1 /* $OpenBSD: pax.c,v 1.14 1998/09/20 02:22:22 millert Exp $ */
2 /* $NetBSD: pax.c,v 1.5 1996/03/26 23:54:20 mrg Exp $ */
5 * Copyright (c) 1992 Keith Muller.
6 * Copyright (c) 1992, 1993
7 * The Regents of the University of California. All rights reserved.
9 * This code is derived from software contributed to Berkeley by
10 * Keith Muller of the University of California, San Diego.
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. All advertising materials mentioning features or use of this software
21 * must display the following acknowledgement:
22 * This product includes software developed by the University of
23 * California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 * may be used to endorse or promote products derived from this software
26 * without specific prior written permission.
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
42 static char copyright[] =
43 "@(#) Copyright (c) 1992, 1993\n\
44 The Regents of the University of California. All rights reserved.\n";
49 static char sccsid[] = "@(#)pax.c 8.2 (Berkeley) 4/18/94";
51 static char rcsid[] = "$OpenBSD: pax.c,v 1.14 1998/09/20 02:22:22 millert Exp $";
56 #include <sys/types.h>
57 #include <sys/param.h>
60 #include <sys/resource.h>
69 static int gen_init __P((void));
72 * PAX main routines, general globals and some simple start up routines
76 * Variables that can be accessed by any routine within pax
78 int act = DEFOP; /* read/write/append/copy */
79 FSUB *frmt = NULL; /* archive format type */
80 int cflag; /* match all EXCEPT pattern/file */
81 int cwdfd; /* starting cwd */
82 int dflag; /* directory member match only */
83 int iflag; /* interactive file/archive rename */
84 int kflag; /* do not overwrite existing files */
85 int lflag; /* use hard links when possible */
86 int nflag; /* select first archive member match */
87 int tflag; /* restore access time after read */
88 int uflag; /* ignore older modification time files */
89 int vflag; /* produce verbose output */
90 int zflag; /* use gzip */
91 int Dflag; /* same as uflag except inode change time */
92 int Hflag; /* follow command line symlinks (write only) */
93 int Lflag; /* follow symlinks when writing */
94 int Xflag; /* archive files with same device id only */
95 int Yflag; /* same as Dflg except after name mode */
96 int Zflag; /* same as uflg except after name mode */
97 int vfpart; /* is partial verbose output in progress */
98 int patime = 1; /* preserve file access time */
99 int pmtime = 1; /* preserve file modification times */
100 int nodirs; /* do not create directories as needed */
101 int pmode; /* preserve file mode bits */
102 int pids; /* preserve file uid/gid */
103 int rmleadslash = 0; /* remove leading '/' from pathnames */
104 int exit_val; /* exit value */
105 int docrc; /* check/create file crc */
106 char *dirptr; /* destination dir in a copy */
107 char *ltmfrmt; /* -v locale time format (if any) */
108 char *argv0; /* root of argv[0] */
109 sigset_t s_mask; /* signal mask for cleanup critical sect */
110 FILE *listf = stderr; /* file pointer to print file list to */
113 * PAX - Portable Archive Interchange
115 * A utility to read, write, and write lists of the members of archive
116 * files and copy directory hierarchies. A variety of archive formats
117 * are supported (some are described in POSIX 1003.1 10.1):
119 * ustar - 10.1.1 extended tar interchange format
120 * cpio - 10.1.2 extended cpio interchange format
121 * tar - old BSD 4.3 tar format
122 * binary cpio - old cpio with binary header format
123 * sysVR4 cpio - with and without CRC
125 * This version is a superset of IEEE Std 1003.2b-d3
127 * Summary of Extensions to the IEEE Standard:
129 * 1 READ ENHANCEMENTS
130 * 1.1 Operations which read archives will continue to operate even when
131 * processing archives which may be damaged, truncated, or fail to meet
132 * format specs in several different ways. Damaged sections of archives
133 * are detected and avoided if possible. Attempts will be made to resync
134 * archive read operations even with badly damaged media.
135 * 1.2 Blocksize requirements are not strictly enforced on archive read.
136 * Tapes which have variable sized records can be read without errors.
137 * 1.3 The user can specify via the non-standard option flag -E if error
138 * resync operation should stop on a media error, try a specified number
139 * of times to correct, or try to correct forever.
140 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks
141 * of all zeros will be restored with holes appropriate for the target
143 * 1.5 The user is notified whenever something is found during archive
144 * read operations which violates spec (but the read will continue).
145 * 1.6 Multiple archive volumes can be read and may span over different
147 * 1.7 Rigidly restores all file attributes exactly as they are stored on the
149 * 1.8 Modification change time ranges can be specified via multiple -T
150 * options. These allow a user to select files whose modification time
151 * lies within a specific time range.
152 * 1.9 Files can be selected based on owner (user name or uid) via one or more
154 * 1.10 Files can be selected based on group (group name or gid) via one o
156 * 1.11 File modification time can be checked against exisiting file after
157 * name modification (-Z)
159 * 2 WRITE ENHANCEMENTS
160 * 2.1 Write operation will stop instead of allowing a user to create a flawed
161 * flawed archive (due to any problem).
162 * 2.2 Archives writtens by pax are forced to strictly conform to both the
163 * archive and pax the spceific format specifications.
164 * 2.3 Blocking size and format is rigidly enforced on writes.
165 * 2.4 Formats which may exhibit header overflow problems (they have fields
166 * too small for large file systems, such as inode number storage), use
167 * routines designed to repair this problem. These techniques still
168 * conform to both pax and format specifications, but no longer truncate
169 * these fields. This removes any restrictions on using these archive
170 * formats on large file systems.
171 * 2.5 Multiple archive volumes can be written and may span over different
173 * 2.6 A archive volume record limit allows the user to specify the number
174 * of bytes stored on an archive volume. When reached the user is
175 * prompted for the next archive volume. This is specified with the
176 * non-standard -B flag. THe limit is rounded up to the next blocksize.
177 * 2.7 All archive padding during write use zero filled sections. This makes
178 * it much easier to pull data out of flawed archive during read
180 * 2.8 Access time reset with the -t applies to all file nodes (including
182 * 2.9 Symbolic links can be followed with -L (optional in the spec).
183 * 2.10 Modification or inode change time ranges can be specified via
184 * multiple -T options. These allow a user to select files whose
185 * modification or inode change time lies within a specific time range.
186 * 2.11 Files can be selected based on owner (user name or uid) via one or more
188 * 2.12 Files can be selected based on group (group name or gid) via one o
190 * 2.13 Symlinks which appear on the command line can be followed (without
191 * following other symlinks; -H flag)
193 * 3 COPY ENHANCEMENTS
194 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes
195 * into zero filled blocks. The file copy is created with holes which are
196 * appropriate for the target filesystem
197 * 3.2 Access time as well as modification time on copied file trees can be
198 * preserved with the appropriate -p options.
199 * 3.3 Access time reset with the -t applies to all file nodes (including
201 * 3.4 Symbolic links can be followed with -L (optional in the spec).
202 * 3.5 Modification or inode change time ranges can be specified via
203 * multiple -T options. These allow a user to select files whose
204 * modification or inode change time lies within a specific time range.
205 * 3.6 Files can be selected based on owner (user name or uid) via one or more
207 * 3.7 Files can be selected based on group (group name or gid) via one o
209 * 3.8 Symlinks which appear on the command line can be followed (without
210 * following other symlinks; -H flag)
211 * 3.9 File inode change time can be checked against exisiting file before
212 * name modification (-D)
213 * 3.10 File inode change time can be checked against exisiting file after
214 * name modification (-Y)
215 * 3.11 File modification time can be checked against exisiting file after
216 * name modification (-Z)
218 * 4 GENERAL ENHANCEMENTS
219 * 4.1 Internal structure is designed to isolate format dependent and
220 * independent functions. Formats are selected via a format driver table.
221 * This encourages the addition of new archive formats by only having to
222 * write those routines which id, read and write the archive header.
227 * parse options, set up and operate as specified by the user.
228 * any operational flaw will set exit_val to non-zero
229 * Return: 0 if ok, 1 otherwise
234 main(int argc, char **argv)
243 * Keep a reference to cwd, so we can always come back home.
245 cwdfd = open(".", O_RDONLY);
247 syswarn(0, errno, "Can't open current working directory.");
252 * parse options, determine operational mode, general init
255 if ((gen_init() < 0) || (tty_init() < 0))
259 * select a primary operation mode
284 * when interrupted we try to do whatever delayed processing we can.
285 * This is not critical, but we really ought to limit our damage when we
286 * are aborted by the user.
293 sig_cleanup(int which_sig)
296 sig_cleanup(which_sig)
301 * restore modes and times for any dirs we may have created
302 * or any dirs we may have read. Set vflag and vfpart so the user
303 * will clearly see the message on a line by itself.
306 if (which_sig == SIGXCPU)
307 paxwarn(0, "Cpu time limit reached, cleaning up.");
309 paxwarn(0, "Signal caught, cleaning up.");
320 * general setup routines. Not all are required, but they really help
321 * when dealing with a medium to large sized archives.
332 struct rlimit reslimit;
333 struct sigaction n_hand;
334 struct sigaction o_hand;
337 * Really needed to handle large archives. We can run out of memory for
338 * internal tables really fast when we have a whole lot of files...
340 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){
341 reslimit.rlim_cur = reslimit.rlim_max;
342 (void)setrlimit(RLIMIT_DATA , &reslimit);
346 * should file size limits be waived? if the os limits us, this is
347 * needed if we want to write a large archive
349 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){
350 reslimit.rlim_cur = reslimit.rlim_max;
351 (void)setrlimit(RLIMIT_FSIZE , &reslimit);
355 * increase the size the stack can grow to
357 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){
358 reslimit.rlim_cur = reslimit.rlim_max;
359 (void)setrlimit(RLIMIT_STACK , &reslimit);
363 * not really needed, but doesn't hurt
365 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){
366 reslimit.rlim_cur = reslimit.rlim_max;
367 (void)setrlimit(RLIMIT_RSS , &reslimit);
371 * Handle posix locale
373 * set user defines time printing format for -v option
375 ltmfrmt = getenv("LC_TIME");
378 * signal handling to reset stored directory times and modes. Since
379 * we deal with broken pipes via failed writes we ignore it. We also
380 * deal with any file size limit thorugh failed writes. Cpu time
381 * limits are caught and a cleanup is forced.
383 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) ||
384 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) ||
385 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) ||
386 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) {
387 paxwarn(1, "Unable to set up signal mask");
390 memset(&n_hand, 0, sizeof n_hand);
391 n_hand.sa_mask = s_mask;
393 n_hand.sa_handler = sig_cleanup;
395 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) &&
396 (o_hand.sa_handler == SIG_IGN) &&
397 (sigaction(SIGHUP, &o_hand, &o_hand) < 0))
400 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) &&
401 (o_hand.sa_handler == SIG_IGN) &&
402 (sigaction(SIGTERM, &o_hand, &o_hand) < 0))
405 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) &&
406 (o_hand.sa_handler == SIG_IGN) &&
407 (sigaction(SIGINT, &o_hand, &o_hand) < 0))
410 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) &&
411 (o_hand.sa_handler == SIG_IGN) &&
412 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0))
415 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) &&
416 (o_hand.sa_handler == SIG_IGN) &&
417 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0))
420 n_hand.sa_handler = SIG_IGN;
421 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) ||
422 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0))
427 syswarn(1, errno, "Unable to set up signal handler");