LXR busybox/archival/libunarchive/get_header

   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
  17typedef uint32_t aliased_uint32_t FIX_ALIASING;
  18typedef off_t    aliased_off_t    FIX_ALIASING;
  19
  20
  21/* NB: _DESTROYS_ str[len] character! */
  22static unsigned long long getOctal(char *str, int len)
  23{
  24        unsigned long long v;
  25        char *end;
  26        /* NB: leading spaces are allowed. Using strtoull to handle that.
  27         * The downside is that we accept e.g. "-123" too :(
  28         */
  29        str[len] = '\0';
  30        v = strtoull(str, &end, 8);
  31        /* std: "Each numeric field is terminated by one or more
  32         * <space> or NUL characters". We must support ' '! */
  33        if (*end != '\0' && *end != ' ') {
  34                int8_t first = str[0];
  35                if (!(first & 0x80))
  36                        bb_error_msg_and_die("corrupted octal value in tar header");
  37                /*
  38                 * GNU tar uses "base-256 encoding" for very large numbers.
  39                 * Encoding is binary, with highest bit always set as a marker
  40                 * and sign in next-highest bit:
  41                 * 80 00 .. 00 - zero
  42                 * bf ff .. ff - largest positive number
  43                 * ff ff .. ff - minus 1
  44                 * c0 00 .. 00 - smallest negative number
  45                 *
  46                 * Example of tar file with 8914993153 (0x213600001) byte file.
  47                 * Field starts at offset 7c:
  48                 * 00070  30 30 30 00 30 30 30 30  30 30 30 00 80 00 00 00  |000.0000000.....|
  49                 * 00080  00 00 00 02 13 60 00 01  31 31 31 32 30 33 33 36  |.....`..11120336|
  50                 *
  51                 * NB: tarballs with NEGATIVE unix times encoded that way were seen!
  52                 */
  53                v = first;
  54                /* Sign-extend using 6th bit: */
  55                v <<= sizeof(unsigned long long)*8 - 7;
  56                v = (long long)v >> (sizeof(unsigned long long)*8 - 7);
  57                while (--len != 0)
  58                        v = (v << 8) + (unsigned char) *str++;
  59        }
  60        return v;
  61}
  62#define GET_OCTAL(a) getOctal((a), sizeof(a))
  63
  64#if ENABLE_FEATURE_TAR_SELINUX
  65/* Scan a PAX header for SELinux contexts, via "RHT.security.selinux" keyword.
  66 * This is what Red Hat's patched version of tar uses.
  67 */
  68# define SELINUX_CONTEXT_KEYWORD "RHT.security.selinux"
  69static char *get_selinux_sctx_from_pax_hdr(archive_handle_t *archive_handle, unsigned sz)
  70{
  71        char *buf, *p;
  72        char *result;
  73
  74        p = buf = xmalloc(sz + 1);
  75        /* prevent bb_strtou from running off the buffer */
  76        buf[sz] = '\0';
  77        xread(archive_handle->src_fd, buf, sz);
  78        archive_handle->offset += sz;
  79
  80        result = NULL;
  81        while (sz != 0) {
  82                char *end, *value;
  83                unsigned len;
  84
  85                /* Every record has this format: "LEN NAME=VALUE\n" */
  86                len = bb_strtou(p, &end, 10);
  87                /* expect errno to be EINVAL, because the character
  88                 * following the digits should be a space
  89                 */
  90                p += len;
  91                sz -= len;
  92                if ((int)sz < 0
  93                 || len == 0
  94                 || errno != EINVAL
  95                 || *end != ' '
  96                ) {
  97                        bb_error_msg("malformed extended header, skipped");
  98                        // More verbose version:
  99                        //bb_error_msg("malformed extended header at %"OFF_FMT"d, skipped",
 100                        //              archive_handle->offset - (sz + len));
 101                        break;
 102                }
 103                /* overwrite the terminating newline with NUL
 104                 * (we do not bother to check that it *was* a newline)
 105                 */
 106                p[-1] = '\0';
 107                /* Is it selinux security context? */
 108                value = end + 1;
 109                if (strncmp(value, SELINUX_CONTEXT_KEYWORD"=", sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1) == 0) {
 110                        value += sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1;
 111                        result = xstrdup(value);
 112                        break;
 113                }
 114        }
 115
 116        free(buf);
 117        return result;
 118}
 119#endif
 120
 121char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 122{
 123        file_header_t *file_header = archive_handle->file_header;
 124        struct tar_header_t tar;
 125        char *cp;
 126        int i, sum_u, sum;
 127#if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 128        int sum_s;
 129#endif
 130        int parse_names;
 131
 132        /* Our "private data" */
 133#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 134# define p_longname (archive_handle->tar__longname)
 135# define p_linkname (archive_handle->tar__linkname)
 136#else
 137# define p_longname 0
 138# define p_linkname 0
 139#endif
 140
 141#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS || ENABLE_FEATURE_TAR_SELINUX
 142 again:
 143#endif
 144        /* Align header */
 145        data_align(archive_handle, 512);
 146
 147 again_after_align:
 148
 149#if ENABLE_DESKTOP || ENABLE_FEATURE_TAR_AUTODETECT
 150        /* to prevent misdetection of bz2 sig */
 151        *(aliased_uint32_t*)&tar = 0;
 152        i = full_read(archive_handle->src_fd, &tar, 512);
 153        /* If GNU tar sees EOF in above read, it says:
 154         * "tar: A lone zero block at N", where N = kilobyte
 155         * where EOF was met (not EOF block, actual EOF!),
 156         * and exits with EXIT_SUCCESS.
 157         * We will mimic exit(EXIT_SUCCESS), although we will not mimic
 158         * the message and we don't check whether we indeed
 159         * saw zero block directly before this. */
 160        if (i == 0) {
 161                xfunc_error_retval = 0;
 162 short_read:
 163                bb_error_msg_and_die("short read");
 164        }
 165        if (i != 512) {
 166                IF_FEATURE_TAR_AUTODETECT(goto autodetect;)
 167                goto short_read;
 168        }
 169
 170#else
 171        i = 512;
 172        xread(archive_handle->src_fd, &tar, i);
 173#endif
 174        archive_handle->offset += i;
 175
 176        /* If there is no filename its an empty header */
 177        if (tar.name[0] == 0 && tar.prefix[0] == 0) {
 178                if (archive_handle->tar__end) {
 179                        /* Second consecutive empty header - end of archive.
 180                         * Read until the end to empty the pipe from gz or bz2
 181                         */
 182                        while (full_read(archive_handle->src_fd, &tar, 512) == 512)
 183                                continue;
 184                        return EXIT_FAILURE;
 185                }
 186                archive_handle->tar__end = 1;
 187                return EXIT_SUCCESS;
 188        }
 189        archive_handle->tar__end = 0;
 190
 191        /* Check header has valid magic, "ustar" is for the proper tar,
 192         * five NULs are for the old tar format  */
 193        if (strncmp(tar.magic, "ustar", 5) != 0
 194         && (!ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 195             || memcmp(tar.magic, "\0\0\0\0", 5) != 0)
 196        ) {
 197#if ENABLE_FEATURE_TAR_AUTODETECT
 198                char FAST_FUNC (*get_header_ptr)(archive_handle_t *);
 199                uint16_t magic2;
 200
 201 autodetect:
 202                magic2 = *(uint16_t*)tar.name;
 203                /* tar gz/bz autodetect: check for gz/bz2 magic.
 204                 * If we see the magic, and it is the very first block,
 205                 * we can switch to get_header_tar_gz/bz2/lzma().
 206                 * Needs seekable fd. I wish recv(MSG_PEEK) works
 207                 * on any fd... */
 208# if ENABLE_FEATURE_SEAMLESS_GZ
 209                if (magic2 == GZIP_MAGIC) {
 210                        get_header_ptr = get_header_tar_gz;
 211                } else
 212# endif
 213# if ENABLE_FEATURE_SEAMLESS_BZ2
 214                if (magic2 == BZIP2_MAGIC
 215                 && tar.name[2] == 'h' && isdigit(tar.name[3])
 216                ) { /* bzip2 */
 217                        get_header_ptr = get_header_tar_bz2;
 218                } else
 219# endif
 220# if ENABLE_FEATURE_SEAMLESS_XZ
 221                //TODO: if (magic2 == XZ_MAGIC1)...
 222                //else
 223# endif
 224                        goto err;
 225                /* Two different causes for lseek() != 0:
 226                 * unseekable fd (would like to support that too, but...),
 227                 * or not first block (false positive, it's not .gz/.bz2!) */
 228                if (lseek(archive_handle->src_fd, -i, SEEK_CUR) != 0)
 229                        goto err;
 230                while (get_header_ptr(archive_handle) == EXIT_SUCCESS)
 231                        continue;
 232                return EXIT_FAILURE;
 233 err:
 234#endif /* FEATURE_TAR_AUTODETECT */
 235                bb_error_msg_and_die("invalid tar magic");
 236        }
 237
 238        /* Do checksum on headers.
 239         * POSIX says that checksum is done on unsigned bytes, but
 240         * Sun and HP-UX gets it wrong... more details in
 241         * GNU tar source. */
 242#if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 243        sum_s = ' ' * sizeof(tar.chksum);
 244#endif
 245        sum_u = ' ' * sizeof(tar.chksum);
 246        for (i = 0; i < 148; i++) {
 247                sum_u += ((unsigned char*)&tar)[i];
 248#if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 249                sum_s += ((signed char*)&tar)[i];
 250#endif
 251        }
 252        for (i = 156; i < 512; i++) {
 253                sum_u += ((unsigned char*)&tar)[i];
 254#if ENABLE_FEATURE_TAR_OLDSUN_COMPATIBILITY
 255                sum_s += ((signed char*)&tar)[i];
 256#endif
 257        }
 258        /* This field does not need special treatment (getOctal) */
 259        {
 260                char *endp; /* gcc likes temp var for &endp */
 261                sum = strtoul(tar.chksum, &endp, 8);
 262                if ((*endp != '\0' && *endp != ' ')
 263                 || (sum_u != sum IF_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum))
 264                ) {
 265                        bb_error_msg_and_die("invalid tar header checksum");
 266                }
 267        }
 268        /* don't use xstrtoul, tar.chksum may have leading spaces */
 269        sum = strtoul(tar.chksum, NULL, 8);
 270        if (sum_u != sum IF_FEATURE_TAR_OLDSUN_COMPATIBILITY(&& sum_s != sum)) {
 271                bb_error_msg_and_die("invalid tar header checksum");
 272        }
 273
 274        /* 0 is reserved for high perf file, treat as normal file */
 275        if (!tar.typeflag) tar.typeflag = '0';
 276        parse_names = (tar.typeflag >= '0' && tar.typeflag <= '7');
 277
 278        /* getOctal trashes subsequent field, therefore we call it
 279         * on fields in reverse order */
 280        if (tar.devmajor[0]) {
 281                char t = tar.prefix[0];
 282                /* we trash prefix[0] here, but we DO need it later! */
 283                unsigned minor = GET_OCTAL(tar.devminor);
 284                unsigned major = GET_OCTAL(tar.devmajor);
 285                file_header->device = makedev(major, minor);
 286                tar.prefix[0] = t;
 287        }
 288        file_header->link_target = NULL;
 289        if (!p_linkname && parse_names && tar.linkname[0]) {
 290                file_header->link_target = xstrndup(tar.linkname, sizeof(tar.linkname));
 291                /* FIXME: what if we have non-link object with link_target? */
 292                /* Will link_target be free()ed? */
 293        }
 294#if ENABLE_FEATURE_TAR_UNAME_GNAME
 295        file_header->tar__uname = tar.uname[0] ? xstrndup(tar.uname, sizeof(tar.uname)) : NULL;
 296        file_header->tar__gname = tar.gname[0] ? xstrndup(tar.gname, sizeof(tar.gname)) : NULL;
 297#endif
 298        file_header->mtime = GET_OCTAL(tar.mtime);
 299        file_header->size = GET_OCTAL(tar.size);
 300        file_header->gid = GET_OCTAL(tar.gid);
 301        file_header->uid = GET_OCTAL(tar.uid);
 302        /* Set bits 0-11 of the files mode */
 303        file_header->mode = 07777 & GET_OCTAL(tar.mode);
 304
 305        file_header->name = NULL;
 306        if (!p_longname && parse_names) {
 307                /* we trash mode[0] here, it's ok */
 308                //tar.name[sizeof(tar.name)] = '\0'; - gcc 4.3.0 would complain
 309                tar.mode[0] = '\0';
 310                if (tar.prefix[0]) {
 311                        /* and padding[0] */
 312                        //tar.prefix[sizeof(tar.prefix)] = '\0'; - gcc 4.3.0 would complain
 313                        tar.padding[0] = '\0';
 314                        file_header->name = concat_path_file(tar.prefix, tar.name);
 315                } else
 316                        file_header->name = xstrdup(tar.name);
 317        }
 318
 319        /* Set bits 12-15 of the files mode */
 320        /* (typeflag was not trashed because chksum does not use getOctal) */
 321        switch (tar.typeflag) {
 322        /* busybox identifies hard links as being regular files with 0 size and a link name */
 323        case '1':
 324                file_header->mode |= S_IFREG;
 325                break;
 326        case '7':
 327        /* case 0: */
 328        case '0':
 329#if ENABLE_FEATURE_TAR_OLDGNU_COMPATIBILITY
 330                if (last_char_is(file_header->name, '/')) {
 331                        goto set_dir;
 332                }
 333#endif
 334                file_header->mode |= S_IFREG;
 335                break;
 336        case '2':
 337                file_header->mode |= S_IFLNK;
 338                /* have seen tarballs with size field containing
 339                 * the size of the link target's name */
 340 size0:
 341                file_header->size = 0;
 342                break;
 343        case '3':
 344                file_header->mode |= S_IFCHR;
 345                goto size0; /* paranoia */
 346        case '4':
 347                file_header->mode |= S_IFBLK;
 348                goto size0;
 349        case '5':
 350 IF_FEATURE_TAR_OLDGNU_COMPATIBILITY(set_dir:)
 351                file_header->mode |= S_IFDIR;
 352                goto size0;
 353        case '6':
 354                file_header->mode |= S_IFIFO;
 355                goto size0;
 356#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 357        case 'L':
 358                /* free: paranoia: tar with several consecutive longnames */
 359                free(p_longname);
 360                /* For paranoia reasons we allocate extra NUL char */
 361                p_longname = xzalloc(file_header->size + 1);
 362                /* We read ASCIZ string, including NUL */
 363                xread(archive_handle->src_fd, p_longname, file_header->size);
 364                archive_handle->offset += file_header->size;
 365                /* return get_header_tar(archive_handle); */
 366                /* gcc 4.1.1 didn't optimize it into jump */
 367                /* so we will do it ourself, this also saves stack */
 368                goto again;
 369        case 'K':
 370                free(p_linkname);
 371                p_linkname = xzalloc(file_header->size + 1);
 372                xread(archive_handle->src_fd, p_linkname, file_header->size);
 373                archive_handle->offset += file_header->size;
 374                /* return get_header_tar(archive_handle); */
 375                goto again;
 376        case 'D':       /* GNU dump dir */
 377        case 'M':       /* Continuation of multi volume archive */
 378        case 'N':       /* Old GNU for names > 100 characters */
 379        case 'S':       /* Sparse file */
 380        case 'V':       /* Volume header */
 381#endif
 382#if !ENABLE_FEATURE_TAR_SELINUX
 383        case 'g':       /* pax global header */
 384        case 'x':       /* pax extended header */
 385#else
 386 skip_ext_hdr:
 387#endif
 388        {
 389                off_t sz;
 390                bb_error_msg("warning: skipping header '%c'", tar.typeflag);
 391                sz = (file_header->size + 511) & ~(off_t)511;
 392                archive_handle->offset += sz;
 393                sz >>= 9; /* sz /= 512 but w/o contortions for signed div */
 394                while (sz--)
 395                        xread(archive_handle->src_fd, &tar, 512);
 396                /* return get_header_tar(archive_handle); */
 397                goto again_after_align;
 398        }
 399#if ENABLE_FEATURE_TAR_SELINUX
 400        case 'g':       /* pax global header */
 401        case 'x': {     /* pax extended header */
 402                char **pp;
 403                if ((uoff_t)file_header->size > 0xfffff) /* paranoia */
 404                        goto skip_ext_hdr;
 405                pp = (tar.typeflag == 'g') ? &archive_handle->tar__global_sctx : &archive_handle->tar__next_file_sctx;
 406                free(*pp);
 407                *pp = get_selinux_sctx_from_pax_hdr(archive_handle, file_header->size);
 408                goto again;
 409        }
 410#endif
 411        default:
 412                bb_error_msg_and_die("unknown typeflag: 0x%x", tar.typeflag);
 413        }
 414
 415#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 416        if (p_longname) {
 417                file_header->name = p_longname;
 418                p_longname = NULL;
 419        }
 420        if (p_linkname) {
 421                file_header->link_target = p_linkname;
 422                p_linkname = NULL;
 423        }
 424#endif
 425        if (strncmp(file_header->name, "/../"+1, 3) == 0
 426         || strstr(file_header->name, "/../")
 427        ) {
 428                bb_error_msg_and_die("name with '..' encountered: '%s'",
 429                                file_header->name);
 430        }
 431
 432        /* Strip trailing '/' in directories */
 433        /* Must be done after mode is set as '/' is used to check if it's a directory */
 434        cp = last_char_is(file_header->name, '/');
 435
 436        if (archive_handle->filter(archive_handle) == EXIT_SUCCESS) {
 437                archive_handle->action_header(/*archive_handle->*/ file_header);
 438                /* Note that we kill the '/' only after action_header() */
 439                /* (like GNU tar 1.15.1: verbose mode outputs "dir/dir/") */
 440                if (cp)
 441                        *cp = '\0';
 442                archive_handle->action_data(archive_handle);
 443                if (archive_handle->accept || archive_handle->reject)
 444                        llist_add_to(&archive_handle->passed, file_header->name);
 445                else /* Caller isn't interested in list of unpacked files */
 446                        free(file_header->name);
 447        } else {
 448                data_skip(archive_handle);
 449                free(file_header->name);
 450        }
 451        archive_handle->offset += file_header->size;
 452
 453        free(file_header->link_target);
 454        /* Do not free(file_header->name)!
 455         * It might be inserted in archive_handle->passed - see above */
 456#if ENABLE_FEATURE_TAR_UNAME_GNAME
 457        free(file_header->tar__uname);
 458        free(file_header->tar__gname);
 459#endif
 460        return EXIT_SUCCESS;
 461}
 462