1 /*-
2 * Copyright (c) 2003-2007 Tim Kientzle
3 * Copyright (c) 2008 Joerg Sonnenberger
4 * Copyright (c) 2011-2012 Michihiro NAKAJIMA
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 #include "archive_platform.h"
29 __FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_mtree.c 201165 2009-12-29 05:52:13Z kientzle $");
30
31 #ifdef HAVE_SYS_STAT_H
32 #include <sys/stat.h>
33 #endif
34 #ifdef HAVE_ERRNO_H
35 #include <errno.h>
36 #endif
37 #ifdef HAVE_FCNTL_H
38 #include <fcntl.h>
39 #endif
40 #include <stddef.h>
41 /* #include <stdint.h> */ /* See archive_platform.h */
42 #ifdef HAVE_STDLIB_H
43 #include <stdlib.h>
44 #endif
45 #ifdef HAVE_STRING_H
46 #include <string.h>
47 #endif
48 #ifdef HAVE_CTYPE_H
49 #include <ctype.h>
50 #endif
51
52 #include "archive.h"
53 #include "archive_entry.h"
54 #include "archive_entry_private.h"
55 #include "archive_private.h"
56 #include "archive_rb.h"
57 #include "archive_read_private.h"
58 #include "archive_string.h"
59 #include "archive_pack_dev.h"
60
61 #ifndef O_BINARY
62 #define O_BINARY 0
63 #endif
64 #ifndef O_CLOEXEC
65 #define O_CLOEXEC 0
66 #endif
67
68 #define MTREE_HAS_DEVICE 0x0001
69 #define MTREE_HAS_FFLAGS 0x0002
70 #define MTREE_HAS_GID 0x0004
71 #define MTREE_HAS_GNAME 0x0008
72 #define MTREE_HAS_MTIME 0x0010
73 #define MTREE_HAS_NLINK 0x0020
74 #define MTREE_HAS_PERM 0x0040
75 #define MTREE_HAS_SIZE 0x0080
76 #define MTREE_HAS_TYPE 0x0100
77 #define MTREE_HAS_UID 0x0200
78 #define MTREE_HAS_UNAME 0x0400
79
80 #define MTREE_HAS_OPTIONAL 0x0800
81 #define MTREE_HAS_NOCHANGE 0x1000 /* FreeBSD specific */
82
83 #define MAX_LINE_LEN (1024 * 1024)
84
85 struct mtree_option {
86 struct mtree_option *next;
87 char *value;
88 };
89
90 struct mtree_entry {
91 struct archive_rb_node rbnode;
92 struct mtree_entry *next_dup;
93 struct mtree_entry *next;
94 struct mtree_option *options;
95 char *name;
96 char full;
97 char used;
98 };
99
100 struct mtree {
101 struct archive_string line;
102 size_t buffsize;
103 char *buff;
104 int64_t offset;
105 int fd;
106 int archive_format;
107 const char *archive_format_name;
108 struct mtree_entry *entries;
109 struct mtree_entry *this_entry;
110 struct archive_rb_tree entry_rbtree;
111 struct archive_string current_dir;
112 struct archive_string contents_name;
113
114 struct archive_entry_linkresolver *resolver;
115 struct archive_rb_tree rbtree;
116
117 int64_t cur_size;
118 char checkfs;
119 };
120
121 static int bid_keycmp(const char *, const char *, ssize_t);
122 static int cleanup(struct archive_read *);
123 static int detect_form(struct archive_read *, int *);
124 static int mtree_bid(struct archive_read *, int);
125 static int parse_file(struct archive_read *, struct archive_entry *,
126 struct mtree *, struct mtree_entry *, int *);
127 static void parse_escapes(char *, struct mtree_entry *);
128 static int parse_line(struct archive_read *, struct archive_entry *,
129 struct mtree *, struct mtree_entry *, int *);
130 static int parse_keyword(struct archive_read *, struct mtree *,
131 struct archive_entry *, struct mtree_option *, int *);
132 static int read_data(struct archive_read *a,
133 const void **buff, size_t *size, int64_t *offset);
134 static ssize_t readline(struct archive_read *, struct mtree *, char **, ssize_t);
135 static int skip(struct archive_read *a);
136 static int read_header(struct archive_read *,
137 struct archive_entry *);
138 static int64_t mtree_atol(char **, int base);
139 #ifndef HAVE_STRNLEN
140 static size_t mtree_strnlen(const char *, size_t);
141 #endif
142
143 /*
144 * There's no standard for TIME_T_MAX/TIME_T_MIN. So we compute them
145 * here. TODO: Move this to configure time, but be careful
146 * about cross-compile environments.
147 */
148 static int64_t
get_time_t_max(void)149 get_time_t_max(void)
150 {
151 #if defined(TIME_T_MAX)
152 return TIME_T_MAX;
153 #else
154 /* ISO C allows time_t to be a floating-point type,
155 but POSIX requires an integer type. The following
156 should work on any system that follows the POSIX
157 conventions. */
158 if (((time_t)0) < ((time_t)-1)) {
159 /* Time_t is unsigned */
160 return (~(time_t)0);
161 } else {
162 /* Time_t is signed. */
163 /* Assume it's the same as int64_t or int32_t */
164 if (sizeof(time_t) == sizeof(int64_t)) {
165 return (time_t)INT64_MAX;
166 } else {
167 return (time_t)INT32_MAX;
168 }
169 }
170 #endif
171 }
172
173 static int64_t
get_time_t_min(void)174 get_time_t_min(void)
175 {
176 #if defined(TIME_T_MIN)
177 return TIME_T_MIN;
178 #else
179 if (((time_t)0) < ((time_t)-1)) {
180 /* Time_t is unsigned */
181 return (time_t)0;
182 } else {
183 /* Time_t is signed. */
184 if (sizeof(time_t) == sizeof(int64_t)) {
185 return (time_t)INT64_MIN;
186 } else {
187 return (time_t)INT32_MIN;
188 }
189 }
190 #endif
191 }
192
193 #ifdef HAVE_STRNLEN
194 #define mtree_strnlen(a,b) strnlen(a,b)
195 #else
196 static size_t
mtree_strnlen(const char * p,size_t maxlen)197 mtree_strnlen(const char *p, size_t maxlen)
198 {
199 size_t i;
200
201 for (i = 0; i <= maxlen; i++) {
202 if (p[i] == 0)
203 break;
204 }
205 if (i > maxlen)
206 return (-1);/* invalid */
207 return (i);
208 }
209 #endif
210
211 static int
archive_read_format_mtree_options(struct archive_read * a,const char * key,const char * val)212 archive_read_format_mtree_options(struct archive_read *a,
213 const char *key, const char *val)
214 {
215 struct mtree *mtree;
216
217 mtree = (struct mtree *)(a->format->data);
218 if (strcmp(key, "checkfs") == 0) {
219 /* Allows to read information missing from the mtree from the file system */
220 if (val == NULL || val[0] == 0) {
221 mtree->checkfs = 0;
222 } else {
223 mtree->checkfs = 1;
224 }
225 return (ARCHIVE_OK);
226 }
227
228 /* Note: The "warn" return is just to inform the options
229 * supervisor that we didn't handle it. It will generate
230 * a suitable error if no one used this option. */
231 return (ARCHIVE_WARN);
232 }
233
234 static void
free_options(struct mtree_option * head)235 free_options(struct mtree_option *head)
236 {
237 struct mtree_option *next;
238
239 for (; head != NULL; head = next) {
240 next = head->next;
241 free(head->value);
242 free(head);
243 }
244 }
245
246 static int
mtree_cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)247 mtree_cmp_node(const struct archive_rb_node *n1,
248 const struct archive_rb_node *n2)
249 {
250 const struct mtree_entry *e1 = (const struct mtree_entry *)n1;
251 const struct mtree_entry *e2 = (const struct mtree_entry *)n2;
252
253 return (strcmp(e1->name, e2->name));
254 }
255
256 static int
mtree_cmp_key(const struct archive_rb_node * n,const void * key)257 mtree_cmp_key(const struct archive_rb_node *n, const void *key)
258 {
259 const struct mtree_entry *e = (const struct mtree_entry *)n;
260
261 return (strcmp(e->name, key));
262 }
263
264 int
archive_read_support_format_mtree(struct archive * _a)265 archive_read_support_format_mtree(struct archive *_a)
266 {
267 static const struct archive_rb_tree_ops rb_ops = {
268 mtree_cmp_node, mtree_cmp_key,
269 };
270 struct archive_read *a = (struct archive_read *)_a;
271 struct mtree *mtree;
272 int r;
273
274 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
275 ARCHIVE_STATE_NEW, "archive_read_support_format_mtree");
276
277 mtree = (struct mtree *)calloc(1, sizeof(*mtree));
278 if (mtree == NULL) {
279 archive_set_error(&a->archive, ENOMEM,
280 "Can't allocate mtree data");
281 return (ARCHIVE_FATAL);
282 }
283 mtree->checkfs = 0;
284 mtree->fd = -1;
285
286 __archive_rb_tree_init(&mtree->rbtree, &rb_ops);
287
288 r = __archive_read_register_format(a, mtree, "mtree",
289 mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL);
290
291 if (r != ARCHIVE_OK)
292 free(mtree);
293 return (ARCHIVE_OK);
294 }
295
296 static int
cleanup(struct archive_read * a)297 cleanup(struct archive_read *a)
298 {
299 struct mtree *mtree;
300 struct mtree_entry *p, *q;
301
302 mtree = (struct mtree *)(a->format->data);
303
304 p = mtree->entries;
305 while (p != NULL) {
306 q = p->next;
307 free(p->name);
308 free_options(p->options);
309 free(p);
310 p = q;
311 }
312 archive_string_free(&mtree->line);
313 archive_string_free(&mtree->current_dir);
314 archive_string_free(&mtree->contents_name);
315 archive_entry_linkresolver_free(mtree->resolver);
316
317 free(mtree->buff);
318 free(mtree);
319 (a->format->data) = NULL;
320 return (ARCHIVE_OK);
321 }
322
323 static ssize_t
get_line_size(const char * b,ssize_t avail,ssize_t * nlsize)324 get_line_size(const char *b, ssize_t avail, ssize_t *nlsize)
325 {
326 ssize_t len;
327
328 len = 0;
329 while (len < avail) {
330 switch (*b) {
331 case '\0':/* Non-ascii character or control character. */
332 if (nlsize != NULL)
333 *nlsize = 0;
334 return (-1);
335 case '\r':
336 if (avail-len > 1 && b[1] == '\n') {
337 if (nlsize != NULL)
338 *nlsize = 2;
339 return (len+2);
340 }
341 /* FALL THROUGH */
342 case '\n':
343 if (nlsize != NULL)
344 *nlsize = 1;
345 return (len+1);
346 default:
347 b++;
348 len++;
349 break;
350 }
351 }
352 if (nlsize != NULL)
353 *nlsize = 0;
354 return (avail);
355 }
356
357 /*
358 * <---------------- ravail --------------------->
359 * <-- diff ------> <--- avail ----------------->
360 * <---- len ----------->
361 * | Previous lines | line being parsed nl extra |
362 * ^
363 * b
364 *
365 */
366 static ssize_t
next_line(struct archive_read * a,const char ** b,ssize_t * avail,ssize_t * ravail,ssize_t * nl)367 next_line(struct archive_read *a,
368 const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl)
369 {
370 ssize_t len;
371 int quit;
372
373 quit = 0;
374 if (*avail == 0) {
375 *nl = 0;
376 len = 0;
377 } else
378 len = get_line_size(*b, *avail, nl);
379 /*
380 * Read bytes more while it does not reach the end of line.
381 */
382 while (*nl == 0 && len == *avail && !quit) {
383 ssize_t diff = *ravail - *avail;
384 size_t nbytes_req = (*ravail+1023) & ~1023U;
385 ssize_t tested;
386
387 /*
388 * Place an arbitrary limit on the line length.
389 * mtree is almost free-form input and without line length limits,
390 * it can consume a lot of memory.
391 */
392 if (len >= MAX_LINE_LEN)
393 return (-1);
394
395 /* Increase reading bytes if it is not enough to at least
396 * new two lines. */
397 if (nbytes_req < (size_t)*ravail + 160)
398 nbytes_req <<= 1;
399
400 *b = __archive_read_ahead(a, nbytes_req, avail);
401 if (*b == NULL) {
402 if (*ravail >= *avail)
403 return (0);
404 /* Reading bytes reaches the end of file. */
405 *b = __archive_read_ahead(a, *avail, avail);
406 quit = 1;
407 }
408 *ravail = *avail;
409 *b += diff;
410 *avail -= diff;
411 tested = len;/* Skip some bytes we already determined. */
412 len = get_line_size(*b + len, *avail - len, nl);
413 if (len >= 0)
414 len += tested;
415 }
416 return (len);
417 }
418
419 /*
420 * Compare characters with a mtree keyword.
421 * Returns the length of a mtree keyword if matched.
422 * Returns 0 if not matched.
423 */
424 static int
bid_keycmp(const char * p,const char * key,ssize_t len)425 bid_keycmp(const char *p, const char *key, ssize_t len)
426 {
427 int match_len = 0;
428
429 while (len > 0 && *p && *key) {
430 if (*p == *key) {
431 --len;
432 ++p;
433 ++key;
434 ++match_len;
435 continue;
436 }
437 return (0);/* Not match */
438 }
439 if (*key != '\0')
440 return (0);/* Not match */
441
442 /* A following character should be specified characters */
443 if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' ||
444 p[0] == '\n' || p[0] == '\r' ||
445 (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')))
446 return (match_len);
447 return (0);/* Not match */
448 }
449
450 /*
451 * Test whether the characters 'p' has is mtree keyword.
452 * Returns the length of a detected keyword.
453 * Returns 0 if any keywords were not found.
454 */
455 static int
bid_keyword(const char * p,ssize_t len)456 bid_keyword(const char *p, ssize_t len)
457 {
458 static const char * const keys_c[] = {
459 "content", "contents", "cksum", NULL
460 };
461 static const char * const keys_df[] = {
462 "device", "flags", NULL
463 };
464 static const char * const keys_g[] = {
465 "gid", "gname", NULL
466 };
467 static const char * const keys_il[] = {
468 "ignore", "inode", "link", NULL
469 };
470 static const char * const keys_m[] = {
471 "md5", "md5digest", "mode", NULL
472 };
473 static const char * const keys_no[] = {
474 "nlink", "nochange", "optional", NULL
475 };
476 static const char * const keys_r[] = {
477 "resdevice", "rmd160", "rmd160digest", NULL
478 };
479 static const char * const keys_s[] = {
480 "sha1", "sha1digest",
481 "sha256", "sha256digest",
482 "sha384", "sha384digest",
483 "sha512", "sha512digest",
484 "size", NULL
485 };
486 static const char * const keys_t[] = {
487 "tags", "time", "type", NULL
488 };
489 static const char * const keys_u[] = {
490 "uid", "uname", NULL
491 };
492 const char * const *keys;
493 int i;
494
495 switch (*p) {
496 case 'c': keys = keys_c; break;
497 case 'd': case 'f': keys = keys_df; break;
498 case 'g': keys = keys_g; break;
499 case 'i': case 'l': keys = keys_il; break;
500 case 'm': keys = keys_m; break;
501 case 'n': case 'o': keys = keys_no; break;
502 case 'r': keys = keys_r; break;
503 case 's': keys = keys_s; break;
504 case 't': keys = keys_t; break;
505 case 'u': keys = keys_u; break;
506 default: return (0);/* Unknown key */
507 }
508
509 for (i = 0; keys[i] != NULL; i++) {
510 int l = bid_keycmp(p, keys[i], len);
511 if (l > 0)
512 return (l);
513 }
514 return (0);/* Unknown key */
515 }
516
517 /*
518 * Test whether there is a set of mtree keywords.
519 * Returns the number of keyword.
520 * Returns -1 if we got incorrect sequence.
521 * This function expects a set of "<space characters>keyword=value".
522 * When "unset" is specified, expects a set of "<space characters>keyword".
523 */
524 static int
bid_keyword_list(const char * p,ssize_t len,int unset,int last_is_path)525 bid_keyword_list(const char *p, ssize_t len, int unset, int last_is_path)
526 {
527 int l;
528 int keycnt = 0;
529
530 while (len > 0 && *p) {
531 int blank = 0;
532
533 /* Test whether there are blank characters in the line. */
534 while (len >0 && (*p == ' ' || *p == '\t')) {
535 ++p;
536 --len;
537 blank = 1;
538 }
539 if (*p == '\n' || *p == '\r')
540 break;
541 if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))
542 break;
543 if (!blank && !last_is_path) /* No blank character. */
544 return (-1);
545 if (last_is_path && len == 0)
546 return (keycnt);
547
548 if (unset) {
549 l = bid_keycmp(p, "all", len);
550 if (l > 0)
551 return (1);
552 }
553 /* Test whether there is a correct key in the line. */
554 l = bid_keyword(p, len);
555 if (l == 0)
556 return (-1);/* Unknown keyword was found. */
557 p += l;
558 len -= l;
559 keycnt++;
560
561 /* Skip value */
562 if (*p == '=') {
563 int value = 0;
564 ++p;
565 --len;
566 while (len > 0 && *p != ' ' && *p != '\t') {
567 ++p;
568 --len;
569 value = 1;
570 }
571 /* A keyword should have a its value unless
572 * "/unset" operation. */
573 if (!unset && value == 0)
574 return (-1);
575 }
576 }
577 return (keycnt);
578 }
579
580 static int
bid_entry(const char * p,ssize_t len,ssize_t nl,int * last_is_path)581 bid_entry(const char *p, ssize_t len, ssize_t nl, int *last_is_path)
582 {
583 int f = 0;
584 static const unsigned char safe_char[256] = {
585 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */
586 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */
587 /* !"$%&'()*+,-./ EXCLUSION:( )(#) */
588 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */
589 /* 0123456789:;<>? EXCLUSION:(=) */
590 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */
591 /* @ABCDEFGHIJKLMNO */
592 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */
593 /* PQRSTUVWXYZ[\]^_ */
594 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */
595 /* `abcdefghijklmno */
596 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */
597 /* pqrstuvwxyz{|}~ */
598 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */
599 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */
600 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */
601 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */
602 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */
603 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */
604 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */
605 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */
606 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */
607 };
608 ssize_t ll;
609 const char *pp = p;
610 const char * const pp_end = pp + len;
611
612 *last_is_path = 0;
613 /*
614 * Skip the path-name which is quoted.
615 */
616 for (;pp < pp_end; ++pp) {
617 if (!safe_char[*(const unsigned char *)pp]) {
618 if (*pp != ' ' && *pp != '\t' && *pp != '\r'
619 && *pp != '\n')
620 f = 0;
621 break;
622 }
623 f = 1;
624 }
625 ll = pp_end - pp;
626
627 /* If a path-name was not found at the first, try to check
628 * a mtree format(a.k.a form D) ``NetBSD's mtree -D'' creates,
629 * which places the path-name at the last. */
630 if (f == 0) {
631 const char *pb = p + len - nl;
632 int name_len = 0;
633 int slash;
634
635 /* The form D accepts only a single line for an entry. */
636 if (pb-2 >= p &&
637 pb[-1] == '\\' && (pb[-2] == ' ' || pb[-2] == '\t'))
638 return (-1);
639 if (pb-1 >= p && pb[-1] == '\\')
640 return (-1);
641
642 slash = 0;
643 while (p <= --pb && *pb != ' ' && *pb != '\t') {
644 if (!safe_char[*(const unsigned char *)pb])
645 return (-1);
646 name_len++;
647 /* The pathname should have a slash in this
648 * format. */
649 if (*pb == '/')
650 slash = 1;
651 }
652 if (name_len == 0 || slash == 0)
653 return (-1);
654 /* If '/' is placed at the first in this field, this is not
655 * a valid filename. */
656 if (pb[1] == '/')
657 return (-1);
658 ll = len - nl - name_len;
659 pp = p;
660 *last_is_path = 1;
661 }
662
663 return (bid_keyword_list(pp, ll, 0, *last_is_path));
664 }
665
666 #define MAX_BID_ENTRY 3
667
668 static int
mtree_bid(struct archive_read * a,int best_bid)669 mtree_bid(struct archive_read *a, int best_bid)
670 {
671 const char *signature = "#mtree";
672 const char *p;
673
674 (void)best_bid; /* UNUSED */
675
676 /* Now let's look at the actual header and see if it matches. */
677 p = __archive_read_ahead(a, strlen(signature), NULL);
678 if (p == NULL)
679 return (-1);
680
681 if (memcmp(p, signature, strlen(signature)) == 0)
682 return (8 * (int)strlen(signature));
683
684 /*
685 * There is not a mtree signature. Let's try to detect mtree format.
686 */
687 return (detect_form(a, NULL));
688 }
689
690 static int
detect_form(struct archive_read * a,int * is_form_d)691 detect_form(struct archive_read *a, int *is_form_d)
692 {
693 const char *p;
694 ssize_t avail, ravail;
695 ssize_t len, nl;
696 int entry_cnt = 0, multiline = 0;
697 int form_D = 0;/* The archive is generated by `NetBSD mtree -D'
698 * (In this source we call it `form D') . */
699
700 if (is_form_d != NULL)
701 *is_form_d = 0;
702 p = __archive_read_ahead(a, 1, &avail);
703 if (p == NULL)
704 return (-1);
705 ravail = avail;
706 for (;;) {
707 len = next_line(a, &p, &avail, &ravail, &nl);
708 /* The terminal character of the line should be
709 * a new line character, '\r\n' or '\n'. */
710 if (len <= 0 || nl == 0)
711 break;
712 if (!multiline) {
713 /* Leading whitespace is never significant,
714 * ignore it. */
715 while (len > 0 && (*p == ' ' || *p == '\t')) {
716 ++p;
717 --avail;
718 --len;
719 }
720 /* Skip comment or empty line. */
721 if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') {
722 p += len;
723 avail -= len;
724 continue;
725 }
726 } else {
727 /* A continuance line; the terminal
728 * character of previous line was '\' character. */
729 if (bid_keyword_list(p, len, 0, 0) <= 0)
730 break;
731 if (p[len-nl-1] != '\\') {
732 if (multiline == 1 &&
733 ++entry_cnt >= MAX_BID_ENTRY)
734 break;
735 multiline = 0;
736 }
737 p += len;
738 avail -= len;
739 continue;
740 }
741 if (p[0] != '/') {
742 int last_is_path, keywords;
743
744 keywords = bid_entry(p, len, nl, &last_is_path);
745 if (keywords >= 0) {
746 if (form_D == 0) {
747 if (last_is_path)
748 form_D = 1;
749 else if (keywords > 0)
750 /* This line is not `form D'. */
751 form_D = -1;
752 } else if (form_D == 1) {
753 if (!last_is_path && keywords > 0)
754 /* This this is not `form D'
755 * and We cannot accept mixed
756 * format. */
757 break;
758 }
759 if (!last_is_path && p[len-nl-1] == '\\')
760 /* This line continues. */
761 multiline = 1;
762 else {
763 /* We've got plenty of correct lines
764 * to assume that this file is a mtree
765 * format. */
766 if (++entry_cnt >= MAX_BID_ENTRY)
767 break;
768 }
769 } else
770 break;
771 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
772 if (bid_keyword_list(p+4, len-4, 0, 0) <= 0)
773 break;
774 /* This line continues. */
775 if (p[len-nl-1] == '\\')
776 multiline = 2;
777 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
778 if (bid_keyword_list(p+6, len-6, 1, 0) <= 0)
779 break;
780 /* This line continues. */
781 if (p[len-nl-1] == '\\')
782 multiline = 2;
783 } else
784 break;
785
786 /* Test next line. */
787 p += len;
788 avail -= len;
789 }
790 if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) {
791 if (is_form_d != NULL) {
792 if (form_D == 1)
793 *is_form_d = 1;
794 }
795 return (32);
796 }
797
798 return (0);
799 }
800
801 /*
802 * The extended mtree format permits multiple lines specifying
803 * attributes for each file. For those entries, only the last line
804 * is actually used. Practically speaking, that means we have
805 * to read the entire mtree file into memory up front.
806 *
807 * The parsing is done in two steps. First, it is decided if a line
808 * changes the global defaults and if it is, processed accordingly.
809 * Otherwise, the options of the line are merged with the current
810 * global options.
811 */
812 static int
add_option(struct archive_read * a,struct mtree_option ** global,const char * value,size_t len)813 add_option(struct archive_read *a, struct mtree_option **global,
814 const char *value, size_t len)
815 {
816 struct mtree_option *opt;
817
818 if ((opt = malloc(sizeof(*opt))) == NULL) {
819 archive_set_error(&a->archive, errno, "Can't allocate memory");
820 return (ARCHIVE_FATAL);
821 }
822 if ((opt->value = malloc(len + 1)) == NULL) {
823 free(opt);
824 archive_set_error(&a->archive, errno, "Can't allocate memory");
825 return (ARCHIVE_FATAL);
826 }
827 memcpy(opt->value, value, len);
828 opt->value[len] = '\0';
829 opt->next = *global;
830 *global = opt;
831 return (ARCHIVE_OK);
832 }
833
834 static void
remove_option(struct mtree_option ** global,const char * value,size_t len)835 remove_option(struct mtree_option **global, const char *value, size_t len)
836 {
837 struct mtree_option *iter, *last;
838
839 last = NULL;
840 for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
841 if (strncmp(iter->value, value, len) == 0 &&
842 (iter->value[len] == '\0' ||
843 iter->value[len] == '='))
844 break;
845 }
846 if (iter == NULL)
847 return;
848 if (last == NULL)
849 *global = iter->next;
850 else
851 last->next = iter->next;
852
853 free(iter->value);
854 free(iter);
855 }
856
857 static int
process_global_set(struct archive_read * a,struct mtree_option ** global,const char * line)858 process_global_set(struct archive_read *a,
859 struct mtree_option **global, const char *line)
860 {
861 const char *next, *eq;
862 size_t len;
863 int r;
864
865 line += 4;
866 for (;;) {
867 next = line + strspn(line, " \t\r\n");
868 if (*next == '\0')
869 return (ARCHIVE_OK);
870 line = next;
871 next = line + strcspn(line, " \t\r\n");
872 eq = strchr(line, '=');
873 if (eq > next)
874 len = next - line;
875 else
876 len = eq - line;
877
878 remove_option(global, line, len);
879 r = add_option(a, global, line, next - line);
880 if (r != ARCHIVE_OK)
881 return (r);
882 line = next;
883 }
884 }
885
886 static int
process_global_unset(struct archive_read * a,struct mtree_option ** global,const char * line)887 process_global_unset(struct archive_read *a,
888 struct mtree_option **global, const char *line)
889 {
890 const char *next;
891 size_t len;
892
893 line += 6;
894 if (strchr(line, '=') != NULL) {
895 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
896 "/unset shall not contain `='");
897 return ARCHIVE_FATAL;
898 }
899
900 for (;;) {
901 next = line + strspn(line, " \t\r\n");
902 if (*next == '\0')
903 return (ARCHIVE_OK);
904 line = next;
905 len = strcspn(line, " \t\r\n");
906
907 if (len == 3 && strncmp(line, "all", 3) == 0) {
908 free_options(*global);
909 *global = NULL;
910 } else {
911 remove_option(global, line, len);
912 }
913
914 line += len;
915 }
916 }
917
918 static int
process_add_entry(struct archive_read * a,struct mtree * mtree,struct mtree_option ** global,const char * line,ssize_t line_len,struct mtree_entry ** last_entry,int is_form_d)919 process_add_entry(struct archive_read *a, struct mtree *mtree,
920 struct mtree_option **global, const char *line, ssize_t line_len,
921 struct mtree_entry **last_entry, int is_form_d)
922 {
923 struct mtree_entry *entry;
924 struct mtree_option *iter;
925 const char *next, *eq, *name, *end;
926 size_t name_len, len;
927 int r, i;
928
929 if ((entry = malloc(sizeof(*entry))) == NULL) {
930 archive_set_error(&a->archive, errno, "Can't allocate memory");
931 return (ARCHIVE_FATAL);
932 }
933 entry->next = NULL;
934 entry->options = NULL;
935 entry->name = NULL;
936 entry->used = 0;
937 entry->full = 0;
938
939 /* Add this entry to list. */
940 if (*last_entry == NULL)
941 mtree->entries = entry;
942 else
943 (*last_entry)->next = entry;
944 *last_entry = entry;
945
946 if (is_form_d) {
947 /* Filename is last item on line. */
948 /* Adjust line_len to trim trailing whitespace */
949 while (line_len > 0) {
950 char last_character = line[line_len - 1];
951 if (last_character == '\r'
952 || last_character == '\n'
953 || last_character == '\t'
954 || last_character == ' ') {
955 line_len--;
956 } else {
957 break;
958 }
959 }
960 /* Name starts after the last whitespace separator */
961 name = line;
962 for (i = 0; i < line_len; i++) {
963 if (line[i] == '\r'
964 || line[i] == '\n'
965 || line[i] == '\t'
966 || line[i] == ' ') {
967 name = line + i + 1;
968 }
969 }
970 name_len = line + line_len - name;
971 end = name;
972 } else {
973 /* Filename is first item on line */
974 name_len = strcspn(line, " \t\r\n");
975 name = line;
976 line += name_len;
977 end = line + line_len;
978 }
979 /* name/name_len is the name within the line. */
980 /* line..end brackets the entire line except the name */
981
982 if ((entry->name = malloc(name_len + 1)) == NULL) {
983 archive_set_error(&a->archive, errno, "Can't allocate memory");
984 return (ARCHIVE_FATAL);
985 }
986
987 memcpy(entry->name, name, name_len);
988 entry->name[name_len] = '\0';
989 parse_escapes(entry->name, entry);
990
991 entry->next_dup = NULL;
992 if (entry->full) {
993 if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) {
994 struct mtree_entry *alt;
995 alt = (struct mtree_entry *)__archive_rb_tree_find_node(
996 &mtree->rbtree, entry->name);
997 while (alt->next_dup)
998 alt = alt->next_dup;
999 alt->next_dup = entry;
1000 }
1001 }
1002
1003 for (iter = *global; iter != NULL; iter = iter->next) {
1004 r = add_option(a, &entry->options, iter->value,
1005 strlen(iter->value));
1006 if (r != ARCHIVE_OK)
1007 return (r);
1008 }
1009
1010 for (;;) {
1011 next = line + strspn(line, " \t\r\n");
1012 if (*next == '\0')
1013 return (ARCHIVE_OK);
1014 if (next >= end)
1015 return (ARCHIVE_OK);
1016 line = next;
1017 next = line + strcspn(line, " \t\r\n");
1018 eq = strchr(line, '=');
1019 if (eq == NULL || eq > next)
1020 len = next - line;
1021 else
1022 len = eq - line;
1023
1024 remove_option(&entry->options, line, len);
1025 r = add_option(a, &entry->options, line, next - line);
1026 if (r != ARCHIVE_OK)
1027 return (r);
1028 line = next;
1029 }
1030 }
1031
1032 static int
read_mtree(struct archive_read * a,struct mtree * mtree)1033 read_mtree(struct archive_read *a, struct mtree *mtree)
1034 {
1035 ssize_t len;
1036 uintmax_t counter;
1037 char *p, *s;
1038 struct mtree_option *global;
1039 struct mtree_entry *last_entry;
1040 int r, is_form_d;
1041
1042 mtree->archive_format = ARCHIVE_FORMAT_MTREE;
1043 mtree->archive_format_name = "mtree";
1044
1045 global = NULL;
1046 last_entry = NULL;
1047
1048 (void)detect_form(a, &is_form_d);
1049
1050 for (counter = 1; ; ++counter) {
1051 r = ARCHIVE_OK;
1052 len = readline(a, mtree, &p, 65536);
1053 if (len == 0) {
1054 mtree->this_entry = mtree->entries;
1055 free_options(global);
1056 return (ARCHIVE_OK);
1057 }
1058 if (len < 0) {
1059 free_options(global);
1060 return ((int)len);
1061 }
1062 /* Leading whitespace is never significant, ignore it. */
1063 while (*p == ' ' || *p == '\t') {
1064 ++p;
1065 --len;
1066 }
1067 /* Skip content lines and blank lines. */
1068 if (*p == '#')
1069 continue;
1070 if (*p == '\r' || *p == '\n' || *p == '\0')
1071 continue;
1072 /* Non-printable characters are not allowed */
1073 for (s = p;s < p + len - 1; s++) {
1074 if (!isprint((unsigned char)*s)) {
1075 r = ARCHIVE_FATAL;
1076 break;
1077 }
1078 }
1079 if (r != ARCHIVE_OK)
1080 break;
1081 if (*p != '/') {
1082 r = process_add_entry(a, mtree, &global, p, len,
1083 &last_entry, is_form_d);
1084 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
1085 if (p[4] != ' ' && p[4] != '\t')
1086 break;
1087 r = process_global_set(a, &global, p);
1088 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
1089 if (p[6] != ' ' && p[6] != '\t')
1090 break;
1091 r = process_global_unset(a, &global, p);
1092 } else
1093 break;
1094
1095 if (r != ARCHIVE_OK) {
1096 free_options(global);
1097 return r;
1098 }
1099 }
1100
1101 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1102 "Can't parse line %ju", counter);
1103 free_options(global);
1104 return (ARCHIVE_FATAL);
1105 }
1106
1107 /*
1108 * Read in the entire mtree file into memory on the first request.
1109 * Then use the next unused file to satisfy each header request.
1110 */
1111 static int
read_header(struct archive_read * a,struct archive_entry * entry)1112 read_header(struct archive_read *a, struct archive_entry *entry)
1113 {
1114 struct mtree *mtree;
1115 char *p;
1116 int r, use_next;
1117
1118 mtree = (struct mtree *)(a->format->data);
1119
1120 if (mtree->fd >= 0) {
1121 close(mtree->fd);
1122 mtree->fd = -1;
1123 }
1124
1125 if (mtree->entries == NULL) {
1126 mtree->resolver = archive_entry_linkresolver_new();
1127 if (mtree->resolver == NULL)
1128 return ARCHIVE_FATAL;
1129 archive_entry_linkresolver_set_strategy(mtree->resolver,
1130 ARCHIVE_FORMAT_MTREE);
1131 r = read_mtree(a, mtree);
1132 if (r != ARCHIVE_OK)
1133 return (r);
1134 }
1135
1136 a->archive.archive_format = mtree->archive_format;
1137 a->archive.archive_format_name = mtree->archive_format_name;
1138
1139 for (;;) {
1140 if (mtree->this_entry == NULL)
1141 return (ARCHIVE_EOF);
1142 if (strcmp(mtree->this_entry->name, "..") == 0) {
1143 mtree->this_entry->used = 1;
1144 if (archive_strlen(&mtree->current_dir) > 0) {
1145 /* Roll back current path. */
1146 p = mtree->current_dir.s
1147 + mtree->current_dir.length - 1;
1148 while (p >= mtree->current_dir.s && *p != '/')
1149 --p;
1150 if (p >= mtree->current_dir.s)
1151 --p;
1152 mtree->current_dir.length
1153 = p - mtree->current_dir.s + 1;
1154 }
1155 }
1156 if (!mtree->this_entry->used) {
1157 use_next = 0;
1158 r = parse_file(a, entry, mtree, mtree->this_entry,
1159 &use_next);
1160 if (use_next == 0)
1161 return (r);
1162 }
1163 mtree->this_entry = mtree->this_entry->next;
1164 }
1165 }
1166
1167 /*
1168 * A single file can have multiple lines contribute specifications.
1169 * Parse as many lines as necessary, then pull additional information
1170 * from a backing file on disk as necessary.
1171 */
1172 static int
parse_file(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mentry,int * use_next)1173 parse_file(struct archive_read *a, struct archive_entry *entry,
1174 struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
1175 {
1176 const char *path;
1177 struct stat st_storage, *st;
1178 struct mtree_entry *mp;
1179 struct archive_entry *sparse_entry;
1180 int r = ARCHIVE_OK, r1, parsed_kws;
1181
1182 mentry->used = 1;
1183
1184 /* Initialize reasonable defaults. */
1185 archive_entry_set_filetype(entry, AE_IFREG);
1186 archive_entry_set_size(entry, 0);
1187 archive_string_empty(&mtree->contents_name);
1188
1189 /* Parse options from this line. */
1190 parsed_kws = 0;
1191 r = parse_line(a, entry, mtree, mentry, &parsed_kws);
1192
1193 if (mentry->full) {
1194 archive_entry_copy_pathname(entry, mentry->name);
1195 /*
1196 * "Full" entries are allowed to have multiple lines
1197 * and those lines aren't required to be adjacent. We
1198 * don't support multiple lines for "relative" entries
1199 * nor do we make any attempt to merge data from
1200 * separate "relative" and "full" entries. (Merging
1201 * "relative" and "full" entries would require dealing
1202 * with pathname canonicalization, which is a very
1203 * tricky subject.)
1204 */
1205 mp = (struct mtree_entry *)__archive_rb_tree_find_node(
1206 &mtree->rbtree, mentry->name);
1207 for (; mp; mp = mp->next_dup) {
1208 if (mp->full && !mp->used) {
1209 /* Later lines override earlier ones. */
1210 mp->used = 1;
1211 r1 = parse_line(a, entry, mtree, mp, &parsed_kws);
1212 if (r1 < r)
1213 r = r1;
1214 }
1215 }
1216 } else {
1217 /*
1218 * Relative entries require us to construct
1219 * the full path and possibly update the
1220 * current directory.
1221 */
1222 size_t n = archive_strlen(&mtree->current_dir);
1223 if (n > 0)
1224 archive_strcat(&mtree->current_dir, "/");
1225 archive_strcat(&mtree->current_dir, mentry->name);
1226 archive_entry_copy_pathname(entry, mtree->current_dir.s);
1227 if (archive_entry_filetype(entry) != AE_IFDIR)
1228 mtree->current_dir.length = n;
1229 }
1230
1231 if (mtree->checkfs) {
1232 /*
1233 * Try to open and stat the file to get the real size
1234 * and other file info. It would be nice to avoid
1235 * this here so that getting a listing of an mtree
1236 * wouldn't require opening every referenced contents
1237 * file. But then we wouldn't know the actual
1238 * contents size, so I don't see a really viable way
1239 * around this. (Also, we may want to someday pull
1240 * other unspecified info from the contents file on
1241 * disk.)
1242 */
1243 mtree->fd = -1;
1244 if (archive_strlen(&mtree->contents_name) > 0)
1245 path = mtree->contents_name.s;
1246 else
1247 path = archive_entry_pathname(entry);
1248
1249 if (archive_entry_filetype(entry) == AE_IFREG ||
1250 archive_entry_filetype(entry) == AE_IFDIR) {
1251 mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC);
1252 __archive_ensure_cloexec_flag(mtree->fd);
1253 if (mtree->fd == -1 &&
1254 (errno != ENOENT ||
1255 archive_strlen(&mtree->contents_name) > 0)) {
1256 archive_set_error(&a->archive, errno,
1257 "Can't open %s", path);
1258 r = ARCHIVE_WARN;
1259 }
1260 }
1261
1262 st = &st_storage;
1263 if (mtree->fd >= 0) {
1264 if (fstat(mtree->fd, st) == -1) {
1265 archive_set_error(&a->archive, errno,
1266 "Could not fstat %s", path);
1267 r = ARCHIVE_WARN;
1268 /* If we can't stat it, don't keep it open. */
1269 close(mtree->fd);
1270 mtree->fd = -1;
1271 st = NULL;
1272 }
1273 } else if (lstat(path, st) == -1) {
1274 st = NULL;
1275 }
1276
1277 /*
1278 * Check for a mismatch between the type in the specification
1279 * and the type of the contents object on disk.
1280 */
1281 if (st != NULL) {
1282 if (((st->st_mode & S_IFMT) == S_IFREG &&
1283 archive_entry_filetype(entry) == AE_IFREG)
1284 #ifdef S_IFLNK
1285 ||((st->st_mode & S_IFMT) == S_IFLNK &&
1286 archive_entry_filetype(entry) == AE_IFLNK)
1287 #endif
1288 #ifdef S_IFSOCK
1289 ||((st->st_mode & S_IFSOCK) == S_IFSOCK &&
1290 archive_entry_filetype(entry) == AE_IFSOCK)
1291 #endif
1292 #ifdef S_IFCHR
1293 ||((st->st_mode & S_IFMT) == S_IFCHR &&
1294 archive_entry_filetype(entry) == AE_IFCHR)
1295 #endif
1296 #ifdef S_IFBLK
1297 ||((st->st_mode & S_IFMT) == S_IFBLK &&
1298 archive_entry_filetype(entry) == AE_IFBLK)
1299 #endif
1300 ||((st->st_mode & S_IFMT) == S_IFDIR &&
1301 archive_entry_filetype(entry) == AE_IFDIR)
1302 #ifdef S_IFIFO
1303 ||((st->st_mode & S_IFMT) == S_IFIFO &&
1304 archive_entry_filetype(entry) == AE_IFIFO)
1305 #endif
1306 ) {
1307 /* Types match. */
1308 } else {
1309 /* Types don't match; bail out gracefully. */
1310 if (mtree->fd >= 0)
1311 close(mtree->fd);
1312 mtree->fd = -1;
1313 if (parsed_kws & MTREE_HAS_OPTIONAL) {
1314 /* It's not an error for an optional
1315 * entry to not match disk. */
1316 *use_next = 1;
1317 } else if (r == ARCHIVE_OK) {
1318 archive_set_error(&a->archive,
1319 ARCHIVE_ERRNO_MISC,
1320 "mtree specification has different"
1321 " type for %s",
1322 archive_entry_pathname(entry));
1323 r = ARCHIVE_WARN;
1324 }
1325 return (r);
1326 }
1327 }
1328
1329 /*
1330 * If there is a contents file on disk, pick some of the
1331 * metadata from that file. For most of these, we only
1332 * set it from the contents if it wasn't already parsed
1333 * from the specification.
1334 */
1335 if (st != NULL) {
1336 if (((parsed_kws & MTREE_HAS_DEVICE) == 0 ||
1337 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) &&
1338 (archive_entry_filetype(entry) == AE_IFCHR ||
1339 archive_entry_filetype(entry) == AE_IFBLK))
1340 archive_entry_set_rdev(entry, st->st_rdev);
1341 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME))
1342 == 0 ||
1343 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1344 archive_entry_set_gid(entry, st->st_gid);
1345 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME))
1346 == 0 ||
1347 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1348 archive_entry_set_uid(entry, st->st_uid);
1349 if ((parsed_kws & MTREE_HAS_MTIME) == 0 ||
1350 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) {
1351 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
1352 archive_entry_set_mtime(entry, st->st_mtime,
1353 st->st_mtimespec.tv_nsec);
1354 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
1355 archive_entry_set_mtime(entry, st->st_mtime,
1356 st->st_mtim.tv_nsec);
1357 #elif HAVE_STRUCT_STAT_ST_MTIME_N
1358 archive_entry_set_mtime(entry, st->st_mtime,
1359 st->st_mtime_n);
1360 #elif HAVE_STRUCT_STAT_ST_UMTIME
1361 archive_entry_set_mtime(entry, st->st_mtime,
1362 st->st_umtime*1000);
1363 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
1364 archive_entry_set_mtime(entry, st->st_mtime,
1365 st->st_mtime_usec*1000);
1366 #else
1367 archive_entry_set_mtime(entry, st->st_mtime, 0);
1368 #endif
1369 }
1370 if ((parsed_kws & MTREE_HAS_NLINK) == 0 ||
1371 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1372 archive_entry_set_nlink(entry, st->st_nlink);
1373 if ((parsed_kws & MTREE_HAS_PERM) == 0 ||
1374 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1375 archive_entry_set_perm(entry, st->st_mode);
1376 if ((parsed_kws & MTREE_HAS_SIZE) == 0 ||
1377 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1378 archive_entry_set_size(entry, st->st_size);
1379 archive_entry_set_ino(entry, st->st_ino);
1380 archive_entry_set_dev(entry, st->st_dev);
1381
1382 archive_entry_linkify(mtree->resolver, &entry,
1383 &sparse_entry);
1384 } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
1385 /*
1386 * Couldn't open the entry, stat it or the on-disk type
1387 * didn't match. If this entry is optional, just
1388 * ignore it and read the next header entry.
1389 */
1390 *use_next = 1;
1391 return ARCHIVE_OK;
1392 }
1393 }
1394
1395 mtree->cur_size = archive_entry_size(entry);
1396 mtree->offset = 0;
1397
1398 return r;
1399 }
1400
1401 /*
1402 * Each line contains a sequence of keywords.
1403 */
1404 static int
parse_line(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mp,int * parsed_kws)1405 parse_line(struct archive_read *a, struct archive_entry *entry,
1406 struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
1407 {
1408 struct mtree_option *iter;
1409 int r = ARCHIVE_OK, r1;
1410
1411 for (iter = mp->options; iter != NULL; iter = iter->next) {
1412 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
1413 if (r1 < r)
1414 r = r1;
1415 }
1416 if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) {
1417 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1418 "Missing type keyword in mtree specification");
1419 return (ARCHIVE_WARN);
1420 }
1421 return (r);
1422 }
1423
1424 /*
1425 * Device entries have one of the following forms:
1426 * - raw dev_t
1427 * - format,major,minor[,subdevice]
1428 * When parsing succeeded, `pdev' will contain the appropriate dev_t value.
1429 */
1430
1431 /* strsep() is not in C90, but strcspn() is. */
1432 /* Taken from http://unixpapa.com/incnote/string.html */
1433 static char *
la_strsep(char ** sp,const char * sep)1434 la_strsep(char **sp, const char *sep)
1435 {
1436 char *p, *s;
1437 if (sp == NULL || *sp == NULL || **sp == '\0')
1438 return(NULL);
1439 s = *sp;
1440 p = s + strcspn(s, sep);
1441 if (*p != '\0')
1442 *p++ = '\0';
1443 *sp = p;
1444 return(s);
1445 }
1446
1447 static int
parse_device(dev_t * pdev,struct archive * a,char * val)1448 parse_device(dev_t *pdev, struct archive *a, char *val)
1449 {
1450 #define MAX_PACK_ARGS 3
1451 unsigned long numbers[MAX_PACK_ARGS];
1452 char *p, *dev;
1453 int argc;
1454 pack_t *pack;
1455 dev_t result;
1456 const char *error = NULL;
1457
1458 memset(pdev, 0, sizeof(*pdev));
1459 if ((dev = strchr(val, ',')) != NULL) {
1460 /*
1461 * Device's major/minor are given in a specified format.
1462 * Decode and pack it accordingly.
1463 */
1464 *dev++ = '\0';
1465 if ((pack = pack_find(val)) == NULL) {
1466 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1467 "Unknown format `%s'", val);
1468 return ARCHIVE_WARN;
1469 }
1470 argc = 0;
1471 while ((p = la_strsep(&dev, ",")) != NULL) {
1472 if (*p == '\0') {
1473 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1474 "Missing number");
1475 return ARCHIVE_WARN;
1476 }
1477 if (argc >= MAX_PACK_ARGS) {
1478 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1479 "Too many arguments");
1480 return ARCHIVE_WARN;
1481 }
1482 numbers[argc++] = (unsigned long)mtree_atol(&p, 0);
1483 }
1484 if (argc < 2) {
1485 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1486 "Not enough arguments");
1487 return ARCHIVE_WARN;
1488 }
1489 result = (*pack)(argc, numbers, &error);
1490 if (error != NULL) {
1491 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1492 "%s", error);
1493 return ARCHIVE_WARN;
1494 }
1495 } else {
1496 /* file system raw value. */
1497 result = (dev_t)mtree_atol(&val, 0);
1498 }
1499 *pdev = result;
1500 return ARCHIVE_OK;
1501 #undef MAX_PACK_ARGS
1502 }
1503
1504 static int
parse_hex_nibble(char c)1505 parse_hex_nibble(char c)
1506 {
1507 if (c >= '0' && c <= '9')
1508 return c - '0';
1509 if (c >= 'a' && c <= 'f')
1510 return 10 + c - 'a';
1511 #if 0
1512 /* XXX: Is uppercase something we should support? */
1513 if (c >= 'A' && c <= 'F')
1514 return 10 + c - 'A';
1515 #endif
1516
1517 return -1;
1518 }
1519
1520 static int
parse_digest(struct archive_read * a,struct archive_entry * entry,const char * digest,int type)1521 parse_digest(struct archive_read *a, struct archive_entry *entry,
1522 const char *digest, int type)
1523 {
1524 unsigned char digest_buf[64];
1525 int high, low;
1526 size_t i, j, len;
1527
1528 switch (type) {
1529 case ARCHIVE_ENTRY_DIGEST_MD5:
1530 len = sizeof(entry->digest.md5);
1531 break;
1532 case ARCHIVE_ENTRY_DIGEST_RMD160:
1533 len = sizeof(entry->digest.rmd160);
1534 break;
1535 case ARCHIVE_ENTRY_DIGEST_SHA1:
1536 len = sizeof(entry->digest.sha1);
1537 break;
1538 case ARCHIVE_ENTRY_DIGEST_SHA256:
1539 len = sizeof(entry->digest.sha256);
1540 break;
1541 case ARCHIVE_ENTRY_DIGEST_SHA384:
1542 len = sizeof(entry->digest.sha384);
1543 break;
1544 case ARCHIVE_ENTRY_DIGEST_SHA512:
1545 len = sizeof(entry->digest.sha512);
1546 break;
1547 default:
1548 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1549 "Internal error: Unknown digest type");
1550 return ARCHIVE_FATAL;
1551 }
1552
1553 if (len > sizeof(digest_buf)) {
1554 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1555 "Internal error: Digest storage too large");
1556 return ARCHIVE_FATAL;
1557 }
1558
1559 len *= 2;
1560
1561 if (mtree_strnlen(digest, len+1) != len) {
1562 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1563 "incorrect digest length, ignoring");
1564 return ARCHIVE_WARN;
1565 }
1566
1567 for (i = 0, j = 0; i < len; i += 2, j++) {
1568 high = parse_hex_nibble(digest[i]);
1569 low = parse_hex_nibble(digest[i+1]);
1570 if (high == -1 || low == -1) {
1571 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1572 "invalid digest data, ignoring");
1573 return ARCHIVE_WARN;
1574 }
1575
1576 digest_buf[j] = high << 4 | low;
1577 }
1578
1579 return archive_entry_set_digest(entry, type, digest_buf);
1580 }
1581
1582 /*
1583 * Parse a single keyword and its value.
1584 */
1585 static int
parse_keyword(struct archive_read * a,struct mtree * mtree,struct archive_entry * entry,struct mtree_option * opt,int * parsed_kws)1586 parse_keyword(struct archive_read *a, struct mtree *mtree,
1587 struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws)
1588 {
1589 char *val, *key;
1590
1591 key = opt->value;
1592
1593 if (*key == '\0')
1594 return (ARCHIVE_OK);
1595
1596 if (strcmp(key, "nochange") == 0) {
1597 *parsed_kws |= MTREE_HAS_NOCHANGE;
1598 return (ARCHIVE_OK);
1599 }
1600 if (strcmp(key, "optional") == 0) {
1601 *parsed_kws |= MTREE_HAS_OPTIONAL;
1602 return (ARCHIVE_OK);
1603 }
1604 if (strcmp(key, "ignore") == 0) {
1605 /*
1606 * The mtree processing is not recursive, so
1607 * recursion will only happen for explicitly listed
1608 * entries.
1609 */
1610 return (ARCHIVE_OK);
1611 }
1612
1613 val = strchr(key, '=');
1614 if (val == NULL) {
1615 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1616 "Malformed attribute \"%s\" (%d)", key, key[0]);
1617 return (ARCHIVE_WARN);
1618 }
1619
1620 *val = '\0';
1621 ++val;
1622
1623 switch (key[0]) {
1624 case 'c':
1625 if (strcmp(key, "content") == 0
1626 || strcmp(key, "contents") == 0) {
1627 parse_escapes(val, NULL);
1628 archive_strcpy(&mtree->contents_name, val);
1629 return (ARCHIVE_OK);
1630 }
1631 if (strcmp(key, "cksum") == 0)
1632 return (ARCHIVE_OK);
1633 break;
1634 case 'd':
1635 if (strcmp(key, "device") == 0) {
1636 /* stat(2) st_rdev field, e.g. the major/minor IDs
1637 * of a char/block special file */
1638 int r;
1639 dev_t dev;
1640
1641 *parsed_kws |= MTREE_HAS_DEVICE;
1642 r = parse_device(&dev, &a->archive, val);
1643 if (r == ARCHIVE_OK)
1644 archive_entry_set_rdev(entry, dev);
1645 return r;
1646 }
1647 break;
1648 case 'f':
1649 if (strcmp(key, "flags") == 0) {
1650 *parsed_kws |= MTREE_HAS_FFLAGS;
1651 archive_entry_copy_fflags_text(entry, val);
1652 return (ARCHIVE_OK);
1653 }
1654 break;
1655 case 'g':
1656 if (strcmp(key, "gid") == 0) {
1657 *parsed_kws |= MTREE_HAS_GID;
1658 archive_entry_set_gid(entry, mtree_atol(&val, 10));
1659 return (ARCHIVE_OK);
1660 }
1661 if (strcmp(key, "gname") == 0) {
1662 *parsed_kws |= MTREE_HAS_GNAME;
1663 archive_entry_copy_gname(entry, val);
1664 return (ARCHIVE_OK);
1665 }
1666 break;
1667 case 'i':
1668 if (strcmp(key, "inode") == 0) {
1669 archive_entry_set_ino(entry, mtree_atol(&val, 10));
1670 return (ARCHIVE_OK);
1671 }
1672 break;
1673 case 'l':
1674 if (strcmp(key, "link") == 0) {
1675 parse_escapes(val, NULL);
1676 archive_entry_copy_symlink(entry, val);
1677 return (ARCHIVE_OK);
1678 }
1679 break;
1680 case 'm':
1681 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) {
1682 return parse_digest(a, entry, val,
1683 ARCHIVE_ENTRY_DIGEST_MD5);
1684 }
1685 if (strcmp(key, "mode") == 0) {
1686 if (val[0] < '0' || val[0] > '7') {
1687 archive_set_error(&a->archive,
1688 ARCHIVE_ERRNO_FILE_FORMAT,
1689 "Symbolic or non-octal mode \"%s\" unsupported", val);
1690 return (ARCHIVE_WARN);
1691 }
1692 *parsed_kws |= MTREE_HAS_PERM;
1693 archive_entry_set_perm(entry, (mode_t)mtree_atol(&val, 8));
1694 return (ARCHIVE_OK);
1695 }
1696 break;
1697 case 'n':
1698 if (strcmp(key, "nlink") == 0) {
1699 *parsed_kws |= MTREE_HAS_NLINK;
1700 archive_entry_set_nlink(entry,
1701 (unsigned int)mtree_atol(&val, 10));
1702 return (ARCHIVE_OK);
1703 }
1704 break;
1705 case 'r':
1706 if (strcmp(key, "resdevice") == 0) {
1707 /* stat(2) st_dev field, e.g. the device ID where the
1708 * inode resides */
1709 int r;
1710 dev_t dev;
1711
1712 r = parse_device(&dev, &a->archive, val);
1713 if (r == ARCHIVE_OK)
1714 archive_entry_set_dev(entry, dev);
1715 return r;
1716 }
1717 if (strcmp(key, "rmd160") == 0 ||
1718 strcmp(key, "rmd160digest") == 0) {
1719 return parse_digest(a, entry, val,
1720 ARCHIVE_ENTRY_DIGEST_RMD160);
1721 }
1722 break;
1723 case 's':
1724 if (strcmp(key, "sha1") == 0 ||
1725 strcmp(key, "sha1digest") == 0) {
1726 return parse_digest(a, entry, val,
1727 ARCHIVE_ENTRY_DIGEST_SHA1);
1728 }
1729 if (strcmp(key, "sha256") == 0 ||
1730 strcmp(key, "sha256digest") == 0) {
1731 return parse_digest(a, entry, val,
1732 ARCHIVE_ENTRY_DIGEST_SHA256);
1733 }
1734 if (strcmp(key, "sha384") == 0 ||
1735 strcmp(key, "sha384digest") == 0) {
1736 return parse_digest(a, entry, val,
1737 ARCHIVE_ENTRY_DIGEST_SHA384);
1738 }
1739 if (strcmp(key, "sha512") == 0 ||
1740 strcmp(key, "sha512digest") == 0) {
1741 return parse_digest(a, entry, val,
1742 ARCHIVE_ENTRY_DIGEST_SHA512);
1743 }
1744 if (strcmp(key, "size") == 0) {
1745 archive_entry_set_size(entry, mtree_atol(&val, 10));
1746 return (ARCHIVE_OK);
1747 }
1748 break;
1749 case 't':
1750 if (strcmp(key, "tags") == 0) {
1751 /*
1752 * Comma delimited list of tags.
1753 * Ignore the tags for now, but the interface
1754 * should be extended to allow inclusion/exclusion.
1755 */
1756 return (ARCHIVE_OK);
1757 }
1758 if (strcmp(key, "time") == 0) {
1759 int64_t m;
1760 int64_t my_time_t_max = get_time_t_max();
1761 int64_t my_time_t_min = get_time_t_min();
1762 long ns = 0;
1763
1764 *parsed_kws |= MTREE_HAS_MTIME;
1765 m = mtree_atol(&val, 10);
1766 /* Replicate an old mtree bug:
1767 * 123456789.1 represents 123456789
1768 * seconds and 1 nanosecond. */
1769 if (*val == '.') {
1770 ++val;
1771 ns = (long)mtree_atol(&val, 10);
1772 if (ns < 0)
1773 ns = 0;
1774 else if (ns > 999999999)
1775 ns = 999999999;
1776 }
1777 if (m > my_time_t_max)
1778 m = my_time_t_max;
1779 else if (m < my_time_t_min)
1780 m = my_time_t_min;
1781 archive_entry_set_mtime(entry, (time_t)m, ns);
1782 return (ARCHIVE_OK);
1783 }
1784 if (strcmp(key, "type") == 0) {
1785 switch (val[0]) {
1786 case 'b':
1787 if (strcmp(val, "block") == 0) {
1788 *parsed_kws |= MTREE_HAS_TYPE;
1789 archive_entry_set_filetype(entry,
1790 AE_IFBLK);
1791 return (ARCHIVE_OK);
1792 }
1793 break;
1794 case 'c':
1795 if (strcmp(val, "char") == 0) {
1796 *parsed_kws |= MTREE_HAS_TYPE;
1797 archive_entry_set_filetype(entry,
1798 AE_IFCHR);
1799 return (ARCHIVE_OK);
1800 }
1801 break;
1802 case 'd':
1803 if (strcmp(val, "dir") == 0) {
1804 *parsed_kws |= MTREE_HAS_TYPE;
1805 archive_entry_set_filetype(entry,
1806 AE_IFDIR);
1807 return (ARCHIVE_OK);
1808 }
1809 break;
1810 case 'f':
1811 if (strcmp(val, "fifo") == 0) {
1812 *parsed_kws |= MTREE_HAS_TYPE;
1813 archive_entry_set_filetype(entry,
1814 AE_IFIFO);
1815 return (ARCHIVE_OK);
1816 }
1817 if (strcmp(val, "file") == 0) {
1818 *parsed_kws |= MTREE_HAS_TYPE;
1819 archive_entry_set_filetype(entry,
1820 AE_IFREG);
1821 return (ARCHIVE_OK);
1822 }
1823 break;
1824 case 'l':
1825 if (strcmp(val, "link") == 0) {
1826 *parsed_kws |= MTREE_HAS_TYPE;
1827 archive_entry_set_filetype(entry,
1828 AE_IFLNK);
1829 return (ARCHIVE_OK);
1830 }
1831 break;
1832 default:
1833 break;
1834 }
1835 archive_set_error(&a->archive,
1836 ARCHIVE_ERRNO_FILE_FORMAT,
1837 "Unrecognized file type \"%s\"; "
1838 "assuming \"file\"", val);
1839 archive_entry_set_filetype(entry, AE_IFREG);
1840 return (ARCHIVE_WARN);
1841 }
1842 break;
1843 case 'u':
1844 if (strcmp(key, "uid") == 0) {
1845 *parsed_kws |= MTREE_HAS_UID;
1846 archive_entry_set_uid(entry, mtree_atol(&val, 10));
1847 return (ARCHIVE_OK);
1848 }
1849 if (strcmp(key, "uname") == 0) {
1850 *parsed_kws |= MTREE_HAS_UNAME;
1851 archive_entry_copy_uname(entry, val);
1852 return (ARCHIVE_OK);
1853 }
1854 break;
1855 default:
1856 break;
1857 }
1858 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1859 "Unrecognized key %s=%s", key, val);
1860 return (ARCHIVE_WARN);
1861 }
1862
1863 static int
read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1864 read_data(struct archive_read *a, const void **buff, size_t *size,
1865 int64_t *offset)
1866 {
1867 size_t bytes_to_read;
1868 ssize_t bytes_read;
1869 struct mtree *mtree;
1870
1871 mtree = (struct mtree *)(a->format->data);
1872 if (mtree->fd < 0) {
1873 *buff = NULL;
1874 *offset = 0;
1875 *size = 0;
1876 return (ARCHIVE_EOF);
1877 }
1878 if (mtree->buff == NULL) {
1879 mtree->buffsize = 64 * 1024;
1880 mtree->buff = malloc(mtree->buffsize);
1881 if (mtree->buff == NULL) {
1882 archive_set_error(&a->archive, ENOMEM,
1883 "Can't allocate memory");
1884 return (ARCHIVE_FATAL);
1885 }
1886 }
1887
1888 *buff = mtree->buff;
1889 *offset = mtree->offset;
1890 if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset)
1891 bytes_to_read = (size_t)(mtree->cur_size - mtree->offset);
1892 else
1893 bytes_to_read = mtree->buffsize;
1894 bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
1895 if (bytes_read < 0) {
1896 archive_set_error(&a->archive, errno, "Can't read");
1897 return (ARCHIVE_WARN);
1898 }
1899 if (bytes_read == 0) {
1900 *size = 0;
1901 return (ARCHIVE_EOF);
1902 }
1903 mtree->offset += bytes_read;
1904 *size = bytes_read;
1905 return (ARCHIVE_OK);
1906 }
1907
1908 /* Skip does nothing except possibly close the contents file. */
1909 static int
skip(struct archive_read * a)1910 skip(struct archive_read *a)
1911 {
1912 struct mtree *mtree;
1913
1914 mtree = (struct mtree *)(a->format->data);
1915 if (mtree->fd >= 0) {
1916 close(mtree->fd);
1917 mtree->fd = -1;
1918 }
1919 return (ARCHIVE_OK);
1920 }
1921
1922 /*
1923 * Since parsing backslash sequences always makes strings shorter,
1924 * we can always do this conversion in-place.
1925 */
1926 static void
parse_escapes(char * src,struct mtree_entry * mentry)1927 parse_escapes(char *src, struct mtree_entry *mentry)
1928 {
1929 char *dest = src;
1930 char c;
1931
1932 if (mentry != NULL && strcmp(src, ".") == 0)
1933 mentry->full = 1;
1934
1935 while (*src != '\0') {
1936 c = *src++;
1937 if (c == '/' && mentry != NULL)
1938 mentry->full = 1;
1939 if (c == '\\') {
1940 switch (src[0]) {
1941 case '0':
1942 if (src[1] < '0' || src[1] > '7') {
1943 c = 0;
1944 ++src;
1945 break;
1946 }
1947 /* FALLTHROUGH */
1948 case '1':
1949 case '2':
1950 case '3':
1951 if (src[1] >= '0' && src[1] <= '7' &&
1952 src[2] >= '0' && src[2] <= '7') {
1953 c = (src[0] - '0') << 6;
1954 c |= (src[1] - '0') << 3;
1955 c |= (src[2] - '0');
1956 src += 3;
1957 }
1958 break;
1959 case 'a':
1960 c = '\a';
1961 ++src;
1962 break;
1963 case 'b':
1964 c = '\b';
1965 ++src;
1966 break;
1967 case 'f':
1968 c = '\f';
1969 ++src;
1970 break;
1971 case 'n':
1972 c = '\n';
1973 ++src;
1974 break;
1975 case 'r':
1976 c = '\r';
1977 ++src;
1978 break;
1979 case 's':
1980 c = ' ';
1981 ++src;
1982 break;
1983 case 't':
1984 c = '\t';
1985 ++src;
1986 break;
1987 case 'v':
1988 c = '\v';
1989 ++src;
1990 break;
1991 case '\\':
1992 c = '\\';
1993 ++src;
1994 break;
1995 }
1996 }
1997 *dest++ = c;
1998 }
1999 *dest = '\0';
2000 }
2001
2002 /* Parse a hex digit. */
2003 static int
parsedigit(char c)2004 parsedigit(char c)
2005 {
2006 if (c >= '0' && c <= '9')
2007 return c - '0';
2008 else if (c >= 'a' && c <= 'f')
2009 return c - 'a';
2010 else if (c >= 'A' && c <= 'F')
2011 return c - 'A';
2012 else
2013 return -1;
2014 }
2015
2016 /*
2017 * Note that this implementation does not (and should not!) obey
2018 * locale settings; you cannot simply substitute strtol here, since
2019 * it does obey locale.
2020 */
2021 static int64_t
mtree_atol(char ** p,int base)2022 mtree_atol(char **p, int base)
2023 {
2024 int64_t l, limit;
2025 int digit, last_digit_limit;
2026
2027 if (base == 0) {
2028 if (**p != '0')
2029 base = 10;
2030 else if ((*p)[1] == 'x' || (*p)[1] == 'X') {
2031 *p += 2;
2032 base = 16;
2033 } else {
2034 base = 8;
2035 }
2036 }
2037
2038 if (**p == '-') {
2039 limit = INT64_MIN / base;
2040 last_digit_limit = -(INT64_MIN % base);
2041 ++(*p);
2042
2043 l = 0;
2044 digit = parsedigit(**p);
2045 while (digit >= 0 && digit < base) {
2046 if (l < limit || (l == limit && digit >= last_digit_limit))
2047 return INT64_MIN;
2048 l = (l * base) - digit;
2049 digit = parsedigit(*++(*p));
2050 }
2051 return l;
2052 } else {
2053 limit = INT64_MAX / base;
2054 last_digit_limit = INT64_MAX % base;
2055
2056 l = 0;
2057 digit = parsedigit(**p);
2058 while (digit >= 0 && digit < base) {
2059 if (l > limit || (l == limit && digit > last_digit_limit))
2060 return INT64_MAX;
2061 l = (l * base) + digit;
2062 digit = parsedigit(*++(*p));
2063 }
2064 return l;
2065 }
2066 }
2067
2068 /*
2069 * Returns length of line (including trailing newline)
2070 * or negative on error. 'start' argument is updated to
2071 * point to first character of line.
2072 */
2073 static ssize_t
readline(struct archive_read * a,struct mtree * mtree,char ** start,ssize_t limit)2074 readline(struct archive_read *a, struct mtree *mtree, char **start,
2075 ssize_t limit)
2076 {
2077 ssize_t bytes_read;
2078 ssize_t total_size = 0;
2079 ssize_t find_off = 0;
2080 const void *t;
2081 void *nl;
2082 char *u;
2083
2084 /* Accumulate line in a line buffer. */
2085 for (;;) {
2086 /* Read some more. */
2087 t = __archive_read_ahead(a, 1, &bytes_read);
2088 if (t == NULL)
2089 return (0);
2090 if (bytes_read < 0)
2091 return (ARCHIVE_FATAL);
2092 nl = memchr(t, '\n', bytes_read);
2093 /* If we found '\n', trim the read to end exactly there. */
2094 if (nl != NULL) {
2095 bytes_read = ((const char *)nl) - ((const char *)t) + 1;
2096 }
2097 if (total_size + bytes_read + 1 > limit) {
2098 archive_set_error(&a->archive,
2099 ARCHIVE_ERRNO_FILE_FORMAT,
2100 "Line too long");
2101 return (ARCHIVE_FATAL);
2102 }
2103 if (archive_string_ensure(&mtree->line,
2104 total_size + bytes_read + 1) == NULL) {
2105 archive_set_error(&a->archive, ENOMEM,
2106 "Can't allocate working buffer");
2107 return (ARCHIVE_FATAL);
2108 }
2109 /* Append new bytes to string. */
2110 memcpy(mtree->line.s + total_size, t, bytes_read);
2111 __archive_read_consume(a, bytes_read);
2112 total_size += bytes_read;
2113 mtree->line.s[total_size] = '\0';
2114
2115 for (u = mtree->line.s + find_off; *u; ++u) {
2116 if (u[0] == '\n') {
2117 /* Ends with unescaped newline. */
2118 *start = mtree->line.s;
2119 return total_size;
2120 } else if (u[0] == '#') {
2121 /* Ends with comment sequence #...\n */
2122 if (nl == NULL) {
2123 /* But we've not found the \n yet */
2124 break;
2125 }
2126 } else if (u[0] == '\\') {
2127 if (u[1] == '\n') {
2128 /* Trim escaped newline. */
2129 total_size -= 2;
2130 mtree->line.s[total_size] = '\0';
2131 break;
2132 } else if (u[1] != '\0') {
2133 /* Skip the two-char escape sequence */
2134 ++u;
2135 }
2136 }
2137 }
2138 find_off = u - mtree->line.s;
2139 }
2140 }
2141