2 * Copyright (c) 2003-2007 Tim Kientzle
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
15 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 #include "archive_platform.h"
27 __FBSDID("$FreeBSD$");
32 /* #include <stdint.h> */ /* See archive_platform.h */
43 #include "archive_entry.h"
44 #include "archive_private.h"
45 #include "archive_read_private.h"
46 #include "archive_string.h"
49 * An overview of ISO 9660 format:
51 * Each disk is laid out as follows:
52 * * 32k reserved for private use
53 * * Volume descriptor table. Each volume descriptor
54 * is 2k and specifies basic format information.
55 * The "Primary Volume Descriptor" (PVD) is defined by the
56 * standard and should always be present; other volume
57 * descriptors include various vendor-specific extensions.
58 * * Files and directories. Each file/dir is specified by
59 * an "extent" (starting sector and length in bytes).
60 * Dirs are just files with directory records packed one
61 * after another. The PVD contains a single dir entry
62 * specifying the location of the root directory. Everything
63 * else follows from there.
65 * This module works by first reading the volume descriptors, then
66 * building a list of directory entries, sorted by starting
67 * sector. At each step, I look for the earliest dir entry that
68 * hasn't yet been read, seek forward to that location and read
69 * that entry. If it's a dir, I slurp in the new dir entries and
70 * add them to the heap; if it's a regular file, I return the
71 * corresponding archive_entry and wait for the client to request
72 * the file body. This strategy allows us to read most compliant
73 * CDs with a single pass through the data, as required by libarchive.
76 /* Structure of on-disk primary volume descriptor. */
77 #define PVD_type_offset 0
78 #define PVD_type_size 1
79 #define PVD_id_offset (PVD_type_offset + PVD_type_size)
81 #define PVD_version_offset (PVD_id_offset + PVD_id_size)
82 #define PVD_version_size 1
83 #define PVD_reserved1_offset (PVD_version_offset + PVD_version_size)
84 #define PVD_reserved1_size 1
85 #define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size)
86 #define PVD_system_id_size 32
87 #define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size)
88 #define PVD_volume_id_size 32
89 #define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size)
90 #define PVD_reserved2_size 8
91 #define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size)
92 #define PVD_volume_space_size_size 8
93 #define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size)
94 #define PVD_reserved3_size 32
95 #define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size)
96 #define PVD_volume_set_size_size 4
97 #define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size)
98 #define PVD_volume_sequence_number_size 4
99 #define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size)
100 #define PVD_logical_block_size_size 4
101 #define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size)
102 #define PVD_path_table_size_size 8
103 #define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size)
104 #define PVD_type_1_path_table_size 4
105 #define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size)
106 #define PVD_opt_type_1_path_table_size 4
107 #define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size)
108 #define PVD_type_m_path_table_size 4
109 #define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size)
110 #define PVD_opt_type_m_path_table_size 4
111 #define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size)
112 #define PVD_root_directory_record_size 34
113 #define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size)
114 #define PVD_volume_set_id_size 128
115 #define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size)
116 #define PVD_publisher_id_size 128
117 #define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size)
118 #define PVD_preparer_id_size 128
119 #define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size)
120 #define PVD_application_id_size 128
121 #define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size)
122 #define PVD_copyright_file_id_size 37
123 #define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size)
124 #define PVD_abstract_file_id_size 37
125 #define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size)
126 #define PVD_bibliographic_file_id_size 37
127 #define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size)
128 #define PVD_creation_date_size 17
129 #define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size)
130 #define PVD_modification_date_size 17
131 #define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size)
132 #define PVD_expiration_date_size 17
133 #define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size)
134 #define PVD_effective_date_size 17
135 #define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size)
136 #define PVD_file_structure_version_size 1
137 #define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size)
138 #define PVD_reserved4_size 1
139 #define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size)
140 #define PVD_application_data_size 512
141 #define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size)
142 #define PVD_reserved5_size (2048 - PVD_reserved5_offset)
144 /* TODO: It would make future maintenance easier to just hardcode the
145 * above values. In particular, ECMA119 states the offsets as part of
146 * the standard. That would eliminate the need for the following check.*/
147 #if PVD_reserved5_offset != 1395
148 #error PVD offset and size definitions are wrong.
151 /* Structure of an on-disk directory record. */
152 /* Note: ISO9660 stores each multi-byte integer twice, once in
153 * each byte order. The sizes here are the size of just one
154 * of the two integers. (This is why the offset of a field isn't
155 * the same as the offset+size of the previous field.) */
156 #define DR_length_offset 0
157 #define DR_length_size 1
158 #define DR_ext_attr_length_offset 1
159 #define DR_ext_attr_length_size 1
160 #define DR_extent_offset 2
161 #define DR_extent_size 4
162 #define DR_size_offset 10
163 #define DR_size_size 4
164 #define DR_date_offset 18
165 #define DR_date_size 7
166 #define DR_flags_offset 25
167 #define DR_flags_size 1
168 #define DR_file_unit_size_offset 26
169 #define DR_file_unit_size_size 1
170 #define DR_interleave_offset 27
171 #define DR_interleave_size 1
172 #define DR_volume_sequence_number_offset 28
173 #define DR_volume_sequence_number_size 2
174 #define DR_name_len_offset 32
175 #define DR_name_len_size 1
176 #define DR_name_offset 33
182 /* In-memory storage for a directory record. */
184 struct file_info *parent;
186 uint64_t offset; /* Offset on disk. */
187 uint64_t size; /* File size in bytes. */
188 uint64_t ce_offset; /* Offset of CE */
189 uint64_t ce_size; /* Size of CE */
190 time_t birthtime; /* File created time. */
191 time_t mtime; /* File last modified time. */
192 time_t atime; /* File last accessed time. */
193 time_t ctime; /* File attribute change time. */
194 uint64_t rdev; /* Device number */
200 struct archive_string name; /* Pathname */
201 char name_continues; /* Non-zero if name continues */
202 struct archive_string symlink;
203 char symlink_continues; /* Non-zero if link continues */
209 #define ISO9660_MAGIC 0x96609660
210 struct archive_string pathname;
211 char seenRockridge; /* Set true if RR extensions are used. */
212 unsigned char suspOffset;
214 uint64_t previous_offset;
215 uint64_t previous_size;
216 struct archive_string previous_pathname;
218 /* TODO: Make this a heap for fast inserts and deletions. */
219 struct file_info **pending_files;
220 int pending_files_allocated;
221 int pending_files_used;
223 uint64_t current_position;
224 ssize_t logical_block_size;
225 uint64_t volume_size; /* Total size of volume in bytes. */
227 off_t entry_sparse_offset;
228 int64_t entry_bytes_remaining;
231 static void add_entry(struct iso9660 *iso9660, struct file_info *file);
232 static int archive_read_format_iso9660_bid(struct archive_read *);
233 static int archive_read_format_iso9660_cleanup(struct archive_read *);
234 static int archive_read_format_iso9660_read_data(struct archive_read *,
235 const void **, size_t *, off_t *);
236 static int archive_read_format_iso9660_read_data_skip(struct archive_read *);
237 static int archive_read_format_iso9660_read_header(struct archive_read *,
238 struct archive_entry *);
239 static const char *build_pathname(struct archive_string *, struct file_info *);
241 static void dump_isodirrec(FILE *, const unsigned char *isodirrec);
243 static time_t time_from_tm(struct tm *);
244 static time_t isodate17(const unsigned char *);
245 static time_t isodate7(const unsigned char *);
246 static int isPVD(struct iso9660 *, const unsigned char *);
247 static struct file_info *next_entry(struct iso9660 *);
248 static int next_entry_seek(struct archive_read *a, struct iso9660 *iso9660,
249 struct file_info **pfile);
250 static struct file_info *
251 parse_file_info(struct iso9660 *iso9660,
252 struct file_info *parent, const unsigned char *isodirrec);
253 static void parse_rockridge(struct iso9660 *iso9660,
254 struct file_info *file, const unsigned char *start,
255 const unsigned char *end);
256 static void parse_rockridge_NM1(struct file_info *,
257 const unsigned char *, int);
258 static void parse_rockridge_SL1(struct file_info *,
259 const unsigned char *, int);
260 static void parse_rockridge_TF1(struct file_info *,
261 const unsigned char *, int);
262 static void release_file(struct iso9660 *, struct file_info *);
263 static unsigned toi(const void *p, int n);
266 archive_read_support_format_iso9660(struct archive *_a)
268 struct archive_read *a = (struct archive_read *)_a;
269 struct iso9660 *iso9660;
272 iso9660 = (struct iso9660 *)malloc(sizeof(*iso9660));
273 if (iso9660 == NULL) {
274 archive_set_error(&a->archive, ENOMEM, "Can't allocate iso9660 data");
275 return (ARCHIVE_FATAL);
277 memset(iso9660, 0, sizeof(*iso9660));
278 iso9660->magic = ISO9660_MAGIC;
280 r = __archive_read_register_format(a,
282 archive_read_format_iso9660_bid,
283 archive_read_format_iso9660_read_header,
284 archive_read_format_iso9660_read_data,
285 archive_read_format_iso9660_read_data_skip,
286 archive_read_format_iso9660_cleanup);
288 if (r != ARCHIVE_OK) {
297 archive_read_format_iso9660_bid(struct archive_read *a)
299 struct iso9660 *iso9660;
302 const unsigned char *p;
305 iso9660 = (struct iso9660 *)(a->format->data);
308 * Skip the first 32k (reserved area) and get the first
309 * 8 sectors of the volume descriptor table. Of course,
310 * if the I/O layer gives us more, we'll take it.
312 bytes_read = (a->decompressor->read_ahead)(a, &h, 32768 + 8*2048);
313 if (bytes_read < 32768 + 8*2048)
315 p = (const unsigned char *)h;
317 /* Skip the reserved area. */
321 /* Check each volume descriptor to locate the PVD. */
322 for (; bytes_read > 2048; bytes_read -= 2048, p += 2048) {
323 bid = isPVD(iso9660, p);
326 if (*p == '\177') /* End-of-volume-descriptor marker. */
330 /* We didn't find a valid PVD; return a bid of zero. */
335 isPVD(struct iso9660 *iso9660, const unsigned char *h)
337 struct file_info *file;
340 /* Type of the Primary Volume Descriptor must be 1. */
341 if (h[PVD_type_offset] != 1)
344 /* ID must be "CD001" */
345 if (memcmp(h + PVD_id_offset, "CD001", 5) != 0)
348 /* PVD version must be 1. */
349 if (h[PVD_version_offset] != 1)
352 /* Reserved field must be 0. */
353 if (h[PVD_reserved1_offset] != 0)
356 /* Reserved field must be 0. */
357 for (i = 0; i < PVD_reserved2_size; ++i)
358 if (h[PVD_reserved2_offset + i] != 0)
361 /* Reserved field must be 0. */
362 for (i = 0; i < PVD_reserved3_size; ++i)
363 if (h[PVD_reserved3_offset + i] != 0)
366 /* Logical block size must be > 0. */
367 /* I've looked at Ecma 119 and can't find any stronger
368 * restriction on this field. */
369 iso9660->logical_block_size = toi(h + PVD_logical_block_size_offset, 2);
370 if (iso9660->logical_block_size <= 0)
373 iso9660->volume_size = iso9660->logical_block_size
374 * (uint64_t)toi(h + PVD_volume_space_size_offset, 4);
376 /* File structure version must be 1 for ISO9660/ECMA119. */
377 if (h[PVD_file_structure_version_offset] != 1)
381 /* Reserved field must be 0. */
382 for (i = 0; i < PVD_reserved4_size; ++i)
383 if (h[PVD_reserved4_offset + i] != 0)
386 /* Reserved field must be 0. */
387 for (i = 0; i < PVD_reserved5_size; ++i)
388 if (h[PVD_reserved5_offset + i] != 0)
391 /* XXX TODO: Check other values for sanity; reject more
392 * malformed PVDs. XXX */
394 /* Store the root directory in the pending list. */
395 file = parse_file_info(iso9660, NULL, h + PVD_root_directory_record_offset);
396 add_entry(iso9660, file);
401 archive_read_format_iso9660_read_header(struct archive_read *a,
402 struct archive_entry *entry)
404 struct iso9660 *iso9660;
405 struct file_info *file;
409 iso9660 = (struct iso9660 *)(a->format->data);
411 if (!a->archive.archive_format) {
412 a->archive.archive_format = ARCHIVE_FORMAT_ISO9660;
413 a->archive.archive_format_name = "ISO9660";
416 /* Get the next entry that appears after the current offset. */
417 r = next_entry_seek(a, iso9660, &file);
421 iso9660->entry_bytes_remaining = file->size;
422 iso9660->entry_sparse_offset = 0; /* Offset for sparse-file-aware clients. */
424 if (file->offset + file->size > iso9660->volume_size) {
425 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
426 "File is beyond end-of-media: %s", file->name);
427 iso9660->entry_bytes_remaining = 0;
428 iso9660->entry_sparse_offset = 0;
429 release_file(iso9660, file);
430 return (ARCHIVE_WARN);
433 /* Set up the entry structure with information about this entry. */
434 archive_entry_set_mode(entry, file->mode);
435 archive_entry_set_uid(entry, file->uid);
436 archive_entry_set_gid(entry, file->gid);
437 archive_entry_set_nlink(entry, file->nlinks);
438 archive_entry_set_ino(entry, file->inode);
439 /* archive_entry_set_birthtime(entry, file->birthtime, 0); */
440 archive_entry_set_mtime(entry, file->mtime, 0);
441 archive_entry_set_ctime(entry, file->ctime, 0);
442 archive_entry_set_atime(entry, file->atime, 0);
443 /* N.B.: Rock Ridge supports 64-bit device numbers. */
444 archive_entry_set_rdev(entry, (dev_t)file->rdev);
445 archive_entry_set_size(entry, iso9660->entry_bytes_remaining);
446 archive_string_empty(&iso9660->pathname);
447 archive_entry_set_pathname(entry,
448 build_pathname(&iso9660->pathname, file));
449 if (file->symlink.s != NULL)
450 archive_entry_copy_symlink(entry, file->symlink.s);
452 /* If this entry points to the same data as the previous
453 * entry, convert this into a hardlink to that entry.
454 * But don't bother for zero-length files. */
455 if (file->offset == iso9660->previous_offset
456 && file->size == iso9660->previous_size
458 archive_entry_set_hardlink(entry,
459 iso9660->previous_pathname.s);
460 iso9660->entry_bytes_remaining = 0;
461 iso9660->entry_sparse_offset = 0;
462 release_file(iso9660, file);
466 /* If the offset is before our current position, we can't
467 * seek backwards to extract it, so issue a warning. */
468 if (file->offset < iso9660->current_position) {
469 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
470 "Ignoring out-of-order file");
471 iso9660->entry_bytes_remaining = 0;
472 iso9660->entry_sparse_offset = 0;
473 release_file(iso9660, file);
474 return (ARCHIVE_WARN);
477 iso9660->previous_size = file->size;
478 iso9660->previous_offset = file->offset;
479 archive_strcpy(&iso9660->previous_pathname, iso9660->pathname.s);
481 /* If this is a directory, read in all of the entries right now. */
482 if (archive_entry_filetype(entry) == AE_IFDIR) {
483 while (iso9660->entry_bytes_remaining > 0) {
485 const unsigned char *p;
486 ssize_t step = iso9660->logical_block_size;
487 if (step > iso9660->entry_bytes_remaining)
488 step = iso9660->entry_bytes_remaining;
489 bytes_read = (a->decompressor->read_ahead)(a, &block, step);
490 if (bytes_read < step) {
491 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
492 "Failed to read full block when scanning ISO9660 directory list");
493 release_file(iso9660, file);
494 return (ARCHIVE_FATAL);
496 if (bytes_read > step)
498 (a->decompressor->consume)(a, bytes_read);
499 iso9660->current_position += bytes_read;
500 iso9660->entry_bytes_remaining -= bytes_read;
501 for (p = (const unsigned char *)block;
502 *p != 0 && p < (const unsigned char *)block + bytes_read;
504 struct file_info *child;
506 /* Skip '.' entry. */
507 if (*(p + DR_name_len_offset) == 1
508 && *(p + DR_name_offset) == '\0')
510 /* Skip '..' entry. */
511 if (*(p + DR_name_len_offset) == 1
512 && *(p + DR_name_offset) == '\001')
514 child = parse_file_info(iso9660, file, p);
515 add_entry(iso9660, child);
516 if (iso9660->seenRockridge) {
517 a->archive.archive_format =
518 ARCHIVE_FORMAT_ISO9660_ROCKRIDGE;
519 a->archive.archive_format_name =
520 "ISO9660 with Rockridge extensions";
526 release_file(iso9660, file);
531 archive_read_format_iso9660_read_data_skip(struct archive_read *a)
533 /* Because read_next_header always does an explicit skip
534 * to the next entry, we don't need to do anything here. */
535 (void)a; /* UNUSED */
540 archive_read_format_iso9660_read_data(struct archive_read *a,
541 const void **buff, size_t *size, off_t *offset)
544 struct iso9660 *iso9660;
546 iso9660 = (struct iso9660 *)(a->format->data);
547 if (iso9660->entry_bytes_remaining <= 0) {
550 *offset = iso9660->entry_sparse_offset;
551 return (ARCHIVE_EOF);
554 bytes_read = (a->decompressor->read_ahead)(a, buff, 1);
556 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
557 "Truncated input file");
559 return (ARCHIVE_FATAL);
560 if (bytes_read > iso9660->entry_bytes_remaining)
561 bytes_read = iso9660->entry_bytes_remaining;
563 *offset = iso9660->entry_sparse_offset;
564 iso9660->entry_sparse_offset += bytes_read;
565 iso9660->entry_bytes_remaining -= bytes_read;
566 iso9660->current_position += bytes_read;
567 (a->decompressor->consume)(a, bytes_read);
572 archive_read_format_iso9660_cleanup(struct archive_read *a)
574 struct iso9660 *iso9660;
575 struct file_info *file;
577 iso9660 = (struct iso9660 *)(a->format->data);
578 while ((file = next_entry(iso9660)) != NULL)
579 release_file(iso9660, file);
580 archive_string_free(&iso9660->pathname);
581 archive_string_free(&iso9660->previous_pathname);
582 if (iso9660->pending_files)
583 free(iso9660->pending_files);
585 (a->format->data) = NULL;
590 * This routine parses a single ISO directory record, makes sense
591 * of any extensions, and stores the result in memory.
593 static struct file_info *
594 parse_file_info(struct iso9660 *iso9660, struct file_info *parent,
595 const unsigned char *isodirrec)
597 struct file_info *file;
601 /* TODO: Sanity check that name_len doesn't exceed length, etc. */
603 /* Create a new file entry and copy data from the ISO dir record. */
604 file = (struct file_info *)malloc(sizeof(*file));
607 memset(file, 0, sizeof(*file));
608 file->parent = parent;
611 file->offset = toi(isodirrec + DR_extent_offset, DR_extent_size)
612 * iso9660->logical_block_size;
613 file->size = toi(isodirrec + DR_size_offset, DR_size_size);
614 file->mtime = isodate7(isodirrec + DR_date_offset);
615 file->ctime = file->atime = file->mtime;
616 name_len = (size_t)*(const unsigned char *)(isodirrec + DR_name_len_offset);
617 archive_strncpy(&file->name, isodirrec + DR_name_offset, name_len);
618 flags = *(isodirrec + DR_flags_offset);
620 file->mode = AE_IFDIR | 0700;
622 file->mode = AE_IFREG | 0400;
624 /* Rockridge extensions overwrite information from above. */
626 const unsigned char *rr_start, *rr_end;
627 rr_end = (const unsigned char *)isodirrec
628 + *(isodirrec + DR_length_offset);
629 rr_start = (const unsigned char *)(isodirrec + DR_name_offset
631 if ((name_len & 1) == 0)
633 rr_start += iso9660->suspOffset;
634 parse_rockridge(iso9660, file, rr_start, rr_end);
638 /* DEBUGGING: Warn about attributes I don't yet fully support. */
639 if ((flags & ~0x02) != 0) {
640 fprintf(stderr, "\n ** Unrecognized flag: ");
641 dump_isodirrec(stderr, isodirrec);
642 fprintf(stderr, "\n");
643 } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) {
644 fprintf(stderr, "\n ** Unrecognized sequence number: ");
645 dump_isodirrec(stderr, isodirrec);
646 fprintf(stderr, "\n");
647 } else if (*(isodirrec + DR_file_unit_size_offset) != 0) {
648 fprintf(stderr, "\n ** Unexpected file unit size: ");
649 dump_isodirrec(stderr, isodirrec);
650 fprintf(stderr, "\n");
651 } else if (*(isodirrec + DR_interleave_offset) != 0) {
652 fprintf(stderr, "\n ** Unexpected interleave: ");
653 dump_isodirrec(stderr, isodirrec);
654 fprintf(stderr, "\n");
655 } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) {
656 fprintf(stderr, "\n ** Unexpected extended attribute length: ");
657 dump_isodirrec(stderr, isodirrec);
658 fprintf(stderr, "\n");
665 add_entry(struct iso9660 *iso9660, struct file_info *file)
667 /* Expand our pending files list as necessary. */
668 if (iso9660->pending_files_used >= iso9660->pending_files_allocated) {
669 struct file_info **new_pending_files;
670 int new_size = iso9660->pending_files_allocated * 2;
672 if (iso9660->pending_files_allocated < 1024)
674 /* Overflow might keep us from growing the list. */
675 if (new_size <= iso9660->pending_files_allocated)
676 __archive_errx(1, "Out of memory");
677 new_pending_files = (struct file_info **)malloc(new_size * sizeof(new_pending_files[0]));
678 if (new_pending_files == NULL)
679 __archive_errx(1, "Out of memory");
680 memcpy(new_pending_files, iso9660->pending_files,
681 iso9660->pending_files_allocated * sizeof(new_pending_files[0]));
682 if (iso9660->pending_files != NULL)
683 free(iso9660->pending_files);
684 iso9660->pending_files = new_pending_files;
685 iso9660->pending_files_allocated = new_size;
688 iso9660->pending_files[iso9660->pending_files_used++] = file;
692 parse_rockridge(struct iso9660 *iso9660, struct file_info *file,
693 const unsigned char *p, const unsigned char *end)
695 (void)iso9660; /* UNUSED */
697 while (p + 4 < end /* Enough space for another entry. */
698 && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */
699 && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */
700 && p[2] >= 4 /* Sanity-check length. */
701 && p + p[2] <= end) { /* Sanity-check length. */
702 const unsigned char *data = p + 4;
703 int data_length = p[2] - 4;
707 * Yes, each 'if' here does test p[0] again.
708 * Otherwise, the fall-through handling to catch
709 * unsupported extensions doesn't work.
713 if (p[0] == 'C' && p[1] == 'E') {
714 if (version == 1 && data_length == 24) {
716 * CE extension comprises:
717 * 8 byte sector containing extension
718 * 8 byte offset w/in above sector
719 * 8 byte length of continuation
721 file->ce_offset = (uint64_t)toi(data, 4)
722 * iso9660->logical_block_size
724 file->ce_size = toi(data + 16, 4);
725 /* If the result is rediculous,
727 if (file->ce_offset + file->ce_size
728 > iso9660->volume_size) {
737 if (p[0] == 'N' && p[1] == 'M') {
739 parse_rockridge_NM1(file,
745 if (p[0] == 'P' && p[1] == 'D') {
747 * PD extension is padding;
748 * contents are always ignored.
752 if (p[0] == 'P' && p[1] == 'N') {
753 if (version == 1 && data_length == 16) {
754 file->rdev = toi(data,4);
756 file->rdev |= toi(data + 8, 4);
760 if (p[0] == 'P' && p[1] == 'X') {
762 * PX extension comprises:
764 * 8 bytes for nlinks,
770 if (data_length >= 8)
773 if (data_length >= 16)
776 if (data_length >= 24)
779 if (data_length >= 32)
782 if (data_length >= 40)
790 if (p[0] == 'R' && p[1] == 'R' && version == 1) {
791 iso9660->seenRockridge = 1;
793 * RR extension comprises:
794 * one byte flag value
796 /* TODO: Handle RR extension. */
801 if (p[0] == 'S' && p[1] == 'L') {
803 parse_rockridge_SL1(file,
807 if (p[0] == 'S' && p[1] == 'P'
808 && version == 1 && data_length == 3
809 && data[0] == (unsigned char)'\xbe'
810 && data[1] == (unsigned char)'\xef') {
812 * SP extension stores the suspOffset
813 * (Number of bytes to skip between
814 * filename and SUSP records.)
815 * It is mandatory by the SUSP standard
818 * It allows SUSP to coexist with
819 * non-SUSP uses of the System
820 * Use Area by placing non-SUSP data
823 * TODO: Add a check for 'SP' in
824 * first directory entry, disable all SUSP
825 * processing if not found.
827 iso9660->suspOffset = data[2];
830 if (p[0] == 'S' && p[1] == 'T'
831 && data_length == 0 && version == 1) {
833 * ST extension marks end of this
834 * block of SUSP entries.
836 * It allows SUSP to coexist with
837 * non-SUSP uses of the System
838 * Use Area by placing non-SUSP data
844 if (p[0] == 'T' && p[1] == 'F') {
846 parse_rockridge_TF1(file,
852 /* The FALLTHROUGHs above leave us here for
853 * any unsupported extension. */
856 const unsigned char *t;
857 fprintf(stderr, "\nUnsupported RRIP extension for %s\n", file->name.s);
858 fprintf(stderr, " %c%c(%d):", p[0], p[1], data_length);
859 for (t = data; t < data + data_length && t < data + 16; t++)
860 fprintf(stderr, " %02x", *t);
861 fprintf(stderr, "\n");
874 parse_rockridge_NM1(struct file_info *file, const unsigned char *data,
877 if (!file->name_continues)
878 archive_string_empty(&file->name);
879 file->name_continues = 0;
883 * NM version 1 extension comprises:
884 * 1 byte flag, value is one of:
885 * = 0: remainder is name
886 * = 1: remainder is name, next NM entry continues name
889 * = 32: Implementation specific
890 * All other values are reserved.
896 archive_strncat(&file->name, data + 1, data_length - 1);
901 archive_strncat(&file->name, data + 1, data_length - 1);
902 file->name_continues = 1;
905 archive_strcat(&file->name, ".");
908 archive_strcat(&file->name, "..");
917 parse_rockridge_TF1(struct file_info *file, const unsigned char *data,
922 * TF extension comprises:
924 * create time (optional)
925 * modify time (optional)
926 * access time (optional)
927 * attribute time (optional)
928 * Time format and presence of fields
929 * is controlled by flag bits.
937 /* Use 17-byte time format. */
938 if ((flag & 1) && data_length >= 17) {
940 file->birthtime = isodate17(data);
944 if ((flag & 2) && data_length >= 17) {
946 file->mtime = isodate17(data);
950 if ((flag & 4) && data_length >= 17) {
952 file->atime = isodate17(data);
956 if ((flag & 8) && data_length >= 17) {
957 /* Attribute change time. */
958 file->ctime = isodate17(data);
963 /* Use 7-byte time format. */
964 if ((flag & 1) && data_length >= 7) {
966 file->birthtime = isodate17(data);
970 if ((flag & 2) && data_length >= 7) {
972 file->mtime = isodate7(data);
976 if ((flag & 4) && data_length >= 7) {
978 file->atime = isodate7(data);
982 if ((flag & 8) && data_length >= 7) {
983 /* Attribute change time. */
984 file->ctime = isodate7(data);
992 parse_rockridge_SL1(struct file_info *file, const unsigned char *data,
995 int component_continues = 1;
997 if (!file->symlink_continues)
998 archive_string_empty(&file->symlink);
1000 archive_strcat(&file->symlink, "/");
1001 file->symlink_continues = 0;
1004 * Defined flag values:
1005 * 0: This is the last SL record for this symbolic link
1006 * 1: this symbolic link field continues in next SL entry
1007 * All other values are reserved.
1009 if (data_length < 1)
1015 file->symlink_continues = 1;
1020 ++data; /* Skip flag byte. */
1024 * SL extension body stores "components".
1025 * Basically, this is a complicated way of storing
1026 * a POSIX path. It also interferes with using
1027 * symlinks for storing non-path data. <sigh>
1029 * Each component is 2 bytes (flag and length)
1030 * possibly followed by name data.
1032 while (data_length >= 2) {
1033 unsigned char flag = *data++;
1034 unsigned char nlen = *data++;
1037 if (!component_continues)
1038 archive_strcat(&file->symlink, "/");
1039 component_continues = 0;
1042 case 0: /* Usual case, this is text. */
1043 if (data_length < nlen)
1045 archive_strncat(&file->symlink,
1046 (const char *)data, nlen);
1048 case 0x01: /* Text continues in next component. */
1049 if (data_length < nlen)
1051 archive_strncat(&file->symlink,
1052 (const char *)data, nlen);
1053 component_continues = 1;
1055 case 0x02: /* Current dir. */
1056 archive_strcat(&file->symlink, ".");
1058 case 0x04: /* Parent dir. */
1059 archive_strcat(&file->symlink, "..");
1061 case 0x08: /* Root of filesystem. */
1062 archive_string_empty(&file->symlink);
1063 archive_strcat(&file->symlink, "/");
1065 case 0x10: /* Undefined (historically "volume root" */
1066 archive_string_empty(&file->symlink);
1067 archive_strcat(&file->symlink, "ROOT");
1069 case 0x20: /* Undefined (historically "hostname") */
1070 archive_strcat(&file->symlink, "hostname");
1073 /* TODO: issue a warning ? */
1077 data_length -= nlen;
1083 release_file(struct iso9660 *iso9660, struct file_info *file)
1085 struct file_info *parent;
1087 if (file->refcount == 0) {
1088 parent = file->parent;
1089 archive_string_free(&file->name);
1090 archive_string_free(&file->symlink);
1092 if (parent != NULL) {
1094 release_file(iso9660, parent);
1100 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660,
1101 struct file_info **pfile)
1103 struct file_info *file;
1108 *pfile = file = next_entry(iso9660);
1110 return (ARCHIVE_EOF);
1112 /* CE area precedes actual file data? Ignore it. */
1113 if (file->ce_offset > file->offset) {
1115 fprintf(stderr, " *** Discarding CE data.\n");
1117 file->ce_offset = 0;
1121 /* Don't waste time seeking for zero-length bodies. */
1122 if (file->size == 0) {
1123 file->offset = iso9660->current_position;
1126 /* If CE exists, find and read it now. */
1127 if (file->ce_offset > 0)
1128 offset = file->ce_offset;
1130 offset = file->offset;
1132 /* Seek forward to the start of the entry. */
1133 if (iso9660->current_position < offset) {
1134 off_t step = offset - iso9660->current_position;
1136 bytes_read = (a->decompressor->skip)(a, step);
1138 return (bytes_read);
1139 iso9660->current_position = offset;
1142 /* We found body of file; handle it now. */
1143 if (offset == file->offset)
1144 return (ARCHIVE_OK);
1146 /* Found CE? Process it and push the file back onto list. */
1147 if (offset == file->ce_offset) {
1149 ssize_t size = file->ce_size;
1151 const unsigned char *rr_start;
1153 file->ce_offset = 0;
1155 bytes_read = (a->decompressor->read_ahead)(a, &p, size);
1156 if (bytes_read > size)
1158 rr_start = (const unsigned char *)p;
1159 parse_rockridge(iso9660, file, rr_start,
1160 rr_start + bytes_read);
1161 (a->decompressor->consume)(a, bytes_read);
1162 iso9660->current_position += bytes_read;
1163 add_entry(iso9660, file);
1168 static struct file_info *
1169 next_entry(struct iso9660 *iso9660)
1172 uint64_t least_end_offset;
1174 struct file_info *r;
1176 if (iso9660->pending_files_used < 1)
1179 /* Assume the first file in the list is the earliest on disk. */
1181 least_end_offset = iso9660->pending_files[0]->offset
1182 + iso9660->pending_files[0]->size;
1184 /* Now, try to find an earlier one. */
1185 for (i = 0; i < iso9660->pending_files_used; i++) {
1186 /* Use the position of the file *end* as our comparison. */
1187 uint64_t end_offset = iso9660->pending_files[i]->offset
1188 + iso9660->pending_files[i]->size;
1189 if (iso9660->pending_files[i]->ce_offset > 0
1190 && iso9660->pending_files[i]->ce_offset < iso9660->pending_files[i]->offset)
1191 end_offset = iso9660->pending_files[i]->ce_offset
1192 + iso9660->pending_files[i]->ce_size;
1193 if (least_end_offset > end_offset) {
1195 least_end_offset = end_offset;
1198 r = iso9660->pending_files[least_index];
1199 iso9660->pending_files[least_index]
1200 = iso9660->pending_files[--iso9660->pending_files_used];
1205 toi(const void *p, int n)
1207 const unsigned char *v = (const unsigned char *)p;
1209 return v[0] + 256 * toi(v + 1, n - 1);
1216 isodate7(const unsigned char *v)
1220 memset(&tm, 0, sizeof(tm));
1222 tm.tm_mon = v[1] - 1;
1227 /* v[6] is the signed timezone offset, in 1/4-hour increments. */
1228 offset = ((const signed char *)v)[6];
1229 if (offset > -48 && offset < 52) {
1230 tm.tm_hour -= offset / 4;
1231 tm.tm_min -= (offset % 4) * 15;
1233 return (time_from_tm(&tm));
1237 isodate17(const unsigned char *v)
1241 memset(&tm, 0, sizeof(tm));
1242 tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100
1243 + (v[2] - '0') * 10 + (v[3] - '0')
1245 tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0');
1246 tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0');
1247 tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0');
1248 tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0');
1249 tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0');
1250 /* v[16] is the signed timezone offset, in 1/4-hour increments. */
1251 offset = ((const signed char *)v)[16];
1252 if (offset > -48 && offset < 52) {
1253 tm.tm_hour -= offset / 4;
1254 tm.tm_min -= (offset % 4) * 15;
1256 return (time_from_tm(&tm));
1260 time_from_tm(struct tm *t)
1263 /* Use platform timegm() if available. */
1266 /* Else use direct calculation using POSIX assumptions. */
1267 /* First, fix up tm_yday based on the year/month/day. */
1269 /* Then we can compute timegm() from first principles. */
1270 return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600
1271 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000
1272 + ((t->tm_year - 69) / 4) * 86400 -
1273 ((t->tm_year - 1) / 100) * 86400
1274 + ((t->tm_year + 299) / 400) * 86400);
1279 build_pathname(struct archive_string *as, struct file_info *file)
1281 if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) {
1282 build_pathname(as, file->parent);
1283 archive_strcat(as, "/");
1285 if (archive_strlen(&file->name) == 0)
1286 archive_strcat(as, ".");
1288 archive_string_concat(as, &file->name);
1294 dump_isodirrec(FILE *out, const unsigned char *isodirrec)
1296 fprintf(out, " l %d,",
1297 toi(isodirrec + DR_length_offset, DR_length_size));
1298 fprintf(out, " a %d,",
1299 toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size));
1300 fprintf(out, " ext 0x%x,",
1301 toi(isodirrec + DR_extent_offset, DR_extent_size));
1302 fprintf(out, " s %d,",
1303 toi(isodirrec + DR_size_offset, DR_extent_size));
1304 fprintf(out, " f 0x%02x,",
1305 toi(isodirrec + DR_flags_offset, DR_flags_size));
1306 fprintf(out, " u %d,",
1307 toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size));
1308 fprintf(out, " ilv %d,",
1309 toi(isodirrec + DR_interleave_offset, DR_interleave_size));
1310 fprintf(out, " seq %d,",
1311 toi(isodirrec + DR_volume_sequence_number_offset, DR_volume_sequence_number_size));
1312 fprintf(out, " nl %d:",
1313 toi(isodirrec + DR_name_len_offset, DR_name_len_size));
1314 fprintf(out, " `%.*s'",
1315 toi(isodirrec + DR_name_len_offset, DR_name_len_size), isodirrec + DR_name_offset);