 1044 "No memory for multi extent"); 1045 __archive_read_consume(a, skip_size); 1046 return (ARCHIVE_FATAL); 1047 } 1048 con->offset = child->offset; 1049 con->size = child->size; 1050 con->next = NULL; 1051 *multi->contents.last = con; 1052 multi->contents.last = &(con->next); 1053 if (multi == child) { 1054 if (add_entry(a, iso9660, child) 1055 != ARCHIVE_OK) 1056 return (ARCHIVE_FATAL); 1057 } else { 1058 multi->size += child->size; 1059 if (!child->multi_extent) 1060 multi = NULL; 1061 } 1062 } else 1063 if (add_entry(a, iso9660, child) != ARCHIVE_OK) 1064 return (ARCHIVE_FATAL); 1065 } 1066 } 1067 1068 __archive_read_consume(a, skip_size); 1069 1070 /* Read data which recorded by RRIP "CE" extension. */ 1071 if (read_CE(a, iso9660) != ARCHIVE_OK) 1072 return (ARCHIVE_FATAL); 1073 1074 return (ARCHIVE_OK); 1075 } 1076 1077 static int 1078 choose_volume(struct archive_read *a, struct iso9660 *iso9660) 1079 { 1080 struct file_info *file; 1081 int64_t skipsize; 1082 struct vd *vd; 1083 const void *block; 1084 char seenJoliet; 1085 1086 vd = &(iso9660->primary); 1087 if (!iso9660->opt_support_joliet) 1088 iso9660->seenJoliet = 0; 1089 if (iso9660->seenJoliet && 1090 vd->location > iso9660->joliet.location) 1091 /* This condition is unlikely; by way of caution. */ 1092 vd = &(iso9660->joliet); 1093 1094 skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; 1095 skipsize = __archive_read_consume(a, skipsize); 1096 if (skipsize < 0) 1097 return ((int)skipsize); 1098 iso9660->current_position = skipsize; 1099 1100 block = __archive_read_ahead(a, vd->size, NULL); 1101 if (block == NULL) { 1102 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1103 "Failed to read full block when scanning " 1104 "ISO9660 directory list"); 1105 return (ARCHIVE_FATAL); 1106 } 1107 1108 /* 1109 * While reading Root Directory, flag seenJoliet must be zero to 1110 * avoid converting special name 0x00(Current Directory) and 1111 * next byte to UCS2. 1112 */ 1113 seenJoliet = iso9660->seenJoliet;/* Save flag. */ 1114 iso9660->seenJoliet = 0; 1115 file = parse_file_info(a, NULL, block); 1116 if (file == NULL) 1117 return (ARCHIVE_FATAL); 1118 iso9660->seenJoliet = seenJoliet; 1119 1120 /* 1121 * If the iso image has both RockRidge and Joliet, we preferentially 1122 * use RockRidge Extensions rather than Joliet ones. 1123 */ 1124 if (vd == &(iso9660->primary) && iso9660->seenRockridge 1125 && iso9660->seenJoliet) 1126 iso9660->seenJoliet = 0; 1127 1128 if (vd == &(iso9660->primary) && !iso9660->seenRockridge 1129 && iso9660->seenJoliet) { 1130 /* Switch reading data from primary to joliet. */ 1131 vd = &(iso9660->joliet); 1132 skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; 1133 skipsize -= iso9660->current_position; 1134 skipsize = __archive_read_consume(a, skipsize); 1135 if (skipsize < 0) 1136 return ((int)skipsize); 1137 iso9660->current_position += skipsize; 1138 1139 block = __archive_read_ahead(a, vd->size, NULL); 1140 if (block == NULL) { 1141 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1142 "Failed to read full block when scanning " 1143 "ISO9660 directory list"); 1144 return (ARCHIVE_FATAL); 1145 } 1146 iso9660->seenJoliet = 0; 1147 file = parse_file_info(a, NULL, block); 1148 if (file == NULL) 1149 return (ARCHIVE_FATAL); 1150 iso9660->seenJoliet = seenJoliet; 1151 } 1152 1153 /* Store the root directory in the pending list. */ 1154 if (add_entry(a, iso9660, file) != ARCHIVE_OK) 1155 return (ARCHIVE_FATAL); 1156 if (iso9660->seenRockridge) { 1157 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; 1158 a->archive.archive_format_name = 1159 "ISO9660 with Rockridge extensions"; 1160 } 1161 1162 return (ARCHIVE_OK); 1163 } 1164 1165 static int 1166 archive_read_format_iso9660_read_header(struct archive_read *a, 1167 struct archive_entry *entry) 1168 { 1169 struct iso9660 *iso9660; 1170 struct file_info *file; 1171 int r, rd_r = ARCHIVE_OK; 1172 1173 iso9660 = (struct iso9660 *)(a->format->data); 1174 1175 if (!a->archive.archive_format) { 1176 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; 1177 a->archive.archive_format_name = "ISO9660"; 1178 } 1179 1180 if (iso9660->current_position == 0) { 1181 r = choose_volume(a, iso9660); 1182 if (r != ARCHIVE_OK) 1183 return (r); 1184 } 1185 1186 file = NULL;/* Eliminate a warning. */ 1187 /* Get the next entry that appears after the current offset. */ 1188 r = next_entry_seek(a, iso9660, &file); 1189 if (r != ARCHIVE_OK) 1190 return (r); 1191 1192 if (iso9660->seenJoliet) { 1193 /* 1194 * Convert UTF-16BE of a filename to local locale MBS 1195 * and store the result into a filename field. 1196 */ 1197 if (iso9660->sconv_utf16be == NULL) { 1198 iso9660->sconv_utf16be = 1199 archive_string_conversion_from_charset( 1200 &(a->archive), "UTF-16BE", 1); 1201 if (iso9660->sconv_utf16be == NULL) 1202 /* Couldn't allocate memory */ 1203 return (ARCHIVE_FATAL); 1204 } 1205 if (iso9660->utf16be_path == NULL) { 1206 iso9660->utf16be_path = malloc(UTF16_NAME_MAX); 1207 if (iso9660->utf16be_path == NULL) { 1208 archive_set_error(&a->archive, ENOMEM, 1209 "No memory"); 1210 return (ARCHIVE_FATAL); 1211 } 1212 } 1213 if (iso9660->utf16be_previous_path == NULL) { 1214 iso9660->utf16be_previous_path = malloc(UTF16_NAME_MAX); 1215 if (iso9660->utf16be_previous_path == NULL) { 1216 archive_set_error(&a->archive, ENOMEM, 1217 "No memory"); 1218 return (ARCHIVE_FATAL); 1219 } 1220 } 1221 1222 iso9660->utf16be_path_len = 0; 1223 if (build_pathname_utf16be(iso9660->utf16be_path, 1224 UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) { 1225 archive_set_error(&a->archive, 1226 ARCHIVE_ERRNO_FILE_FORMAT, 1227 "Pathname is too long"); 1228 return (ARCHIVE_FATAL); 1229 } 1230 1231 r = archive_entry_copy_pathname_l(entry, 1232 (const char *)iso9660->utf16be_path, 1233 iso9660->utf16be_path_len, 1234 iso9660->sconv_utf16be); 1235 if (r != 0) { 1236 if (errno == ENOMEM) { 1237 archive_set_error(&a->archive, ENOMEM, 1238 "No memory for Pathname"); 1239 return (ARCHIVE_FATAL); 1240 } 1241 archive_set_error(&a->archive, 1242 ARCHIVE_ERRNO_FILE_FORMAT, 1243 "Pathname cannot be converted " 1244 "from %s to current locale.", 1245 archive_string_conversion_charset_name( 1246 iso9660->sconv_utf16be)); 1247 1248 rd_r = ARCHIVE_WARN; 1249 } 1250 } else { 1251 const char *path = build_pathname(&iso9660->pathname, file, 0); 1252 if (path == NULL) { 1253 archive_set_error(&a->archive, 1254 ARCHIVE_ERRNO_FILE_FORMAT, 1255 "Pathname is too long"); 1256 return (ARCHIVE_FATAL); 1257 } else { 1258 archive_string_empty(&iso9660->pathname); 1259 archive_entry_set_pathname(entry, path); 1260 } 1261 } 1262 1263 iso9660->entry_bytes_remaining = file->size; 1264 /* Offset for sparse-file-aware clients. */ 1265 iso9660->entry_sparse_offset = 0; 1266 1267 if (file->offset + file->size > iso9660->volume_size) { 1268 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1269 "File is beyond end-of-media: %s", 1270 archive_entry_pathname(entry)); 1271 iso9660->entry_bytes_remaining = 0; 1272 return (ARCHIVE_WARN); 1273 } 1274 1275 /* Set up the entry structure with information about this entry. */ 1276 archive_entry_set_mode(entry, file->mode); 1277 archive_entry_set_uid(entry, file->uid); 1278 archive_entry_set_gid(entry, file->gid); 1279 archive_entry_set_nlink(entry, file->nlinks); 1280 if (file->birthtime_is_set) 1281 archive_entry_set_birthtime(entry, file->birthtime, 0); 1282 else 1283 archive_entry_unset_birthtime(entry); 1284 archive_entry_set_mtime(entry, file->mtime, 0); 1285 archive_entry_set_ctime(entry, file->ctime, 0); 1286 archive_entry_set_atime(entry, file->atime, 0); 1287 /* N.B.: Rock Ridge supports 64-bit device numbers. */ 1288 archive_entry_set_rdev(entry, (dev_t)file->rdev); 1289 archive_entry_set_size(entry, iso9660->entry_bytes_remaining); 1290 if (file->symlink.s != NULL) 1291 archive_entry_copy_symlink(entry, file->symlink.s); 1292 1293 /* Note: If the input isn't seekable, we can't rewind to 1294 * return the same body again, so if the next entry refers to 1295 * the same data, we have to return it as a hardlink to the 1296 * original entry. */ 1297 if (file->number != -1 && 1298 file->number == iso9660->previous_number) { 1299 if (iso9660->seenJoliet) { 1300 r = archive_entry_copy_hardlink_l(entry, 1301 (const char *)iso9660->utf16be_previous_path, 1302 iso9660->utf16be_previous_path_len, 1303 iso9660->sconv_utf16be); 1304 if (r != 0) { 1305 if (errno == ENOMEM) { 1306 archive_set_error(&a->archive, ENOMEM, 1307 "No memory for Linkname"); 1308 return (ARCHIVE_FATAL); 1309 } 1310 archive_set_error(&a->archive, 1311 ARCHIVE_ERRNO_FILE_FORMAT, 1312 "Linkname cannot be converted " 1313 "from %s to current locale.", 1314 archive_string_conversion_charset_name( 1315 iso9660->sconv_utf16be)); 1316 rd_r = ARCHIVE_WARN; 1317 } 1318 } else 1319 archive_entry_set_hardlink(entry, 1320 iso9660->previous_pathname.s); 1321 archive_entry_unset_size(entry); 1322 iso9660->entry_bytes_remaining = 0; 1323 return (rd_r); 1324 } 1325 1326 if ((file->mode & AE_IFMT) != AE_IFDIR && 1327 file->offset < iso9660->current_position) { 1328 int64_t r64; 1329 1330 r64 = __archive_read_seek(a, file->offset, SEEK_SET); 1331 if (r64 != (int64_t)file->offset) { 1332 /* We can't seek backwards to extract it, so issue 1333 * a warning. Note that this can only happen if 1334 * this entry was added to the heap after we passed 1335 * this offset, that is, only if the directory 1336 * mentioning this entry is later than the body of 1337 * the entry. Such layouts are very unusual; most 1338 * ISO9660 writers lay out and record all directory 1339 * information first, then store all file bodies. */ 1340 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1341 "Ignoring out-of-order file @%jx (%s) %jd < %jd", 1342 (intmax_t)file->number, 1343 iso9660->pathname.s, 1344 (intmax_t)file->offset, 1345 (intmax_t)iso9660->current_position); 1346 iso9660->entry_bytes_remaining = 0; 1347 return (ARCHIVE_WARN); 1348 } 1349 iso9660->current_position = (uint64_t)r64; 1350 } 1351 1352 /* Initialize zisofs variables. */ 1353 iso9660->entry_zisofs.pz = file->pz; 1354 if (file->pz) { 1355 #ifdef HAVE_ZLIB_H 1356 struct zisofs *zisofs; 1357 1358 zisofs = &iso9660->entry_zisofs; 1359 zisofs->initialized = 0; 1360 zisofs->pz_log2_bs = file->pz_log2_bs; 1361 zisofs->pz_uncompressed_size = file->pz_uncompressed_size; 1362 zisofs->pz_offset = 0; 1363 zisofs->header_avail = 0; 1364 zisofs->header_passed = 0; 1365 zisofs->block_pointers_avail = 0; 1366 #endif 1367 archive_entry_set_size(entry, file->pz_uncompressed_size); 1368 } 1369 1370 iso9660->previous_number = file->number; 1371 if (iso9660->seenJoliet) { 1372 memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path, 1373 iso9660->utf16be_path_len); 1374 iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len; 1375 } else 1376 archive_strcpy( 1377 &iso9660->previous_pathname, iso9660->pathname.s); 1378 1379 /* Reset entry_bytes_remaining if the file is multi extent. */ 1380 iso9660->entry_content = file->contents.first; 1381 if (iso9660->entry_content != NULL) 1382 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1383 1384 if (archive_entry_filetype(entry) == AE_IFDIR) { 1385 /* Overwrite nlinks by proper link number which is 1386 * calculated from number of sub directories. */ 1387 archive_entry_set_nlink(entry, 2 + file->subdirs); 1388 /* Directory data has been read completely. */ 1389 iso9660->entry_bytes_remaining = 0; 1390 } 1391 1392 if (rd_r != ARCHIVE_OK) 1393 return (rd_r); 1394 return (ARCHIVE_OK); 1395 } 1396 1397 static int 1398 archive_read_format_iso9660_read_data_skip(struct archive_read *a) 1399 { 1400 /* Because read_next_header always does an explicit skip 1401 * to the next entry, we don't need to do anything here. */ 1402 (void)a; /* UNUSED */ 1403 return (ARCHIVE_OK); 1404 } 1405 1406 #ifdef HAVE_ZLIB_H 1407 1408 static int 1409 zisofs_read_data(struct archive_read *a, 1410 const void **buff, size_t *size, int64_t *offset) 1411 { 1412 struct iso9660 *iso9660; 1413 struct zisofs *zisofs; 1414 const unsigned char *p; 1415 size_t avail; 1416 ssize_t bytes_read; 1417 size_t uncompressed_size; 1418 int r; 1419 1420 iso9660 = (struct iso9660 *)(a->format->data); 1421 zisofs = &iso9660->entry_zisofs; 1422 1423 p = __archive_read_ahead(a, 1, &bytes_read); 1424 if (bytes_read <= 0) { 1425 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1426 "Truncated zisofs file body"); 1427 return (ARCHIVE_FATAL); 1428 } 1429 if (bytes_read > iso9660->entry_bytes_remaining) 1430 bytes_read = (ssize_t)iso9660->entry_bytes_remaining; 1431 avail = bytes_read; 1432 uncompressed_size = 0; 1433 1434 if (!zisofs->initialized) { 1435 size_t ceil, xsize; 1436 1437 /* Allocate block pointers buffer. */ 1438 ceil = (size_t)((zisofs->pz_uncompressed_size + 1439 (((int64_t)1) << zisofs->pz_log2_bs) - 1) 1440 >> zisofs->pz_log2_bs); 1441 xsize = (ceil + 1) * 4; 1442 if (zisofs->block_pointers_alloc < xsize) { 1443 size_t alloc; 1444 1445 if (zisofs->block_pointers != NULL) 1446 free(zisofs->block_pointers); 1447 alloc = ((xsize >> 10) + 1) << 10; 1448 zisofs->block_pointers = malloc(alloc); 1449 if (zisofs->block_pointers == NULL) { 1450 archive_set_error(&a->archive, ENOMEM, 1451 "No memory for zisofs decompression"); 1452 return (ARCHIVE_FATAL); 1453 } 1454 zisofs->block_pointers_alloc = alloc; 1455 } 1456 zisofs->block_pointers_size = xsize; 1457 1458 /* Allocate uncompressed data buffer. */ 1459 xsize = (size_t)1UL << zisofs->pz_log2_bs; 1460 if (zisofs->uncompressed_buffer_size < xsize) { 1461 if (zisofs->uncompressed_buffer != NULL) 1462 free(zisofs->uncompressed_buffer); 1463 zisofs->uncompressed_buffer = malloc(xsize); 1464 if (zisofs->uncompressed_buffer == NULL) { 1465 archive_set_error(&a->archive, ENOMEM, 1466 "No memory for zisofs decompression"); 1467 return (ARCHIVE_FATAL); 1468 } 1469 } 1470 zisofs->uncompressed_buffer_size = xsize; 1471 1472 /* 1473 * Read the file header, and check the magic code of zisofs. 1474 */ 1475 if (zisofs->header_avail < sizeof(zisofs->header)) { 1476 xsize = sizeof(zisofs->header) - zisofs->header_avail; 1477 if (avail < xsize) 1478 xsize = avail; 1479 memcpy(zisofs->header + zisofs->header_avail, p, xsize); 1480 zisofs->header_avail += xsize; 1481 avail -= xsize; 1482 p += xsize; 1483 } 1484 if (!zisofs->header_passed && 1485 zisofs->header_avail == sizeof(zisofs->header)) { 1486 int err = 0; 1487 1488 if (memcmp(zisofs->header, zisofs_magic, 1489 sizeof(zisofs_magic)) != 0) 1490 err = 1; 1491 if (archive_le32dec(zisofs->header + 8) 1492 != zisofs->pz_uncompressed_size) 1493 err = 1; 1494 if (zisofs->header[12] != 4) 1495 err = 1; 1496 if (zisofs->header[13] != zisofs->pz_log2_bs) 1497 err = 1; 1498 if (err) { 1499 archive_set_error(&a->archive, 1500 ARCHIVE_ERRNO_FILE_FORMAT, 1501 "Illegal zisofs file body"); 1502 return (ARCHIVE_FATAL); 1503 } 1504 zisofs->header_passed = 1; 1505 } 1506 /* 1507 * Read block pointers. 1508 */ 1509 if (zisofs->header_passed && 1510 zisofs->block_pointers_avail < zisofs->block_pointers_size) { 1511 xsize = zisofs->block_pointers_size 1512 - zisofs->block_pointers_avail; 1513 if (avail < xsize) 1514 xsize = avail; 1515 memcpy(zisofs->block_pointers 1516 + zisofs->block_pointers_avail, p, xsize); 1517 zisofs->block_pointers_avail += xsize; 1518 avail -= xsize; 1519 p += xsize; 1520 if (zisofs->block_pointers_avail 1521 == zisofs->block_pointers_size) { 1522 /* We've got all block pointers and initialize 1523 * related variables. */ 1524 zisofs->block_off = 0; 1525 zisofs->block_avail = 0; 1526 /* Complete a initialization */ 1527 zisofs->initialized = 1; 1528 } 1529 } 1530 1531 if (!zisofs->initialized) 1532 goto next_data; /* We need more data. */ 1533 } 1534 1535 /* 1536 * Get block offsets from block pointers. 1537 */ 1538 if (zisofs->block_avail == 0) { 1539 uint32_t bst, bed; 1540 1541 if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { 1542 /* There isn't a pair of offsets. */ 1543 archive_set_error(&a->archive, 1544 ARCHIVE_ERRNO_FILE_FORMAT, 1545 "Illegal zisofs block pointers"); 1546 return (ARCHIVE_FATAL); 1547 } 1548 bst = archive_le32dec( 1549 zisofs->block_pointers + zisofs->block_off); 1550 if (bst != zisofs->pz_offset + (bytes_read - avail)) { 1551 /* TODO: Should we seek offset of current file 1552 * by bst ? */ 1553 archive_set_error(&a->archive, 1554 ARCHIVE_ERRNO_FILE_FORMAT, 1555 "Illegal zisofs block pointers(cannot seek)"); 1556 return (ARCHIVE_FATAL); 1557 } 1558 bed = archive_le32dec( 1559 zisofs->block_pointers + zisofs->block_off + 4); 1560 if (bed < bst) { 1561 archive_set_error(&a->archive, 1562 ARCHIVE_ERRNO_FILE_FORMAT, 1563 "Illegal zisofs block pointers"); 1564 return (ARCHIVE_FATAL); 1565 } 1566 zisofs->block_avail = bed - bst; 1567 zisofs->block_off += 4; 1568 1569 /* Initialize compression library for new block. */ 1570 if (zisofs->stream_valid) 1571 r = inflateReset(&zisofs->stream); 1572 else 1573 r = inflateInit(&zisofs->stream); 1574 if (r != Z_OK) { 1575 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1576 "Can't initialize zisofs decompression."); 1577 return (ARCHIVE_FATAL); 1578 } 1579 zisofs->stream_valid = 1; 1580 zisofs->stream.total_in = 0; 1581 zisofs->stream.total_out = 0; 1582 } 1583 1584 /* 1585 * Make uncompressed data. 1586 */ 1587 if (zisofs->block_avail == 0) { 1588 memset(zisofs->uncompressed_buffer, 0, 1589 zisofs->uncompressed_buffer_size); 1590 uncompressed_size = zisofs->uncompressed_buffer_size; 1591 } else { 1592 zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; 1593 if (avail > zisofs->block_avail) 1594 zisofs->stream.avail_in = zisofs->block_avail; 1595 else 1596 zisofs->stream.avail_in = (uInt)avail; 1597 zisofs->stream.next_out = zisofs->uncompressed_buffer; 1598 zisofs->stream.avail_out = 1599 (uInt)zisofs->uncompressed_buffer_size; 1600 1601 r = inflate(&zisofs->stream, 0); 1602 switch (r) { 1603 case Z_OK: /* Decompressor made some progress.*/ 1604 case Z_STREAM_END: /* Found end of stream. */ 1605 break; 1606 default: 1607 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1608 "zisofs decompression failed (%d)", r); 1609 return (ARCHIVE_FATAL); 1610 } 1611 uncompressed_size = 1612 zisofs->uncompressed_buffer_size - zisofs->stream.avail_out; 1613 avail -= zisofs->stream.next_in - p; 1614 zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p); 1615 } 1616 next_data: 1617 bytes_read -= avail; 1618 *buff = zisofs->uncompressed_buffer; 1619 *size = uncompressed_size; 1620 *offset = iso9660->entry_sparse_offset; 1621 iso9660->entry_sparse_offset += uncompressed_size; 1622 iso9660->entry_bytes_remaining -= bytes_read; 1623 iso9660->current_position += bytes_read; 1624 zisofs->pz_offset += (uint32_t)bytes_read; 1625 iso9660->entry_bytes_unconsumed += bytes_read; 1626 1627 return (ARCHIVE_OK); 1628 } 1629 1630 #else /* HAVE_ZLIB_H */ 1631 1632 static int 1633 zisofs_read_data(struct archive_read *a, 1634 const void **buff, size_t *size, int64_t *offset) 1635 { 1636 1637 (void)buff;/* UNUSED */ 1638 (void)size;/* UNUSED */ 1639 (void)offset;/* UNUSED */ 1640 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1641 "zisofs is not supported on this platform."); 1642 return (ARCHIVE_FAILED); 1643 } 1644 1645 #endif /* HAVE_ZLIB_H */ 1646 1647 static int 1648 archive_read_format_iso9660_read_data(struct archive_read *a, 1649 const void **buff, size_t *size, int64_t *offset) 1650 { 1651 ssize_t bytes_read; 1652 struct iso9660 *iso9660; 1653 1654 iso9660 = (struct iso9660 *)(a->format->data); 1655 1656 if (iso9660->entry_bytes_unconsumed) { 1657 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 1658 iso9660->entry_bytes_unconsumed = 0; 1659 } 1660 1661 if (iso9660->entry_bytes_remaining <= 0) { 1662 if (iso9660->entry_content != NULL) 1663 iso9660->entry_content = iso9660->entry_content->next; 1664 if (iso9660->entry_content == NULL) { 1665 *buff = NULL; 1666 *size = 0; 1667 *offset = iso9660->entry_sparse_offset; 1668 return (ARCHIVE_EOF); 1669 } 1670 /* Seek forward to the start of the entry. */ 1671 if (iso9660->current_position < iso9660->entry_content->offset) { 1672 int64_t step; 1673 1674 step = iso9660->entry_content->offset - 1675 iso9660->current_position; 1676 step = __archive_read_consume(a, step); 1677 if (step < 0) 1678 return ((int)step); 1679 iso9660->current_position = 1680 iso9660->entry_content->offset; 1681 } 1682 if (iso9660->entry_content->offset < iso9660->current_position) { 1683 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1684 "Ignoring out-of-order file (%s) %jd < %jd", 1685 iso9660->pathname.s, 1686 (intmax_t)iso9660->entry_content->offset, 1687 (intmax_t)iso9660->current_position); 1688 *buff = NULL; 1689 *size = 0; 1690 *offset = iso9660->entry_sparse_offset; 1691 return (ARCHIVE_WARN); 1692 } 1693 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1694 } 1695 if (iso9660->entry_zisofs.pz) 1696 return (zisofs_read_data(a, buff, size, offset)); 1697 1698 *buff = __archive_read_ahead(a, 1, &bytes_read); 1699 if (bytes_read == 0) 1700 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1701 "Truncated input file"); 1702 if (*buff == NULL) 1703 return (ARCHIVE_FATAL); 1704 if (bytes_read > iso9660->entry_bytes_remaining) 1705 bytes_read = (ssize_t)iso9660->entry_bytes_remaining; 1706 *size = bytes_read; 1707 *offset = iso9660->entry_sparse_offset; 1708 iso9660->entry_sparse_offset += bytes_read; 1709 iso9660->entry_bytes_remaining -= bytes_read; 1710 iso9660->entry_bytes_unconsumed = bytes_read; 1711 iso9660->current_position += bytes_read; 1712 return (ARCHIVE_OK); 1713 } 1714 1715 static int 1716 archive_read_format_iso9660_cleanup(struct archive_read *a) 1717 { 1718 struct iso9660 *iso9660; 1719 int r = ARCHIVE_OK; 1720 1721 iso9660 = (struct iso9660 *)(a->format->data); 1722 release_files(iso9660); 1723 free(iso9660->read_ce_req.reqs); 1724 archive_string_free(&iso9660->pathname); 1725 archive_string_free(&iso9660->previous_pathname); 1726 if (iso9660->pending_files.files) 1727 free(iso9660->pending_files.files); 1728 #ifdef HAVE_ZLIB_H 1729 free(iso9660->entry_zisofs.uncompressed_buffer); 1730 free(iso9660->entry_zisofs.block_pointers); 1731 if (iso9660->entry_zisofs.stream_valid) { 1732 if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) { 1733 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1734 "Failed to clean up zlib decompressor"); 1735 r = ARCHIVE_FATAL; 1736 } 1737 } 1738 #endif 1739 free(iso9660->utf16be_path); 1740 free(iso9660->utf16be_previous_path); 1741 free(iso9660); 1742 (a->format->data) = NULL; 1743 return (r); 1744 } 1745 1746 /* 1747 * This routine parses a single ISO directory record, makes sense 1748 * of any extensions, and stores the result in memory. 1749 */ 1750 static struct file_info * 1751 parse_file_info(struct archive_read *a, struct file_info *parent, 1752 const unsigned char *isodirrec) 1753 { 1754 struct iso9660 *iso9660; 1755 struct file_info *file, *filep; 1756 size_t name_len; 1757 const unsigned char *rr_start, *rr_end; 1758 const unsigned char *p; 1759 size_t dr_len; 1760 uint64_t fsize, offset; 1761 int32_t location; 1762 int flags; 1763 1764 iso9660 = (struct iso9660 *)(a->format->data); 1765 1766 dr_len = (size_t)isodirrec[DR_length_offset]; 1767 name_len = (size_t)isodirrec[DR_name_len_offset]; 1768 location = archive_le32dec(isodirrec + DR_extent_offset); 1769 fsize = toi(isodirrec + DR_size_offset, DR_size_size); 1770 /* Sanity check that dr_len needs at least 34. */ 1771 if (dr_len < 34) { 1772 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1773 "Invalid length of directory record"); 1774 return (NULL); 1775 } 1776 /* Sanity check that name_len doesn't exceed dr_len. */ 1777 if (dr_len - 33 < name_len || name_len == 0) { 1778 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1779 "Invalid length of file identifier"); 1780 return (NULL); 1781 } 1782 /* Sanity check that location doesn't exceed volume block. 1783 * Don't check lower limit of location; it's possibility 1784 * the location has negative value when file type is symbolic 1785 * link or file size is zero. As far as I know latest mkisofs 1786 * do that. 1787 */ 1788 if (location > 0 && 1789 (location + ((fsize + iso9660->logical_block_size -1) 1790 / iso9660->logical_block_size)) 1791 > (uint32_t)iso9660->volume_block) { 1792 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1793 "Invalid location of extent of file"); 1794 return (NULL); 1795 } 1796 /* Sanity check that location doesn't have a negative value 1797 * when the file is not empty. it's too large. */ 1798 if (fsize != 0 && location < 0) { 1799 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1800 "Invalid location of extent of file"); 1801 return (NULL); 1802 } 1803 1804 /* Sanity check that this entry does not create a cycle. */ 1805 offset = iso9660->logical_block_size * (uint64_t)location; 1806 for (filep = parent; filep != NULL; filep = filep->parent) { 1807 if (filep->offset == offset) { 1808 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1809 "Directory structure contains loop"); 1810 return (NULL); 1811 } 1812 } 1813 1814 /* Create a new file entry and copy data from the ISO dir record. */ 1815 file = (struct file_info *)calloc(1, sizeof(*file)); 1816 if (file == NULL) { 1817 archive_set_error(&a->archive, ENOMEM, 1818 "No memory for file entry"); 1819 return (NULL); 1820 } 1821 file->parent = parent; 1822 file->offset = offset; 1823 file->size = fsize; 1824 file->mtime = isodate7(isodirrec + DR_date_offset); 1825 file->ctime = file->atime = file->mtime; 1826 file->rede_files.first = NULL; 1827 file->rede_files.last = &(file->rede_files.first); 1828 1829 p = isodirrec + DR_name_offset; 1830 /* Rockridge extensions (if any) follow name. Compute this 1831 * before fidgeting the name_len below. */ 1832 rr_start = p + name_len + (name_len & 1 ? 0 : 1); 1833 rr_end = isodirrec + dr_len; 1834 1835 if (iso9660->seenJoliet) { 1836 /* Joliet names are max 64 chars (128 bytes) according to spec, 1837 * but genisoimage/mkisofs allows recording longer Joliet 1838 * names which are 103 UCS2 characters(206 bytes) by their 1839 * option '-joliet-long'. 1840 */ 1841 if (name_len > 206) 1842 name_len = 206; 1843 name_len &= ~1; 1844 1845 /* trim trailing first version and dot from filename. 1846 * 1847 * Remember we were in UTF-16BE land! 1848 * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both 1849 * 16 bits big endian characters on Joliet. 1850 * 1851 * TODO: sanitize filename? 1852 * Joliet allows any UCS-2 char except: 1853 * *, /, :, ;, ? and \. 1854 */ 1855 /* Chop off trailing ';1' from files. */ 1856 if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';' 1857 && p[name_len-2] == 0 && p[name_len-1] == '1') 1858 name_len -= 4; 1859 #if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */ 1860 /* Chop off trailing '.' from filenames. */ 1861 if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.') 1862 name_len -= 2; 1863 #endif 1864 if ((file->utf16be_name = malloc(name_len)) == NULL) { 1865 archive_set_error(&a->archive, ENOMEM, 1866 "No memory for file name"); 1867 goto fail; 1868 } 1869 memcpy(file->utf16be_name, p, name_len); 1870 file->utf16be_bytes = name_len; 1871 } else { 1872 /* Chop off trailing ';1' from files. */ 1873 if (name_len > 2 && p[name_len - 2] == ';' && 1874 p[name_len - 1] == '1') 1875 name_len -= 2; 1876 /* Chop off trailing '.' from filenames. */ 1877 if (name_len > 1 && p[name_len - 1] == '.') 1878 --name_len; 1879 1880 archive_strncpy(&file->name, (const char *)p, name_len); 1881 } 1882 1883 flags = isodirrec[DR_flags_offset]; 1884 if (flags & 0x02) 1885 file->mode = AE_IFDIR | 0700; 1886 else 1887 file->mode = AE_IFREG | 0400; 1888 if (flags & 0x80) 1889 file->multi_extent = 1; 1890 else 1891 file->multi_extent = 0; 1892 /* 1893 * Use a location for the file number, which is treated as an inode 1894 * number to find out hardlink target. If Rockridge extensions is 1895 * being used, the file number will be overwritten by FILE SERIAL 1896 * NUMBER of RRIP "PX" extension. 1897 * Note: Old mkisofs did not record that FILE SERIAL NUMBER 1898 * in ISO images. 1899 * Note2: xorriso set 0 to the location of a symlink file. 1900 */ 1901 if (file->size == 0 && location >= 0) { 1902 /* If file->size is zero, its location points wrong place, 1903 * and so we should not use it for the file number. 1904 * When the location has negative value, it can be used 1905 * for the file number. 1906 */ 1907 file->number = -1; 1908 /* Do not appear before any directory entries. */ 1909 file->offset = -1; 1910 } else 1911 file->number = (int64_t)(uint32_t)location; 1912 1913 /* Rockridge extensions overwrite information from above. */ 1914 if (iso9660->opt_support_rockridge) { 1915 if (parent == NULL && rr_end - rr_start >= 7) { 1916 p = rr_start; 1917 if (memcmp(p, "SP\x07\x01\xbe\xef", 6) == 0) { 1918 /* 1919 * SP extension stores the suspOffset 1920 * (Number of bytes to skip between 1921 * filename and SUSP records.) 1922 * It is mandatory by the SUSP standard 1923 * (IEEE 1281). 1924 * 1925 * It allows SUSP to coexist with 1926 * non-SUSP uses of the System 1927 * Use Area by placing non-SUSP data 1928 * before SUSP data. 1929 * 1930 * SP extension must be in the root 1931 * directory entry, disable all SUSP 1932 * processing if not found. 1933 */ 1934 iso9660->suspOffset = p[6]; 1935 iso9660->seenSUSP = 1; 1936 rr_start += 7; 1937 } 1938 } 1939 if (iso9660->seenSUSP) { 1940 int r; 1941 1942 file->name_continues = 0; 1943 file->symlink_continues = 0; 1944 rr_start += iso9660->suspOffset; 1945 r = parse_rockridge(a, file, rr_start, rr_end); 1946 if (r != ARCHIVE_OK) 1947 goto fail; 1948 /* 1949 * A file size of symbolic link files in ISO images 1950 * made by makefs is not zero and its location is 1951 * the same as those of next regular file. That is 1952 * the same as hard like file and it causes unexpected 1953 * error. 1954 */ 1955 if (file->size > 0 && 1956 (file->mode & AE_IFMT) == AE_IFLNK) { 1957 file->size = 0; 1958 file->number = -1; 1959 file->offset = -1; 1960 } 1961 } else 1962 /* If there isn't SUSP, disable parsing 1963 * rock ridge extensions. */ 1964 iso9660->opt_support_rockridge = 0; 1965 } 1966 1967 file->nlinks = 1;/* Reset nlink. we'll calculate it later. */ 1968 /* Tell file's parent how many children that parent has. */ 1969 if (parent != NULL && (flags & 0x02)) 1970 parent->subdirs++; 1971 1972 if (iso9660->seenRockridge) { 1973 if (parent != NULL && parent->parent == NULL && 1974 (flags & 0x02) && iso9660->rr_moved == NULL && 1975 file->name.s && 1976 (strcmp(file->name.s, "rr_moved") == 0 || 1977 strcmp(file->name.s, ".rr_moved") == 0)) { 1978 iso9660->rr_moved = file; 1979 file->rr_moved = 1; 1980 file->rr_moved_has_re_only = 1; 1981 file->re = 0; 1982 parent->subdirs--; 1983 } else if (file->re) { 1984 /* 1985 * Sanity check: file's parent is rr_moved. 1986 */ 1987 if (parent == NULL || parent->rr_moved == 0) { 1988 archive_set_error(&a->archive, 1989 ARCHIVE_ERRNO_MISC, 1990 "Invalid Rockridge RE"); 1991 goto fail; 1992 } 1993 /* 1994 * Sanity check: file does not have "CL" extension. 1995 */ 1996 if (file->cl_offset) { 1997 archive_set_error(&a->archive, 1998 ARCHIVE_ERRNO_MISC, 1999 "Invalid Rockridge RE and CL"); 2000 goto fail; 2001 } 2002 /* 2003 * Sanity check: The file type must be a directory. 2004 */ 2005 if ((flags & 0x02) == 0) { 2006 archive_set_error(&a->archive, 2007 ARCHIVE_ERRNO_MISC, 2008 "Invalid Rockridge RE"); 2009 goto fail; 2010 } 2011 } else if (parent != NULL && parent->rr_moved) 2012 file->rr_moved_has_re_only = 0; 2013 else if (parent != NULL && (flags & 0x02) && 2014 (parent->re || parent->re_descendant)) 2015 file->re_descendant = 1; 2016 if (file->cl_offset) { 2017 struct file_info *r; 2018 2019 if (parent == NULL || parent->parent == NULL) { 2020 archive_set_error(&a->archive, 2021 ARCHIVE_ERRNO_MISC, 2022 "Invalid Rockridge CL"); 2023 goto fail; 2024 } 2025 /* 2026 * Sanity check: The file type must be a regular file. 2027 */ 2028 if ((flags & 0x02) != 0) { 2029 archive_set_error(&a->archive, 2030 ARCHIVE_ERRNO_MISC, 2031 "Invalid Rockridge CL"); 2032 goto fail; 2033 } 2034 parent->subdirs++; 2035 /* Overwrite an offset and a number of this "CL" entry 2036 * to appear before other dirs. "+1" to those is to 2037 * make sure to appear after "RE" entry which this 2038 * "CL" entry should be connected with. */ 2039 file->offset = file->number = file->cl_offset + 1; 2040 2041 /* 2042 * Sanity check: cl_offset does not point at its 2043 * the parents or itself. 2044 */ 2045 for (r = parent; r; r = r->parent) { 2046 if (r->offset == file->cl_offset) { 2047 archive_set_error(&a->archive, 2048 ARCHIVE_ERRNO_MISC, 2049 "Invalid Rockridge CL"); 2050 goto fail; 2051 } 2052 } 2053 if (file->cl_offset == file->offset || 2054 parent->rr_moved) { 2055 archive_set_error(&a->archive, 2056 ARCHIVE_ERRNO_MISC, 2057 "Invalid Rockridge CL"); 2058 goto fail; 2059 } 2060 } 2061 } 2062 2063 #if DEBUG 2064 /* DEBUGGING: Warn about attributes I don't yet fully support. */ 2065 if ((flags & ~0x02) != 0) { 2066 fprintf(stderr, "\n ** Unrecognized flag: "); 2067 dump_isodirrec(stderr, isodirrec); 2068 fprintf(stderr, "\n"); 2069 } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { 2070 fprintf(stderr, "\n ** Unrecognized sequence number: "); 2071 dump_isodirrec(stderr, isodirrec); 2072 fprintf(stderr, "\n"); 2073 } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { 2074 fprintf(stderr, "\n ** Unexpected file unit size: "); 2075 dump_isodirrec(stderr, isodirrec); 2076 fprintf(stderr, "\n"); 2077 } else if (*(isodirrec + DR_interleave_offset) != 0) { 2078 fprintf(stderr, "\n ** Unexpected interleave: "); 2079 dump_isodirrec(stderr, isodirrec); 2080 fprintf(stderr, "\n"); 2081 } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { 2082 fprintf(stderr, "\n ** Unexpected extended attribute length: "); 2083 dump_isodirrec(stderr, isodirrec); 2084 fprintf(stderr, "\n"); 2085 } 2086 #endif 2087 register_file(iso9660, file); 2088 return (file); 2089 fail: 2090 archive_string_free(&file->name); 2091 free(file); 2092 return (NULL); 2093 } 2094 2095 static int 2096 parse_rockridge(struct archive_read *a, struct file_info *file, 2097 const unsigned char *p, const unsigned char *end) 2098 { 2099 struct iso9660 *iso9660; 2100 2101 iso9660 = (struct iso9660 *)(a->format->data); 2102 2103 while (p + 4 <= end /* Enough space for another entry. */ 2104 && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ 2105 && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ 2106 && p[2] >= 4 /* Sanity-check length. */ 2107 && p + p[2] <= end) { /* Sanity-check length. */ 2108 const unsigned char *data = p + 4; 2109 int data_length = p[2] - 4; 2110 int version = p[3]; 2111 2112 switch(p[0]) { 2113 case 'C': 2114 if (p[1] == 'E') { 2115 if (version == 1 && data_length == 24) { 2116 /* 2117 * CE extension comprises: 2118 * 8 byte sector containing extension 2119 * 8 byte offset w/in above sector 2120 * 8 byte length of continuation 2121 */ 2122 int32_t location = 2123 archive_le32dec(data); 2124 file->ce_offset = 2125 archive_le32dec(data+8); 2126 file->ce_size = 2127 archive_le32dec(data+16); 2128 if (register_CE(a, location, file) 2129 != ARCHIVE_OK) 2130 return (ARCHIVE_FATAL); 2131 } 2132 } 2133 else if (p[1] == 'L') { 2134 if (version == 1 && data_length == 8) { 2135 file->cl_offset = (uint64_t) 2136 iso9660->logical_block_size * 2137 (uint64_t)archive_le32dec(data); 2138 iso9660->seenRockridge = 1; 2139 } 2140 } 2141 break; 2142 case 'N': 2143 if (p[1] == 'M') { 2144 if (version == 1) { 2145 parse_rockridge_NM1(file, 2146 data, data_length); 2147 iso9660->seenRockridge = 1; 2148 } 2149 } 2150 break; 2151 case 'P': 2152 /* 2153 * PD extension is padding; 2154 * contents are always ignored. 2155 * 2156 * PL extension won't appear; 2157 * contents are always ignored. 2158 */ 2159 if (p[1] == 'N') { 2160 if (version == 1 && data_length == 16) { 2161 file->rdev = toi(data,4); 2162 file->rdev <<= 32; 2163 file->rdev |= toi(data + 8, 4); 2164 iso9660->seenRockridge = 1; 2165 } 2166 } 2167 else if (p[1] == 'X') { 2168 /* 2169 * PX extension comprises: 2170 * 8 bytes for mode, 2171 * 8 bytes for nlinks, 2172 * 8 bytes for uid, 2173 * 8 bytes for gid, 2174 * 8 bytes for inode. 2175 */ 2176 if (version == 1) { 2177 if (data_length >= 8) 2178 file->mode 2179 = toi(data, 4); 2180 if (data_length >= 16) 2181 file->nlinks 2182 = toi(data + 8, 4); 2183 if (data_length >= 24) 2184 file->uid 2185 = toi(data + 16, 4); 2186 if (data_length >= 32) 2187 file->gid 2188 = toi(data + 24, 4); 2189 if (data_length >= 40) 2190 file->number 2191 = toi(data + 32, 4); 2192 iso9660->seenRockridge = 1; 2193 } 2194 } 2195 break; 2196 case 'R': 2197 if (p[1] == 'E' && version == 1) { 2198 file->re = 1; 2199 iso9660->seenRockridge = 1; 2200 } 2201 else if (p[1] == 'R' && version == 1) { 2202 /* 2203 * RR extension comprises: 2204 * one byte flag value 2205 * This extension is obsolete, 2206 * so contents are always ignored. 2207 */ 2208 } 2209 break; 2210 case 'S': 2211 if (p[1] == 'L') { 2212 if (version == 1) { 2213 parse_rockridge_SL1(file, 2214 data, data_length); 2215 iso9660->seenRockridge = 1; 2216 } 2217 } 2218 else if (p[1] == 'T' 2219 && data_length == 0 && version == 1) { 2220 /* 2221 * ST extension marks end of this 2222 * block of SUSP entries. 2223 * 2224 * It allows SUSP to coexist with 2225 * non-SUSP uses of the System 2226 * Use Area by placing non-SUSP data 2227 * after SUSP data. 2228 */ 2229 iso9660->seenSUSP = 0; 2230 iso9660->seenRockridge = 0; 2231 return (ARCHIVE_OK); 2232 } 2233 break; 2234 case 'T': 2235 if (p[1] == 'F') { 2236 if (version == 1) { 2237 parse_rockridge_TF1(file, 2238 data, data_length); 2239 iso9660->seenRockridge = 1; 2240 } 2241 } 2242 break; 2243 case 'Z': 2244 if (p[1] == 'F') { 2245 if (version == 1) 2246 parse_rockridge_ZF1(file, 2247 data, data_length); 2248 } 2249 break; 2250 default: 2251 break; 2252 } 2253 2254 p += p[2]; 2255 } 2256 return (ARCHIVE_OK); 2257 } 2258 2259 static int 2260 register_CE(struct archive_read *a, int32_t location, 2261 struct file_info *file) 2262 { 2263 struct iso9660 *iso9660; 2264 struct read_ce_queue *heap; 2265 struct read_ce_req *p; 2266 uint64_t offset, parent_offset; 2267 int hole, parent; 2268 2269 iso9660 = (struct iso9660 *)(a->format->data); 2270 offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size; 2271 if (((file->mode & AE_IFMT) == AE_IFREG && 2272 offset >= file->offset) || 2273 offset < iso9660->current_position || 2274 (((uint64_t)file->ce_offset) + file->ce_size) 2275 > (uint64_t)iso9660->logical_block_size || 2276 offset + file->ce_offset + file->ce_size 2277 > iso9660->volume_size) { 2278 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2279 "Invalid parameter in SUSP \"CE\" extension"); 2280 return (ARCHIVE_FATAL); 2281 } 2282 2283 /* Expand our CE list as necessary. */ 2284 heap = &(iso9660->read_ce_req); 2285 if (heap->cnt >= heap->allocated) { 2286 int new_size; 2287 2288 if (heap->allocated < 16) 2289 new_size = 16; 2290 else 2291 new_size = heap->allocated * 2; 2292 /* Overflow might keep us from growing the list. */ 2293 if (new_size <= heap->allocated) { 2294 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2295 return (ARCHIVE_FATAL); 2296 } 2297 p = calloc(new_size, sizeof(p[0])); 2298 if (p == NULL) { 2299 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2300 return (ARCHIVE_FATAL); 2301 } 2302 if (heap->reqs != NULL) { 2303 memcpy(p, heap->reqs, heap->cnt * sizeof(*p)); 2304 free(heap->reqs); 2305 } 2306 heap->reqs = p; 2307 heap->allocated = new_size; 2308 } 2309 2310 /* 2311 * Start with hole at end, walk it up tree to find insertion point. 2312 */ 2313 hole = heap->cnt++; 2314 while (hole > 0) { 2315 parent = (hole - 1)/2; 2316 parent_offset = heap->reqs[parent].offset; 2317 if (offset >= parent_offset) { 2318 heap->reqs[hole].offset = offset; 2319 heap->reqs[hole].file = file; 2320 return (ARCHIVE_OK); 2321 } 2322 /* Move parent into hole <==> move hole up tree. */ 2323 heap->reqs[hole] = heap->reqs[parent]; 2324 hole = parent; 2325 } 2326 heap->reqs[0].offset = offset; 2327 heap->reqs[0].file = file; 2328 return (ARCHIVE_OK); 2329 } 2330 2331 static void 2332 next_CE(struct read_ce_queue *heap) 2333 { 2334 uint64_t a_offset, b_offset, c_offset; 2335 int a, b, c; 2336 struct read_ce_req tmp; 2337 2338 if (heap->cnt < 1) 2339 return; 2340 2341 /* 2342 * Move the last item in the heap to the root of the tree 2343 */ 2344 heap->reqs[0] = heap->reqs[--(heap->cnt)]; 2345 2346 /* 2347 * Rebalance the heap. 2348 */ 2349 a = 0; /* Starting element and its offset */ 2350 a_offset = heap->reqs[a].offset; 2351 for (;;) { 2352 b = a + a + 1; /* First child */ 2353 if (b >= heap->cnt) 2354 return; 2355 b_offset = heap->reqs[b].offset; 2356 c = b + 1; /* Use second child if it is smaller. */ 2357 if (c < heap->cnt) { 2358 c_offset = heap->reqs[c].offset; 2359 if (c_offset < b_offset) { 2360 b = c; 2361 b_offset = c_offset; 2362 } 2363 } 2364 if (a_offset <= b_offset) 2365 return; 2366 tmp = heap->reqs[a]; 2367 heap->reqs[a] = heap->reqs[b]; 2368 heap->reqs[b] = tmp; 2369 a = b; 2370 } 2371 } 2372 2373 2374 static int 2375 read_CE(struct archive_read *a, struct iso9660 *iso9660) 2376 { 2377 struct read_ce_queue *heap; 2378 const unsigned char *b, *p, *end; 2379 struct file_info *file; 2380 size_t step; 2381 int r; 2382 2383 /* Read data which RRIP "CE" extension points. */ 2384 heap = &(iso9660->read_ce_req); 2385 step = iso9660->logical_block_size; 2386 while (heap->cnt && 2387 heap->reqs[0].offset == iso9660->current_position) { 2388 b = __archive_read_ahead(a, step, NULL); 2389 if (b == NULL) { 2390 archive_set_error(&a->archive, 2391 ARCHIVE_ERRNO_MISC, 2392 "Failed to read full block when scanning " 2393 "ISO9660 directory list"); 2394 return (ARCHIVE_FATAL); 2395 } 2396 do { 2397 file = heap->reqs[0].file; 2398 if (file->ce_offset + file->ce_size > step) { 2399 archive_set_error(&a->archive, 2400 ARCHIVE_ERRNO_FILE_FORMAT, 2401 "Malformed CE information"); 2402 return (ARCHIVE_FATAL); 2403 } 2404 p = b + file->ce_offset; 2405 end = p + file->ce_size; 2406 next_CE(heap); 2407 r = parse_rockridge(a, file, p, end); 2408 if (r != ARCHIVE_OK) 2409 return (ARCHIVE_FATAL); 2410 } while (heap->cnt && 2411 heap->reqs[0].offset == iso9660->current_position); 2412 /* NOTE: Do not move this consume's code to front of 2413 * do-while loop. Registration of nested CE extension 2414 * might cause error because of current position. */ 2415 __archive_read_consume(a, step); 2416 iso9660->current_position += step; 2417 } 2418 return (ARCHIVE_OK); 2419 } 2420 2421 static void 2422 parse_rockridge_NM1(struct file_info *file, 2423 const unsigned char *data, int data_length) 2424 { 2425 if (!file->name_continues) 2426 archive_string_empty(&file->name); 2427 file->name_continues = 0; 2428 if (data_length < 1) 2429 return; 2430 /* 2431 * NM version 1 extension comprises: 2432 * 1 byte flag, value is one of: 2433 * = 0: remainder is name 2434 * = 1: remainder is name, next NM entry continues name 2435 * = 2: "." 2436 * = 4: ".." 2437 * = 32: Implementation specific 2438 * All other values are reserved. 2439 */ 2440 switch(data[0]) { 2441 case 0: 2442 if (data_length < 2) 2443 return; 2444 archive_strncat(&file->name, 2445 (const char *)data + 1, data_length - 1); 2446 break; 2447 case 1: 2448 if (data_length < 2) 2449 return; 2450 archive_strncat(&file->name, 2451 (const char *)data + 1, data_length - 1); 2452 file->name_continues = 1; 2453 break; 2454 case 2: 2455 archive_strcat(&file->name, "."); 2456 break; 2457 case 4: 2458 archive_strcat(&file->name, ".."); 2459 break; 2460 default: 2461 return; 2462 } 2463 2464 } 2465 2466 static void 2467 parse_rockridge_TF1(struct file_info *file, const unsigned char *data, 2468 int data_length) 2469 { 2470 char flag; 2471 /* 2472 * TF extension comprises: 2473 * one byte flag 2474 * create time (optional) 2475 * modify time (optional) 2476 * access time (optional) 2477 * attribute time (optional) 2478 * Time format and presence of fields 2479 * is controlled by flag bits. 2480 */ 2481 if (data_length < 1) 2482 return; 2483 flag = data[0]; 2484 ++data; 2485 --data_length; 2486 if (flag & 0x80) { 2487 /* Use 17-byte time format. */ 2488 if ((flag & 1) && data_length >= 17) { 2489 /* Create time. */ 2490 file->birthtime_is_set = 1; 2491 file->birthtime = isodate17(data); 2492 data += 17; 2493 data_length -= 17; 2494 } 2495 if ((flag & 2) && data_length >= 17) { 2496 /* Modify time. */ 2497 file->mtime = isodate17(data); 2498 data += 17; 2499 data_length -= 17; 2500 } 2501 if ((flag & 4) && data_length >= 17) { 2502 /* Access time. */ 2503 file->atime = isodate17(data); 2504 data += 17; 2505 data_length -= 17; 2506 } 2507 if ((flag & 8) && data_length >= 17) { 2508 /* Attribute change time. */ 2509 file->ctime = isodate17(data); 2510 } 2511 } else { 2512 /* Use 7-byte time format. */ 2513 if ((flag & 1) && data_length >= 7) { 2514 /* Create time. */ 2515 file->birthtime_is_set = 1; 2516 file->birthtime = isodate7(data); 2517 data += 7; 2518 data_length -= 7; 2519 } 2520 if ((flag & 2) && data_length >= 7) { 2521 /* Modify time. */ 2522 file->mtime = isodate7(data); 2523 data += 7; 2524 data_length -= 7; 2525 } 2526 if ((flag & 4) && data_length >= 7) { 2527 /* Access time. */ 2528 file->atime = isodate7(data); 2529 data += 7; 2530 data_length -= 7; 2531 } 2532 if ((flag & 8) && data_length >= 7) { 2533 /* Attribute change time. */ 2534 file->ctime = isodate7(data); 2535 } 2536 } 2537 } 2538 2539 static void 2540 parse_rockridge_SL1(struct file_info *file, const unsigned char *data, 2541 int data_length) 2542 { 2543 const char *separator = ""; 2544 2545 if (!file->symlink_continues || file->symlink.length < 1) 2546 archive_string_empty(&file->symlink); 2547 file->symlink_continues = 0; 2548 2549 /* 2550 * Defined flag values: 2551 * 0: This is the last SL record for this symbolic link 2552 * 1: this symbolic link field continues in next SL entry 2553 * All other values are reserved. 2554 */ 2555 if (data_length < 1) 2556 return; 2557 switch(*data) { 2558 case 0: 2559 break; 2560 case 1: 2561 file->symlink_continues = 1; 2562 break; 2563 default: 2564 return; 2565 } 2566 ++data; /* Skip flag byte. */ 2567 --data_length; 2568 2569 /* 2570 * SL extension body stores "components". 2571 * Basically, this is a complicated way of storing 2572 * a POSIX path. It also interferes with using 2573 * symlinks for storing non-path data. <sigh> 2574 * 2575 * Each component is 2 bytes (flag and length) 2576 * possibly followed by name data. 2577 */ 2578 while (data_length >= 2) { 2579 unsigned char flag = *data++; 2580 unsigned char nlen = *data++; 2581 data_length -= 2; 2582 2583 archive_strcat(&file->symlink, separator); 2584 separator = "/"; 2585 2586 switch(flag) { 2587 case 0: /* Usual case, this is text. */ 2588 if (data_length < nlen) 2589 return; 2590 archive_strncat(&file->symlink, 2591 (const char *)data, nlen); 2592 break; 2593 case 0x01: /* Text continues in next component. */ 2594 if (data_length < nlen) 2595 return; 2596 archive_strncat(&file->symlink, 2597 (const char *)data, nlen); 2598 separator = ""; 2599 break; 2600 case 0x02: /* Current dir. */ 2601 archive_strcat(&file->symlink, "."); 2602 break; 2603 case 0x04: /* Parent dir. */ 2604 archive_strcat(&file->symlink, ".."); 2605 break; 2606 case 0x08: /* Root of filesystem. */ 2607 archive_strcat(&file->symlink, "/"); 2608 separator = ""; 2609 break; 2610 case 0x10: /* Undefined (historically "volume root" */ 2611 archive_string_empty(&file->symlink); 2612 archive_strcat(&file->symlink, "ROOT"); 2613 break; 2614 case 0x20: /* Undefined (historically "hostname") */ 2615 archive_strcat(&file->symlink, "hostname"); 2616 break; 2617 default: 2618 /* TODO: issue a warning ? */ 2619 return; 2620 } 2621 data += nlen; 2622 data_length -= nlen; 2623 } 2624 } 2625 2626 static void 2627 parse_rockridge_ZF1(struct file_info *file, const unsigned char *data, 2628 int data_length) 2629 { 2630 2631 if (data[0] == 0x70 && data[1] == 0x7a && data_length == 12) { 2632 /* paged zlib */ 2633 file->pz = 1; 2634 file->pz_log2_bs = data[3]; 2635 file->pz_uncompressed_size = archive_le32dec(&data[4]); 2636 } 2637 } 2638 2639 static void 2640 register_file(struct iso9660 *iso9660, struct file_info *file) 2641 { 2642 2643 file->use_next = iso9660->use_files; 2644 iso9660->use_files = file; 2645 } 2646 2647 static void 2648 release_files(struct iso9660 *iso9660) 2649 { 2650 struct content *con, *connext; 2651 struct file_info *file; 2652 2653 file = iso9660->use_files; 2654 while (file != NULL) { 2655 struct file_info *next = file->use_next; 2656 2657 archive_string_free(&file->name); 2658 archive_string_free(&file->symlink); 2659 free(file->utf16be_name); 2660 con = file->contents.first; 2661 while (con != NULL) { 2662 connext = con->next; 2663 free(con); 2664 con = connext; 2665 } 2666 free(file); 2667 file = next; 2668 } 2669 } 2670 2671 static int 2672 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, 2673 struct file_info **pfile) 2674 { 2675 struct file_info *file; 2676 int r; 2677 2678 r = next_cache_entry(a, iso9660, pfile); 2679 if (r != ARCHIVE_OK) 2680 return (r); 2681 file = *pfile; 2682 2683 /* Don't waste time seeking for zero-length bodies. */ 2684 if (file->size == 0) 2685 file->offset = iso9660->current_position; 2686 2687 /* flush any remaining bytes from the last round to ensure 2688 * we're positioned */ 2689 if (iso9660->entry_bytes_unconsumed) { 2690 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 2691 iso9660->entry_bytes_unconsumed = 0; 2692 } 2693 2694 /* Seek forward to the start of the entry. */ 2695 if (iso9660->current_position < file->offset) { 2696 int64_t step; 2697 2698 step = file->offset - iso9660->current_position; 2699 step = __archive_read_consume(a, step); 2700 if (step < 0) 2701 return ((int)step); 2702 iso9660->current_position = file->offset; 2703 } 2704 2705 /* We found body of file; handle it now. */ 2706 return (ARCHIVE_OK); 2707 } 2708 2709 static int 2710 next_cache_entry(struct archive_read *a, struct iso9660 *iso9660, 2711 struct file_info **pfile) 2712 { 2713 struct file_info *file; 2714 struct { 2715 struct file_info *first; 2716 struct file_info **last; 2717 } empty_files; 2718 int64_t number; 2719 int count; 2720 2721 file = cache_get_entry(iso9660); 2722 if (file != NULL) { 2723 *pfile = file; 2724 return (ARCHIVE_OK); 2725 } 2726 2727 for (;;) { 2728 struct file_info *re, *d; 2729 2730 *pfile = file = next_entry(iso9660); 2731 if (file == NULL) { 2732 /* 2733 * If directory entries all which are descendant of 2734 * rr_moved are still remaining, expose their. 2735 */ 2736 if (iso9660->re_files.first != NULL && 2737 iso9660->rr_moved != NULL && 2738 iso9660->rr_moved->rr_moved_has_re_only) 2739 /* Expose "rr_moved" entry. */ 2740 cache_add_entry(iso9660, iso9660->rr_moved); 2741 while ((re = re_get_entry(iso9660)) != NULL) { 2742 /* Expose its descendant dirs. */ 2743 while ((d = rede_get_entry(re)) != NULL) 2744 cache_add_entry(iso9660, d); 2745 } 2746 if (iso9660->cache_files.first != NULL) 2747 return (next_cache_entry(a, iso9660, pfile)); 2748 return (ARCHIVE_EOF); 2749 } 2750 2751 if (file->cl_offset) { 2752 struct file_info *first_re = NULL; 2753 int nexted_re = 0; 2754 2755 /* 2756 * Find "RE" dir for the current file, which 2757 * has "CL" flag. 2758 */ 2759 while ((re = re_get_entry(iso9660)) 2760 != first_re) { 2761 if (first_re == NULL) 2762 first_re = re; 2763 if (re->offset == file->cl_offset) { 2764 re->parent->subdirs--; 2765 re->parent = file->parent; 2766 re->re = 0; 2767 if (re->parent->re_descendant) { 2768 nexted_re = 1; 2769 re->re_descendant = 1; 2770 if (rede_add_entry(re) < 0) 2771 goto fatal_rr; 2772 /* Move a list of descendants 2773 * to a new ancestor. */ 2774 while ((d = rede_get_entry( 2775 re)) != NULL) 2776 if (rede_add_entry(d) 2777 < 0) 2778 goto fatal_rr; 2779 break; 2780 } 2781 /* Replace the current file 2782 * with "RE" dir */ 2783 *pfile = file = re; 2784 /* Expose its descendant */ 2785 while ((d = rede_get_entry( 2786 file)) != NULL) 2787 cache_add_entry( 2788 iso9660, d); 2789 break; 2790 } else 2791 re_add_entry(iso9660, re); 2792 } 2793 if (nexted_re) { 2794 /* 2795 * Do not expose this at this time 2796 * because we have not gotten its full-path 2797 * name yet. 2798 */ 2799 continue; 2800 } 2801 } else if ((file->mode & AE_IFMT) == AE_IFDIR) { 2802 int r; 2803 2804 /* Read file entries in this dir. */ 2805 r = read_children(a, file); 2806 if (r != ARCHIVE_OK) 2807 return (r); 2808 2809 /* 2810 * Handle a special dir of Rockridge extensions, 2811 * "rr_moved". 2812 */ 2813 if (file->rr_moved) { 2814 /* 2815 * If this has only the subdirectories which 2816 * have "RE" flags, do not expose at this time. 2817 */ 2818 if (file->rr_moved_has_re_only) 2819 continue; 2820 /* Otherwise expose "rr_moved" entry. */ 2821 } else if (file->re) { 2822 /* 2823 * Do not expose this at this time 2824 * because we have not gotten its full-path 2825 * name yet. 2826 */ 2827 re_add_entry(iso9660, file); 2828 continue; 2829 } else if (file->re_descendant) { 2830 /* 2831 * If the top level "RE" entry of this entry 2832 * is not exposed, we, accordingly, should not 2833 * expose this entry at this time because 2834 * we cannot make its proper full-path name. 2835 */ 2836 if (rede_add_entry(file) == 0) 2837 continue; 2838 /* Otherwise we can expose this entry because 2839 * it seems its top level "RE" has already been 2840 * exposed. */ 2841 } 2842 } 2843 break; 2844 } 2845 2846 if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1) 2847 return (ARCHIVE_OK); 2848 2849 count = 0; 2850 number = file->number; 2851 iso9660->cache_files.first = NULL; 2852 iso9660->cache_files.last = &(iso9660->cache_files.first); 2853 empty_files.first = NULL; 2854 empty_files.last = &empty_files.first; 2855 /* Collect files which has the same file serial number. 2856 * Peek pending_files so that file which number is different 2857 * is not put back. */ 2858 while (iso9660->pending_files.used > 0 && 2859 (iso9660->pending_files.files[0]->number == -1 || 2860 iso9660->pending_files.files[0]->number == number)) { 2861 if (file->number == -1) { 2862 /* This file has the same offset 2863 * but it's wrong offset which empty files 2864 * and symlink files have. 2865 * NOTE: This wrong offset was recorded by 2866 * old mkisofs utility. If ISO images is 2867 * created by latest mkisofs, this does not 2868 * happen. 2869 */ 2870 file->next = NULL; 2871 *empty_files.last = file; 2872 empty_files.last = &(file->next); 2873 } else { 2874 count++; 2875 cache_add_entry(iso9660, file); 2876 } 2877 file = next_entry(iso9660); 2878 } 2879 2880 if (count == 0) { 2881 *pfile = file; 2882 return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2883 } 2884 if (file->number == -1) { 2885 file->next = NULL; 2886 *empty_files.last = file; 2887 empty_files.last = &(file->next); 2888 } else { 2889 count++; 2890 cache_add_entry(iso9660, file); 2891 } 2892 2893 if (count > 1) { 2894 /* The count is the same as number of hardlink, 2895 * so much so that each nlinks of files in cache_file 2896 * is overwritten by value of the count. 2897 */ 2898 for (file = iso9660->cache_files.first; 2899 file != NULL; file = file->next) 2900 file->nlinks = count; 2901 } 2902 /* If there are empty files, that files are added 2903 * to the tail of the cache_files. */ 2904 if (empty_files.first != NULL) { 2905 *iso9660->cache_files.last = empty_files.first; 2906 iso9660->cache_files.last = empty_files.last; 2907 } 2908 *pfile = cache_get_entry(iso9660); 2909 return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2910 2911 fatal_rr: 2912 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2913 "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of " 2914 "Rockridge extensions: current position = %jd, CL offset = %jd", 2915 (intmax_t)iso9660->current_position, (intmax_t)file->cl_offset); 2916 return (ARCHIVE_FATAL); 2917 } 2918 2919 static inline void 2920 re_add_entry(struct iso9660 *iso9660, struct file_info *file) 2921 { 2922 file->re_next = NULL; 2923 *iso9660->re_files.last = file; 2924 iso9660->re_files.last = &(file->re_next); 2925 } 2926 2927 static inline struct file_info * 2928 re_get_entry(struct iso9660 *iso9660) 2929 { 2930 struct file_info *file; 2931 2932 if ((file = iso9660->re_files.first) != NULL) { 2933 iso9660->re_files.first = file->re_next; 2934 if (iso9660->re_files.first == NULL) 2935 iso9660->re_files.last = 2936 &(iso9660->re_files.first); 2937 } 2938 return (file); 2939 } 2940 2941 static inline int 2942 rede_add_entry(struct file_info *file) 2943 { 2944 struct file_info *re; 2945 2946 /* 2947 * Find "RE" entry. 2948 */ 2949 re = file->parent; 2950 while (re != NULL && !re->re) 2951 re = re->parent; 2952 if (re == NULL) 2953 return (-1); 2954 2955 file->re_next = NULL; 2956 *re->rede_files.last = file; 2957 re->rede_files.last = &(file->re_next); 2958 return (0); 2959 } 2960 2961 static inline struct file_info * 2962 rede_get_entry(struct file_info *re) 2963 { 2964 struct file_info *file; 2965 2966 if ((file = re->rede_files.first) != NULL) { 2967 re->rede_files.first = file->re_next; 2968 if (re->rede_files.first == NULL) 2969 re->rede_files.last = 2970 &(re->rede_files.first); 2971 } 2972 return (file); 2973 } 2974 2975 static inline void 2976 cache_add_entry(struct iso9660 *iso9660, struct file_info *file) 2977 { 2978 file->next = NULL; 2979 *iso9660->cache_files.last = file; 2980 iso9660->cache_files.last = &(file->next); 2981 } 2982 2983 static inline struct file_info * 2984 cache_get_entry(struct iso9660 *iso9660) 2985 { 2986 struct file_info *file; 2987 2988 if ((file = iso9660->cache_files.first) != NULL) { 2989 iso9660->cache_files.first = file->next; 2990 if (iso9660->cache_files.first == NULL) 2991 iso9660->cache_files.last = 2992 &(iso9660->cache_files.first); 2993 } 2994 return (file); 2995 } 2996 2997 static int 2998 heap_add_entry(struct archive_read *a, struct heap_queue *heap, 2999 struct file_info *file, uint64_t key) 3000 { 3001 uint64_t file_key, parent_key; 3002 int hole, parent; 3003 3004 /* Expand our pending files list as necessary. */ 3005 if (heap->used >= heap->allocated) { 3006 struct file_info **new_pending_files; 3007 int new_size = heap->allocated * 2; 3008 3009 if (heap->allocated < 1024) 3010 new_size = 1024; 3011 /* Overflow might keep us from growing the list. */ 3012 if (new_size <= heap->allocated) { 3013 archive_set_error(&a->archive, 3014 ENOMEM, "Out of memory"); 3015 return (ARCHIVE_FATAL); 3016 } 3017 new_pending_files = (struct file_info **) 3018 malloc(new_size * sizeof(new_pending_files[0])); 3019 if (new_pending_files == NULL) { 3020 archive_set_error(&a->archive, 3021 ENOMEM, "Out of memory"); 3022 return (ARCHIVE_FATAL); 3023 } 3024 if (heap->allocated) 3025 memcpy(new_pending_files, heap->files, 3026 heap->allocated * sizeof(new_pending_files[0])); 3027 if (heap->files != NULL) 3028 free(heap->files); 3029 heap->files = new_pending_files; 3030 heap->allocated = new_size; 3031 } 3032 3033 file_key = file->key = key; 3034 3035 /* 3036 * Start with hole at end, walk it up tree to find insertion point. 3037 */ 3038 hole = heap->used++; 3039 while (hole > 0) { 3040 parent = (hole - 1)/2; 3041 parent_key = heap->files[parent]->key; 3042 if (file_key >= parent_key) { 3043 heap->files[hole] = file; 3044 return (ARCHIVE_OK); 3045 } 3046 /* Move parent into hole <==> move hole up tree. */ 3047 heap->files[hole] = heap->files[parent]; 3048 hole = parent; 3049 } 3050 heap->files[0] = file; 3051 3052 return (ARCHIVE_OK); 3053 } 3054 3055 static struct file_info * 3056 heap_get_entry(struct heap_queue *heap) 3057 { 3058 uint64_t a_key, b_key, c_key; 3059 int a, b, c; 3060 struct file_info *r, *tmp; 3061 3062 if (heap->used < 1) 3063 return (NULL); 3064 3065 /* 3066 * The first file in the list is the earliest; we'll return this. 3067 */ 3068 r = heap->files[0]; 3069 3070 /* 3071 * Move the last item in the heap to the root of the tree 3072 */ 3073 heap->files[0] = heap->files[--(heap->used)]; 3074 3075 /* 3076 * Rebalance the heap. 3077 */ 3078 a = 0; /* Starting element and its heap key */ 3079 a_key = heap->files[a]->key; 3080 for (;;) { 3081 b = a + a + 1; /* First child */ 3082 if (b >= heap->used) 3083 return (r); 3084 b_key = heap->files[b]->key; 3085 c = b + 1; /* Use second child if it is smaller. */ 3086 if (c < heap->used) { 3087 c_key = heap->files[c]->key; 3088 if (c_key < b_key) { 3089 b = c; 3090 b_key = c_key; 3091 } 3092 } 3093 if (a_key <= b_key) 3094 return (r); 3095 tmp = heap->files[a]; 3096 heap->files[a] = heap->files[b]; 3097 heap->files[b] = tmp; 3098 a = b; 3099 } 3100 } 3101 3102 static unsigned int 3103 toi(const void *p, int n) 3104 { 3105 const unsigned char *v = (const unsigned char *)p; 3106 if (n > 1) 3107 return v[0] + 256 * toi(v + 1, n - 1); 3108 if (n == 1) 3109 return v[0]; 3110 return (0); 3111 } 3112 3113 static time_t 3114 isodate7(const unsigned char *v) 3115 { 3116 struct tm tm; 3117 int offset; 3118 time_t t; 3119 3120 memset(&tm, 0, sizeof(tm)); 3121 tm.tm_year = v[0]; 3122 tm.tm_mon = v[1] - 1; 3123 tm.tm_mday = v[2]; 3124 tm.tm_hour = v[3]; 3125 tm.tm_min = v[4]; 3126 tm.tm_sec = v[5]; 3127 /* v[6] is the signed timezone offset, in 1/4-hour increments. */ 3128 offset = ((const signed char *)v)[6]; 3129 if (offset > -48 && offset < 52) { 3130 tm.tm_hour -= offset / 4; 3131 tm.tm_min -= (offset % 4) * 15; 3132 } 3133 t = time_from_tm(&tm); 3134 if (t == (time_t)-1) 3135 return ((time_t)0); 3136 return (t); 3137 } 3138 3139 static time_t 3140 isodate17(const unsigned char *v) 3141 { 3142 struct tm tm; 3143 int offset; 3144 time_t t; 3145 3146 memset(&tm, 0, sizeof(tm)); 3147 tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 3148 + (v[2] - '0') * 10 + (v[3] - '0') 3149 - 1900; 3150 tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); 3151 tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); 3152 tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); 3153 tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); 3154 tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); 3155 /* v[16] is the signed timezone offset, in 1/4-hour increments. */ 3156 offset = ((const signed char *)v)[16]; 3157 if (offset > -48 && offset < 52) { 3158 tm.tm_hour -= offset / 4; 3159 tm.tm_min -= (offset % 4) * 15; 3160 } 3161 t = time_from_tm(&tm); 3162 if (t == (time_t)-1) 3163 return ((time_t)0); 3164 return (t); 3165 } 3166 3167 static time_t 3168 time_from_tm(struct tm *t) 3169 { 3170 #if HAVE_TIMEGM 3171 /* Use platform timegm() if available. */ 3172 return (timegm(t)); 3173 #elif HAVE__MKGMTIME64 3174 return (_mkgmtime64(t)); 3175 #else 3176 /* Else use direct calculation using POSIX assumptions. */ 3177 /* First, fix up tm_yday based on the year/month/day. */ 3178 if (mktime(t) == (time_t)-1) 3179 return ((time_t)-1); 3180 /* Then we can compute timegm() from first principles. */ 3181 return (t->tm_sec 3182 + t->tm_min * 60 3183 + t->tm_hour * 3600 3184 + t->tm_yday * 86400 3185 + (t->tm_year - 70) * 31536000 3186 + ((t->tm_year - 69) / 4) * 86400 3187 - ((t->tm_year - 1) / 100) * 86400 3188 + ((t->tm_year + 299) / 400) * 86400); 3189 #endif 3190 } 3191 3192 static const char * 3193 build_pathname(struct archive_string *as, struct file_info *file, int depth) 3194 { 3195 // Plain ISO9660 only allows 8 dir levels; if we get 3196 // to 1000, then something is very, very wrong. 3197 if (depth > 1000) { 3198 return NULL; 3199 } 3200 if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { 3201 if (build_pathname(as, file->parent, depth + 1) == NULL) { 3202 return NULL; 3203 } 3204 archive_strcat(as, "/"); 3205 } 3206 if (archive_strlen(&file->name) == 0) 3207 archive_strcat(as, "."); 3208 else 3209 archive_string_concat(as, &file->name); 3210 return (as->s); 3211 } 3212 3213 static int 3214 build_pathname_utf16be(unsigned char *p, size_t max, size_t *len, 3215 struct file_info *file) 3216 { 3217 if (file->parent != NULL && file->parent->utf16be_bytes > 0) { 3218 if (build_pathname_utf16be(p, max, len, file->parent) != 0) 3219 return (-1); 3220 p[*len] = 0; 3221 p[*len + 1] = '/'; 3222 *len += 2; 3223 } 3224 if (file->utf16be_bytes == 0) { 3225 if (*len + 2 > max) 3226 return (-1);/* Path is too long! */ 3227 p[*len] = 0; 3228 p[*len + 1] = '.'; 3229 *len += 2; 3230 } else { 3231 if (*len + file->utf16be_bytes > max) 3232 return (-1);/* Path is too long! */ 3233 memcpy(p + *len, file->utf16be_name, file->utf16be_bytes); 3234 *len += file->utf16be_bytes; 3235 } 3236 return (0); 3237 } 3238 3239 #if DEBUG 3240 static void 3241 dump_isodirrec(FILE *out, const unsigned char *isodirrec) 3242 { 3243 fprintf(out, " l %d,", 3244 toi(isodirrec + DR_length_offset, DR_length_size)); 3245 fprintf(out, " a %d,", 3246 toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); 3247 fprintf(out, " ext 0x%x,", 3248 toi(isodirrec + DR_extent_offset, DR_extent_size)); 3249 fprintf(out, " s %d,", 3250 toi(isodirrec + DR_size_offset, DR_extent_size)); 3251 fprintf(out, " f 0x%x,", 3252 toi(isodirrec + DR_flags_offset, DR_flags_size)); 3253 fprintf(out, " u %d,", 3254 toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); 3255 fprintf(out, " ilv %d,", 3256 toi(isodirrec + DR_interleave_offset, DR_interleave_size)); 3257 fprintf(out, " seq %d,", 3258 toi(isodirrec + DR_volume_sequence_number_offset, 3259 DR_volume_sequence_number_size)); 3260 fprintf(out, " nl %d:", 3261 toi(isodirrec + DR_name_len_offset, DR_name_len_size)); 3262 fprintf(out, " `%.*s'", 3263 toi(isodirrec + DR_name_len_offset, DR_name_len_size), 3264 isodirrec + DR_name_offset); 3265 } 3266 #endif 3267