More minor IPI work.
[dragonfly/vkernel-mp.git] / bin / pax / pax.c
blob3efe421f39f5af08561ca53fd49e74dd343d0ce4
1 /*-
2 * Copyright (c) 1992 Keith Muller.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
6 * This code is derived from software contributed to Berkeley by
7 * Keith Muller of the University of California, San Diego.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by the University of
20 * California, Berkeley and its contributors.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * SUCH DAMAGE.
37 * @(#) Copyright (c) 1992, 1993 The Regents of the University of California. All rights reserved.
38 * @(#)pax.c 8.2 (Berkeley) 4/18/94
39 * $FreeBSD: src/bin/pax/pax.c,v 1.13.2.4 2002/11/07 15:29:53 imp Exp $
40 * $DragonFly: src/bin/pax/pax.c,v 1.6 2006/09/27 23:58:08 pavalos Exp $
43 #include <sys/types.h>
44 #include <sys/stat.h>
45 #include <sys/time.h>
46 #include <sys/resource.h>
47 #include <err.h>
48 #include <errno.h>
49 #include <fcntl.h>
50 #include <locale.h>
51 #include <paths.h>
52 #include <signal.h>
53 #include <stdio.h>
54 #include <stdlib.h>
55 #include <string.h>
56 #include <unistd.h>
57 #include "pax.h"
58 #include "extern.h"
59 static int gen_init (void);
62 * PAX main routines, general globals and some simple start up routines
66 * Variables that can be accessed by any routine within pax
68 int act = DEFOP; /* read/write/append/copy */
69 FSUB *frmt = NULL; /* archive format type */
70 int cflag; /* match all EXCEPT pattern/file */
71 int cwdfd; /* starting cwd */
72 int dflag; /* directory member match only */
73 int iflag; /* interactive file/archive rename */
74 int kflag; /* do not overwrite existing files */
75 int lflag; /* use hard links when possible */
76 int nflag; /* select first archive member match */
77 int tflag; /* restore access time after read */
78 int uflag; /* ignore older modification time files */
79 int vflag; /* produce verbose output */
80 int Dflag; /* same as uflag except inode change time */
81 int Hflag; /* follow command line symlinks (write only) */
82 int Lflag; /* follow symlinks when writing */
83 int Xflag; /* archive files with same device id only */
84 int Yflag; /* same as Dflag except after name mode */
85 int Zflag; /* same as uflag except after name mode */
86 int vfpart; /* is partial verbose output in progress */
87 int patime = 1; /* preserve file access time */
88 int pmtime = 1; /* preserve file modification times */
89 int nodirs; /* do not create directories as needed */
90 int pmode; /* preserve file mode bits */
91 int pids; /* preserve file uid/gid */
92 int rmleadslash = 0; /* remove leading '/' from pathnames */
93 int exit_val; /* exit value */
94 int docrc; /* check/create file crc */
95 char *dirptr; /* destination dir in a copy */
96 char *argv0; /* root of argv[0] */
97 sigset_t s_mask; /* signal mask for cleanup critical sect */
98 FILE *listf; /* file pointer to print file list to */
99 char *tempfile; /* tempfile to use for mkstemp(3) */
100 char *tempbase; /* basename of tempfile to use for mkstemp(3) */
103 * PAX - Portable Archive Interchange
105 * A utility to read, write, and write lists of the members of archive
106 * files and copy directory hierarchies. A variety of archive formats
107 * are supported (some are described in POSIX 1003.1 10.1):
109 * ustar - 10.1.1 extended tar interchange format
110 * cpio - 10.1.2 extended cpio interchange format
111 * tar - old BSD 4.3 tar format
112 * binary cpio - old cpio with binary header format
113 * sysVR4 cpio - with and without CRC
115 * This version is a superset of IEEE Std 1003.2b-d3
117 * Summary of Extensions to the IEEE Standard:
119 * 1 READ ENHANCEMENTS
120 * 1.1 Operations which read archives will continue to operate even when
121 * processing archives which may be damaged, truncated, or fail to meet
122 * format specs in several different ways. Damaged sections of archives
123 * are detected and avoided if possible. Attempts will be made to resync
124 * archive read operations even with badly damaged media.
125 * 1.2 Blocksize requirements are not strictly enforced on archive read.
126 * Tapes which have variable sized records can be read without errors.
127 * 1.3 The user can specify via the non-standard option flag -E if error
128 * resync operation should stop on a media error, try a specified number
129 * of times to correct, or try to correct forever.
130 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks
131 * of all zeros will be restored with holes appropriate for the target
132 * filesystem
133 * 1.5 The user is notified whenever something is found during archive
134 * read operations which violates spec (but the read will continue).
135 * 1.6 Multiple archive volumes can be read and may span over different
136 * archive devices
137 * 1.7 Rigidly restores all file attributes exactly as they are stored on the
138 * archive.
139 * 1.8 Modification change time ranges can be specified via multiple -T
140 * options. These allow a user to select files whose modification time
141 * lies within a specific time range.
142 * 1.9 Files can be selected based on owner (user name or uid) via one or more
143 * -U options.
144 * 1.10 Files can be selected based on group (group name or gid) via one o
145 * more -G options.
146 * 1.11 File modification time can be checked against existing file after
147 * name modification (-Z)
149 * 2 WRITE ENHANCEMENTS
150 * 2.1 Write operation will stop instead of allowing a user to create a flawed
151 * flawed archive (due to any problem).
152 * 2.2 Archives written by pax are forced to strictly conform to both the
153 * archive and pax the specific format specifications.
154 * 2.3 Blocking size and format is rigidly enforced on writes.
155 * 2.4 Formats which may exhibit header overflow problems (they have fields
156 * too small for large file systems, such as inode number storage), use
157 * routines designed to repair this problem. These techniques still
158 * conform to both pax and format specifications, but no longer truncate
159 * these fields. This removes any restrictions on using these archive
160 * formats on large file systems.
161 * 2.5 Multiple archive volumes can be written and may span over different
162 * archive devices
163 * 2.6 A archive volume record limit allows the user to specify the number
164 * of bytes stored on an archive volume. When reached the user is
165 * prompted for the next archive volume. This is specified with the
166 * non-standard -B flag. The limit is rounded up to the next blocksize.
167 * 2.7 All archive padding during write use zero filled sections. This makes
168 * it much easier to pull data out of flawed archive during read
169 * operations.
170 * 2.8 Access time reset with the -t applies to all file nodes (including
171 * directories).
172 * 2.9 Symbolic links can be followed with -L (optional in the spec).
173 * 2.10 Modification or inode change time ranges can be specified via
174 * multiple -T options. These allow a user to select files whose
175 * modification or inode change time lies within a specific time range.
176 * 2.11 Files can be selected based on owner (user name or uid) via one or more
177 * -U options.
178 * 2.12 Files can be selected based on group (group name or gid) via one o
179 * more -G options.
180 * 2.13 Symlinks which appear on the command line can be followed (without
181 * following other symlinks; -H flag)
183 * 3 COPY ENHANCEMENTS
184 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes
185 * into zero filled blocks. The file copy is created with holes which are
186 * appropriate for the target filesystem
187 * 3.2 Access time as well as modification time on copied file trees can be
188 * preserved with the appropriate -p options.
189 * 3.3 Access time reset with the -t applies to all file nodes (including
190 * directories).
191 * 3.4 Symbolic links can be followed with -L (optional in the spec).
192 * 3.5 Modification or inode change time ranges can be specified via
193 * multiple -T options. These allow a user to select files whose
194 * modification or inode change time lies within a specific time range.
195 * 3.6 Files can be selected based on owner (user name or uid) via one or more
196 * -U options.
197 * 3.7 Files can be selected based on group (group name or gid) via one o
198 * more -G options.
199 * 3.8 Symlinks which appear on the command line can be followed (without
200 * following other symlinks; -H flag)
201 * 3.9 File inode change time can be checked against existing file before
202 * name modification (-D)
203 * 3.10 File inode change time can be checked against existing file after
204 * name modification (-Y)
205 * 3.11 File modification time can be checked against existing file after
206 * name modification (-Z)
208 * 4 GENERAL ENHANCEMENTS
209 * 4.1 Internal structure is designed to isolate format dependent and
210 * independent functions. Formats are selected via a format driver table.
211 * This encourages the addition of new archive formats by only having to
212 * write those routines which id, read and write the archive header.
216 * main()
217 * parse options, set up and operate as specified by the user.
218 * any operational flaw will set exit_val to non-zero
219 * Return: 0 if ok, 1 otherwise
223 main(int argc, char **argv)
225 char *tmpdir;
226 size_t tdlen;
228 setlocale(LC_ALL, "");
229 listf = stderr;
231 * Keep a reference to cwd, so we can always come back home.
233 cwdfd = open(".", O_RDONLY);
234 if (cwdfd < 0) {
235 syswarn(0, errno, "Can't open current working directory.");
236 return(exit_val);
240 * Where should we put temporary files?
242 if ((tmpdir = getenv("TMPDIR")) == NULL || *tmpdir == '\0')
243 tmpdir = _PATH_TMP;
244 tdlen = strlen(tmpdir);
245 while(tdlen > 0 && tmpdir[tdlen - 1] == '/')
246 tdlen--;
247 tempfile = malloc(tdlen + 1 + sizeof(_TFILE_BASE));
248 if (tempfile == NULL) {
249 paxwarn(1, "Cannot allocate memory for temp file name.");
250 return(exit_val);
252 if (tdlen)
253 memcpy(tempfile, tmpdir, tdlen);
254 tempbase = tempfile + tdlen;
255 *tempbase++ = '/';
258 * parse options, determine operational mode, general init
260 options(argc, argv);
261 if ((gen_init() < 0) || (tty_init() < 0))
262 return(exit_val);
265 * select a primary operation mode
267 switch(act) {
268 case EXTRACT:
269 extract();
270 break;
271 case ARCHIVE:
272 archive();
273 break;
274 case APPND:
275 if (gzip_program != NULL)
276 err(1, "can not gzip while appending");
277 append();
278 break;
279 case COPY:
280 copy();
281 break;
282 default:
283 case LIST:
284 list();
285 break;
287 return(exit_val);
291 * sig_cleanup()
292 * when interrupted we try to do whatever delayed processing we can.
293 * This is not critical, but we really ought to limit our damage when we
294 * are aborted by the user.
295 * Return:
296 * never....
299 void
300 sig_cleanup(int which_sig)
303 * restore modes and times for any dirs we may have created
304 * or any dirs we may have read. Set vflag and vfpart so the user
305 * will clearly see the message on a line by itself.
307 vflag = vfpart = 1;
308 if (which_sig == SIGXCPU)
309 paxwarn(0, "Cpu time limit reached, cleaning up.");
310 else
311 paxwarn(0, "Signal caught, cleaning up.");
313 ar_close();
314 proc_dir();
315 if (tflag)
316 atdir_end();
317 exit(1);
321 * gen_init()
322 * general setup routines. Not all are required, but they really help
323 * when dealing with a medium to large sized archives.
326 static int
327 gen_init(void)
329 struct rlimit reslimit;
330 struct sigaction n_hand;
331 struct sigaction o_hand;
334 * Really needed to handle large archives. We can run out of memory for
335 * internal tables really fast when we have a whole lot of files...
337 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){
338 reslimit.rlim_cur = reslimit.rlim_max;
339 setrlimit(RLIMIT_DATA , &reslimit);
343 * should file size limits be waived? if the os limits us, this is
344 * needed if we want to write a large archive
346 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){
347 reslimit.rlim_cur = reslimit.rlim_max;
348 setrlimit(RLIMIT_FSIZE , &reslimit);
352 * increase the size the stack can grow to
354 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){
355 reslimit.rlim_cur = reslimit.rlim_max;
356 setrlimit(RLIMIT_STACK , &reslimit);
360 * not really needed, but doesn't hurt
362 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){
363 reslimit.rlim_cur = reslimit.rlim_max;
364 setrlimit(RLIMIT_RSS , &reslimit);
368 * signal handling to reset stored directory times and modes. Since
369 * we deal with broken pipes via failed writes we ignore it. We also
370 * deal with any file size limit through failed writes. Cpu time
371 * limits are caught and a cleanup is forced.
373 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) ||
374 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) ||
375 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) ||
376 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) {
377 paxwarn(1, "Unable to set up signal mask");
378 return(-1);
380 memset(&n_hand, 0, sizeof n_hand);
381 n_hand.sa_mask = s_mask;
382 n_hand.sa_flags = 0;
383 n_hand.sa_handler = sig_cleanup;
385 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) &&
386 (o_hand.sa_handler == SIG_IGN) &&
387 (sigaction(SIGHUP, &o_hand, &o_hand) < 0))
388 goto out;
390 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) &&
391 (o_hand.sa_handler == SIG_IGN) &&
392 (sigaction(SIGTERM, &o_hand, &o_hand) < 0))
393 goto out;
395 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) &&
396 (o_hand.sa_handler == SIG_IGN) &&
397 (sigaction(SIGINT, &o_hand, &o_hand) < 0))
398 goto out;
400 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) &&
401 (o_hand.sa_handler == SIG_IGN) &&
402 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0))
403 goto out;
405 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) &&
406 (o_hand.sa_handler == SIG_IGN) &&
407 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0))
408 goto out;
410 n_hand.sa_handler = SIG_IGN;
411 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) ||
412 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0))
413 goto out;
414 return(0);
416 out:
417 syswarn(1, errno, "Unable to set up signal handler");
418 return(-1);