u-tools/apps/busybox/archival/libunarchive/get_header_tar.c

   1 /* vi: set sw=4 ts=4: */
   2 /* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
   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 "unarchive.h"
  16
  17 /* NB: _DESTROYS_ str[len] character! */
  18 static unsigned long long getOctal(char *str, int len)
  19 {
  20         unsigned long long v;
  21         /* NB: leading spaces are allowed. Using strtoull to handle that.
  22          * The downside is that we accept e.g. "-123" too :)
  23          */
  24         str[len] = '\0';
  25         v = strtoull(str, &str, 8);
  26         if (*str && (!ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY || *str != ' '))
  27                 bb_error_msg_and_die("corrupted octal value in tar header");
  28         return v;
  29 }
  30 #define GET_OCTAL(a) getOctal((a), sizeof(a))
  31
  32 void BUG_tar_header_size(void);
  33 char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
  34 {
  35         file_header_t *file_header = archive_handle->file_header;
  36         struct {
  37                 /* ustar header, Posix 1003.1 */
  38                 char name[100];     /*   0-99 */
  39                 char mode[8];       /* 100-107 */
  40                 char uid[8];        /* 108-115 */
  41                 char gid[8];        /* 116-123 */
  42                 char size[12];      /* 124-135 */
  43                 char mtime[12];     /* 136-147 */
  44                 char chksum[8];     /* 148-155 */
  45                 char typeflag;      /* 156-156 */
  46                 char linkname[100]; /* 157-256 */
  47                 /* POSIX:   "ustar" NUL "00" */
  48                 /* GNU tar: "ustar  " NUL */
  49                 /* Normally it's defined as magic[6] followed by
  50                  * version[2], but we put them together to simplify code
  51                  */
  52                 char magic[8];      /* 257-264 */
  53                 char uname[32];     /* 265-296 */
  54                 char gname[32];     /* 297-328 */
  55                 char devmajor[8];   /* 329-336 */
  56                 char devminor[8];   /* 337-344 */
  57                 char prefix[155];   /* 345-499 */
  58                 char padding[12];   /* 500-512 */
  59         } tar;
  60         char *cp;
  61         int i, sum_u, sum;
  62 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
  63         int sum_s;
  64 #endif
  65         int parse_names;
  66
  67         /* Our "private data" */
  68 #define p_end (*(smallint *)(&archive_handle->ah_priv[0]))
  69 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
  70 #define p_longname (*(char* *)(&archive_handle->ah_priv[1]))
  71 #define p_linkname (*(char* *)(&archive_handle->ah_priv[2]))
  72 #else
  73 #define p_longname 0
  74 #define p_linkname 0
  75 #endif
  76 //      if (!archive_handle->ah_priv_inited) {
  77 //              archive_handle->ah_priv_inited = 1;
  78 //              p_end = 0;
  79 //              USE_FEATURE_TAR_GNU_EXTENSIONS(p_longname = NULL;)
  80 //              USE_FEATURE_TAR_GNU_EXTENSIONS(p_linkname = NULL;)
  81 //      }
  82
  83         if (sizeof(tar) != 512)
  84                 BUG_tar_header_size();
  85
  86 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
  87  again:
  88 #endif
  89         /* Align header */
  90         data_align(archive_handle, 512);
  91
  92  again_after_align:
  93
  94 #if ENABLE_DESKTOP
  95         /* to prevent misdetection of bz2 sig */
  96         *(uint32_t*)(&tar) = 0;
  97         i = full_read(archive_handle->src_fd, &tar, 512);
  98         /* If GNU tar sees EOF in above read, it says:
  99          * "tar: A lone zero block at N", where N = kilobyte
 100          * where EOF was met (not EOF block, actual EOF!),
 101          * and exits with EXIT_SUCCESS.
 102          * We will mimic exit(EXIT_SUCCESS), although we will not mimic
 103          * the message and we don't check whether we indeed
 104          * saw zero block directly before this. */
 105         if (i == 0) {
 106                 xfunc_error_retval = 0;
 107  short_read:
 108                 bb_error_msg_and_die("short read");
 109         }
 110         if (i != 512) {
 111                 USE_FEATURE_TAR_AUTODETECT(goto autodetect;)
 112                 goto short_read;
 113         }
 114
 115 #else
 116         i = 512;
 117         xread(archive_handle->src_fd, &tar, i);
 118 #endif
 119         archive_handle->offset += i;
 120
 121         /* If there is no filename its an empty header */
 122         if (tar.name[0] == 0 && tar.prefix[0] == 0) {
 123                 if (p_end) {
 124                         /* Second consecutive empty header - end of archive.
 125                          * Read until the end to empty the pipe from gz or bz2
 126                          */
 127                         while (full_read(archive_handle->src_fd, &tar, 512) == 512)
 128                                 continue;
 129                         return EXIT_FAILURE;
 130                 }
 131                 p_end = 1;
 132                 return EXIT_SUCCESS;
 133         }
 134         p_end = 0;
 135
 136         /* Check header has valid magic, "ustar" is for the proper tar,
 137          * five NULs are for the old tar format  */
 138         if (strncmp(tar.magic, "ustar", 5) != 0
 139          && (!ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 140              || memcmp(tar.magic, "\0\0\0\0", 5) != 0)
 141         ) {
 142 #if ENABLE_FEATURE_TAR_AUTODETECT
 143                 char FAST_FUNC (*get_header_ptr)(archive_handle_t *);
 144
 145  USE_DESKTOP(autodetect:)
 146                 /* tar gz/bz autodetect: check for gz/bz2 magic.
 147                  * If we see the magic, and it is the very first block,
 148                  * we can switch to get_header_tar_gz/bz2/lzma().
 149                  * Needs seekable fd. I wish recv(MSG_PEEK) works
 150                  * on any fd... */
 151 #if ENABLE_FEATURE_SEAMLESS_GZ
 152                 if (tar.name[0] == 0x1f && tar.name[1] == (char)0x8b) { /* gzip */
 153                         get_header_ptr = get_header_tar_gz;
 154                 } else
 155 #endif
 156 #if ENABLE_FEATURE_SEAMLESS_BZ2
 157                 if (tar.name[0] == 'B' && tar.name[1] == 'Z'
 158                  && tar.name[2] == 'h' && isdigit(tar.name[3])
 159                 ) { /* bzip2 */
 160                         get_header_ptr = get_header_tar_bz2;
 161                 } else
 162 #endif
 163                         goto err;
 164                 /* Two different causes for lseek() != 0:
 165                  * unseekable fd (would like to support that too, but...),
 166                  * or not first block (false positive, it's not .gz/.bz2!) */
 167                 if (lseek(archive_handle->src_fd, -i, SEEK_CUR) != 0)
 168                         goto err;
 169                 while (get_header_ptr(archive_handle) == EXIT_SUCCESS)
 170                         continue;
 171                 return EXIT_FAILURE;
 172  err:
 173 #endif /* FEATURE_TAR_AUTODETECT */
 174                 bb_error_msg_and_die("invalid tar magic");
 175         }
 176
 177         /* Do checksum on headers.
 178          * POSIX says that checksum is done on unsigned bytes, but
 179          * Sun and HP-UX gets it wrong... more details in
 180          * GNU tar source. */
 181 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 182         sum_s = ' ' * sizeof(tar.chksum);
 183 #endif
 184         sum_u = ' ' * sizeof(tar.chksum);
 185         for (i = 0; i < 148; i++) {
 186                 sum_u += ((unsigned char*)&tar)[i];
 187 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 188                 sum_s += ((signed char*)&tar)[i];
 189 #endif
 190         }
 191         for (i = 156; i < 512; i++) {
 192                 sum_u += ((unsigned char*)&tar)[i];
 193 #if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 194                 sum_s += ((signed char*)&tar)[i];
 195 #endif
 196         }
 197 #if ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 198         sum = strtoul(tar.chksum, &cp, 8);
 199         if ((*cp && *cp != ' ')
 200          || (sum_u != sum USE_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum))
 201         ) {
 202                 bb_error_msg_and_die("invalid tar header checksum");
 203         }
 204 #else
 205         /* This field does not need special treatment (getOctal) */
 206         sum = xstrtoul(tar.chksum, 8);
 207         if (sum_u != sum USE_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum)) {
 208                 bb_error_msg_and_die("invalid tar header checksum");
 209         }
 210 #endif
 211
 212         /* 0 is reserved for high perf file, treat as normal file */
 213         if (!tar.typeflag) tar.typeflag = '0';
 214         parse_names = (tar.typeflag >= '0' && tar.typeflag <= '7');
 215
 216         /* getOctal trashes subsequent field, therefore we call it
 217          * on fields in reverse order */
 218         if (tar.devmajor[0]) {
 219                 char t = tar.prefix[0];
 220                 /* we trash prefix[0] here, but we DO need it later! */
 221                 unsigned minor = GET_OCTAL(tar.devminor);
 222                 unsigned major = GET_OCTAL(tar.devmajor);
 223                 file_header->device = makedev(major, minor);
 224                 tar.prefix[0] = t;
 225         }
 226         file_header->link_target = NULL;
 227         if (!p_linkname && parse_names && tar.linkname[0]) {
 228                 file_header->link_target = xstrndup(tar.linkname, sizeof(tar.linkname));
 229                 /* FIXME: what if we have non-link object with link_target? */
 230                 /* Will link_target be free()ed? */
 231         }
 232 #if ENABLE_FEATURE_TAR_UNAME_GNAME
 233         file_header->uname = tar.uname[0] ? xstrndup(tar.uname, sizeof(tar.uname)) : NULL;
 234         file_header->gname = tar.gname[0] ? xstrndup(tar.gname, sizeof(tar.gname)) : NULL;
 235 #endif
 236         file_header->mtime = GET_OCTAL(tar.mtime);
 237         file_header->size = GET_OCTAL(tar.size);
 238         file_header->gid = GET_OCTAL(tar.gid);
 239         file_header->uid = GET_OCTAL(tar.uid);
 240         /* Set bits 0-11 of the files mode */
 241         file_header->mode = 07777 & GET_OCTAL(tar.mode);
 242
 243         file_header->name = NULL;
 244         if (!p_longname && parse_names) {
 245                 /* we trash mode[0] here, it's ok */
 246                 //tar.name[sizeof(tar.name)] = '\0'; - gcc 4.3.0 would complain
 247                 tar.mode[0] = '\0';
 248                 if (tar.prefix[0]) {
 249                         /* and padding[0] */
 250                         //tar.prefix[sizeof(tar.prefix)] = '\0'; - gcc 4.3.0 would complain
 251                         tar.padding[0] = '\0';
 252                         file_header->name = concat_path_file(tar.prefix, tar.name);
 253                 } else
 254                         file_header->name = xstrdup(tar.name);
 255         }
 256
 257         /* Set bits 12-15 of the files mode */
 258         /* (typeflag was not trashed because chksum does not use getOctal) */
 259         switch (tar.typeflag) {
 260         /* busybox identifies hard links as being regular files with 0 size and a link name */
 261         case '1':
 262                 file_header->mode |= S_IFREG;
 263                 break;
 264         case '7':
 265         /* case 0: */
 266         case '0':
 267 #if ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 268                 if (last_char_is(file_header->name, '/')) {
 269                         goto set_dir;
 270                 }
 271 #endif
 272                 file_header->mode |= S_IFREG;
 273                 break;
 274         case '2':
 275                 file_header->mode |= S_IFLNK;
 276                 /* have seen tarballs with size field containing
 277                  * the size of the link target's name */
 278  size0:
 279                 file_header->size = 0;
 280                 break;
 281         case '3':
 282                 file_header->mode |= S_IFCHR;
 283                 goto size0; /* paranoia */
 284         case '4':
 285                 file_header->mode |= S_IFBLK;
 286                 goto size0;
 287         case '5':
 288  USE_FEATURE_TAR_OLDGNU_COMPATIBILITY(set_dir:)
 289                 file_header->mode |= S_IFDIR;
 290                 goto size0;
 291         case '6':
 292                 file_header->mode |= S_IFIFO;
 293                 goto size0;
 294 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 295         case 'L':
 296                 /* free: paranoia: tar with several consecutive longnames */
 297                 free(p_longname);
 298                 /* For paranoia reasons we allocate extra NUL char */
 299                 p_longname = xzalloc(file_header->size + 1);
 300                 /* We read ASCIZ string, including NUL */
 301                 xread(archive_handle->src_fd, p_longname, file_header->size);
 302                 archive_handle->offset += file_header->size;
 303                 /* return get_header_tar(archive_handle); */
 304                 /* gcc 4.1.1 didn't optimize it into jump */
 305                 /* so we will do it ourself, this also saves stack */
 306                 goto again;
 307         case 'K':
 308                 free(p_linkname);
 309                 p_linkname = xzalloc(file_header->size + 1);
 310                 xread(archive_handle->src_fd, p_linkname, file_header->size);
 311                 archive_handle->offset += file_header->size;
 312                 /* return get_header_tar(archive_handle); */
 313                 goto again;
 314         case 'D':       /* GNU dump dir */
 315         case 'M':       /* Continuation of multi volume archive */
 316         case 'N':       /* Old GNU for names > 100 characters */
 317         case 'S':       /* Sparse file */
 318         case 'V':       /* Volume header */
 319 #endif
 320         case 'g':       /* pax global header */
 321         case 'x': {     /* pax extended header */
 322                 off_t sz;
 323                 bb_error_msg("warning: skipping header '%c'", tar.typeflag);
 324                 sz = (file_header->size + 511) & ~(off_t)511;
 325                 archive_handle->offset += sz;
 326                 sz >>= 9; /* sz /= 512 but w/o contortions for signed div */
 327                 while (sz--)
 328                         xread(archive_handle->src_fd, &tar, 512);
 329                 /* return get_header_tar(archive_handle); */
 330                 goto again_after_align;
 331         }
 332         default:
 333                 bb_error_msg_and_die("unknown typeflag: 0x%x", tar.typeflag);
 334         }
 335
 336 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 337         if (p_longname) {
 338                 file_header->name = p_longname;
 339                 p_longname = NULL;
 340         }
 341         if (p_linkname) {
 342                 file_header->link_target = p_linkname;
 343                 p_linkname = NULL;
 344         }
 345 #endif
 346         if (strncmp(file_header->name, "/../"+1, 3) == 0
 347          || strstr(file_header->name, "/../")
 348         ) {
 349                 bb_error_msg_and_die("name with '..' encountered: '%s'",
 350                                 file_header->name);
 351         }
 352
 353         /* Strip trailing '/' in directories */
 354         /* Must be done after mode is set as '/' is used to check if it's a directory */
 355         cp = last_char_is(file_header->name, '/');
 356
 357         if (archive_handle->filter(archive_handle) == EXIT_SUCCESS) {
 358                 archive_handle->action_header(/*archive_handle->*/ file_header);
 359                 /* Note that we kill the '/' only after action_header() */
 360                 /* (like GNU tar 1.15.1: verbose mode outputs "dir/dir/") */
 361                 if (cp) *cp = '\0';
 362                 archive_handle->ah_flags |= ARCHIVE_EXTRACT_QUIET;
 363                 archive_handle->action_data(archive_handle);
 364                 llist_add_to(&(archive_handle->passed), file_header->name);
 365         } else {
 366                 data_skip(archive_handle);
 367                 free(file_header->name);
 368         }
 369         archive_handle->offset += file_header->size;
 370
 371         free(file_header->link_target);
 372         /* Do not free(file_header->name)! (why?) */
 373 #if ENABLE_FEATURE_TAR_UNAME_GNAME
 374         free(file_header->uname);
 375         free(file_header->gname);
 376 #endif
 377         return EXIT_SUCCESS;
 378 }