archive_read_support_format_mtree.c revision 368708
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: stable/10/contrib/libarchive/libarchive/archive_read_support_format_mtree.c 368708 2020-12-16 22:25:40Z mm $"); 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 85struct mtree_option { 86 struct mtree_option *next; 87 char *value; 88}; 89 90struct 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 100struct 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 121static int bid_keycmp(const char *, const char *, ssize_t); 122static int cleanup(struct archive_read *); 123static int detect_form(struct archive_read *, int *); 124static int mtree_bid(struct archive_read *, int); 125static int parse_file(struct archive_read *, struct archive_entry *, 126 struct mtree *, struct mtree_entry *, int *); 127static void parse_escapes(char *, struct mtree_entry *); 128static int parse_line(struct archive_read *, struct archive_entry *, 129 struct mtree *, struct mtree_entry *, int *); 130static int parse_keyword(struct archive_read *, struct mtree *, 131 struct archive_entry *, struct mtree_option *, int *); 132static int read_data(struct archive_read *a, 133 const void **buff, size_t *size, int64_t *offset); 134static ssize_t readline(struct archive_read *, struct mtree *, char **, ssize_t); 135static int skip(struct archive_read *a); 136static int read_header(struct archive_read *, 137 struct archive_entry *); 138static int64_t mtree_atol(char **, int base); 139 140/* 141 * There's no standard for TIME_T_MAX/TIME_T_MIN. So we compute them 142 * here. TODO: Move this to configure time, but be careful 143 * about cross-compile environments. 144 */ 145static int64_t 146get_time_t_max(void) 147{ 148#if defined(TIME_T_MAX) 149 return TIME_T_MAX; 150#else 151 /* ISO C allows time_t to be a floating-point type, 152 but POSIX requires an integer type. The following 153 should work on any system that follows the POSIX 154 conventions. */ 155 if (((time_t)0) < ((time_t)-1)) { 156 /* Time_t is unsigned */ 157 return (~(time_t)0); 158 } else { 159 /* Time_t is signed. */ 160 /* Assume it's the same as int64_t or int32_t */ 161 if (sizeof(time_t) == sizeof(int64_t)) { 162 return (time_t)INT64_MAX; 163 } else { 164 return (time_t)INT32_MAX; 165 } 166 } 167#endif 168} 169 170static int64_t 171get_time_t_min(void) 172{ 173#if defined(TIME_T_MIN) 174 return TIME_T_MIN; 175#else 176 if (((time_t)0) < ((time_t)-1)) { 177 /* Time_t is unsigned */ 178 return (time_t)0; 179 } else { 180 /* Time_t is signed. */ 181 if (sizeof(time_t) == sizeof(int64_t)) { 182 return (time_t)INT64_MIN; 183 } else { 184 return (time_t)INT32_MIN; 185 } 186 } 187#endif 188} 189 190static int 191archive_read_format_mtree_options(struct archive_read *a, 192 const char *key, const char *val) 193{ 194 struct mtree *mtree; 195 196 mtree = (struct mtree *)(a->format->data); 197 if (strcmp(key, "checkfs") == 0) { 198 /* Allows to read information missing from the mtree from the file system */ 199 if (val == NULL || val[0] == 0) { 200 mtree->checkfs = 0; 201 } else { 202 mtree->checkfs = 1; 203 } 204 return (ARCHIVE_OK); 205 } 206 207 /* Note: The "warn" return is just to inform the options 208 * supervisor that we didn't handle it. It will generate 209 * a suitable error if no one used this option. */ 210 return (ARCHIVE_WARN); 211} 212 213static void 214free_options(struct mtree_option *head) 215{ 216 struct mtree_option *next; 217 218 for (; head != NULL; head = next) { 219 next = head->next; 220 free(head->value); 221 free(head); 222 } 223} 224 225static int 226mtree_cmp_node(const struct archive_rb_node *n1, 227 const struct archive_rb_node *n2) 228{ 229 const struct mtree_entry *e1 = (const struct mtree_entry *)n1; 230 const struct mtree_entry *e2 = (const struct mtree_entry *)n2; 231 232 return (strcmp(e1->name, e2->name)); 233} 234 235static int 236mtree_cmp_key(const struct archive_rb_node *n, const void *key) 237{ 238 const struct mtree_entry *e = (const struct mtree_entry *)n; 239 240 return (strcmp(e->name, key)); 241} 242 243int 244archive_read_support_format_mtree(struct archive *_a) 245{ 246 static const struct archive_rb_tree_ops rb_ops = { 247 mtree_cmp_node, mtree_cmp_key, 248 }; 249 struct archive_read *a = (struct archive_read *)_a; 250 struct mtree *mtree; 251 int r; 252 253 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 254 ARCHIVE_STATE_NEW, "archive_read_support_format_mtree"); 255 256 mtree = (struct mtree *)calloc(1, sizeof(*mtree)); 257 if (mtree == NULL) { 258 archive_set_error(&a->archive, ENOMEM, 259 "Can't allocate mtree data"); 260 return (ARCHIVE_FATAL); 261 } 262 mtree->checkfs = 0; 263 mtree->fd = -1; 264 265 __archive_rb_tree_init(&mtree->rbtree, &rb_ops); 266 267 r = __archive_read_register_format(a, mtree, "mtree", 268 mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL); 269 270 if (r != ARCHIVE_OK) 271 free(mtree); 272 return (ARCHIVE_OK); 273} 274 275static int 276cleanup(struct archive_read *a) 277{ 278 struct mtree *mtree; 279 struct mtree_entry *p, *q; 280 281 mtree = (struct mtree *)(a->format->data); 282 283 p = mtree->entries; 284 while (p != NULL) { 285 q = p->next; 286 free(p->name); 287 free_options(p->options); 288 free(p); 289 p = q; 290 } 291 archive_string_free(&mtree->line); 292 archive_string_free(&mtree->current_dir); 293 archive_string_free(&mtree->contents_name); 294 archive_entry_linkresolver_free(mtree->resolver); 295 296 free(mtree->buff); 297 free(mtree); 298 (a->format->data) = NULL; 299 return (ARCHIVE_OK); 300} 301 302static ssize_t 303get_line_size(const char *b, ssize_t avail, ssize_t *nlsize) 304{ 305 ssize_t len; 306 307 len = 0; 308 while (len < avail) { 309 switch (*b) { 310 case '\0':/* Non-ascii character or control character. */ 311 if (nlsize != NULL) 312 *nlsize = 0; 313 return (-1); 314 case '\r': 315 if (avail-len > 1 && b[1] == '\n') { 316 if (nlsize != NULL) 317 *nlsize = 2; 318 return (len+2); 319 } 320 /* FALL THROUGH */ 321 case '\n': 322 if (nlsize != NULL) 323 *nlsize = 1; 324 return (len+1); 325 default: 326 b++; 327 len++; 328 break; 329 } 330 } 331 if (nlsize != NULL) 332 *nlsize = 0; 333 return (avail); 334} 335 336/* 337 * <---------------- ravail ---------------------> 338 * <-- diff ------> <--- avail -----------------> 339 * <---- len -----------> 340 * | Previous lines | line being parsed nl extra | 341 * ^ 342 * b 343 * 344 */ 345static ssize_t 346next_line(struct archive_read *a, 347 const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl) 348{ 349 ssize_t len; 350 int quit; 351 352 quit = 0; 353 if (*avail == 0) { 354 *nl = 0; 355 len = 0; 356 } else 357 len = get_line_size(*b, *avail, nl); 358 /* 359 * Read bytes more while it does not reach the end of line. 360 */ 361 while (*nl == 0 && len == *avail && !quit) { 362 ssize_t diff = *ravail - *avail; 363 size_t nbytes_req = (*ravail+1023) & ~1023U; 364 ssize_t tested; 365 366 /* 367 * Place an arbitrary limit on the line length. 368 * mtree is almost free-form input and without line length limits, 369 * it can consume a lot of memory. 370 */ 371 if (len >= MAX_LINE_LEN) 372 return (-1); 373 374 /* Increase reading bytes if it is not enough to at least 375 * new two lines. */ 376 if (nbytes_req < (size_t)*ravail + 160) 377 nbytes_req <<= 1; 378 379 *b = __archive_read_ahead(a, nbytes_req, avail); 380 if (*b == NULL) { 381 if (*ravail >= *avail) 382 return (0); 383 /* Reading bytes reaches the end of file. */ 384 *b = __archive_read_ahead(a, *avail, avail); 385 quit = 1; 386 } 387 *ravail = *avail; 388 *b += diff; 389 *avail -= diff; 390 tested = len;/* Skip some bytes we already determinated. */ 391 len = get_line_size(*b + len, *avail - len, nl); 392 if (len >= 0) 393 len += tested; 394 } 395 return (len); 396} 397 398/* 399 * Compare characters with a mtree keyword. 400 * Returns the length of a mtree keyword if matched. 401 * Returns 0 if not matched. 402 */ 403static int 404bid_keycmp(const char *p, const char *key, ssize_t len) 405{ 406 int match_len = 0; 407 408 while (len > 0 && *p && *key) { 409 if (*p == *key) { 410 --len; 411 ++p; 412 ++key; 413 ++match_len; 414 continue; 415 } 416 return (0);/* Not match */ 417 } 418 if (*key != '\0') 419 return (0);/* Not match */ 420 421 /* A following character should be specified characters */ 422 if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' || 423 p[0] == '\n' || p[0] == '\r' || 424 (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))) 425 return (match_len); 426 return (0);/* Not match */ 427} 428 429/* 430 * Test whether the characters 'p' has is mtree keyword. 431 * Returns the length of a detected keyword. 432 * Returns 0 if any keywords were not found. 433 */ 434static int 435bid_keyword(const char *p, ssize_t len) 436{ 437 static const char * const keys_c[] = { 438 "content", "contents", "cksum", NULL 439 }; 440 static const char * const keys_df[] = { 441 "device", "flags", NULL 442 }; 443 static const char * const keys_g[] = { 444 "gid", "gname", NULL 445 }; 446 static const char * const keys_il[] = { 447 "ignore", "inode", "link", NULL 448 }; 449 static const char * const keys_m[] = { 450 "md5", "md5digest", "mode", NULL 451 }; 452 static const char * const keys_no[] = { 453 "nlink", "nochange", "optional", NULL 454 }; 455 static const char * const keys_r[] = { 456 "resdevice", "rmd160", "rmd160digest", NULL 457 }; 458 static const char * const keys_s[] = { 459 "sha1", "sha1digest", 460 "sha256", "sha256digest", 461 "sha384", "sha384digest", 462 "sha512", "sha512digest", 463 "size", NULL 464 }; 465 static const char * const keys_t[] = { 466 "tags", "time", "type", NULL 467 }; 468 static const char * const keys_u[] = { 469 "uid", "uname", NULL 470 }; 471 const char * const *keys; 472 int i; 473 474 switch (*p) { 475 case 'c': keys = keys_c; break; 476 case 'd': case 'f': keys = keys_df; break; 477 case 'g': keys = keys_g; break; 478 case 'i': case 'l': keys = keys_il; break; 479 case 'm': keys = keys_m; break; 480 case 'n': case 'o': keys = keys_no; break; 481 case 'r': keys = keys_r; break; 482 case 's': keys = keys_s; break; 483 case 't': keys = keys_t; break; 484 case 'u': keys = keys_u; break; 485 default: return (0);/* Unknown key */ 486 } 487 488 for (i = 0; keys[i] != NULL; i++) { 489 int l = bid_keycmp(p, keys[i], len); 490 if (l > 0) 491 return (l); 492 } 493 return (0);/* Unknown key */ 494} 495 496/* 497 * Test whether there is a set of mtree keywords. 498 * Returns the number of keyword. 499 * Returns -1 if we got incorrect sequence. 500 * This function expects a set of "<space characters>keyword=value". 501 * When "unset" is specified, expects a set of "<space characters>keyword". 502 */ 503static int 504bid_keyword_list(const char *p, ssize_t len, int unset, int last_is_path) 505{ 506 int l; 507 int keycnt = 0; 508 509 while (len > 0 && *p) { 510 int blank = 0; 511 512 /* Test whether there are blank characters in the line. */ 513 while (len >0 && (*p == ' ' || *p == '\t')) { 514 ++p; 515 --len; 516 blank = 1; 517 } 518 if (*p == '\n' || *p == '\r') 519 break; 520 if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')) 521 break; 522 if (!blank && !last_is_path) /* No blank character. */ 523 return (-1); 524 if (last_is_path && len == 0) 525 return (keycnt); 526 527 if (unset) { 528 l = bid_keycmp(p, "all", len); 529 if (l > 0) 530 return (1); 531 } 532 /* Test whether there is a correct key in the line. */ 533 l = bid_keyword(p, len); 534 if (l == 0) 535 return (-1);/* Unknown keyword was found. */ 536 p += l; 537 len -= l; 538 keycnt++; 539 540 /* Skip value */ 541 if (*p == '=') { 542 int value = 0; 543 ++p; 544 --len; 545 while (len > 0 && *p != ' ' && *p != '\t') { 546 ++p; 547 --len; 548 value = 1; 549 } 550 /* A keyword should have a its value unless 551 * "/unset" operation. */ 552 if (!unset && value == 0) 553 return (-1); 554 } 555 } 556 return (keycnt); 557} 558 559static int 560bid_entry(const char *p, ssize_t len, ssize_t nl, int *last_is_path) 561{ 562 int f = 0; 563 static const unsigned char safe_char[256] = { 564 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ 565 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ 566 /* !"$%&'()*+,-./ EXCLUSION:( )(#) */ 567 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ 568 /* 0123456789:;<>? EXCLUSION:(=) */ 569 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ 570 /* @ABCDEFGHIJKLMNO */ 571 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ 572 /* PQRSTUVWXYZ[\]^_ */ 573 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */ 574 /* `abcdefghijklmno */ 575 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ 576 /* pqrstuvwxyz{|}~ */ 577 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ 578 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 579 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 580 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 581 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 582 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 583 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 584 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 585 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ 586 }; 587 ssize_t ll; 588 const char *pp = p; 589 const char * const pp_end = pp + len; 590 591 *last_is_path = 0; 592 /* 593 * Skip the path-name which is quoted. 594 */ 595 for (;pp < pp_end; ++pp) { 596 if (!safe_char[*(const unsigned char *)pp]) { 597 if (*pp != ' ' && *pp != '\t' && *pp != '\r' 598 && *pp != '\n') 599 f = 0; 600 break; 601 } 602 f = 1; 603 } 604 ll = pp_end - pp; 605 606 /* If a path-name was not found at the first, try to check 607 * a mtree format(a.k.a form D) ``NetBSD's mtree -D'' creates, 608 * which places the path-name at the last. */ 609 if (f == 0) { 610 const char *pb = p + len - nl; 611 int name_len = 0; 612 int slash; 613 614 /* The form D accepts only a single line for an entry. */ 615 if (pb-2 >= p && 616 pb[-1] == '\\' && (pb[-2] == ' ' || pb[-2] == '\t')) 617 return (-1); 618 if (pb-1 >= p && pb[-1] == '\\') 619 return (-1); 620 621 slash = 0; 622 while (p <= --pb && *pb != ' ' && *pb != '\t') { 623 if (!safe_char[*(const unsigned char *)pb]) 624 return (-1); 625 name_len++; 626 /* The pathname should have a slash in this 627 * format. */ 628 if (*pb == '/') 629 slash = 1; 630 } 631 if (name_len == 0 || slash == 0) 632 return (-1); 633 /* If '/' is placed at the first in this field, this is not 634 * a valid filename. */ 635 if (pb[1] == '/') 636 return (-1); 637 ll = len - nl - name_len; 638 pp = p; 639 *last_is_path = 1; 640 } 641 642 return (bid_keyword_list(pp, ll, 0, *last_is_path)); 643} 644 645#define MAX_BID_ENTRY 3 646 647static int 648mtree_bid(struct archive_read *a, int best_bid) 649{ 650 const char *signature = "#mtree"; 651 const char *p; 652 653 (void)best_bid; /* UNUSED */ 654 655 /* Now let's look at the actual header and see if it matches. */ 656 p = __archive_read_ahead(a, strlen(signature), NULL); 657 if (p == NULL) 658 return (-1); 659 660 if (memcmp(p, signature, strlen(signature)) == 0) 661 return (8 * (int)strlen(signature)); 662 663 /* 664 * There is not a mtree signature. Let's try to detect mtree format. 665 */ 666 return (detect_form(a, NULL)); 667} 668 669static int 670detect_form(struct archive_read *a, int *is_form_d) 671{ 672 const char *p; 673 ssize_t avail, ravail; 674 ssize_t detected_bytes = 0, len, nl; 675 int entry_cnt = 0, multiline = 0; 676 int form_D = 0;/* The archive is generated by `NetBSD mtree -D' 677 * (In this source we call it `form D') . */ 678 679 if (is_form_d != NULL) 680 *is_form_d = 0; 681 p = __archive_read_ahead(a, 1, &avail); 682 if (p == NULL) 683 return (-1); 684 ravail = avail; 685 for (;;) { 686 len = next_line(a, &p, &avail, &ravail, &nl); 687 /* The terminal character of the line should be 688 * a new line character, '\r\n' or '\n'. */ 689 if (len <= 0 || nl == 0) 690 break; 691 if (!multiline) { 692 /* Leading whitespace is never significant, 693 * ignore it. */ 694 while (len > 0 && (*p == ' ' || *p == '\t')) { 695 ++p; 696 --avail; 697 --len; 698 } 699 /* Skip comment or empty line. */ 700 if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') { 701 p += len; 702 avail -= len; 703 continue; 704 } 705 } else { 706 /* A continuance line; the terminal 707 * character of previous line was '\' character. */ 708 if (bid_keyword_list(p, len, 0, 0) <= 0) 709 break; 710 if (multiline == 1) 711 detected_bytes += len; 712 if (p[len-nl-1] != '\\') { 713 if (multiline == 1 && 714 ++entry_cnt >= MAX_BID_ENTRY) 715 break; 716 multiline = 0; 717 } 718 p += len; 719 avail -= len; 720 continue; 721 } 722 if (p[0] != '/') { 723 int last_is_path, keywords; 724 725 keywords = bid_entry(p, len, nl, &last_is_path); 726 if (keywords >= 0) { 727 detected_bytes += len; 728 if (form_D == 0) { 729 if (last_is_path) 730 form_D = 1; 731 else if (keywords > 0) 732 /* This line is not `form D'. */ 733 form_D = -1; 734 } else if (form_D == 1) { 735 if (!last_is_path && keywords > 0) 736 /* This this is not `form D' 737 * and We cannot accept mixed 738 * format. */ 739 break; 740 } 741 if (!last_is_path && p[len-nl-1] == '\\') 742 /* This line continues. */ 743 multiline = 1; 744 else { 745 /* We've got plenty of correct lines 746 * to assume that this file is a mtree 747 * format. */ 748 if (++entry_cnt >= MAX_BID_ENTRY) 749 break; 750 } 751 } else 752 break; 753 } else if (len > 4 && strncmp(p, "/set", 4) == 0) { 754 if (bid_keyword_list(p+4, len-4, 0, 0) <= 0) 755 break; 756 /* This line continues. */ 757 if (p[len-nl-1] == '\\') 758 multiline = 2; 759 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { 760 if (bid_keyword_list(p+6, len-6, 1, 0) <= 0) 761 break; 762 /* This line continues. */ 763 if (p[len-nl-1] == '\\') 764 multiline = 2; 765 } else 766 break; 767 768 /* Test next line. */ 769 p += len; 770 avail -= len; 771 } 772 if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) { 773 if (is_form_d != NULL) { 774 if (form_D == 1) 775 *is_form_d = 1; 776 } 777 return (32); 778 } 779 780 return (0); 781} 782 783/* 784 * The extended mtree format permits multiple lines specifying 785 * attributes for each file. For those entries, only the last line 786 * is actually used. Practically speaking, that means we have 787 * to read the entire mtree file into memory up front. 788 * 789 * The parsing is done in two steps. First, it is decided if a line 790 * changes the global defaults and if it is, processed accordingly. 791 * Otherwise, the options of the line are merged with the current 792 * global options. 793 */ 794static int 795add_option(struct archive_read *a, struct mtree_option **global, 796 const char *value, size_t len) 797{ 798 struct mtree_option *opt; 799 800 if ((opt = malloc(sizeof(*opt))) == NULL) { 801 archive_set_error(&a->archive, errno, "Can't allocate memory"); 802 return (ARCHIVE_FATAL); 803 } 804 if ((opt->value = malloc(len + 1)) == NULL) { 805 free(opt); 806 archive_set_error(&a->archive, errno, "Can't allocate memory"); 807 return (ARCHIVE_FATAL); 808 } 809 memcpy(opt->value, value, len); 810 opt->value[len] = '\0'; 811 opt->next = *global; 812 *global = opt; 813 return (ARCHIVE_OK); 814} 815 816static void 817remove_option(struct mtree_option **global, const char *value, size_t len) 818{ 819 struct mtree_option *iter, *last; 820 821 last = NULL; 822 for (iter = *global; iter != NULL; last = iter, iter = iter->next) { 823 if (strncmp(iter->value, value, len) == 0 && 824 (iter->value[len] == '\0' || 825 iter->value[len] == '=')) 826 break; 827 } 828 if (iter == NULL) 829 return; 830 if (last == NULL) 831 *global = iter->next; 832 else 833 last->next = iter->next; 834 835 free(iter->value); 836 free(iter); 837} 838 839static int 840process_global_set(struct archive_read *a, 841 struct mtree_option **global, const char *line) 842{ 843 const char *next, *eq; 844 size_t len; 845 int r; 846 847 line += 4; 848 for (;;) { 849 next = line + strspn(line, " \t\r\n"); 850 if (*next == '\0') 851 return (ARCHIVE_OK); 852 line = next; 853 next = line + strcspn(line, " \t\r\n"); 854 eq = strchr(line, '='); 855 if (eq > next) 856 len = next - line; 857 else 858 len = eq - line; 859 860 remove_option(global, line, len); 861 r = add_option(a, global, line, next - line); 862 if (r != ARCHIVE_OK) 863 return (r); 864 line = next; 865 } 866} 867 868static int 869process_global_unset(struct archive_read *a, 870 struct mtree_option **global, const char *line) 871{ 872 const char *next; 873 size_t len; 874 875 line += 6; 876 if (strchr(line, '=') != NULL) { 877 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, 878 "/unset shall not contain `='"); 879 return ARCHIVE_FATAL; 880 } 881 882 for (;;) { 883 next = line + strspn(line, " \t\r\n"); 884 if (*next == '\0') 885 return (ARCHIVE_OK); 886 line = next; 887 len = strcspn(line, " \t\r\n"); 888 889 if (len == 3 && strncmp(line, "all", 3) == 0) { 890 free_options(*global); 891 *global = NULL; 892 } else { 893 remove_option(global, line, len); 894 } 895 896 line += len; 897 } 898} 899 900static int 901process_add_entry(struct archive_read *a, struct mtree *mtree, 902 struct mtree_option **global, const char *line, ssize_t line_len, 903 struct mtree_entry **last_entry, int is_form_d) 904{ 905 struct mtree_entry *entry; 906 struct mtree_option *iter; 907 const char *next, *eq, *name, *end; 908 size_t name_len, len; 909 int r, i; 910 911 if ((entry = malloc(sizeof(*entry))) == NULL) { 912 archive_set_error(&a->archive, errno, "Can't allocate memory"); 913 return (ARCHIVE_FATAL); 914 } 915 entry->next = NULL; 916 entry->options = NULL; 917 entry->name = NULL; 918 entry->used = 0; 919 entry->full = 0; 920 921 /* Add this entry to list. */ 922 if (*last_entry == NULL) 923 mtree->entries = entry; 924 else 925 (*last_entry)->next = entry; 926 *last_entry = entry; 927 928 if (is_form_d) { 929 /* Filename is last item on line. */ 930 /* Adjust line_len to trim trailing whitespace */ 931 while (line_len > 0) { 932 char last_character = line[line_len - 1]; 933 if (last_character == '\r' 934 || last_character == '\n' 935 || last_character == '\t' 936 || last_character == ' ') { 937 line_len--; 938 } else { 939 break; 940 } 941 } 942 /* Name starts after the last whitespace separator */ 943 name = line; 944 for (i = 0; i < line_len; i++) { 945 if (line[i] == '\r' 946 || line[i] == '\n' 947 || line[i] == '\t' 948 || line[i] == ' ') { 949 name = line + i + 1; 950 } 951 } 952 name_len = line + line_len - name; 953 end = name; 954 } else { 955 /* Filename is first item on line */ 956 name_len = strcspn(line, " \t\r\n"); 957 name = line; 958 line += name_len; 959 end = line + line_len; 960 } 961 /* name/name_len is the name within the line. */ 962 /* line..end brackets the entire line except the name */ 963 964 if ((entry->name = malloc(name_len + 1)) == NULL) { 965 archive_set_error(&a->archive, errno, "Can't allocate memory"); 966 return (ARCHIVE_FATAL); 967 } 968 969 memcpy(entry->name, name, name_len); 970 entry->name[name_len] = '\0'; 971 parse_escapes(entry->name, entry); 972 973 entry->next_dup = NULL; 974 if (entry->full) { 975 if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) { 976 struct mtree_entry *alt; 977 alt = (struct mtree_entry *)__archive_rb_tree_find_node( 978 &mtree->rbtree, entry->name); 979 while (alt->next_dup) 980 alt = alt->next_dup; 981 alt->next_dup = entry; 982 } 983 } 984 985 for (iter = *global; iter != NULL; iter = iter->next) { 986 r = add_option(a, &entry->options, iter->value, 987 strlen(iter->value)); 988 if (r != ARCHIVE_OK) 989 return (r); 990 } 991 992 for (;;) { 993 next = line + strspn(line, " \t\r\n"); 994 if (*next == '\0') 995 return (ARCHIVE_OK); 996 if (next >= end) 997 return (ARCHIVE_OK); 998 line = next; 999 next = line + strcspn(line, " \t\r\n"); 1000 eq = strchr(line, '='); 1001 if (eq == NULL || eq > next) 1002 len = next - line; 1003 else 1004 len = eq - line; 1005 1006 remove_option(&entry->options, line, len); 1007 r = add_option(a, &entry->options, line, next - line); 1008 if (r != ARCHIVE_OK) 1009 return (r); 1010 line = next; 1011 } 1012} 1013 1014static int 1015read_mtree(struct archive_read *a, struct mtree *mtree) 1016{ 1017 ssize_t len; 1018 uintmax_t counter; 1019 char *p, *s; 1020 struct mtree_option *global; 1021 struct mtree_entry *last_entry; 1022 int r, is_form_d; 1023 1024 mtree->archive_format = ARCHIVE_FORMAT_MTREE; 1025 mtree->archive_format_name = "mtree"; 1026 1027 global = NULL; 1028 last_entry = NULL; 1029 1030 (void)detect_form(a, &is_form_d); 1031 1032 for (counter = 1; ; ++counter) { 1033 r = ARCHIVE_OK; 1034 len = readline(a, mtree, &p, 65536); 1035 if (len == 0) { 1036 mtree->this_entry = mtree->entries; 1037 free_options(global); 1038 return (ARCHIVE_OK); 1039 } 1040 if (len < 0) { 1041 free_options(global); 1042 return ((int)len); 1043 } 1044 /* Leading whitespace is never significant, ignore it. */ 1045 while (*p == ' ' || *p == '\t') { 1046 ++p; 1047 --len; 1048 } 1049 /* Skip content lines and blank lines. */ 1050 if (*p == '#') 1051 continue; 1052 if (*p == '\r' || *p == '\n' || *p == '\0') 1053 continue; 1054 /* Non-printable characters are not allowed */ 1055 for (s = p;s < p + len - 1; s++) { 1056 if (!isprint(*s)) { 1057 r = ARCHIVE_FATAL; 1058 break; 1059 } 1060 } 1061 if (r != ARCHIVE_OK) 1062 break; 1063 if (*p != '/') { 1064 r = process_add_entry(a, mtree, &global, p, len, 1065 &last_entry, is_form_d); 1066 } else if (len > 4 && strncmp(p, "/set", 4) == 0) { 1067 if (p[4] != ' ' && p[4] != '\t') 1068 break; 1069 r = process_global_set(a, &global, p); 1070 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { 1071 if (p[6] != ' ' && p[6] != '\t') 1072 break; 1073 r = process_global_unset(a, &global, p); 1074 } else 1075 break; 1076 1077 if (r != ARCHIVE_OK) { 1078 free_options(global); 1079 return r; 1080 } 1081 } 1082 1083 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1084 "Can't parse line %ju", counter); 1085 free_options(global); 1086 return (ARCHIVE_FATAL); 1087} 1088 1089/* 1090 * Read in the entire mtree file into memory on the first request. 1091 * Then use the next unused file to satisfy each header request. 1092 */ 1093static int 1094read_header(struct archive_read *a, struct archive_entry *entry) 1095{ 1096 struct mtree *mtree; 1097 char *p; 1098 int r, use_next; 1099 1100 mtree = (struct mtree *)(a->format->data); 1101 1102 if (mtree->fd >= 0) { 1103 close(mtree->fd); 1104 mtree->fd = -1; 1105 } 1106 1107 if (mtree->entries == NULL) { 1108 mtree->resolver = archive_entry_linkresolver_new(); 1109 if (mtree->resolver == NULL) 1110 return ARCHIVE_FATAL; 1111 archive_entry_linkresolver_set_strategy(mtree->resolver, 1112 ARCHIVE_FORMAT_MTREE); 1113 r = read_mtree(a, mtree); 1114 if (r != ARCHIVE_OK) 1115 return (r); 1116 } 1117 1118 a->archive.archive_format = mtree->archive_format; 1119 a->archive.archive_format_name = mtree->archive_format_name; 1120 1121 for (;;) { 1122 if (mtree->this_entry == NULL) 1123 return (ARCHIVE_EOF); 1124 if (strcmp(mtree->this_entry->name, "..") == 0) { 1125 mtree->this_entry->used = 1; 1126 if (archive_strlen(&mtree->current_dir) > 0) { 1127 /* Roll back current path. */ 1128 p = mtree->current_dir.s 1129 + mtree->current_dir.length - 1; 1130 while (p >= mtree->current_dir.s && *p != '/') 1131 --p; 1132 if (p >= mtree->current_dir.s) 1133 --p; 1134 mtree->current_dir.length 1135 = p - mtree->current_dir.s + 1; 1136 } 1137 } 1138 if (!mtree->this_entry->used) { 1139 use_next = 0; 1140 r = parse_file(a, entry, mtree, mtree->this_entry, 1141 &use_next); 1142 if (use_next == 0) 1143 return (r); 1144 } 1145 mtree->this_entry = mtree->this_entry->next; 1146 } 1147} 1148 1149/* 1150 * A single file can have multiple lines contribute specifications. 1151 * Parse as many lines as necessary, then pull additional information 1152 * from a backing file on disk as necessary. 1153 */ 1154static int 1155parse_file(struct archive_read *a, struct archive_entry *entry, 1156 struct mtree *mtree, struct mtree_entry *mentry, int *use_next) 1157{ 1158 const char *path; 1159 struct stat st_storage, *st; 1160 struct mtree_entry *mp; 1161 struct archive_entry *sparse_entry; 1162 int r = ARCHIVE_OK, r1, parsed_kws; 1163 1164 mentry->used = 1; 1165 1166 /* Initialize reasonable defaults. */ 1167 archive_entry_set_filetype(entry, AE_IFREG); 1168 archive_entry_set_size(entry, 0); 1169 archive_string_empty(&mtree->contents_name); 1170 1171 /* Parse options from this line. */ 1172 parsed_kws = 0; 1173 r = parse_line(a, entry, mtree, mentry, &parsed_kws); 1174 1175 if (mentry->full) { 1176 archive_entry_copy_pathname(entry, mentry->name); 1177 /* 1178 * "Full" entries are allowed to have multiple lines 1179 * and those lines aren't required to be adjacent. We 1180 * don't support multiple lines for "relative" entries 1181 * nor do we make any attempt to merge data from 1182 * separate "relative" and "full" entries. (Merging 1183 * "relative" and "full" entries would require dealing 1184 * with pathname canonicalization, which is a very 1185 * tricky subject.) 1186 */ 1187 mp = (struct mtree_entry *)__archive_rb_tree_find_node( 1188 &mtree->rbtree, mentry->name); 1189 for (; mp; mp = mp->next_dup) { 1190 if (mp->full && !mp->used) { 1191 /* Later lines override earlier ones. */ 1192 mp->used = 1; 1193 r1 = parse_line(a, entry, mtree, mp, &parsed_kws); 1194 if (r1 < r) 1195 r = r1; 1196 } 1197 } 1198 } else { 1199 /* 1200 * Relative entries require us to construct 1201 * the full path and possibly update the 1202 * current directory. 1203 */ 1204 size_t n = archive_strlen(&mtree->current_dir); 1205 if (n > 0) 1206 archive_strcat(&mtree->current_dir, "/"); 1207 archive_strcat(&mtree->current_dir, mentry->name); 1208 archive_entry_copy_pathname(entry, mtree->current_dir.s); 1209 if (archive_entry_filetype(entry) != AE_IFDIR) 1210 mtree->current_dir.length = n; 1211 } 1212 1213 if (mtree->checkfs) { 1214 /* 1215 * Try to open and stat the file to get the real size 1216 * and other file info. It would be nice to avoid 1217 * this here so that getting a listing of an mtree 1218 * wouldn't require opening every referenced contents 1219 * file. But then we wouldn't know the actual 1220 * contents size, so I don't see a really viable way 1221 * around this. (Also, we may want to someday pull 1222 * other unspecified info from the contents file on 1223 * disk.) 1224 */ 1225 mtree->fd = -1; 1226 if (archive_strlen(&mtree->contents_name) > 0) 1227 path = mtree->contents_name.s; 1228 else 1229 path = archive_entry_pathname(entry); 1230 1231 if (archive_entry_filetype(entry) == AE_IFREG || 1232 archive_entry_filetype(entry) == AE_IFDIR) { 1233 mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC); 1234 __archive_ensure_cloexec_flag(mtree->fd); 1235 if (mtree->fd == -1 && 1236 (errno != ENOENT || 1237 archive_strlen(&mtree->contents_name) > 0)) { 1238 archive_set_error(&a->archive, errno, 1239 "Can't open %s", path); 1240 r = ARCHIVE_WARN; 1241 } 1242 } 1243 1244 st = &st_storage; 1245 if (mtree->fd >= 0) { 1246 if (fstat(mtree->fd, st) == -1) { 1247 archive_set_error(&a->archive, errno, 1248 "Could not fstat %s", path); 1249 r = ARCHIVE_WARN; 1250 /* If we can't stat it, don't keep it open. */ 1251 close(mtree->fd); 1252 mtree->fd = -1; 1253 st = NULL; 1254 } 1255 } else if (lstat(path, st) == -1) { 1256 st = NULL; 1257 } 1258 1259 /* 1260 * Check for a mismatch between the type in the specification 1261 * and the type of the contents object on disk. 1262 */ 1263 if (st != NULL) { 1264 if (((st->st_mode & S_IFMT) == S_IFREG && 1265 archive_entry_filetype(entry) == AE_IFREG) 1266#ifdef S_IFLNK 1267 ||((st->st_mode & S_IFMT) == S_IFLNK && 1268 archive_entry_filetype(entry) == AE_IFLNK) 1269#endif 1270#ifdef S_IFSOCK 1271 ||((st->st_mode & S_IFSOCK) == S_IFSOCK && 1272 archive_entry_filetype(entry) == AE_IFSOCK) 1273#endif 1274#ifdef S_IFCHR 1275 ||((st->st_mode & S_IFMT) == S_IFCHR && 1276 archive_entry_filetype(entry) == AE_IFCHR) 1277#endif 1278#ifdef S_IFBLK 1279 ||((st->st_mode & S_IFMT) == S_IFBLK && 1280 archive_entry_filetype(entry) == AE_IFBLK) 1281#endif 1282 ||((st->st_mode & S_IFMT) == S_IFDIR && 1283 archive_entry_filetype(entry) == AE_IFDIR) 1284#ifdef S_IFIFO 1285 ||((st->st_mode & S_IFMT) == S_IFIFO && 1286 archive_entry_filetype(entry) == AE_IFIFO) 1287#endif 1288 ) { 1289 /* Types match. */ 1290 } else { 1291 /* Types don't match; bail out gracefully. */ 1292 if (mtree->fd >= 0) 1293 close(mtree->fd); 1294 mtree->fd = -1; 1295 if (parsed_kws & MTREE_HAS_OPTIONAL) { 1296 /* It's not an error for an optional 1297 * entry to not match disk. */ 1298 *use_next = 1; 1299 } else if (r == ARCHIVE_OK) { 1300 archive_set_error(&a->archive, 1301 ARCHIVE_ERRNO_MISC, 1302 "mtree specification has different" 1303 " type for %s", 1304 archive_entry_pathname(entry)); 1305 r = ARCHIVE_WARN; 1306 } 1307 return (r); 1308 } 1309 } 1310 1311 /* 1312 * If there is a contents file on disk, pick some of the 1313 * metadata from that file. For most of these, we only 1314 * set it from the contents if it wasn't already parsed 1315 * from the specification. 1316 */ 1317 if (st != NULL) { 1318 if (((parsed_kws & MTREE_HAS_DEVICE) == 0 || 1319 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) && 1320 (archive_entry_filetype(entry) == AE_IFCHR || 1321 archive_entry_filetype(entry) == AE_IFBLK)) 1322 archive_entry_set_rdev(entry, st->st_rdev); 1323 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) 1324 == 0 || 1325 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) 1326 archive_entry_set_gid(entry, st->st_gid); 1327 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) 1328 == 0 || 1329 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) 1330 archive_entry_set_uid(entry, st->st_uid); 1331 if ((parsed_kws & MTREE_HAS_MTIME) == 0 || 1332 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) { 1333#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC 1334 archive_entry_set_mtime(entry, st->st_mtime, 1335 st->st_mtimespec.tv_nsec); 1336#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC 1337 archive_entry_set_mtime(entry, st->st_mtime, 1338 st->st_mtim.tv_nsec); 1339#elif HAVE_STRUCT_STAT_ST_MTIME_N 1340 archive_entry_set_mtime(entry, st->st_mtime, 1341 st->st_mtime_n); 1342#elif HAVE_STRUCT_STAT_ST_UMTIME 1343 archive_entry_set_mtime(entry, st->st_mtime, 1344 st->st_umtime*1000); 1345#elif HAVE_STRUCT_STAT_ST_MTIME_USEC 1346 archive_entry_set_mtime(entry, st->st_mtime, 1347 st->st_mtime_usec*1000); 1348#else 1349 archive_entry_set_mtime(entry, st->st_mtime, 0); 1350#endif 1351 } 1352 if ((parsed_kws & MTREE_HAS_NLINK) == 0 || 1353 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) 1354 archive_entry_set_nlink(entry, st->st_nlink); 1355 if ((parsed_kws & MTREE_HAS_PERM) == 0 || 1356 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) 1357 archive_entry_set_perm(entry, st->st_mode); 1358 if ((parsed_kws & MTREE_HAS_SIZE) == 0 || 1359 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) 1360 archive_entry_set_size(entry, st->st_size); 1361 archive_entry_set_ino(entry, st->st_ino); 1362 archive_entry_set_dev(entry, st->st_dev); 1363 1364 archive_entry_linkify(mtree->resolver, &entry, 1365 &sparse_entry); 1366 } else if (parsed_kws & MTREE_HAS_OPTIONAL) { 1367 /* 1368 * Couldn't open the entry, stat it or the on-disk type 1369 * didn't match. If this entry is optional, just 1370 * ignore it and read the next header entry. 1371 */ 1372 *use_next = 1; 1373 return ARCHIVE_OK; 1374 } 1375 } 1376 1377 mtree->cur_size = archive_entry_size(entry); 1378 mtree->offset = 0; 1379 1380 return r; 1381} 1382 1383/* 1384 * Each line contains a sequence of keywords. 1385 */ 1386static int 1387parse_line(struct archive_read *a, struct archive_entry *entry, 1388 struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws) 1389{ 1390 struct mtree_option *iter; 1391 int r = ARCHIVE_OK, r1; 1392 1393 for (iter = mp->options; iter != NULL; iter = iter->next) { 1394 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws); 1395 if (r1 < r) 1396 r = r1; 1397 } 1398 if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) { 1399 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1400 "Missing type keyword in mtree specification"); 1401 return (ARCHIVE_WARN); 1402 } 1403 return (r); 1404} 1405 1406/* 1407 * Device entries have one of the following forms: 1408 * - raw dev_t 1409 * - format,major,minor[,subdevice] 1410 * When parsing succeeded, `pdev' will contain the appropriate dev_t value. 1411 */ 1412 1413/* strsep() is not in C90, but strcspn() is. */ 1414/* Taken from http://unixpapa.com/incnote/string.html */ 1415static char * 1416la_strsep(char **sp, const char *sep) 1417{ 1418 char *p, *s; 1419 if (sp == NULL || *sp == NULL || **sp == '\0') 1420 return(NULL); 1421 s = *sp; 1422 p = s + strcspn(s, sep); 1423 if (*p != '\0') 1424 *p++ = '\0'; 1425 *sp = p; 1426 return(s); 1427} 1428 1429static int 1430parse_device(dev_t *pdev, struct archive *a, char *val) 1431{ 1432#define MAX_PACK_ARGS 3 1433 unsigned long numbers[MAX_PACK_ARGS]; 1434 char *p, *dev; 1435 int argc; 1436 pack_t *pack; 1437 dev_t result; 1438 const char *error = NULL; 1439 1440 memset(pdev, 0, sizeof(*pdev)); 1441 if ((dev = strchr(val, ',')) != NULL) { 1442 /* 1443 * Device's major/minor are given in a specified format. 1444 * Decode and pack it accordingly. 1445 */ 1446 *dev++ = '\0'; 1447 if ((pack = pack_find(val)) == NULL) { 1448 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, 1449 "Unknown format `%s'", val); 1450 return ARCHIVE_WARN; 1451 } 1452 argc = 0; 1453 while ((p = la_strsep(&dev, ",")) != NULL) { 1454 if (*p == '\0') { 1455 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, 1456 "Missing number"); 1457 return ARCHIVE_WARN; 1458 } 1459 if (argc >= MAX_PACK_ARGS) { 1460 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, 1461 "Too many arguments"); 1462 return ARCHIVE_WARN; 1463 } 1464 numbers[argc++] = (unsigned long)mtree_atol(&p, 0); 1465 } 1466 if (argc < 2) { 1467 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, 1468 "Not enough arguments"); 1469 return ARCHIVE_WARN; 1470 } 1471 result = (*pack)(argc, numbers, &error); 1472 if (error != NULL) { 1473 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, 1474 "%s", error); 1475 return ARCHIVE_WARN; 1476 } 1477 } else { 1478 /* file system raw value. */ 1479 result = (dev_t)mtree_atol(&val, 0); 1480 } 1481 *pdev = result; 1482 return ARCHIVE_OK; 1483#undef MAX_PACK_ARGS 1484} 1485 1486static int 1487parse_hex_nibble(char c) 1488{ 1489 if (c >= '0' && c <= '9') 1490 return c - '0'; 1491 if (c >= 'a' && c <= 'f') 1492 return 10 + c - 'a'; 1493#if 0 1494 /* XXX: Is uppercase something we should support? */ 1495 if (c >= 'A' && c <= 'F') 1496 return 10 + c - 'A'; 1497#endif 1498 1499 return -1; 1500} 1501 1502static int 1503parse_digest(struct archive_read *a, struct archive_entry *entry, 1504 const char *digest, int type) 1505{ 1506 unsigned char digest_buf[64]; 1507 int high, low; 1508 size_t i, j, len; 1509 1510 switch (type) { 1511 case ARCHIVE_ENTRY_DIGEST_MD5: 1512 len = sizeof(entry->digest.md5); 1513 break; 1514 case ARCHIVE_ENTRY_DIGEST_RMD160: 1515 len = sizeof(entry->digest.rmd160); 1516 break; 1517 case ARCHIVE_ENTRY_DIGEST_SHA1: 1518 len = sizeof(entry->digest.sha1); 1519 break; 1520 case ARCHIVE_ENTRY_DIGEST_SHA256: 1521 len = sizeof(entry->digest.sha256); 1522 break; 1523 case ARCHIVE_ENTRY_DIGEST_SHA384: 1524 len = sizeof(entry->digest.sha384); 1525 break; 1526 case ARCHIVE_ENTRY_DIGEST_SHA512: 1527 len = sizeof(entry->digest.sha512); 1528 break; 1529 default: 1530 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1531 "Internal error: Unknown digest type"); 1532 return ARCHIVE_FATAL; 1533 } 1534 1535 if (len > sizeof(digest_buf)) { 1536 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, 1537 "Internal error: Digest storage too large"); 1538 return ARCHIVE_FATAL; 1539 } 1540 1541 len *= 2; 1542 1543 if (strnlen(digest, len+1) != len) { 1544 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1545 "incorrect digest length, ignoring"); 1546 return ARCHIVE_WARN; 1547 } 1548 1549 for (i = 0, j = 0; i < len; i += 2, j++) { 1550 high = parse_hex_nibble(digest[i]); 1551 low = parse_hex_nibble(digest[i+1]); 1552 if (high == -1 || low == -1) { 1553 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1554 "invalid digest data, ignoring"); 1555 return ARCHIVE_WARN; 1556 } 1557 1558 digest_buf[j] = high << 4 | low; 1559 } 1560 1561 return archive_entry_set_digest(entry, type, digest_buf); 1562} 1563 1564/* 1565 * Parse a single keyword and its value. 1566 */ 1567static int 1568parse_keyword(struct archive_read *a, struct mtree *mtree, 1569 struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws) 1570{ 1571 char *val, *key; 1572 1573 key = opt->value; 1574 1575 if (*key == '\0') 1576 return (ARCHIVE_OK); 1577 1578 if (strcmp(key, "nochange") == 0) { 1579 *parsed_kws |= MTREE_HAS_NOCHANGE; 1580 return (ARCHIVE_OK); 1581 } 1582 if (strcmp(key, "optional") == 0) { 1583 *parsed_kws |= MTREE_HAS_OPTIONAL; 1584 return (ARCHIVE_OK); 1585 } 1586 if (strcmp(key, "ignore") == 0) { 1587 /* 1588 * The mtree processing is not recursive, so 1589 * recursion will only happen for explicitly listed 1590 * entries. 1591 */ 1592 return (ARCHIVE_OK); 1593 } 1594 1595 val = strchr(key, '='); 1596 if (val == NULL) { 1597 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1598 "Malformed attribute \"%s\" (%d)", key, key[0]); 1599 return (ARCHIVE_WARN); 1600 } 1601 1602 *val = '\0'; 1603 ++val; 1604 1605 switch (key[0]) { 1606 case 'c': 1607 if (strcmp(key, "content") == 0 1608 || strcmp(key, "contents") == 0) { 1609 parse_escapes(val, NULL); 1610 archive_strcpy(&mtree->contents_name, val); 1611 break; 1612 } 1613 if (strcmp(key, "cksum") == 0) 1614 break; 1615 __LA_FALLTHROUGH; 1616 case 'd': 1617 if (strcmp(key, "device") == 0) { 1618 /* stat(2) st_rdev field, e.g. the major/minor IDs 1619 * of a char/block special file */ 1620 int r; 1621 dev_t dev; 1622 1623 *parsed_kws |= MTREE_HAS_DEVICE; 1624 r = parse_device(&dev, &a->archive, val); 1625 if (r == ARCHIVE_OK) 1626 archive_entry_set_rdev(entry, dev); 1627 return r; 1628 } 1629 __LA_FALLTHROUGH; 1630 case 'f': 1631 if (strcmp(key, "flags") == 0) { 1632 *parsed_kws |= MTREE_HAS_FFLAGS; 1633 archive_entry_copy_fflags_text(entry, val); 1634 break; 1635 } 1636 __LA_FALLTHROUGH; 1637 case 'g': 1638 if (strcmp(key, "gid") == 0) { 1639 *parsed_kws |= MTREE_HAS_GID; 1640 archive_entry_set_gid(entry, mtree_atol(&val, 10)); 1641 break; 1642 } 1643 if (strcmp(key, "gname") == 0) { 1644 *parsed_kws |= MTREE_HAS_GNAME; 1645 archive_entry_copy_gname(entry, val); 1646 break; 1647 } 1648 __LA_FALLTHROUGH; 1649 case 'i': 1650 if (strcmp(key, "inode") == 0) { 1651 archive_entry_set_ino(entry, mtree_atol(&val, 10)); 1652 break; 1653 } 1654 __LA_FALLTHROUGH; 1655 case 'l': 1656 if (strcmp(key, "link") == 0) { 1657 archive_entry_copy_symlink(entry, val); 1658 break; 1659 } 1660 __LA_FALLTHROUGH; 1661 case 'm': 1662 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) { 1663 return parse_digest(a, entry, val, 1664 ARCHIVE_ENTRY_DIGEST_MD5); 1665 } 1666 if (strcmp(key, "mode") == 0) { 1667 if (val[0] >= '0' && val[0] <= '7') { 1668 *parsed_kws |= MTREE_HAS_PERM; 1669 archive_entry_set_perm(entry, 1670 (mode_t)mtree_atol(&val, 8)); 1671 } else { 1672 archive_set_error(&a->archive, 1673 ARCHIVE_ERRNO_FILE_FORMAT, 1674 "Symbolic or non-octal mode \"%s\" unsupported", val); 1675 return ARCHIVE_WARN; 1676 } 1677 break; 1678 } 1679 __LA_FALLTHROUGH; 1680 case 'n': 1681 if (strcmp(key, "nlink") == 0) { 1682 *parsed_kws |= MTREE_HAS_NLINK; 1683 archive_entry_set_nlink(entry, 1684 (unsigned int)mtree_atol(&val, 10)); 1685 break; 1686 } 1687 __LA_FALLTHROUGH; 1688 case 'r': 1689 if (strcmp(key, "resdevice") == 0) { 1690 /* stat(2) st_dev field, e.g. the device ID where the 1691 * inode resides */ 1692 int r; 1693 dev_t dev; 1694 1695 r = parse_device(&dev, &a->archive, val); 1696 if (r == ARCHIVE_OK) 1697 archive_entry_set_dev(entry, dev); 1698 return r; 1699 } 1700 if (strcmp(key, "rmd160") == 0 || 1701 strcmp(key, "rmd160digest") == 0) { 1702 return parse_digest(a, entry, val, 1703 ARCHIVE_ENTRY_DIGEST_RMD160); 1704 } 1705 __LA_FALLTHROUGH; 1706 case 's': 1707 if (strcmp(key, "sha1") == 0 || 1708 strcmp(key, "sha1digest") == 0) { 1709 return parse_digest(a, entry, val, 1710 ARCHIVE_ENTRY_DIGEST_SHA1); 1711 } 1712 if (strcmp(key, "sha256") == 0 || 1713 strcmp(key, "sha256digest") == 0) { 1714 return parse_digest(a, entry, val, 1715 ARCHIVE_ENTRY_DIGEST_SHA256); 1716 } 1717 if (strcmp(key, "sha384") == 0 || 1718 strcmp(key, "sha384digest") == 0) { 1719 return parse_digest(a, entry, val, 1720 ARCHIVE_ENTRY_DIGEST_SHA384); 1721 } 1722 if (strcmp(key, "sha512") == 0 || 1723 strcmp(key, "sha512digest") == 0) { 1724 return parse_digest(a, entry, val, 1725 ARCHIVE_ENTRY_DIGEST_SHA512); 1726 } 1727 if (strcmp(key, "size") == 0) { 1728 archive_entry_set_size(entry, mtree_atol(&val, 10)); 1729 break; 1730 } 1731 __LA_FALLTHROUGH; 1732 case 't': 1733 if (strcmp(key, "tags") == 0) { 1734 /* 1735 * Comma delimited list of tags. 1736 * Ignore the tags for now, but the interface 1737 * should be extended to allow inclusion/exclusion. 1738 */ 1739 break; 1740 } 1741 if (strcmp(key, "time") == 0) { 1742 int64_t m; 1743 int64_t my_time_t_max = get_time_t_max(); 1744 int64_t my_time_t_min = get_time_t_min(); 1745 long ns = 0; 1746 1747 *parsed_kws |= MTREE_HAS_MTIME; 1748 m = mtree_atol(&val, 10); 1749 /* Replicate an old mtree bug: 1750 * 123456789.1 represents 123456789 1751 * seconds and 1 nanosecond. */ 1752 if (*val == '.') { 1753 ++val; 1754 ns = (long)mtree_atol(&val, 10); 1755 if (ns < 0) 1756 ns = 0; 1757 else if (ns > 999999999) 1758 ns = 999999999; 1759 } 1760 if (m > my_time_t_max) 1761 m = my_time_t_max; 1762 else if (m < my_time_t_min) 1763 m = my_time_t_min; 1764 archive_entry_set_mtime(entry, (time_t)m, ns); 1765 break; 1766 } 1767 if (strcmp(key, "type") == 0) { 1768 switch (val[0]) { 1769 case 'b': 1770 if (strcmp(val, "block") == 0) { 1771 archive_entry_set_filetype(entry, AE_IFBLK); 1772 break; 1773 } 1774 __LA_FALLTHROUGH; 1775 case 'c': 1776 if (strcmp(val, "char") == 0) { 1777 archive_entry_set_filetype(entry, 1778 AE_IFCHR); 1779 break; 1780 } 1781 __LA_FALLTHROUGH; 1782 case 'd': 1783 if (strcmp(val, "dir") == 0) { 1784 archive_entry_set_filetype(entry, 1785 AE_IFDIR); 1786 break; 1787 } 1788 __LA_FALLTHROUGH; 1789 case 'f': 1790 if (strcmp(val, "fifo") == 0) { 1791 archive_entry_set_filetype(entry, 1792 AE_IFIFO); 1793 break; 1794 } 1795 if (strcmp(val, "file") == 0) { 1796 archive_entry_set_filetype(entry, 1797 AE_IFREG); 1798 break; 1799 } 1800 __LA_FALLTHROUGH; 1801 case 'l': 1802 if (strcmp(val, "link") == 0) { 1803 archive_entry_set_filetype(entry, 1804 AE_IFLNK); 1805 break; 1806 } 1807 __LA_FALLTHROUGH; 1808 default: 1809 archive_set_error(&a->archive, 1810 ARCHIVE_ERRNO_FILE_FORMAT, 1811 "Unrecognized file type \"%s\"; " 1812 "assuming \"file\"", val); 1813 archive_entry_set_filetype(entry, AE_IFREG); 1814 return (ARCHIVE_WARN); 1815 } 1816 *parsed_kws |= MTREE_HAS_TYPE; 1817 break; 1818 } 1819 __LA_FALLTHROUGH; 1820 case 'u': 1821 if (strcmp(key, "uid") == 0) { 1822 *parsed_kws |= MTREE_HAS_UID; 1823 archive_entry_set_uid(entry, mtree_atol(&val, 10)); 1824 break; 1825 } 1826 if (strcmp(key, "uname") == 0) { 1827 *parsed_kws |= MTREE_HAS_UNAME; 1828 archive_entry_copy_uname(entry, val); 1829 break; 1830 } 1831 __LA_FALLTHROUGH; 1832 default: 1833 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, 1834 "Unrecognized key %s=%s", key, val); 1835 return (ARCHIVE_WARN); 1836 } 1837 return (ARCHIVE_OK); 1838} 1839 1840static int 1841read_data(struct archive_read *a, const void **buff, size_t *size, 1842 int64_t *offset) 1843{ 1844 size_t bytes_to_read; 1845 ssize_t bytes_read; 1846 struct mtree *mtree; 1847 1848 mtree = (struct mtree *)(a->format->data); 1849 if (mtree->fd < 0) { 1850 *buff = NULL; 1851 *offset = 0; 1852 *size = 0; 1853 return (ARCHIVE_EOF); 1854 } 1855 if (mtree->buff == NULL) { 1856 mtree->buffsize = 64 * 1024; 1857 mtree->buff = malloc(mtree->buffsize); 1858 if (mtree->buff == NULL) { 1859 archive_set_error(&a->archive, ENOMEM, 1860 "Can't allocate memory"); 1861 return (ARCHIVE_FATAL); 1862 } 1863 } 1864 1865 *buff = mtree->buff; 1866 *offset = mtree->offset; 1867 if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset) 1868 bytes_to_read = (size_t)(mtree->cur_size - mtree->offset); 1869 else 1870 bytes_to_read = mtree->buffsize; 1871 bytes_read = read(mtree->fd, mtree->buff, bytes_to_read); 1872 if (bytes_read < 0) { 1873 archive_set_error(&a->archive, errno, "Can't read"); 1874 return (ARCHIVE_WARN); 1875 } 1876 if (bytes_read == 0) { 1877 *size = 0; 1878 return (ARCHIVE_EOF); 1879 } 1880 mtree->offset += bytes_read; 1881 *size = bytes_read; 1882 return (ARCHIVE_OK); 1883} 1884 1885/* Skip does nothing except possibly close the contents file. */ 1886static int 1887skip(struct archive_read *a) 1888{ 1889 struct mtree *mtree; 1890 1891 mtree = (struct mtree *)(a->format->data); 1892 if (mtree->fd >= 0) { 1893 close(mtree->fd); 1894 mtree->fd = -1; 1895 } 1896 return (ARCHIVE_OK); 1897} 1898 1899/* 1900 * Since parsing backslash sequences always makes strings shorter, 1901 * we can always do this conversion in-place. 1902 */ 1903static void 1904parse_escapes(char *src, struct mtree_entry *mentry) 1905{ 1906 char *dest = src; 1907 char c; 1908 1909 if (mentry != NULL && strcmp(src, ".") == 0) 1910 mentry->full = 1; 1911 1912 while (*src != '\0') { 1913 c = *src++; 1914 if (c == '/' && mentry != NULL) 1915 mentry->full = 1; 1916 if (c == '\\') { 1917 switch (src[0]) { 1918 case '0': 1919 if (src[1] < '0' || src[1] > '7') { 1920 c = 0; 1921 ++src; 1922 break; 1923 } 1924 /* FALLTHROUGH */ 1925 case '1': 1926 case '2': 1927 case '3': 1928 if (src[1] >= '0' && src[1] <= '7' && 1929 src[2] >= '0' && src[2] <= '7') { 1930 c = (src[0] - '0') << 6; 1931 c |= (src[1] - '0') << 3; 1932 c |= (src[2] - '0'); 1933 src += 3; 1934 } 1935 break; 1936 case 'a': 1937 c = '\a'; 1938 ++src; 1939 break; 1940 case 'b': 1941 c = '\b'; 1942 ++src; 1943 break; 1944 case 'f': 1945 c = '\f'; 1946 ++src; 1947 break; 1948 case 'n': 1949 c = '\n'; 1950 ++src; 1951 break; 1952 case 'r': 1953 c = '\r'; 1954 ++src; 1955 break; 1956 case 's': 1957 c = ' '; 1958 ++src; 1959 break; 1960 case 't': 1961 c = '\t'; 1962 ++src; 1963 break; 1964 case 'v': 1965 c = '\v'; 1966 ++src; 1967 break; 1968 case '\\': 1969 c = '\\'; 1970 ++src; 1971 break; 1972 } 1973 } 1974 *dest++ = c; 1975 } 1976 *dest = '\0'; 1977} 1978 1979/* Parse a hex digit. */ 1980static int 1981parsedigit(char c) 1982{ 1983 if (c >= '0' && c <= '9') 1984 return c - '0'; 1985 else if (c >= 'a' && c <= 'f') 1986 return c - 'a'; 1987 else if (c >= 'A' && c <= 'F') 1988 return c - 'A'; 1989 else 1990 return -1; 1991} 1992 1993/* 1994 * Note that this implementation does not (and should not!) obey 1995 * locale settings; you cannot simply substitute strtol here, since 1996 * it does obey locale. 1997 */ 1998static int64_t 1999mtree_atol(char **p, int base) 2000{ 2001 int64_t l, limit; 2002 int digit, last_digit_limit; 2003 2004 if (base == 0) { 2005 if (**p != '0') 2006 base = 10; 2007 else if ((*p)[1] == 'x' || (*p)[1] == 'X') { 2008 *p += 2; 2009 base = 16; 2010 } else { 2011 base = 8; 2012 } 2013 } 2014 2015 if (**p == '-') { 2016 limit = INT64_MIN / base; 2017 last_digit_limit = INT64_MIN % base; 2018 ++(*p); 2019 2020 l = 0; 2021 digit = parsedigit(**p); 2022 while (digit >= 0 && digit < base) { 2023 if (l < limit || (l == limit && digit > last_digit_limit)) 2024 return INT64_MIN; 2025 l = (l * base) - digit; 2026 digit = parsedigit(*++(*p)); 2027 } 2028 return l; 2029 } else { 2030 limit = INT64_MAX / base; 2031 last_digit_limit = INT64_MAX % base; 2032 2033 l = 0; 2034 digit = parsedigit(**p); 2035 while (digit >= 0 && digit < base) { 2036 if (l > limit || (l == limit && digit > last_digit_limit)) 2037 return INT64_MAX; 2038 l = (l * base) + digit; 2039 digit = parsedigit(*++(*p)); 2040 } 2041 return l; 2042 } 2043} 2044 2045/* 2046 * Returns length of line (including trailing newline) 2047 * or negative on error. 'start' argument is updated to 2048 * point to first character of line. 2049 */ 2050static ssize_t 2051readline(struct archive_read *a, struct mtree *mtree, char **start, 2052 ssize_t limit) 2053{ 2054 ssize_t bytes_read; 2055 ssize_t total_size = 0; 2056 ssize_t find_off = 0; 2057 const void *t; 2058 void *nl; 2059 char *u; 2060 2061 /* Accumulate line in a line buffer. */ 2062 for (;;) { 2063 /* Read some more. */ 2064 t = __archive_read_ahead(a, 1, &bytes_read); 2065 if (t == NULL) 2066 return (0); 2067 if (bytes_read < 0) 2068 return (ARCHIVE_FATAL); 2069 nl = memchr(t, '\n', bytes_read); 2070 /* If we found '\n', trim the read to end exactly there. */ 2071 if (nl != NULL) { 2072 bytes_read = ((const char *)nl) - ((const char *)t) + 1; 2073 } 2074 if (total_size + bytes_read + 1 > limit) { 2075 archive_set_error(&a->archive, 2076 ARCHIVE_ERRNO_FILE_FORMAT, 2077 "Line too long"); 2078 return (ARCHIVE_FATAL); 2079 } 2080 if (archive_string_ensure(&mtree->line, 2081 total_size + bytes_read + 1) == NULL) { 2082 archive_set_error(&a->archive, ENOMEM, 2083 "Can't allocate working buffer"); 2084 return (ARCHIVE_FATAL); 2085 } 2086 /* Append new bytes to string. */ 2087 memcpy(mtree->line.s + total_size, t, bytes_read); 2088 __archive_read_consume(a, bytes_read); 2089 total_size += bytes_read; 2090 mtree->line.s[total_size] = '\0'; 2091 2092 for (u = mtree->line.s + find_off; *u; ++u) { 2093 if (u[0] == '\n') { 2094 /* Ends with unescaped newline. */ 2095 *start = mtree->line.s; 2096 return total_size; 2097 } else if (u[0] == '#') { 2098 /* Ends with comment sequence #...\n */ 2099 if (nl == NULL) { 2100 /* But we've not found the \n yet */ 2101 break; 2102 } 2103 } else if (u[0] == '\\') { 2104 if (u[1] == '\n') { 2105 /* Trim escaped newline. */ 2106 total_size -= 2; 2107 mtree->line.s[total_size] = '\0'; 2108 break; 2109 } else if (u[1] != '\0') { 2110 /* Skip the two-char escape sequence */ 2111 ++u; 2112 } 2113 } 2114 } 2115 find_off = u - mtree->line.s; 2116 } 2117} 2118