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 determinated. */
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 detected_bytes = 0, 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 (multiline == 1)
732 detected_bytes += len;
733 if (p[len-nl-1] != '\\') {
734 if (multiline == 1 &&
735 ++entry_cnt >= MAX_BID_ENTRY)
736 break;
737 multiline = 0;
738 }
739 p += len;
740 avail -= len;
741 continue;
742 }
743 if (p[0] != '/') {
744 int last_is_path, keywords;
745
746 keywords = bid_entry(p, len, nl, &last_is_path);
747 if (keywords >= 0) {
748 detected_bytes += len;
749 if (form_D == 0) {
750 if (last_is_path)
751 form_D = 1;
752 else if (keywords > 0)
753 /* This line is not `form D'. */
754 form_D = -1;
755 } else if (form_D == 1) {
756 if (!last_is_path && keywords > 0)
757 /* This this is not `form D'
758 * and We cannot accept mixed
759 * format. */
760 break;
761 }
762 if (!last_is_path && p[len-nl-1] == '\\')
763 /* This line continues. */
764 multiline = 1;
765 else {
766 /* We've got plenty of correct lines
767 * to assume that this file is a mtree
768 * format. */
769 if (++entry_cnt >= MAX_BID_ENTRY)
770 break;
771 }
772 } else
773 break;
774 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
775 if (bid_keyword_list(p+4, len-4, 0, 0) <= 0)
776 break;
777 /* This line continues. */
778 if (p[len-nl-1] == '\\')
779 multiline = 2;
780 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
781 if (bid_keyword_list(p+6, len-6, 1, 0) <= 0)
782 break;
783 /* This line continues. */
784 if (p[len-nl-1] == '\\')
785 multiline = 2;
786 } else
787 break;
788
789 /* Test next line. */
790 p += len;
791 avail -= len;
792 }
793 if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) {
794 if (is_form_d != NULL) {
795 if (form_D == 1)
796 *is_form_d = 1;
797 }
798 return (32);
799 }
800
801 return (0);
802 }
803
804 /*
805 * The extended mtree format permits multiple lines specifying
806 * attributes for each file. For those entries, only the last line
807 * is actually used. Practically speaking, that means we have
808 * to read the entire mtree file into memory up front.
809 *
810 * The parsing is done in two steps. First, it is decided if a line
811 * changes the global defaults and if it is, processed accordingly.
812 * Otherwise, the options of the line are merged with the current
813 * global options.
814 */
815 static int
add_option(struct archive_read * a,struct mtree_option ** global,const char * value,size_t len)816 add_option(struct archive_read *a, struct mtree_option **global,
817 const char *value, size_t len)
818 {
819 struct mtree_option *opt;
820
821 if ((opt = malloc(sizeof(*opt))) == NULL) {
822 archive_set_error(&a->archive, errno, "Can't allocate memory");
823 return (ARCHIVE_FATAL);
824 }
825 if ((opt->value = malloc(len + 1)) == NULL) {
826 free(opt);
827 archive_set_error(&a->archive, errno, "Can't allocate memory");
828 return (ARCHIVE_FATAL);
829 }
830 memcpy(opt->value, value, len);
831 opt->value[len] = '\0';
832 opt->next = *global;
833 *global = opt;
834 return (ARCHIVE_OK);
835 }
836
837 static void
remove_option(struct mtree_option ** global,const char * value,size_t len)838 remove_option(struct mtree_option **global, const char *value, size_t len)
839 {
840 struct mtree_option *iter, *last;
841
842 last = NULL;
843 for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
844 if (strncmp(iter->value, value, len) == 0 &&
845 (iter->value[len] == '\0' ||
846 iter->value[len] == '='))
847 break;
848 }
849 if (iter == NULL)
850 return;
851 if (last == NULL)
852 *global = iter->next;
853 else
854 last->next = iter->next;
855
856 free(iter->value);
857 free(iter);
858 }
859
860 static int
process_global_set(struct archive_read * a,struct mtree_option ** global,const char * line)861 process_global_set(struct archive_read *a,
862 struct mtree_option **global, const char *line)
863 {
864 const char *next, *eq;
865 size_t len;
866 int r;
867
868 line += 4;
869 for (;;) {
870 next = line + strspn(line, " \t\r\n");
871 if (*next == '\0')
872 return (ARCHIVE_OK);
873 line = next;
874 next = line + strcspn(line, " \t\r\n");
875 eq = strchr(line, '=');
876 if (eq > next)
877 len = next - line;
878 else
879 len = eq - line;
880
881 remove_option(global, line, len);
882 r = add_option(a, global, line, next - line);
883 if (r != ARCHIVE_OK)
884 return (r);
885 line = next;
886 }
887 }
888
889 static int
process_global_unset(struct archive_read * a,struct mtree_option ** global,const char * line)890 process_global_unset(struct archive_read *a,
891 struct mtree_option **global, const char *line)
892 {
893 const char *next;
894 size_t len;
895
896 line += 6;
897 if (strchr(line, '=') != NULL) {
898 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
899 "/unset shall not contain `='");
900 return ARCHIVE_FATAL;
901 }
902
903 for (;;) {
904 next = line + strspn(line, " \t\r\n");
905 if (*next == '\0')
906 return (ARCHIVE_OK);
907 line = next;
908 len = strcspn(line, " \t\r\n");
909
910 if (len == 3 && strncmp(line, "all", 3) == 0) {
911 free_options(*global);
912 *global = NULL;
913 } else {
914 remove_option(global, line, len);
915 }
916
917 line += len;
918 }
919 }
920
921 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)922 process_add_entry(struct archive_read *a, struct mtree *mtree,
923 struct mtree_option **global, const char *line, ssize_t line_len,
924 struct mtree_entry **last_entry, int is_form_d)
925 {
926 struct mtree_entry *entry;
927 struct mtree_option *iter;
928 const char *next, *eq, *name, *end;
929 size_t name_len, len;
930 int r, i;
931
932 if ((entry = malloc(sizeof(*entry))) == NULL) {
933 archive_set_error(&a->archive, errno, "Can't allocate memory");
934 return (ARCHIVE_FATAL);
935 }
936 entry->next = NULL;
937 entry->options = NULL;
938 entry->name = NULL;
939 entry->used = 0;
940 entry->full = 0;
941
942 /* Add this entry to list. */
943 if (*last_entry == NULL)
944 mtree->entries = entry;
945 else
946 (*last_entry)->next = entry;
947 *last_entry = entry;
948
949 if (is_form_d) {
950 /* Filename is last item on line. */
951 /* Adjust line_len to trim trailing whitespace */
952 while (line_len > 0) {
953 char last_character = line[line_len - 1];
954 if (last_character == '\r'
955 || last_character == '\n'
956 || last_character == '\t'
957 || last_character == ' ') {
958 line_len--;
959 } else {
960 break;
961 }
962 }
963 /* Name starts after the last whitespace separator */
964 name = line;
965 for (i = 0; i < line_len; i++) {
966 if (line[i] == '\r'
967 || line[i] == '\n'
968 || line[i] == '\t'
969 || line[i] == ' ') {
970 name = line + i + 1;
971 }
972 }
973 name_len = line + line_len - name;
974 end = name;
975 } else {
976 /* Filename is first item on line */
977 name_len = strcspn(line, " \t\r\n");
978 name = line;
979 line += name_len;
980 end = line + line_len;
981 }
982 /* name/name_len is the name within the line. */
983 /* line..end brackets the entire line except the name */
984
985 if ((entry->name = malloc(name_len + 1)) == NULL) {
986 archive_set_error(&a->archive, errno, "Can't allocate memory");
987 return (ARCHIVE_FATAL);
988 }
989
990 memcpy(entry->name, name, name_len);
991 entry->name[name_len] = '\0';
992 parse_escapes(entry->name, entry);
993
994 entry->next_dup = NULL;
995 if (entry->full) {
996 if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) {
997 struct mtree_entry *alt;
998 alt = (struct mtree_entry *)__archive_rb_tree_find_node(
999 &mtree->rbtree, entry->name);
1000 while (alt->next_dup)
1001 alt = alt->next_dup;
1002 alt->next_dup = entry;
1003 }
1004 }
1005
1006 for (iter = *global; iter != NULL; iter = iter->next) {
1007 r = add_option(a, &entry->options, iter->value,
1008 strlen(iter->value));
1009 if (r != ARCHIVE_OK)
1010 return (r);
1011 }
1012
1013 for (;;) {
1014 next = line + strspn(line, " \t\r\n");
1015 if (*next == '\0')
1016 return (ARCHIVE_OK);
1017 if (next >= end)
1018 return (ARCHIVE_OK);
1019 line = next;
1020 next = line + strcspn(line, " \t\r\n");
1021 eq = strchr(line, '=');
1022 if (eq == NULL || eq > next)
1023 len = next - line;
1024 else
1025 len = eq - line;
1026
1027 remove_option(&entry->options, line, len);
1028 r = add_option(a, &entry->options, line, next - line);
1029 if (r != ARCHIVE_OK)
1030 return (r);
1031 line = next;
1032 }
1033 }
1034
1035 static int
read_mtree(struct archive_read * a,struct mtree * mtree)1036 read_mtree(struct archive_read *a, struct mtree *mtree)
1037 {
1038 ssize_t len;
1039 uintmax_t counter;
1040 char *p, *s;
1041 struct mtree_option *global;
1042 struct mtree_entry *last_entry;
1043 int r, is_form_d;
1044
1045 mtree->archive_format = ARCHIVE_FORMAT_MTREE;
1046 mtree->archive_format_name = "mtree";
1047
1048 global = NULL;
1049 last_entry = NULL;
1050
1051 (void)detect_form(a, &is_form_d);
1052
1053 for (counter = 1; ; ++counter) {
1054 r = ARCHIVE_OK;
1055 len = readline(a, mtree, &p, 65536);
1056 if (len == 0) {
1057 mtree->this_entry = mtree->entries;
1058 free_options(global);
1059 return (ARCHIVE_OK);
1060 }
1061 if (len < 0) {
1062 free_options(global);
1063 return ((int)len);
1064 }
1065 /* Leading whitespace is never significant, ignore it. */
1066 while (*p == ' ' || *p == '\t') {
1067 ++p;
1068 --len;
1069 }
1070 /* Skip content lines and blank lines. */
1071 if (*p == '#')
1072 continue;
1073 if (*p == '\r' || *p == '\n' || *p == '\0')
1074 continue;
1075 /* Non-printable characters are not allowed */
1076 for (s = p;s < p + len - 1; s++) {
1077 if (!isprint(*s)) {
1078 r = ARCHIVE_FATAL;
1079 break;
1080 }
1081 }
1082 if (r != ARCHIVE_OK)
1083 break;
1084 if (*p != '/') {
1085 r = process_add_entry(a, mtree, &global, p, len,
1086 &last_entry, is_form_d);
1087 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
1088 if (p[4] != ' ' && p[4] != '\t')
1089 break;
1090 r = process_global_set(a, &global, p);
1091 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
1092 if (p[6] != ' ' && p[6] != '\t')
1093 break;
1094 r = process_global_unset(a, &global, p);
1095 } else
1096 break;
1097
1098 if (r != ARCHIVE_OK) {
1099 free_options(global);
1100 return r;
1101 }
1102 }
1103
1104 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1105 "Can't parse line %ju", counter);
1106 free_options(global);
1107 return (ARCHIVE_FATAL);
1108 }
1109
1110 /*
1111 * Read in the entire mtree file into memory on the first request.
1112 * Then use the next unused file to satisfy each header request.
1113 */
1114 static int
read_header(struct archive_read * a,struct archive_entry * entry)1115 read_header(struct archive_read *a, struct archive_entry *entry)
1116 {
1117 struct mtree *mtree;
1118 char *p;
1119 int r, use_next;
1120
1121 mtree = (struct mtree *)(a->format->data);
1122
1123 if (mtree->fd >= 0) {
1124 close(mtree->fd);
1125 mtree->fd = -1;
1126 }
1127
1128 if (mtree->entries == NULL) {
1129 mtree->resolver = archive_entry_linkresolver_new();
1130 if (mtree->resolver == NULL)
1131 return ARCHIVE_FATAL;
1132 archive_entry_linkresolver_set_strategy(mtree->resolver,
1133 ARCHIVE_FORMAT_MTREE);
1134 r = read_mtree(a, mtree);
1135 if (r != ARCHIVE_OK)
1136 return (r);
1137 }
1138
1139 a->archive.archive_format = mtree->archive_format;
1140 a->archive.archive_format_name = mtree->archive_format_name;
1141
1142 for (;;) {
1143 if (mtree->this_entry == NULL)
1144 return (ARCHIVE_EOF);
1145 if (strcmp(mtree->this_entry->name, "..") == 0) {
1146 mtree->this_entry->used = 1;
1147 if (archive_strlen(&mtree->current_dir) > 0) {
1148 /* Roll back current path. */
1149 p = mtree->current_dir.s
1150 + mtree->current_dir.length - 1;
1151 while (p >= mtree->current_dir.s && *p != '/')
1152 --p;
1153 if (p >= mtree->current_dir.s)
1154 --p;
1155 mtree->current_dir.length
1156 = p - mtree->current_dir.s + 1;
1157 }
1158 }
1159 if (!mtree->this_entry->used) {
1160 use_next = 0;
1161 r = parse_file(a, entry, mtree, mtree->this_entry,
1162 &use_next);
1163 if (use_next == 0)
1164 return (r);
1165 }
1166 mtree->this_entry = mtree->this_entry->next;
1167 }
1168 }
1169
1170 /*
1171 * A single file can have multiple lines contribute specifications.
1172 * Parse as many lines as necessary, then pull additional information
1173 * from a backing file on disk as necessary.
1174 */
1175 static int
parse_file(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mentry,int * use_next)1176 parse_file(struct archive_read *a, struct archive_entry *entry,
1177 struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
1178 {
1179 const char *path;
1180 struct stat st_storage, *st;
1181 struct mtree_entry *mp;
1182 struct archive_entry *sparse_entry;
1183 int r = ARCHIVE_OK, r1, parsed_kws;
1184
1185 mentry->used = 1;
1186
1187 /* Initialize reasonable defaults. */
1188 archive_entry_set_filetype(entry, AE_IFREG);
1189 archive_entry_set_size(entry, 0);
1190 archive_string_empty(&mtree->contents_name);
1191
1192 /* Parse options from this line. */
1193 parsed_kws = 0;
1194 r = parse_line(a, entry, mtree, mentry, &parsed_kws);
1195
1196 if (mentry->full) {
1197 archive_entry_copy_pathname(entry, mentry->name);
1198 /*
1199 * "Full" entries are allowed to have multiple lines
1200 * and those lines aren't required to be adjacent. We
1201 * don't support multiple lines for "relative" entries
1202 * nor do we make any attempt to merge data from
1203 * separate "relative" and "full" entries. (Merging
1204 * "relative" and "full" entries would require dealing
1205 * with pathname canonicalization, which is a very
1206 * tricky subject.)
1207 */
1208 mp = (struct mtree_entry *)__archive_rb_tree_find_node(
1209 &mtree->rbtree, mentry->name);
1210 for (; mp; mp = mp->next_dup) {
1211 if (mp->full && !mp->used) {
1212 /* Later lines override earlier ones. */
1213 mp->used = 1;
1214 r1 = parse_line(a, entry, mtree, mp, &parsed_kws);
1215 if (r1 < r)
1216 r = r1;
1217 }
1218 }
1219 } else {
1220 /*
1221 * Relative entries require us to construct
1222 * the full path and possibly update the
1223 * current directory.
1224 */
1225 size_t n = archive_strlen(&mtree->current_dir);
1226 if (n > 0)
1227 archive_strcat(&mtree->current_dir, "/");
1228 archive_strcat(&mtree->current_dir, mentry->name);
1229 archive_entry_copy_pathname(entry, mtree->current_dir.s);
1230 if (archive_entry_filetype(entry) != AE_IFDIR)
1231 mtree->current_dir.length = n;
1232 }
1233
1234 if (mtree->checkfs) {
1235 /*
1236 * Try to open and stat the file to get the real size
1237 * and other file info. It would be nice to avoid
1238 * this here so that getting a listing of an mtree
1239 * wouldn't require opening every referenced contents
1240 * file. But then we wouldn't know the actual
1241 * contents size, so I don't see a really viable way
1242 * around this. (Also, we may want to someday pull
1243 * other unspecified info from the contents file on
1244 * disk.)
1245 */
1246 mtree->fd = -1;
1247 if (archive_strlen(&mtree->contents_name) > 0)
1248 path = mtree->contents_name.s;
1249 else
1250 path = archive_entry_pathname(entry);
1251
1252 if (archive_entry_filetype(entry) == AE_IFREG ||
1253 archive_entry_filetype(entry) == AE_IFDIR) {
1254 mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC);
1255 __archive_ensure_cloexec_flag(mtree->fd);
1256 if (mtree->fd == -1 &&
1257 (errno != ENOENT ||
1258 archive_strlen(&mtree->contents_name) > 0)) {
1259 archive_set_error(&a->archive, errno,
1260 "Can't open %s", path);
1261 r = ARCHIVE_WARN;
1262 }
1263 }
1264
1265 st = &st_storage;
1266 if (mtree->fd >= 0) {
1267 if (fstat(mtree->fd, st) == -1) {
1268 archive_set_error(&a->archive, errno,
1269 "Could not fstat %s", path);
1270 r = ARCHIVE_WARN;
1271 /* If we can't stat it, don't keep it open. */
1272 close(mtree->fd);
1273 mtree->fd = -1;
1274 st = NULL;
1275 }
1276 } else if (lstat(path, st) == -1) {
1277 st = NULL;
1278 }
1279
1280 /*
1281 * Check for a mismatch between the type in the specification
1282 * and the type of the contents object on disk.
1283 */
1284 if (st != NULL) {
1285 if (((st->st_mode & S_IFMT) == S_IFREG &&
1286 archive_entry_filetype(entry) == AE_IFREG)
1287 #ifdef S_IFLNK
1288 ||((st->st_mode & S_IFMT) == S_IFLNK &&
1289 archive_entry_filetype(entry) == AE_IFLNK)
1290 #endif
1291 #ifdef S_IFSOCK
1292 ||((st->st_mode & S_IFSOCK) == S_IFSOCK &&
1293 archive_entry_filetype(entry) == AE_IFSOCK)
1294 #endif
1295 #ifdef S_IFCHR
1296 ||((st->st_mode & S_IFMT) == S_IFCHR &&
1297 archive_entry_filetype(entry) == AE_IFCHR)
1298 #endif
1299 #ifdef S_IFBLK
1300 ||((st->st_mode & S_IFMT) == S_IFBLK &&
1301 archive_entry_filetype(entry) == AE_IFBLK)
1302 #endif
1303 ||((st->st_mode & S_IFMT) == S_IFDIR &&
1304 archive_entry_filetype(entry) == AE_IFDIR)
1305 #ifdef S_IFIFO
1306 ||((st->st_mode & S_IFMT) == S_IFIFO &&
1307 archive_entry_filetype(entry) == AE_IFIFO)
1308 #endif
1309 ) {
1310 /* Types match. */
1311 } else {
1312 /* Types don't match; bail out gracefully. */
1313 if (mtree->fd >= 0)
1314 close(mtree->fd);
1315 mtree->fd = -1;
1316 if (parsed_kws & MTREE_HAS_OPTIONAL) {
1317 /* It's not an error for an optional
1318 * entry to not match disk. */
1319 *use_next = 1;
1320 } else if (r == ARCHIVE_OK) {
1321 archive_set_error(&a->archive,
1322 ARCHIVE_ERRNO_MISC,
1323 "mtree specification has different"
1324 " type for %s",
1325 archive_entry_pathname(entry));
1326 r = ARCHIVE_WARN;
1327 }
1328 return (r);
1329 }
1330 }
1331
1332 /*
1333 * If there is a contents file on disk, pick some of the
1334 * metadata from that file. For most of these, we only
1335 * set it from the contents if it wasn't already parsed
1336 * from the specification.
1337 */
1338 if (st != NULL) {
1339 if (((parsed_kws & MTREE_HAS_DEVICE) == 0 ||
1340 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) &&
1341 (archive_entry_filetype(entry) == AE_IFCHR ||
1342 archive_entry_filetype(entry) == AE_IFBLK))
1343 archive_entry_set_rdev(entry, st->st_rdev);
1344 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME))
1345 == 0 ||
1346 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1347 archive_entry_set_gid(entry, st->st_gid);
1348 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME))
1349 == 0 ||
1350 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1351 archive_entry_set_uid(entry, st->st_uid);
1352 if ((parsed_kws & MTREE_HAS_MTIME) == 0 ||
1353 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) {
1354 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
1355 archive_entry_set_mtime(entry, st->st_mtime,
1356 st->st_mtimespec.tv_nsec);
1357 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
1358 archive_entry_set_mtime(entry, st->st_mtime,
1359 st->st_mtim.tv_nsec);
1360 #elif HAVE_STRUCT_STAT_ST_MTIME_N
1361 archive_entry_set_mtime(entry, st->st_mtime,
1362 st->st_mtime_n);
1363 #elif HAVE_STRUCT_STAT_ST_UMTIME
1364 archive_entry_set_mtime(entry, st->st_mtime,
1365 st->st_umtime*1000);
1366 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
1367 archive_entry_set_mtime(entry, st->st_mtime,
1368 st->st_mtime_usec*1000);
1369 #else
1370 archive_entry_set_mtime(entry, st->st_mtime, 0);
1371 #endif
1372 }
1373 if ((parsed_kws & MTREE_HAS_NLINK) == 0 ||
1374 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1375 archive_entry_set_nlink(entry, st->st_nlink);
1376 if ((parsed_kws & MTREE_HAS_PERM) == 0 ||
1377 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1378 archive_entry_set_perm(entry, st->st_mode);
1379 if ((parsed_kws & MTREE_HAS_SIZE) == 0 ||
1380 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1381 archive_entry_set_size(entry, st->st_size);
1382 archive_entry_set_ino(entry, st->st_ino);
1383 archive_entry_set_dev(entry, st->st_dev);
1384
1385 archive_entry_linkify(mtree->resolver, &entry,
1386 &sparse_entry);
1387 } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
1388 /*
1389 * Couldn't open the entry, stat it or the on-disk type
1390 * didn't match. If this entry is optional, just
1391 * ignore it and read the next header entry.
1392 */
1393 *use_next = 1;
1394 return ARCHIVE_OK;
1395 }
1396 }
1397
1398 mtree->cur_size = archive_entry_size(entry);
1399 mtree->offset = 0;
1400
1401 return r;
1402 }
1403
1404 /*
1405 * Each line contains a sequence of keywords.
1406 */
1407 static int
parse_line(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mp,int * parsed_kws)1408 parse_line(struct archive_read *a, struct archive_entry *entry,
1409 struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
1410 {
1411 struct mtree_option *iter;
1412 int r = ARCHIVE_OK, r1;
1413
1414 for (iter = mp->options; iter != NULL; iter = iter->next) {
1415 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
1416 if (r1 < r)
1417 r = r1;
1418 }
1419 if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) {
1420 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1421 "Missing type keyword in mtree specification");
1422 return (ARCHIVE_WARN);
1423 }
1424 return (r);
1425 }
1426
1427 /*
1428 * Device entries have one of the following forms:
1429 * - raw dev_t
1430 * - format,major,minor[,subdevice]
1431 * When parsing succeeded, `pdev' will contain the appropriate dev_t value.
1432 */
1433
1434 /* strsep() is not in C90, but strcspn() is. */
1435 /* Taken from http://unixpapa.com/incnote/string.html */
1436 static char *
la_strsep(char ** sp,const char * sep)1437 la_strsep(char **sp, const char *sep)
1438 {
1439 char *p, *s;
1440 if (sp == NULL || *sp == NULL || **sp == '\0')
1441 return(NULL);
1442 s = *sp;
1443 p = s + strcspn(s, sep);
1444 if (*p != '\0')
1445 *p++ = '\0';
1446 *sp = p;
1447 return(s);
1448 }
1449
1450 static int
parse_device(dev_t * pdev,struct archive * a,char * val)1451 parse_device(dev_t *pdev, struct archive *a, char *val)
1452 {
1453 #define MAX_PACK_ARGS 3
1454 unsigned long numbers[MAX_PACK_ARGS];
1455 char *p, *dev;
1456 int argc;
1457 pack_t *pack;
1458 dev_t result;
1459 const char *error = NULL;
1460
1461 memset(pdev, 0, sizeof(*pdev));
1462 if ((dev = strchr(val, ',')) != NULL) {
1463 /*
1464 * Device's major/minor are given in a specified format.
1465 * Decode and pack it accordingly.
1466 */
1467 *dev++ = '\0';
1468 if ((pack = pack_find(val)) == NULL) {
1469 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1470 "Unknown format `%s'", val);
1471 return ARCHIVE_WARN;
1472 }
1473 argc = 0;
1474 while ((p = la_strsep(&dev, ",")) != NULL) {
1475 if (*p == '\0') {
1476 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1477 "Missing number");
1478 return ARCHIVE_WARN;
1479 }
1480 if (argc >= MAX_PACK_ARGS) {
1481 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1482 "Too many arguments");
1483 return ARCHIVE_WARN;
1484 }
1485 numbers[argc++] = (unsigned long)mtree_atol(&p, 0);
1486 }
1487 if (argc < 2) {
1488 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1489 "Not enough arguments");
1490 return ARCHIVE_WARN;
1491 }
1492 result = (*pack)(argc, numbers, &error);
1493 if (error != NULL) {
1494 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1495 "%s", error);
1496 return ARCHIVE_WARN;
1497 }
1498 } else {
1499 /* file system raw value. */
1500 result = (dev_t)mtree_atol(&val, 0);
1501 }
1502 *pdev = result;
1503 return ARCHIVE_OK;
1504 #undef MAX_PACK_ARGS
1505 }
1506
1507 static int
parse_hex_nibble(char c)1508 parse_hex_nibble(char c)
1509 {
1510 if (c >= '0' && c <= '9')
1511 return c - '0';
1512 if (c >= 'a' && c <= 'f')
1513 return 10 + c - 'a';
1514 #if 0
1515 /* XXX: Is uppercase something we should support? */
1516 if (c >= 'A' && c <= 'F')
1517 return 10 + c - 'A';
1518 #endif
1519
1520 return -1;
1521 }
1522
1523 static int
parse_digest(struct archive_read * a,struct archive_entry * entry,const char * digest,int type)1524 parse_digest(struct archive_read *a, struct archive_entry *entry,
1525 const char *digest, int type)
1526 {
1527 unsigned char digest_buf[64];
1528 int high, low;
1529 size_t i, j, len;
1530
1531 switch (type) {
1532 case ARCHIVE_ENTRY_DIGEST_MD5:
1533 len = sizeof(entry->digest.md5);
1534 break;
1535 case ARCHIVE_ENTRY_DIGEST_RMD160:
1536 len = sizeof(entry->digest.rmd160);
1537 break;
1538 case ARCHIVE_ENTRY_DIGEST_SHA1:
1539 len = sizeof(entry->digest.sha1);
1540 break;
1541 case ARCHIVE_ENTRY_DIGEST_SHA256:
1542 len = sizeof(entry->digest.sha256);
1543 break;
1544 case ARCHIVE_ENTRY_DIGEST_SHA384:
1545 len = sizeof(entry->digest.sha384);
1546 break;
1547 case ARCHIVE_ENTRY_DIGEST_SHA512:
1548 len = sizeof(entry->digest.sha512);
1549 break;
1550 default:
1551 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1552 "Internal error: Unknown digest type");
1553 return ARCHIVE_FATAL;
1554 }
1555
1556 if (len > sizeof(digest_buf)) {
1557 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1558 "Internal error: Digest storage too large");
1559 return ARCHIVE_FATAL;
1560 }
1561
1562 len *= 2;
1563
1564 if (mtree_strnlen(digest, len+1) != len) {
1565 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1566 "incorrect digest length, ignoring");
1567 return ARCHIVE_WARN;
1568 }
1569
1570 for (i = 0, j = 0; i < len; i += 2, j++) {
1571 high = parse_hex_nibble(digest[i]);
1572 low = parse_hex_nibble(digest[i+1]);
1573 if (high == -1 || low == -1) {
1574 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1575 "invalid digest data, ignoring");
1576 return ARCHIVE_WARN;
1577 }
1578
1579 digest_buf[j] = high << 4 | low;
1580 }
1581
1582 return archive_entry_set_digest(entry, type, digest_buf);
1583 }
1584
1585 /*
1586 * Parse a single keyword and its value.
1587 */
1588 static int
parse_keyword(struct archive_read * a,struct mtree * mtree,struct archive_entry * entry,struct mtree_option * opt,int * parsed_kws)1589 parse_keyword(struct archive_read *a, struct mtree *mtree,
1590 struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws)
1591 {
1592 char *val, *key;
1593
1594 key = opt->value;
1595
1596 if (*key == '\0')
1597 return (ARCHIVE_OK);
1598
1599 if (strcmp(key, "nochange") == 0) {
1600 *parsed_kws |= MTREE_HAS_NOCHANGE;
1601 return (ARCHIVE_OK);
1602 }
1603 if (strcmp(key, "optional") == 0) {
1604 *parsed_kws |= MTREE_HAS_OPTIONAL;
1605 return (ARCHIVE_OK);
1606 }
1607 if (strcmp(key, "ignore") == 0) {
1608 /*
1609 * The mtree processing is not recursive, so
1610 * recursion will only happen for explicitly listed
1611 * entries.
1612 */
1613 return (ARCHIVE_OK);
1614 }
1615
1616 val = strchr(key, '=');
1617 if (val == NULL) {
1618 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1619 "Malformed attribute \"%s\" (%d)", key, key[0]);
1620 return (ARCHIVE_WARN);
1621 }
1622
1623 *val = '\0';
1624 ++val;
1625
1626 switch (key[0]) {
1627 case 'c':
1628 if (strcmp(key, "content") == 0
1629 || strcmp(key, "contents") == 0) {
1630 parse_escapes(val, NULL);
1631 archive_strcpy(&mtree->contents_name, val);
1632 break;
1633 }
1634 if (strcmp(key, "cksum") == 0)
1635 break;
1636 __LA_FALLTHROUGH;
1637 case 'd':
1638 if (strcmp(key, "device") == 0) {
1639 /* stat(2) st_rdev field, e.g. the major/minor IDs
1640 * of a char/block special file */
1641 int r;
1642 dev_t dev;
1643
1644 *parsed_kws |= MTREE_HAS_DEVICE;
1645 r = parse_device(&dev, &a->archive, val);
1646 if (r == ARCHIVE_OK)
1647 archive_entry_set_rdev(entry, dev);
1648 return r;
1649 }
1650 __LA_FALLTHROUGH;
1651 case 'f':
1652 if (strcmp(key, "flags") == 0) {
1653 *parsed_kws |= MTREE_HAS_FFLAGS;
1654 archive_entry_copy_fflags_text(entry, val);
1655 break;
1656 }
1657 __LA_FALLTHROUGH;
1658 case 'g':
1659 if (strcmp(key, "gid") == 0) {
1660 *parsed_kws |= MTREE_HAS_GID;
1661 archive_entry_set_gid(entry, mtree_atol(&val, 10));
1662 break;
1663 }
1664 if (strcmp(key, "gname") == 0) {
1665 *parsed_kws |= MTREE_HAS_GNAME;
1666 archive_entry_copy_gname(entry, val);
1667 break;
1668 }
1669 __LA_FALLTHROUGH;
1670 case 'i':
1671 if (strcmp(key, "inode") == 0) {
1672 archive_entry_set_ino(entry, mtree_atol(&val, 10));
1673 break;
1674 }
1675 __LA_FALLTHROUGH;
1676 case 'l':
1677 if (strcmp(key, "link") == 0) {
1678 archive_entry_copy_symlink(entry, val);
1679 break;
1680 }
1681 __LA_FALLTHROUGH;
1682 case 'm':
1683 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) {
1684 return parse_digest(a, entry, val,
1685 ARCHIVE_ENTRY_DIGEST_MD5);
1686 }
1687 if (strcmp(key, "mode") == 0) {
1688 if (val[0] >= '0' && val[0] <= '7') {
1689 *parsed_kws |= MTREE_HAS_PERM;
1690 archive_entry_set_perm(entry,
1691 (mode_t)mtree_atol(&val, 8));
1692 } else {
1693 archive_set_error(&a->archive,
1694 ARCHIVE_ERRNO_FILE_FORMAT,
1695 "Symbolic or non-octal mode \"%s\" unsupported", val);
1696 return ARCHIVE_WARN;
1697 }
1698 break;
1699 }
1700 __LA_FALLTHROUGH;
1701 case 'n':
1702 if (strcmp(key, "nlink") == 0) {
1703 *parsed_kws |= MTREE_HAS_NLINK;
1704 archive_entry_set_nlink(entry,
1705 (unsigned int)mtree_atol(&val, 10));
1706 break;
1707 }
1708 __LA_FALLTHROUGH;
1709 case 'r':
1710 if (strcmp(key, "resdevice") == 0) {
1711 /* stat(2) st_dev field, e.g. the device ID where the
1712 * inode resides */
1713 int r;
1714 dev_t dev;
1715
1716 r = parse_device(&dev, &a->archive, val);
1717 if (r == ARCHIVE_OK)
1718 archive_entry_set_dev(entry, dev);
1719 return r;
1720 }
1721 if (strcmp(key, "rmd160") == 0 ||
1722 strcmp(key, "rmd160digest") == 0) {
1723 return parse_digest(a, entry, val,
1724 ARCHIVE_ENTRY_DIGEST_RMD160);
1725 }
1726 __LA_FALLTHROUGH;
1727 case 's':
1728 if (strcmp(key, "sha1") == 0 ||
1729 strcmp(key, "sha1digest") == 0) {
1730 return parse_digest(a, entry, val,
1731 ARCHIVE_ENTRY_DIGEST_SHA1);
1732 }
1733 if (strcmp(key, "sha256") == 0 ||
1734 strcmp(key, "sha256digest") == 0) {
1735 return parse_digest(a, entry, val,
1736 ARCHIVE_ENTRY_DIGEST_SHA256);
1737 }
1738 if (strcmp(key, "sha384") == 0 ||
1739 strcmp(key, "sha384digest") == 0) {
1740 return parse_digest(a, entry, val,
1741 ARCHIVE_ENTRY_DIGEST_SHA384);
1742 }
1743 if (strcmp(key, "sha512") == 0 ||
1744 strcmp(key, "sha512digest") == 0) {
1745 return parse_digest(a, entry, val,
1746 ARCHIVE_ENTRY_DIGEST_SHA512);
1747 }
1748 if (strcmp(key, "size") == 0) {
1749 archive_entry_set_size(entry, mtree_atol(&val, 10));
1750 break;
1751 }
1752 __LA_FALLTHROUGH;
1753 case 't':
1754 if (strcmp(key, "tags") == 0) {
1755 /*
1756 * Comma delimited list of tags.
1757 * Ignore the tags for now, but the interface
1758 * should be extended to allow inclusion/exclusion.
1759 */
1760 break;
1761 }
1762 if (strcmp(key, "time") == 0) {
1763 int64_t m;
1764 int64_t my_time_t_max = get_time_t_max();
1765 int64_t my_time_t_min = get_time_t_min();
1766 long ns = 0;
1767
1768 *parsed_kws |= MTREE_HAS_MTIME;
1769 m = mtree_atol(&val, 10);
1770 /* Replicate an old mtree bug:
1771 * 123456789.1 represents 123456789
1772 * seconds and 1 nanosecond. */
1773 if (*val == '.') {
1774 ++val;
1775 ns = (long)mtree_atol(&val, 10);
1776 if (ns < 0)
1777 ns = 0;
1778 else if (ns > 999999999)
1779 ns = 999999999;
1780 }
1781 if (m > my_time_t_max)
1782 m = my_time_t_max;
1783 else if (m < my_time_t_min)
1784 m = my_time_t_min;
1785 archive_entry_set_mtime(entry, (time_t)m, ns);
1786 break;
1787 }
1788 if (strcmp(key, "type") == 0) {
1789 switch (val[0]) {
1790 case 'b':
1791 if (strcmp(val, "block") == 0) {
1792 archive_entry_set_filetype(entry, AE_IFBLK);
1793 break;
1794 }
1795 __LA_FALLTHROUGH;
1796 case 'c':
1797 if (strcmp(val, "char") == 0) {
1798 archive_entry_set_filetype(entry,
1799 AE_IFCHR);
1800 break;
1801 }
1802 __LA_FALLTHROUGH;
1803 case 'd':
1804 if (strcmp(val, "dir") == 0) {
1805 archive_entry_set_filetype(entry,
1806 AE_IFDIR);
1807 break;
1808 }
1809 __LA_FALLTHROUGH;
1810 case 'f':
1811 if (strcmp(val, "fifo") == 0) {
1812 archive_entry_set_filetype(entry,
1813 AE_IFIFO);
1814 break;
1815 }
1816 if (strcmp(val, "file") == 0) {
1817 archive_entry_set_filetype(entry,
1818 AE_IFREG);
1819 break;
1820 }
1821 __LA_FALLTHROUGH;
1822 case 'l':
1823 if (strcmp(val, "link") == 0) {
1824 archive_entry_set_filetype(entry,
1825 AE_IFLNK);
1826 break;
1827 }
1828 __LA_FALLTHROUGH;
1829 default:
1830 archive_set_error(&a->archive,
1831 ARCHIVE_ERRNO_FILE_FORMAT,
1832 "Unrecognized file type \"%s\"; "
1833 "assuming \"file\"", val);
1834 archive_entry_set_filetype(entry, AE_IFREG);
1835 return (ARCHIVE_WARN);
1836 }
1837 *parsed_kws |= MTREE_HAS_TYPE;
1838 break;
1839 }
1840 __LA_FALLTHROUGH;
1841 case 'u':
1842 if (strcmp(key, "uid") == 0) {
1843 *parsed_kws |= MTREE_HAS_UID;
1844 archive_entry_set_uid(entry, mtree_atol(&val, 10));
1845 break;
1846 }
1847 if (strcmp(key, "uname") == 0) {
1848 *parsed_kws |= MTREE_HAS_UNAME;
1849 archive_entry_copy_uname(entry, val);
1850 break;
1851 }
1852 __LA_FALLTHROUGH;
1853 default:
1854 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1855 "Unrecognized key %s=%s", key, val);
1856 return (ARCHIVE_WARN);
1857 }
1858 return (ARCHIVE_OK);
1859 }
1860
1861 static int
read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1862 read_data(struct archive_read *a, const void **buff, size_t *size,
1863 int64_t *offset)
1864 {
1865 size_t bytes_to_read;
1866 ssize_t bytes_read;
1867 struct mtree *mtree;
1868
1869 mtree = (struct mtree *)(a->format->data);
1870 if (mtree->fd < 0) {
1871 *buff = NULL;
1872 *offset = 0;
1873 *size = 0;
1874 return (ARCHIVE_EOF);
1875 }
1876 if (mtree->buff == NULL) {
1877 mtree->buffsize = 64 * 1024;
1878 mtree->buff = malloc(mtree->buffsize);
1879 if (mtree->buff == NULL) {
1880 archive_set_error(&a->archive, ENOMEM,
1881 "Can't allocate memory");
1882 return (ARCHIVE_FATAL);
1883 }
1884 }
1885
1886 *buff = mtree->buff;
1887 *offset = mtree->offset;
1888 if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset)
1889 bytes_to_read = (size_t)(mtree->cur_size - mtree->offset);
1890 else
1891 bytes_to_read = mtree->buffsize;
1892 bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
1893 if (bytes_read < 0) {
1894 archive_set_error(&a->archive, errno, "Can't read");
1895 return (ARCHIVE_WARN);
1896 }
1897 if (bytes_read == 0) {
1898 *size = 0;
1899 return (ARCHIVE_EOF);
1900 }
1901 mtree->offset += bytes_read;
1902 *size = bytes_read;
1903 return (ARCHIVE_OK);
1904 }
1905
1906 /* Skip does nothing except possibly close the contents file. */
1907 static int
skip(struct archive_read * a)1908 skip(struct archive_read *a)
1909 {
1910 struct mtree *mtree;
1911
1912 mtree = (struct mtree *)(a->format->data);
1913 if (mtree->fd >= 0) {
1914 close(mtree->fd);
1915 mtree->fd = -1;
1916 }
1917 return (ARCHIVE_OK);
1918 }
1919
1920 /*
1921 * Since parsing backslash sequences always makes strings shorter,
1922 * we can always do this conversion in-place.
1923 */
1924 static void
parse_escapes(char * src,struct mtree_entry * mentry)1925 parse_escapes(char *src, struct mtree_entry *mentry)
1926 {
1927 char *dest = src;
1928 char c;
1929
1930 if (mentry != NULL && strcmp(src, ".") == 0)
1931 mentry->full = 1;
1932
1933 while (*src != '\0') {
1934 c = *src++;
1935 if (c == '/' && mentry != NULL)
1936 mentry->full = 1;
1937 if (c == '\\') {
1938 switch (src[0]) {
1939 case '0':
1940 if (src[1] < '0' || src[1] > '7') {
1941 c = 0;
1942 ++src;
1943 break;
1944 }
1945 /* FALLTHROUGH */
1946 case '1':
1947 case '2':
1948 case '3':
1949 if (src[1] >= '0' && src[1] <= '7' &&
1950 src[2] >= '0' && src[2] <= '7') {
1951 c = (src[0] - '0') << 6;
1952 c |= (src[1] - '0') << 3;
1953 c |= (src[2] - '0');
1954 src += 3;
1955 }
1956 break;
1957 case 'a':
1958 c = '\a';
1959 ++src;
1960 break;
1961 case 'b':
1962 c = '\b';
1963 ++src;
1964 break;
1965 case 'f':
1966 c = '\f';
1967 ++src;
1968 break;
1969 case 'n':
1970 c = '\n';
1971 ++src;
1972 break;
1973 case 'r':
1974 c = '\r';
1975 ++src;
1976 break;
1977 case 's':
1978 c = ' ';
1979 ++src;
1980 break;
1981 case 't':
1982 c = '\t';
1983 ++src;
1984 break;
1985 case 'v':
1986 c = '\v';
1987 ++src;
1988 break;
1989 case '\\':
1990 c = '\\';
1991 ++src;
1992 break;
1993 }
1994 }
1995 *dest++ = c;
1996 }
1997 *dest = '\0';
1998 }
1999
2000 /* Parse a hex digit. */
2001 static int
parsedigit(char c)2002 parsedigit(char c)
2003 {
2004 if (c >= '0' && c <= '9')
2005 return c - '0';
2006 else if (c >= 'a' && c <= 'f')
2007 return c - 'a';
2008 else if (c >= 'A' && c <= 'F')
2009 return c - 'A';
2010 else
2011 return -1;
2012 }
2013
2014 /*
2015 * Note that this implementation does not (and should not!) obey
2016 * locale settings; you cannot simply substitute strtol here, since
2017 * it does obey locale.
2018 */
2019 static int64_t
mtree_atol(char ** p,int base)2020 mtree_atol(char **p, int base)
2021 {
2022 int64_t l, limit;
2023 int digit, last_digit_limit;
2024
2025 if (base == 0) {
2026 if (**p != '0')
2027 base = 10;
2028 else if ((*p)[1] == 'x' || (*p)[1] == 'X') {
2029 *p += 2;
2030 base = 16;
2031 } else {
2032 base = 8;
2033 }
2034 }
2035
2036 if (**p == '-') {
2037 limit = INT64_MIN / base;
2038 last_digit_limit = INT64_MIN % base;
2039 ++(*p);
2040
2041 l = 0;
2042 digit = parsedigit(**p);
2043 while (digit >= 0 && digit < base) {
2044 if (l < limit || (l == limit && digit > last_digit_limit))
2045 return INT64_MIN;
2046 l = (l * base) - digit;
2047 digit = parsedigit(*++(*p));
2048 }
2049 return l;
2050 } else {
2051 limit = INT64_MAX / base;
2052 last_digit_limit = INT64_MAX % base;
2053
2054 l = 0;
2055 digit = parsedigit(**p);
2056 while (digit >= 0 && digit < base) {
2057 if (l > limit || (l == limit && digit > last_digit_limit))
2058 return INT64_MAX;
2059 l = (l * base) + digit;
2060 digit = parsedigit(*++(*p));
2061 }
2062 return l;
2063 }
2064 }
2065
2066 /*
2067 * Returns length of line (including trailing newline)
2068 * or negative on error. 'start' argument is updated to
2069 * point to first character of line.
2070 */
2071 static ssize_t
readline(struct archive_read * a,struct mtree * mtree,char ** start,ssize_t limit)2072 readline(struct archive_read *a, struct mtree *mtree, char **start,
2073 ssize_t limit)
2074 {
2075 ssize_t bytes_read;
2076 ssize_t total_size = 0;
2077 ssize_t find_off = 0;
2078 const void *t;
2079 void *nl;
2080 char *u;
2081
2082 /* Accumulate line in a line buffer. */
2083 for (;;) {
2084 /* Read some more. */
2085 t = __archive_read_ahead(a, 1, &bytes_read);
2086 if (t == NULL)
2087 return (0);
2088 if (bytes_read < 0)
2089 return (ARCHIVE_FATAL);
2090 nl = memchr(t, '\n', bytes_read);
2091 /* If we found '\n', trim the read to end exactly there. */
2092 if (nl != NULL) {
2093 bytes_read = ((const char *)nl) - ((const char *)t) + 1;
2094 }
2095 if (total_size + bytes_read + 1 > limit) {
2096 archive_set_error(&a->archive,
2097 ARCHIVE_ERRNO_FILE_FORMAT,
2098 "Line too long");
2099 return (ARCHIVE_FATAL);
2100 }
2101 if (archive_string_ensure(&mtree->line,
2102 total_size + bytes_read + 1) == NULL) {
2103 archive_set_error(&a->archive, ENOMEM,
2104 "Can't allocate working buffer");
2105 return (ARCHIVE_FATAL);
2106 }
2107 /* Append new bytes to string. */
2108 memcpy(mtree->line.s + total_size, t, bytes_read);
2109 __archive_read_consume(a, bytes_read);
2110 total_size += bytes_read;
2111 mtree->line.s[total_size] = '\0';
2112
2113 for (u = mtree->line.s + find_off; *u; ++u) {
2114 if (u[0] == '\n') {
2115 /* Ends with unescaped newline. */
2116 *start = mtree->line.s;
2117 return total_size;
2118 } else if (u[0] == '#') {
2119 /* Ends with comment sequence #...\n */
2120 if (nl == NULL) {
2121 /* But we've not found the \n yet */
2122 break;
2123 }
2124 } else if (u[0] == '\\') {
2125 if (u[1] == '\n') {
2126 /* Trim escaped newline. */
2127 total_size -= 2;
2128 mtree->line.s[total_size] = '\0';
2129 break;
2130 } else if (u[1] != '\0') {
2131 /* Skip the two-char escape sequence */
2132 ++u;
2133 }
2134 }
2135 }
2136 find_off = u - mtree->line.s;
2137 }
2138 }
2139