1 /* vi: set sw=4 ts=4: */
3 * Mini tar implementation for busybox
5 * Modified to use common extraction code used by ar, cpio, dpkg-deb, dpkg
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.)
12 * Copyright (C) 1999-2004 by Erik Andersen <andersen@codepoet.org>
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 GPL has been granted.
20 * Based in part on the tar implementation from busybox-0.28
21 * Copyright (C) 1995 Bruce Perens
23 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
27 //config: bool "tar (40 kb)"
30 //config: tar is an archiving program. It's commonly used with gzip to
31 //config: create compressed archives. It's probably the most widely used
32 //config: UNIX archive program.
34 //config:config FEATURE_TAR_LONG_OPTIONS
35 //config: bool "Enable long options"
37 //config: depends on TAR && LONG_OPTS
39 //config:config FEATURE_TAR_CREATE
40 //config: bool "Enable -c (archive creation)"
42 //config: depends on TAR
44 //config:config FEATURE_TAR_AUTODETECT
45 //config: bool "Autodetect compressed tarballs"
47 //config: depends on TAR && (FEATURE_SEAMLESS_Z || FEATURE_SEAMLESS_GZ || FEATURE_SEAMLESS_BZ2 || FEATURE_SEAMLESS_LZMA || FEATURE_SEAMLESS_XZ)
49 //config: With this option tar can automatically detect compressed
50 //config: tarballs. Currently it works only on files (not pipes etc).
52 //config:config FEATURE_TAR_FROM
53 //config: bool "Enable -X (exclude from) and -T (include from) options)"
55 //config: depends on TAR
57 //config: If you enable this option you'll be able to specify
58 //config: a list of files to include or exclude from an archive.
60 //config:config FEATURE_TAR_OLDGNU_COMPATIBILITY
61 //config: bool "Support old tar header format"
63 //config: depends on TAR || DPKG
65 //config: This option is required to unpack archives created in
66 //config: the old GNU format; help to kill this old format by
67 //config: repacking your ancient archives with the new format.
69 //config:config FEATURE_TAR_OLDSUN_COMPATIBILITY
70 //config: bool "Enable untarring of tarballs with checksums produced by buggy Sun tar"
72 //config: depends on TAR || DPKG
74 //config: This option is required to unpack archives created by some old
75 //config: version of Sun's tar (it was calculating checksum using signed
76 //config: arithmetic). It is said to be fixed in newer Sun tar, but "old"
77 //config: tarballs still exist.
79 //config:config FEATURE_TAR_GNU_EXTENSIONS
80 //config: bool "Support GNU tar extensions (long filenames)"
82 //config: depends on TAR || DPKG
84 //config:config FEATURE_TAR_TO_COMMAND
85 //config: bool "Support writing to an external program (--to-command)"
87 //config: depends on TAR && FEATURE_TAR_LONG_OPTIONS
89 //config: If you enable this option you'll be able to instruct tar to send
90 //config: the contents of each extracted file to the standard input of an
91 //config: external program.
93 //config:config FEATURE_TAR_UNAME_GNAME
94 //config: bool "Enable use of user and group names"
96 //config: depends on TAR
98 //config: Enable use of user and group names in tar. This affects contents
99 //config: listings (-t) and preserving permissions when unpacking (-p).
100 //config: +200 bytes.
102 //config:config FEATURE_TAR_NOPRESERVE_TIME
103 //config: bool "Enable -m (do not preserve time) GNU option"
105 //config: depends on TAR
107 //config:config FEATURE_TAR_SELINUX
108 //config: bool "Support extracting SELinux labels"
110 //config: depends on TAR && SELINUX
112 //config: With this option busybox supports restoring SELinux labels
113 //config: when extracting files from tar archives.
115 //applet:IF_TAR(APPLET(tar, BB_DIR_BIN, BB_SUID_DROP))
116 //kbuild:lib-$(CONFIG_TAR) += tar.o
120 #include "common_bufsiz.h"
121 #include "bb_archive.h"
122 /* FIXME: Stop using this non-standard feature */
123 #ifndef FNM_LEADING_DIR
124 # define FNM_LEADING_DIR 0
128 # define DBG(fmt, ...) bb_error_msg("%s: " fmt, __func__, ## __VA_ARGS__)
130 # define DBG(...) ((void)0)
132 #define DBG_OPTION_PARSING 0
135 #define block_buf bb_common_bufsiz1
136 #define INIT_G() do { setup_common_bufsiz(); } while (0)
139 #if ENABLE_FEATURE_TAR_CREATE
142 ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are
143 ** the only functions that deal with the HardLinkInfo structure.
144 ** Even these functions use the xxxHardLinkInfo() functions.
146 typedef struct HardLinkInfo {
147 struct HardLinkInfo *next; /* Next entry in list */
148 dev_t dev; /* Device number */
149 ino_t ino; /* Inode number */
150 // short linkCount; /* (Hard) Link Count */
151 char name[1]; /* Start of filename (must be last) */
154 /* Some info to be carried along when creating a new tarball */
155 typedef struct TarBallInfo {
156 int tarFd; /* Open-for-write file descriptor
158 int verboseFlag; /* Whether to print extra stuff or not */
159 const llist_t *excludeList; /* List of files to not include */
160 HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */
161 HardLinkInfo *hlInfo; /* Hard Link Info for the current file */
162 //TODO: save only st_dev + st_ino
163 struct stat tarFileStatBuf; /* Stat info for the tarball, letting
164 * us know the inode and device that the
165 * tarball lives, so we can avoid trying
166 * to include the tarball into itself */
169 /* A nice enum with all the possible tar file content types */
171 REGTYPE = '0', /* regular file */
172 REGTYPE0 = '\0', /* regular file (ancient bug compat) */
173 LNKTYPE = '1', /* hard link */
174 SYMTYPE = '2', /* symbolic link */
175 CHRTYPE = '3', /* character special */
176 BLKTYPE = '4', /* block special */
177 DIRTYPE = '5', /* directory */
178 FIFOTYPE = '6', /* FIFO special */
179 CONTTYPE = '7', /* reserved */
180 GNULONGLINK = 'K', /* GNU long (>100 chars) link name */
181 GNULONGNAME = 'L', /* GNU long (>100 chars) file name */
184 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
185 static void addHardLinkInfo(HardLinkInfo **hlInfoHeadPtr,
186 struct stat *statbuf,
187 const char *fileName)
189 /* Note: hlInfoHeadPtr can never be NULL! */
190 HardLinkInfo *hlInfo;
192 hlInfo = xmalloc(sizeof(HardLinkInfo) + strlen(fileName));
193 hlInfo->next = *hlInfoHeadPtr;
194 *hlInfoHeadPtr = hlInfo;
195 hlInfo->dev = statbuf->st_dev;
196 hlInfo->ino = statbuf->st_ino;
197 // hlInfo->linkCount = statbuf->st_nlink;
198 strcpy(hlInfo->name, fileName);
201 static void freeHardLinkInfo(HardLinkInfo **hlInfoHeadPtr)
203 HardLinkInfo *hlInfo;
204 HardLinkInfo *hlInfoNext;
207 hlInfo = *hlInfoHeadPtr;
209 hlInfoNext = hlInfo->next;
213 *hlInfoHeadPtr = NULL;
217 /* Might be faster (and bigger) if the dev/ino were stored in numeric order ;) */
218 static HardLinkInfo *findHardLinkInfo(HardLinkInfo *hlInfo, struct stat *statbuf)
221 if (statbuf->st_ino == hlInfo->ino
222 && statbuf->st_dev == hlInfo->dev
224 DBG("found hardlink:'%s'", hlInfo->name);
227 hlInfo = hlInfo->next;
232 /* Put an octal string into the specified buffer.
233 * The number is zero padded and possibly null terminated.
234 * Stores low-order bits only if whole value does not fit. */
235 static void putOctal(char *cp, int len, off_t value)
237 char tempBuffer[sizeof(off_t)*3 + 1];
238 char *tempString = tempBuffer;
241 width = sprintf(tempBuffer, "%0*"OFF_FMT"o", len, value);
242 tempString += (width - len);
244 /* If string has leading zeroes, we can drop one */
245 /* and field will have trailing '\0' */
246 /* (increases chances of compat with other tars) */
247 if (tempString[0] == '0')
250 /* Copy the string to the field */
251 memcpy(cp, tempString, len);
253 #define PUT_OCTAL(a, b) putOctal((a), sizeof(a), (b))
255 static void chksum_and_xwrite(int fd, struct tar_header_t* hp)
257 /* POSIX says that checksum is done on unsigned bytes
258 * (Sun and HP-UX gets it wrong... more details in
260 const unsigned char *cp;
263 strcpy(hp->magic, "ustar ");
265 /* Calculate and store the checksum (i.e., the sum of all of the bytes of
266 * the header). The checksum field must be filled with blanks for the
267 * calculation. The checksum field is formatted differently from the
268 * other fields: it has 6 digits, a null, then a space -- rather than
269 * digits, followed by a null like the other fields... */
270 memset(hp->chksum, ' ', sizeof(hp->chksum));
271 cp = (const unsigned char *) hp;
274 do { chksum += *cp++; } while (--size);
275 putOctal(hp->chksum, sizeof(hp->chksum)-1, chksum);
277 /* Now write the header out to disk */
278 xwrite(fd, hp, sizeof(*hp));
281 static void replace_symlink_placeholders(llist_t *list)
286 target = list->data + strlen(list->data) + 1;
287 if (symlink(target, list->data)) {
289 bb_error_msg_and_die("can't create %slink '%s' to '%s'",
298 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
299 static void writeLongname(int fd, int type, const char *name, int dir)
301 static const struct {
302 char mode[8]; /* 100-107 */
303 char uid[8]; /* 108-115 */
304 char gid[8]; /* 116-123 */
305 char size[12]; /* 124-135 */
306 char mtime[12]; /* 136-147 */
314 struct tar_header_t header;
317 dir = !!dir; /* normalize: 0/1 */
318 size = strlen(name) + 1 + dir; /* GNU tar uses strlen+1 */
319 /* + dir: account for possible '/' */
321 memset(&header, 0, sizeof(header));
322 strcpy(header.name, "././@LongLink");
323 memcpy(header.mode, prefilled.mode, sizeof(prefilled));
324 PUT_OCTAL(header.size, size);
325 header.typeflag = type;
326 chksum_and_xwrite(fd, &header);
328 /* Write filename[/] and pad the block. */
329 /* dir=0: writes 'name<NUL>', pads */
330 /* dir=1: writes 'name', writes '/<NUL>', pads */
332 xwrite(fd, name, size - dir);
333 xwrite(fd, "/", dir);
334 size = (-size) & (TAR_BLOCK_SIZE-1);
335 memset(&header, 0, size);
336 xwrite(fd, &header, size);
340 /* Write out a tar header for the specified file/directory/whatever */
341 static int writeTarHeader(struct TarBallInfo *tbInfo,
342 const char *header_name, const char *fileName, struct stat *statbuf)
344 struct tar_header_t header;
346 memset(&header, 0, sizeof(header));
348 strncpy(header.name, header_name, sizeof(header.name));
350 /* POSIX says to mask mode with 07777. */
351 PUT_OCTAL(header.mode, statbuf->st_mode & 07777);
352 PUT_OCTAL(header.uid, statbuf->st_uid);
353 PUT_OCTAL(header.gid, statbuf->st_gid);
354 memset(header.size, '0', sizeof(header.size)-1); /* Regular file size is handled later */
355 /* users report that files with negative st_mtime cause trouble, so: */
356 PUT_OCTAL(header.mtime, statbuf->st_mtime >= 0 ? statbuf->st_mtime : 0);
358 /* Enter the user and group names */
359 safe_strncpy(header.uname, get_cached_username(statbuf->st_uid), sizeof(header.uname));
360 safe_strncpy(header.gname, get_cached_groupname(statbuf->st_gid), sizeof(header.gname));
362 if (tbInfo->hlInfo) {
363 /* This is a hard link */
364 header.typeflag = LNKTYPE;
365 strncpy(header.linkname, tbInfo->hlInfo->name,
366 sizeof(header.linkname));
367 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
368 /* Write out long linkname if needed */
369 if (header.linkname[sizeof(header.linkname)-1])
370 writeLongname(tbInfo->tarFd, GNULONGLINK,
371 tbInfo->hlInfo->name, 0);
373 } else if (S_ISLNK(statbuf->st_mode)) {
374 char *lpath = xmalloc_readlink_or_warn(fileName);
377 header.typeflag = SYMTYPE;
378 strncpy(header.linkname, lpath, sizeof(header.linkname));
379 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
380 /* Write out long linkname if needed */
381 if (header.linkname[sizeof(header.linkname)-1])
382 writeLongname(tbInfo->tarFd, GNULONGLINK, lpath, 0);
384 /* If it is larger than 100 bytes, bail out */
385 if (header.linkname[sizeof(header.linkname)-1]) {
387 bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported");
392 } else if (S_ISDIR(statbuf->st_mode)) {
393 header.typeflag = DIRTYPE;
394 /* Append '/' only if there is a space for it */
395 if (!header.name[sizeof(header.name)-1])
396 header.name[strlen(header.name)] = '/';
397 } else if (S_ISCHR(statbuf->st_mode)) {
398 header.typeflag = CHRTYPE;
399 PUT_OCTAL(header.devmajor, major(statbuf->st_rdev));
400 PUT_OCTAL(header.devminor, minor(statbuf->st_rdev));
401 } else if (S_ISBLK(statbuf->st_mode)) {
402 header.typeflag = BLKTYPE;
403 PUT_OCTAL(header.devmajor, major(statbuf->st_rdev));
404 PUT_OCTAL(header.devminor, minor(statbuf->st_rdev));
405 } else if (S_ISFIFO(statbuf->st_mode)) {
406 header.typeflag = FIFOTYPE;
407 } else if (S_ISREG(statbuf->st_mode)) {
408 /* header.size field is 12 bytes long */
409 /* Does octal-encoded size fit? */
410 uoff_t filesize = statbuf->st_size;
411 if (sizeof(filesize) <= 4
412 || filesize <= (uoff_t)0777777777777LL
414 PUT_OCTAL(header.size, filesize);
416 /* Does base256-encoded size fit?
417 * It always does unless off_t is wider than 64 bits.
419 else if (ENABLE_FEATURE_TAR_GNU_EXTENSIONS
420 #if ULLONG_MAX > 0xffffffffffffffffLL /* 2^64-1 */
421 && (filesize <= 0x3fffffffffffffffffffffffLL)
424 /* GNU tar uses "base-256 encoding" for very large numbers.
425 * Encoding is binary, with highest bit always set as a marker
426 * and sign in next-highest bit:
428 * bf ff .. ff - largest positive number
429 * ff ff .. ff - minus 1
430 * c0 00 .. 00 - smallest negative number
432 char *p8 = header.size + sizeof(header.size);
434 *--p8 = (uint8_t)filesize;
436 } while (p8 != header.size);
439 bb_error_msg_and_die("can't store file '%s' "
440 "of size %"OFF_FMT"u, aborting",
441 fileName, statbuf->st_size);
443 header.typeflag = REGTYPE;
445 bb_error_msg("%s: unknown file type", fileName);
449 #if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
450 /* Write out long name if needed */
451 /* (we, like GNU tar, output long linkname *before* long name) */
452 if (header.name[sizeof(header.name)-1])
453 writeLongname(tbInfo->tarFd, GNULONGNAME,
454 header_name, S_ISDIR(statbuf->st_mode));
457 /* Now write the header out to disk */
458 chksum_and_xwrite(tbInfo->tarFd, &header);
460 /* Now do the verbose thing (or not) */
461 if (tbInfo->verboseFlag) {
464 /* If archive goes to stdout, verbose goes to stderr */
465 if (tbInfo->tarFd == STDOUT_FILENO)
467 /* GNU "tar cvvf" prints "extended" listing a-la "ls -l" */
468 /* We don't have such excesses here: for us "v" == "vv" */
469 /* '/' is probably a GNUism */
470 fprintf(vbFd, "%s%s\n", header_name,
471 S_ISDIR(statbuf->st_mode) ? "/" : "");
477 #if ENABLE_FEATURE_TAR_FROM
478 static int exclude_file(const llist_t *excluded_files, const char *file)
480 while (excluded_files) {
481 if (excluded_files->data[0] == '/') {
482 if (fnmatch(excluded_files->data, file,
483 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
488 for (p = file; p[0] != '\0'; p++) {
489 if ((p == file || p[-1] == '/')
491 && fnmatch(excluded_files->data, p,
492 FNM_PATHNAME | FNM_LEADING_DIR) == 0
498 excluded_files = excluded_files->link;
504 # define exclude_file(excluded_files, file) 0
507 static int FAST_FUNC writeFileToTarball(const char *fileName, struct stat *statbuf,
508 void *userData, int depth UNUSED_PARAM)
510 struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData;
511 const char *header_name;
512 int inputFileFd = -1;
514 DBG("writeFileToTarball('%s')", fileName);
516 /* Strip leading '/' and such (must be before memorizing hardlink's name) */
517 header_name = strip_unsafe_prefix(fileName);
519 if (header_name[0] == '\0')
522 /* It is against the rules to archive a socket */
523 if (S_ISSOCK(statbuf->st_mode)) {
524 bb_error_msg("%s: socket ignored", fileName);
529 * Check to see if we are dealing with a hard link.
531 * Treat the first occurrence of a given dev/inode as a file while
532 * treating any additional occurrences as hard links. This is done
533 * by adding the file information to the HardLinkInfo linked list.
535 tbInfo->hlInfo = NULL;
536 if (!S_ISDIR(statbuf->st_mode) && statbuf->st_nlink > 1) {
537 DBG("'%s': st_nlink > 1", header_name);
538 tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf);
539 if (tbInfo->hlInfo == NULL) {
540 DBG("'%s': addHardLinkInfo", header_name);
541 addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, header_name);
545 /* It is a bad idea to store the archive we are in the process of creating,
546 * so check the device and inode to be sure that this particular file isn't
548 if (tbInfo->tarFileStatBuf.st_dev == statbuf->st_dev
549 && tbInfo->tarFileStatBuf.st_ino == statbuf->st_ino
551 bb_error_msg("%s: file is the archive; skipping", fileName);
555 if (exclude_file(tbInfo->excludeList, header_name))
558 #if !ENABLE_FEATURE_TAR_GNU_EXTENSIONS
559 if (strlen(header_name) >= NAME_SIZE) {
560 bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported");
565 /* Is this a regular file? */
566 if (tbInfo->hlInfo == NULL && S_ISREG(statbuf->st_mode)) {
567 /* open the file we want to archive, and make sure all is well */
568 inputFileFd = open_or_warn(fileName, O_RDONLY);
569 if (inputFileFd < 0) {
574 /* Add an entry to the tarball */
575 if (writeTarHeader(tbInfo, header_name, fileName, statbuf) == FALSE) {
579 /* If it was a regular file, write out the body */
580 if (inputFileFd >= 0) {
582 /* Write the file to the archive. */
583 /* We record size into header first, */
584 /* and then write out file. If file shrinks in between, */
585 /* tar will be corrupted. So we don't allow for that. */
586 /* NB: GNU tar 1.16 warns and pads with zeroes */
587 /* or even seeks back and updates header */
588 bb_copyfd_exact_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
590 ////readSize = bb_copyfd_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
591 ////if (readSize != statbuf->st_size && readSize >= 0) {
592 //// bb_error_msg_and_die("short read from %s, aborting", fileName);
595 /* Check that file did not grow in between? */
596 /* if (safe_read(inputFileFd, 1) == 1) warn but continue? */
600 /* Pad the file up to the tar block size */
601 /* (a few tricks here in the name of code size) */
602 readSize = (-(int)statbuf->st_size) & (TAR_BLOCK_SIZE-1);
603 memset(block_buf, 0, readSize);
604 xwrite(tbInfo->tarFd, block_buf, readSize);
610 #if SEAMLESS_COMPRESSION
611 /* Don't inline: vfork scares gcc and pessimizes code */
612 static void NOINLINE vfork_compressor(int tar_fd, const char *gzip)
616 // On Linux, vfork never unpauses parent early, although standard
617 // allows for that. Do we want to waste bytes checking for it?
618 # define WAIT_FOR_CHILD 0
619 volatile int vfork_exec_errno = 0;
620 struct fd_pair gzipDataPipe;
622 struct fd_pair gzipStatusPipe;
623 xpiped_pair(gzipStatusPipe);
625 xpiped_pair(gzipDataPipe);
627 signal(SIGPIPE, SIG_IGN); /* we only want EPIPE on errors */
633 /* NB: close _first_, then move fds! */
634 close(gzipDataPipe.wr);
636 close(gzipStatusPipe.rd);
637 /* gzipStatusPipe.wr will close only on exec -
638 * parent waits for this close to happen */
639 fcntl(gzipStatusPipe.wr, F_SETFD, FD_CLOEXEC);
641 xmove_fd(gzipDataPipe.rd, 0);
643 /* exec gzip/bzip2 program/applet */
644 BB_EXECLP(gzip, gzip, "-f", (char *)0);
645 vfork_exec_errno = errno;
650 xmove_fd(gzipDataPipe.wr, tar_fd);
651 close(gzipDataPipe.rd);
653 close(gzipStatusPipe.wr);
658 /* Wait until child execs (or fails to) */
659 n = full_read(gzipStatusPipe.rd, &buf, 1);
660 if (n < 0 /* && errno == EAGAIN */)
661 continue; /* try it again */
663 close(gzipStatusPipe.rd);
665 if (vfork_exec_errno) {
666 errno = vfork_exec_errno;
667 bb_perror_msg_and_die("can't execute '%s'", gzip);
670 #endif /* SEAMLESS_COMPRESSION */
673 #if !SEAMLESS_COMPRESSION
674 /* Do not pass gzip flag to writeTarFile() */
675 #define writeTarFile(tar_fd, verboseFlag, recurseFlags, include, exclude, gzip) \
676 writeTarFile(tar_fd, verboseFlag, recurseFlags, include, exclude)
678 /* gcc 4.2.1 inlines it, making code bigger */
679 static NOINLINE int writeTarFile(int tar_fd, int verboseFlag,
680 int recurseFlags, const llist_t *include,
681 const llist_t *exclude, const char *gzip)
683 int errorFlag = FALSE;
684 struct TarBallInfo tbInfo;
686 tbInfo.hlInfoHead = NULL;
687 tbInfo.tarFd = tar_fd;
688 tbInfo.verboseFlag = verboseFlag;
690 /* Store the stat info for the tarball's file, so
691 * can avoid including the tarball into itself.... */
692 xfstat(tbInfo.tarFd, &tbInfo.tarFileStatBuf, "can't stat tar file");
694 #if SEAMLESS_COMPRESSION
696 vfork_compressor(tbInfo.tarFd, gzip);
699 tbInfo.excludeList = exclude;
701 /* Read the directory/files and iterate over them one at a time */
703 if (!recursive_action(include->data, recurseFlags,
704 writeFileToTarball, writeFileToTarball, &tbInfo, 0)
708 include = include->link;
710 /* Write two empty blocks to the end of the archive */
711 memset(block_buf, 0, 2*TAR_BLOCK_SIZE);
712 xwrite(tbInfo.tarFd, block_buf, 2*TAR_BLOCK_SIZE);
714 /* To be pedantically correct, we would check if the tarball
715 * is smaller than 20 tar blocks, and pad it if it was smaller,
716 * but that isn't necessary for GNU tar interoperability, and
717 * so is considered a waste of space */
719 /* Close so the child process (if any) will exit */
722 /* Hang up the tools, close up shop, head home */
723 if (ENABLE_FEATURE_CLEAN_UP)
724 freeHardLinkInfo(&tbInfo.hlInfoHead);
727 bb_error_msg("error exit delayed from previous errors");
729 #if SEAMLESS_COMPRESSION
732 if (safe_waitpid(-1, &status, 0) == -1)
733 bb_perror_msg("waitpid");
734 else if (!WIFEXITED(status) || WEXITSTATUS(status))
735 /* gzip was killed or has exited with nonzero! */
741 #else /* !FEATURE_TAR_CREATE */
742 # define writeTarFile(...) 0
745 #if ENABLE_FEATURE_TAR_FROM
746 static llist_t *append_file_list_to_list(llist_t *list)
750 llist_t *newlist = NULL;
753 src_stream = xfopen_stdin(llist_pop(&list));
754 while ((line = xmalloc_fgetline(src_stream)) != NULL) {
755 /* kill trailing '/' unless the string is just "/" */
756 char *cp = last_char_is(line, '/');
759 llist_add_to_end(&newlist, line);
767 //usage:#define tar_trivial_usage
768 //usage: "-[" IF_FEATURE_TAR_CREATE("c") "xt"
769 //usage: IF_FEATURE_SEAMLESS_Z("Z")
770 //usage: IF_FEATURE_SEAMLESS_GZ("z")
771 //usage: IF_FEATURE_SEAMLESS_XZ("J")
772 //usage: IF_FEATURE_SEAMLESS_BZ2("j")
773 //usage: IF_FEATURE_SEAMLESS_LZMA("a")
774 //usage: IF_FEATURE_TAR_CREATE("h")
775 //usage: IF_FEATURE_TAR_NOPRESERVE_TIME("m")
777 //usage: IF_FEATURE_TAR_FROM("[-X FILE] [-T FILE] ")
778 //usage: "[-f TARFILE] [-C DIR] [FILE]..."
779 //usage:#define tar_full_usage "\n\n"
780 //usage: IF_FEATURE_TAR_CREATE("Create, extract, ")
781 //usage: IF_NOT_FEATURE_TAR_CREATE("Extract ")
782 //usage: "or list files from a tar file\n"
783 //usage: "\nOperation:"
784 //usage: IF_FEATURE_TAR_CREATE(
785 //usage: "\n c Create"
787 //usage: "\n x Extract"
789 //usage: "\n f Name of TARFILE ('-' for stdin/out)"
790 //usage: "\n C Change to DIR before operation"
791 //usage: "\n v Verbose"
792 //usage: IF_FEATURE_SEAMLESS_Z(
793 //usage: "\n Z (De)compress using compress"
795 //usage: IF_FEATURE_SEAMLESS_GZ(
796 //usage: "\n z (De)compress using gzip"
798 //usage: IF_FEATURE_SEAMLESS_XZ(
799 //usage: "\n J (De)compress using xz"
801 //usage: IF_FEATURE_SEAMLESS_BZ2(
802 //usage: "\n j (De)compress using bzip2"
804 //usage: IF_FEATURE_SEAMLESS_LZMA(
805 //usage: "\n a (De)compress using lzma"
807 //usage: "\n O Extract to stdout"
808 //usage: IF_FEATURE_TAR_CREATE(
809 //usage: "\n h Follow symlinks"
811 //usage: IF_FEATURE_TAR_NOPRESERVE_TIME(
812 //usage: "\n m Don't restore mtime"
814 //usage: IF_FEATURE_TAR_FROM(
815 //usage: IF_FEATURE_TAR_LONG_OPTIONS(
816 //usage: "\n exclude File to exclude"
818 //usage: "\n X File with names to exclude"
819 //usage: "\n T File with names to include"
822 //usage:#define tar_example_usage
823 //usage: "$ zcat /tmp/tarball.tar.gz | tar -xf -\n"
824 //usage: "$ tar -cf /tmp/tarball.tar /usr/local\n"
826 // Supported but aren't in --help:
828 // p same-permissions
832 // no-same-permissions
834 //IF_FEATURE_TAR_TO_COMMAND(
840 IF_FEATURE_TAR_CREATE( OPTBIT_CREATE ,)
841 IF_FEATURE_TAR_CREATE( OPTBIT_DEREFERENCE ,)
842 IF_FEATURE_SEAMLESS_BZ2( OPTBIT_BZIP2 ,)
843 IF_FEATURE_SEAMLESS_LZMA(OPTBIT_LZMA ,)
844 IF_FEATURE_TAR_FROM( OPTBIT_INCLUDE_FROM,)
845 IF_FEATURE_TAR_FROM( OPTBIT_EXCLUDE_FROM,)
846 IF_FEATURE_SEAMLESS_GZ( OPTBIT_GZIP ,)
847 IF_FEATURE_SEAMLESS_XZ( OPTBIT_XZ ,) // 16th bit
848 IF_FEATURE_SEAMLESS_Z( OPTBIT_COMPRESS ,)
849 IF_FEATURE_TAR_NOPRESERVE_TIME(OPTBIT_NOPRESERVE_TIME,)
850 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
851 OPTBIT_STRIP_COMPONENTS,
853 IF_FEATURE_TAR_TO_COMMAND(OPTBIT_2COMMAND ,)
854 OPTBIT_NUMERIC_OWNER,
855 OPTBIT_NOPRESERVE_PERM,
858 OPT_TEST = 1 << 0, // t
859 OPT_EXTRACT = 1 << 1, // x
860 OPT_BASEDIR = 1 << 2, // C
861 OPT_TARNAME = 1 << 3, // f
862 OPT_2STDOUT = 1 << 4, // O
863 OPT_NOPRESERVE_OWNER = 1 << 5, // o == no-same-owner
865 OPT_VERBOSE = 1 << 7, // v
866 OPT_KEEP_OLD = 1 << 8, // k
867 OPT_CREATE = IF_FEATURE_TAR_CREATE( (1 << OPTBIT_CREATE )) + 0, // c
868 OPT_DEREFERENCE = IF_FEATURE_TAR_CREATE( (1 << OPTBIT_DEREFERENCE )) + 0, // h
869 OPT_BZIP2 = IF_FEATURE_SEAMLESS_BZ2( (1 << OPTBIT_BZIP2 )) + 0, // j
870 OPT_LZMA = IF_FEATURE_SEAMLESS_LZMA((1 << OPTBIT_LZMA )) + 0, // a
871 OPT_INCLUDE_FROM = IF_FEATURE_TAR_FROM( (1 << OPTBIT_INCLUDE_FROM)) + 0, // T
872 OPT_EXCLUDE_FROM = IF_FEATURE_TAR_FROM( (1 << OPTBIT_EXCLUDE_FROM)) + 0, // X
873 OPT_GZIP = IF_FEATURE_SEAMLESS_GZ( (1 << OPTBIT_GZIP )) + 0, // z
874 OPT_XZ = IF_FEATURE_SEAMLESS_XZ( (1 << OPTBIT_XZ )) + 0, // J
875 OPT_COMPRESS = IF_FEATURE_SEAMLESS_Z( (1 << OPTBIT_COMPRESS )) + 0, // Z
876 OPT_NOPRESERVE_TIME = IF_FEATURE_TAR_NOPRESERVE_TIME((1 << OPTBIT_NOPRESERVE_TIME)) + 0, // m
877 OPT_STRIP_COMPONENTS = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_STRIP_COMPONENTS)) + 0, // strip-components
878 OPT_NORECURSION = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_NORECURSION )) + 0, // no-recursion
879 OPT_2COMMAND = IF_FEATURE_TAR_TO_COMMAND( (1 << OPTBIT_2COMMAND )) + 0, // to-command
880 OPT_NUMERIC_OWNER = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_NUMERIC_OWNER )) + 0, // numeric-owner
881 OPT_NOPRESERVE_PERM = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_NOPRESERVE_PERM)) + 0, // no-same-permissions
882 OPT_OVERWRITE = IF_FEATURE_TAR_LONG_OPTIONS((1 << OPTBIT_OVERWRITE )) + 0, // overwrite
884 OPT_ANY_COMPRESS = (OPT_BZIP2 | OPT_LZMA | OPT_GZIP | OPT_XZ | OPT_COMPRESS),
886 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
887 static const char tar_longopts[] ALIGN1 =
888 "list\0" No_argument "t"
889 "extract\0" No_argument "x"
890 "directory\0" Required_argument "C"
891 "file\0" Required_argument "f"
892 "to-stdout\0" No_argument "O"
893 /* do not restore owner */
894 /* Note: GNU tar handles 'o' as no-same-owner only on extract,
895 * on create, 'o' is --old-archive. We do not support --old-archive. */
896 "no-same-owner\0" No_argument "o"
897 "same-permissions\0" No_argument "p"
898 "verbose\0" No_argument "v"
899 "keep-old\0" No_argument "k"
900 # if ENABLE_FEATURE_TAR_CREATE
901 "create\0" No_argument "c"
902 "dereference\0" No_argument "h"
904 # if ENABLE_FEATURE_SEAMLESS_BZ2
905 "bzip2\0" No_argument "j"
907 # if ENABLE_FEATURE_SEAMLESS_LZMA
908 "lzma\0" No_argument "a"
910 # if ENABLE_FEATURE_TAR_FROM
911 "files-from\0" Required_argument "T"
912 "exclude-from\0" Required_argument "X"
914 # if ENABLE_FEATURE_SEAMLESS_GZ
915 "gzip\0" No_argument "z"
917 # if ENABLE_FEATURE_SEAMLESS_XZ
918 "xz\0" No_argument "J"
920 # if ENABLE_FEATURE_SEAMLESS_Z
921 "compress\0" No_argument "Z"
923 # if ENABLE_FEATURE_TAR_NOPRESERVE_TIME
924 "touch\0" No_argument "m"
926 "strip-components\0" Required_argument "\xf9"
927 "no-recursion\0" No_argument "\xfa"
928 # if ENABLE_FEATURE_TAR_TO_COMMAND
929 "to-command\0" Required_argument "\xfb"
931 /* use numeric uid/gid from tar header, not textual */
932 "numeric-owner\0" No_argument "\xfc"
933 /* do not restore mode */
934 "no-same-permissions\0" No_argument "\xfd"
935 /* on unpack, open with O_TRUNC and !O_EXCL */
936 "overwrite\0" No_argument "\xfe"
937 /* --exclude takes next bit position in option mask, */
938 /* therefore we have to put it _after_ --no-same-permissions */
939 # if ENABLE_FEATURE_TAR_FROM
940 "exclude\0" Required_argument "\xff"
945 int tar_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
946 int tar_main(int argc UNUSED_PARAM, char **argv)
948 archive_handle_t *tar_handle;
949 char *base_dir = NULL;
950 const char *tar_filename = "-";
953 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
954 llist_t *excludes = NULL;
958 /* Initialise default values */
959 tar_handle = init_handle();
960 tar_handle->ah_flags = ARCHIVE_CREATE_LEADING_DIRS
961 | ARCHIVE_RESTORE_DATE
962 | ARCHIVE_UNLINK_OLD;
964 /* Apparently only root's tar preserves perms (see bug 3844) */
966 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM;
968 /* Prepend '-' to the first argument if required */
969 opt_complementary = "--:" // first arg is options
970 "tt:vv:" // count -t,-v
971 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
972 "\xff::" // --exclude=PATTERN is a list
974 IF_FEATURE_TAR_CREATE("c:") "t:x:" // at least one of these is reqd
975 IF_FEATURE_TAR_CREATE("c--tx:t--cx:x--ct") // mutually exclusive
976 IF_NOT_FEATURE_TAR_CREATE("t--x:x--t") // mutually exclusive
977 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
978 ":\xf9+" // --strip-components=NUM
981 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
982 applet_long_options = tar_longopts;
985 /* Lie to buildroot when it starts asking stupid questions. */
986 if (argv[1] && strcmp(argv[1], "--version") == 0) {
987 // Output of 'tar --version' examples:
988 // tar (GNU tar) 1.15.1
989 // tar (GNU tar) 1.25
990 // bsdtar 2.8.3 - libarchive 2.8.3
991 puts("tar (busybox) " BB_VER);
994 if (argv[1] && argv[1][0] != '-') {
996 * 1st argument without dash handles options with parameters
997 * differently from dashed one: it takes *next argv[i]*
998 * as paramenter even if there are more chars in 1st argument:
999 * "tar fx TARFILE" - "x" is not taken as f's param
1000 * but is interpreted as -x option
1001 * "tar -xf TARFILE" - dashed equivalent of the above
1002 * "tar -fx ..." - "x" is taken as f's param
1003 * getopt32 wouldn't handle 1st command correctly.
1004 * Unfortunately, people do use such commands.
1005 * We massage argv[1] to work around it by moving 'f'
1006 * to the end of the string.
1007 * More contrived "tar fCx TARFILE DIR" still fails,
1008 * but such commands are much less likely to be used.
1010 char *f = strchr(argv[1], 'f');
1012 while (f[1] != '\0') {
1020 opt = getopt32(argv,
1022 IF_FEATURE_TAR_CREATE( "ch" )
1023 IF_FEATURE_SEAMLESS_BZ2( "j" )
1024 IF_FEATURE_SEAMLESS_LZMA("a" )
1025 IF_FEATURE_TAR_FROM( "T:*X:*")
1026 IF_FEATURE_SEAMLESS_GZ( "z" )
1027 IF_FEATURE_SEAMLESS_XZ( "J" )
1028 IF_FEATURE_SEAMLESS_Z( "Z" )
1029 IF_FEATURE_TAR_NOPRESERVE_TIME("m")
1030 IF_FEATURE_TAR_LONG_OPTIONS("\xf9:") // --strip-components
1031 , &base_dir // -C dir
1032 , &tar_filename // -f filename
1033 IF_FEATURE_TAR_FROM(, &(tar_handle->accept)) // T
1034 IF_FEATURE_TAR_FROM(, &(tar_handle->reject)) // X
1035 #if ENABLE_FEATURE_TAR_LONG_OPTIONS
1036 , &tar_handle->tar__strip_components // --strip-components
1038 IF_FEATURE_TAR_TO_COMMAND(, &(tar_handle->tar__to_command)) // --to-command
1039 #if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM
1040 , &excludes // --exclude
1042 , &verboseFlag // combined count for -t and -v
1043 , &verboseFlag // combined count for -t and -v
1045 #if DBG_OPTION_PARSING
1046 bb_error_msg("opt: 0x%08x", opt);
1047 # define showopt(o) bb_error_msg("opt & %s(%x): %x", #o, o, opt & o);
1049 showopt(OPT_EXTRACT );
1050 showopt(OPT_BASEDIR );
1051 showopt(OPT_TARNAME );
1052 showopt(OPT_2STDOUT );
1053 showopt(OPT_NOPRESERVE_OWNER);
1055 showopt(OPT_VERBOSE );
1056 showopt(OPT_KEEP_OLD );
1057 showopt(OPT_CREATE );
1058 showopt(OPT_DEREFERENCE );
1059 showopt(OPT_BZIP2 );
1061 showopt(OPT_INCLUDE_FROM );
1062 showopt(OPT_EXCLUDE_FROM );
1065 showopt(OPT_COMPRESS );
1066 showopt(OPT_NOPRESERVE_TIME );
1067 showopt(OPT_STRIP_COMPONENTS);
1068 showopt(OPT_NORECURSION );
1069 showopt(OPT_2COMMAND );
1070 showopt(OPT_NUMERIC_OWNER );
1071 showopt(OPT_NOPRESERVE_PERM );
1072 showopt(OPT_OVERWRITE );
1073 showopt(OPT_ANY_COMPRESS );
1074 bb_error_msg("base_dir:'%s'", base_dir);
1075 bb_error_msg("tar_filename:'%s'", tar_filename);
1076 bb_error_msg("verboseFlag:%d", verboseFlag);
1077 bb_error_msg("tar_handle->tar__to_command:'%s'", tar_handle->tar__to_command);
1078 bb_error_msg("tar_handle->tar__strip_components:%u", tar_handle->tar__strip_components);
1085 tar_handle->action_header = header_verbose_list;
1086 if (verboseFlag == 1)
1087 tar_handle->action_header = header_list;
1089 if (opt & OPT_EXTRACT)
1090 tar_handle->action_data = data_extract_all;
1092 if (opt & OPT_2STDOUT)
1093 tar_handle->action_data = data_extract_to_stdout;
1095 if (opt & OPT_2COMMAND) {
1096 putenv((char*)"TAR_FILETYPE=f");
1097 signal(SIGPIPE, SIG_IGN);
1098 tar_handle->action_data = data_extract_to_command;
1099 IF_FEATURE_TAR_TO_COMMAND(tar_handle->tar__to_command_shell = xstrdup(get_shell_name());)
1102 if (opt & OPT_KEEP_OLD)
1103 tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD;
1105 if (opt & OPT_NUMERIC_OWNER)
1106 tar_handle->ah_flags |= ARCHIVE_NUMERIC_OWNER;
1108 if (opt & OPT_NOPRESERVE_OWNER)
1109 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_OWNER;
1111 if (opt & OPT_NOPRESERVE_PERM)
1112 tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM;
1114 if (opt & OPT_OVERWRITE) {
1115 tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD;
1116 tar_handle->ah_flags |= ARCHIVE_O_TRUNC;
1119 if (opt & OPT_NOPRESERVE_TIME)
1120 tar_handle->ah_flags &= ~ARCHIVE_RESTORE_DATE;
1122 #if ENABLE_FEATURE_TAR_FROM
1123 tar_handle->reject = append_file_list_to_list(tar_handle->reject);
1124 # if ENABLE_FEATURE_TAR_LONG_OPTIONS
1125 /* Append excludes to reject */
1127 llist_t *next = excludes->link;
1128 excludes->link = tar_handle->reject;
1129 tar_handle->reject = excludes;
1133 tar_handle->accept = append_file_list_to_list(tar_handle->accept);
1136 /* Setup an array of filenames to work with */
1137 /* TODO: This is the same as in ar, make a separate function? */
1139 /* kill trailing '/' unless the string is just "/" */
1140 char *cp = last_char_is(*argv, '/');
1143 llist_add_to_end(&tar_handle->accept, *argv);
1147 if (tar_handle->accept || tar_handle->reject)
1148 tar_handle->filter = filter_accept_reject_list;
1150 /* Open the tar file */
1152 int tar_fd = STDIN_FILENO;
1153 int flags = O_RDONLY;
1155 if (opt & OPT_CREATE) {
1156 /* Make sure there is at least one file to tar up */
1157 if (tar_handle->accept == NULL)
1158 bb_error_msg_and_die("empty archive");
1160 tar_fd = STDOUT_FILENO;
1161 /* Mimicking GNU tar 1.15.1: */
1162 flags = O_WRONLY | O_CREAT | O_TRUNC;
1165 if (LONE_DASH(tar_filename)) {
1166 tar_handle->src_fd = tar_fd;
1167 tar_handle->seek = seek_by_read;
1169 if (ENABLE_FEATURE_TAR_AUTODETECT
1170 && flags == O_RDONLY
1171 && !(opt & OPT_ANY_COMPRESS)
1173 tar_handle->src_fd = open_zipped(tar_filename, /*fail_if_not_compressed:*/ 0);
1174 if (tar_handle->src_fd < 0)
1175 bb_perror_msg_and_die("can't open '%s'", tar_filename);
1177 tar_handle->src_fd = xopen(tar_filename, flags);
1185 //if (SEAMLESS_COMPRESSION)
1186 // /* We need to know whether child (gzip/bzip/etc) exits abnormally */
1187 // signal(SIGCHLD, check_errors_in_children);
1189 #if ENABLE_FEATURE_TAR_CREATE
1190 /* Create an archive */
1191 if (opt & OPT_CREATE) {
1192 # if SEAMLESS_COMPRESSION
1193 const char *zipMode = NULL;
1194 if (opt & OPT_COMPRESS)
1195 zipMode = "compress";
1198 if (opt & OPT_BZIP2)
1205 /* NB: writeTarFile() closes tar_handle->src_fd */
1206 return writeTarFile(tar_handle->src_fd, verboseFlag,
1207 (opt & OPT_DEREFERENCE ? ACTION_FOLLOWLINKS : 0)
1208 | (opt & OPT_NORECURSION ? 0 : ACTION_RECURSE),
1210 tar_handle->reject, zipMode);
1214 if (opt & OPT_ANY_COMPRESS) {
1215 USE_FOR_MMU(IF_DESKTOP(long long) int FAST_FUNC (*xformer)(transformer_state_t *xstate);)
1216 USE_FOR_NOMMU(const char *xformer_prog;)
1218 if (opt & OPT_COMPRESS) {
1219 USE_FOR_MMU(IF_FEATURE_SEAMLESS_Z(xformer = unpack_Z_stream;))
1220 USE_FOR_NOMMU(xformer_prog = "uncompress";)
1222 if (opt & OPT_GZIP) {
1223 USE_FOR_MMU(IF_FEATURE_SEAMLESS_GZ(xformer = unpack_gz_stream;))
1224 USE_FOR_NOMMU(xformer_prog = "gunzip";)
1226 if (opt & OPT_BZIP2) {
1227 USE_FOR_MMU(IF_FEATURE_SEAMLESS_BZ2(xformer = unpack_bz2_stream;))
1228 USE_FOR_NOMMU(xformer_prog = "bunzip2";)
1230 if (opt & OPT_LZMA) {
1231 USE_FOR_MMU(IF_FEATURE_SEAMLESS_LZMA(xformer = unpack_lzma_stream;))
1232 USE_FOR_NOMMU(xformer_prog = "unlzma";)
1235 USE_FOR_MMU(IF_FEATURE_SEAMLESS_XZ(xformer = unpack_xz_stream;))
1236 USE_FOR_NOMMU(xformer_prog = "unxz";)
1239 fork_transformer_with_sig(tar_handle->src_fd, xformer, xformer_prog);
1240 /* Can't lseek over pipes */
1241 tar_handle->seek = seek_by_read;
1242 /*tar_handle->offset = 0; - already is */
1245 /* Zero processed headers (== empty file) is not a valid tarball.
1246 * We (ab)use bb_got_signal as exitcode here,
1247 * because check_errors_in_children() uses _it_ as error indicator.
1249 bb_got_signal = EXIT_FAILURE;
1251 while (get_header_tar(tar_handle) == EXIT_SUCCESS)
1252 bb_got_signal = EXIT_SUCCESS; /* saw at least one header, good */
1254 replace_symlink_placeholders(tar_handle->symlink_placeholders);
1256 /* Check that every file that should have been extracted was */
1257 while (tar_handle->accept) {
1258 if (!find_list_entry(tar_handle->reject, tar_handle->accept->data)
1259 && !find_list_entry(tar_handle->passed, tar_handle->accept->data)
1261 bb_error_msg_and_die("%s: not found in archive",
1262 tar_handle->accept->data);
1264 tar_handle->accept = tar_handle->accept->link;
1266 if (ENABLE_FEATURE_CLEAN_UP /* && tar_handle->src_fd != STDIN_FILENO */)
1267 close(tar_handle->src_fd);
1269 if (SEAMLESS_COMPRESSION || OPT_COMPRESS) {
1270 /* Set bb_got_signal to 1 if a child died with !0 exitcode */
1271 check_errors_in_children(0);
1274 return bb_got_signal;