1 /* vi: set sw=4 ts=4: */
3 * Mini tar implementation for busybox
5 * Modifed to use common extraction code used by ar, cpio, dpkg-deb, dpkg
6 * Glenn McGrath <bug1@optushome.com.au>
8 * Note, that as of BusyBox-0.43, tar has been completely rewritten from the
9 * ground up. It still has remnents of the old code lying about, but it is
10 * very different now (i.e., cleaner, less global variables, etc.)
12 * Copyright (C) 1999,2000 by Lineo, inc. and Erik Andersen
13 * Copyright (C) 1999-2002 by Erik Andersen <andersee@debian.org>
15 * Based in part in the tar implementation in sash
16 * Copyright (c) 1999 by David I. Bell
17 * Permission is granted to use, distribute, or modify this source,
18 * provided that this copyright notice remains intact.
19 * Permission to distribute sash derived code under the GPL has been granted.
21 * Based in part on the tar implementation from busybox-0.28
22 * Copyright (C) 1995 Bruce Perens
23 * This is free software under the GNU General Public License.
25 * This program is free software; you can redistribute it and/or modify
26 * it under the terms of the GNU General Public License as published by
27 * the Free Software Foundation; either version 2 of the License, or
28 * (at your option) any later version.
30 * This program is distributed in the hope that it will be useful,
31 * but WITHOUT ANY WARRANTY; without even the implied warranty of
32 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
33 * General Public License for more details.
35 * You should have received a copy of the GNU General Public License
36 * along with this program; if not, write to the Free Software
37 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
52 #include <sys/socket.h>
53 #include "unarchive.h"
56 #ifdef CONFIG_FEATURE_TAR_CREATE
58 /* Tar file constants */
59 # define TAR_MAGIC "ustar" /* ustar and a null */
60 # define TAR_VERSION " " /* Be compatable with GNU tar format */
63 # define MAJOR(dev) (((dev)>>8)&0xff)
64 # define MINOR(dev) ((dev)&0xff)
67 static const int TAR_BLOCK_SIZE = 512;
68 static const int TAR_MAGIC_LEN = 6;
69 static const int TAR_VERSION_LEN = 2;
71 /* POSIX tar Header Block, from POSIX 1003.1-1990 */
72 enum { NAME_SIZE = 100 }; /* because gcc won't let me use 'static const int' */
73 struct TarHeader { /* byte offset */
74 char name[NAME_SIZE]; /* 0-99 */
75 char mode[8]; /* 100-107 */
76 char uid[8]; /* 108-115 */
77 char gid[8]; /* 116-123 */
78 char size[12]; /* 124-135 */
79 char mtime[12]; /* 136-147 */
80 char chksum[8]; /* 148-155 */
81 char typeflag; /* 156-156 */
82 char linkname[NAME_SIZE]; /* 157-256 */
83 char magic[6]; /* 257-262 */
84 char version[2]; /* 263-264 */
85 char uname[32]; /* 265-296 */
86 char gname[32]; /* 297-328 */
87 char devmajor[8]; /* 329-336 */
88 char devminor[8]; /* 337-344 */
89 char prefix[155]; /* 345-499 */
90 char padding[12]; /* 500-512 (pad to exactly the TAR_BLOCK_SIZE) */
92 typedef struct TarHeader TarHeader;
95 ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are
96 ** the only functions that deal with the HardLinkInfo structure.
97 ** Even these functions use the xxxHardLinkInfo() functions.
99 typedef struct HardLinkInfo HardLinkInfo;
100 struct HardLinkInfo {
101 HardLinkInfo *next; /* Next entry in list */
102 dev_t dev; /* Device number */
103 ino_t ino; /* Inode number */
104 short linkCount; /* (Hard) Link Count */
105 char name[1]; /* Start of filename (must be last) */
108 /* Some info to be carried along when creating a new tarball */
110 char *fileName; /* File name of the tarball */
111 int tarFd; /* Open-for-write file descriptor
113 struct stat statBuf; /* Stat info for the tarball, letting
114 us know the inode and device that the
115 tarball lives, so we can avoid trying
116 to include the tarball into itself */
117 int verboseFlag; /* Whether to print extra stuff or not */
118 const llist_t *excludeList; /* List of files to not include */
119 HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */
120 HardLinkInfo *hlInfo; /* Hard Link Info for the current file */
122 typedef struct TarBallInfo TarBallInfo;
124 /* A nice enum with all the possible tar file content types */
126 REGTYPE = '0', /* regular file */
127 REGTYPE0 = '\0', /* regular file (ancient bug compat) */
128 LNKTYPE = '1', /* hard link */
129 SYMTYPE = '2', /* symbolic link */
130 CHRTYPE = '3', /* character special */
131 BLKTYPE = '4', /* block special */
132 DIRTYPE = '5', /* directory */
133 FIFOTYPE = '6', /* FIFO special */
134 CONTTYPE = '7', /* reserved */
135 GNULONGLINK = 'K', /* GNU long (>100 chars) link name */
136 GNULONGNAME = 'L', /* GNU long (>100 chars) file name */
138 typedef enum TarFileType TarFileType;
140 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
141 static inline void addHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr,
142 struct stat *statbuf,
145 /* Note: hlInfoHeadPtr can never be NULL! */
146 HardLinkInfo *hlInfo;
148 hlInfo = (HardLinkInfo *) xmalloc(sizeof(HardLinkInfo) + strlen(name));
149 hlInfo->next = *hlInfoHeadPtr;
150 *hlInfoHeadPtr = hlInfo;
151 hlInfo->dev = statbuf->st_dev;
152 hlInfo->ino = statbuf->st_ino;
153 hlInfo->linkCount = statbuf->st_nlink;
154 strcpy(hlInfo->name, name);
157 static void freeHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr)
159 HardLinkInfo *hlInfo = NULL;
160 HardLinkInfo *hlInfoNext = NULL;
163 hlInfo = *hlInfoHeadPtr;
165 hlInfoNext = hlInfo->next;
169 *hlInfoHeadPtr = NULL;
174 /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */
175 static inline HardLinkInfo *findHardLinkInfo(HardLinkInfo * hlInfo, struct stat *statbuf)
178 if ((statbuf->st_ino == hlInfo->ino) && (statbuf->st_dev == hlInfo->dev))
180 hlInfo = hlInfo->next;
185 /* Put an octal string into the specified buffer.
186 * The number is zero and space padded and possibly null padded.
187 * Returns TRUE if successful. */
188 static int putOctal(char *cp, int len, long value)
192 char *tempString = tempBuffer;
194 /* Create a string of the specified length with an initial space,
195 * leading zeroes and the octal number, and a trailing null. */
196 sprintf(tempString, "%0*lo", len - 1, value);
198 /* If the string is too large, suppress the leading space. */
199 tempLength = strlen(tempString) + 1;
200 if (tempLength > len) {
205 /* If the string is still too large, suppress the trailing null. */
206 if (tempLength > len)
209 /* If the string is still too large, fail. */
210 if (tempLength > len)
213 /* Copy the string to the field. */
214 memcpy(cp, tempString, len);
219 /* Write out a tar header for the specified file/directory/whatever */
220 static inline int writeTarHeader(struct TarBallInfo *tbInfo,
221 const char *header_name,
222 const char *real_name, struct stat *statbuf)
225 struct TarHeader header;
226 const unsigned char *cp = (const unsigned char *) &header;
227 ssize_t size = sizeof(struct TarHeader);
229 memset(&header, 0, size);
231 strncpy(header.name, header_name, sizeof(header.name));
233 putOctal(header.mode, sizeof(header.mode), statbuf->st_mode);
234 putOctal(header.uid, sizeof(header.uid), statbuf->st_uid);
235 putOctal(header.gid, sizeof(header.gid), statbuf->st_gid);
236 putOctal(header.size, sizeof(header.size), 0); /* Regular file size is handled later */
237 putOctal(header.mtime, sizeof(header.mtime), statbuf->st_mtime);
238 strncpy(header.magic, TAR_MAGIC TAR_VERSION,
239 TAR_MAGIC_LEN + TAR_VERSION_LEN);
241 /* Enter the user and group names (default to root if it fails) */
242 if (my_getpwuid(header.uname, statbuf->st_uid) == NULL)
243 strcpy(header.uname, "root");
244 if (my_getgrgid(header.gname, statbuf->st_gid) == NULL)
245 strcpy(header.gname, "root");
247 if (tbInfo->hlInfo) {
248 /* This is a hard link */
249 header.typeflag = LNKTYPE;
250 strncpy(header.linkname, tbInfo->hlInfo->name,
251 sizeof(header.linkname));
252 } else if (S_ISLNK(statbuf->st_mode)) {
253 char *lpath = xreadlink(real_name);
255 if (!lpath) /* Already printed err msg inside xreadlink() */
257 header.typeflag = SYMTYPE;
258 strncpy(header.linkname, lpath, sizeof(header.linkname));
260 } else if (S_ISDIR(statbuf->st_mode)) {
261 header.typeflag = DIRTYPE;
262 strncat(header.name, "/", sizeof(header.name));
263 } else if (S_ISCHR(statbuf->st_mode)) {
264 header.typeflag = CHRTYPE;
265 putOctal(header.devmajor, sizeof(header.devmajor),
266 MAJOR(statbuf->st_rdev));
267 putOctal(header.devminor, sizeof(header.devminor),
268 MINOR(statbuf->st_rdev));
269 } else if (S_ISBLK(statbuf->st_mode)) {
270 header.typeflag = BLKTYPE;
271 putOctal(header.devmajor, sizeof(header.devmajor),
272 MAJOR(statbuf->st_rdev));
273 putOctal(header.devminor, sizeof(header.devminor),
274 MINOR(statbuf->st_rdev));
275 } else if (S_ISFIFO(statbuf->st_mode)) {
276 header.typeflag = FIFOTYPE;
277 } else if (S_ISREG(statbuf->st_mode)) {
278 header.typeflag = REGTYPE;
279 putOctal(header.size, sizeof(header.size), statbuf->st_size);
281 error_msg("%s: Unknown file type", real_name);
285 /* Calculate and store the checksum (i.e., the sum of all of the bytes of
286 * the header). The checksum field must be filled with blanks for the
287 * calculation. The checksum field is formatted differently from the
288 * other fields: it has [6] digits, a null, then a space -- rather than
289 * digits, followed by a null like the other fields... */
290 memset(header.chksum, ' ', sizeof(header.chksum));
291 cp = (const unsigned char *) &header;
294 putOctal(header.chksum, 7, chksum);
296 /* Now write the header out to disk */
298 full_write(tbInfo->tarFd, (char *) &header,
299 sizeof(struct TarHeader))) < 0) {
300 error_msg(io_error, real_name);
303 /* Pad the header up to the tar block size */
304 for (; size < TAR_BLOCK_SIZE; size++) {
305 write(tbInfo->tarFd, "\0", 1);
307 /* Now do the verbose thing (or not) */
309 if (tbInfo->verboseFlag) {
312 if (tbInfo->verboseFlag == 2) /* If the archive goes to stdout, verbose to stderr */
314 fprintf(vbFd, "%s\n", header.name);
320 # if defined CONFIG_FEATURE_TAR_EXCLUDE
321 static inline int exclude_file(const llist_t *excluded_files, const char *file)
323 if (excluded_files == NULL) {
327 while (excluded_files) {
328 if (excluded_files->data[0] == '/') {
329 if (fnmatch(excluded_files->data, file,
330 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
335 for (p = file; p[0] != '\0'; p++) {
336 if ((p == file || p[-1] == '/') && p[0] != '/' &&
337 fnmatch(excluded_files->data, p,
338 FNM_PATHNAME | FNM_LEADING_DIR) == 0)
342 excluded_files = excluded_files->link;
349 static int writeFileToTarball(const char *fileName, struct stat *statbuf,
352 struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData;
353 const char *header_name;
356 ** Check to see if we are dealing with a hard link.
358 ** Treat the first occurance of a given dev/inode as a file while
359 ** treating any additional occurances as hard links. This is done
360 ** by adding the file information to the HardLinkInfo linked list.
362 tbInfo->hlInfo = NULL;
363 if (statbuf->st_nlink > 1) {
364 tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf);
365 if (tbInfo->hlInfo == NULL)
366 addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, fileName);
369 /* It is against the rules to archive a socket */
370 if (S_ISSOCK(statbuf->st_mode)) {
371 error_msg("%s: socket ignored", fileName);
375 /* It is a bad idea to store the archive we are in the process of creating,
376 * so check the device and inode to be sure that this particular file isn't
378 if (tbInfo->statBuf.st_dev == statbuf->st_dev &&
379 tbInfo->statBuf.st_ino == statbuf->st_ino) {
380 error_msg("%s: file is the archive; skipping", fileName);
384 header_name = fileName;
385 while (header_name[0] == '/') {
386 static int alreadyWarned = FALSE;
388 if (alreadyWarned == FALSE) {
389 error_msg("Removing leading '/' from member names");
390 alreadyWarned = TRUE;
395 if (strlen(fileName) >= NAME_SIZE) {
396 error_msg(name_longer_than_foo, NAME_SIZE);
400 if (header_name[0] == '\0')
403 # if defined CONFIG_FEATURE_TAR_EXCLUDE
404 if (exclude_file(tbInfo->excludeList, header_name)) {
407 # endif /* CONFIG_FEATURE_TAR_EXCLUDE */
409 if (writeTarHeader(tbInfo, header_name, fileName, statbuf) == FALSE) {
413 /* Now, if the file is a regular file, copy it out to the tarball */
414 if ((tbInfo->hlInfo == NULL)
415 && (S_ISREG(statbuf->st_mode))) {
418 ssize_t size = 0, readSize = 0;
420 /* open the file we want to archive, and make sure all is well */
421 if ((inputFileFd = open(fileName, O_RDONLY)) < 0) {
422 perror_msg("%s: Cannot open", fileName);
426 /* write the file to the archive */
427 while ((size = full_read(inputFileFd, buffer, sizeof(buffer))) > 0) {
428 if (full_write(tbInfo->tarFd, buffer, size) != size) {
429 /* Output file seems to have a problem */
430 error_msg(io_error, fileName);
436 error_msg(io_error, fileName);
439 /* Pad the file up to the tar block size */
440 for (; (readSize % TAR_BLOCK_SIZE) != 0; readSize++) {
441 write(tbInfo->tarFd, "\0", 1);
449 static inline int writeTarFile(const char *tarName, const int verboseFlag,
450 const llist_t *include, const llist_t *exclude, const int gzip)
452 #ifdef CONFIG_FEATURE_TAR_GZIP
453 int gzipDataPipe[2] = { -1, -1 };
454 int gzipStatusPipe[2] = { -1, -1 };
456 volatile int vfork_exec_errno = 0;
459 int errorFlag = FALSE;
461 struct TarBallInfo tbInfo;
463 tbInfo.hlInfoHead = NULL;
465 /* Make sure there is at least one file to tar up. */
466 if (include == NULL) {
467 error_msg_and_die("Cowardly refusing to create an empty archive");
470 /* Open the tar file for writing. */
471 if (tarName == NULL || (tarName[0] == '-' && tarName[1] == '\0')) {
472 tbInfo.tarFd = fileno(stdout);
473 tbInfo.verboseFlag = verboseFlag ? 2 : 0;
475 tbInfo.tarFd = open(tarName, O_WRONLY | O_CREAT | O_TRUNC, 0644);
476 tbInfo.verboseFlag = verboseFlag ? 1 : 0;
479 if (tbInfo.tarFd < 0) {
480 perror_msg("%s: Cannot open", tarName);
481 freeHardLinkInfo(&tbInfo.hlInfoHead);
485 /* Store the stat info for the tarball's file, so
486 * can avoid including the tarball into itself.... */
487 if (fstat(tbInfo.tarFd, &tbInfo.statBuf) < 0)
488 error_msg_and_die(io_error, tarName);
490 #ifdef CONFIG_FEATURE_TAR_GZIP
492 if (pipe(gzipDataPipe) < 0 || pipe(gzipStatusPipe) < 0) {
493 perror_msg_and_die("Failed to create gzip pipe");
496 signal(SIGPIPE, SIG_IGN); /* we only want EPIPE on errors */
499 /* Avoid vfork clobbering */
507 dup2(gzipDataPipe[0], 0);
508 close(gzipDataPipe[1]);
510 if (tbInfo.tarFd != 1);
511 dup2(tbInfo.tarFd, 1);
513 close(gzipStatusPipe[0]);
514 fcntl(gzipStatusPipe[1], F_SETFD, FD_CLOEXEC); /* close on exec shows sucess */
516 execl("/bin/gzip", "gzip", "-f", 0);
517 vfork_exec_errno = errno;
519 close(gzipStatusPipe[1]);
521 } else if (gzipPid > 0) {
522 close(gzipDataPipe[0]);
523 close(gzipStatusPipe[1]);
528 int n = read(gzipStatusPipe[0], &buf, 1);
530 if (n == 0 && vfork_exec_errno != 0) {
531 errno = vfork_exec_errno;
532 perror_msg_and_die("Could not exec gzip process");
533 } else if ((n < 0) && (errno == EAGAIN || errno == EINTR))
534 continue; /* try it again */
537 close(gzipStatusPipe[0]);
539 tbInfo.tarFd = gzipDataPipe[1];
541 perror_msg_and_die("Failed to vfork gzip process");
546 tbInfo.excludeList = exclude;
548 /* Read the directory/files and iterate over them one at a time */
550 if (!recursive_action(include->data, TRUE, FALSE, FALSE,
551 writeFileToTarball, writeFileToTarball,
555 include = include->link;
557 /* Write two empty blocks to the end of the archive */
558 for (size = 0; size < (2 * TAR_BLOCK_SIZE); size++) {
559 write(tbInfo.tarFd, "\0", 1);
562 /* To be pedantically correct, we would check if the tarball
563 * is smaller than 20 tar blocks, and pad it if it was smaller,
564 * but that isn't necessary for GNU tar interoperability, and
565 * so is considered a waste of space */
567 /* Hang up the tools, close up shop, head home */
570 error_msg("Error exit delayed from previous errors");
572 freeHardLinkInfo(&tbInfo.hlInfoHead);
574 #ifdef CONFIG_FEATURE_TAR_GZIP
575 if (gzip && gzipPid) {
576 if (waitpid(gzipPid, NULL, 0) == -1)
577 printf("Couldnt wait ?");
583 #endif /* tar_create */
585 #ifdef CONFIG_FEATURE_TAR_EXCLUDE
586 static const llist_t *append_file_list_to_list(const char *filename, const llist_t *list)
588 FILE *src_stream = xfopen(filename, "r");
590 while((line = get_line_from_file(src_stream)) != NULL) {
592 list = add_to_list(list, line);
602 #define CTX_EXTRACT 4
604 int tar_main(int argc, char **argv)
606 char (*get_header_ptr)(archive_handle_t *) = get_header_tar;
607 archive_handle_t *tar_handle;
609 char *base_dir = NULL;
610 const char *tar_filename = "-";
611 unsigned char ctx_flag = 0;
617 /* Prepend '-' to the first argument if required */
618 if (argv[1][0] != '-') {
619 char *tmp = xmalloc(strlen(argv[1]) + 2);
621 strcpy(tmp + 1, argv[1]);
625 /* Initialise default values */
626 tar_handle = init_handle();
627 tar_handle->flags = ARCHIVE_CREATE_LEADING_DIRS;
629 while ((opt = getopt(argc, argv, "cjtxT:X:C:f:Opvz")) != -1) {
631 /* One and only one of these is required */
632 #ifdef CONFIG_FEATURE_TAR_CREATE
634 ctx_flag |= CTX_CREATE;
638 ctx_flag |= CTX_TEST;
639 if ((tar_handle->action_header == header_list) ||
640 (tar_handle->action_header == header_verbose_list)) {
641 tar_handle->action_header = header_verbose_list;
643 tar_handle->action_header = header_list;
647 ctx_flag |= CTX_EXTRACT;
648 tar_handle->action_data = data_extract_all;
651 /* These are optional */
652 /* Exclude or Include files listed in <filename> */
653 #ifdef CONFIG_FEATURE_TAR_EXCLUDE
656 append_file_list_to_list(optarg, tar_handle->reject);
660 /* by default a list is an include list */
662 case 'C': /* Change to dir <optarg> */
665 case 'f': /* archive filename */
666 tar_filename = optarg;
668 case 'O': /* To stdout */
669 tar_handle->action_data = data_extract_to_stdout;
672 tar_handle->flags |= ARCHIVE_PRESERVE_DATE;
675 if ((tar_handle->action_header == header_list) ||
676 (tar_handle->action_header == header_verbose_list)) {
677 tar_handle->action_header = header_verbose_list;
679 tar_handle->action_header = header_list;
682 #ifdef CONFIG_FEATURE_TAR_GZIP
684 get_header_ptr = get_header_tar_gz;
687 #ifdef CONFIG_FEATURE_TAR_BZIP2
689 get_header_ptr = get_header_tar_bz2;
697 /* Check one and only one context option was given */
698 if ((ctx_flag != CTX_CREATE) && (ctx_flag != CTX_TEST) && (ctx_flag != CTX_EXTRACT)) {
702 /* Check if we are reading from stdin */
703 if ((argv[optind]) && (*argv[optind] == '-')) {
704 /* Default is to read from stdin, so just skip to next arg */
708 /* Setup an array of filenames to work with */
709 /* TODO: This is the same as in ar, seperate function ? */
710 while (optind < argc) {
711 tar_handle->accept = add_to_list(tar_handle->accept, argv[optind]);
715 if ((tar_handle->accept) || (tar_handle->reject)) {
716 tar_handle->filter = filter_accept_reject_list;
719 #ifdef CONFIG_FEATURE_TAR_CREATE
720 /* create an archive */
721 if (ctx_flag == CTX_CREATE) {
722 int verboseFlag = FALSE;
723 int gzipFlag = FALSE;
725 # ifdef CONFIG_FEATURE_TAR_GZIP
726 if (get_header_ptr == get_header_tar_gz) {
729 # endif /* CONFIG_FEATURE_TAR_GZIP */
731 if (tar_handle->action_header == header_verbose_list) {
734 writeTarFile(tar_filename, verboseFlag, tar_handle->accept,
735 tar_handle->reject, gzipFlag);
737 #endif /* CONFIG_FEATURE_TAR_CREATE */
739 if ((tar_filename[0] == '-') && (tar_filename[1] == '\0')) {
740 tar_handle->src_fd = fileno(stdin);
741 tar_handle->seek = seek_by_char;
743 tar_handle->src_fd = xopen(tar_filename, O_RDONLY);
746 if ((base_dir) && (chdir(base_dir))) {
747 perror_msg_and_die("Couldnt chdir");
750 while (get_header_ptr(tar_handle) == EXIT_SUCCESS);
752 /* Ckeck that every file that should have been extracted was */
753 while (tar_handle->accept) {
754 if (find_list_entry(tar_handle->reject, tar_handle->accept->data) == NULL) {
755 if (find_list_entry(tar_handle->passed, tar_handle->accept->data) == NULL) {
756 error_msg_and_die("%s: Not found in archive\n", tar_handle->accept->data);
759 tar_handle->accept = tar_handle->accept->link;
763 #ifdef CONFIG_FEATURE_CLEAN_UP
764 if (tar_handle->src_fd != fileno(stdin)) {
765 close(tar_handle->src_fd);
767 #endif /* CONFIG_FEATURE_CLEAN_UP */
769 return(EXIT_SUCCESS);