xref: /freebsd/contrib/libarchive/libarchive/archive_read_support_format_iso9660.c (revision 185becb1e1bd2657c156f78aeb52edac05ba5fb5)
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 
30 #ifdef HAVE_ERRNO_H
31 #include <errno.h>
32 #endif
33 /* #include <stdint.h> */ /* See archive_platform.h */
34 #include <stdio.h>
35 #ifdef HAVE_STDLIB_H
36 #include <stdlib.h>
37 #endif
38 #ifdef HAVE_STRING_H
39 #include <string.h>
40 #endif
41 #include <time.h>
42 #ifdef HAVE_ZLIB_H
43 #include <zlib.h>
44 #endif
45 
46 #include "archive.h"
47 #include "archive_endian.h"
48 #include "archive_entry.h"
49 #include "archive_entry_locale.h"
50 #include "archive_integer.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 	uint64_t	 ce_processed_end;/* End offset of processed CE.	*/
273 	char		 rr_moved;	/* Flag to rr_moved.		*/
274 	char		 rr_moved_has_re_only;
275 	char		 re;		/* Having RRIP "RE" extension.	*/
276 	char		 re_descendant;
277 	uint64_t	 cl_offset;	/* Having RRIP "CL" extension.	*/
278 	int		 time_is_set;	/* Bitmask indicating which times are known */
279 	time_t		 birthtime;	/* File created time.		*/
280 	time_t		 mtime;		/* File last modified time.	*/
281 	time_t		 atime;		/* File last accessed time.	*/
282 	time_t		 ctime;		/* File attribute change time.	*/
283 	uint64_t	 rdev;		/* Device number.		*/
284 	mode_t		 mode;
285 	uid_t		 uid;
286 	gid_t		 gid;
287 	int64_t		 number;
288 	int		 nlinks;
289 	struct archive_string name; /* Pathname */
290 	unsigned char	*utf16be_name;
291 	size_t		 utf16be_bytes;
292 	char		 name_continues; /* Non-zero if name continues */
293 	struct archive_string symlink;
294 	char		 symlink_continues; /* Non-zero if link continues */
295 	/* Set 1 if this file compressed by paged zlib(zisofs) */
296 	int		 pz;
297 	int		 pz_log2_bs; /* Log2 of block size */
298 	uint64_t	 pz_uncompressed_size;
299 	/* Set 1 if this file is multi extent. */
300 	int		 multi_extent;
301 	struct {
302 		struct content	*first;
303 		struct content	**last;
304 	} contents;
305 	struct {
306 		struct file_info	*first;
307 		struct file_info	**last;
308 	} rede_files;
309 };
310 
311 #define BIRTHTIME_IS_SET 1
312 #define MTIME_IS_SET 2
313 #define ATIME_IS_SET 4
314 #define CTIME_IS_SET 8
315 
316 struct heap_queue {
317 	struct file_info **files;
318 	size_t		 allocated;
319 	size_t		 used;
320 };
321 
322 struct iso9660 {
323 	int	magic;
324 #define ISO9660_MAGIC   0x96609660
325 
326 	int opt_support_joliet;
327 	int opt_support_rockridge;
328 
329 	struct archive_string pathname;
330 	char	seenRockridge;	/* Set true if RR extensions are used. */
331 	char	seenSUSP;	/* Set true if SUSP is being used. */
332 	char	seenJoliet;
333 
334 	unsigned char	suspOffset;
335 	struct file_info *rr_moved;
336 	struct read_ce_queue {
337 		struct read_ce_req {
338 			uint64_t	 offset;/* Offset of CE on disk. */
339 			struct file_info *file;
340 		}		*reqs;
341 		size_t		 cnt;
342 		size_t		 allocated;
343 	}	read_ce_req;
344 
345 	int64_t		previous_number;
346 	struct archive_string previous_pathname;
347 
348 	struct file_info		*use_files;
349 	struct heap_queue		 pending_files;
350 	struct {
351 		struct file_info	*first;
352 		struct file_info	**last;
353 	}	cache_files;
354 	struct {
355 		struct file_info	*first;
356 		struct file_info	**last;
357 	}	re_files;
358 
359 	uint64_t current_position;
360 	ssize_t	logical_block_size;
361 	uint64_t volume_size; /* Total size of volume in bytes. */
362 	int32_t  volume_block;/* Total size of volume in logical blocks. */
363 
364 	struct vd {
365 		int		location;	/* Location of Extent.	*/
366 		uint32_t	size;
367 	} primary, joliet;
368 
369 	int64_t	entry_sparse_offset;
370 	int64_t	entry_bytes_remaining;
371 	size_t  entry_bytes_unconsumed;
372 	struct zisofs	 entry_zisofs;
373 	struct content	*entry_content;
374 	struct archive_string_conv *sconv_utf16be;
375 	/*
376 	 * Buffers for a full pathname in UTF-16BE in Joliet extensions.
377 	 */
378 #define UTF16_NAME_MAX	1024
379 	unsigned char *utf16be_path;
380 	size_t		 utf16be_path_len;
381 	unsigned char *utf16be_previous_path;
382 	size_t		 utf16be_previous_path_len;
383 };
384 
385 static int	archive_read_format_iso9660_bid(struct archive_read *, int);
386 static int	archive_read_format_iso9660_options(struct archive_read *,
387 		    const char *, const char *);
388 static int	archive_read_format_iso9660_cleanup(struct archive_read *);
389 static int	archive_read_format_iso9660_read_data(struct archive_read *,
390 		    const void **, size_t *, int64_t *);
391 static int	archive_read_format_iso9660_read_data_skip(struct archive_read *);
392 static int	archive_read_format_iso9660_read_header(struct archive_read *,
393 		    struct archive_entry *);
394 static const char *build_pathname(struct archive_string *, struct file_info *, int);
395 static int	build_pathname_utf16be(unsigned char *, size_t, size_t *,
396 		    struct file_info *);
397 #if DEBUG
398 static void	dump_isodirrec(FILE *, const unsigned char *isodirrec);
399 #endif
400 static time_t	time_from_tm(struct tm *);
401 static time_t	isodate17(const unsigned char *);
402 static int	isodate17_valid(const unsigned char *);
403 static time_t	isodate7(const unsigned char *);
404 static int	isodate7_valid(const unsigned char *);
405 static int	isBootRecord(const unsigned char *);
406 static int	isVolumePartition(struct iso9660 *, const unsigned char *);
407 static int	isVDSetTerminator(const unsigned char *);
408 static int	isJolietSVD(struct iso9660 *, const unsigned char *);
409 static int	isSVD(const unsigned char *);
410 static int	isEVD(const unsigned char *);
411 static int	isPVD(struct iso9660 *, const unsigned char *);
412 static int	isRootDirectoryRecord(const unsigned char *);
413 static int	isValid723Integer(const unsigned char *);
414 static int	isValid733Integer(const unsigned char *);
415 static int	next_cache_entry(struct archive_read *, struct iso9660 *,
416 		    struct file_info **);
417 static int	next_entry_seek(struct archive_read *, struct iso9660 *,
418 		    struct file_info **);
419 static struct file_info *
420 		parse_file_info(struct archive_read *a,
421 		    struct file_info *parent, const unsigned char *isodirrec,
422 		    size_t reclen);
423 static int	parse_rockridge(struct archive_read *a,
424 		    struct file_info *file, const unsigned char *start,
425 		    const unsigned char *end);
426 static int	register_CE(struct archive_read *a, int32_t location,
427 		    struct file_info *file);
428 static int	read_CE(struct archive_read *a, struct iso9660 *iso9660);
429 static void	parse_rockridge_NM1(struct file_info *,
430 		    const unsigned char *, int);
431 static void	parse_rockridge_SL1(struct file_info *,
432 		    const unsigned char *, int);
433 static void	parse_rockridge_TF1(struct file_info *,
434 		    const unsigned char *, int);
435 static void	parse_rockridge_ZF1(struct file_info *,
436 		    const unsigned char *, int);
437 static void	register_file(struct iso9660 *, struct file_info *);
438 static void	release_files(struct iso9660 *);
439 static inline void re_add_entry(struct iso9660 *, struct file_info *);
440 static inline struct file_info * re_get_entry(struct iso9660 *);
441 static inline int rede_add_entry(struct file_info *);
442 static inline struct file_info * rede_get_entry(struct file_info *);
443 static inline void cache_add_entry(struct iso9660 *iso9660,
444 		    struct file_info *file);
445 static inline struct file_info *cache_get_entry(struct iso9660 *iso9660);
446 static int	heap_add_entry(struct archive_read *a, struct heap_queue *heap,
447 		    struct file_info *file, uint64_t key);
448 static struct file_info *heap_get_entry(struct heap_queue *heap);
449 
450 #define add_entry(arch, iso9660, file)	\
451 	heap_add_entry(arch, &((iso9660)->pending_files), file, file->offset)
452 #define next_entry(iso9660)		\
453 	heap_get_entry(&((iso9660)->pending_files))
454 
455 int
archive_read_support_format_iso9660(struct archive * _a)456 archive_read_support_format_iso9660(struct archive *_a)
457 {
458 	struct archive_read *a = (struct archive_read *)_a;
459 	struct iso9660 *iso9660;
460 	int r;
461 
462 	archive_check_magic(_a, ARCHIVE_READ_MAGIC,
463 	    ARCHIVE_STATE_NEW, "archive_read_support_format_iso9660");
464 
465 	iso9660 = calloc(1, sizeof(*iso9660));
466 	if (iso9660 == NULL) {
467 		archive_set_error(&a->archive, ENOMEM,
468 		    "Can't allocate iso9660 data");
469 		return (ARCHIVE_FATAL);
470 	}
471 	iso9660->magic = ISO9660_MAGIC;
472 	iso9660->cache_files.first = NULL;
473 	iso9660->cache_files.last = &(iso9660->cache_files.first);
474 	iso9660->re_files.first = NULL;
475 	iso9660->re_files.last = &(iso9660->re_files.first);
476 	/* Enable to support Joliet extensions by default.	*/
477 	iso9660->opt_support_joliet = 1;
478 	/* Enable to support Rock Ridge extensions by default.	*/
479 	iso9660->opt_support_rockridge = 1;
480 
481 	r = __archive_read_register_format(a,
482 	    iso9660,
483 	    "iso9660",
484 	    archive_read_format_iso9660_bid,
485 	    archive_read_format_iso9660_options,
486 	    archive_read_format_iso9660_read_header,
487 	    archive_read_format_iso9660_read_data,
488 	    archive_read_format_iso9660_read_data_skip,
489 	    NULL,
490 	    archive_read_format_iso9660_cleanup,
491 	    NULL,
492 	    NULL);
493 
494 	if (r != ARCHIVE_OK) {
495 		free(iso9660);
496 		return (r);
497 	}
498 	return (ARCHIVE_OK);
499 }
500 
501 
502 static int
archive_read_format_iso9660_bid(struct archive_read * a,int best_bid)503 archive_read_format_iso9660_bid(struct archive_read *a, int best_bid)
504 {
505 	struct iso9660 *iso9660 = a->format->data;
506 	ssize_t bytes_read;
507 	const unsigned char *p;
508 	int seenTerminator;
509 
510 	/* If there's already a better bid than we can ever
511 	   make, don't bother testing. */
512 	if (best_bid > 48)
513 		return (-1);
514 
515 	/*
516 	 * Skip the first 32k (reserved area) and get the first
517 	 * 8 sectors of the volume descriptor table.  Of course,
518 	 * if the I/O layer gives us more, we'll take it.
519 	 */
520 #define RESERVED_AREA	(SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE)
521 	p = __archive_read_ahead(a,
522 	    RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE,
523 	    &bytes_read);
524 	if (p == NULL)
525 	    return (-1);
526 
527 	/* Skip the reserved area. */
528 	bytes_read -= RESERVED_AREA;
529 	p += RESERVED_AREA;
530 
531 	/* Check each volume descriptor. */
532 	seenTerminator = 0;
533 	for (; bytes_read > LOGICAL_BLOCK_SIZE;
534 	    bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) {
535 		/* Do not handle undefined Volume Descriptor Type. */
536 		if (p[0] >= 4 && p[0] <= 254)
537 			return (0);
538 		/* Standard Identifier must be "CD001" */
539 		if (memcmp(p + 1, "CD001", 5) != 0)
540 			return (0);
541 		if (isPVD(iso9660, p))
542 			continue;
543 		if (!iso9660->joliet.location) {
544 			if (isJolietSVD(iso9660, p))
545 				continue;
546 		}
547 		if (isBootRecord(p))
548 			continue;
549 		if (isEVD(p))
550 			continue;
551 		if (isSVD(p))
552 			continue;
553 		if (isVolumePartition(iso9660, p))
554 			continue;
555 		if (isVDSetTerminator(p)) {
556 			seenTerminator = 1;
557 			break;
558 		}
559 		return (0);
560 	}
561 	/*
562 	 * ISO 9660 format must have Primary Volume Descriptor and
563 	 * Volume Descriptor Set Terminator.
564 	 */
565 	if (seenTerminator && iso9660->primary.location > 16)
566 		return (48);
567 
568 	/* We didn't find a valid PVD; return a bid of zero. */
569 	return (0);
570 }
571 
572 static int
archive_read_format_iso9660_options(struct archive_read * a,const char * key,const char * val)573 archive_read_format_iso9660_options(struct archive_read *a,
574 		const char *key, const char *val)
575 {
576 	struct iso9660 *iso9660 = a->format->data;
577 
578 	if (strcmp(key, "joliet") == 0) {
579 		if (val == NULL || strcmp(val, "off") == 0 ||
580 				strcmp(val, "ignore") == 0 ||
581 				strcmp(val, "disable") == 0 ||
582 				strcmp(val, "0") == 0)
583 			iso9660->opt_support_joliet = 0;
584 		else
585 			iso9660->opt_support_joliet = 1;
586 		return (ARCHIVE_OK);
587 	}
588 	if (strcmp(key, "rockridge") == 0 ||
589 	    strcmp(key, "Rockridge") == 0) {
590 		iso9660->opt_support_rockridge = val != NULL;
591 		return (ARCHIVE_OK);
592 	}
593 
594 	/* Note: The "warn" return is just to inform the options
595 	 * supervisor that we didn't handle it.  It will generate
596 	 * a suitable error if no one used this option. */
597 	return (ARCHIVE_WARN);
598 }
599 
600 static int
isNull(const unsigned char * h,unsigned offset,unsigned bytes)601 isNull(const unsigned char *h, unsigned offset, unsigned bytes)
602 {
603 	if (bytes == 0)
604 		return (1);
605 
606 	/*
607 	 * If the first byte is zero and every byte equals the following
608 	 * byte, the entire range is zero.
609 	 */
610 	return (h[offset] == 0 &&
611 	    (bytes == 1 ||
612 	    memcmp(h + offset, h + offset + 1, bytes - 1) == 0));
613 }
614 
615 static int
isBootRecord(const unsigned char * h)616 isBootRecord(const unsigned char *h)
617 {
618 
619 	/* Type of the Volume Descriptor Boot Record must be 0. */
620 	if (h[0] != 0)
621 		return (0);
622 
623 	/* Volume Descriptor Version must be 1. */
624 	if (h[6] != 1)
625 		return (0);
626 
627 	return (1);
628 }
629 
630 static int
isVolumePartition(struct iso9660 * iso9660,const unsigned char * h)631 isVolumePartition(struct iso9660 *iso9660, const unsigned char *h)
632 {
633 	int32_t location;
634 
635 	/* Type of the Volume Partition Descriptor must be 3. */
636 	if (h[0] != 3)
637 		return (0);
638 
639 	/* Volume Descriptor Version must be 1. */
640 	if (h[6] != 1)
641 		return (0);
642 	/* Unused Field */
643 	if (h[7] != 0)
644 		return (0);
645 
646 	location = archive_le32dec(h + 72);
647 	if (location <= SYSTEM_AREA_BLOCK ||
648 	    location >= iso9660->volume_block)
649 		return (0);
650 	if ((uint32_t)location != archive_be32dec(h + 76))
651 		return (0);
652 
653 	return (1);
654 }
655 
656 static int
isVDSetTerminator(const unsigned char * h)657 isVDSetTerminator(const unsigned char *h)
658 {
659 
660 	/* Type of the Volume Descriptor Set Terminator must be 255. */
661 	if (h[0] != 255)
662 		return (0);
663 
664 	/* Volume Descriptor Version must be 1. */
665 	if (h[6] != 1)
666 		return (0);
667 
668 	/* Reserved field must be 0. */
669 	if (!isNull(h, 7, 2048-7))
670 		return (0);
671 
672 	return (1);
673 }
674 
675 static int
isJolietSVD(struct iso9660 * iso9660,const unsigned char * h)676 isJolietSVD(struct iso9660 *iso9660, const unsigned char *h)
677 {
678 	const unsigned char *p;
679 	ssize_t logical_block_size;
680 	int32_t volume_block;
681 
682 	/* Check if current sector is a kind of Supplementary Volume
683 	 * Descriptor. */
684 	if (!isSVD(h))
685 		return (0);
686 
687 	/* FIXME: do more validations according to joliet spec. */
688 
689 	/* check if this SVD contains joliet extension! */
690 	p = h + SVD_escape_sequences_offset;
691 	/* N.B. Joliet spec says p[1] == '\\', but.... */
692 	if (p[0] == '%' && p[1] == '/') {
693 		int level = 0;
694 
695 		if (p[2] == '@')
696 			level = 1;
697 		else if (p[2] == 'C')
698 			level = 2;
699 		else if (p[2] == 'E')
700 			level = 3;
701 		else /* not joliet */
702 			return (0);
703 
704 		iso9660->seenJoliet = level;
705 
706 	} else /* not joliet */
707 		return (0);
708 
709 	logical_block_size =
710 	    archive_le16dec(h + SVD_logical_block_size_offset);
711 	volume_block = archive_le32dec(h + SVD_volume_space_size_offset);
712 
713 	iso9660->logical_block_size = logical_block_size;
714 	iso9660->volume_block = volume_block;
715 	iso9660->volume_size = logical_block_size * (uint64_t)volume_block;
716 	/* Read Root Directory Record in Volume Descriptor. */
717 	p = h + SVD_root_directory_record_offset;
718 	iso9660->joliet.location = archive_le32dec(p + DR_extent_offset);
719 	iso9660->joliet.size = archive_le32dec(p + DR_size_offset);
720 
721 	return (48);
722 }
723 
724 static int
isSVD(const unsigned char * h)725 isSVD(const unsigned char *h)
726 {
727 	const unsigned char *p;
728 	ssize_t logical_block_size;
729 	int32_t volume_block;
730 	int32_t location;
731 
732 
733 	/* Type 2 means it's a SVD. */
734 	if (h[SVD_type_offset] != 2)
735 		return (0);
736 
737 	/* Reserved field must be 0. */
738 	if (!isNull(h, SVD_reserved1_offset, SVD_reserved1_size))
739 		return (0);
740 	if (!isNull(h, SVD_reserved2_offset, SVD_reserved2_size))
741 		return (0);
742 	if (!isNull(h, SVD_reserved3_offset, SVD_reserved3_size))
743 		return (0);
744 
745 	/* File structure version must be 1 for ISO9660/ECMA119. */
746 	if (h[SVD_file_structure_version_offset] != 1)
747 		return (0);
748 
749 	logical_block_size =
750 	    archive_le16dec(h + SVD_logical_block_size_offset);
751 	if (logical_block_size <= 0)
752 		return (0);
753 
754 	volume_block = archive_le32dec(h + SVD_volume_space_size_offset);
755 	if (volume_block <= SYSTEM_AREA_BLOCK+4)
756 		return (0);
757 
758 	/* Location of Occurrence of Type L Path Table must be
759 	 * available location,
760 	 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
761 	location = archive_le32dec(h+SVD_type_L_path_table_offset);
762 	if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block)
763 		return (0);
764 
765 	/* The Type M Path Table must be at a valid location (WinISO
766 	 * and probably other programs omit this, so we allow zero)
767 	 *
768 	 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
769 	location = archive_be32dec(h+SVD_type_M_path_table_offset);
770 	if ((location > 0 && location < SYSTEM_AREA_BLOCK+2)
771 	    || location >= volume_block)
772 		return (0);
773 
774 	/* Read Root Directory Record in Volume Descriptor. */
775 	p = h + SVD_root_directory_record_offset;
776 	if (!isRootDirectoryRecord(p)) {
777 		return (0);
778 	}
779 
780 	return (48);
781 }
782 
783 static int
isEVD(const unsigned char * h)784 isEVD(const unsigned char *h)
785 {
786 	const unsigned char *p;
787 	ssize_t logical_block_size;
788 	int32_t volume_block;
789 	int32_t location;
790 
791 
792 	/* Type of the Enhanced Volume Descriptor must be 2. */
793 	if (h[PVD_type_offset] != 2)
794 		return (0);
795 
796 	/* EVD version must be 2. */
797 	if (h[PVD_version_offset] != 2)
798 		return (0);
799 
800 	/* Reserved field must be 0. */
801 	if (h[PVD_reserved1_offset] != 0)
802 		return (0);
803 
804 	/* Reserved field must be 0. */
805 	if (!isNull(h, PVD_reserved2_offset, PVD_reserved2_size))
806 		return (0);
807 
808 	/* Reserved field must be 0. */
809 	if (!isNull(h, PVD_reserved3_offset, PVD_reserved3_size))
810 		return (0);
811 
812 	/* Logical block size must be > 0. */
813 	/* I've looked at Ecma 119 and can't find any stronger
814 	 * restriction on this field. */
815 	logical_block_size =
816 	    archive_le16dec(h + PVD_logical_block_size_offset);
817 	if (logical_block_size <= 0)
818 		return (0);
819 
820 	volume_block =
821 	    archive_le32dec(h + PVD_volume_space_size_offset);
822 	if (volume_block <= SYSTEM_AREA_BLOCK+4)
823 		return (0);
824 
825 	/* File structure version must be 2 for ISO9660:1999. */
826 	if (h[PVD_file_structure_version_offset] != 2)
827 		return (0);
828 
829 	/* Location of Occurrence of Type L Path Table must be
830 	 * available location,
831 	 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
832 	location = archive_le32dec(h+PVD_type_1_path_table_offset);
833 	if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block)
834 		return (0);
835 
836 	/* Location of Occurrence of Type M Path Table must be
837 	 * available location,
838 	 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
839 	location = archive_be32dec(h+PVD_type_m_path_table_offset);
840 	if ((location > 0 && location < SYSTEM_AREA_BLOCK+2)
841 	    || location >= volume_block)
842 		return (0);
843 
844 	/* Reserved field must be 0. */
845 	if (!isNull(h, PVD_reserved4_offset, PVD_reserved4_size))
846 		return (0);
847 
848 	/* Reserved field must be 0. */
849 	if (!isNull(h, PVD_reserved5_offset, PVD_reserved5_size))
850 		return (0);
851 
852 	/* Read Root Directory Record in Volume Descriptor. */
853 	p = h + PVD_root_directory_record_offset;
854 	if (!isRootDirectoryRecord(p)) {
855 		return (0);
856 	}
857 
858 	return (48);
859 }
860 
861 static int
isPVD(struct iso9660 * iso9660,const unsigned char * h)862 isPVD(struct iso9660 *iso9660, const unsigned char *h)
863 {
864 	const unsigned char *p;
865 	ssize_t logical_block_size;
866 	int32_t volume_block;
867 	int32_t location;
868 	int i;
869 
870 	/* Type of the Primary Volume Descriptor must be 1. */
871 	if (h[PVD_type_offset] != 1)
872 		return (0);
873 
874 	/* PVD version must be 1. */
875 	if (h[PVD_version_offset] != 1)
876 		return (0);
877 
878 	/* Reserved field must be 0. */
879 	if (h[PVD_reserved1_offset] != 0)
880 		return (0);
881 
882 	/* Reserved field must be 0. */
883 	if (!isNull(h, PVD_reserved2_offset, PVD_reserved2_size))
884 		return (0);
885 
886 	/* Volume space size must be encoded according to 7.3.3 */
887 	if (!isValid733Integer(h + PVD_volume_space_size_offset)) {
888 		return (0);
889 	}
890 	volume_block = archive_le32dec(h + PVD_volume_space_size_offset);
891 	if (volume_block <= SYSTEM_AREA_BLOCK+4)
892 		return (0);
893 
894 	/* Reserved field must be 0. */
895 	if (!isNull(h, PVD_reserved3_offset, PVD_reserved3_size))
896 		return (0);
897 
898 	/* Volume set size must be encoded according to 7.2.3 */
899 	if (!isValid723Integer(h + PVD_volume_set_size_offset)) {
900 		return (0);
901 	}
902 
903 	/* Volume sequence number must be encoded according to 7.2.3 */
904 	if (!isValid723Integer(h + PVD_volume_sequence_number_offset)) {
905 		return (0);
906 	}
907 
908 	/* Logical block size must be > 0. */
909 	/* I've looked at Ecma 119 and can't find any stronger
910 	 * restriction on this field. */
911 	if (!isValid723Integer(h + PVD_logical_block_size_offset)) {
912 		return (0);
913 	}
914 	logical_block_size =
915 	    archive_le16dec(h + PVD_logical_block_size_offset);
916 	if (logical_block_size <= 0)
917 		return (0);
918 
919 	/* Path Table size must be encoded according to 7.3.3 */
920 	if (!isValid733Integer(h + PVD_path_table_size_offset)) {
921 		return (0);
922 	}
923 
924 	/* File structure version must be 1 for ISO9660/ECMA119. */
925 	if (h[PVD_file_structure_version_offset] != 1)
926 		return (0);
927 
928 	/* Location of Occurrence of Type L Path Table must be
929 	 * available location,
930 	 * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
931 	location = archive_le32dec(h+PVD_type_1_path_table_offset);
932 	if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block)
933 		return (0);
934 
935 	/* The Type M Path Table must also be at a valid location
936 	 * (although ECMA 119 requires a Type M Path Table, WinISO and
937 	 * probably other programs omit it, so we permit a zero here)
938 	 *
939 	 * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */
940 	location = archive_be32dec(h+PVD_type_m_path_table_offset);
941 	if ((location > 0 && location < SYSTEM_AREA_BLOCK+2)
942 	    || location >= volume_block)
943 		return (0);
944 
945 	/* Reserved field must be 0. */
946 	/* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */
947 	for (i = 0; i < PVD_reserved4_size; ++i)
948 		if (h[PVD_reserved4_offset + i] != 0
949 		    && h[PVD_reserved4_offset + i] != 0x20)
950 			return (0);
951 
952 	/* Reserved field must be 0. */
953 	if (!isNull(h, PVD_reserved5_offset, PVD_reserved5_size))
954 		return (0);
955 
956 	/* XXX TODO: Check other values for sanity; reject more
957 	 * malformed PVDs. XXX */
958 
959 	/* Read Root Directory Record in Volume Descriptor. */
960 	p = h + PVD_root_directory_record_offset;
961 	if (!isRootDirectoryRecord(p)) {
962 		return (0);
963 	}
964 
965 	if (!iso9660->primary.location) {
966 		iso9660->logical_block_size = logical_block_size;
967 		iso9660->volume_block = volume_block;
968 		iso9660->volume_size =
969 		    logical_block_size * (uint64_t)volume_block;
970 		iso9660->primary.location =
971 		    archive_le32dec(p + DR_extent_offset);
972 		iso9660->primary.size = archive_le32dec(p + DR_size_offset);
973 	}
974 
975 	return (48);
976 }
977 
978 static int
isRootDirectoryRecord(const unsigned char * p)979 isRootDirectoryRecord(const unsigned char *p) {
980 	int flags;
981 
982 	/* ECMA119/ISO9660 requires that the root directory record be _exactly_ 34 bytes.
983 	 * However, we've seen images that have root directory records up to 68 bytes. */
984 	if (p[DR_length_offset] < 34 || p[DR_length_offset] > 68) {
985 		return (0);
986 	}
987 
988 	/* The root directory location must be a 7.3.3 32-bit integer. */
989 	if (!isValid733Integer(p + DR_extent_offset)) {
990 		return (0);
991 	}
992 
993 	/* The root directory size must be a 7.3.3 integer. */
994 	if (!isValid733Integer(p + DR_size_offset)) {
995 		return (0);
996 	}
997 
998 	/* According to the standard, certain bits must be one or zero:
999 	 * Bit 1: must be 1 (this is a directory)
1000 	 * Bit 2: must be 0 (not an associated file)
1001 	 * Bit 3: must be 0 (doesn't use extended attribute record)
1002 	 * Bit 7: must be 0 (final directory record for this file)
1003 	 */
1004 	flags = p[DR_flags_offset];
1005 	if ((flags & 0x8E) != 0x02) {
1006 		return (0);
1007 	}
1008 
1009 	/* Volume sequence number must be a 7.2.3 integer. */
1010 	if (!isValid723Integer(p + DR_volume_sequence_number_offset)) {
1011 		return (0);
1012 	}
1013 
1014 	/* Root directory name is a single zero byte... */
1015 	if (p[DR_name_len_offset] != 1 || p[DR_name_offset] != 0) {
1016 		return (0);
1017 	}
1018 
1019 	/* Nothing looked wrong, so let's accept it. */
1020 	return (1);
1021 }
1022 
1023 static int
read_children(struct archive_read * a,struct file_info * parent)1024 read_children(struct archive_read *a, struct file_info *parent)
1025 {
1026 	struct iso9660 *iso9660 = a->format->data;
1027 	const unsigned char *b, *p;
1028 	struct file_info *multi;
1029 	size_t step, skip_size;
1030 
1031 	/* flush any remaining bytes from the last round to ensure
1032 	 * we're positioned */
1033 	if (iso9660->entry_bytes_unconsumed) {
1034 		__archive_read_consume(a, iso9660->entry_bytes_unconsumed);
1035 		iso9660->entry_bytes_unconsumed = 0;
1036 	}
1037 	if (iso9660->current_position > parent->offset) {
1038 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1039 		    "Ignoring out-of-order directory (%s) %jd > %jd",
1040 		    parent->name.s,
1041 		    (intmax_t)iso9660->current_position,
1042 		    (intmax_t)parent->offset);
1043 		return (ARCHIVE_WARN);
1044 	}
1045 	if (parent->offset + parent->size > iso9660->volume_size) {
1046 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1047 		    "Directory is beyond end-of-media: %s",
1048 		    parent->name.s);
1049 		return (ARCHIVE_WARN);
1050 	}
1051 	if (iso9660->current_position < parent->offset) {
1052 		int64_t skipsize;
1053 
1054 		skipsize = parent->offset - iso9660->current_position;
1055 		skipsize = __archive_read_consume(a, skipsize);
1056 		if (skipsize < 0)
1057 			return ((int)skipsize);
1058 		iso9660->current_position = parent->offset;
1059 	}
1060 
1061 	step = (size_t)(((parent->size + iso9660->logical_block_size -1) /
1062 	    iso9660->logical_block_size) * iso9660->logical_block_size);
1063 	b = __archive_read_ahead(a, step, NULL);
1064 	if (b == NULL) {
1065 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1066 		    "Failed to read full block when scanning "
1067 		    "ISO9660 directory list");
1068 		return (ARCHIVE_FATAL);
1069 	}
1070 	iso9660->current_position += step;
1071 	multi = NULL;
1072 	skip_size = step;
1073 	while (step) {
1074 		p = b;
1075 		b += iso9660->logical_block_size;
1076 		step -= iso9660->logical_block_size;
1077 		for (; p < b && b - p > DR_name_offset && *p != 0 && *p <= b - p;
1078 			p += *p) {
1079 			struct file_info *child;
1080 
1081 			/* N.B.: these special directory identifiers
1082 			 * are 8 bit "values" even on a
1083 			 * Joliet CD with UCS-2 (16bit) encoding.
1084 			 */
1085 
1086 			/* Skip '.' entry. */
1087 			if (*(p + DR_name_len_offset) == 1
1088 			    && *(p + DR_name_offset) == '\0')
1089 				continue;
1090 			/* Skip '..' entry. */
1091 			if (*(p + DR_name_len_offset) == 1
1092 			    && *(p + DR_name_offset) == '\001')
1093 				continue;
1094 			child = parse_file_info(a, parent, p, b - p);
1095 			if (child == NULL) {
1096 				__archive_read_consume(a, skip_size);
1097 				return (ARCHIVE_FATAL);
1098 			}
1099 			if (child->cl_offset == 0 &&
1100 			    (child->multi_extent || multi != NULL)) {
1101 				struct content *con;
1102 
1103 				if (multi == NULL) {
1104 					multi = child;
1105 					multi->contents.first = NULL;
1106 					multi->contents.last =
1107 					    &(multi->contents.first);
1108 				}
1109 				con = malloc(sizeof(struct content));
1110 				if (con == NULL) {
1111 					archive_set_error(
1112 					    &a->archive, ENOMEM,
1113 					    "No memory for multi extent");
1114 					__archive_read_consume(a, skip_size);
1115 					return (ARCHIVE_FATAL);
1116 				}
1117 				con->offset = child->offset;
1118 				con->size = child->size;
1119 				con->next = NULL;
1120 				*multi->contents.last = con;
1121 				multi->contents.last = &(con->next);
1122 				if (multi == child) {
1123 					if (add_entry(a, iso9660, child)
1124 					    != ARCHIVE_OK)
1125 						return (ARCHIVE_FATAL);
1126 				} else {
1127 					multi->size += child->size;
1128 					if (!child->multi_extent)
1129 						multi = NULL;
1130 				}
1131 			} else
1132 				if (add_entry(a, iso9660, child) != ARCHIVE_OK)
1133 					return (ARCHIVE_FATAL);
1134 		}
1135 	}
1136 
1137 	__archive_read_consume(a, skip_size);
1138 
1139 	/* Read data which recorded by RRIP "CE" extension. */
1140 	if (read_CE(a, iso9660) != ARCHIVE_OK)
1141 		return (ARCHIVE_FATAL);
1142 
1143 	return (ARCHIVE_OK);
1144 }
1145 
1146 static int
choose_volume(struct archive_read * a,struct iso9660 * iso9660)1147 choose_volume(struct archive_read *a, struct iso9660 *iso9660)
1148 {
1149 	struct file_info *file;
1150 	int64_t skipsize;
1151 	struct vd *vd;
1152 	const void *block;
1153 	char seenJoliet;
1154 
1155 	vd = &(iso9660->primary);
1156 	if (!iso9660->opt_support_joliet)
1157 		iso9660->seenJoliet = 0;
1158 	if (iso9660->seenJoliet &&
1159 		vd->location > iso9660->joliet.location)
1160 		/* This condition is unlikely; by way of caution. */
1161 		vd = &(iso9660->joliet);
1162 
1163 	skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location;
1164 	skipsize = __archive_read_consume(a, skipsize);
1165 	if (skipsize < 0)
1166 		return ((int)skipsize);
1167 	iso9660->current_position = skipsize;
1168 
1169 	block = __archive_read_ahead(a, vd->size, NULL);
1170 	if (block == NULL) {
1171 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1172 		    "Failed to read full block when scanning "
1173 		    "ISO9660 directory list");
1174 		return (ARCHIVE_FATAL);
1175 	}
1176 
1177 	/*
1178 	 * While reading Root Directory, flag seenJoliet must be zero to
1179 	 * avoid converting special name 0x00(Current Directory) and
1180 	 * next byte to UCS2.
1181 	 */
1182 	seenJoliet = iso9660->seenJoliet;/* Save flag. */
1183 	iso9660->seenJoliet = 0;
1184 	file = parse_file_info(a, NULL, block, vd->size);
1185 	if (file == NULL)
1186 		return (ARCHIVE_FATAL);
1187 	iso9660->seenJoliet = seenJoliet;
1188 
1189 	/*
1190 	 * If the iso image has both RockRidge and Joliet, we preferentially
1191 	 * use RockRidge Extensions rather than Joliet ones.
1192 	 */
1193 	if (vd == &(iso9660->primary) && iso9660->seenRockridge
1194 	    && iso9660->seenJoliet)
1195 		iso9660->seenJoliet = 0;
1196 
1197 	if (vd == &(iso9660->primary) && !iso9660->seenRockridge
1198 	    && iso9660->seenJoliet) {
1199 		/* Switch reading data from primary to joliet. */
1200 		vd = &(iso9660->joliet);
1201 		skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location;
1202 		skipsize -= iso9660->current_position;
1203 		skipsize = __archive_read_consume(a, skipsize);
1204 		if (skipsize < 0)
1205 			return ((int)skipsize);
1206 		iso9660->current_position += skipsize;
1207 
1208 		block = __archive_read_ahead(a, vd->size, NULL);
1209 		if (block == NULL) {
1210 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1211 			    "Failed to read full block when scanning "
1212 			    "ISO9660 directory list");
1213 			return (ARCHIVE_FATAL);
1214 		}
1215 		iso9660->seenJoliet = 0;
1216 		file = parse_file_info(a, NULL, block, vd->size);
1217 		if (file == NULL)
1218 			return (ARCHIVE_FATAL);
1219 		iso9660->seenJoliet = seenJoliet;
1220 	}
1221 
1222 	/* Store the root directory in the pending list. */
1223 	if (add_entry(a, iso9660, file) != ARCHIVE_OK)
1224 		return (ARCHIVE_FATAL);
1225 	if (iso9660->seenRockridge) {
1226 		a->archive.archive_format = ARCHIVE_FORMAT_ISO9660_ROCKRIDGE;
1227 		a->archive.archive_format_name =
1228 		    "ISO9660 with Rockridge extensions";
1229 	}
1230 
1231 	return (ARCHIVE_OK);
1232 }
1233 
1234 static int
archive_read_format_iso9660_read_header(struct archive_read * a,struct archive_entry * entry)1235 archive_read_format_iso9660_read_header(struct archive_read *a,
1236     struct archive_entry *entry)
1237 {
1238 	struct iso9660 *iso9660 = a->format->data;
1239 	struct file_info *file;
1240 	int r, rd_r = ARCHIVE_OK;
1241 
1242 	if (!a->archive.archive_format) {
1243 		a->archive.archive_format = ARCHIVE_FORMAT_ISO9660;
1244 		a->archive.archive_format_name = "ISO9660";
1245 	}
1246 
1247 	if (iso9660->current_position == 0) {
1248 		r = choose_volume(a, iso9660);
1249 		if (r != ARCHIVE_OK)
1250 			return (r);
1251 	}
1252 
1253 	file = NULL;/* Eliminate a warning. */
1254 	/* Get the next entry that appears after the current offset. */
1255 	r = next_entry_seek(a, iso9660, &file);
1256 	if (r != ARCHIVE_OK)
1257 		return (r);
1258 
1259 	if (iso9660->seenJoliet) {
1260 		/*
1261 		 * Convert UTF-16BE of a filename to local locale MBS
1262 		 * and store the result into a filename field.
1263 		 */
1264 		if (iso9660->sconv_utf16be == NULL) {
1265 			iso9660->sconv_utf16be =
1266 			    archive_string_conversion_from_charset(
1267 				&(a->archive), "UTF-16BE", 1);
1268 			if (iso9660->sconv_utf16be == NULL)
1269 				/* Couldn't allocate memory */
1270 				return (ARCHIVE_FATAL);
1271 		}
1272 		if (iso9660->utf16be_path == NULL) {
1273 			iso9660->utf16be_path = malloc(UTF16_NAME_MAX);
1274 			if (iso9660->utf16be_path == NULL) {
1275 				archive_set_error(&a->archive, ENOMEM,
1276 				    "No memory");
1277 				return (ARCHIVE_FATAL);
1278 			}
1279 		}
1280 		if (iso9660->utf16be_previous_path == NULL) {
1281 			iso9660->utf16be_previous_path = calloc(1, UTF16_NAME_MAX);
1282 			if (iso9660->utf16be_previous_path == NULL) {
1283 				archive_set_error(&a->archive, ENOMEM,
1284 				    "No memory");
1285 				return (ARCHIVE_FATAL);
1286 			}
1287 		}
1288 
1289 		iso9660->utf16be_path_len = 0;
1290 		if (build_pathname_utf16be(iso9660->utf16be_path,
1291 		    UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) {
1292 			archive_set_error(&a->archive,
1293 			    ARCHIVE_ERRNO_FILE_FORMAT,
1294 			    "Pathname is too long");
1295 			return (ARCHIVE_FATAL);
1296 		}
1297 
1298 		r = archive_entry_copy_pathname_l(entry,
1299 		    (const char *)iso9660->utf16be_path,
1300 		    iso9660->utf16be_path_len,
1301 		    iso9660->sconv_utf16be);
1302 		if (r != 0) {
1303 			if (errno == ENOMEM) {
1304 				archive_set_error(&a->archive, ENOMEM,
1305 				    "No memory for Pathname");
1306 				return (ARCHIVE_FATAL);
1307 			}
1308 			archive_set_error(&a->archive,
1309 			    ARCHIVE_ERRNO_FILE_FORMAT,
1310 			    "Pathname cannot be converted "
1311 			    "from %s to current locale",
1312 			    archive_string_conversion_charset_name(
1313 			      iso9660->sconv_utf16be));
1314 
1315 			rd_r = ARCHIVE_WARN;
1316 		}
1317 	} else {
1318 		const char *path = build_pathname(&iso9660->pathname, file, 0);
1319 		if (path == NULL) {
1320 			archive_set_error(&a->archive,
1321 			    ARCHIVE_ERRNO_FILE_FORMAT,
1322 			    "Pathname is too long");
1323 			return (ARCHIVE_FATAL);
1324 		} else {
1325 			archive_string_empty(&iso9660->pathname);
1326 			archive_entry_set_pathname(entry, path);
1327 		}
1328 	}
1329 
1330 	iso9660->entry_bytes_remaining = file->size;
1331 	/* Offset for sparse-file-aware clients. */
1332 	iso9660->entry_sparse_offset = 0;
1333 
1334 	if (file->offset + file->size > iso9660->volume_size) {
1335 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1336 		    "File is beyond end-of-media: %s",
1337 		    archive_entry_pathname(entry));
1338 		iso9660->entry_bytes_remaining = 0;
1339 		return (ARCHIVE_WARN);
1340 	}
1341 
1342 	/* Set up the entry structure with information about this entry. */
1343 	archive_entry_set_mode(entry, file->mode);
1344 	archive_entry_set_uid(entry, file->uid);
1345 	archive_entry_set_gid(entry, file->gid);
1346 	archive_entry_set_nlink(entry, file->nlinks);
1347 	if ((file->time_is_set & BIRTHTIME_IS_SET))
1348 		archive_entry_set_birthtime(entry, file->birthtime, 0);
1349 	else
1350 		archive_entry_unset_birthtime(entry);
1351 	if ((file->time_is_set & MTIME_IS_SET))
1352 		archive_entry_set_mtime(entry, file->mtime, 0);
1353 	else
1354 		archive_entry_unset_mtime(entry);
1355 	if ((file->time_is_set & CTIME_IS_SET))
1356 		archive_entry_set_ctime(entry, file->ctime, 0);
1357 	else
1358 		archive_entry_unset_ctime(entry);
1359 	if ((file->time_is_set & ATIME_IS_SET))
1360 		archive_entry_set_atime(entry, file->atime, 0);
1361 	else
1362 		archive_entry_unset_atime(entry);
1363 	/* N.B.: Rock Ridge supports 64-bit device numbers. */
1364 	archive_entry_set_rdev(entry, (dev_t)file->rdev);
1365 	archive_entry_set_size(entry, iso9660->entry_bytes_remaining);
1366 	if (file->symlink.s != NULL)
1367 		archive_entry_copy_symlink(entry, file->symlink.s);
1368 
1369 	/* Note: If the input isn't seekable, we can't rewind to
1370 	 * return the same body again, so if the next entry refers to
1371 	 * the same data, we have to return it as a hardlink to the
1372 	 * original entry. */
1373 	if (file->number != -1 &&
1374 	    file->number == iso9660->previous_number) {
1375 		if (iso9660->seenJoliet) {
1376 			r = archive_entry_copy_hardlink_l(entry,
1377 			    (const char *)iso9660->utf16be_previous_path,
1378 			    iso9660->utf16be_previous_path_len,
1379 			    iso9660->sconv_utf16be);
1380 			if (r != 0) {
1381 				if (errno == ENOMEM) {
1382 					archive_set_error(&a->archive, ENOMEM,
1383 					    "No memory for Linkname");
1384 					return (ARCHIVE_FATAL);
1385 				}
1386 				archive_set_error(&a->archive,
1387 				    ARCHIVE_ERRNO_FILE_FORMAT,
1388 				    "Linkname cannot be converted "
1389 				    "from %s to current locale",
1390 				    archive_string_conversion_charset_name(
1391 				      iso9660->sconv_utf16be));
1392 				rd_r = ARCHIVE_WARN;
1393 			}
1394 		} else
1395 			archive_entry_set_hardlink(entry,
1396 			    iso9660->previous_pathname.s);
1397 		archive_entry_unset_size(entry);
1398 		iso9660->entry_bytes_remaining = 0;
1399 		return (rd_r);
1400 	}
1401 
1402 	if ((file->mode & AE_IFMT) != AE_IFDIR &&
1403 	    file->offset < iso9660->current_position) {
1404 		int64_t r64;
1405 
1406 		r64 = __archive_read_seek(a, file->offset, SEEK_SET);
1407 		if (r64 != (int64_t)file->offset) {
1408 			/* We can't seek backwards to extract it, so issue
1409 			 * a warning.  Note that this can only happen if
1410 			 * this entry was added to the heap after we passed
1411 			 * this offset, that is, only if the directory
1412 			 * mentioning this entry is later than the body of
1413 			 * the entry. Such layouts are very unusual; most
1414 			 * ISO9660 writers lay out and record all directory
1415 			 * information first, then store all file bodies. */
1416 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1417 			    "Ignoring out-of-order file @%jx (%s) %jd < %jd",
1418 			    (uintmax_t)file->number,
1419 			    iso9660->pathname.s,
1420 			    (intmax_t)file->offset,
1421 			    (intmax_t)iso9660->current_position);
1422 			iso9660->entry_bytes_remaining = 0;
1423 			return (ARCHIVE_WARN);
1424 		}
1425 		iso9660->current_position = (uint64_t)r64;
1426 	}
1427 
1428 	/* Initialize zisofs variables. */
1429 	iso9660->entry_zisofs.pz = file->pz;
1430 	if (file->pz) {
1431 #ifdef HAVE_ZLIB_H
1432 		struct zisofs  *zisofs;
1433 
1434 		zisofs = &iso9660->entry_zisofs;
1435 		zisofs->initialized = 0;
1436 		zisofs->pz_log2_bs = file->pz_log2_bs;
1437 		zisofs->pz_uncompressed_size = file->pz_uncompressed_size;
1438 		zisofs->pz_offset = 0;
1439 		zisofs->header_avail = 0;
1440 		zisofs->header_passed = 0;
1441 		zisofs->block_pointers_avail = 0;
1442 #endif
1443 		archive_entry_set_size(entry, file->pz_uncompressed_size);
1444 	}
1445 
1446 	iso9660->previous_number = file->number;
1447 	if (iso9660->seenJoliet) {
1448 		memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path,
1449 		    iso9660->utf16be_path_len);
1450 		iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len;
1451 	} else
1452 		archive_strcpy(
1453 		    &iso9660->previous_pathname, iso9660->pathname.s);
1454 
1455 	/* Reset entry_bytes_remaining if the file is multi extent. */
1456 	iso9660->entry_content = file->contents.first;
1457 	if (iso9660->entry_content != NULL)
1458 		iso9660->entry_bytes_remaining = iso9660->entry_content->size;
1459 
1460 	if (archive_entry_filetype(entry) == AE_IFDIR) {
1461 		/* Overwrite nlinks by proper link number which is
1462 		 * calculated from number of sub directories. */
1463 		archive_entry_set_nlink(entry, 2 + file->subdirs);
1464 		/* Directory data has been read completely. */
1465 		iso9660->entry_bytes_remaining = 0;
1466 	}
1467 
1468 	if (rd_r != ARCHIVE_OK)
1469 		return (rd_r);
1470 	return (ARCHIVE_OK);
1471 }
1472 
1473 static int
archive_read_format_iso9660_read_data_skip(struct archive_read * a)1474 archive_read_format_iso9660_read_data_skip(struct archive_read *a)
1475 {
1476 	/* Because read_next_header always does an explicit skip
1477 	 * to the next entry, we don't need to do anything here. */
1478 	(void)a; /* UNUSED */
1479 	return (ARCHIVE_OK);
1480 }
1481 
1482 #ifdef HAVE_ZLIB_H
1483 
1484 static int
zisofs_read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1485 zisofs_read_data(struct archive_read *a,
1486     const void **buff, size_t *size, int64_t *offset)
1487 {
1488 	struct iso9660 *iso9660 = a->format->data;
1489 	struct zisofs  *zisofs;
1490 	const unsigned char *p;
1491 	size_t avail;
1492 	ssize_t bytes_read;
1493 	size_t uncompressed_size;
1494 	int r;
1495 
1496 	zisofs = &iso9660->entry_zisofs;
1497 
1498 	p = __archive_read_ahead(a, 1, &bytes_read);
1499 	if (bytes_read <= 0) {
1500 		archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1501 		    "Truncated zisofs file body");
1502 		return (ARCHIVE_FATAL);
1503 	}
1504 	if (bytes_read > iso9660->entry_bytes_remaining)
1505 		bytes_read = (ssize_t)iso9660->entry_bytes_remaining;
1506 	avail = bytes_read;
1507 	uncompressed_size = 0;
1508 
1509 	if (!zisofs->initialized) {
1510 		size_t ceil, xsize;
1511 
1512 		/* Allocate block pointers buffer. */
1513 		ceil = (size_t)((zisofs->pz_uncompressed_size +
1514 			(((int64_t)1) << zisofs->pz_log2_bs) - 1)
1515 			>> zisofs->pz_log2_bs);
1516 		xsize = (ceil + 1) * 4;
1517 		if (zisofs->block_pointers_alloc < xsize) {
1518 			size_t alloc;
1519 
1520 			if (zisofs->block_pointers != NULL)
1521 				free(zisofs->block_pointers);
1522 			alloc = ((xsize >> 10) + 1) << 10;
1523 			zisofs->block_pointers = malloc(alloc);
1524 			if (zisofs->block_pointers == NULL) {
1525 				archive_set_error(&a->archive, ENOMEM,
1526 				    "No memory for zisofs decompression");
1527 				return (ARCHIVE_FATAL);
1528 			}
1529 			zisofs->block_pointers_alloc = alloc;
1530 		}
1531 		zisofs->block_pointers_size = xsize;
1532 
1533 		/* Allocate uncompressed data buffer. */
1534 		xsize = (size_t)1UL << zisofs->pz_log2_bs;
1535 		if (zisofs->uncompressed_buffer_size < xsize) {
1536 			if (zisofs->uncompressed_buffer != NULL)
1537 				free(zisofs->uncompressed_buffer);
1538 			zisofs->uncompressed_buffer = malloc(xsize);
1539 			if (zisofs->uncompressed_buffer == NULL) {
1540 				archive_set_error(&a->archive, ENOMEM,
1541 				    "No memory for zisofs decompression");
1542 				return (ARCHIVE_FATAL);
1543 			}
1544 		}
1545 		zisofs->uncompressed_buffer_size = xsize;
1546 
1547 		/*
1548 		 * Read the file header, and check the magic code of zisofs.
1549 		 */
1550 		if (zisofs->header_avail < sizeof(zisofs->header)) {
1551 			xsize = sizeof(zisofs->header) - zisofs->header_avail;
1552 			if (avail < xsize)
1553 				xsize = avail;
1554 			memcpy(zisofs->header + zisofs->header_avail, p, xsize);
1555 			zisofs->header_avail += xsize;
1556 			avail -= xsize;
1557 			p += xsize;
1558 		}
1559 		if (!zisofs->header_passed &&
1560 		    zisofs->header_avail == sizeof(zisofs->header)) {
1561 			int err = 0;
1562 
1563 			if (memcmp(zisofs->header, zisofs_magic,
1564 			    sizeof(zisofs_magic)) != 0)
1565 				err = 1;
1566 			if (archive_le32dec(zisofs->header + 8)
1567 			    != zisofs->pz_uncompressed_size)
1568 				err = 1;
1569 			if (zisofs->header[12] != 4)
1570 				err = 1;
1571 			if (zisofs->header[13] != zisofs->pz_log2_bs)
1572 				err = 1;
1573 			if (err) {
1574 				archive_set_error(&a->archive,
1575 				    ARCHIVE_ERRNO_FILE_FORMAT,
1576 				    "Illegal zisofs file body");
1577 				return (ARCHIVE_FATAL);
1578 			}
1579 			zisofs->header_passed = 1;
1580 		}
1581 		/*
1582 		 * Read block pointers.
1583 		 */
1584 		if (zisofs->header_passed &&
1585 		    zisofs->block_pointers_avail < zisofs->block_pointers_size) {
1586 			xsize = zisofs->block_pointers_size
1587 			    - zisofs->block_pointers_avail;
1588 			if (avail < xsize)
1589 				xsize = avail;
1590 			memcpy(zisofs->block_pointers
1591 			    + zisofs->block_pointers_avail, p, xsize);
1592 			zisofs->block_pointers_avail += xsize;
1593 			avail -= xsize;
1594 			p += xsize;
1595 		    	if (zisofs->block_pointers_avail
1596 			    == zisofs->block_pointers_size) {
1597 				/* We've got all block pointers and initialize
1598 				 * related variables.	*/
1599 				zisofs->block_off = 0;
1600 				zisofs->block_avail = 0;
1601 				/* Complete a initialization */
1602 				zisofs->initialized = 1;
1603 			}
1604 		}
1605 
1606 		if (!zisofs->initialized)
1607 			goto next_data; /* We need more data. */
1608 	}
1609 
1610 	/*
1611 	 * Get block offsets from block pointers.
1612 	 */
1613 	if (zisofs->block_avail == 0) {
1614 		uint32_t bst, bed;
1615 
1616 		if (zisofs->block_off + 4 >= zisofs->block_pointers_size) {
1617 			/* There isn't a pair of offsets. */
1618 			archive_set_error(&a->archive,
1619 			    ARCHIVE_ERRNO_FILE_FORMAT,
1620 			    "Illegal zisofs block pointers");
1621 			return (ARCHIVE_FATAL);
1622 		}
1623 		bst = archive_le32dec(
1624 		    zisofs->block_pointers + zisofs->block_off);
1625 		if (bst != zisofs->pz_offset + (bytes_read - avail)) {
1626 			/* TODO: Should we seek offset of current file
1627 			 * by bst ? */
1628 			archive_set_error(&a->archive,
1629 			    ARCHIVE_ERRNO_FILE_FORMAT,
1630 			    "Illegal zisofs block pointers(cannot seek)");
1631 			return (ARCHIVE_FATAL);
1632 		}
1633 		bed = archive_le32dec(
1634 		    zisofs->block_pointers + zisofs->block_off + 4);
1635 		if (bed < bst) {
1636 			archive_set_error(&a->archive,
1637 			    ARCHIVE_ERRNO_FILE_FORMAT,
1638 			    "Illegal zisofs block pointers");
1639 			return (ARCHIVE_FATAL);
1640 		}
1641 		zisofs->block_avail = bed - bst;
1642 		zisofs->block_off += 4;
1643 
1644 		/* Initialize compression library for new block. */
1645 		if (zisofs->stream_valid)
1646 			r = inflateReset(&zisofs->stream);
1647 		else
1648 			r = inflateInit(&zisofs->stream);
1649 		if (r != Z_OK) {
1650 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1651 			    "Can't initialize zisofs decompression");
1652 			return (ARCHIVE_FATAL);
1653 		}
1654 		zisofs->stream_valid = 1;
1655 		zisofs->stream.total_in = 0;
1656 		zisofs->stream.total_out = 0;
1657 	}
1658 
1659 	/*
1660 	 * Make uncompressed data.
1661 	 */
1662 	if (zisofs->block_avail == 0) {
1663 		memset(zisofs->uncompressed_buffer, 0,
1664 		    zisofs->uncompressed_buffer_size);
1665 		uncompressed_size = zisofs->uncompressed_buffer_size;
1666 	} else {
1667 		zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p;
1668 		if (avail > zisofs->block_avail)
1669 			zisofs->stream.avail_in = zisofs->block_avail;
1670 		else
1671 			zisofs->stream.avail_in = (uInt)avail;
1672 		zisofs->stream.next_out = zisofs->uncompressed_buffer;
1673 		zisofs->stream.avail_out =
1674 		    (uInt)zisofs->uncompressed_buffer_size;
1675 
1676 		r = inflate(&zisofs->stream, 0);
1677 		switch (r) {
1678 		case Z_OK: /* Decompressor made some progress.*/
1679 		case Z_STREAM_END: /* Found end of stream. */
1680 			break;
1681 		default:
1682 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1683 			    "zisofs decompression failed (%d)", r);
1684 			return (ARCHIVE_FATAL);
1685 		}
1686 		uncompressed_size =
1687 		    zisofs->uncompressed_buffer_size - zisofs->stream.avail_out;
1688 		avail -= zisofs->stream.next_in - p;
1689 		zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p);
1690 	}
1691 next_data:
1692 	bytes_read -= avail;
1693 	*buff = zisofs->uncompressed_buffer;
1694 	*size = uncompressed_size;
1695 	*offset = iso9660->entry_sparse_offset;
1696 	iso9660->entry_sparse_offset += uncompressed_size;
1697 	iso9660->entry_bytes_remaining -= bytes_read;
1698 	iso9660->current_position += bytes_read;
1699 	zisofs->pz_offset += (uint32_t)bytes_read;
1700 	iso9660->entry_bytes_unconsumed += bytes_read;
1701 
1702 	return (ARCHIVE_OK);
1703 }
1704 
1705 #else /* HAVE_ZLIB_H */
1706 
1707 static int
zisofs_read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1708 zisofs_read_data(struct archive_read *a,
1709     const void **buff, size_t *size, int64_t *offset)
1710 {
1711 
1712 	(void)buff;/* UNUSED */
1713 	(void)size;/* UNUSED */
1714 	(void)offset;/* UNUSED */
1715 	archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1716 	    "zisofs is not supported on this platform");
1717 	return (ARCHIVE_FAILED);
1718 }
1719 
1720 #endif /* HAVE_ZLIB_H */
1721 
1722 static int
archive_read_format_iso9660_read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1723 archive_read_format_iso9660_read_data(struct archive_read *a,
1724     const void **buff, size_t *size, int64_t *offset)
1725 {
1726 	struct iso9660 *iso9660 = a->format->data;
1727 	ssize_t bytes_read;
1728 
1729 	if (iso9660->entry_bytes_unconsumed) {
1730 		__archive_read_consume(a, iso9660->entry_bytes_unconsumed);
1731 		iso9660->entry_bytes_unconsumed = 0;
1732 	}
1733 
1734 	if (iso9660->entry_bytes_remaining <= 0) {
1735 		if (iso9660->entry_content != NULL)
1736 			iso9660->entry_content = iso9660->entry_content->next;
1737 		if (iso9660->entry_content == NULL) {
1738 			*buff = NULL;
1739 			*size = 0;
1740 			*offset = iso9660->entry_sparse_offset;
1741 			return (ARCHIVE_EOF);
1742 		}
1743 		/* Seek forward to the start of the entry. */
1744 		if (iso9660->current_position < iso9660->entry_content->offset) {
1745 			int64_t step;
1746 
1747 			step = iso9660->entry_content->offset -
1748 			    iso9660->current_position;
1749 			step = __archive_read_consume(a, step);
1750 			if (step < 0)
1751 				return ((int)step);
1752 			iso9660->current_position =
1753 			    iso9660->entry_content->offset;
1754 		}
1755 		if (iso9660->entry_content->offset < iso9660->current_position) {
1756 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1757 			    "Ignoring out-of-order file (%s) %jd < %jd",
1758 			    iso9660->pathname.s,
1759 			    (intmax_t)iso9660->entry_content->offset,
1760 			    (intmax_t)iso9660->current_position);
1761 			*buff = NULL;
1762 			*size = 0;
1763 			*offset = iso9660->entry_sparse_offset;
1764 			return (ARCHIVE_WARN);
1765 		}
1766 		iso9660->entry_bytes_remaining = iso9660->entry_content->size;
1767 	}
1768 	if (iso9660->entry_zisofs.pz)
1769 		return (zisofs_read_data(a, buff, size, offset));
1770 
1771 	*buff = __archive_read_ahead(a, 1, &bytes_read);
1772 	if (bytes_read == 0)
1773 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1774 		    "Truncated input file");
1775 	if (*buff == NULL)
1776 		return (ARCHIVE_FATAL);
1777 	if (bytes_read > iso9660->entry_bytes_remaining)
1778 		bytes_read = (ssize_t)iso9660->entry_bytes_remaining;
1779 	*size = bytes_read;
1780 	*offset = iso9660->entry_sparse_offset;
1781 	iso9660->entry_sparse_offset += bytes_read;
1782 	iso9660->entry_bytes_remaining -= bytes_read;
1783 	iso9660->entry_bytes_unconsumed = bytes_read;
1784 	iso9660->current_position += bytes_read;
1785 	return (ARCHIVE_OK);
1786 }
1787 
1788 static int
archive_read_format_iso9660_cleanup(struct archive_read * a)1789 archive_read_format_iso9660_cleanup(struct archive_read *a)
1790 {
1791 	struct iso9660 *iso9660 = a->format->data;
1792 	int r = ARCHIVE_OK;
1793 
1794 	release_files(iso9660);
1795 	free(iso9660->read_ce_req.reqs);
1796 	archive_string_free(&iso9660->pathname);
1797 	archive_string_free(&iso9660->previous_pathname);
1798 	free(iso9660->pending_files.files);
1799 #ifdef HAVE_ZLIB_H
1800 	free(iso9660->entry_zisofs.uncompressed_buffer);
1801 	free(iso9660->entry_zisofs.block_pointers);
1802 	if (iso9660->entry_zisofs.stream_valid) {
1803 		if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) {
1804 			archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1805 			    "Failed to clean up zlib decompressor");
1806 			r = ARCHIVE_FATAL;
1807 		}
1808 	}
1809 #endif
1810 	free(iso9660->utf16be_path);
1811 	free(iso9660->utf16be_previous_path);
1812 	free(iso9660);
1813 	a->format->data = NULL;
1814 	return (r);
1815 }
1816 
1817 /*
1818  * This routine parses a single ISO directory record, makes sense
1819  * of any extensions, and stores the result in memory.
1820  */
1821 static struct file_info *
parse_file_info(struct archive_read * a,struct file_info * parent,const unsigned char * isodirrec,size_t reclen)1822 parse_file_info(struct archive_read *a, struct file_info *parent,
1823     const unsigned char *isodirrec, size_t reclen)
1824 {
1825 	struct iso9660 *iso9660 = a->format->data;
1826 	struct file_info *file, *filep;
1827 	size_t name_len;
1828 	const unsigned char *rr_start, *rr_end;
1829 	const unsigned char *p;
1830 	size_t dr_len = 0;
1831 	uint64_t fsize, offset;
1832 	int32_t location;
1833 	int flags;
1834 
1835 	if (reclen != 0)
1836 		dr_len = (size_t)isodirrec[DR_length_offset];
1837 	/*
1838 	 * Sanity check that reclen is not zero and dr_len is greater than
1839 	 * reclen but at least 34
1840 	 */
1841 	if (reclen == 0 || reclen < dr_len || dr_len < 34) {
1842 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1843 			"Invalid length of directory record");
1844 		return (NULL);
1845 	}
1846 	name_len = (size_t)isodirrec[DR_name_len_offset];
1847 	location = archive_le32dec(isodirrec + DR_extent_offset);
1848 	fsize = archive_le32dec(isodirrec + DR_size_offset);
1849 	/* Sanity check that name_len doesn't exceed dr_len. */
1850 	if (dr_len - 33 < name_len || name_len == 0) {
1851 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1852 		    "Invalid length of file identifier");
1853 		return (NULL);
1854 	}
1855 	/* Sanity check that location doesn't exceed volume block.
1856 	 * Don't check lower limit of location; it's possibility
1857 	 * the location has negative value when file type is symbolic
1858 	 * link or file size is zero. As far as I know latest mkisofs
1859 	 * do that.
1860 	 */
1861 	if (location > 0 &&
1862 	    (location + ((fsize + iso9660->logical_block_size -1)
1863 	       / iso9660->logical_block_size))
1864 			> (uint32_t)iso9660->volume_block) {
1865 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1866 		    "Invalid location of extent of file");
1867 		return (NULL);
1868 	}
1869 	/* Sanity check that location doesn't have a negative value
1870 	 * when the file is not empty. it's too large. */
1871 	if (fsize != 0 && location < 0) {
1872 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
1873 		    "Invalid location of extent of file");
1874 		return (NULL);
1875 	}
1876 
1877 	/* Sanity check that this entry does not create a cycle. */
1878 	offset = iso9660->logical_block_size * (uint64_t)location;
1879 	for (filep = parent; filep != NULL; filep = filep->parent) {
1880 		if (filep->offset == offset) {
1881 			archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1882 			    "Directory structure contains loop");
1883 			return (NULL);
1884 		}
1885 	}
1886 
1887 	/* Create a new file entry and copy data from the ISO dir record. */
1888 	file = calloc(1, sizeof(*file));
1889 	if (file == NULL) {
1890 		archive_set_error(&a->archive, ENOMEM,
1891 		    "No memory for file entry");
1892 		return (NULL);
1893 	}
1894 	file->parent = parent;
1895 	file->offset = offset;
1896 	file->size = fsize;
1897 	if (isodate7_valid(isodirrec + DR_date_offset)) {
1898 		file->time_is_set |= MTIME_IS_SET | ATIME_IS_SET | CTIME_IS_SET;
1899 		file->mtime = isodate7(isodirrec + DR_date_offset);
1900 		file->ctime = file->atime = file->mtime;
1901 	}
1902 	file->rede_files.first = NULL;
1903 	file->rede_files.last = &(file->rede_files.first);
1904 
1905 	p = isodirrec + DR_name_offset;
1906 	/* Rockridge extensions (if any) follow name.  Compute this
1907 	 * before fidgeting the name_len below. */
1908 	rr_start = p + name_len + (name_len & 1 ? 0 : 1);
1909 	rr_end = isodirrec + dr_len;
1910 
1911 	if (iso9660->seenJoliet) {
1912 		/* Joliet names are max 64 chars (128 bytes) according to spec,
1913 		 * but genisoimage/mkisofs allows recording longer Joliet
1914 		 * names which are 103 UCS2 characters(206 bytes) by their
1915 		 * option '-joliet-long'.
1916 		 */
1917 		if (name_len > 206)
1918 			name_len = 206;
1919 		name_len &= ~1;
1920 
1921 		/* trim trailing first version and dot from filename.
1922 		 *
1923 		 * Remember we were in UTF-16BE land!
1924 		 * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both
1925 		 * 16 bits big endian characters on Joliet.
1926 		 *
1927 		 * TODO: sanitize filename?
1928 		 *       Joliet allows any UCS-2 char except:
1929 		 *       *, /, :, ;, ? and \.
1930 		 */
1931 		/* Chop off trailing ';1' from files. */
1932 		if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';'
1933 		    && p[name_len-2] == 0 && p[name_len-1] == '1')
1934 			name_len -= 4;
1935 #if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */
1936 		/* Chop off trailing '.' from filenames. */
1937 		if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.')
1938 			name_len -= 2;
1939 #endif
1940 		if ((file->utf16be_name = malloc(name_len)) == NULL) {
1941 			archive_set_error(&a->archive, ENOMEM,
1942 			    "No memory for file name");
1943 			goto fail;
1944 		}
1945 		memcpy(file->utf16be_name, p, name_len);
1946 		file->utf16be_bytes = name_len;
1947 	} else {
1948 		/* Chop off trailing ';1' from files. */
1949 		if (name_len > 2 && p[name_len - 2] == ';' &&
1950 				p[name_len - 1] == '1')
1951 			name_len -= 2;
1952 		/* Chop off trailing '.' from filenames. */
1953 		if (name_len > 1 && p[name_len - 1] == '.')
1954 			--name_len;
1955 
1956 		archive_strncpy(&file->name, (const char *)p, name_len);
1957 	}
1958 
1959 	flags = isodirrec[DR_flags_offset];
1960 	if (flags & 0x02)
1961 		file->mode = AE_IFDIR | 0700;
1962 	else
1963 		file->mode = AE_IFREG | 0400;
1964 	if (flags & 0x80)
1965 		file->multi_extent = 1;
1966 	else
1967 		file->multi_extent = 0;
1968 	/*
1969 	 * Use a location for the file number, which is treated as an inode
1970 	 * number to find out hardlink target. If Rockridge extensions is
1971 	 * being used, the file number will be overwritten by FILE SERIAL
1972 	 * NUMBER of RRIP "PX" extension.
1973 	 * Note: Old mkisofs did not record that FILE SERIAL NUMBER
1974 	 * in ISO images.
1975 	 * Note2: xorriso set 0 to the location of a symlink file.
1976 	 */
1977 	if (file->size == 0 && location >= 0) {
1978 		/* If file->size is zero, its location points wrong place,
1979 		 * and so we should not use it for the file number.
1980 		 * When the location has negative value, it can be used
1981 		 * for the file number.
1982 		 */
1983 		file->number = -1;
1984 		/* Do not appear before any directory entries. */
1985 		file->offset = -1;
1986 	} else
1987 		file->number = (int64_t)(uint32_t)location;
1988 
1989 	/* Rockridge extensions overwrite information from above. */
1990 	if (iso9660->opt_support_rockridge) {
1991 		if (parent == NULL && rr_end - rr_start >= 7) {
1992 			p = rr_start;
1993 			if (memcmp(p, "SP\x07\x01\xbe\xef", 6) == 0) {
1994 				/*
1995 				 * SP extension stores the suspOffset
1996 				 * (Number of bytes to skip between
1997 				 * filename and SUSP records.)
1998 				 * It is mandatory by the SUSP standard
1999 				 * (IEEE 1281).
2000 				 *
2001 				 * It allows SUSP to coexist with
2002 				 * non-SUSP uses of the System
2003 				 * Use Area by placing non-SUSP data
2004 				 * before SUSP data.
2005 				 *
2006 				 * SP extension must be in the root
2007 				 * directory entry, disable all SUSP
2008 				 * processing if not found.
2009 				 */
2010 				iso9660->suspOffset = p[6];
2011 				iso9660->seenSUSP = 1;
2012 				rr_start += 7;
2013 			}
2014 		}
2015 		if (iso9660->seenSUSP) {
2016 			int r;
2017 
2018 			file->name_continues = 0;
2019 			file->symlink_continues = 0;
2020 			rr_start += iso9660->suspOffset;
2021 			r = parse_rockridge(a, file, rr_start, rr_end);
2022 			if (r != ARCHIVE_OK)
2023 				goto fail;
2024 			/*
2025 			 * A file size of symbolic link files in ISO images
2026 			 * made by makefs is not zero and its location is
2027 			 * the same as those of next regular file. That is
2028 			 * the same as hard like file and it causes unexpected
2029 			 * error.
2030 			 */
2031 			if (file->size > 0 &&
2032 			    (file->mode & AE_IFMT) == AE_IFLNK) {
2033 				file->size = 0;
2034 				file->number = -1;
2035 				file->offset = -1;
2036 			}
2037 		} else
2038 			/* If there isn't SUSP, disable parsing
2039 			 * rock ridge extensions. */
2040 			iso9660->opt_support_rockridge = 0;
2041 	}
2042 
2043 	file->nlinks = 1;/* Reset nlink. we'll calculate it later. */
2044 	/* Tell file's parent how many children that parent has. */
2045 	if (parent != NULL && (flags & 0x02))
2046 		parent->subdirs++;
2047 
2048 	if (iso9660->seenRockridge) {
2049 		if (parent != NULL && parent->parent == NULL &&
2050 		    (flags & 0x02) && iso9660->rr_moved == NULL &&
2051 		    file->name.s &&
2052 		    (strcmp(file->name.s, "rr_moved") == 0 ||
2053 		     strcmp(file->name.s, ".rr_moved") == 0)) {
2054 			iso9660->rr_moved = file;
2055 			file->rr_moved = 1;
2056 			file->rr_moved_has_re_only = 1;
2057 			file->re = 0;
2058 			parent->subdirs--;
2059 		} else if (file->re) {
2060 			/*
2061 			 * Sanity check: file's parent is rr_moved.
2062 			 */
2063 			if (parent == NULL || parent->rr_moved == 0) {
2064 				archive_set_error(&a->archive,
2065 				    ARCHIVE_ERRNO_MISC,
2066 				    "Invalid Rockridge RE");
2067 				goto fail;
2068 			}
2069 			/*
2070 			 * Sanity check: file does not have "CL" extension.
2071 			 */
2072 			if (file->cl_offset) {
2073 				archive_set_error(&a->archive,
2074 				    ARCHIVE_ERRNO_MISC,
2075 				    "Invalid Rockridge RE and CL");
2076 				goto fail;
2077 			}
2078 			/*
2079 			 * Sanity check: The file type must be a directory.
2080 			 */
2081 			if ((flags & 0x02) == 0) {
2082 				archive_set_error(&a->archive,
2083 				    ARCHIVE_ERRNO_MISC,
2084 				    "Invalid Rockridge RE");
2085 				goto fail;
2086 			}
2087 		} else if (parent != NULL && parent->rr_moved)
2088 			file->rr_moved_has_re_only = 0;
2089 		else if (parent != NULL && (flags & 0x02) &&
2090 		    (parent->re || parent->re_descendant))
2091 			file->re_descendant = 1;
2092 		if (file->cl_offset) {
2093 			struct file_info *r;
2094 
2095 			if (parent == NULL || parent->parent == NULL) {
2096 				archive_set_error(&a->archive,
2097 				    ARCHIVE_ERRNO_MISC,
2098 				    "Invalid Rockridge CL");
2099 				goto fail;
2100 			}
2101 			/*
2102 			 * Sanity check: The file type must be a regular file.
2103 			 */
2104 			if ((flags & 0x02) != 0) {
2105 				archive_set_error(&a->archive,
2106 				    ARCHIVE_ERRNO_MISC,
2107 				    "Invalid Rockridge CL");
2108 				goto fail;
2109 			}
2110 			parent->subdirs++;
2111 			/* Overwrite an offset and a number of this "CL" entry
2112 			 * to appear before other dirs. "+1" to those is to
2113 			 * make sure to appear after "RE" entry which this
2114 			 * "CL" entry should be connected with. */
2115 			file->offset = file->number = file->cl_offset + 1;
2116 
2117 			/*
2118 			 * Sanity check: cl_offset does not point at its
2119 			 * the parents or itself.
2120 			 */
2121 			for (r = parent; r; r = r->parent) {
2122 				if (r->offset == file->cl_offset) {
2123 					archive_set_error(&a->archive,
2124 					    ARCHIVE_ERRNO_MISC,
2125 					    "Invalid Rockridge CL");
2126 					goto fail;
2127 				}
2128 			}
2129 			if (file->cl_offset == file->offset ||
2130 			    parent->rr_moved) {
2131 				archive_set_error(&a->archive,
2132 				    ARCHIVE_ERRNO_MISC,
2133 				    "Invalid Rockridge CL");
2134 				goto fail;
2135 			}
2136 		}
2137 	}
2138 
2139 #if DEBUG
2140 	/* DEBUGGING: Warn about attributes I don't yet fully support. */
2141 	if ((flags & ~0x02) != 0) {
2142 		fprintf(stderr, "\n ** Unrecognized flag: ");
2143 		dump_isodirrec(stderr, isodirrec);
2144 		fprintf(stderr, "\n");
2145 	} else if (archive_le16dec(isodirrec + DR_volume_sequence_number_offset) != 1) {
2146 		fprintf(stderr, "\n ** Unrecognized sequence number: ");
2147 		dump_isodirrec(stderr, isodirrec);
2148 		fprintf(stderr, "\n");
2149 	} else if (*(isodirrec + DR_file_unit_size_offset) != 0) {
2150 		fprintf(stderr, "\n ** Unexpected file unit size: ");
2151 		dump_isodirrec(stderr, isodirrec);
2152 		fprintf(stderr, "\n");
2153 	} else if (*(isodirrec + DR_interleave_offset) != 0) {
2154 		fprintf(stderr, "\n ** Unexpected interleave: ");
2155 		dump_isodirrec(stderr, isodirrec);
2156 		fprintf(stderr, "\n");
2157 	} else if (*(isodirrec + DR_ext_attr_length_offset) != 0) {
2158 		fprintf(stderr, "\n ** Unexpected extended attribute length: ");
2159 		dump_isodirrec(stderr, isodirrec);
2160 		fprintf(stderr, "\n");
2161 	}
2162 #endif
2163 	register_file(iso9660, file);
2164 	return (file);
2165 fail:
2166 	archive_string_free(&file->name);
2167 	free(file);
2168 	return (NULL);
2169 }
2170 
2171 static int
parse_rockridge(struct archive_read * a,struct file_info * file,const unsigned char * p,const unsigned char * end)2172 parse_rockridge(struct archive_read *a, struct file_info *file,
2173     const unsigned char *p, const unsigned char *end)
2174 {
2175 	struct iso9660 *iso9660 = a->format->data;
2176 	int entry_seen = 0;
2177 
2178 	while (p + 4 <= end  /* Enough space for another entry. */
2179 	    && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */
2180 	    && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */
2181 	    && p[2] >= 4 /* Sanity-check length. */
2182 	    && p + p[2] <= end) { /* Sanity-check length. */
2183 		const unsigned char *data = p + 4;
2184 		int data_length = p[2] - 4;
2185 		int version = p[3];
2186 
2187 		switch(p[0]) {
2188 		case 'C':
2189 			if (p[1] == 'E') {
2190 				if (version == 1 && data_length == 24) {
2191 					/*
2192 					 * CE extension comprises:
2193 					 *   8 byte sector containing extension
2194 					 *   8 byte offset w/in above sector
2195 					 *   8 byte length of continuation
2196 					 */
2197 					int32_t location =
2198 					    archive_le32dec(data);
2199 					file->ce_offset =
2200 					    archive_le32dec(data+8);
2201 					file->ce_size =
2202 					    archive_le32dec(data+16);
2203 					if (register_CE(a, location, file)
2204 					    != ARCHIVE_OK)
2205 						return (ARCHIVE_FATAL);
2206 				}
2207 			}
2208 			else if (p[1] == 'L') {
2209 				if (version == 1 && data_length == 8) {
2210 					file->cl_offset = (uint64_t)
2211 					    iso9660->logical_block_size *
2212 					    (uint64_t)archive_le32dec(data);
2213 					iso9660->seenRockridge = 1;
2214 				}
2215 			}
2216 			break;
2217 		case 'N':
2218 			if (p[1] == 'M') {
2219 				if (version == 1) {
2220 					parse_rockridge_NM1(file,
2221 					    data, data_length);
2222 					iso9660->seenRockridge = 1;
2223 				}
2224 			}
2225 			break;
2226 		case 'P':
2227 			/*
2228 			 * PD extension is padding;
2229 			 * contents are always ignored.
2230 			 *
2231 			 * PL extension won't appear;
2232 			 * contents are always ignored.
2233 			 */
2234 			if (p[1] == 'N') {
2235 				if (version == 1 && data_length == 16) {
2236 					file->rdev = archive_le32dec(data);
2237 					file->rdev <<= 32;
2238 					file->rdev |=
2239 					    archive_le32dec(data + 8);
2240 					iso9660->seenRockridge = 1;
2241 				}
2242 			}
2243 			else if (p[1] == 'X') {
2244 				/*
2245 				 * PX extension comprises:
2246 				 *   8 bytes for mode,
2247 				 *   8 bytes for nlinks,
2248 				 *   8 bytes for uid,
2249 				 *   8 bytes for gid,
2250 				 *   8 bytes for inode.
2251 				 */
2252 				if (version == 1) {
2253 					if (data_length >= 8)
2254 						file->mode = (__LA_MODE_T)
2255 						    archive_le32dec(data);
2256 					if (data_length >= 16)
2257 						file->nlinks =
2258 						    archive_le32dec(data + 8);
2259 					if (data_length >= 24)
2260 						file->uid =
2261 						    archive_le32dec(data + 16);
2262 					if (data_length >= 32)
2263 						file->gid =
2264 						    archive_le32dec(data + 24);
2265 					if (data_length >= 40)
2266 						file->number =
2267 						    archive_le32dec(data + 32);
2268 					iso9660->seenRockridge = 1;
2269 				}
2270 			}
2271 			break;
2272 		case 'R':
2273 			if (p[1] == 'E' && version == 1) {
2274 				file->re = 1;
2275 				iso9660->seenRockridge = 1;
2276 			}
2277 			else if (p[1] == 'R' && version == 1) {
2278 				/*
2279 				 * RR extension comprises:
2280 				 *    one byte flag value
2281 				 * This extension is obsolete,
2282 				 * so contents are always ignored.
2283 				 */
2284 			}
2285 			break;
2286 		case 'S':
2287 			if (p[1] == 'L') {
2288 				if (version == 1) {
2289 					parse_rockridge_SL1(file,
2290 					    data, data_length);
2291 					iso9660->seenRockridge = 1;
2292 				}
2293 			}
2294 			else if (p[1] == 'T'
2295 			    && data_length == 0 && version == 1) {
2296 				/*
2297 				 * ST extension marks end of this
2298 				 * block of SUSP entries.
2299 				 *
2300 				 * It allows SUSP to coexist with
2301 				 * non-SUSP uses of the System
2302 				 * Use Area by placing non-SUSP data
2303 				 * after SUSP data.
2304 				 */
2305 				iso9660->seenSUSP = 0;
2306 				iso9660->seenRockridge = 0;
2307 				return (ARCHIVE_OK);
2308 			}
2309 			break;
2310 		case 'T':
2311 			if (p[1] == 'F') {
2312 				if (version == 1) {
2313 					parse_rockridge_TF1(file,
2314 					    data, data_length);
2315 					iso9660->seenRockridge = 1;
2316 				}
2317 			}
2318 			break;
2319 		case 'Z':
2320 			if (p[1] == 'F') {
2321 				if (version == 1)
2322 					parse_rockridge_ZF1(file,
2323 					    data, data_length);
2324 			}
2325 			break;
2326 		default:
2327 			break;
2328 		}
2329 
2330 		p += p[2];
2331 		entry_seen = 1;
2332 	}
2333 
2334 	if (entry_seen)
2335 		return (ARCHIVE_OK);
2336 	else {
2337 		archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
2338 				  "Tried to parse Rockridge extensions, but none found");
2339 		return (ARCHIVE_WARN);
2340 	}
2341 }
2342 
2343 static int
register_CE(struct archive_read * a,int32_t location,struct file_info * file)2344 register_CE(struct archive_read *a, int32_t location,
2345     struct file_info *file)
2346 {
2347 	struct iso9660 *iso9660 = a->format->data;
2348 	struct read_ce_queue *heap;
2349 	uint64_t offset, parent_offset;
2350 	size_t hole, parent;
2351 
2352 	offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size;
2353 	if (((file->mode & AE_IFMT) == AE_IFREG &&
2354 	    offset >= file->offset) ||
2355 	    offset < iso9660->current_position ||
2356 	    (((uint64_t)file->ce_offset) + file->ce_size)
2357 	      > (uint64_t)iso9660->logical_block_size ||
2358 	    offset + file->ce_offset + file->ce_size
2359 		  > iso9660->volume_size) {
2360 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2361 		    "Invalid parameter in SUSP \"CE\" extension");
2362 		return (ARCHIVE_FATAL);
2363 	}
2364 
2365 	/* Expand our CE list as necessary. */
2366 	heap = &(iso9660->read_ce_req);
2367 	if (heap->cnt >= heap->allocated) {
2368 		struct read_ce_req *p;
2369 		size_t new_size;
2370 
2371 		if (heap->allocated < 16)
2372 			new_size = 16;
2373 		else if (archive_ckd_mul_size(&new_size, heap->allocated, 2)) {
2374 			/* Overflow keeps us from growing the list. */
2375 			archive_set_error(&a->archive, ENOMEM, "Out of memory");
2376 			return (ARCHIVE_FATAL);
2377 		}
2378 		p = calloc(new_size, sizeof(p[0]));
2379 		if (p == NULL) {
2380 			archive_set_error(&a->archive, ENOMEM, "Out of memory");
2381 			return (ARCHIVE_FATAL);
2382 		}
2383 		if (heap->reqs != NULL) {
2384 			memcpy(p, heap->reqs, heap->cnt * sizeof(*p));
2385 			free(heap->reqs);
2386 		}
2387 		heap->reqs = p;
2388 		heap->allocated = new_size;
2389 	}
2390 
2391 	/*
2392 	 * Start with hole at end, walk it up tree to find insertion point.
2393 	 */
2394 	hole = heap->cnt++;
2395 	while (hole > 0) {
2396 		parent = (hole - 1)/2;
2397 		parent_offset = heap->reqs[parent].offset;
2398 		if (offset >= parent_offset) {
2399 			heap->reqs[hole].offset = offset;
2400 			heap->reqs[hole].file = file;
2401 			return (ARCHIVE_OK);
2402 		}
2403 		/* Move parent into hole <==> move hole up tree. */
2404 		heap->reqs[hole] = heap->reqs[parent];
2405 		hole = parent;
2406 	}
2407 	heap->reqs[0].offset = offset;
2408 	heap->reqs[0].file = file;
2409 	return (ARCHIVE_OK);
2410 }
2411 
2412 static void
next_CE(struct read_ce_queue * heap)2413 next_CE(struct read_ce_queue *heap)
2414 {
2415 	uint64_t a_offset, b_offset, c_offset;
2416 	size_t a, b, c;
2417 	struct read_ce_req tmp;
2418 
2419 	if (heap->cnt < 1)
2420 		return;
2421 
2422 	/*
2423 	 * Move the last item in the heap to the root of the tree
2424 	 */
2425 	heap->reqs[0] = heap->reqs[--(heap->cnt)];
2426 
2427 	/*
2428 	 * Rebalance the heap.
2429 	 */
2430 	a = 0; /* Starting element and its offset */
2431 	a_offset = heap->reqs[a].offset;
2432 	for (;;) {
2433 		b = a + a + 1; /* First child */
2434 		if (b >= heap->cnt)
2435 			return;
2436 		b_offset = heap->reqs[b].offset;
2437 		c = b + 1; /* Use second child if it is smaller. */
2438 		if (c < heap->cnt) {
2439 			c_offset = heap->reqs[c].offset;
2440 			if (c_offset < b_offset) {
2441 				b = c;
2442 				b_offset = c_offset;
2443 			}
2444 		}
2445 		if (a_offset <= b_offset)
2446 			return;
2447 		tmp = heap->reqs[a];
2448 		heap->reqs[a] = heap->reqs[b];
2449 		heap->reqs[b] = tmp;
2450 		a = b;
2451 	}
2452 }
2453 
2454 
2455 static int
read_CE(struct archive_read * a,struct iso9660 * iso9660)2456 read_CE(struct archive_read *a, struct iso9660 *iso9660)
2457 {
2458 	struct read_ce_queue *heap;
2459 	const unsigned char *b, *p, *end;
2460 	struct file_info *file;
2461 	size_t step;
2462 	uint64_t ce_start, ce_end;
2463 	int r;
2464 
2465 	/* Read data which RRIP "CE" extension points. */
2466 	heap = &(iso9660->read_ce_req);
2467 	step = iso9660->logical_block_size;
2468 	while (heap->cnt &&
2469 	    heap->reqs[0].offset == iso9660->current_position) {
2470 		b = __archive_read_ahead(a, step, NULL);
2471 		if (b == NULL) {
2472 			archive_set_error(&a->archive,
2473 			    ARCHIVE_ERRNO_MISC,
2474 			    "Failed to read full block when scanning "
2475 			    "ISO9660 directory list");
2476 			return (ARCHIVE_FATAL);
2477 		}
2478 		do {
2479 			file = heap->reqs[0].file;
2480 			if (file->ce_offset + file->ce_size > step) {
2481 				archive_set_error(&a->archive,
2482 				    ARCHIVE_ERRNO_FILE_FORMAT,
2483 				    "Malformed CE information");
2484 				return (ARCHIVE_FATAL);
2485 			}
2486 			ce_start = heap->reqs[0].offset + file->ce_offset;
2487 			ce_end = ce_start + file->ce_size;
2488 			if (ce_start < file->ce_processed_end) {
2489 				archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
2490 				    "Invalid parameter in SUSP \"CE\" extension");
2491 				return (ARCHIVE_FATAL);
2492 			}
2493 			p = b + file->ce_offset;
2494 			end = p + file->ce_size;
2495 			file->ce_processed_end = ce_end;
2496 			next_CE(heap);
2497 			r = parse_rockridge(a, file, p, end);
2498 			if (r != ARCHIVE_OK)
2499 				return (ARCHIVE_FATAL);
2500 		} while (heap->cnt &&
2501 		    heap->reqs[0].offset == iso9660->current_position);
2502 		/* NOTE: Do not move this consume's code to front of
2503 		 * do-while loop. Registration of nested CE extension
2504 		 * might cause error because of current position. */
2505 		__archive_read_consume(a, step);
2506 		iso9660->current_position += step;
2507 	}
2508 	return (ARCHIVE_OK);
2509 }
2510 
2511 static void
parse_rockridge_NM1(struct file_info * file,const unsigned char * data,int data_length)2512 parse_rockridge_NM1(struct file_info *file,
2513 		    const unsigned char *data, int data_length)
2514 {
2515 	if (!file->name_continues)
2516 		archive_string_empty(&file->name);
2517 	file->name_continues = 0;
2518 	if (data_length < 1)
2519 		return;
2520 	/*
2521 	 * NM version 1 extension comprises:
2522 	 *   1 byte flag, value is one of:
2523 	 *     = 0: remainder is name
2524 	 *     = 1: remainder is name, next NM entry continues name
2525 	 *     = 2: "."
2526 	 *     = 4: ".."
2527 	 *     = 32: Implementation specific
2528 	 *     All other values are reserved.
2529 	 */
2530 	switch(data[0]) {
2531 	case 0:
2532 		if (data_length < 2)
2533 			return;
2534 		archive_strncat(&file->name,
2535 		    (const char *)data + 1, data_length - 1);
2536 		break;
2537 	case 1:
2538 		if (data_length < 2)
2539 			return;
2540 		archive_strncat(&file->name,
2541 		    (const char *)data + 1, data_length - 1);
2542 		file->name_continues = 1;
2543 		break;
2544 	case 2:
2545 		archive_strcat(&file->name, ".");
2546 		break;
2547 	case 4:
2548 		archive_strcat(&file->name, "..");
2549 		break;
2550 	default:
2551 		return;
2552 	}
2553 
2554 }
2555 
2556 static void
parse_rockridge_TF1(struct file_info * file,const unsigned char * data,int data_length)2557 parse_rockridge_TF1(struct file_info *file, const unsigned char *data,
2558     int data_length)
2559 {
2560 	char flag;
2561 	/*
2562 	 * TF extension comprises:
2563 	 *   one byte flag
2564 	 *   create time (optional)
2565 	 *   modify time (optional)
2566 	 *   access time (optional)
2567 	 *   attribute time (optional)
2568 	 *  Time format and presence of fields
2569 	 *  is controlled by flag bits.
2570 	 */
2571 	if (data_length < 1)
2572 		return;
2573 	flag = data[0];
2574 	++data;
2575 	--data_length;
2576 	if (flag & 0x80) {
2577 		/* Use 17-byte time format. */
2578 		if ((flag & 1) && data_length >= 17) {
2579 			/* Create time. */
2580 			if (isodate17_valid(data)) {
2581 				file->time_is_set |= BIRTHTIME_IS_SET;
2582 				file->birthtime = isodate17(data);
2583 			}
2584 			data += 17;
2585 			data_length -= 17;
2586 		}
2587 		if ((flag & 2) && data_length >= 17) {
2588 			/* Modify time. */
2589 			if (isodate17_valid(data)) {
2590 				file->time_is_set |= MTIME_IS_SET;
2591 				file->mtime = isodate17(data);
2592 			}
2593 			data += 17;
2594 			data_length -= 17;
2595 		}
2596 		if ((flag & 4) && data_length >= 17) {
2597 			/* Access time. */
2598 			if (isodate17_valid(data)) {
2599 				file->time_is_set |= ATIME_IS_SET;
2600 				file->atime = isodate17(data);
2601 			}
2602 			data += 17;
2603 			data_length -= 17;
2604 		}
2605 		if ((flag & 8) && data_length >= 17) {
2606 			/* Attribute change time. */
2607 			if (isodate17_valid(data)) {
2608 				file->time_is_set |= CTIME_IS_SET;
2609 				file->ctime = isodate17(data);
2610 			}
2611 		}
2612 	} else {
2613 		/* Use 7-byte time format. */
2614 		if ((flag & 1) && data_length >= 7) {
2615 			/* Create time. */
2616 			if (isodate7_valid(data)) {
2617 				file->time_is_set |= BIRTHTIME_IS_SET;
2618 				file->birthtime = isodate7(data);
2619 			}
2620 			data += 7;
2621 			data_length -= 7;
2622 		}
2623 		if ((flag & 2) && data_length >= 7) {
2624 			/* Modify time. */
2625 			if (isodate7_valid(data)) {
2626 				file->time_is_set |= MTIME_IS_SET;
2627 				file->mtime = isodate7(data);
2628 			}
2629 			data += 7;
2630 			data_length -= 7;
2631 		}
2632 		if ((flag & 4) && data_length >= 7) {
2633 			/* Access time. */
2634 			if (isodate7_valid(data)) {
2635 				file->time_is_set |= ATIME_IS_SET;
2636 				file->atime = isodate7(data);
2637 			}
2638 			data += 7;
2639 			data_length -= 7;
2640 		}
2641 		if ((flag & 8) && data_length >= 7) {
2642 			/* Attribute change time. */
2643 			if (isodate7_valid(data)) {
2644 				file->time_is_set |= CTIME_IS_SET;
2645 				file->ctime = isodate7(data);
2646 			}
2647 		}
2648 	}
2649 }
2650 
2651 static void
parse_rockridge_SL1(struct file_info * file,const unsigned char * data,int data_length)2652 parse_rockridge_SL1(struct file_info *file, const unsigned char *data,
2653     int data_length)
2654 {
2655 	const char *separator = "";
2656 
2657 	if (!file->symlink_continues || file->symlink.length < 1)
2658 		archive_string_empty(&file->symlink);
2659 	file->symlink_continues = 0;
2660 
2661 	/*
2662 	 * Defined flag values:
2663 	 *  0: This is the last SL record for this symbolic link
2664 	 *  1: this symbolic link field continues in next SL entry
2665 	 *  All other values are reserved.
2666 	 */
2667 	if (data_length < 1)
2668 		return;
2669 	switch(*data) {
2670 	case 0:
2671 		break;
2672 	case 1:
2673 		file->symlink_continues = 1;
2674 		break;
2675 	default:
2676 		return;
2677 	}
2678 	++data;  /* Skip flag byte. */
2679 	--data_length;
2680 
2681 	/*
2682 	 * SL extension body stores "components".
2683 	 * Basically, this is a complicated way of storing
2684 	 * a POSIX path.  It also interferes with using
2685 	 * symlinks for storing non-path data. <sigh>
2686 	 *
2687 	 * Each component is 2 bytes (flag and length)
2688 	 * possibly followed by name data.
2689 	 */
2690 	while (data_length >= 2) {
2691 		unsigned char flag = *data++;
2692 		unsigned char nlen = *data++;
2693 		data_length -= 2;
2694 
2695 		archive_strcat(&file->symlink, separator);
2696 		separator = "/";
2697 
2698 		switch(flag) {
2699 		case 0: /* Usual case, this is text. */
2700 			if (data_length < nlen)
2701 				return;
2702 			archive_strncat(&file->symlink,
2703 			    (const char *)data, nlen);
2704 			break;
2705 		case 0x01: /* Text continues in next component. */
2706 			if (data_length < nlen)
2707 				return;
2708 			archive_strncat(&file->symlink,
2709 			    (const char *)data, nlen);
2710 			separator = "";
2711 			break;
2712 		case 0x02: /* Current dir. */
2713 			archive_strcat(&file->symlink, ".");
2714 			break;
2715 		case 0x04: /* Parent dir. */
2716 			archive_strcat(&file->symlink, "..");
2717 			break;
2718 		case 0x08: /* Root of filesystem. */
2719 			archive_strcat(&file->symlink, "/");
2720 			separator = "";
2721 			break;
2722 		case 0x10: /* Undefined (historically "volume root" */
2723 			archive_string_empty(&file->symlink);
2724 			archive_strcat(&file->symlink, "ROOT");
2725 			break;
2726 		case 0x20: /* Undefined (historically "hostname") */
2727 			archive_strcat(&file->symlink, "hostname");
2728 			break;
2729 		default:
2730 			/* TODO: issue a warning ? */
2731 			return;
2732 		}
2733 		data += nlen;
2734 		data_length -= nlen;
2735 	}
2736 }
2737 
2738 static void
parse_rockridge_ZF1(struct file_info * file,const unsigned char * data,int data_length)2739 parse_rockridge_ZF1(struct file_info *file, const unsigned char *data,
2740     int data_length)
2741 {
2742 
2743 	if (data_length == 12 && data[0] == 0x70 && data[1] == 0x7a) {
2744         /* paged zlib */
2745         file->pz = 1;
2746         file->pz_log2_bs = data[3];
2747         if (file->pz_log2_bs < 15 || file->pz_log2_bs > 17) {
2748             /* TODO: Return an error here instead of silently
2749              * disabling zisofs. That requires propagating an
2750              * error return through parse_rockridge() and its
2751              * callers. */
2752             file->pz = 0;
2753             return;
2754         }
2755         file->pz_uncompressed_size = archive_le32dec(&data[4]);
2756     }
2757 }
2758 
2759 static void
register_file(struct iso9660 * iso9660,struct file_info * file)2760 register_file(struct iso9660 *iso9660, struct file_info *file)
2761 {
2762 
2763 	file->use_next = iso9660->use_files;
2764 	iso9660->use_files = file;
2765 }
2766 
2767 static void
release_files(struct iso9660 * iso9660)2768 release_files(struct iso9660 *iso9660)
2769 {
2770 	struct content *con, *connext;
2771 	struct file_info *file;
2772 
2773 	file = iso9660->use_files;
2774 	while (file != NULL) {
2775 		struct file_info *next = file->use_next;
2776 
2777 		archive_string_free(&file->name);
2778 		archive_string_free(&file->symlink);
2779 		free(file->utf16be_name);
2780 		con = file->contents.first;
2781 		while (con != NULL) {
2782 			connext = con->next;
2783 			free(con);
2784 			con = connext;
2785 		}
2786 		free(file);
2787 		file = next;
2788 	}
2789 }
2790 
2791 static int
next_entry_seek(struct archive_read * a,struct iso9660 * iso9660,struct file_info ** pfile)2792 next_entry_seek(struct archive_read *a, struct iso9660 *iso9660,
2793     struct file_info **pfile)
2794 {
2795 	struct file_info *file;
2796 	int r;
2797 
2798 	r = next_cache_entry(a, iso9660, pfile);
2799 	if (r != ARCHIVE_OK)
2800 		return (r);
2801 	file = *pfile;
2802 
2803 	/* Don't waste time seeking for zero-length bodies. */
2804 	if (file->size == 0)
2805 		file->offset = iso9660->current_position;
2806 
2807 	/* flush any remaining bytes from the last round to ensure
2808 	 * we're positioned */
2809 	if (iso9660->entry_bytes_unconsumed) {
2810 		__archive_read_consume(a, iso9660->entry_bytes_unconsumed);
2811 		iso9660->entry_bytes_unconsumed = 0;
2812 	}
2813 
2814 	/* Seek forward to the start of the entry. */
2815 	if (iso9660->current_position < file->offset) {
2816 		int64_t step;
2817 
2818 		step = file->offset - iso9660->current_position;
2819 		step = __archive_read_consume(a, step);
2820 		if (step < 0)
2821 			return ((int)step);
2822 		iso9660->current_position = file->offset;
2823 	}
2824 
2825 	/* We found body of file; handle it now. */
2826 	return (ARCHIVE_OK);
2827 }
2828 
2829 static int
next_cache_entry(struct archive_read * a,struct iso9660 * iso9660,struct file_info ** pfile)2830 next_cache_entry(struct archive_read *a, struct iso9660 *iso9660,
2831     struct file_info **pfile)
2832 {
2833 	struct file_info *file;
2834 	struct {
2835 		struct file_info	*first;
2836 		struct file_info	**last;
2837 	}	empty_files;
2838 	int64_t number;
2839 	int count;
2840 
2841 	file = cache_get_entry(iso9660);
2842 	if (file != NULL) {
2843 		*pfile = file;
2844 		return (ARCHIVE_OK);
2845 	}
2846 
2847 	for (;;) {
2848 		struct file_info *re, *d;
2849 
2850 		*pfile = file = next_entry(iso9660);
2851 		if (file == NULL) {
2852 			/*
2853 			 * If directory entries all which are descendant of
2854 			 * rr_moved are still remaining, expose their.
2855 			 */
2856 			if (iso9660->re_files.first != NULL &&
2857 			    iso9660->rr_moved != NULL &&
2858 			    iso9660->rr_moved->rr_moved_has_re_only)
2859 				/* Expose "rr_moved" entry. */
2860 				cache_add_entry(iso9660, iso9660->rr_moved);
2861 			while ((re = re_get_entry(iso9660)) != NULL) {
2862 				/* Expose its descendant dirs. */
2863 				while ((d = rede_get_entry(re)) != NULL)
2864 					cache_add_entry(iso9660, d);
2865 			}
2866 			if (iso9660->cache_files.first != NULL)
2867 				return (next_cache_entry(a, iso9660, pfile));
2868 			return (ARCHIVE_EOF);
2869 		}
2870 
2871 		if (file->cl_offset) {
2872 			struct file_info *first_re = NULL;
2873 			int nexted_re = 0;
2874 
2875 			/*
2876 			 * Find "RE" dir for the current file, which
2877 			 * has "CL" flag.
2878 			 */
2879 			while ((re = re_get_entry(iso9660))
2880 			    != first_re) {
2881 				if (first_re == NULL)
2882 					first_re = re;
2883 				if (re->offset == file->cl_offset) {
2884 					re->parent->subdirs--;
2885 					re->parent = file->parent;
2886 					re->re = 0;
2887 					if (re->parent->re_descendant) {
2888 						nexted_re = 1;
2889 						re->re_descendant = 1;
2890 						if (rede_add_entry(re) < 0)
2891 							goto fatal_rr;
2892 						/* Move a list of descendants
2893 						 * to a new ancestor. */
2894 						while ((d = rede_get_entry(
2895 						    re)) != NULL)
2896 							if (rede_add_entry(d)
2897 							    < 0)
2898 								goto fatal_rr;
2899 						break;
2900 					}
2901 					/* Replace the current file
2902 					 * with "RE" dir */
2903 					*pfile = file = re;
2904 					/* Expose its descendant */
2905 					while ((d = rede_get_entry(
2906 					    file)) != NULL)
2907 						cache_add_entry(
2908 						    iso9660, d);
2909 					break;
2910 				} else
2911 					re_add_entry(iso9660, re);
2912 			}
2913 			if (nexted_re) {
2914 				/*
2915 				 * Do not expose this at this time
2916 				 * because we have not gotten its full-path
2917 				 * name yet.
2918 				 */
2919 				continue;
2920 			}
2921 		} else if ((file->mode & AE_IFMT) == AE_IFDIR) {
2922 			int r;
2923 
2924 			/* Read file entries in this dir. */
2925 			r = read_children(a, file);
2926 			if (r != ARCHIVE_OK)
2927 				return (r);
2928 
2929 			/*
2930 			 * Handle a special dir of Rockridge extensions,
2931 			 * "rr_moved".
2932 			 */
2933 			if (file->rr_moved) {
2934 				/*
2935 				 * If this has only the subdirectories which
2936 				 * have "RE" flags, do not expose at this time.
2937 				 */
2938 				if (file->rr_moved_has_re_only)
2939 					continue;
2940 				/* Otherwise expose "rr_moved" entry. */
2941 			} else if (file->re) {
2942 				/*
2943 				 * Do not expose this at this time
2944 				 * because we have not gotten its full-path
2945 				 * name yet.
2946 				 */
2947 				re_add_entry(iso9660, file);
2948 				continue;
2949 			} else if (file->re_descendant) {
2950 				/*
2951 				 * If the top level "RE" entry of this entry
2952 				 * is not exposed, we, accordingly, should not
2953 				 * expose this entry at this time because
2954 				 * we cannot make its proper full-path name.
2955 				 */
2956 				if (rede_add_entry(file) == 0)
2957 					continue;
2958 				/* Otherwise we can expose this entry because
2959 				 * it seems its top level "RE" has already been
2960 				 * exposed. */
2961 			}
2962 		}
2963 		break;
2964 	}
2965 
2966 	if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1)
2967 		return (ARCHIVE_OK);
2968 
2969 	count = 0;
2970 	number = file->number;
2971 	iso9660->cache_files.first = NULL;
2972 	iso9660->cache_files.last = &(iso9660->cache_files.first);
2973 	empty_files.first = NULL;
2974 	empty_files.last = &empty_files.first;
2975 	/* Collect files which has the same file serial number.
2976 	 * Peek pending_files so that file which number is different
2977 	 * is not put back. */
2978 	while (iso9660->pending_files.used > 0 &&
2979 	    (iso9660->pending_files.files[0]->number == -1 ||
2980 	     iso9660->pending_files.files[0]->number == number)) {
2981 		if (file->number == -1) {
2982 			/* This file has the same offset
2983 			 * but it's wrong offset which empty files
2984 			 * and symlink files have.
2985 			 * NOTE: This wrong offset was recorded by
2986 			 * old mkisofs utility. If ISO images is
2987 			 * created by latest mkisofs, this does not
2988 			 * happen.
2989 			 */
2990 			file->next = NULL;
2991 			*empty_files.last = file;
2992 			empty_files.last = &(file->next);
2993 		} else {
2994 			count++;
2995 			cache_add_entry(iso9660, file);
2996 		}
2997 		file = next_entry(iso9660);
2998 	}
2999 
3000 	if (count == 0) {
3001 		*pfile = file;
3002 		return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK);
3003 	}
3004 	if (file->number == -1) {
3005 		file->next = NULL;
3006 		*empty_files.last = file;
3007 		empty_files.last = &(file->next);
3008 	} else {
3009 		count++;
3010 		cache_add_entry(iso9660, file);
3011 	}
3012 
3013 	if (count > 1) {
3014 		/* The count is the same as number of hardlink,
3015 		 * so much so that each nlinks of files in cache_file
3016 		 * is overwritten by value of the count.
3017 		 */
3018 		for (file = iso9660->cache_files.first;
3019 		    file != NULL; file = file->next)
3020 			file->nlinks = count;
3021 	}
3022 	/* If there are empty files, that files are added
3023 	 * to the tail of the cache_files. */
3024 	if (empty_files.first != NULL) {
3025 		*iso9660->cache_files.last = empty_files.first;
3026 		iso9660->cache_files.last = empty_files.last;
3027 	}
3028 	*pfile = cache_get_entry(iso9660);
3029 	return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK);
3030 
3031 fatal_rr:
3032 	archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
3033 	    "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of "
3034 	    "Rockridge extensions: current position = %jd, CL offset = %jd",
3035 	    (intmax_t)iso9660->current_position, (intmax_t)file->cl_offset);
3036 	return (ARCHIVE_FATAL);
3037 }
3038 
3039 static inline void
re_add_entry(struct iso9660 * iso9660,struct file_info * file)3040 re_add_entry(struct iso9660 *iso9660, struct file_info *file)
3041 {
3042 	file->re_next = NULL;
3043 	*iso9660->re_files.last = file;
3044 	iso9660->re_files.last = &(file->re_next);
3045 }
3046 
3047 static inline struct file_info *
re_get_entry(struct iso9660 * iso9660)3048 re_get_entry(struct iso9660 *iso9660)
3049 {
3050 	struct file_info *file;
3051 
3052 	if ((file = iso9660->re_files.first) != NULL) {
3053 		iso9660->re_files.first = file->re_next;
3054 		if (iso9660->re_files.first == NULL)
3055 			iso9660->re_files.last =
3056 			    &(iso9660->re_files.first);
3057 	}
3058 	return (file);
3059 }
3060 
3061 static inline int
rede_add_entry(struct file_info * file)3062 rede_add_entry(struct file_info *file)
3063 {
3064 	struct file_info *re;
3065 
3066 	/*
3067 	 * Find "RE" entry.
3068 	 */
3069 	re = file->parent;
3070 	while (re != NULL && !re->re)
3071 		re = re->parent;
3072 	if (re == NULL)
3073 		return (-1);
3074 
3075 	file->re_next = NULL;
3076 	*re->rede_files.last = file;
3077 	re->rede_files.last = &(file->re_next);
3078 	return (0);
3079 }
3080 
3081 static inline struct file_info *
rede_get_entry(struct file_info * re)3082 rede_get_entry(struct file_info *re)
3083 {
3084 	struct file_info *file;
3085 
3086 	if ((file = re->rede_files.first) != NULL) {
3087 		re->rede_files.first = file->re_next;
3088 		if (re->rede_files.first == NULL)
3089 			re->rede_files.last =
3090 			    &(re->rede_files.first);
3091 	}
3092 	return (file);
3093 }
3094 
3095 static inline void
cache_add_entry(struct iso9660 * iso9660,struct file_info * file)3096 cache_add_entry(struct iso9660 *iso9660, struct file_info *file)
3097 {
3098 	file->next = NULL;
3099 	*iso9660->cache_files.last = file;
3100 	iso9660->cache_files.last = &(file->next);
3101 }
3102 
3103 static inline struct file_info *
cache_get_entry(struct iso9660 * iso9660)3104 cache_get_entry(struct iso9660 *iso9660)
3105 {
3106 	struct file_info *file;
3107 
3108 	if ((file = iso9660->cache_files.first) != NULL) {
3109 		iso9660->cache_files.first = file->next;
3110 		if (iso9660->cache_files.first == NULL)
3111 			iso9660->cache_files.last =
3112 			    &(iso9660->cache_files.first);
3113 	}
3114 	return (file);
3115 }
3116 
3117 static int
heap_add_entry(struct archive_read * a,struct heap_queue * heap,struct file_info * file,uint64_t key)3118 heap_add_entry(struct archive_read *a, struct heap_queue *heap,
3119     struct file_info *file, uint64_t key)
3120 {
3121 	uint64_t file_key, parent_key;
3122 	size_t hole, parent;
3123 
3124 	/* Reserve 16 bits for possible key collisions (needed for linked items) */
3125 	/* For ISO files with more than 65535 entries, reordering will still occur */
3126 	key <<= 16;
3127 	key += heap->used & 0xFFFF;
3128 
3129 	/* Expand our pending files list as necessary. */
3130 	if (heap->used >= heap->allocated) {
3131 		struct file_info **new_pending_files;
3132 		size_t new_size;
3133 
3134 		if (heap->allocated < 1024)
3135 			new_size = 1024;
3136 		else if (archive_ckd_mul_size(&new_size, heap->allocated, 2)) {
3137 			/* Overflow keeps us from growing the list. */
3138 			archive_set_error(&a->archive,
3139 			    ENOMEM, "Out of memory");
3140 			return (ARCHIVE_FATAL);
3141 		}
3142 		new_pending_files = (struct file_info **)
3143 		    calloc(new_size, sizeof(new_pending_files[0]));
3144 		if (new_pending_files == NULL) {
3145 			archive_set_error(&a->archive,
3146 			    ENOMEM, "Out of memory");
3147 			return (ARCHIVE_FATAL);
3148 		}
3149 		if (heap->allocated)
3150 			memcpy(new_pending_files, heap->files,
3151 			    heap->allocated * sizeof(new_pending_files[0]));
3152 		free(heap->files);
3153 		heap->files = new_pending_files;
3154 		heap->allocated = new_size;
3155 	}
3156 
3157 	file_key = file->key = key;
3158 
3159 	/*
3160 	 * Start with hole at end, walk it up tree to find insertion point.
3161 	 */
3162 	hole = heap->used++;
3163 	while (hole > 0) {
3164 		parent = (hole - 1) / 2;
3165 		parent_key = heap->files[parent]->key;
3166 		if (file_key >= parent_key) {
3167 			heap->files[hole] = file;
3168 			return (ARCHIVE_OK);
3169 		}
3170 		/* Move parent into hole <==> move hole up tree. */
3171 		heap->files[hole] = heap->files[parent];
3172 		hole = parent;
3173 	}
3174 	heap->files[0] = file;
3175 
3176 	return (ARCHIVE_OK);
3177 }
3178 
3179 static struct file_info *
heap_get_entry(struct heap_queue * heap)3180 heap_get_entry(struct heap_queue *heap)
3181 {
3182 	uint64_t a_key, b_key, c_key;
3183 	size_t a, b, c;
3184 	struct file_info *r, *tmp;
3185 
3186 	if (heap->used < 1)
3187 		return (NULL);
3188 
3189 	/*
3190 	 * The first file in the list is the earliest; we'll return this.
3191 	 */
3192 	r = heap->files[0];
3193 
3194 	/*
3195 	 * Move the last item in the heap to the root of the tree
3196 	 */
3197 	heap->files[0] = heap->files[--(heap->used)];
3198 
3199 	/*
3200 	 * Rebalance the heap.
3201 	 */
3202 	a = 0; /* Starting element and its heap key */
3203 	a_key = heap->files[a]->key;
3204 	for (;;) {
3205 		b = a + a + 1; /* First child */
3206 		if (b >= heap->used)
3207 			return (r);
3208 		b_key = heap->files[b]->key;
3209 		c = b + 1; /* Use second child if it is smaller. */
3210 		if (c < heap->used) {
3211 			c_key = heap->files[c]->key;
3212 			if (c_key < b_key) {
3213 				b = c;
3214 				b_key = c_key;
3215 			}
3216 		}
3217 		if (a_key <= b_key)
3218 			return (r);
3219 		tmp = heap->files[a];
3220 		heap->files[a] = heap->files[b];
3221 		heap->files[b] = tmp;
3222 		a = b;
3223 	}
3224 }
3225 
3226 /*
3227  * ECMA119/ISO9660 stores multi-byte integers in one of
3228  * three different formats:
3229  *  * Little-endian (specified in section 7.2.1 and 7.3.1)
3230  *  * Big-endian (specified in section 7.2.2 and 7.3.2)
3231  *  * Both (specified in section 7.2.3 and 7.3.3)
3232  *
3233  * For values that follow section 7.2.3 (16-bit) or 7.3.3 (32-bit), we
3234  * can check that the little-endian and big-endian forms agree with
3235  * each other.  This helps us avoid trying to decode files that are
3236  * not really ISO images.
3237  */
3238 static int
isValid723Integer(const unsigned char * p)3239 isValid723Integer(const unsigned char *p) {
3240 	return (p[0] == p[3] && p[1] == p[2]);
3241 }
3242 
3243 static int
isValid733Integer(const unsigned char * p)3244 isValid733Integer(const unsigned char *p)
3245 {
3246 	return (p[0] == p[7]
3247 		&& p[1] == p[6]
3248 		&& p[2] == p[5]
3249 		&& p[3] == p[4]);
3250 }
3251 
3252 static int
isodate7_valid(const unsigned char * v)3253 isodate7_valid(const unsigned char *v)
3254 {
3255 	int year = v[0];
3256 	int month = v[1];
3257 	int day = v[2];
3258 	int hour = v[3];
3259 	int minute = v[4];
3260 	int second = v[5];
3261 	int gmt_off = (signed char)v[6];
3262 
3263 	/* ECMA-119 9.1.5 "If all seven values are zero, it shall mean
3264 	 * that the date is unspecified" */
3265 	if (year == 0
3266 	    && month == 0
3267 	    && day == 0
3268 	    && hour == 0
3269 	    && minute == 0
3270 	    && second == 0
3271 	    && gmt_off == 0)
3272 		return 0;
3273 	/*
3274 	 * Sanity-test each individual field
3275 	 */
3276 	/* Year can have any value */
3277 	/* Month must be 1-12 */
3278 	if (month < 1 || month > 12)
3279 		return 0;
3280 	/* Day must be 1-31 */
3281 	if (day < 1 || day > 31)
3282 		return 0;
3283 	/* Hour must be 0-23 */
3284 	if (hour > 23)
3285 		return 0;
3286 	/* Minute must be 0-59 */
3287 	if (minute > 59)
3288 		return 0;
3289 	/* second must be 0-59 according to ECMA-119 9.1.5 */
3290 	/* BUT: we should probably allow for the time being in UTC, which
3291 	   allows up to 61 seconds in a minute in certain cases */
3292 	if (second > 61)
3293 		return 0;
3294 	/* Offset from GMT must be -48 to +52 */
3295 	if (gmt_off < -48 || gmt_off > +52)
3296 		return 0;
3297 
3298 	/* All tests pass, this is OK */
3299 	return 1;
3300 }
3301 
3302 static time_t
isodate7(const unsigned char * v)3303 isodate7(const unsigned char *v)
3304 {
3305 	struct tm tm;
3306 	int offset;
3307 	time_t t;
3308 
3309 	memset(&tm, 0, sizeof(tm));
3310 	tm.tm_year = v[0];
3311 	tm.tm_mon = v[1] - 1;
3312 	tm.tm_mday = v[2];
3313 	tm.tm_hour = v[3];
3314 	tm.tm_min = v[4];
3315 	tm.tm_sec = v[5];
3316 	/* v[6] is the signed timezone offset, in 1/4-hour increments. */
3317 	offset = ((const signed char *)v)[6];
3318 	if (offset > -48 && offset < 52) {
3319 		tm.tm_hour -= offset / 4;
3320 		tm.tm_min -= (offset % 4) * 15;
3321 	}
3322 	t = time_from_tm(&tm);
3323 	if (t == (time_t)-1)
3324 		return ((time_t)0);
3325 	return (t);
3326 }
3327 
3328 static int
isodate17_valid(const unsigned char * v)3329 isodate17_valid(const unsigned char *v)
3330 {
3331 	/* First 16 bytes are all ASCII digits */
3332 	for (int i = 0; i < 16; i++) {
3333 		if (v[i] < '0' || v[i] > '9')
3334 			return 0;
3335 	}
3336 
3337 	int year = (v[0] - '0') * 1000 + (v[1] - '0') * 100
3338 		+ (v[2] - '0') * 10 + (v[3] - '0');
3339 	int month = (v[4] - '0') * 10 + (v[5] - '0');
3340 	int day = (v[6] - '0') * 10 + (v[7] - '0');
3341 	int hour = (v[8] - '0') * 10 + (v[9] - '0');
3342 	int minute = (v[10] - '0') * 10 + (v[11] - '0');
3343 	int second = (v[12] - '0') * 10 + (v[13] - '0');
3344 	int hundredths = (v[14] - '0') * 10 + (v[15] - '0');
3345 	int gmt_off = (signed char)v[16];
3346 
3347 	if (year == 0 && month == 0 && day == 0
3348 	    && hour == 0 && minute == 0 && second == 0
3349 	    && hundredths == 0 && gmt_off == 0)
3350 		return 0;
3351 	/*
3352 	 * Sanity-test each individual field
3353 	 */
3354 
3355 	/* Year must be 1900-2300 */
3356 	/* (Not specified in ECMA-119, but these seem
3357 	   like reasonable limits. */
3358 	if (year < 1900 || year > 2300)
3359 		return 0;
3360 	/* Month must be 1-12 */
3361 	if (month < 1 || month > 12)
3362 		return 0;
3363 	/* Day must be 1-31 */
3364 	if (day < 1 || day > 31)
3365 		return 0;
3366 	/* Hour must be 0-23 */
3367 	if (hour > 23)
3368 		return 0;
3369 	/* Minute must be 0-59 */
3370 	if (minute > 59)
3371 		return 0;
3372 	/* second must be 0-59 according to ECMA-119 9.1.5 */
3373 	/* BUT: we should probably allow for the time being in UTC, which
3374 	   allows up to 61 seconds in a minute in certain cases */
3375 	if (second > 61)
3376 		return 0;
3377 	/* Hundredths must be 0-99 */
3378 	if (hundredths > 99)
3379 		return 0;
3380 	/* Offset from GMT must be -48 to +52 */
3381 	if (gmt_off < -48 || gmt_off > +52)
3382 		return 0;
3383 
3384 	/* All tests pass, this is OK */
3385 	return 1;
3386 
3387 }
3388 
3389 static time_t
isodate17(const unsigned char * v)3390 isodate17(const unsigned char *v)
3391 {
3392 	struct tm tm;
3393 	int offset;
3394 	time_t t;
3395 
3396 	memset(&tm, 0, sizeof(tm));
3397 	tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100
3398 	    + (v[2] - '0') * 10 + (v[3] - '0')
3399 	    - 1900;
3400 	tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0') - 1;
3401 	tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0');
3402 	tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0');
3403 	tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0');
3404 	tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0');
3405 	/* v[16] is the signed timezone offset, in 1/4-hour increments. */
3406 	offset = ((const signed char *)v)[16];
3407 	if (offset > -48 && offset < 52) {
3408 		tm.tm_hour -= offset / 4;
3409 		tm.tm_min -= (offset % 4) * 15;
3410 	}
3411 	t = time_from_tm(&tm);
3412 	if (t == (time_t)-1)
3413 		return ((time_t)0);
3414 	return (t);
3415 }
3416 
3417 static time_t
time_from_tm(struct tm * t)3418 time_from_tm(struct tm *t)
3419 {
3420 #if HAVE__MKGMTIME
3421         return _mkgmtime(t);
3422 #elif HAVE_TIMEGM
3423         /* Use platform timegm() if available. */
3424         return (timegm(t));
3425 #else
3426         /* Else use direct calculation using POSIX assumptions. */
3427         /* First, fix up tm_yday based on the year/month/day. */
3428         if (mktime(t) == (time_t)-1)
3429                 return ((time_t)-1);
3430         /* Then we can compute timegm() from first principles. */
3431         return (t->tm_sec
3432             + t->tm_min * 60
3433             + t->tm_hour * 3600
3434             + t->tm_yday * 86400
3435             + (t->tm_year - 70) * 31536000
3436             + ((t->tm_year - 69) / 4) * 86400
3437             - ((t->tm_year - 1) / 100) * 86400
3438             + ((t->tm_year + 299) / 400) * 86400);
3439 #endif
3440 }
3441 
3442 static const char *
build_pathname(struct archive_string * as,struct file_info * file,int depth)3443 build_pathname(struct archive_string *as, struct file_info *file, int depth)
3444 {
3445 	// Plain ISO9660 only allows 8 dir levels; if we get
3446 	// to 1000, then something is very, very wrong.
3447 	if (depth > 1000) {
3448 		return NULL;
3449 	}
3450 	if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) {
3451 		if (build_pathname(as, file->parent, depth + 1) == NULL) {
3452 			return NULL;
3453 		}
3454 		archive_strcat(as, "/");
3455 	}
3456 	if (archive_strlen(&file->name) == 0)
3457 		archive_strcat(as, ".");
3458 	else
3459 		archive_string_concat(as, &file->name);
3460 	return (as->s);
3461 }
3462 
3463 static int
build_pathname_utf16be(unsigned char * p,size_t max,size_t * len,struct file_info * file)3464 build_pathname_utf16be(unsigned char *p, size_t max, size_t *len,
3465     struct file_info *file)
3466 {
3467 	if (file->parent != NULL && file->parent->utf16be_bytes > 0) {
3468 		if (build_pathname_utf16be(p, max, len, file->parent) != 0)
3469 			return (-1);
3470 		if (*len + 2 > max)
3471 			return (-1);/* Path is too long! */
3472 		p[*len] = 0;
3473 		p[*len + 1] = '/';
3474 		*len += 2;
3475 	}
3476 	if (file->utf16be_bytes == 0) {
3477 		if (*len + 2 > max)
3478 			return (-1);/* Path is too long! */
3479 		p[*len] = 0;
3480 		p[*len + 1] = '.';
3481 		*len += 2;
3482 	} else {
3483 		if (*len + file->utf16be_bytes > max)
3484 			return (-1);/* Path is too long! */
3485 		memcpy(p + *len, file->utf16be_name, file->utf16be_bytes);
3486 		*len += file->utf16be_bytes;
3487 	}
3488 	return (0);
3489 }
3490 
3491 #if DEBUG
3492 static void
dump_isodirrec(FILE * out,const unsigned char * isodirrec)3493 dump_isodirrec(FILE *out, const unsigned char *isodirrec)
3494 {
3495 	fprintf(out, " l %d,",
3496 	    isodirrec[DR_length_offset]);
3497 	fprintf(out, " a %d,",
3498 	    isodirrec[DR_ext_attr_length_offset]);
3499 	fprintf(out, " ext 0x%x,",
3500 	    archive_le32dec(isodirrec + DR_extent_offset));
3501 	fprintf(out, " s %d,",
3502 	    archive_le32dec(isodirrec + DR_size_offset));
3503 	fprintf(out, " f 0x%x,",
3504 	    isodirrec[DR_flags_offset]);
3505 	fprintf(out, " u %d,",
3506 	    isodirrec[DR_file_unit_size_offset]);
3507 	fprintf(out, " ilv %d,",
3508 	    isodirrec[DR_interleave_offset]);
3509 	fprintf(out, " seq %d,",
3510 	    archive_le16dec(isodirrec + DR_volume_sequence_number_offset));
3511 	fprintf(out, " nl %d:",
3512 	    isodirrec[DR_name_len_offset]);
3513 	fprintf(out, " `%.*s'",
3514 	    isodirrec[DR_name_len_offset], isodirrec + DR_name_offset);
3515 }
3516 #endif
3517