archival/tar.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini tar implementation for busybox
   4  *
   5  * Modified to use common extraction code used by ar, cpio, dpkg-deb, dpkg
   6  *  by Glenn McGrath
   7  *
   8  * Note, that as of BusyBox-0.43, tar has been completely rewritten from the
   9  * ground up.  It still has remnants of the old code lying about, but it is
  10  * very different now (i.e., cleaner, less global variables, etc.)
  11  *
  12  * Copyright (C) 1999-2004 by Erik Andersen <andersen@codepoet.org>
  13  *
  14  * Based in part in the tar implementation in sash
  15  *  Copyright (c) 1999 by David I. Bell
  16  *  Permission is granted to use, distribute, or modify this source,
  17  *  provided that this copyright notice remains intact.
  18  *  Permission to distribute sash derived code under the GPL has been granted.
  19  *
  20  * Based in part on the tar implementation from busybox-0.28
  21  *  Copyright (C) 1995 Bruce Perens
  22  *
  23  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
  24  */
  25
  26 #include <fnmatch.h>
  27 #include "libbb.h"
  28 #include "unarchive.h"
  29 /* FIXME: Stop using this non-standard feature */
  30 #ifndef FNM_LEADING_DIR
  31 # define FNM_LEADING_DIR 0
  32 #endif
  33
  34
  35 //#define DBG(fmt, ...) bb_error_msg("%s: " fmt, __func__, ## __VA_ARGS__)
  36 #define DBG(...) ((void)0)
  37
  38
  39 #define block_buf bb_common_bufsiz1
  40
  41
  42 #if !ENABLE_FEATURE_SEAMLESS_GZ && !ENABLE_FEATURE_SEAMLESS_BZ2
  43 /* Do not pass gzip flag to writeTarFile() */
  44 #define writeTarFile(tar_fd, verboseFlag, dereferenceFlag, include, exclude, gzip) \
  45         writeTarFile(tar_fd, verboseFlag, dereferenceFlag, include, exclude)
  46 #endif
  47
  48
  49 #if ENABLE_FEATURE_TAR_CREATE
  50
  51 /* Tar file constants  */
  52
  53 #define TAR_BLOCK_SIZE          512
  54
  55 /* POSIX tar Header Block, from POSIX 1003.1-1990  */
  56 #define NAME_SIZE      100
  57 #define NAME_SIZE_STR "100"
  58 typedef struct TarHeader {       /* byte offset */
  59         char name[NAME_SIZE];     /*   0-99 */
  60         char mode[8];             /* 100-107 */
  61         char uid[8];              /* 108-115 */
  62         char gid[8];              /* 116-123 */
  63         char size[12];            /* 124-135 */
  64         char mtime[12];           /* 136-147 */
  65         char chksum[8];           /* 148-155 */
  66         char typeflag;            /* 156-156 */
  67         char linkname[NAME_SIZE]; /* 157-256 */
  68         /* POSIX:   "ustar" NUL "00" */
  69         /* GNU tar: "ustar  " NUL */
  70         /* Normally it's defined as magic[6] followed by
  71          * version[2], but we put them together to save code.
  72          */
  73         char magic[8];            /* 257-264 */
  74         char uname[32];           /* 265-296 */
  75         char gname[32];           /* 297-328 */
  76         char devmajor[8];         /* 329-336 */
  77         char devminor[8];         /* 337-344 */
  78         char prefix[155];         /* 345-499 */
  79         char padding[12];         /* 500-512 (pad to exactly TAR_BLOCK_SIZE) */
  80 } TarHeader;
  81
  82 /*
  83 ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are
  84 ** the only functions that deal with the HardLinkInfo structure.
  85 ** Even these functions use the xxxHardLinkInfo() functions.
  86 */
  87 typedef struct HardLinkInfo {
  88         struct HardLinkInfo *next; /* Next entry in list */
  89         dev_t dev;                 /* Device number */
  90         ino_t ino;                 /* Inode number */
  91 //      short linkCount;           /* (Hard) Link Count */
  92         char name[1];              /* Start of filename (must be last) */
  93 } HardLinkInfo;
  94
  95 /* Some info to be carried along when creating a new tarball */
  96 typedef struct TarBallInfo {
  97         int tarFd;                      /* Open-for-write file descriptor
  98                                          * for the tarball */
  99         int verboseFlag;                /* Whether to print extra stuff or not */
 100         const llist_t *excludeList;     /* List of files to not include */
 101         HardLinkInfo *hlInfoHead;       /* Hard Link Tracking Information */
 102         HardLinkInfo *hlInfo;           /* Hard Link Info for the current file */
 103 //TODO: save only st_dev + st_ino
 104         struct stat tarFileStatBuf;     /* Stat info for the tarball, letting
 105                                          * us know the inode and device that the
 106                                          * tarball lives, so we can avoid trying
 107                                          * to include the tarball into itself */
 108 } TarBallInfo;
 109
 110 /* A nice enum with all the possible tar file content types */
 111 enum {
 112         REGTYPE = '0',          /* regular file */
 113         REGTYPE0 = '\0',        /* regular file (ancient bug compat) */
 114         LNKTYPE = '1',          /* hard link */
 115         SYMTYPE = '2',          /* symbolic link */
 116         CHRTYPE = '3',          /* character special */
 117         BLKTYPE = '4',          /* block special */
 118         DIRTYPE = '5',          /* directory */
 119         FIFOTYPE = '6',         /* FIFO special */
 120         CONTTYPE = '7',         /* reserved */
 121         GNULONGLINK = 'K',      /* GNU long (>100 chars) link name */
 122         GNULONGNAME = 'L',      /* GNU long (>100 chars) file name */
 123 };
 124
 125 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
 126 static void addHardLinkInfo(HardLinkInfo **hlInfoHeadPtr,
 127                                         struct stat *statbuf,
 128                                         const char *fileName)
 129 {
 130         /* Note: hlInfoHeadPtr can never be NULL! */
 131         HardLinkInfo *hlInfo;
 132
 133         hlInfo = xmalloc(sizeof(HardLinkInfo) + strlen(fileName));
 134         hlInfo->next = *hlInfoHeadPtr;
 135         *hlInfoHeadPtr = hlInfo;
 136         hlInfo->dev = statbuf->st_dev;
 137         hlInfo->ino = statbuf->st_ino;
 138 //      hlInfo->linkCount = statbuf->st_nlink;
 139         strcpy(hlInfo->name, fileName);
 140 }
 141
 142 static void freeHardLinkInfo(HardLinkInfo **hlInfoHeadPtr)
 143 {
 144         HardLinkInfo *hlInfo;
 145         HardLinkInfo *hlInfoNext;
 146
 147         if (hlInfoHeadPtr) {
 148                 hlInfo = *hlInfoHeadPtr;
 149                 while (hlInfo) {
 150                         hlInfoNext = hlInfo->next;
 151                         free(hlInfo);
 152                         hlInfo = hlInfoNext;
 153                 }
 154                 *hlInfoHeadPtr = NULL;
 155         }
 156 }
 157
 158 /* Might be faster (and bigger) if the dev/ino were stored in numeric order ;) */
 159 static HardLinkInfo *findHardLinkInfo(HardLinkInfo *hlInfo, struct stat *statbuf)
 160 {
 161         while (hlInfo) {
 162                 if (statbuf->st_ino == hlInfo->ino
 163                  && statbuf->st_dev == hlInfo->dev
 164                 ) {
 165                         DBG("found hardlink:'%s'", hlInfo->name);
 166                         break;
 167                 }
 168                 hlInfo = hlInfo->next;
 169         }
 170         return hlInfo;
 171 }
 172
 173 /* Put an octal string into the specified buffer.
 174  * The number is zero padded and possibly null terminated.
 175  * Stores low-order bits only if whole value does not fit. */
 176 static void putOctal(char *cp, int len, off_t value)
 177 {
 178         char tempBuffer[sizeof(off_t)*3 + 1];
 179         char *tempString = tempBuffer;
 180         int width;
 181
 182         width = sprintf(tempBuffer, "%0*"OFF_FMT"o", len, value);
 183         tempString += (width - len);
 184
 185         /* If string has leading zeroes, we can drop one */
 186         /* and field will have trailing '\0' */
 187         /* (increases chances of compat with other tars) */
 188         if (tempString[0] == '0')
 189                 tempString++;
 190
 191         /* Copy the string to the field */
 192         memcpy(cp, tempString, len);
 193 }
 194 #define PUT_OCTAL(a, b) putOctal((a), sizeof(a), (b))
 195
 196 static void chksum_and_xwrite(int fd, struct TarHeader* hp)
 197 {
 198         /* POSIX says that checksum is done on unsigned bytes
 199          * (Sun and HP-UX gets it wrong... more details in
 200          * GNU tar source) */
 201         const unsigned char *cp;
 202         int chksum, size;
 203
 204         strcpy(hp->magic, "ustar  ");
 205
 206         /* Calculate and store the checksum (i.e., the sum of all of the bytes of
 207          * the header).  The checksum field must be filled with blanks for the
 208          * calculation.  The checksum field is formatted differently from the
 209          * other fields: it has 6 digits, a null, then a space -- rather than
 210          * digits, followed by a null like the other fields... */
 211         memset(hp->chksum, ' ', sizeof(hp->chksum));
 212         cp = (const unsigned char *) hp;
 213         chksum = 0;
 214         size = sizeof(*hp);
 215         do { chksum += *cp++; } while (--size);
 216         putOctal(hp->chksum, sizeof(hp->chksum)-1, chksum);
 217
 218         /* Now write the header out to disk */
 219         xwrite(fd, hp, sizeof(*hp));
 220 }
 221
 222 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 223 static void writeLongname(int fd, int type, const char *name, int dir)
 224 {
 225         static const struct {
 226                 char mode[8];             /* 100-107 */
 227                 char uid[8];              /* 108-115 */
 228                 char gid[8];              /* 116-123 */
 229                 char size[12];            /* 124-135 */
 230                 char mtime[12];           /* 136-147 */
 231         } prefilled = {
 232                 "0000000",
 233                 "0000000",
 234                 "0000000",
 235                 "00000000000",
 236                 "00000000000",
 237         };
 238         struct TarHeader header;
 239         int size;
 240
 241         dir = !!dir; /* normalize: 0/1 */
 242         size = strlen(name) + 1 + dir; /* GNU tar uses strlen+1 */
 243         /* + dir: account for possible '/' */
 244
 245         memset(&header, 0, sizeof(header));
 246         strcpy(header.name, "././@LongLink");
 247         memcpy(header.mode, prefilled.mode, sizeof(prefilled));
 248         PUT_OCTAL(header.size, size);
 249         header.typeflag = type;
 250         chksum_and_xwrite(fd, &header);
 251
 252         /* Write filename[/] and pad the block. */
 253         /* dir=0: writes 'name<NUL>', pads */
 254         /* dir=1: writes 'name', writes '/<NUL>', pads */
 255         dir *= 2;
 256         xwrite(fd, name, size - dir);
 257         xwrite(fd, "/", dir);
 258         size = (-size) & (TAR_BLOCK_SIZE-1);
 259         memset(&header, 0, size);
 260         xwrite(fd, &header, size);
 261 }
 262 #endif
 263
 264 /* Write out a tar header for the specified file/directory/whatever */
 265 void BUG_tar_header_size(void);
 266 static int writeTarHeader(struct TarBallInfo *tbInfo,
 267                 const char *header_name, const char *fileName, struct stat *statbuf)
 268 {
 269         struct TarHeader header;
 270
 271         if (sizeof(header) != 512)
 272                 BUG_tar_header_size();
 273
 274         memset(&header, 0, sizeof(struct TarHeader));
 275
 276         strncpy(header.name, header_name, sizeof(header.name));
 277
 278         /* POSIX says to mask mode with 07777. */
 279         PUT_OCTAL(header.mode, statbuf->st_mode & 07777);
 280         PUT_OCTAL(header.uid, statbuf->st_uid);
 281         PUT_OCTAL(header.gid, statbuf->st_gid);
 282         memset(header.size, '0', sizeof(header.size)-1); /* Regular file size is handled later */
 283         PUT_OCTAL(header.mtime, statbuf->st_mtime);
 284
 285         /* Enter the user and group names */
 286         safe_strncpy(header.uname, get_cached_username(statbuf->st_uid), sizeof(header.uname));
 287         safe_strncpy(header.gname, get_cached_groupname(statbuf->st_gid), sizeof(header.gname));
 288
 289         if (tbInfo->hlInfo) {
 290                 /* This is a hard link */
 291                 header.typeflag = LNKTYPE;
 292                 strncpy(header.linkname, tbInfo->hlInfo->name,
 293                                 sizeof(header.linkname));
 294 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 295                 /* Write out long linkname if needed */
 296                 if (header.linkname[sizeof(header.linkname)-1])
 297                         writeLongname(tbInfo->tarFd, GNULONGLINK,
 298                                         tbInfo->hlInfo->name, 0);
 299 #endif
 300         } else if (S_ISLNK(statbuf->st_mode)) {
 301                 char *lpath = xmalloc_readlink_or_warn(fileName);
 302                 if (!lpath)
 303                         return FALSE;
 304                 header.typeflag = SYMTYPE;
 305                 strncpy(header.linkname, lpath, sizeof(header.linkname));
 306 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 307                 /* Write out long linkname if needed */
 308                 if (header.linkname[sizeof(header.linkname)-1])
 309                         writeLongname(tbInfo->tarFd, GNULONGLINK, lpath, 0);
 310 #else
 311                 /* If it is larger than 100 bytes, bail out */
 312                 if (header.linkname[sizeof(header.linkname)-1]) {
 313                         free(lpath);
 314                         bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported");
 315                         return FALSE;
 316                 }
 317 #endif
 318                 free(lpath);
 319         } else if (S_ISDIR(statbuf->st_mode)) {
 320                 header.typeflag = DIRTYPE;
 321                 /* Append '/' only if there is a space for it */
 322                 if (!header.name[sizeof(header.name)-1])
 323                         header.name[strlen(header.name)] = '/';
 324         } else if (S_ISCHR(statbuf->st_mode)) {
 325                 header.typeflag = CHRTYPE;
 326                 PUT_OCTAL(header.devmajor, major(statbuf->st_rdev));
 327                 PUT_OCTAL(header.devminor, minor(statbuf->st_rdev));
 328         } else if (S_ISBLK(statbuf->st_mode)) {
 329                 header.typeflag = BLKTYPE;
 330                 PUT_OCTAL(header.devmajor, major(statbuf->st_rdev));
 331                 PUT_OCTAL(header.devminor, minor(statbuf->st_rdev));
 332         } else if (S_ISFIFO(statbuf->st_mode)) {
 333                 header.typeflag = FIFOTYPE;
 334         } else if (S_ISREG(statbuf->st_mode)) {
 335                 if (sizeof(statbuf->st_size) > 4
 336                  && statbuf->st_size > (off_t)0777777777777LL
 337                 ) {
 338                         bb_error_msg_and_die("can't store file '%s' "
 339                                 "of size %"OFF_FMT"u, aborting",
 340                                 fileName, statbuf->st_size);
 341                 }
 342                 header.typeflag = REGTYPE;
 343                 PUT_OCTAL(header.size, statbuf->st_size);
 344         } else {
 345                 bb_error_msg("%s: unknown file type", fileName);
 346                 return FALSE;
 347         }
 348
 349 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 350         /* Write out long name if needed */
 351         /* (we, like GNU tar, output long linkname *before* long name) */
 352         if (header.name[sizeof(header.name)-1])
 353                 writeLongname(tbInfo->tarFd, GNULONGNAME,
 354                                 header_name, S_ISDIR(statbuf->st_mode));
 355 #endif
 356
 357         /* Now write the header out to disk */
 358         chksum_and_xwrite(tbInfo->tarFd, &header);
 359
 360         /* Now do the verbose thing (or not) */
 361         if (tbInfo->verboseFlag) {
 362                 FILE *vbFd = stdout;
 363
 364                 /* If archive goes to stdout, verbose goes to stderr */
 365                 if (tbInfo->tarFd == STDOUT_FILENO)
 366                         vbFd = stderr;
 367                 /* GNU "tar cvvf" prints "extended" listing a-la "ls -l" */
 368                 /* We don't have such excesses here: for us "v" == "vv" */
 369                 /* '/' is probably a GNUism */
 370                 fprintf(vbFd, "%s%s\n", header_name,
 371                                 S_ISDIR(statbuf->st_mode) ? "/" : "");
 372         }
 373
 374         return TRUE;
 375 }
 376
 377 #if ENABLE_FEATURE_TAR_FROM
 378 static int exclude_file(const llist_t *excluded_files, const char *file)
 379 {
 380         while (excluded_files) {
 381                 if (excluded_files->data[0] == '/') {
 382                         if (fnmatch(excluded_files->data, file,
 383                                         FNM_PATHNAME | FNM_LEADING_DIR) == 0)
 384                                 return 1;
 385                 } else {
 386                         const char *p;
 387
 388                         for (p = file; p[0] != '\0'; p++) {
 389                                 if ((p == file || p[-1] == '/')
 390                                  && p[0] != '/'
 391                                  && fnmatch(excluded_files->data, p,
 392                                                 FNM_PATHNAME | FNM_LEADING_DIR) == 0
 393                                 ) {
 394                                         return 1;
 395                                 }
 396                         }
 397                 }
 398                 excluded_files = excluded_files->link;
 399         }
 400
 401         return 0;
 402 }
 403 #else
 404 # define exclude_file(excluded_files, file) 0
 405 #endif
 406
 407 static int FAST_FUNC writeFileToTarball(const char *fileName, struct stat *statbuf,
 408                         void *userData, int depth UNUSED_PARAM)
 409 {
 410         struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData;
 411         const char *header_name;
 412         int inputFileFd = -1;
 413
 414         DBG("writeFileToTarball('%s')", fileName);
 415
 416         /* Strip leading '/' (must be before memorizing hardlink's name) */
 417         header_name = fileName;
 418         while (header_name[0] == '/') {
 419                 static smallint warned;
 420
 421                 if (!warned) {
 422                         bb_error_msg("removing leading '/' from member names");
 423                         warned = 1;
 424                 }
 425                 header_name++;
 426         }
 427
 428         if (header_name[0] == '\0')
 429                 return TRUE;
 430
 431         /* It is against the rules to archive a socket */
 432         if (S_ISSOCK(statbuf->st_mode)) {
 433                 bb_error_msg("%s: socket ignored", fileName);
 434                 return TRUE;
 435         }
 436
 437         /*
 438          * Check to see if we are dealing with a hard link.
 439          * If so -
 440          * Treat the first occurance of a given dev/inode as a file while
 441          * treating any additional occurances as hard links.  This is done
 442          * by adding the file information to the HardLinkInfo linked list.
 443          */
 444         tbInfo->hlInfo = NULL;
 445         if (!S_ISDIR(statbuf->st_mode) && statbuf->st_nlink > 1) {
 446                 DBG("'%s': st_nlink > 1", header_name);
 447                 tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf);
 448                 if (tbInfo->hlInfo == NULL) {
 449                         DBG("'%s': addHardLinkInfo", header_name);
 450                         addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, header_name);
 451                 }
 452         }
 453
 454         /* It is a bad idea to store the archive we are in the process of creating,
 455          * so check the device and inode to be sure that this particular file isn't
 456          * the new tarball */
 457         if (tbInfo->tarFileStatBuf.st_dev == statbuf->st_dev
 458          && tbInfo->tarFileStatBuf.st_ino == statbuf->st_ino
 459         ) {
 460                 bb_error_msg("%s: file is the archive; skipping", fileName);
 461                 return TRUE;
 462         }
 463
 464         if (exclude_file(tbInfo->excludeList, header_name))
 465                 return SKIP;
 466
 467 #if !ENABLE_FEATURE_TAR_GNU_EXTENSIONS
 468         if (strlen(header_name) >= NAME_SIZE) {
 469                 bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported");
 470                 return TRUE;
 471         }
 472 #endif
 473
 474         /* Is this a regular file? */
 475         if (tbInfo->hlInfo == NULL && S_ISREG(statbuf->st_mode)) {
 476                 /* open the file we want to archive, and make sure all is well */
 477                 inputFileFd = open_or_warn(fileName, O_RDONLY);
 478                 if (inputFileFd < 0) {
 479                         return FALSE;
 480                 }
 481         }
 482
 483         /* Add an entry to the tarball */
 484         if (writeTarHeader(tbInfo, header_name, fileName, statbuf) == FALSE) {
 485                 return FALSE;
 486         }
 487
 488         /* If it was a regular file, write out the body */
 489         if (inputFileFd >= 0) {
 490                 size_t readSize;
 491                 /* Write the file to the archive. */
 492                 /* We record size into header first, */
 493                 /* and then write out file. If file shrinks in between, */
 494                 /* tar will be corrupted. So we don't allow for that. */
 495                 /* NB: GNU tar 1.16 warns and pads with zeroes */
 496                 /* or even seeks back and updates header */
 497                 bb_copyfd_exact_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
 498                 ////off_t readSize;
 499                 ////readSize = bb_copyfd_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
 500                 ////if (readSize != statbuf->st_size && readSize >= 0) {
 501                 ////    bb_error_msg_and_die("short read from %s, aborting", fileName);
 502                 ////}
 503
 504                 /* Check that file did not grow in between? */
 505                 /* if (safe_read(inputFileFd, 1) == 1) warn but continue? */
 506
 507                 close(inputFileFd);
 508
 509                 /* Pad the file up to the tar block size */
 510                 /* (a few tricks here in the name of code size) */
 511                 readSize = (-(int)statbuf->st_size) & (TAR_BLOCK_SIZE-1);
 512                 memset(block_buf, 0, readSize);
 513                 xwrite(tbInfo->tarFd, block_buf, readSize);
 514         }
 515
 516         return TRUE;
 517 }
 518
 519 #if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2
 520 # if !(ENABLE_FEATURE_SEAMLESS_GZ && ENABLE_FEATURE_SEAMLESS_BZ2)
 521 #  define vfork_compressor(tar_fd, gzip) vfork_compressor(tar_fd)
 522 # endif
 523 /* Don't inline: vfork scares gcc and pessimizes code */
 524 static void NOINLINE vfork_compressor(int tar_fd, int gzip)
 525 {
 526         pid_t gzipPid;
 527 # if ENABLE_FEATURE_SEAMLESS_GZ && ENABLE_FEATURE_SEAMLESS_BZ2
 528         const char *zip_exec = (gzip == 1) ? "gzip" : "bzip2";
 529 # elif ENABLE_FEATURE_SEAMLESS_GZ
 530         const char *zip_exec = "gzip";
 531 # else /* only ENABLE_FEATURE_SEAMLESS_BZ2 */
 532         const char *zip_exec = "bzip2";
 533 # endif
 534         // On Linux, vfork never unpauses parent early, although standard
 535         // allows for that. Do we want to waste bytes checking for it?
 536 # define WAIT_FOR_CHILD 0
 537         volatile int vfork_exec_errno = 0;
 538         struct fd_pair gzipDataPipe;
 539 # if WAIT_FOR_CHILD
 540         struct fd_pair gzipStatusPipe;
 541         xpiped_pair(gzipStatusPipe);
 542 # endif
 543         xpiped_pair(gzipDataPipe);
 544
 545         signal(SIGPIPE, SIG_IGN); /* we only want EPIPE on errors */
 546
 547 # if defined(__GNUC__) && __GNUC__
 548         /* Avoid vfork clobbering */
 549         (void) &zip_exec;
 550 # endif
 551
 552         gzipPid = vfork();
 553         if (gzipPid < 0)
 554                 bb_perror_msg_and_die("vfork");
 555
 556         if (gzipPid == 0) {
 557                 /* child */
 558                 /* NB: close _first_, then move fds! */
 559                 close(gzipDataPipe.wr);
 560 # if WAIT_FOR_CHILD
 561                 close(gzipStatusPipe.rd);
 562                 /* gzipStatusPipe.wr will close only on exec -
 563                  * parent waits for this close to happen */
 564                 fcntl(gzipStatusPipe.wr, F_SETFD, FD_CLOEXEC);
 565 # endif
 566                 xmove_fd(gzipDataPipe.rd, 0);
 567                 xmove_fd(tar_fd, 1);
 568                 /* exec gzip/bzip2 program/applet */
 569                 BB_EXECLP(zip_exec, zip_exec, "-f", NULL);
 570                 vfork_exec_errno = errno;
 571                 _exit(EXIT_FAILURE);
 572         }
 573
 574         /* parent */
 575         xmove_fd(gzipDataPipe.wr, tar_fd);
 576         close(gzipDataPipe.rd);
 577 # if WAIT_FOR_CHILD
 578         close(gzipStatusPipe.wr);
 579         while (1) {
 580                 char buf;
 581                 int n;
 582
 583                 /* Wait until child execs (or fails to) */
 584                 n = full_read(gzipStatusPipe.rd, &buf, 1);
 585                 if (n < 0 /* && errno == EAGAIN */)
 586                         continue;       /* try it again */
 587         }
 588         close(gzipStatusPipe.rd);
 589 # endif
 590         if (vfork_exec_errno) {
 591                 errno = vfork_exec_errno;
 592                 bb_perror_msg_and_die("can't execute '%s'", zip_exec);
 593         }
 594 }
 595 #endif /* ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 */
 596
 597
 598 /* gcc 4.2.1 inlines it, making code bigger */
 599 static NOINLINE int writeTarFile(int tar_fd, int verboseFlag,
 600         int dereferenceFlag, const llist_t *include,
 601         const llist_t *exclude, int gzip)
 602 {
 603         int errorFlag = FALSE;
 604         struct TarBallInfo tbInfo;
 605
 606         tbInfo.hlInfoHead = NULL;
 607         tbInfo.tarFd = tar_fd;
 608         tbInfo.verboseFlag = verboseFlag;
 609
 610         /* Store the stat info for the tarball's file, so
 611          * can avoid including the tarball into itself....  */
 612         if (fstat(tbInfo.tarFd, &tbInfo.tarFileStatBuf) < 0)
 613                 bb_perror_msg_and_die("can't stat tar file");
 614
 615 #if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2
 616         if (gzip)
 617                 vfork_compressor(tbInfo.tarFd, gzip);
 618 #endif
 619
 620         tbInfo.excludeList = exclude;
 621
 622         /* Read the directory/files and iterate over them one at a time */
 623         while (include) {
 624                 if (!recursive_action(include->data, ACTION_RECURSE |
 625                                 (dereferenceFlag ? ACTION_FOLLOWLINKS : 0),
 626                                 writeFileToTarball, writeFileToTarball, &tbInfo, 0)
 627                 ) {
 628                         errorFlag = TRUE;
 629                 }
 630                 include = include->link;
 631         }
 632         /* Write two empty blocks to the end of the archive */
 633         memset(block_buf, 0, 2*TAR_BLOCK_SIZE);
 634         xwrite(tbInfo.tarFd, block_buf, 2*TAR_BLOCK_SIZE);
 635
 636         /* To be pedantically correct, we would check if the tarball
 637          * is smaller than 20 tar blocks, and pad it if it was smaller,
 638          * but that isn't necessary for GNU tar interoperability, and
 639          * so is considered a waste of space */
 640
 641         /* Close so the child process (if any) will exit */
 642         close(tbInfo.tarFd);
 643
 644         /* Hang up the tools, close up shop, head home */
 645         if (ENABLE_FEATURE_CLEAN_UP)
 646                 freeHardLinkInfo(&tbInfo.hlInfoHead);
 647
 648         if (errorFlag)
 649                 bb_error_msg("error exit delayed from previous errors");
 650
 651 #if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2
 652         if (gzip) {
 653                 int status;
 654                 if (safe_waitpid(-1, &status, 0) == -1)
 655                         bb_perror_msg("waitpid");
 656                 else if (!WIFEXITED(status) || WEXITSTATUS(status))
 657                         /* gzip was killed or has exited with nonzero! */
 658                         errorFlag = TRUE;
 659         }
 660 #endif
 661         return errorFlag;
 662 }
 663 #else
 664 int writeTarFile(int tar_fd, int verboseFlag,
 665         int dereferenceFlag, const llist_t *include,
 666         const llist_t *exclude, int gzip);
 667 #endif /* FEATURE_TAR_CREATE */
 668
 669 #if ENABLE_FEATURE_TAR_FROM
 670 static llist_t *append_file_list_to_list(llist_t *list)
 671 {
 672         FILE *src_stream;
 673         char *line;
 674         llist_t *newlist = NULL;
 675
 676         while (list) {
 677                 src_stream = xfopen_for_read(llist_pop(&list));
 678                 while ((line = xmalloc_fgetline(src_stream)) != NULL) {
 679                         /* kill trailing '/' unless the string is just "/" */
 680                         char *cp = last_char_is(line, '/');
 681                         if (cp > line)
 682                                 *cp = '\0';
 683                         llist_add_to(&newlist, line);
 684                 }
 685                 fclose(src_stream);
 686         }
 687         return newlist;
 688 }
 689 #else
 690 # define append_file_list_to_list(x) 0
 691 #endif
 692
 693 #if ENABLE_FEATURE_SEAMLESS_Z
 694 static char FAST_FUNC get_header_tar_Z(archive_handle_t *archive_handle)
 695 {
 696         /* Can't lseek over pipes */
 697         archive_handle->seek = seek_by_read;
 698
 699         /* do the decompression, and cleanup */
 700         if (xread_char(archive_handle->src_fd) != 0x1f
 701          || xread_char(archive_handle->src_fd) != 0x9d
 702         ) {
 703                 bb_error_msg_and_die("invalid magic");
 704         }
 705
 706         open_transformer(archive_handle->src_fd, unpack_Z_stream, "uncompress");
 707         archive_handle->offset = 0;
 708         while (get_header_tar(archive_handle) == EXIT_SUCCESS)
 709                 continue;
 710
 711         /* Can only do one file at a time */
 712         return EXIT_FAILURE;
 713 }
 714 #else
 715 # define get_header_tar_Z NULL
 716 #endif
 717
 718 #ifdef CHECK_FOR_CHILD_EXITCODE
 719 /* Looks like it isn't needed - tar detects malformed (truncated)
 720  * archive if e.g. bunzip2 fails */
 721 static int child_error;
 722
 723 static void handle_SIGCHLD(int status)
 724 {
 725         /* Actually, 'status' is a signo. We reuse it for other needs */
 726
 727         /* Wait for any child without blocking */
 728         if (wait_any_nohang(&status) < 0)
 729                 /* wait failed?! I'm confused... */
 730                 return;
 731
 732         if (WIFEXITED(status) && WEXITSTATUS(status) == 0)
 733                 /* child exited with 0 */
 734                 return;
 735         /* Cannot happen?
 736         if (!WIFSIGNALED(status) && !WIFEXITED(status)) return; */
 737         child_error = 1;
 738 }
 739 #endif
 740
 741 enum {
 742         OPTBIT_KEEP_OLD = 8,
 743         IF_FEATURE_TAR_CREATE(   OPTBIT_CREATE      ,)
 744         IF_FEATURE_TAR_CREATE(   OPTBIT_DEREFERENCE ,)
 745         IF_FEATURE_SEAMLESS_BZ2( OPTBIT_BZIP2       ,)
 746         IF_FEATURE_SEAMLESS_LZMA(OPTBIT_LZMA        ,)
 747         IF_FEATURE_TAR_FROM(     OPTBIT_INCLUDE_FROM,)
 748         IF_FEATURE_TAR_FROM(     OPTBIT_EXCLUDE_FROM,)
 749         IF_FEATURE_SEAMLESS_GZ(  OPTBIT_GZIP        ,)
 750         IF_FEATURE_SEAMLESS_Z(   OPTBIT_COMPRESS    ,) // 16th bit
 751 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
 752         OPTBIT_NUMERIC_OWNER,
 753         OPTBIT_NOPRESERVE_PERM,
 754         OPTBIT_OVERWRITE,
 755 #endif
 756         OPT_TEST         = 1 << 0, // t
 757         OPT_EXTRACT      = 1 << 1, // x
 758         OPT_BASEDIR      = 1 << 2, // C
 759         OPT_TARNAME      = 1 << 3, // f
 760         OPT_2STDOUT      = 1 << 4, // O
 761         OPT_NOPRESERVE_OWNER = 1 << 5, // o == no-same-owner
 762         OPT_P            = 1 << 6, // p
 763         OPT_VERBOSE      = 1 << 7, // v
 764         OPT_KEEP_OLD     = 1 << 8, // k
 765         OPT_CREATE       = IF_FEATURE_TAR_CREATE(   (1 << OPTBIT_CREATE      )) + 0, // c
 766         OPT_DEREFERENCE  = IF_FEATURE_TAR_CREATE(   (1 << OPTBIT_DEREFERENCE )) + 0, // h
 767         OPT_BZIP2        = IF_FEATURE_SEAMLESS_BZ2( (1 << OPTBIT_BZIP2       )) + 0, // j
 768         OPT_LZMA         = IF_FEATURE_SEAMLESS_LZMA((1 << OPTBIT_LZMA        )) + 0, // a
 769         OPT_INCLUDE_FROM = IF_FEATURE_TAR_FROM(     (1 << OPTBIT_INCLUDE_FROM)) + 0, // T
 770         OPT_EXCLUDE_FROM = IF_FEATURE_TAR_FROM(     (1 << OPTBIT_EXCLUDE_FROM)) + 0, // X
 771         OPT_GZIP         = IF_FEATURE_SEAMLESS_GZ(  (1 << OPTBIT_GZIP        )) + 0, // z
 772         OPT_COMPRESS     = IF_FEATURE_SEAMLESS_Z(   (1 << OPTBIT_COMPRESS    )) + 0, // Z
 773         OPT_NUMERIC_OWNER   = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_NUMERIC_OWNER  )) + 0, // numeric-owner
 774         OPT_NOPRESERVE_PERM = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_NOPRESERVE_PERM)) + 0, // no-same-permissions
 775         OPT_OVERWRITE       = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_OVERWRITE      )) + 0, // overwrite
 776 };
 777 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
 778 static const char tar_longopts[] ALIGN1 =
 779         "list\0"                No_argument       "t"
 780         "extract\0"             No_argument       "x"
 781         "directory\0"           Required_argument "C"
 782         "file\0"                Required_argument "f"
 783         "to-stdout\0"           No_argument       "O"
 784         /* do not restore owner */
 785         /* Note: GNU tar handles 'o' as no-same-owner only on extract,
 786          * on create, 'o' is --old-archive. We do not support --old-archive. */
 787         "no-same-owner\0"       No_argument       "o"
 788         "same-permissions\0"    No_argument       "p"
 789         "verbose\0"             No_argument       "v"
 790         "keep-old\0"            No_argument       "k"
 791 # if ENABLE_FEATURE_TAR_CREATE
 792         "create\0"              No_argument       "c"
 793         "dereference\0"         No_argument       "h"
 794 # endif
 795 # if ENABLE_FEATURE_SEAMLESS_BZ2
 796         "bzip2\0"               No_argument       "j"
 797 # endif
 798 # if ENABLE_FEATURE_SEAMLESS_LZMA
 799         "lzma\0"                No_argument       "a"
 800 # endif
 801 # if ENABLE_FEATURE_TAR_FROM
 802         "files-from\0"          Required_argument "T"
 803         "exclude-from\0"        Required_argument "X"
 804 # endif
 805 # if ENABLE_FEATURE_SEAMLESS_GZ
 806         "gzip\0"                No_argument       "z"
 807 # endif
 808 # if ENABLE_FEATURE_SEAMLESS_Z
 809         "compress\0"            No_argument       "Z"
 810 # endif
 811         /* use numeric uid/gid from tar header, not textual */
 812         "numeric-owner\0"       No_argument       "\xfc"
 813         /* do not restore mode */
 814         "no-same-permissions\0" No_argument       "\xfd"
 815         /* on unpack, open with O_TRUNC and !O_EXCL */
 816         "overwrite\0"           No_argument       "\xfe"
 817         /* --exclude takes next bit position in option mask, */
 818         /* therefore we have to put it _after_ --no-same-permissions */
 819 # if ENABLE_FEATURE_TAR_FROM
 820         "exclude\0"             Required_argument "\xff"
 821 # endif
 822         ;
 823 #endif
 824
 825 int tar_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 826 int tar_main(int argc UNUSED_PARAM, char **argv)
 827 {
 828         char FAST_FUNC (*get_header_ptr)(archive_handle_t *) = get_header_tar;
 829         archive_handle_t *tar_handle;
 830         char *base_dir = NULL;
 831         const char *tar_filename = "-";
 832         unsigned opt;
 833         int verboseFlag = 0;
 834 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
 835         llist_t *excludes = NULL;
 836 #endif
 837
 838         /* Initialise default values */
 839         tar_handle = init_handle();
 840         tar_handle->ah_flags = ARCHIVE_CREATE_LEADING_DIRS
 841                              | ARCHIVE_RESTORE_DATE
 842                              | ARCHIVE_UNLINK_OLD;
 843
 844         /* Apparently only root's tar preserves perms (see bug 3844) */
 845         if (getuid() != 0)
 846                 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM;
 847
 848         /* Prepend '-' to the first argument if required */
 849         opt_complementary = "--:" // first arg is options
 850                 "tt:vv:" // count -t,-v
 851                 "?:" // bail out with usage instead of error return
 852                 "X::T::" // cumulative lists
 853 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
 854                 "\xff::" // cumulative lists for --exclude
 855 #endif
 856                 IF_FEATURE_TAR_CREATE("c:") "t:x:" // at least one of these is reqd
 857                 IF_FEATURE_TAR_CREATE("c--tx:t--cx:x--ct") // mutually exclusive
 858                 IF_NOT_FEATURE_TAR_CREATE("t--x:x--t"); // mutually exclusive
 859 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
 860         applet_long_options = tar_longopts;
 861 #endif
 862 #if ENABLE_DESKTOP
 863         if (argv[1] && argv[1][0] != '-') {
 864                 /* Compat:
 865                  * 1st argument without dash handles options with parameters
 866                  * differently from dashed one: it takes *next argv[i]*
 867                  * as paramenter even if there are more chars in 1st argument:
 868                  *  "tar fx TARFILE" - "x" is not taken as f's param
 869                  *  but is interpreted as -x option
 870                  *  "tar -xf TARFILE" - dashed equivalent of the above
 871                  *  "tar -fx ..." - "x" is taken as f's param
 872                  * getopt32 wouldn't handle 1st command correctly.
 873                  * Unfortunately, people do use such commands.
 874                  * We massage argv[1] to work around it by moving 'f'
 875                  * to the end of the string.
 876                  * More contrived "tar fCx TARFILE DIR" still fails,
 877                  * but such commands are much less likely to be used.
 878                  */
 879                 char *f = strchr(argv[1], 'f');
 880                 if (f) {
 881                         while (f[1] != '\0') {
 882                                 *f = f[1];
 883                                 f++;
 884                         }
 885                         *f = 'f';
 886                 }
 887         }
 888 #endif
 889         opt = getopt32(argv,
 890                 "txC:f:Oopvk"
 891                 IF_FEATURE_TAR_CREATE(   "ch"  )
 892                 IF_FEATURE_SEAMLESS_BZ2( "j"   )
 893                 IF_FEATURE_SEAMLESS_LZMA("a"   )
 894                 IF_FEATURE_TAR_FROM(     "T:X:")
 895                 IF_FEATURE_SEAMLESS_GZ(  "z"   )
 896                 IF_FEATURE_SEAMLESS_Z(   "Z"   )
 897                 , &base_dir // -C dir
 898                 , &tar_filename // -f filename
 899                 IF_FEATURE_TAR_FROM(, &(tar_handle->accept)) // T
 900                 IF_FEATURE_TAR_FROM(, &(tar_handle->reject)) // X
 901 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
 902                 , &excludes // --exclude
 903 #endif
 904                 , &verboseFlag // combined count for -t and -v
 905                 , &verboseFlag // combined count for -t and -v
 906                 );
 907         //bb_error_msg("opt:%08x", opt);
 908         argv += optind;
 909
 910         if (verboseFlag) tar_handle->action_header = header_verbose_list;
 911         if (verboseFlag == 1) tar_handle->action_header = header_list;
 912
 913         if (opt & OPT_EXTRACT)
 914                 tar_handle->action_data = data_extract_all;
 915
 916         if (opt & OPT_2STDOUT)
 917                 tar_handle->action_data = data_extract_to_stdout;
 918
 919         if (opt & OPT_KEEP_OLD)
 920                 tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD;
 921
 922         if (opt & OPT_NUMERIC_OWNER)
 923                 tar_handle->ah_flags |= ARCHIVE_NUMERIC_OWNER;
 924
 925         if (opt & OPT_NOPRESERVE_OWNER)
 926                 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_OWNER;
 927
 928         if (opt & OPT_NOPRESERVE_PERM)
 929                 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM;
 930
 931         if (opt & OPT_OVERWRITE) {
 932                 tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD;
 933                 tar_handle->ah_flags |= ARCHIVE_O_TRUNC;
 934         }
 935
 936         if (opt & OPT_GZIP)
 937                 get_header_ptr = get_header_tar_gz;
 938
 939         if (opt & OPT_BZIP2)
 940                 get_header_ptr = get_header_tar_bz2;
 941
 942         if (opt & OPT_LZMA)
 943                 get_header_ptr = get_header_tar_lzma;
 944
 945         if (opt & OPT_COMPRESS)
 946                 get_header_ptr = get_header_tar_Z;
 947
 948 #if ENABLE_FEATURE_TAR_FROM
 949         tar_handle->reject = append_file_list_to_list(tar_handle->reject);
 950 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
 951         /* Append excludes to reject */
 952         while (excludes) {
 953                 llist_t *next = excludes->link;
 954                 excludes->link = tar_handle->reject;
 955                 tar_handle->reject = excludes;
 956                 excludes = next;
 957         }
 958 #endif
 959         tar_handle->accept = append_file_list_to_list(tar_handle->accept);
 960 #endif
 961
 962         /* Setup an array of filenames to work with */
 963         /* TODO: This is the same as in ar, separate function ? */
 964         while (*argv) {
 965                 /* kill trailing '/' unless the string is just "/" */
 966                 char *cp = last_char_is(*argv, '/');
 967                 if (cp > *argv)
 968                         *cp = '\0';
 969                 llist_add_to_end(&tar_handle->accept, *argv);
 970                 argv++;
 971         }
 972
 973         if (tar_handle->accept || tar_handle->reject)
 974                 tar_handle->filter = filter_accept_reject_list;
 975
 976         /* Open the tar file */
 977         {
 978                 int tar_fd = STDIN_FILENO;
 979                 int flags = O_RDONLY;
 980
 981                 if (opt & OPT_CREATE) {
 982                         /* Make sure there is at least one file to tar up.  */
 983                         if (tar_handle->accept == NULL)
 984                                 bb_error_msg_and_die("empty archive");
 985
 986                         tar_fd = STDOUT_FILENO;
 987                         /* Mimicking GNU tar 1.15.1: */
 988                         flags = O_WRONLY | O_CREAT | O_TRUNC;
 989                 }
 990
 991                 if (LONE_DASH(tar_filename)) {
 992                         tar_handle->src_fd = tar_fd;
 993                         tar_handle->seek = seek_by_read;
 994                 } else {
 995                         if (ENABLE_FEATURE_TAR_AUTODETECT && flags == O_RDONLY) {
 996                                 get_header_ptr = get_header_tar;
 997                                 tar_handle->src_fd = open_zipped(tar_filename);
 998                                 if (tar_handle->src_fd < 0)
 999                                         bb_perror_msg_and_die("can't open '%s'", tar_filename);
1000                         } else {
1001                                 tar_handle->src_fd = xopen(tar_filename, flags);
1002                         }
1003                 }
1004         }
1005
1006         if (base_dir)
1007                 xchdir(base_dir);
1008
1009 #ifdef CHECK_FOR_CHILD_EXITCODE
1010         /* We need to know whether child (gzip/bzip/etc) exits abnormally */
1011         signal(SIGCHLD, handle_SIGCHLD);
1012 #endif
1013
1014         /* create an archive */
1015         if (opt & OPT_CREATE) {
1016 #if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2
1017                 int zipMode = 0;
1018                 if (ENABLE_FEATURE_SEAMLESS_GZ && (opt & OPT_GZIP))
1019                         zipMode = 1;
1020                 if (ENABLE_FEATURE_SEAMLESS_BZ2 && (opt & OPT_BZIP2))
1021                         zipMode = 2;
1022 #endif
1023                 /* NB: writeTarFile() closes tar_handle->src_fd */
1024                 return writeTarFile(tar_handle->src_fd, verboseFlag, opt & OPT_DEREFERENCE,
1025                                 tar_handle->accept,
1026                                 tar_handle->reject, zipMode);
1027         }
1028
1029         while (get_header_ptr(tar_handle) == EXIT_SUCCESS)
1030                 continue;
1031
1032         /* Check that every file that should have been extracted was */
1033         while (tar_handle->accept) {
1034                 if (!find_list_entry(tar_handle->reject, tar_handle->accept->data)
1035                  && !find_list_entry(tar_handle->passed, tar_handle->accept->data)
1036                 ) {
1037                         bb_error_msg_and_die("%s: not found in archive",
1038                                 tar_handle->accept->data);
1039                 }
1040                 tar_handle->accept = tar_handle->accept->link;
1041         }
1042         if (ENABLE_FEATURE_CLEAN_UP /* && tar_handle->src_fd != STDIN_FILENO */)
1043                 close(tar_handle->src_fd);
1044
1045         return EXIT_SUCCESS;
1046 }