release/src/router/busybox/archival/libarchive/get_header_tar.c

   1 /* vi: set sw=4 ts=4: */
   2 /* Licensed under GPLv2 or later, see file LICENSE in this source tree.
   3  *
   4  * FIXME:
   5  *    In privileged mode if uname and gname map to a uid and gid then use the
   6  *    mapped value instead of the uid/gid values in tar header
   7  *
   8  * References:
   9  *    GNU tar and star man pages,
  10  *    Opengroup's ustar interchange format,
  11  *    http://www.opengroup.org/onlinepubs/007904975/utilities/pax.html
  12  */
  13
  14 #include "libbb.h"
  15 #include "bb_archive.h"
  16
  17 typedef uint32_t aliased_uint32_t FIX_ALIASING;
  18 typedef off_t    aliased_off_t    FIX_ALIASING;
  19
  20
  21 const char* FAST_FUNC strip_unsafe_prefix(const char *str)
  22 {
  23         const char *cp = str;
  24         while (1) {
  25                 char *cp2;
  26                 if (*cp == '/') {
  27                         cp++;
  28                         continue;
  29                 }
  30                 if (strncmp(cp, "/../"+1, 3) == 0) {
  31                         cp += 3;
  32                         continue;
  33                 }
  34                 cp2 = strstr(cp, "/../");
  35                 if (!cp2)
  36                         break;
  37                 cp = cp2 + 4;
  38         }
  39         if (cp != str) {
  40                 static smallint warned = 0;
  41                 if (!warned) {
  42                         warned = 1;
  43                         bb_error_msg("removing leading '%.*s' from member names",
  44                                 (int)(cp - str), str);
  45                 }
  46         }
  47         return cp;
  48 }
  49
  50 /* NB: _DESTROYS_ str[len] character! */
  51 static unsigned long long getOctal(char *str, int len)
  52 {
  53         unsigned long long v;
  54         char *end;
  55         /* NB: leading spaces are allowed. Using strtoull to handle that.
  56          * The downside is that we accept e.g. "-123" too :(
  57          */
  58         str[len] = '\0';
  59         v = strtoull(str, &end, 8);
  60         /* std: "Each numeric field is terminated by one or more
  61          * <space> or NUL characters". We must support ' '! */
  62         if (*end != '\0' && *end != ' ') {
  63                 int8_t first = str[0];
  64                 if (!(first & 0x80))
  65                         bb_error_msg_and_die("corrupted octal value in tar header");
  66                 /*
  67                  * GNU tar uses "base-256 encoding" for very large numbers.
  68                  * Encoding is binary, with highest bit always set as a marker
  69                  * and sign in next-highest bit:
  70                  * 80 00 .. 00 - zero
  71                  * bf ff .. ff - largest positive number
  72                  * ff ff .. ff - minus 1
  73                  * c0 00 .. 00 - smallest negative number
  74                  *
  75                  * Example of tar file with 8914993153 (0x213600001) byte file.
  76                  * Field starts at offset 7c:
  77                  * 00070  30 30 30 00 30 30 30 30  30 30 30 00 80 00 00 00  |000.0000000.....|
  78                  * 00080  00 00 00 02 13 60 00 01  31 31 31 32 30 33 33 36  |.....`..11120336|
  79                  *
  80                  * NB: tarballs with NEGATIVE unix times encoded that way were seen!
  81                  */
  82                 /* Sign-extend 7bit 'first' to 64bit 'v' (that is, using 6th bit as sign): */
  83                 first <<= 1;
  84                 first >>= 1; /* now 7th bit = 6th bit */
  85                 v = first;   /* sign-extend 8 bits to 64 */
  86                 while (--len != 0)
  87                         v = (v << 8) + (uint8_t) *++str;
  88         }
  89         return v;
  90 }
  91 #define GET_OCTAL(a) getOctal((a), sizeof(a))
  92
  93 /* "global" is 0 or 1 */
  94 static void process_pax_hdr(archive_handle_t *archive_handle, unsigned sz, int global)
  95 {
  96         char *buf, *p;
  97         unsigned blk_sz;
  98
  99         blk_sz = (sz + 511) & (~511);
 100         p = buf = xmalloc(blk_sz + 1);
 101         xread(archive_handle->src_fd, buf, blk_sz);
 102         archive_handle->offset += blk_sz;
 103
 104         /* prevent bb_strtou from running off the buffer */
 105         buf[sz] = '\0';
 106
 107         while (sz != 0) {
 108                 char *end, *value;
 109                 unsigned len;
 110
 111                 /* Every record has this format: "LEN NAME=VALUE\n" */
 112                 len = bb_strtou(p, &end, 10);
 113                 /* expect errno to be EINVAL, because the character
 114                  * following the digits should be a space
 115                  */
 116                 p += len;
 117                 sz -= len;
 118                 if (
 119                 /** (int)sz < 0 - not good enough for huge malicious VALUE of 2^32-1 */
 120                     (int)(sz|len) < 0 /* this works */
 121                  || len == 0
 122                  || errno != EINVAL
 123                  || *end != ' '
 124                 ) {
 125                         bb_error_msg("malformed extended header, skipped");
 126                         // More verbose version:
 127                         //bb_error_msg("malformed extended header at %"OFF_FMT"d, skipped",
 128                         //              archive_handle->offset - (sz + len));
 129                         break;
 130                 }
 131                 /* overwrite the terminating newline with NUL
 132                  * (we do not bother to check that it *was* a newline)
 133                  */
 134                 p[-1] = '\0';
 135                 value = end + 1;
 136
 137 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 138                 if (!global && strncmp(value, "path=", sizeof("path=") - 1) == 0) {
 139                         value += sizeof("path=") - 1;
 140                         free(archive_handle->tar__longname);
 141                         archive_handle->tar__longname = xstrdup(value);
 142                         continue;
 143                 }
 144 #endif
 145
 146 #if ENABLE_FEATURE_TAR_SELINUX
 147                 /* Scan for SELinux contexts, via "RHT.security.selinux" keyword.
 148                  * This is what Red Hat's patched version of tar uses.
 149                  */
 150 # define SELINUX_CONTEXT_KEYWORD "RHT.security.selinux"
 151                 if (strncmp(value, SELINUX_CONTEXT_KEYWORD"=", sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1) == 0) {
 152                         value += sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1;
 153                         free(archive_handle->tar__sctx[global]);
 154                         archive_handle->tar__sctx[global] = xstrdup(value);
 155                         continue;
 156                 }
 157 #endif
 158         }
 159
 160         free(buf);
 161 }
 162
 163 char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 164 {
 165         file_header_t *file_header = archive_handle->file_header;
 166         struct tar_header_t tar;
 167         char *cp;
 168         int i, sum_u, sum;
 169 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 170         int sum_s;
 171 #endif
 172         int parse_names;
 173
 174         /* Our "private data" */
 175 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 176 # define p_longname (archive_handle->tar__longname)
 177 # define p_linkname (archive_handle->tar__linkname)
 178 #else
 179 # define p_longname 0
 180 # define p_linkname 0
 181 #endif
 182
 183 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS || ENABLE_FEATURE_TAR_SELINUX
 184  again:
 185 #endif
 186         /* Align header */
 187         data_align(archive_handle, 512);
 188
 189  again_after_align:
 190
 191 #if ENABLE_DESKTOP || ENABLE_FEATURE_TAR_AUTODETECT
 192         /* to prevent misdetection of bz2 sig */
 193         *(aliased_uint32_t*)&tar = 0;
 194         i = full_read(archive_handle->src_fd, &tar, 512);
 195         /* If GNU tar sees EOF in above read, it says:
 196          * "tar: A lone zero block at N", where N = kilobyte
 197          * where EOF was met (not EOF block, actual EOF!),
 198          * and exits with EXIT_SUCCESS.
 199          * We will mimic exit(EXIT_SUCCESS), although we will not mimic
 200          * the message and we don't check whether we indeed
 201          * saw zero block directly before this. */
 202         if (i == 0) {
 203                 bb_error_msg("short read");
 204                 /* this merely signals end of archive, not exit(1): */
 205                 return EXIT_FAILURE;
 206         }
 207         if (i != 512) {
 208                 IF_FEATURE_TAR_AUTODETECT(goto autodetect;)
 209                 bb_error_msg_and_die("short read");
 210         }
 211
 212 #else
 213         i = 512;
 214         xread(archive_handle->src_fd, &tar, i);
 215 #endif
 216         archive_handle->offset += i;
 217
 218         /* If there is no filename its an empty header */
 219         if (tar.name[0] == 0 && tar.prefix[0] == 0) {
 220                 if (archive_handle->tar__end) {
 221                         /* Second consecutive empty header - end of archive.
 222                          * Read until the end to empty the pipe from gz or bz2
 223                          */
 224                         while (full_read(archive_handle->src_fd, &tar, 512) == 512)
 225                                 continue;
 226                         return EXIT_FAILURE; /* "end of archive" */
 227                 }
 228                 archive_handle->tar__end = 1;
 229                 return EXIT_SUCCESS; /* "decoded one header" */
 230         }
 231         archive_handle->tar__end = 0;
 232
 233         /* Check header has valid magic, "ustar" is for the proper tar,
 234          * five NULs are for the old tar format  */
 235         if (strncmp(tar.magic, "ustar", 5) != 0
 236          && (!ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 237              || memcmp(tar.magic, "\0\0\0\0", 5) != 0)
 238         ) {
 239 #if ENABLE_FEATURE_TAR_AUTODETECT
 240  autodetect:
 241                 /* Two different causes for lseek() != 0:
 242                  * unseekable fd (would like to support that too, but...),
 243                  * or not first block (false positive, it's not .gz/.bz2!) */
 244                 if (lseek(archive_handle->src_fd, -i, SEEK_CUR) != 0)
 245                         goto err;
 246                 if (setup_unzip_on_fd(archive_handle->src_fd, /*fail_if_not_compressed:*/ 0) != 0)
 247  err:
 248                         bb_error_msg_and_die("invalid tar magic");
 249                 archive_handle->offset = 0;
 250                 goto again_after_align;
 251 #endif
 252                 bb_error_msg_and_die("invalid tar magic");
 253         }
 254
 255         /* Do checksum on headers.
 256          * POSIX says that checksum is done on unsigned bytes, but
 257          * Sun and HP-UX gets it wrong... more details in
 258          * GNU tar source. */
 259 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 260         sum_s = ' ' * sizeof(tar.chksum);
 261 #endif
 262         sum_u = ' ' * sizeof(tar.chksum);
 263         for (i = 0; i < 148; i++) {
 264                 sum_u += ((unsigned char*)&tar)[i];
 265 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 266                 sum_s += ((signed char*)&tar)[i];
 267 #endif
 268         }
 269         for (i = 156; i < 512; i++) {
 270                 sum_u += ((unsigned char*)&tar)[i];
 271 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 272                 sum_s += ((signed char*)&tar)[i];
 273 #endif
 274         }
 275         /* This field does not need special treatment (getOctal) */
 276         {
 277                 char *endp; /* gcc likes temp var for &endp */
 278                 sum = strtoul(tar.chksum, &endp, 8);
 279                 if ((*endp != '\0' && *endp != ' ')
 280                  || (sum_u != sum IF_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum))
 281                 ) {
 282                         bb_error_msg_and_die("invalid tar header checksum");
 283                 }
 284         }
 285         /* don't use xstrtoul, tar.chksum may have leading spaces */
 286         sum = strtoul(tar.chksum, NULL, 8);
 287         if (sum_u != sum IF_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum)) {
 288                 bb_error_msg_and_die("invalid tar header checksum");
 289         }
 290
 291         /* 0 is reserved for high perf file, treat as normal file */
 292         if (!tar.typeflag) tar.typeflag = '0';
 293         parse_names = (tar.typeflag >= '0' && tar.typeflag <= '7');
 294
 295         /* getOctal trashes subsequent field, therefore we call it
 296          * on fields in reverse order */
 297         if (tar.devmajor[0]) {
 298                 char t = tar.prefix[0];
 299                 /* we trash prefix[0] here, but we DO need it later! */
 300                 unsigned minor = GET_OCTAL(tar.devminor);
 301                 unsigned major = GET_OCTAL(tar.devmajor);
 302                 file_header->device = makedev(major, minor);
 303                 tar.prefix[0] = t;
 304         }
 305         file_header->link_target = NULL;
 306         if (!p_linkname && parse_names && tar.linkname[0]) {
 307                 file_header->link_target = xstrndup(tar.linkname, sizeof(tar.linkname));
 308                 /* FIXME: what if we have non-link object with link_target? */
 309                 /* Will link_target be free()ed? */
 310         }
 311 #if ENABLE_FEATURE_TAR_UNAME_GNAME
 312         file_header->tar__uname = tar.uname[0] ? xstrndup(tar.uname, sizeof(tar.uname)) : NULL;
 313         file_header->tar__gname = tar.gname[0] ? xstrndup(tar.gname, sizeof(tar.gname)) : NULL;
 314 #endif
 315         file_header->mtime = GET_OCTAL(tar.mtime);
 316         file_header->size = GET_OCTAL(tar.size);
 317         file_header->gid = GET_OCTAL(tar.gid);
 318         file_header->uid = GET_OCTAL(tar.uid);
 319         /* Set bits 0-11 of the files mode */
 320         file_header->mode = 07777 & GET_OCTAL(tar.mode);
 321
 322         file_header->name = NULL;
 323         if (!p_longname && parse_names) {
 324                 /* we trash mode[0] here, it's ok */
 325                 //tar.name[sizeof(tar.name)] = '\0'; - gcc 4.3.0 would complain
 326                 tar.mode[0] = '\0';
 327                 if (tar.prefix[0]) {
 328                         /* and padding[0] */
 329                         //tar.prefix[sizeof(tar.prefix)] = '\0'; - gcc 4.3.0 would complain
 330                         tar.padding[0] = '\0';
 331                         file_header->name = concat_path_file(tar.prefix, tar.name);
 332                 } else
 333                         file_header->name = xstrdup(tar.name);
 334         }
 335
 336         /* Set bits 12-15 of the files mode */
 337         /* (typeflag was not trashed because chksum does not use getOctal) */
 338         switch (tar.typeflag) {
 339         case '1': /* hardlink */
 340                 /* we mark hardlinks as regular files with zero size and a link name */
 341                 file_header->mode |= S_IFREG;
 342                 /* on size of link fields from star(4)
 343                  * ... For tar archives written by pre POSIX.1-1988
 344                  * implementations, the size field usually contains the size of
 345                  * the file and needs to be ignored as no data may follow this
 346                  * header type.  For POSIX.1- 1988 compliant archives, the size
 347                  * field needs to be 0.  For POSIX.1-2001 compliant archives,
 348                  * the size field may be non zero, indicating that file data is
 349                  * included in the archive.
 350                  * i.e; always assume this is zero for safety.
 351                  */
 352                 goto size0;
 353         case '7':
 354         /* case 0: */
 355         case '0':
 356 #if ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 357                 if (last_char_is(file_header->name, '/')) {
 358                         goto set_dir;
 359                 }
 360 #endif
 361                 file_header->mode |= S_IFREG;
 362                 break;
 363         case '2':
 364                 file_header->mode |= S_IFLNK;
 365                 /* have seen tarballs with size field containing
 366                  * the size of the link target's name */
 367  size0:
 368                 file_header->size = 0;
 369                 break;
 370         case '3':
 371                 file_header->mode |= S_IFCHR;
 372                 goto size0; /* paranoia */
 373         case '4':
 374                 file_header->mode |= S_IFBLK;
 375                 goto size0;
 376         case '5':
 377  IF_FEATURE_TAR_OLDGNU_COMPATIBILITY(set_dir:)
 378                 file_header->mode |= S_IFDIR;
 379                 goto size0;
 380         case '6':
 381                 file_header->mode |= S_IFIFO;
 382                 goto size0;
 383 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 384         case 'L':
 385                 /* free: paranoia: tar with several consecutive longnames */
 386                 free(p_longname);
 387                 /* For paranoia reasons we allocate extra NUL char */
 388                 p_longname = xzalloc(file_header->size + 1);
 389                 /* We read ASCIZ string, including NUL */
 390                 xread(archive_handle->src_fd, p_longname, file_header->size);
 391                 archive_handle->offset += file_header->size;
 392                 /* return get_header_tar(archive_handle); */
 393                 /* gcc 4.1.1 didn't optimize it into jump */
 394                 /* so we will do it ourself, this also saves stack */
 395                 goto again;
 396         case 'K':
 397                 free(p_linkname);
 398                 p_linkname = xzalloc(file_header->size + 1);
 399                 xread(archive_handle->src_fd, p_linkname, file_header->size);
 400                 archive_handle->offset += file_header->size;
 401                 /* return get_header_tar(archive_handle); */
 402                 goto again;
 403         case 'D':       /* GNU dump dir */
 404         case 'M':       /* Continuation of multi volume archive */
 405         case 'N':       /* Old GNU for names > 100 characters */
 406         case 'S':       /* Sparse file */
 407         case 'V':       /* Volume header */
 408 #endif
 409         case 'g':       /* pax global header */
 410         case 'x': {     /* pax extended header */
 411                 if ((uoff_t)file_header->size > 0xfffff) /* paranoia */
 412                         goto skip_ext_hdr;
 413                 process_pax_hdr(archive_handle, file_header->size, (tar.typeflag == 'g'));
 414                 goto again_after_align;
 415         }
 416  skip_ext_hdr:
 417         {
 418                 off_t sz;
 419                 bb_error_msg("warning: skipping header '%c'", tar.typeflag);
 420                 sz = (file_header->size + 511) & ~(off_t)511;
 421                 archive_handle->offset += sz;
 422                 sz >>= 9; /* sz /= 512 but w/o contortions for signed div */
 423                 while (sz--)
 424                         xread(archive_handle->src_fd, &tar, 512);
 425                 /* return get_header_tar(archive_handle); */
 426                 goto again_after_align;
 427         }
 428         default:
 429                 bb_error_msg_and_die("unknown typeflag: 0x%x", tar.typeflag);
 430         }
 431
 432 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 433         if (p_longname) {
 434                 file_header->name = p_longname;
 435                 p_longname = NULL;
 436         }
 437         if (p_linkname) {
 438                 file_header->link_target = p_linkname;
 439                 p_linkname = NULL;
 440         }
 441 #endif
 442
 443         /* Everything up to and including last ".." component is stripped */
 444         overlapping_strcpy(file_header->name, strip_unsafe_prefix(file_header->name));
 445
 446         /* Strip trailing '/' in directories */
 447         /* Must be done after mode is set as '/' is used to check if it's a directory */
 448         cp = last_char_is(file_header->name, '/');
 449
 450         if (archive_handle->filter(archive_handle) == EXIT_SUCCESS) {
 451                 archive_handle->action_header(/*archive_handle->*/ file_header);
 452                 /* Note that we kill the '/' only after action_header() */
 453                 /* (like GNU tar 1.15.1: verbose mode outputs "dir/dir/") */
 454                 if (cp)
 455                         *cp = '\0';
 456                 archive_handle->action_data(archive_handle);
 457                 if (archive_handle->accept || archive_handle->reject
 458                  || (archive_handle->ah_flags & ARCHIVE_REMEMBER_NAMES)
 459                 ) {
 460                         llist_add_to(&archive_handle->passed, file_header->name);
 461                 } else /* Caller isn't interested in list of unpacked files */
 462                         free(file_header->name);
 463         } else {
 464                 data_skip(archive_handle);
 465                 free(file_header->name);
 466         }
 467         archive_handle->offset += file_header->size;
 468
 469         free(file_header->link_target);
 470         /* Do not free(file_header->name)!
 471          * It might be inserted in archive_handle->passed - see above */
 472 #if ENABLE_FEATURE_TAR_UNAME_GNAME
 473         free(file_header->tar__uname);
 474         free(file_header->tar__gname);
 475 #endif
 476         return EXIT_SUCCESS; /* "decoded one header" */
 477 }