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