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: head/lib/libarchive/archive_read_support_format_iso9660.c 201246 2009-12-30 05:30:35Z kientzle $"); 30 31 #ifdef HAVE_ERRNO_H 32 #include <errno.h> 33 #endif 34 /* #include <stdint.h> */ /* See archive_platform.h */ 35 #include <stdio.h> 36 #ifdef HAVE_STDLIB_H 37 #include <stdlib.h> 38 #endif 39 #ifdef HAVE_STRING_H 40 #include <string.h> 41 #endif 42 #include <time.h> 43 #ifdef HAVE_ZLIB_H 44 #include <zlib.h> 45 #endif 46 47 #include "archive.h" 48 #include "archive_endian.h" 49 #include "archive_entry.h" 50 #include "archive_entry_locale.h" 51 #include "archive_private.h" 52 #include "archive_read_private.h" 53 #include "archive_string.h" 54 55 /* 56 * An overview of ISO 9660 format: 57 * 58 * Each disk is laid out as follows: 59 * * 32k reserved for private use 60 * * Volume descriptor table. Each volume descriptor 61 * is 2k and specifies basic format information. 62 * The "Primary Volume Descriptor" (PVD) is defined by the 63 * standard and should always be present; other volume 64 * descriptors include various vendor-specific extensions. 65 * * Files and directories. Each file/dir is specified by 66 * an "extent" (starting sector and length in bytes). 67 * Dirs are just files with directory records packed one 68 * after another. The PVD contains a single dir entry 69 * specifying the location of the root directory. Everything 70 * else follows from there. 71 * 72 * This module works by first reading the volume descriptors, then 73 * building a list of directory entries, sorted by starting 74 * sector. At each step, I look for the earliest dir entry that 75 * hasn't yet been read, seek forward to that location and read 76 * that entry. If it's a dir, I slurp in the new dir entries and 77 * add them to the heap; if it's a regular file, I return the 78 * corresponding archive_entry and wait for the client to request 79 * the file body. This strategy allows us to read most compliant 80 * CDs with a single pass through the data, as required by libarchive. 81 */ 82 #define LOGICAL_BLOCK_SIZE 2048 83 #define SYSTEM_AREA_BLOCK 16 84 85 /* Structure of on-disk primary volume descriptor. */ 86 #define PVD_type_offset 0 87 #define PVD_type_size 1 88 #define PVD_id_offset (PVD_type_offset + PVD_type_size) 89 #define PVD_id_size 5 90 #define PVD_version_offset (PVD_id_offset + PVD_id_size) 91 #define PVD_version_size 1 92 #define PVD_reserved1_offset (PVD_version_offset + PVD_version_size) 93 #define PVD_reserved1_size 1 94 #define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size) 95 #define PVD_system_id_size 32 96 #define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size) 97 #define PVD_volume_id_size 32 98 #define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size) 99 #define PVD_reserved2_size 8 100 #define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size) 101 #define PVD_volume_space_size_size 8 102 #define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size) 103 #define PVD_reserved3_size 32 104 #define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size) 105 #define PVD_volume_set_size_size 4 106 #define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size) 107 #define PVD_volume_sequence_number_size 4 108 #define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size) 109 #define PVD_logical_block_size_size 4 110 #define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size) 111 #define PVD_path_table_size_size 8 112 #define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size) 113 #define PVD_type_1_path_table_size 4 114 #define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size) 115 #define PVD_opt_type_1_path_table_size 4 116 #define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size) 117 #define PVD_type_m_path_table_size 4 118 #define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size) 119 #define PVD_opt_type_m_path_table_size 4 120 #define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size) 121 #define PVD_root_directory_record_size 34 122 #define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size) 123 #define PVD_volume_set_id_size 128 124 #define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size) 125 #define PVD_publisher_id_size 128 126 #define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size) 127 #define PVD_preparer_id_size 128 128 #define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size) 129 #define PVD_application_id_size 128 130 #define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size) 131 #define PVD_copyright_file_id_size 37 132 #define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size) 133 #define PVD_abstract_file_id_size 37 134 #define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size) 135 #define PVD_bibliographic_file_id_size 37 136 #define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size) 137 #define PVD_creation_date_size 17 138 #define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size) 139 #define PVD_modification_date_size 17 140 #define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size) 141 #define PVD_expiration_date_size 17 142 #define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size) 143 #define PVD_effective_date_size 17 144 #define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size) 145 #define PVD_file_structure_version_size 1 146 #define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size) 147 #define PVD_reserved4_size 1 148 #define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size) 149 #define PVD_application_data_size 512 150 #define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size) 151 #define PVD_reserved5_size (2048 - PVD_reserved5_offset) 152 153 /* TODO: It would make future maintenance easier to just hardcode the 154 * above values. In particular, ECMA119 states the offsets as part of 155 * the standard. That would eliminate the need for the following check.*/ 156 #if PVD_reserved5_offset != 1395 157 #error PVD offset and size definitions are wrong. 158 #endif 159 160 161 /* Structure of optional on-disk supplementary volume descriptor. */ 162 #define SVD_type_offset 0 163 #define SVD_type_size 1 164 #define SVD_id_offset (SVD_type_offset + SVD_type_size) 165 #define SVD_id_size 5 166 #define SVD_version_offset (SVD_id_offset + SVD_id_size) 167 #define SVD_version_size 1 168 /* ... */ 169 #define SVD_reserved1_offset 72 170 #define SVD_reserved1_size 8 171 #define SVD_volume_space_size_offset 80 172 #define SVD_volume_space_size_size 8 173 #define SVD_escape_sequences_offset (SVD_volume_space_size_offset + SVD_volume_space_size_size) 174 #define SVD_escape_sequences_size 32 175 /* ... */ 176 #define SVD_logical_block_size_offset 128 177 #define SVD_logical_block_size_size 4 178 #define SVD_type_L_path_table_offset 140 179 #define SVD_type_M_path_table_offset 148 180 /* ... */ 181 #define SVD_root_directory_record_offset 156 182 #define SVD_root_directory_record_size 34 183 #define SVD_file_structure_version_offset 881 184 #define SVD_reserved2_offset 882 185 #define SVD_reserved2_size 1 186 #define SVD_reserved3_offset 1395 187 #define SVD_reserved3_size 653 188 /* ... */ 189 /* FIXME: validate correctness of last SVD entry offset. */ 190 191 /* Structure of an on-disk directory record. */ 192 /* Note: ISO9660 stores each multi-byte integer twice, once in 193 * each byte order. The sizes here are the size of just one 194 * of the two integers. (This is why the offset of a field isn't 195 * the same as the offset+size of the previous field.) */ 196 #define DR_length_offset 0 197 #define DR_length_size 1 198 #define DR_ext_attr_length_offset 1 199 #define DR_ext_attr_length_size 1 200 #define DR_extent_offset 2 201 #define DR_extent_size 4 202 #define DR_size_offset 10 203 #define DR_size_size 4 204 #define DR_date_offset 18 205 #define DR_date_size 7 206 #define DR_flags_offset 25 207 #define DR_flags_size 1 208 #define DR_file_unit_size_offset 26 209 #define DR_file_unit_size_size 1 210 #define DR_interleave_offset 27 211 #define DR_interleave_size 1 212 #define DR_volume_sequence_number_offset 28 213 #define DR_volume_sequence_number_size 2 214 #define DR_name_len_offset 32 215 #define DR_name_len_size 1 216 #define DR_name_offset 33 217 218 #ifdef HAVE_ZLIB_H 219 static const unsigned char zisofs_magic[8] = { 220 0x37, 0xE4, 0x53, 0x96, 0xC9, 0xDB, 0xD6, 0x07 221 }; 222 223 struct zisofs { 224 /* Set 1 if this file compressed by paged zlib */ 225 int pz; 226 int pz_log2_bs; /* Log2 of block size */ 227 uint64_t pz_uncompressed_size; 228 229 int initialized; 230 unsigned char *uncompressed_buffer; 231 size_t uncompressed_buffer_size; 232 233 uint32_t pz_offset; 234 unsigned char header[16]; 235 size_t header_avail; 236 int header_passed; 237 unsigned char *block_pointers; 238 size_t block_pointers_alloc; 239 size_t block_pointers_size; 240 size_t block_pointers_avail; 241 size_t block_off; 242 uint32_t block_avail; 243 244 z_stream stream; 245 int stream_valid; 246 }; 247 #else 248 struct zisofs { 249 /* Set 1 if this file compressed by paged zlib */ 250 int pz; 251 }; 252 #endif 253 254 struct content { 255 uint64_t offset;/* Offset on disk. */ 256 uint64_t size; /* File size in bytes. */ 257 struct content *next; 258 }; 259 260 /* In-memory storage for a directory record. */ 261 struct file_info { 262 struct file_info *use_next; 263 struct file_info *parent; 264 struct file_info *next; 265 struct file_info *re_next; 266 int subdirs; 267 uint64_t key; /* Heap Key. */ 268 uint64_t offset; /* Offset on disk. */ 269 uint64_t size; /* File size in bytes. */ 270 uint32_t ce_offset; /* Offset of CE. */ 271 uint32_t ce_size; /* Size of CE. */ 272 char rr_moved; /* Flag to rr_moved. */ 273 char rr_moved_has_re_only; 274 char re; /* Having RRIP "RE" extension. */ 275 char re_descendant; 276 uint64_t cl_offset; /* Having RRIP "CL" extension. */ 277 int birthtime_is_set; 278 time_t birthtime; /* File created time. */ 279 time_t mtime; /* File last modified time. */ 280 time_t atime; /* File last accessed time. */ 281 time_t ctime; /* File attribute change time. */ 282 uint64_t rdev; /* Device number. */ 283 mode_t mode; 284 uid_t uid; 285 gid_t gid; 286 int64_t number; 287 int nlinks; 288 struct archive_string name; /* Pathname */ 289 unsigned char *utf16be_name; 290 size_t utf16be_bytes; 291 char name_continues; /* Non-zero if name continues */ 292 struct archive_string symlink; 293 char symlink_continues; /* Non-zero if link continues */ 294 /* Set 1 if this file compressed by paged zlib(zisofs) */ 295 int pz; 296 int pz_log2_bs; /* Log2 of block size */ 297 uint64_t pz_uncompressed_size; 298 /* Set 1 if this file is multi extent. */ 299 int multi_extent; 300 struct { 301 struct content *first; 302 struct content **last; 303 } contents; 304 struct { 305 struct file_info *first; 306 struct file_info **last; 307 } rede_files; 308 }; 309 310 struct heap_queue { 311 struct file_info **files; 312 int allocated; 313 int used; 314 }; 315 316 struct iso9660 { 317 int magic; 318 #define ISO9660_MAGIC 0x96609660 319 320 int opt_support_joliet; 321 int opt_support_rockridge; 322 323 struct archive_string pathname; 324 char seenRockridge; /* Set true if RR extensions are used. */ 325 char seenSUSP; /* Set true if SUSP is beging used. */ 326 char seenJoliet; 327 328 unsigned char suspOffset; 329 struct file_info *rr_moved; 330 struct read_ce_queue { 331 struct read_ce_req { 332 uint64_t offset;/* Offset of CE on disk. */ 333 struct file_info *file; 334 } *reqs; 335 int cnt; 336 int allocated; 337 } read_ce_req; 338 339 int64_t previous_number; 340 struct archive_string previous_pathname; 341 342 struct file_info *use_files; 343 struct heap_queue pending_files; 344 struct { 345 struct file_info *first; 346 struct file_info **last; 347 } cache_files; 348 struct { 349 struct file_info *first; 350 struct file_info **last; 351 } re_files; 352 353 uint64_t current_position; 354 ssize_t logical_block_size; 355 uint64_t volume_size; /* Total size of volume in bytes. */ 356 int32_t volume_block;/* Total size of volume in logical blocks. */ 357 358 struct vd { 359 int location; /* Location of Extent. */ 360 uint32_t size; 361 } primary, joliet; 362 363 int64_t entry_sparse_offset; 364 int64_t entry_bytes_remaining; 365 size_t entry_bytes_unconsumed; 366 struct zisofs entry_zisofs; 367 struct content *entry_content; 368 struct archive_string_conv *sconv_utf16be; 369 /* 370 * Buffers for a full pathname in UTF-16BE in Joliet extensions. 371 */ 372 #define UTF16_NAME_MAX 1024 373 unsigned char *utf16be_path; 374 size_t utf16be_path_len; 375 unsigned char *utf16be_previous_path; 376 size_t utf16be_previous_path_len; 377 /* Null buufer 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 *); 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); 482 483 if (r != ARCHIVE_OK) { 484 free(iso9660); 485 return (r); 486 } 487 return (ARCHIVE_OK); 488 } 489 490 491 static int 492 archive_read_format_iso9660_bid(struct archive_read *a, int best_bid) 493 { 494 struct iso9660 *iso9660; 495 ssize_t bytes_read; 496 const unsigned char *p; 497 int seenTerminator; 498 499 /* If there's already a better bid than we can ever 500 make, don't bother testing. */ 501 if (best_bid > 48) 502 return (-1); 503 504 iso9660 = (struct iso9660 *)(a->format->data); 505 506 /* 507 * Skip the first 32k (reserved area) and get the first 508 * 8 sectors of the volume descriptor table. Of course, 509 * if the I/O layer gives us more, we'll take it. 510 */ 511 #define RESERVED_AREA (SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE) 512 p = __archive_read_ahead(a, 513 RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE, 514 &bytes_read); 515 if (p == NULL) 516 return (-1); 517 518 /* Skip the reserved area. */ 519 bytes_read -= RESERVED_AREA; 520 p += RESERVED_AREA; 521 522 /* Check each volume descriptor. */ 523 seenTerminator = 0; 524 for (; bytes_read > LOGICAL_BLOCK_SIZE; 525 bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) { 526 /* Do not handle undefined Volume Descriptor Type. */ 527 if (p[0] >= 4 && p[0] <= 254) 528 return (0); 529 /* Standard Identifier must be "CD001" */ 530 if (memcmp(p + 1, "CD001", 5) != 0) 531 return (0); 532 if (isPVD(iso9660, p)) 533 continue; 534 if (!iso9660->joliet.location) { 535 if (isJolietSVD(iso9660, p)) 536 continue; 537 } 538 if (isBootRecord(iso9660, p)) 539 continue; 540 if (isEVD(iso9660, p)) 541 continue; 542 if (isSVD(iso9660, p)) 543 continue; 544 if (isVolumePartition(iso9660, p)) 545 continue; 546 if (isVDSetTerminator(iso9660, p)) { 547 seenTerminator = 1; 548 break; 549 } 550 return (0); 551 } 552 /* 553 * ISO 9660 format must have Primary Volume Descriptor and 554 * Volume Descriptor Set Terminator. 555 */ 556 if (seenTerminator && iso9660->primary.location > 16) 557 return (48); 558 559 /* We didn't find a valid PVD; return a bid of zero. */ 560 return (0); 561 } 562 563 static int 564 archive_read_format_iso9660_options(struct archive_read *a, 565 const char *key, const char *val) 566 { 567 struct iso9660 *iso9660; 568 569 iso9660 = (struct iso9660 *)(a->format->data); 570 571 if (strcmp(key, "joliet") == 0) { 572 if (val == NULL || strcmp(val, "off") == 0 || 573 strcmp(val, "ignore") == 0 || 574 strcmp(val, "disable") == 0 || 575 strcmp(val, "0") == 0) 576 iso9660->opt_support_joliet = 0; 577 else 578 iso9660->opt_support_joliet = 1; 579 return (ARCHIVE_OK); 580 } 581 if (strcmp(key, "rockridge") == 0 || 582 strcmp(key, "Rockridge") == 0) { 583 iso9660->opt_support_rockridge = val != NULL; 584 return (ARCHIVE_OK); 585 } 586 587 /* Note: The "warn" return is just to inform the options 588 * supervisor that we didn't handle it. It will generate 589 * a suitable error if no one used this option. */ 590 return (ARCHIVE_WARN); 591 } 592 593 static int 594 isNull(struct iso9660 *iso9660, const unsigned char *h, unsigned offset, 595 unsigned bytes) 596 { 597 598 while (bytes >= sizeof(iso9660->null)) { 599 if (!memcmp(iso9660->null, h + offset, sizeof(iso9660->null))) 600 return (0); 601 offset += sizeof(iso9660->null); 602 bytes -= sizeof(iso9660->null); 603 } 604 if (bytes) 605 return memcmp(iso9660->null, h + offset, bytes) == 0; 606 else 607 return (1); 608 } 609 610 static int 611 isBootRecord(struct iso9660 *iso9660, const unsigned char *h) 612 { 613 (void)iso9660; /* UNUSED */ 614 615 /* Type of the Volume Descriptor Boot Record must be 0. */ 616 if (h[0] != 0) 617 return (0); 618 619 /* Volume Descriptor Version must be 1. */ 620 if (h[6] != 1) 621 return (0); 622 623 return (1); 624 } 625 626 static int 627 isVolumePartition(struct iso9660 *iso9660, const unsigned char *h) 628 { 629 int32_t location; 630 631 /* Type of the Volume Partition Descriptor must be 3. */ 632 if (h[0] != 3) 633 return (0); 634 635 /* Volume Descriptor Version must be 1. */ 636 if (h[6] != 1) 637 return (0); 638 /* Unused Field */ 639 if (h[7] != 0) 640 return (0); 641 642 location = archive_le32dec(h + 72); 643 if (location <= SYSTEM_AREA_BLOCK || 644 location >= iso9660->volume_block) 645 return (0); 646 if ((uint32_t)location != archive_be32dec(h + 76)) 647 return (0); 648 649 return (1); 650 } 651 652 static int 653 isVDSetTerminator(struct iso9660 *iso9660, const unsigned char *h) 654 { 655 (void)iso9660; /* UNUSED */ 656 657 /* Type of the Volume Descriptor Set Terminator must be 255. */ 658 if (h[0] != 255) 659 return (0); 660 661 /* Volume Descriptor Version must be 1. */ 662 if (h[6] != 1) 663 return (0); 664 665 /* Reserved field must be 0. */ 666 if (!isNull(iso9660, h, 7, 2048-7)) 667 return (0); 668 669 return (1); 670 } 671 672 static int 673 isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) 674 { 675 const unsigned char *p; 676 ssize_t logical_block_size; 677 int32_t volume_block; 678 679 /* Check if current sector is a kind of Supplementary Volume 680 * Descriptor. */ 681 if (!isSVD(iso9660, h)) 682 return (0); 683 684 /* FIXME: do more validations according to joliet spec. */ 685 686 /* check if this SVD contains joliet extension! */ 687 p = h + SVD_escape_sequences_offset; 688 /* N.B. Joliet spec says p[1] == '\\', but.... */ 689 if (p[0] == '%' && p[1] == '/') { 690 int level = 0; 691 692 if (p[2] == '@') 693 level = 1; 694 else if (p[2] == 'C') 695 level = 2; 696 else if (p[2] == 'E') 697 level = 3; 698 else /* not joliet */ 699 return (0); 700 701 iso9660->seenJoliet = level; 702 703 } else /* not joliet */ 704 return (0); 705 706 logical_block_size = 707 archive_le16dec(h + SVD_logical_block_size_offset); 708 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 709 710 iso9660->logical_block_size = logical_block_size; 711 iso9660->volume_block = volume_block; 712 iso9660->volume_size = logical_block_size * (uint64_t)volume_block; 713 /* Read Root Directory Record in Volume Descriptor. */ 714 p = h + SVD_root_directory_record_offset; 715 iso9660->joliet.location = archive_le32dec(p + DR_extent_offset); 716 iso9660->joliet.size = archive_le32dec(p + DR_size_offset); 717 718 return (48); 719 } 720 721 static int 722 isSVD(struct iso9660 *iso9660, const unsigned char *h) 723 { 724 const unsigned char *p; 725 ssize_t logical_block_size; 726 int32_t volume_block; 727 int32_t location; 728 729 (void)iso9660; /* UNUSED */ 730 731 /* Type 2 means it's a SVD. */ 732 if (h[SVD_type_offset] != 2) 733 return (0); 734 735 /* Reserved field must be 0. */ 736 if (!isNull(iso9660, h, SVD_reserved1_offset, SVD_reserved1_size)) 737 return (0); 738 if (!isNull(iso9660, h, SVD_reserved2_offset, SVD_reserved2_size)) 739 return (0); 740 if (!isNull(iso9660, h, SVD_reserved3_offset, SVD_reserved3_size)) 741 return (0); 742 743 /* File structure version must be 1 for ISO9660/ECMA119. */ 744 if (h[SVD_file_structure_version_offset] != 1) 745 return (0); 746 747 logical_block_size = 748 archive_le16dec(h + SVD_logical_block_size_offset); 749 if (logical_block_size <= 0) 750 return (0); 751 752 volume_block = archive_le32dec(h + SVD_volume_space_size_offset); 753 if (volume_block <= SYSTEM_AREA_BLOCK+4) 754 return (0); 755 756 /* Location of Occurrence of Type L Path Table must be 757 * available location, 758 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 759 location = archive_le32dec(h+SVD_type_L_path_table_offset); 760 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 761 return (0); 762 763 /* The Type M Path Table must be at a valid location (WinISO 764 * and probably other programs omit this, so we allow zero) 765 * 766 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 767 location = archive_be32dec(h+SVD_type_M_path_table_offset); 768 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 769 || location >= volume_block) 770 return (0); 771 772 /* Read Root Directory Record in Volume Descriptor. */ 773 p = h + SVD_root_directory_record_offset; 774 if (p[DR_length_offset] != 34) 775 return (0); 776 777 return (48); 778 } 779 780 static int 781 isEVD(struct iso9660 *iso9660, const unsigned char *h) 782 { 783 const unsigned char *p; 784 ssize_t logical_block_size; 785 int32_t volume_block; 786 int32_t location; 787 788 (void)iso9660; /* UNUSED */ 789 790 /* Type of the Enhanced Volume Descriptor must be 2. */ 791 if (h[PVD_type_offset] != 2) 792 return (0); 793 794 /* EVD version must be 2. */ 795 if (h[PVD_version_offset] != 2) 796 return (0); 797 798 /* Reserved field must be 0. */ 799 if (h[PVD_reserved1_offset] != 0) 800 return (0); 801 802 /* Reserved field must be 0. */ 803 if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) 804 return (0); 805 806 /* Reserved field must be 0. */ 807 if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) 808 return (0); 809 810 /* Logical block size must be > 0. */ 811 /* I've looked at Ecma 119 and can't find any stronger 812 * restriction on this field. */ 813 logical_block_size = 814 archive_le16dec(h + PVD_logical_block_size_offset); 815 if (logical_block_size <= 0) 816 return (0); 817 818 volume_block = 819 archive_le32dec(h + PVD_volume_space_size_offset); 820 if (volume_block <= SYSTEM_AREA_BLOCK+4) 821 return (0); 822 823 /* File structure version must be 2 for ISO9660:1999. */ 824 if (h[PVD_file_structure_version_offset] != 2) 825 return (0); 826 827 /* Location of Occurrence of Type L Path Table must be 828 * available location, 829 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 830 location = archive_le32dec(h+PVD_type_1_path_table_offset); 831 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 832 return (0); 833 834 /* Location of Occurrence of Type M Path Table must be 835 * available location, 836 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 837 location = archive_be32dec(h+PVD_type_m_path_table_offset); 838 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 839 || location >= volume_block) 840 return (0); 841 842 /* Reserved field must be 0. */ 843 if (!isNull(iso9660, h, PVD_reserved4_offset, PVD_reserved4_size)) 844 return (0); 845 846 /* Reserved field must be 0. */ 847 if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) 848 return (0); 849 850 /* Read Root Directory Record in Volume Descriptor. */ 851 p = h + PVD_root_directory_record_offset; 852 if (p[DR_length_offset] != 34) 853 return (0); 854 855 return (48); 856 } 857 858 static int 859 isPVD(struct iso9660 *iso9660, const unsigned char *h) 860 { 861 const unsigned char *p; 862 ssize_t logical_block_size; 863 int32_t volume_block; 864 int32_t location; 865 int i; 866 867 /* Type of the Primary Volume Descriptor must be 1. */ 868 if (h[PVD_type_offset] != 1) 869 return (0); 870 871 /* PVD version must be 1. */ 872 if (h[PVD_version_offset] != 1) 873 return (0); 874 875 /* Reserved field must be 0. */ 876 if (h[PVD_reserved1_offset] != 0) 877 return (0); 878 879 /* Reserved field must be 0. */ 880 if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) 881 return (0); 882 883 /* Reserved field must be 0. */ 884 if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) 885 return (0); 886 887 /* Logical block size must be > 0. */ 888 /* I've looked at Ecma 119 and can't find any stronger 889 * restriction on this field. */ 890 logical_block_size = 891 archive_le16dec(h + PVD_logical_block_size_offset); 892 if (logical_block_size <= 0) 893 return (0); 894 895 volume_block = archive_le32dec(h + PVD_volume_space_size_offset); 896 if (volume_block <= SYSTEM_AREA_BLOCK+4) 897 return (0); 898 899 /* File structure version must be 1 for ISO9660/ECMA119. */ 900 if (h[PVD_file_structure_version_offset] != 1) 901 return (0); 902 903 /* Location of Occurrence of Type L Path Table must be 904 * available location, 905 * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 906 location = archive_le32dec(h+PVD_type_1_path_table_offset); 907 if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) 908 return (0); 909 910 /* The Type M Path Table must also be at a valid location 911 * (although ECMA 119 requires a Type M Path Table, WinISO and 912 * probably other programs omit it, so we permit a zero here) 913 * 914 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ 915 location = archive_be32dec(h+PVD_type_m_path_table_offset); 916 if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) 917 || location >= volume_block) 918 return (0); 919 920 /* Reserved field must be 0. */ 921 /* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */ 922 for (i = 0; i < PVD_reserved4_size; ++i) 923 if (h[PVD_reserved4_offset + i] != 0 924 && h[PVD_reserved4_offset + i] != 0x20) 925 return (0); 926 927 /* Reserved field must be 0. */ 928 if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) 929 return (0); 930 931 /* XXX TODO: Check other values for sanity; reject more 932 * malformed PVDs. XXX */ 933 934 /* Read Root Directory Record in Volume Descriptor. */ 935 p = h + PVD_root_directory_record_offset; 936 if (p[DR_length_offset] != 34) 937 return (0); 938 939 if (!iso9660->primary.location) { 940 iso9660->logical_block_size = logical_block_size; 941 iso9660->volume_block = volume_block; 942 iso9660->volume_size = 943 logical_block_size * (uint64_t)volume_block; 944 iso9660->primary.location = 945 archive_le32dec(p + DR_extent_offset); 946 iso9660->primary.size = archive_le32dec(p + DR_size_offset); 947 } 948 949 return (48); 950 } 951 952 static int 953 read_children(struct archive_read *a, struct file_info *parent) 954 { 955 struct iso9660 *iso9660; 956 const unsigned char *b, *p; 957 struct file_info *multi; 958 size_t step, skip_size; 959 960 iso9660 = (struct iso9660 *)(a->format->data); 961 /* flush any remaining bytes from the last round to ensure 962 * we're positioned */ 963 if (iso9660->entry_bytes_unconsumed) { 964 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 965 iso9660->entry_bytes_unconsumed = 0; 966 } 967 if (iso9660->current_position > parent->offset) { 968 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 969 "Ignoring out-of-order directory (%s) %jd > %jd", 970 parent->name.s, 971 (intmax_t)iso9660->current_position, 972 (intmax_t)parent->offset); 973 return (ARCHIVE_WARN); 974 } 975 if (parent->offset + parent->size > iso9660->volume_size) { 976 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 977 "Directory is beyond end-of-media: %s", 978 parent->name.s); 979 return (ARCHIVE_WARN); 980 } 981 if (iso9660->current_position < parent->offset) { 982 int64_t skipsize; 983 984 skipsize = parent->offset - iso9660->current_position; 985 skipsize = __archive_read_consume(a, skipsize); 986 if (skipsize < 0) 987 return ((int)skipsize); 988 iso9660->current_position = parent->offset; 989 } 990 991 step = (size_t)(((parent->size + iso9660->logical_block_size -1) / 992 iso9660->logical_block_size) * iso9660->logical_block_size); 993 b = __archive_read_ahead(a, step, NULL); 994 if (b == NULL) { 995 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 996 "Failed to read full block when scanning " 997 "ISO9660 directory list"); 998 return (ARCHIVE_FATAL); 999 } 1000 iso9660->current_position += step; 1001 multi = NULL; 1002 skip_size = step; 1003 while (step) { 1004 p = b; 1005 b += iso9660->logical_block_size; 1006 step -= iso9660->logical_block_size; 1007 for (; *p != 0 && p < b && p + *p <= b; p += *p) { 1008 struct file_info *child; 1009 1010 /* N.B.: these special directory identifiers 1011 * are 8 bit "values" even on a 1012 * Joliet CD with UCS-2 (16bit) encoding. 1013 */ 1014 1015 /* Skip '.' entry. */ 1016 if (*(p + DR_name_len_offset) == 1 1017 && *(p + DR_name_offset) == '\0') 1018 continue; 1019 /* Skip '..' entry. */ 1020 if (*(p + DR_name_len_offset) == 1 1021 && *(p + DR_name_offset) == '\001') 1022 continue; 1023 child = parse_file_info(a, parent, p); 1024 if (child == NULL) { 1025 __archive_read_consume(a, skip_size); 1026 return (ARCHIVE_FATAL); 1027 } 1028 if (child->cl_offset == 0 && 1029 (child->multi_extent || multi != NULL)) { 1030 struct content *con; 1031 1032 if (multi == NULL) { 1033 multi = child; 1034 multi->contents.first = NULL; 1035 multi->contents.last = 1036 &(multi->contents.first); 1037 } 1038 con = malloc(sizeof(struct content)); 1039 if (con == NULL) { 1040 archive_set_error( 1041 &a->archive, ENOMEM, 1042 "No memory for multi extent"); 1043 __archive_read_consume(a, skip_size); 1044 return (ARCHIVE_FATAL); 1045 } 1046 con->offset = child->offset; 1047 con->size = child->size; 1048 con->next = NULL; 1049 *multi->contents.last = con; 1050 multi->contents.last = &(con->next); 1051 if (multi == child) { 1052 if (add_entry(a, iso9660, child) 1053 != ARCHIVE_OK) 1054 return (ARCHIVE_FATAL); 1055 } else { 1056 multi->size += child->size; 1057 if (!child->multi_extent) 1058 multi = NULL; 1059 } 1060 } else 1061 if (add_entry(a, iso9660, child) != ARCHIVE_OK) 1062 return (ARCHIVE_FATAL); 1063 } 1064 } 1065 1066 __archive_read_consume(a, skip_size); 1067 1068 /* Read data which recorded by RRIP "CE" extension. */ 1069 if (read_CE(a, iso9660) != ARCHIVE_OK) 1070 return (ARCHIVE_FATAL); 1071 1072 return (ARCHIVE_OK); 1073 } 1074 1075 static int 1076 choose_volume(struct archive_read *a, struct iso9660 *iso9660) 1077 { 1078 struct file_info *file; 1079 int64_t skipsize; 1080 struct vd *vd; 1081 const void *block; 1082 char seenJoliet; 1083 1084 vd = &(iso9660->primary); 1085 if (!iso9660->opt_support_joliet) 1086 iso9660->seenJoliet = 0; 1087 if (iso9660->seenJoliet && 1088 vd->location > iso9660->joliet.location) 1089 /* This condition is unlikely; by way of caution. */ 1090 vd = &(iso9660->joliet); 1091 1092 skipsize = LOGICAL_BLOCK_SIZE * vd->location; 1093 skipsize = __archive_read_consume(a, skipsize); 1094 if (skipsize < 0) 1095 return ((int)skipsize); 1096 iso9660->current_position = skipsize; 1097 1098 block = __archive_read_ahead(a, vd->size, NULL); 1099 if (block == NULL) { 1100 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1101 "Failed to read full block when scanning " 1102 "ISO9660 directory list"); 1103 return (ARCHIVE_FATAL); 1104 } 1105 1106 /* 1107 * While reading Root Directory, flag seenJoliet must be zero to 1108 * avoid converting special name 0x00(Current Directory) and 1109 * next byte to UCS2. 1110 */ 1111 seenJoliet = iso9660->seenJoliet;/* Save flag. */ 1112 iso9660->seenJoliet = 0; 1113 file = parse_file_info(a, NULL, block); 1114 if (file == NULL) 1115 return (ARCHIVE_FATAL); 1116 iso9660->seenJoliet = seenJoliet; 1117 1118 /* 1119 * If the iso image has both RockRidge and Joliet, we preferentially 1120 * use RockRidge Extensions rather than Joliet ones. 1121 */ 1122 if (vd == &(iso9660->primary) && iso9660->seenRockridge 1123 && iso9660->seenJoliet) 1124 iso9660->seenJoliet = 0; 1125 1126 if (vd == &(iso9660->primary) && !iso9660->seenRockridge 1127 && iso9660->seenJoliet) { 1128 /* Switch reading data from primary to joliet. */ 1129 vd = &(iso9660->joliet); 1130 skipsize = LOGICAL_BLOCK_SIZE * vd->location; 1131 skipsize -= iso9660->current_position; 1132 skipsize = __archive_read_consume(a, skipsize); 1133 if (skipsize < 0) 1134 return ((int)skipsize); 1135 iso9660->current_position += skipsize; 1136 1137 block = __archive_read_ahead(a, vd->size, NULL); 1138 if (block == NULL) { 1139 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1140 "Failed to read full block when scanning " 1141 "ISO9660 directory list"); 1142 return (ARCHIVE_FATAL); 1143 } 1144 iso9660->seenJoliet = 0; 1145 file = parse_file_info(a, NULL, block); 1146 if (file == NULL) 1147 return (ARCHIVE_FATAL); 1148 iso9660->seenJoliet = seenJoliet; 1149 } 1150 1151 /* Store the root directory in the pending list. */ 1152 if (add_entry(a, iso9660, file) != ARCHIVE_OK) 1153 return (ARCHIVE_FATAL); 1154 if (iso9660->seenRockridge) { 1155 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; 1156 a->archive.archive_format_name = 1157 "ISO9660 with Rockridge extensions"; 1158 } 1159 1160 return (ARCHIVE_OK); 1161 } 1162 1163 static int 1164 archive_read_format_iso9660_read_header(struct archive_read *a, 1165 struct archive_entry *entry) 1166 { 1167 struct iso9660 *iso9660; 1168 struct file_info *file; 1169 int r, rd_r = ARCHIVE_OK; 1170 1171 iso9660 = (struct iso9660 *)(a->format->data); 1172 1173 if (!a->archive.archive_format) { 1174 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; 1175 a->archive.archive_format_name = "ISO9660"; 1176 } 1177 1178 if (iso9660->current_position == 0) { 1179 r = choose_volume(a, iso9660); 1180 if (r != ARCHIVE_OK) 1181 return (r); 1182 } 1183 1184 file = NULL;/* Eliminate a warning. */ 1185 /* Get the next entry that appears after the current offset. */ 1186 r = next_entry_seek(a, iso9660, &file); 1187 if (r != ARCHIVE_OK) 1188 return (r); 1189 1190 if (iso9660->seenJoliet) { 1191 /* 1192 * Convert UTF-16BE of a filename to local locale MBS 1193 * and store the result into a filename field. 1194 */ 1195 if (iso9660->sconv_utf16be == NULL) { 1196 iso9660->sconv_utf16be = 1197 archive_string_conversion_from_charset( 1198 &(a->archive), "UTF-16BE", 1); 1199 if (iso9660->sconv_utf16be == NULL) 1200 /* Coundn't allocate memory */ 1201 return (ARCHIVE_FATAL); 1202 } 1203 if (iso9660->utf16be_path == NULL) { 1204 iso9660->utf16be_path = malloc(UTF16_NAME_MAX); 1205 if (iso9660->utf16be_path == NULL) { 1206 archive_set_error(&a->archive, ENOMEM, 1207 "No memory"); 1208 return (ARCHIVE_FATAL); 1209 } 1210 } 1211 if (iso9660->utf16be_previous_path == NULL) { 1212 iso9660->utf16be_previous_path = malloc(UTF16_NAME_MAX); 1213 if (iso9660->utf16be_previous_path == NULL) { 1214 archive_set_error(&a->archive, ENOMEM, 1215 "No memory"); 1216 return (ARCHIVE_FATAL); 1217 } 1218 } 1219 1220 iso9660->utf16be_path_len = 0; 1221 if (build_pathname_utf16be(iso9660->utf16be_path, 1222 UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) { 1223 archive_set_error(&a->archive, 1224 ARCHIVE_ERRNO_FILE_FORMAT, 1225 "Pathname is too long"); 1226 } 1227 1228 r = archive_entry_copy_pathname_l(entry, 1229 (const char *)iso9660->utf16be_path, 1230 iso9660->utf16be_path_len, 1231 iso9660->sconv_utf16be); 1232 if (r != 0) { 1233 if (errno == ENOMEM) { 1234 archive_set_error(&a->archive, ENOMEM, 1235 "No memory for Pathname"); 1236 return (ARCHIVE_FATAL); 1237 } 1238 archive_set_error(&a->archive, 1239 ARCHIVE_ERRNO_FILE_FORMAT, 1240 "Pathname cannot be converted " 1241 "from %s to current locale.", 1242 archive_string_conversion_charset_name( 1243 iso9660->sconv_utf16be)); 1244 1245 rd_r = ARCHIVE_WARN; 1246 } 1247 } else { 1248 archive_string_empty(&iso9660->pathname); 1249 archive_entry_set_pathname(entry, 1250 build_pathname(&iso9660->pathname, file)); 1251 } 1252 1253 iso9660->entry_bytes_remaining = file->size; 1254 /* Offset for sparse-file-aware clients. */ 1255 iso9660->entry_sparse_offset = 0; 1256 1257 if (file->offset + file->size > iso9660->volume_size) { 1258 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1259 "File is beyond end-of-media: %s", 1260 archive_entry_pathname(entry)); 1261 iso9660->entry_bytes_remaining = 0; 1262 return (ARCHIVE_WARN); 1263 } 1264 1265 /* Set up the entry structure with information about this entry. */ 1266 archive_entry_set_mode(entry, file->mode); 1267 archive_entry_set_uid(entry, file->uid); 1268 archive_entry_set_gid(entry, file->gid); 1269 archive_entry_set_nlink(entry, file->nlinks); 1270 if (file->birthtime_is_set) 1271 archive_entry_set_birthtime(entry, file->birthtime, 0); 1272 else 1273 archive_entry_unset_birthtime(entry); 1274 archive_entry_set_mtime(entry, file->mtime, 0); 1275 archive_entry_set_ctime(entry, file->ctime, 0); 1276 archive_entry_set_atime(entry, file->atime, 0); 1277 /* N.B.: Rock Ridge supports 64-bit device numbers. */ 1278 archive_entry_set_rdev(entry, (dev_t)file->rdev); 1279 archive_entry_set_size(entry, iso9660->entry_bytes_remaining); 1280 if (file->symlink.s != NULL) 1281 archive_entry_copy_symlink(entry, file->symlink.s); 1282 1283 /* Note: If the input isn't seekable, we can't rewind to 1284 * return the same body again, so if the next entry refers to 1285 * the same data, we have to return it as a hardlink to the 1286 * original entry. */ 1287 if (file->number != -1 && 1288 file->number == iso9660->previous_number) { 1289 if (iso9660->seenJoliet) { 1290 r = archive_entry_copy_hardlink_l(entry, 1291 (const char *)iso9660->utf16be_previous_path, 1292 iso9660->utf16be_previous_path_len, 1293 iso9660->sconv_utf16be); 1294 if (r != 0) { 1295 if (errno == ENOMEM) { 1296 archive_set_error(&a->archive, ENOMEM, 1297 "No memory for Linkname"); 1298 return (ARCHIVE_FATAL); 1299 } 1300 archive_set_error(&a->archive, 1301 ARCHIVE_ERRNO_FILE_FORMAT, 1302 "Linkname cannot be converted " 1303 "from %s to current locale.", 1304 archive_string_conversion_charset_name( 1305 iso9660->sconv_utf16be)); 1306 rd_r = ARCHIVE_WARN; 1307 } 1308 } else 1309 archive_entry_set_hardlink(entry, 1310 iso9660->previous_pathname.s); 1311 archive_entry_unset_size(entry); 1312 iso9660->entry_bytes_remaining = 0; 1313 return (rd_r); 1314 } 1315 1316 if ((file->mode & AE_IFMT) != AE_IFDIR && 1317 file->offset < iso9660->current_position) { 1318 int64_t r64; 1319 1320 r64 = __archive_read_seek(a, file->offset, SEEK_SET); 1321 if (r64 != (int64_t)file->offset) { 1322 /* We can't seek backwards to extract it, so issue 1323 * a warning. Note that this can only happen if 1324 * this entry was added to the heap after we passed 1325 * this offset, that is, only if the directory 1326 * mentioning this entry is later than the body of 1327 * the entry. Such layouts are very unusual; most 1328 * ISO9660 writers lay out and record all directory 1329 * information first, then store all file bodies. */ 1330 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1331 "Ignoring out-of-order file @%jx (%s) %jd < %jd", 1332 (intmax_t)file->number, 1333 iso9660->pathname.s, 1334 (intmax_t)file->offset, 1335 (intmax_t)iso9660->current_position); 1336 iso9660->entry_bytes_remaining = 0; 1337 return (ARCHIVE_WARN); 1338 } 1339 iso9660->current_position = (uint64_t)r64; 1340 } 1341 1342 /* Initialize zisofs variables. */ 1343 iso9660->entry_zisofs.pz = file->pz; 1344 if (file->pz) { 1345 #ifdef HAVE_ZLIB_H 1346 struct zisofs *zisofs; 1347 1348 zisofs = &iso9660->entry_zisofs; 1349 zisofs->initialized = 0; 1350 zisofs->pz_log2_bs = file->pz_log2_bs; 1351 zisofs->pz_uncompressed_size = file->pz_uncompressed_size; 1352 zisofs->pz_offset = 0; 1353 zisofs->header_avail = 0; 1354 zisofs->header_passed = 0; 1355 zisofs->block_pointers_avail = 0; 1356 #endif 1357 archive_entry_set_size(entry, file->pz_uncompressed_size); 1358 } 1359 1360 iso9660->previous_number = file->number; 1361 if (iso9660->seenJoliet) { 1362 memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path, 1363 iso9660->utf16be_path_len); 1364 iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len; 1365 } else 1366 archive_strcpy( 1367 &iso9660->previous_pathname, iso9660->pathname.s); 1368 1369 /* Reset entry_bytes_remaining if the file is multi extent. */ 1370 iso9660->entry_content = file->contents.first; 1371 if (iso9660->entry_content != NULL) 1372 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1373 1374 if (archive_entry_filetype(entry) == AE_IFDIR) { 1375 /* Overwrite nlinks by proper link number which is 1376 * calculated from number of sub directories. */ 1377 archive_entry_set_nlink(entry, 2 + file->subdirs); 1378 /* Directory data has been read completely. */ 1379 iso9660->entry_bytes_remaining = 0; 1380 } 1381 1382 if (rd_r != ARCHIVE_OK) 1383 return (rd_r); 1384 return (ARCHIVE_OK); 1385 } 1386 1387 static int 1388 archive_read_format_iso9660_read_data_skip(struct archive_read *a) 1389 { 1390 /* Because read_next_header always does an explicit skip 1391 * to the next entry, we don't need to do anything here. */ 1392 (void)a; /* UNUSED */ 1393 return (ARCHIVE_OK); 1394 } 1395 1396 #ifdef HAVE_ZLIB_H 1397 1398 static int 1399 zisofs_read_data(struct archive_read *a, 1400 const void **buff, size_t *size, int64_t *offset) 1401 { 1402 struct iso9660 *iso9660; 1403 struct zisofs *zisofs; 1404 const unsigned char *p; 1405 size_t avail; 1406 ssize_t bytes_read; 1407 size_t uncompressed_size; 1408 int r; 1409 1410 iso9660 = (struct iso9660 *)(a->format->data); 1411 zisofs = &iso9660->entry_zisofs; 1412 1413 p = __archive_read_ahead(a, 1, &bytes_read); 1414 if (bytes_read <= 0) { 1415 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1416 "Truncated zisofs file body"); 1417 return (ARCHIVE_FATAL); 1418 } 1419 if (bytes_read > iso9660->entry_bytes_remaining) 1420 bytes_read = (ssize_t)iso9660->entry_bytes_remaining; 1421 avail = bytes_read; 1422 uncompressed_size = 0; 1423 1424 if (!zisofs->initialized) { 1425 size_t ceil, xsize; 1426 1427 /* Allocate block pointers buffer. */ 1428 ceil = (size_t)((zisofs->pz_uncompressed_size + 1429 (((int64_t)1) << zisofs->pz_log2_bs) - 1) 1430 >> zisofs->pz_log2_bs); 1431 xsize = (ceil + 1) * 4; 1432 if (zisofs->block_pointers_alloc < xsize) { 1433 size_t alloc; 1434 1435 if (zisofs->block_pointers != NULL) 1436 free(zisofs->block_pointers); 1437 alloc = ((xsize >> 10) + 1) << 10; 1438 zisofs->block_pointers = malloc(alloc); 1439 if (zisofs->block_pointers == NULL) { 1440 archive_set_error(&a->archive, ENOMEM, 1441 "No memory for zisofs decompression"); 1442 return (ARCHIVE_FATAL); 1443 } 1444 zisofs->block_pointers_alloc = alloc; 1445 } 1446 zisofs->block_pointers_size = xsize; 1447 1448 /* Allocate uncompressed data buffer. */ 1449 xsize = (size_t)1UL << zisofs->pz_log2_bs; 1450 if (zisofs->uncompressed_buffer_size < xsize) { 1451 if (zisofs->uncompressed_buffer != NULL) 1452 free(zisofs->uncompressed_buffer); 1453 zisofs->uncompressed_buffer = malloc(xsize); 1454 if (zisofs->uncompressed_buffer == NULL) { 1455 archive_set_error(&a->archive, ENOMEM, 1456 "No memory for zisofs decompression"); 1457 return (ARCHIVE_FATAL); 1458 } 1459 } 1460 zisofs->uncompressed_buffer_size = xsize; 1461 1462 /* 1463 * Read the file header, and check the magic code of zisofs. 1464 */ 1465 if (zisofs->header_avail < sizeof(zisofs->header)) { 1466 xsize = sizeof(zisofs->header) - zisofs->header_avail; 1467 if (avail < xsize) 1468 xsize = avail; 1469 memcpy(zisofs->header + zisofs->header_avail, p, xsize); 1470 zisofs->header_avail += xsize; 1471 avail -= xsize; 1472 p += xsize; 1473 } 1474 if (!zisofs->header_passed && 1475 zisofs->header_avail == sizeof(zisofs->header)) { 1476 int err = 0; 1477 1478 if (memcmp(zisofs->header, zisofs_magic, 1479 sizeof(zisofs_magic)) != 0) 1480 err = 1; 1481 if (archive_le32dec(zisofs->header + 8) 1482 != zisofs->pz_uncompressed_size) 1483 err = 1; 1484 if (zisofs->header[12] != 4) 1485 err = 1; 1486 if (zisofs->header[13] != zisofs->pz_log2_bs) 1487 err = 1; 1488 if (err) { 1489 archive_set_error(&a->archive, 1490 ARCHIVE_ERRNO_FILE_FORMAT, 1491 "Illegal zisofs file body"); 1492 return (ARCHIVE_FATAL); 1493 } 1494 zisofs->header_passed = 1; 1495 } 1496 /* 1497 * Read block pointers. 1498 */ 1499 if (zisofs->header_passed && 1500 zisofs->block_pointers_avail < zisofs->block_pointers_size) { 1501 xsize = zisofs->block_pointers_size 1502 - zisofs->block_pointers_avail; 1503 if (avail < xsize) 1504 xsize = avail; 1505 memcpy(zisofs->block_pointers 1506 + zisofs->block_pointers_avail, p, xsize); 1507 zisofs->block_pointers_avail += xsize; 1508 avail -= xsize; 1509 p += xsize; 1510 if (zisofs->block_pointers_avail 1511 == zisofs->block_pointers_size) { 1512 /* We've got all block pointers and initialize 1513 * related variables. */ 1514 zisofs->block_off = 0; 1515 zisofs->block_avail = 0; 1516 /* Complete a initialization */ 1517 zisofs->initialized = 1; 1518 } 1519 } 1520 1521 if (!zisofs->initialized) 1522 goto next_data; /* We need more data. */ 1523 } 1524 1525 /* 1526 * Get block offsets from block pointers. 1527 */ 1528 if (zisofs->block_avail == 0) { 1529 uint32_t bst, bed; 1530 1531 if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { 1532 /* There isn't a pair of offsets. */ 1533 archive_set_error(&a->archive, 1534 ARCHIVE_ERRNO_FILE_FORMAT, 1535 "Illegal zisofs block pointers"); 1536 return (ARCHIVE_FATAL); 1537 } 1538 bst = archive_le32dec( 1539 zisofs->block_pointers + zisofs->block_off); 1540 if (bst != zisofs->pz_offset + (bytes_read - avail)) { 1541 /* TODO: Should we seek offset of current file 1542 * by bst ? */ 1543 archive_set_error(&a->archive, 1544 ARCHIVE_ERRNO_FILE_FORMAT, 1545 "Illegal zisofs block pointers(cannot seek)"); 1546 return (ARCHIVE_FATAL); 1547 } 1548 bed = archive_le32dec( 1549 zisofs->block_pointers + zisofs->block_off + 4); 1550 if (bed < bst) { 1551 archive_set_error(&a->archive, 1552 ARCHIVE_ERRNO_FILE_FORMAT, 1553 "Illegal zisofs block pointers"); 1554 return (ARCHIVE_FATAL); 1555 } 1556 zisofs->block_avail = bed - bst; 1557 zisofs->block_off += 4; 1558 1559 /* Initialize compression library for new block. */ 1560 if (zisofs->stream_valid) 1561 r = inflateReset(&zisofs->stream); 1562 else 1563 r = inflateInit(&zisofs->stream); 1564 if (r != Z_OK) { 1565 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1566 "Can't initialize zisofs decompression."); 1567 return (ARCHIVE_FATAL); 1568 } 1569 zisofs->stream_valid = 1; 1570 zisofs->stream.total_in = 0; 1571 zisofs->stream.total_out = 0; 1572 } 1573 1574 /* 1575 * Make uncompressed data. 1576 */ 1577 if (zisofs->block_avail == 0) { 1578 memset(zisofs->uncompressed_buffer, 0, 1579 zisofs->uncompressed_buffer_size); 1580 uncompressed_size = zisofs->uncompressed_buffer_size; 1581 } else { 1582 zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; 1583 if (avail > zisofs->block_avail) 1584 zisofs->stream.avail_in = zisofs->block_avail; 1585 else 1586 zisofs->stream.avail_in = (uInt)avail; 1587 zisofs->stream.next_out = zisofs->uncompressed_buffer; 1588 zisofs->stream.avail_out = 1589 (uInt)zisofs->uncompressed_buffer_size; 1590 1591 r = inflate(&zisofs->stream, 0); 1592 switch (r) { 1593 case Z_OK: /* Decompressor made some progress.*/ 1594 case Z_STREAM_END: /* Found end of stream. */ 1595 break; 1596 default: 1597 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1598 "zisofs decompression failed (%d)", r); 1599 return (ARCHIVE_FATAL); 1600 } 1601 uncompressed_size = 1602 zisofs->uncompressed_buffer_size - zisofs->stream.avail_out; 1603 avail -= zisofs->stream.next_in - p; 1604 zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p); 1605 } 1606 next_data: 1607 bytes_read -= avail; 1608 *buff = zisofs->uncompressed_buffer; 1609 *size = uncompressed_size; 1610 *offset = iso9660->entry_sparse_offset; 1611 iso9660->entry_sparse_offset += uncompressed_size; 1612 iso9660->entry_bytes_remaining -= bytes_read; 1613 iso9660->current_position += bytes_read; 1614 zisofs->pz_offset += (uint32_t)bytes_read; 1615 iso9660->entry_bytes_unconsumed += bytes_read; 1616 1617 return (ARCHIVE_OK); 1618 } 1619 1620 #else /* HAVE_ZLIB_H */ 1621 1622 static int 1623 zisofs_read_data(struct archive_read *a, 1624 const void **buff, size_t *size, int64_t *offset) 1625 { 1626 1627 (void)buff;/* UNUSED */ 1628 (void)size;/* UNUSED */ 1629 (void)offset;/* UNUSED */ 1630 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1631 "zisofs is not supported on this platform."); 1632 return (ARCHIVE_FAILED); 1633 } 1634 1635 #endif /* HAVE_ZLIB_H */ 1636 1637 static int 1638 archive_read_format_iso9660_read_data(struct archive_read *a, 1639 const void **buff, size_t *size, int64_t *offset) 1640 { 1641 ssize_t bytes_read; 1642 struct iso9660 *iso9660; 1643 1644 iso9660 = (struct iso9660 *)(a->format->data); 1645 1646 if (iso9660->entry_bytes_unconsumed) { 1647 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 1648 iso9660->entry_bytes_unconsumed = 0; 1649 } 1650 1651 if (iso9660->entry_bytes_remaining <= 0) { 1652 if (iso9660->entry_content != NULL) 1653 iso9660->entry_content = iso9660->entry_content->next; 1654 if (iso9660->entry_content == NULL) { 1655 *buff = NULL; 1656 *size = 0; 1657 *offset = iso9660->entry_sparse_offset; 1658 return (ARCHIVE_EOF); 1659 } 1660 /* Seek forward to the start of the entry. */ 1661 if (iso9660->current_position < iso9660->entry_content->offset) { 1662 int64_t step; 1663 1664 step = iso9660->entry_content->offset - 1665 iso9660->current_position; 1666 step = __archive_read_consume(a, step); 1667 if (step < 0) 1668 return ((int)step); 1669 iso9660->current_position = 1670 iso9660->entry_content->offset; 1671 } 1672 if (iso9660->entry_content->offset < iso9660->current_position) { 1673 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1674 "Ignoring out-of-order file (%s) %jd < %jd", 1675 iso9660->pathname.s, 1676 (intmax_t)iso9660->entry_content->offset, 1677 (intmax_t)iso9660->current_position); 1678 *buff = NULL; 1679 *size = 0; 1680 *offset = iso9660->entry_sparse_offset; 1681 return (ARCHIVE_WARN); 1682 } 1683 iso9660->entry_bytes_remaining = iso9660->entry_content->size; 1684 } 1685 if (iso9660->entry_zisofs.pz) 1686 return (zisofs_read_data(a, buff, size, offset)); 1687 1688 *buff = __archive_read_ahead(a, 1, &bytes_read); 1689 if (bytes_read == 0) 1690 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1691 "Truncated input file"); 1692 if (*buff == NULL) 1693 return (ARCHIVE_FATAL); 1694 if (bytes_read > iso9660->entry_bytes_remaining) 1695 bytes_read = (ssize_t)iso9660->entry_bytes_remaining; 1696 *size = bytes_read; 1697 *offset = iso9660->entry_sparse_offset; 1698 iso9660->entry_sparse_offset += bytes_read; 1699 iso9660->entry_bytes_remaining -= bytes_read; 1700 iso9660->entry_bytes_unconsumed = bytes_read; 1701 iso9660->current_position += bytes_read; 1702 return (ARCHIVE_OK); 1703 } 1704 1705 static int 1706 archive_read_format_iso9660_cleanup(struct archive_read *a) 1707 { 1708 struct iso9660 *iso9660; 1709 int r = ARCHIVE_OK; 1710 1711 iso9660 = (struct iso9660 *)(a->format->data); 1712 release_files(iso9660); 1713 free(iso9660->read_ce_req.reqs); 1714 archive_string_free(&iso9660->pathname); 1715 archive_string_free(&iso9660->previous_pathname); 1716 if (iso9660->pending_files.files) 1717 free(iso9660->pending_files.files); 1718 #ifdef HAVE_ZLIB_H 1719 free(iso9660->entry_zisofs.uncompressed_buffer); 1720 free(iso9660->entry_zisofs.block_pointers); 1721 if (iso9660->entry_zisofs.stream_valid) { 1722 if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) { 1723 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1724 "Failed to clean up zlib decompressor"); 1725 r = ARCHIVE_FATAL; 1726 } 1727 } 1728 #endif 1729 free(iso9660->utf16be_path); 1730 free(iso9660->utf16be_previous_path); 1731 free(iso9660); 1732 (a->format->data) = NULL; 1733 return (r); 1734 } 1735 1736 /* 1737 * This routine parses a single ISO directory record, makes sense 1738 * of any extensions, and stores the result in memory. 1739 */ 1740 static struct file_info * 1741 parse_file_info(struct archive_read *a, struct file_info *parent, 1742 const unsigned char *isodirrec) 1743 { 1744 struct iso9660 *iso9660; 1745 struct file_info *file; 1746 size_t name_len; 1747 const unsigned char *rr_start, *rr_end; 1748 const unsigned char *p; 1749 size_t dr_len; 1750 uint64_t fsize; 1751 int32_t location; 1752 int flags; 1753 1754 iso9660 = (struct iso9660 *)(a->format->data); 1755 1756 dr_len = (size_t)isodirrec[DR_length_offset]; 1757 name_len = (size_t)isodirrec[DR_name_len_offset]; 1758 location = archive_le32dec(isodirrec + DR_extent_offset); 1759 fsize = toi(isodirrec + DR_size_offset, DR_size_size); 1760 /* Sanity check that dr_len needs at least 34. */ 1761 if (dr_len < 34) { 1762 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1763 "Invalid length of directory record"); 1764 return (NULL); 1765 } 1766 /* Sanity check that name_len doesn't exceed dr_len. */ 1767 if (dr_len - 33 < name_len || name_len == 0) { 1768 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1769 "Invalid length of file identifier"); 1770 return (NULL); 1771 } 1772 /* Sanity check that location doesn't exceed volume block. 1773 * Don't check lower limit of location; it's possibility 1774 * the location has negative value when file type is symbolic 1775 * link or file size is zero. As far as I know latest mkisofs 1776 * do that. 1777 */ 1778 if (location > 0 && 1779 (location + ((fsize + iso9660->logical_block_size -1) 1780 / iso9660->logical_block_size)) 1781 > (uint32_t)iso9660->volume_block) { 1782 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1783 "Invalid location of extent of file"); 1784 return (NULL); 1785 } 1786 /* Sanity check that location doesn't have a negative value 1787 * when the file is not empty. it's too large. */ 1788 if (fsize != 0 && location < 0) { 1789 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 1790 "Invalid location of extent of file"); 1791 return (NULL); 1792 } 1793 1794 /* Create a new file entry and copy data from the ISO dir record. */ 1795 file = (struct file_info *)calloc(1, sizeof(*file)); 1796 if (file == NULL) { 1797 archive_set_error(&a->archive, ENOMEM, 1798 "No memory for file entry"); 1799 return (NULL); 1800 } 1801 file->parent = parent; 1802 file->offset = iso9660->logical_block_size * (uint64_t)location; 1803 file->size = fsize; 1804 file->mtime = isodate7(isodirrec + DR_date_offset); 1805 file->ctime = file->atime = file->mtime; 1806 file->rede_files.first = NULL; 1807 file->rede_files.last = &(file->rede_files.first); 1808 1809 p = isodirrec + DR_name_offset; 1810 /* Rockridge extensions (if any) follow name. Compute this 1811 * before fidgeting the name_len below. */ 1812 rr_start = p + name_len + (name_len & 1 ? 0 : 1); 1813 rr_end = isodirrec + dr_len; 1814 1815 if (iso9660->seenJoliet) { 1816 /* Joliet names are max 64 chars (128 bytes) according to spec, 1817 * but genisoimage/mkisofs allows recording longer Joliet 1818 * names which are 103 UCS2 characters(206 bytes) by their 1819 * option '-joliet-long'. 1820 */ 1821 if (name_len > 206) 1822 name_len = 206; 1823 name_len &= ~1; 1824 1825 /* trim trailing first version and dot from filename. 1826 * 1827 * Remember we were in UTF-16BE land! 1828 * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both 1829 * 16 bits big endian characters on Joliet. 1830 * 1831 * TODO: sanitize filename? 1832 * Joliet allows any UCS-2 char except: 1833 * *, /, :, ;, ? and \. 1834 */ 1835 /* Chop off trailing ';1' from files. */ 1836 if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';' 1837 && p[name_len-2] == 0 && p[name_len-1] == '1') 1838 name_len -= 4; 1839 #if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */ 1840 /* Chop off trailing '.' from filenames. */ 1841 if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.') 1842 name_len -= 2; 1843 #endif 1844 if ((file->utf16be_name = malloc(name_len)) == NULL) { 1845 archive_set_error(&a->archive, ENOMEM, 1846 "No memory for file name"); 1847 return (NULL); 1848 } 1849 memcpy(file->utf16be_name, p, name_len); 1850 file->utf16be_bytes = name_len; 1851 } else { 1852 /* Chop off trailing ';1' from files. */ 1853 if (name_len > 2 && p[name_len - 2] == ';' && 1854 p[name_len - 1] == '1') 1855 name_len -= 2; 1856 /* Chop off trailing '.' from filenames. */ 1857 if (name_len > 1 && p[name_len - 1] == '.') 1858 --name_len; 1859 1860 archive_strncpy(&file->name, (const char *)p, name_len); 1861 } 1862 1863 flags = isodirrec[DR_flags_offset]; 1864 if (flags & 0x02) 1865 file->mode = AE_IFDIR | 0700; 1866 else 1867 file->mode = AE_IFREG | 0400; 1868 if (flags & 0x80) 1869 file->multi_extent = 1; 1870 else 1871 file->multi_extent = 0; 1872 /* 1873 * Use a location for the file number, which is treated as an inode 1874 * number to find out hardlink target. If Rockridge extensions is 1875 * being used, the file number will be overwritten by FILE SERIAL 1876 * NUMBER of RRIP "PX" extension. 1877 * Note: Old mkisofs did not record that FILE SERIAL NUMBER 1878 * in ISO images. 1879 * Note2: xorriso set 0 to the location of a symlink file. 1880 */ 1881 if (file->size == 0 && location >= 0) { 1882 /* If file->size is zero, its location points wrong place, 1883 * and so we should not use it for the file number. 1884 * When the location has negative value, it can be used 1885 * for the file number. 1886 */ 1887 file->number = -1; 1888 /* Do not appear before any directory entries. */ 1889 file->offset = -1; 1890 } else 1891 file->number = (int64_t)(uint32_t)location; 1892 1893 /* Rockridge extensions overwrite information from above. */ 1894 if (iso9660->opt_support_rockridge) { 1895 if (parent == NULL && rr_end - rr_start >= 7) { 1896 p = rr_start; 1897 if (memcmp(p, "SP\x07\x01\xbe\xef", 6) == 0) { 1898 /* 1899 * SP extension stores the suspOffset 1900 * (Number of bytes to skip between 1901 * filename and SUSP records.) 1902 * It is mandatory by the SUSP standard 1903 * (IEEE 1281). 1904 * 1905 * It allows SUSP to coexist with 1906 * non-SUSP uses of the System 1907 * Use Area by placing non-SUSP data 1908 * before SUSP data. 1909 * 1910 * SP extension must be in the root 1911 * directory entry, disable all SUSP 1912 * processing if not found. 1913 */ 1914 iso9660->suspOffset = p[6]; 1915 iso9660->seenSUSP = 1; 1916 rr_start += 7; 1917 } 1918 } 1919 if (iso9660->seenSUSP) { 1920 int r; 1921 1922 file->name_continues = 0; 1923 file->symlink_continues = 0; 1924 rr_start += iso9660->suspOffset; 1925 r = parse_rockridge(a, file, rr_start, rr_end); 1926 if (r != ARCHIVE_OK) { 1927 free(file); 1928 return (NULL); 1929 } 1930 /* 1931 * A file size of symbolic link files in ISO images 1932 * made by makefs is not zero and its location is 1933 * the same as those of next regular file. That is 1934 * the same as hard like file and it causes unexpected 1935 * error. 1936 */ 1937 if (file->size > 0 && 1938 (file->mode & AE_IFMT) == AE_IFLNK) { 1939 file->size = 0; 1940 file->number = -1; 1941 file->offset = -1; 1942 } 1943 } else 1944 /* If there isn't SUSP, disable parsing 1945 * rock ridge extensions. */ 1946 iso9660->opt_support_rockridge = 0; 1947 } 1948 1949 file->nlinks = 1;/* Reset nlink. we'll calculate it later. */ 1950 /* Tell file's parent how many children that parent has. */ 1951 if (parent != NULL && (flags & 0x02)) 1952 parent->subdirs++; 1953 1954 if (iso9660->seenRockridge) { 1955 if (parent != NULL && parent->parent == NULL && 1956 (flags & 0x02) && iso9660->rr_moved == NULL && 1957 file->name.s && 1958 (strcmp(file->name.s, "rr_moved") == 0 || 1959 strcmp(file->name.s, ".rr_moved") == 0)) { 1960 iso9660->rr_moved = file; 1961 file->rr_moved = 1; 1962 file->rr_moved_has_re_only = 1; 1963 file->re = 0; 1964 parent->subdirs--; 1965 } else if (file->re) { 1966 /* 1967 * Sanity check: file's parent is rr_moved. 1968 */ 1969 if (parent == NULL || parent->rr_moved == 0) { 1970 archive_set_error(&a->archive, 1971 ARCHIVE_ERRNO_MISC, 1972 "Invalid Rockridge RE"); 1973 return (NULL); 1974 } 1975 /* 1976 * Sanity check: file does not have "CL" extension. 1977 */ 1978 if (file->cl_offset) { 1979 archive_set_error(&a->archive, 1980 ARCHIVE_ERRNO_MISC, 1981 "Invalid Rockridge RE and CL"); 1982 return (NULL); 1983 } 1984 /* 1985 * Sanity check: The file type must be a directory. 1986 */ 1987 if ((flags & 0x02) == 0) { 1988 archive_set_error(&a->archive, 1989 ARCHIVE_ERRNO_MISC, 1990 "Invalid Rockridge RE"); 1991 return (NULL); 1992 } 1993 } else if (parent != NULL && parent->rr_moved) 1994 file->rr_moved_has_re_only = 0; 1995 else if (parent != NULL && (flags & 0x02) && 1996 (parent->re || parent->re_descendant)) 1997 file->re_descendant = 1; 1998 if (file->cl_offset) { 1999 struct file_info *r; 2000 2001 if (parent == NULL || parent->parent == NULL) { 2002 archive_set_error(&a->archive, 2003 ARCHIVE_ERRNO_MISC, 2004 "Invalid Rockridge CL"); 2005 return (NULL); 2006 } 2007 /* 2008 * Sanity check: The file type must be a regular file. 2009 */ 2010 if ((flags & 0x02) != 0) { 2011 archive_set_error(&a->archive, 2012 ARCHIVE_ERRNO_MISC, 2013 "Invalid Rockridge CL"); 2014 return (NULL); 2015 } 2016 parent->subdirs++; 2017 /* Overwrite an offset and a number of this "CL" entry 2018 * to appear before other dirs. "+1" to those is to 2019 * make sure to appear after "RE" entry which this 2020 * "CL" entry should be connected with. */ 2021 file->offset = file->number = file->cl_offset + 1; 2022 2023 /* 2024 * Sanity check: cl_offset does not point at its 2025 * the parents or itself. 2026 */ 2027 for (r = parent; r; r = r->parent) { 2028 if (r->offset == file->cl_offset) { 2029 archive_set_error(&a->archive, 2030 ARCHIVE_ERRNO_MISC, 2031 "Invalid Rockridge CL"); 2032 return (NULL); 2033 } 2034 } 2035 if (file->cl_offset == file->offset || 2036 parent->rr_moved) { 2037 archive_set_error(&a->archive, 2038 ARCHIVE_ERRNO_MISC, 2039 "Invalid Rockridge CL"); 2040 return (NULL); 2041 } 2042 } 2043 } 2044 2045 #if DEBUG 2046 /* DEBUGGING: Warn about attributes I don't yet fully support. */ 2047 if ((flags & ~0x02) != 0) { 2048 fprintf(stderr, "\n ** Unrecognized flag: "); 2049 dump_isodirrec(stderr, isodirrec); 2050 fprintf(stderr, "\n"); 2051 } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { 2052 fprintf(stderr, "\n ** Unrecognized sequence number: "); 2053 dump_isodirrec(stderr, isodirrec); 2054 fprintf(stderr, "\n"); 2055 } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { 2056 fprintf(stderr, "\n ** Unexpected file unit size: "); 2057 dump_isodirrec(stderr, isodirrec); 2058 fprintf(stderr, "\n"); 2059 } else if (*(isodirrec + DR_interleave_offset) != 0) { 2060 fprintf(stderr, "\n ** Unexpected interleave: "); 2061 dump_isodirrec(stderr, isodirrec); 2062 fprintf(stderr, "\n"); 2063 } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { 2064 fprintf(stderr, "\n ** Unexpected extended attribute length: "); 2065 dump_isodirrec(stderr, isodirrec); 2066 fprintf(stderr, "\n"); 2067 } 2068 #endif 2069 register_file(iso9660, file); 2070 return (file); 2071 } 2072 2073 static int 2074 parse_rockridge(struct archive_read *a, struct file_info *file, 2075 const unsigned char *p, const unsigned char *end) 2076 { 2077 struct iso9660 *iso9660; 2078 2079 iso9660 = (struct iso9660 *)(a->format->data); 2080 2081 while (p + 4 <= end /* Enough space for another entry. */ 2082 && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ 2083 && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ 2084 && p[2] >= 4 /* Sanity-check length. */ 2085 && p + p[2] <= end) { /* Sanity-check length. */ 2086 const unsigned char *data = p + 4; 2087 int data_length = p[2] - 4; 2088 int version = p[3]; 2089 2090 switch(p[0]) { 2091 case 'C': 2092 if (p[1] == 'E') { 2093 if (version == 1 && data_length == 24) { 2094 /* 2095 * CE extension comprises: 2096 * 8 byte sector containing extension 2097 * 8 byte offset w/in above sector 2098 * 8 byte length of continuation 2099 */ 2100 int32_t location = 2101 archive_le32dec(data); 2102 file->ce_offset = 2103 archive_le32dec(data+8); 2104 file->ce_size = 2105 archive_le32dec(data+16); 2106 if (register_CE(a, location, file) 2107 != ARCHIVE_OK) 2108 return (ARCHIVE_FATAL); 2109 } 2110 } 2111 else if (p[1] == 'L') { 2112 if (version == 1 && data_length == 8) { 2113 file->cl_offset = (uint64_t) 2114 iso9660->logical_block_size * 2115 (uint64_t)archive_le32dec(data); 2116 iso9660->seenRockridge = 1; 2117 } 2118 } 2119 break; 2120 case 'N': 2121 if (p[1] == 'M') { 2122 if (version == 1) { 2123 parse_rockridge_NM1(file, 2124 data, data_length); 2125 iso9660->seenRockridge = 1; 2126 } 2127 } 2128 break; 2129 case 'P': 2130 /* 2131 * PD extension is padding; 2132 * contents are always ignored. 2133 * 2134 * PL extension won't appear; 2135 * contents are always ignored. 2136 */ 2137 if (p[1] == 'N') { 2138 if (version == 1 && data_length == 16) { 2139 file->rdev = toi(data,4); 2140 file->rdev <<= 32; 2141 file->rdev |= toi(data + 8, 4); 2142 iso9660->seenRockridge = 1; 2143 } 2144 } 2145 else if (p[1] == 'X') { 2146 /* 2147 * PX extension comprises: 2148 * 8 bytes for mode, 2149 * 8 bytes for nlinks, 2150 * 8 bytes for uid, 2151 * 8 bytes for gid, 2152 * 8 bytes for inode. 2153 */ 2154 if (version == 1) { 2155 if (data_length >= 8) 2156 file->mode 2157 = toi(data, 4); 2158 if (data_length >= 16) 2159 file->nlinks 2160 = toi(data + 8, 4); 2161 if (data_length >= 24) 2162 file->uid 2163 = toi(data + 16, 4); 2164 if (data_length >= 32) 2165 file->gid 2166 = toi(data + 24, 4); 2167 if (data_length >= 40) 2168 file->number 2169 = toi(data + 32, 4); 2170 iso9660->seenRockridge = 1; 2171 } 2172 } 2173 break; 2174 case 'R': 2175 if (p[1] == 'E' && version == 1) { 2176 file->re = 1; 2177 iso9660->seenRockridge = 1; 2178 } 2179 else if (p[1] == 'R' && version == 1) { 2180 /* 2181 * RR extension comprises: 2182 * one byte flag value 2183 * This extension is obsolete, 2184 * so contents are always ignored. 2185 */ 2186 } 2187 break; 2188 case 'S': 2189 if (p[1] == 'L') { 2190 if (version == 1) { 2191 parse_rockridge_SL1(file, 2192 data, data_length); 2193 iso9660->seenRockridge = 1; 2194 } 2195 } 2196 else if (p[1] == 'T' 2197 && data_length == 0 && version == 1) { 2198 /* 2199 * ST extension marks end of this 2200 * block of SUSP entries. 2201 * 2202 * It allows SUSP to coexist with 2203 * non-SUSP uses of the System 2204 * Use Area by placing non-SUSP data 2205 * after SUSP data. 2206 */ 2207 iso9660->seenSUSP = 0; 2208 iso9660->seenRockridge = 0; 2209 return (ARCHIVE_OK); 2210 } 2211 break; 2212 case 'T': 2213 if (p[1] == 'F') { 2214 if (version == 1) { 2215 parse_rockridge_TF1(file, 2216 data, data_length); 2217 iso9660->seenRockridge = 1; 2218 } 2219 } 2220 break; 2221 case 'Z': 2222 if (p[1] == 'F') { 2223 if (version == 1) 2224 parse_rockridge_ZF1(file, 2225 data, data_length); 2226 } 2227 break; 2228 default: 2229 break; 2230 } 2231 2232 p += p[2]; 2233 } 2234 return (ARCHIVE_OK); 2235 } 2236 2237 static int 2238 register_CE(struct archive_read *a, int32_t location, 2239 struct file_info *file) 2240 { 2241 struct iso9660 *iso9660; 2242 struct read_ce_queue *heap; 2243 struct read_ce_req *p; 2244 uint64_t offset, parent_offset; 2245 int hole, parent; 2246 2247 iso9660 = (struct iso9660 *)(a->format->data); 2248 offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size; 2249 if (((file->mode & AE_IFMT) == AE_IFREG && 2250 offset >= file->offset) || 2251 offset < iso9660->current_position || 2252 (((uint64_t)file->ce_offset) + file->ce_size) 2253 > (uint64_t)iso9660->logical_block_size || 2254 offset + file->ce_offset + file->ce_size 2255 > iso9660->volume_size) { 2256 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2257 "Invalid parameter in SUSP \"CE\" extension"); 2258 return (ARCHIVE_FATAL); 2259 } 2260 2261 /* Expand our CE list as necessary. */ 2262 heap = &(iso9660->read_ce_req); 2263 if (heap->cnt >= heap->allocated) { 2264 int new_size; 2265 2266 if (heap->allocated < 16) 2267 new_size = 16; 2268 else 2269 new_size = heap->allocated * 2; 2270 /* Overflow might keep us from growing the list. */ 2271 if (new_size <= heap->allocated) { 2272 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2273 return (ARCHIVE_FATAL); 2274 } 2275 p = calloc(new_size, sizeof(p[0])); 2276 if (p == NULL) { 2277 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 2278 return (ARCHIVE_FATAL); 2279 } 2280 if (heap->reqs != NULL) { 2281 memcpy(p, heap->reqs, heap->cnt * sizeof(*p)); 2282 free(heap->reqs); 2283 } 2284 heap->reqs = p; 2285 heap->allocated = new_size; 2286 } 2287 2288 /* 2289 * Start with hole at end, walk it up tree to find insertion point. 2290 */ 2291 hole = heap->cnt++; 2292 while (hole > 0) { 2293 parent = (hole - 1)/2; 2294 parent_offset = heap->reqs[parent].offset; 2295 if (offset >= parent_offset) { 2296 heap->reqs[hole].offset = offset; 2297 heap->reqs[hole].file = file; 2298 return (ARCHIVE_OK); 2299 } 2300 /* Move parent into hole <==> move hole up tree. */ 2301 heap->reqs[hole] = heap->reqs[parent]; 2302 hole = parent; 2303 } 2304 heap->reqs[0].offset = offset; 2305 heap->reqs[0].file = file; 2306 return (ARCHIVE_OK); 2307 } 2308 2309 static void 2310 next_CE(struct read_ce_queue *heap) 2311 { 2312 uint64_t a_offset, b_offset, c_offset; 2313 int a, b, c; 2314 struct read_ce_req tmp; 2315 2316 if (heap->cnt < 1) 2317 return; 2318 2319 /* 2320 * Move the last item in the heap to the root of the tree 2321 */ 2322 heap->reqs[0] = heap->reqs[--(heap->cnt)]; 2323 2324 /* 2325 * Rebalance the heap. 2326 */ 2327 a = 0; /* Starting element and its offset */ 2328 a_offset = heap->reqs[a].offset; 2329 for (;;) { 2330 b = a + a + 1; /* First child */ 2331 if (b >= heap->cnt) 2332 return; 2333 b_offset = heap->reqs[b].offset; 2334 c = b + 1; /* Use second child if it is smaller. */ 2335 if (c < heap->cnt) { 2336 c_offset = heap->reqs[c].offset; 2337 if (c_offset < b_offset) { 2338 b = c; 2339 b_offset = c_offset; 2340 } 2341 } 2342 if (a_offset <= b_offset) 2343 return; 2344 tmp = heap->reqs[a]; 2345 heap->reqs[a] = heap->reqs[b]; 2346 heap->reqs[b] = tmp; 2347 a = b; 2348 } 2349 } 2350 2351 2352 static int 2353 read_CE(struct archive_read *a, struct iso9660 *iso9660) 2354 { 2355 struct read_ce_queue *heap; 2356 const unsigned char *b, *p, *end; 2357 struct file_info *file; 2358 size_t step; 2359 int r; 2360 2361 /* Read data which RRIP "CE" extension points. */ 2362 heap = &(iso9660->read_ce_req); 2363 step = iso9660->logical_block_size; 2364 while (heap->cnt && 2365 heap->reqs[0].offset == iso9660->current_position) { 2366 b = __archive_read_ahead(a, step, NULL); 2367 if (b == NULL) { 2368 archive_set_error(&a->archive, 2369 ARCHIVE_ERRNO_MISC, 2370 "Failed to read full block when scanning " 2371 "ISO9660 directory list"); 2372 return (ARCHIVE_FATAL); 2373 } 2374 do { 2375 file = heap->reqs[0].file; 2376 if (file->ce_offset + file->ce_size > step) { 2377 archive_set_error(&a->archive, 2378 ARCHIVE_ERRNO_FILE_FORMAT, 2379 "Malformed CE information"); 2380 return (ARCHIVE_FATAL); 2381 } 2382 p = b + file->ce_offset; 2383 end = p + file->ce_size; 2384 next_CE(heap); 2385 r = parse_rockridge(a, file, p, end); 2386 if (r != ARCHIVE_OK) 2387 return (ARCHIVE_FATAL); 2388 } while (heap->cnt && 2389 heap->reqs[0].offset == iso9660->current_position); 2390 /* NOTE: Do not move this consume's code to fron of 2391 * do-while loop. Registration of nested CE extension 2392 * might cause error because of current position. */ 2393 __archive_read_consume(a, step); 2394 iso9660->current_position += step; 2395 } 2396 return (ARCHIVE_OK); 2397 } 2398 2399 static void 2400 parse_rockridge_NM1(struct file_info *file, 2401 const unsigned char *data, int data_length) 2402 { 2403 if (!file->name_continues) 2404 archive_string_empty(&file->name); 2405 file->name_continues = 0; 2406 if (data_length < 1) 2407 return; 2408 /* 2409 * NM version 1 extension comprises: 2410 * 1 byte flag, value is one of: 2411 * = 0: remainder is name 2412 * = 1: remainder is name, next NM entry continues name 2413 * = 2: "." 2414 * = 4: ".." 2415 * = 32: Implementation specific 2416 * All other values are reserved. 2417 */ 2418 switch(data[0]) { 2419 case 0: 2420 if (data_length < 2) 2421 return; 2422 archive_strncat(&file->name, 2423 (const char *)data + 1, data_length - 1); 2424 break; 2425 case 1: 2426 if (data_length < 2) 2427 return; 2428 archive_strncat(&file->name, 2429 (const char *)data + 1, data_length - 1); 2430 file->name_continues = 1; 2431 break; 2432 case 2: 2433 archive_strcat(&file->name, "."); 2434 break; 2435 case 4: 2436 archive_strcat(&file->name, ".."); 2437 break; 2438 default: 2439 return; 2440 } 2441 2442 } 2443 2444 static void 2445 parse_rockridge_TF1(struct file_info *file, const unsigned char *data, 2446 int data_length) 2447 { 2448 char flag; 2449 /* 2450 * TF extension comprises: 2451 * one byte flag 2452 * create time (optional) 2453 * modify time (optional) 2454 * access time (optional) 2455 * attribute time (optional) 2456 * Time format and presence of fields 2457 * is controlled by flag bits. 2458 */ 2459 if (data_length < 1) 2460 return; 2461 flag = data[0]; 2462 ++data; 2463 --data_length; 2464 if (flag & 0x80) { 2465 /* Use 17-byte time format. */ 2466 if ((flag & 1) && data_length >= 17) { 2467 /* Create time. */ 2468 file->birthtime_is_set = 1; 2469 file->birthtime = isodate17(data); 2470 data += 17; 2471 data_length -= 17; 2472 } 2473 if ((flag & 2) && data_length >= 17) { 2474 /* Modify time. */ 2475 file->mtime = isodate17(data); 2476 data += 17; 2477 data_length -= 17; 2478 } 2479 if ((flag & 4) && data_length >= 17) { 2480 /* Access time. */ 2481 file->atime = isodate17(data); 2482 data += 17; 2483 data_length -= 17; 2484 } 2485 if ((flag & 8) && data_length >= 17) { 2486 /* Attribute change time. */ 2487 file->ctime = isodate17(data); 2488 } 2489 } else { 2490 /* Use 7-byte time format. */ 2491 if ((flag & 1) && data_length >= 7) { 2492 /* Create time. */ 2493 file->birthtime_is_set = 1; 2494 file->birthtime = isodate7(data); 2495 data += 7; 2496 data_length -= 7; 2497 } 2498 if ((flag & 2) && data_length >= 7) { 2499 /* Modify time. */ 2500 file->mtime = isodate7(data); 2501 data += 7; 2502 data_length -= 7; 2503 } 2504 if ((flag & 4) && data_length >= 7) { 2505 /* Access time. */ 2506 file->atime = isodate7(data); 2507 data += 7; 2508 data_length -= 7; 2509 } 2510 if ((flag & 8) && data_length >= 7) { 2511 /* Attribute change time. */ 2512 file->ctime = isodate7(data); 2513 } 2514 } 2515 } 2516 2517 static void 2518 parse_rockridge_SL1(struct file_info *file, const unsigned char *data, 2519 int data_length) 2520 { 2521 const char *separator = ""; 2522 2523 if (!file->symlink_continues || file->symlink.length < 1) 2524 archive_string_empty(&file->symlink); 2525 file->symlink_continues = 0; 2526 2527 /* 2528 * Defined flag values: 2529 * 0: This is the last SL record for this symbolic link 2530 * 1: this symbolic link field continues in next SL entry 2531 * All other values are reserved. 2532 */ 2533 if (data_length < 1) 2534 return; 2535 switch(*data) { 2536 case 0: 2537 break; 2538 case 1: 2539 file->symlink_continues = 1; 2540 break; 2541 default: 2542 return; 2543 } 2544 ++data; /* Skip flag byte. */ 2545 --data_length; 2546 2547 /* 2548 * SL extension body stores "components". 2549 * Basically, this is a complicated way of storing 2550 * a POSIX path. It also interferes with using 2551 * symlinks for storing non-path data. <sigh> 2552 * 2553 * Each component is 2 bytes (flag and length) 2554 * possibly followed by name data. 2555 */ 2556 while (data_length >= 2) { 2557 unsigned char flag = *data++; 2558 unsigned char nlen = *data++; 2559 data_length -= 2; 2560 2561 archive_strcat(&file->symlink, separator); 2562 separator = "/"; 2563 2564 switch(flag) { 2565 case 0: /* Usual case, this is text. */ 2566 if (data_length < nlen) 2567 return; 2568 archive_strncat(&file->symlink, 2569 (const char *)data, nlen); 2570 break; 2571 case 0x01: /* Text continues in next component. */ 2572 if (data_length < nlen) 2573 return; 2574 archive_strncat(&file->symlink, 2575 (const char *)data, nlen); 2576 separator = ""; 2577 break; 2578 case 0x02: /* Current dir. */ 2579 archive_strcat(&file->symlink, "."); 2580 break; 2581 case 0x04: /* Parent dir. */ 2582 archive_strcat(&file->symlink, ".."); 2583 break; 2584 case 0x08: /* Root of filesystem. */ 2585 archive_strcat(&file->symlink, "/"); 2586 separator = ""; 2587 break; 2588 case 0x10: /* Undefined (historically "volume root" */ 2589 archive_string_empty(&file->symlink); 2590 archive_strcat(&file->symlink, "ROOT"); 2591 break; 2592 case 0x20: /* Undefined (historically "hostname") */ 2593 archive_strcat(&file->symlink, "hostname"); 2594 break; 2595 default: 2596 /* TODO: issue a warning ? */ 2597 return; 2598 } 2599 data += nlen; 2600 data_length -= nlen; 2601 } 2602 } 2603 2604 static void 2605 parse_rockridge_ZF1(struct file_info *file, const unsigned char *data, 2606 int data_length) 2607 { 2608 2609 if (data[0] == 0x70 && data[1] == 0x7a && data_length == 12) { 2610 /* paged zlib */ 2611 file->pz = 1; 2612 file->pz_log2_bs = data[3]; 2613 file->pz_uncompressed_size = archive_le32dec(&data[4]); 2614 } 2615 } 2616 2617 static void 2618 register_file(struct iso9660 *iso9660, struct file_info *file) 2619 { 2620 2621 file->use_next = iso9660->use_files; 2622 iso9660->use_files = file; 2623 } 2624 2625 static void 2626 release_files(struct iso9660 *iso9660) 2627 { 2628 struct content *con, *connext; 2629 struct file_info *file; 2630 2631 file = iso9660->use_files; 2632 while (file != NULL) { 2633 struct file_info *next = file->use_next; 2634 2635 archive_string_free(&file->name); 2636 archive_string_free(&file->symlink); 2637 free(file->utf16be_name); 2638 con = file->contents.first; 2639 while (con != NULL) { 2640 connext = con->next; 2641 free(con); 2642 con = connext; 2643 } 2644 free(file); 2645 file = next; 2646 } 2647 } 2648 2649 static int 2650 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, 2651 struct file_info **pfile) 2652 { 2653 struct file_info *file; 2654 int r; 2655 2656 r = next_cache_entry(a, iso9660, pfile); 2657 if (r != ARCHIVE_OK) 2658 return (r); 2659 file = *pfile; 2660 2661 /* Don't waste time seeking for zero-length bodies. */ 2662 if (file->size == 0) 2663 file->offset = iso9660->current_position; 2664 2665 /* flush any remaining bytes from the last round to ensure 2666 * we're positioned */ 2667 if (iso9660->entry_bytes_unconsumed) { 2668 __archive_read_consume(a, iso9660->entry_bytes_unconsumed); 2669 iso9660->entry_bytes_unconsumed = 0; 2670 } 2671 2672 /* Seek forward to the start of the entry. */ 2673 if (iso9660->current_position < file->offset) { 2674 int64_t step; 2675 2676 step = file->offset - iso9660->current_position; 2677 step = __archive_read_consume(a, step); 2678 if (step < 0) 2679 return ((int)step); 2680 iso9660->current_position = file->offset; 2681 } 2682 2683 /* We found body of file; handle it now. */ 2684 return (ARCHIVE_OK); 2685 } 2686 2687 static int 2688 next_cache_entry(struct archive_read *a, struct iso9660 *iso9660, 2689 struct file_info **pfile) 2690 { 2691 struct file_info *file; 2692 struct { 2693 struct file_info *first; 2694 struct file_info **last; 2695 } empty_files; 2696 int64_t number; 2697 int count; 2698 2699 file = cache_get_entry(iso9660); 2700 if (file != NULL) { 2701 *pfile = file; 2702 return (ARCHIVE_OK); 2703 } 2704 2705 for (;;) { 2706 struct file_info *re, *d; 2707 2708 *pfile = file = next_entry(iso9660); 2709 if (file == NULL) { 2710 /* 2711 * If directory entries all which are descendant of 2712 * rr_moved are stil remaning, expose their. 2713 */ 2714 if (iso9660->re_files.first != NULL && 2715 iso9660->rr_moved != NULL && 2716 iso9660->rr_moved->rr_moved_has_re_only) 2717 /* Expose "rr_moved" entry. */ 2718 cache_add_entry(iso9660, iso9660->rr_moved); 2719 while ((re = re_get_entry(iso9660)) != NULL) { 2720 /* Expose its descendant dirs. */ 2721 while ((d = rede_get_entry(re)) != NULL) 2722 cache_add_entry(iso9660, d); 2723 } 2724 if (iso9660->cache_files.first != NULL) 2725 return (next_cache_entry(a, iso9660, pfile)); 2726 return (ARCHIVE_EOF); 2727 } 2728 2729 if (file->cl_offset) { 2730 struct file_info *first_re = NULL; 2731 int nexted_re = 0; 2732 2733 /* 2734 * Find "RE" dir for the current file, which 2735 * has "CL" flag. 2736 */ 2737 while ((re = re_get_entry(iso9660)) 2738 != first_re) { 2739 if (first_re == NULL) 2740 first_re = re; 2741 if (re->offset == file->cl_offset) { 2742 re->parent->subdirs--; 2743 re->parent = file->parent; 2744 re->re = 0; 2745 if (re->parent->re_descendant) { 2746 nexted_re = 1; 2747 re->re_descendant = 1; 2748 if (rede_add_entry(re) < 0) 2749 goto fatal_rr; 2750 /* Move a list of descendants 2751 * to a new ancestor. */ 2752 while ((d = rede_get_entry( 2753 re)) != NULL) 2754 if (rede_add_entry(d) 2755 < 0) 2756 goto fatal_rr; 2757 break; 2758 } 2759 /* Replace the current file 2760 * with "RE" dir */ 2761 *pfile = file = re; 2762 /* Expose its descendant */ 2763 while ((d = rede_get_entry( 2764 file)) != NULL) 2765 cache_add_entry( 2766 iso9660, d); 2767 break; 2768 } else 2769 re_add_entry(iso9660, re); 2770 } 2771 if (nexted_re) { 2772 /* 2773 * Do not expose this at this time 2774 * because we have not gotten its full-path 2775 * name yet. 2776 */ 2777 continue; 2778 } 2779 } else if ((file->mode & AE_IFMT) == AE_IFDIR) { 2780 int r; 2781 2782 /* Read file entries in this dir. */ 2783 r = read_children(a, file); 2784 if (r != ARCHIVE_OK) 2785 return (r); 2786 2787 /* 2788 * Handle a special dir of Rockridge extensions, 2789 * "rr_moved". 2790 */ 2791 if (file->rr_moved) { 2792 /* 2793 * If this has only the subdirectories which 2794 * have "RE" flags, do not expose at this time. 2795 */ 2796 if (file->rr_moved_has_re_only) 2797 continue; 2798 /* Otherwise expose "rr_moved" entry. */ 2799 } else if (file->re) { 2800 /* 2801 * Do not expose this at this time 2802 * because we have not gotten its full-path 2803 * name yet. 2804 */ 2805 re_add_entry(iso9660, file); 2806 continue; 2807 } else if (file->re_descendant) { 2808 /* 2809 * If the top level "RE" entry of this entry 2810 * is not exposed, we, accordingly, should not 2811 * expose this entry at this time because 2812 * we cannot make its proper full-path name. 2813 */ 2814 if (rede_add_entry(file) == 0) 2815 continue; 2816 /* Otherwise we can expose this entry because 2817 * it seems its top level "RE" has already been 2818 * exposed. */ 2819 } 2820 } 2821 break; 2822 } 2823 2824 if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1) 2825 return (ARCHIVE_OK); 2826 2827 count = 0; 2828 number = file->number; 2829 iso9660->cache_files.first = NULL; 2830 iso9660->cache_files.last = &(iso9660->cache_files.first); 2831 empty_files.first = NULL; 2832 empty_files.last = &empty_files.first; 2833 /* Collect files which has the same file serial number. 2834 * Peek pending_files so that file which number is different 2835 * is not put bak. */ 2836 while (iso9660->pending_files.used > 0 && 2837 (iso9660->pending_files.files[0]->number == -1 || 2838 iso9660->pending_files.files[0]->number == number)) { 2839 if (file->number == -1) { 2840 /* This file has the same offset 2841 * but it's wrong offset which empty files 2842 * and symlink files have. 2843 * NOTE: This wrong offse was recorded by 2844 * old mkisofs utility. If ISO images is 2845 * created by latest mkisofs, this does not 2846 * happen. 2847 */ 2848 file->next = NULL; 2849 *empty_files.last = file; 2850 empty_files.last = &(file->next); 2851 } else { 2852 count++; 2853 cache_add_entry(iso9660, file); 2854 } 2855 file = next_entry(iso9660); 2856 } 2857 2858 if (count == 0) { 2859 *pfile = file; 2860 return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2861 } 2862 if (file->number == -1) { 2863 file->next = NULL; 2864 *empty_files.last = file; 2865 empty_files.last = &(file->next); 2866 } else { 2867 count++; 2868 cache_add_entry(iso9660, file); 2869 } 2870 2871 if (count > 1) { 2872 /* The count is the same as number of hardlink, 2873 * so much so that each nlinks of files in cache_file 2874 * is overwritten by value of the count. 2875 */ 2876 for (file = iso9660->cache_files.first; 2877 file != NULL; file = file->next) 2878 file->nlinks = count; 2879 } 2880 /* If there are empty files, that files are added 2881 * to the tail of the cache_files. */ 2882 if (empty_files.first != NULL) { 2883 *iso9660->cache_files.last = empty_files.first; 2884 iso9660->cache_files.last = empty_files.last; 2885 } 2886 *pfile = cache_get_entry(iso9660); 2887 return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK); 2888 2889 fatal_rr: 2890 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 2891 "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of " 2892 "Rockridge extensions: current position = %jd, CL offset = %jd", 2893 (intmax_t)iso9660->current_position, (intmax_t)file->cl_offset); 2894 return (ARCHIVE_FATAL); 2895 } 2896 2897 static inline void 2898 re_add_entry(struct iso9660 *iso9660, struct file_info *file) 2899 { 2900 file->re_next = NULL; 2901 *iso9660->re_files.last = file; 2902 iso9660->re_files.last = &(file->re_next); 2903 } 2904 2905 static inline struct file_info * 2906 re_get_entry(struct iso9660 *iso9660) 2907 { 2908 struct file_info *file; 2909 2910 if ((file = iso9660->re_files.first) != NULL) { 2911 iso9660->re_files.first = file->re_next; 2912 if (iso9660->re_files.first == NULL) 2913 iso9660->re_files.last = 2914 &(iso9660->re_files.first); 2915 } 2916 return (file); 2917 } 2918 2919 static inline int 2920 rede_add_entry(struct file_info *file) 2921 { 2922 struct file_info *re; 2923 2924 /* 2925 * Find "RE" entry. 2926 */ 2927 re = file->parent; 2928 while (re != NULL && !re->re) 2929 re = re->parent; 2930 if (re == NULL) 2931 return (-1); 2932 2933 file->re_next = NULL; 2934 *re->rede_files.last = file; 2935 re->rede_files.last = &(file->re_next); 2936 return (0); 2937 } 2938 2939 static inline struct file_info * 2940 rede_get_entry(struct file_info *re) 2941 { 2942 struct file_info *file; 2943 2944 if ((file = re->rede_files.first) != NULL) { 2945 re->rede_files.first = file->re_next; 2946 if (re->rede_files.first == NULL) 2947 re->rede_files.last = 2948 &(re->rede_files.first); 2949 } 2950 return (file); 2951 } 2952 2953 static inline void 2954 cache_add_entry(struct iso9660 *iso9660, struct file_info *file) 2955 { 2956 file->next = NULL; 2957 *iso9660->cache_files.last = file; 2958 iso9660->cache_files.last = &(file->next); 2959 } 2960 2961 static inline struct file_info * 2962 cache_get_entry(struct iso9660 *iso9660) 2963 { 2964 struct file_info *file; 2965 2966 if ((file = iso9660->cache_files.first) != NULL) { 2967 iso9660->cache_files.first = file->next; 2968 if (iso9660->cache_files.first == NULL) 2969 iso9660->cache_files.last = 2970 &(iso9660->cache_files.first); 2971 } 2972 return (file); 2973 } 2974 2975 static int 2976 heap_add_entry(struct archive_read *a, struct heap_queue *heap, 2977 struct file_info *file, uint64_t key) 2978 { 2979 uint64_t file_key, parent_key; 2980 int hole, parent; 2981 2982 /* Expand our pending files list as necessary. */ 2983 if (heap->used >= heap->allocated) { 2984 struct file_info **new_pending_files; 2985 int new_size = heap->allocated * 2; 2986 2987 if (heap->allocated < 1024) 2988 new_size = 1024; 2989 /* Overflow might keep us from growing the list. */ 2990 if (new_size <= heap->allocated) { 2991 archive_set_error(&a->archive, 2992 ENOMEM, "Out of memory"); 2993 return (ARCHIVE_FATAL); 2994 } 2995 new_pending_files = (struct file_info **) 2996 malloc(new_size * sizeof(new_pending_files[0])); 2997 if (new_pending_files == NULL) { 2998 archive_set_error(&a->archive, 2999 ENOMEM, "Out of memory"); 3000 return (ARCHIVE_FATAL); 3001 } 3002 memcpy(new_pending_files, heap->files, 3003 heap->allocated * sizeof(new_pending_files[0])); 3004 if (heap->files != NULL) 3005 free(heap->files); 3006 heap->files = new_pending_files; 3007 heap->allocated = new_size; 3008 } 3009 3010 file_key = file->key = key; 3011 3012 /* 3013 * Start with hole at end, walk it up tree to find insertion point. 3014 */ 3015 hole = heap->used++; 3016 while (hole > 0) { 3017 parent = (hole - 1)/2; 3018 parent_key = heap->files[parent]->key; 3019 if (file_key >= parent_key) { 3020 heap->files[hole] = file; 3021 return (ARCHIVE_OK); 3022 } 3023 /* Move parent into hole <==> move hole up tree. */ 3024 heap->files[hole] = heap->files[parent]; 3025 hole = parent; 3026 } 3027 heap->files[0] = file; 3028 3029 return (ARCHIVE_OK); 3030 } 3031 3032 static struct file_info * 3033 heap_get_entry(struct heap_queue *heap) 3034 { 3035 uint64_t a_key, b_key, c_key; 3036 int a, b, c; 3037 struct file_info *r, *tmp; 3038 3039 if (heap->used < 1) 3040 return (NULL); 3041 3042 /* 3043 * The first file in the list is the earliest; we'll return this. 3044 */ 3045 r = heap->files[0]; 3046 3047 /* 3048 * Move the last item in the heap to the root of the tree 3049 */ 3050 heap->files[0] = heap->files[--(heap->used)]; 3051 3052 /* 3053 * Rebalance the heap. 3054 */ 3055 a = 0; /* Starting element and its heap key */ 3056 a_key = heap->files[a]->key; 3057 for (;;) { 3058 b = a + a + 1; /* First child */ 3059 if (b >= heap->used) 3060 return (r); 3061 b_key = heap->files[b]->key; 3062 c = b + 1; /* Use second child if it is smaller. */ 3063 if (c < heap->used) { 3064 c_key = heap->files[c]->key; 3065 if (c_key < b_key) { 3066 b = c; 3067 b_key = c_key; 3068 } 3069 } 3070 if (a_key <= b_key) 3071 return (r); 3072 tmp = heap->files[a]; 3073 heap->files[a] = heap->files[b]; 3074 heap->files[b] = tmp; 3075 a = b; 3076 } 3077 } 3078 3079 static unsigned int 3080 toi(const void *p, int n) 3081 { 3082 const unsigned char *v = (const unsigned char *)p; 3083 if (n > 1) 3084 return v[0] + 256 * toi(v + 1, n - 1); 3085 if (n == 1) 3086 return v[0]; 3087 return (0); 3088 } 3089 3090 static time_t 3091 isodate7(const unsigned char *v) 3092 { 3093 struct tm tm; 3094 int offset; 3095 time_t t; 3096 3097 memset(&tm, 0, sizeof(tm)); 3098 tm.tm_year = v[0]; 3099 tm.tm_mon = v[1] - 1; 3100 tm.tm_mday = v[2]; 3101 tm.tm_hour = v[3]; 3102 tm.tm_min = v[4]; 3103 tm.tm_sec = v[5]; 3104 /* v[6] is the signed timezone offset, in 1/4-hour increments. */ 3105 offset = ((const signed char *)v)[6]; 3106 if (offset > -48 && offset < 52) { 3107 tm.tm_hour -= offset / 4; 3108 tm.tm_min -= (offset % 4) * 15; 3109 } 3110 t = time_from_tm(&tm); 3111 if (t == (time_t)-1) 3112 return ((time_t)0); 3113 return (t); 3114 } 3115 3116 static time_t 3117 isodate17(const unsigned char *v) 3118 { 3119 struct tm tm; 3120 int offset; 3121 time_t t; 3122 3123 memset(&tm, 0, sizeof(tm)); 3124 tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 3125 + (v[2] - '0') * 10 + (v[3] - '0') 3126 - 1900; 3127 tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); 3128 tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); 3129 tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); 3130 tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); 3131 tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); 3132 /* v[16] is the signed timezone offset, in 1/4-hour increments. */ 3133 offset = ((const signed char *)v)[16]; 3134 if (offset > -48 && offset < 52) { 3135 tm.tm_hour -= offset / 4; 3136 tm.tm_min -= (offset % 4) * 15; 3137 } 3138 t = time_from_tm(&tm); 3139 if (t == (time_t)-1) 3140 return ((time_t)0); 3141 return (t); 3142 } 3143 3144 static time_t 3145 time_from_tm(struct tm *t) 3146 { 3147 #if HAVE_TIMEGM 3148 /* Use platform timegm() if available. */ 3149 return (timegm(t)); 3150 #elif HAVE__MKGMTIME64 3151 return (_mkgmtime64(t)); 3152 #else 3153 /* Else use direct calculation using POSIX assumptions. */ 3154 /* First, fix up tm_yday based on the year/month/day. */ 3155 if (mktime(t) == (time_t)-1) 3156 return ((time_t)-1); 3157 /* Then we can compute timegm() from first principles. */ 3158 return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 3159 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 3160 + ((t->tm_year - 69) / 4) * 86400 - 3161 ((t->tm_year - 1) / 100) * 86400 3162 + ((t->tm_year + 299) / 400) * 86400); 3163 #endif 3164 } 3165 3166 static const char * 3167 build_pathname(struct archive_string *as, struct file_info *file) 3168 { 3169 if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { 3170 build_pathname(as, file->parent); 3171 archive_strcat(as, "/"); 3172 } 3173 if (archive_strlen(&file->name) == 0) 3174 archive_strcat(as, "."); 3175 else 3176 archive_string_concat(as, &file->name); 3177 return (as->s); 3178 } 3179 3180 static int 3181 build_pathname_utf16be(unsigned char *p, size_t max, size_t *len, 3182 struct file_info *file) 3183 { 3184 if (file->parent != NULL && file->parent->utf16be_bytes > 0) { 3185 if (build_pathname_utf16be(p, max, len, file->parent) != 0) 3186 return (-1); 3187 p[*len] = 0; 3188 p[*len + 1] = '/'; 3189 *len += 2; 3190 } 3191 if (file->utf16be_bytes == 0) { 3192 if (*len + 2 > max) 3193 return (-1);/* Path is too long! */ 3194 p[*len] = 0; 3195 p[*len + 1] = '.'; 3196 *len += 2; 3197 } else { 3198 if (*len + file->utf16be_bytes > max) 3199 return (-1);/* Path is too long! */ 3200 memcpy(p + *len, file->utf16be_name, file->utf16be_bytes); 3201 *len += file->utf16be_bytes; 3202 } 3203 return (0); 3204 } 3205 3206 #if DEBUG 3207 static void 3208 dump_isodirrec(FILE *out, const unsigned char *isodirrec) 3209 { 3210 fprintf(out, " l %d,", 3211 toi(isodirrec + DR_length_offset, DR_length_size)); 3212 fprintf(out, " a %d,", 3213 toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); 3214 fprintf(out, " ext 0x%x,", 3215 toi(isodirrec + DR_extent_offset, DR_extent_size)); 3216 fprintf(out, " s %d,", 3217 toi(isodirrec + DR_size_offset, DR_extent_size)); 3218 fprintf(out, " f 0x%x,", 3219 toi(isodirrec + DR_flags_offset, DR_flags_size)); 3220 fprintf(out, " u %d,", 3221 toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); 3222 fprintf(out, " ilv %d,", 3223 toi(isodirrec + DR_interleave_offset, DR_interleave_size)); 3224 fprintf(out, " seq %d,", 3225 toi(isodirrec + DR_volume_sequence_number_offset, 3226 DR_volume_sequence_number_size)); 3227 fprintf(out, " nl %d:", 3228 toi(isodirrec + DR_name_len_offset, DR_name_len_size)); 3229 fprintf(out, " `%.*s'", 3230 toi(isodirrec + DR_name_len_offset, DR_name_len_size), 3231 isodirrec + DR_name_offset); 3232 } 3233 #endif 3234