pax.c revision 31666
1/*- 2 * Copyright (c) 1992 Keith Muller. 3 * Copyright (c) 1992, 1993 4 * The Regents of the University of California. All rights reserved. 5 * 6 * This code is derived from software contributed to Berkeley by 7 * Keith Muller of the University of California, San Diego. 8 * 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. 24 * 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. 36 * 37 * $Id: pax.c,v 1.9 1997/08/29 16:12:28 sos Exp $ 38 */ 39 40#ifndef lint 41static char const copyright[] = 42"@(#) Copyright (c) 1992, 1993\n\ 43 The Regents of the University of California. All rights reserved.\n"; 44#endif /* not lint */ 45 46#ifndef lint 47static char const sccsid[] = "@(#)pax.c 8.2 (Berkeley) 4/18/94"; 48#endif /* not lint */ 49 50#include <sys/types.h> 51#include <sys/stat.h> 52#include <sys/time.h> 53#include <sys/resource.h> 54#include <errno.h> 55#include <locale.h> 56#include <signal.h> 57#include <stdio.h> 58#include <stdlib.h> 59#include <unistd.h> 60#include "pax.h" 61#include "extern.h" 62static int gen_init __P((void)); 63 64/* 65 * PAX main routines, general globals and some simple start up routines 66 */ 67 68/* 69 * Variables that can be accessed by any routine within pax 70 */ 71int act = DEFOP; /* read/write/append/copy */ 72FSUB *frmt = NULL; /* archive format type */ 73int cflag; /* match all EXCEPT pattern/file */ 74int dflag; /* directory member match only */ 75int iflag; /* interactive file/archive rename */ 76int kflag; /* do not overwrite existing files */ 77int lflag; /* use hard links when possible */ 78int nflag; /* select first archive member match */ 79int tflag; /* restore access time after read */ 80int uflag; /* ignore older modification time files */ 81int vflag; /* produce verbose output */ 82int Dflag; /* same as uflag except inode change time */ 83int Hflag; /* follow command line symlinks (write only) */ 84int Lflag; /* follow symlinks when writing */ 85int Xflag; /* archive files with same device id only */ 86int Yflag; /* same as Dflg except after name mode */ 87int Zflag; /* same as uflg except after name mode */ 88int vfpart; /* is partial verbose output in progress */ 89int patime = 1; /* preserve file access time */ 90int pmtime = 1; /* preserve file modification times */ 91int pmode; /* preserve file mode bits */ 92int pids; /* preserve file uid/gid */ 93int exit_val; /* exit value */ 94int docrc; /* check/create file crc */ 95char *dirptr; /* destination dir in a copy */ 96char *ltmfrmt; /* -v locale time format (if any) */ 97char *argv0; /* root of argv[0] */ 98sigset_t s_mask; /* signal mask for cleanup critical sect */ 99 100/* 101 * PAX - Portable Archive Interchange 102 * 103 * A utility to read, write, and write lists of the members of archive 104 * files and copy directory hierarchies. A variety of archive formats 105 * are supported (some are described in POSIX 1003.1 10.1): 106 * 107 * ustar - 10.1.1 extended tar interchange format 108 * cpio - 10.1.2 extended cpio interchange format 109 * tar - old BSD 4.3 tar format 110 * binary cpio - old cpio with binary header format 111 * sysVR4 cpio - with and without CRC 112 * 113 * This version is a superset of IEEE Std 1003.2b-d3 114 * 115 * Summary of Extensions to the IEEE Standard: 116 * 117 * 1 READ ENHANCEMENTS 118 * 1.1 Operations which read archives will continue to operate even when 119 * processing archives which may be damaged, truncated, or fail to meet 120 * format specs in several different ways. Damaged sections of archives 121 * are detected and avoided if possible. Attempts will be made to resync 122 * archive read operations even with badly damaged media. 123 * 1.2 Blocksize requirements are not strictly enforced on archive read. 124 * Tapes which have variable sized records can be read without errors. 125 * 1.3 The user can specify via the non-standard option flag -E if error 126 * resync operation should stop on a media error, try a specified number 127 * of times to correct, or try to correct forever. 128 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks 129 * of all zeros will be restored with holes appropriate for the target 130 * filesystem 131 * 1.5 The user is notified whenever something is found during archive 132 * read operations which violates spec (but the read will continue). 133 * 1.6 Multiple archive volumes can be read and may span over different 134 * archive devices 135 * 1.7 Rigidly restores all file attributes exactly as they are stored on the 136 * archive. 137 * 1.8 Modification change time ranges can be specified via multiple -T 138 * options. These allow a user to select files whose modification time 139 * lies within a specific time range. 140 * 1.9 Files can be selected based on owner (user name or uid) via one or more 141 * -U options. 142 * 1.10 Files can be selected based on group (group name or gid) via one o 143 * more -G options. 144 * 1.11 File modification time can be checked against exisiting file after 145 * name modification (-Z) 146 * 147 * 2 WRITE ENHANCEMENTS 148 * 2.1 Write operation will stop instead of allowing a user to create a flawed 149 * flawed archive (due to any problem). 150 * 2.2 Archives writtens by pax are forced to strictly conform to both the 151 * archive and pax the spceific format specifications. 152 * 2.3 Blocking size and format is rigidly enforced on writes. 153 * 2.4 Formats which may exhibit header overflow problems (they have fields 154 * too small for large file systems, such as inode number storage), use 155 * routines designed to repair this problem. These techniques still 156 * conform to both pax and format specifications, but no longer truncate 157 * these fields. This removes any restrictions on using these archive 158 * formats on large file systems. 159 * 2.5 Multiple archive volumes can be written and may span over different 160 * archive devices 161 * 2.6 A archive volume record limit allows the user to specify the number 162 * of bytes stored on an archive volume. When reached the user is 163 * prompted for the next archive volume. This is specified with the 164 * non-standard -B flag. THe limit is rounded up to the next blocksize. 165 * 2.7 All archive padding during write use zero filled sections. This makes 166 * it much easier to pull data out of flawed archive during read 167 * operations. 168 * 2.8 Access time reset with the -t applies to all file nodes (including 169 * directories). 170 * 2.9 Symbolic links can be followed with -L (optional in the spec). 171 * 2.10 Modification or inode change time ranges can be specified via 172 * multiple -T options. These allow a user to select files whose 173 * modification or inode change time lies within a specific time range. 174 * 2.11 Files can be selected based on owner (user name or uid) via one or more 175 * -U options. 176 * 2.12 Files can be selected based on group (group name or gid) via one o 177 * more -G options. 178 * 2.13 Symlinks which appear on the command line can be followed (without 179 * following other symlinks; -H flag) 180 * 181 * 3 COPY ENHANCEMENTS 182 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes 183 * into zero filled blocks. The file copy is created with holes which are 184 * appropriate for the target filesystem 185 * 3.2 Access time as well as modification time on copied file trees can be 186 * preserved with the appropriate -p options. 187 * 3.3 Access time reset with the -t applies to all file nodes (including 188 * directories). 189 * 3.4 Symbolic links can be followed with -L (optional in the spec). 190 * 3.5 Modification or inode change time ranges can be specified via 191 * multiple -T options. These allow a user to select files whose 192 * modification or inode change time lies within a specific time range. 193 * 3.6 Files can be selected based on owner (user name or uid) via one or more 194 * -U options. 195 * 3.7 Files can be selected based on group (group name or gid) via one o 196 * more -G options. 197 * 3.8 Symlinks which appear on the command line can be followed (without 198 * following other symlinks; -H flag) 199 * 3.9 File inode change time can be checked against exisiting file before 200 * name modification (-D) 201 * 3.10 File inode change time can be checked against exisiting file after 202 * name modification (-Y) 203 * 3.11 File modification time can be checked against exisiting file after 204 * name modification (-Z) 205 * 206 * 4 GENERAL ENHANCEMENTS 207 * 4.1 Internal structure is designed to isolate format dependent and 208 * independent functions. Formats are selected via a format driver table. 209 * This encourages the addition of new archive formats by only having to 210 * write those routines which id, read and write the archive header. 211 */ 212 213/* 214 * main() 215 * parse options, set up and operate as specified by the user. 216 * any operational flaw will set exit_val to non-zero 217 * Return: 0 if ok, 1 otherwise 218 */ 219 220#if __STDC__ 221int 222main(int argc, char **argv) 223#else 224int 225main(argc, argv) 226 int argc; 227 char **argv; 228#endif 229{ 230 (void) setlocale(LC_ALL, ""); 231 /* 232 * parse options, determine operational mode, general init 233 */ 234 options(argc, argv); 235 if ((gen_init() < 0) || (tty_init() < 0)) 236 return(exit_val); 237 238 /* 239 * select a primary operation mode 240 */ 241 switch(act) { 242 case EXTRACT: 243 extract(); 244 break; 245 case ARCHIVE: 246 archive(); 247 break; 248 case APPND: 249 append(); 250 break; 251 case COPY: 252 copy(); 253 break; 254 default: 255 case LIST: 256 list(); 257 break; 258 } 259 return(exit_val); 260} 261 262/* 263 * sig_cleanup() 264 * when interrupted we try to do whatever delayed processing we can. 265 * This is not critical, but we really ought to limit our damage when we 266 * are aborted by the user. 267 * Return: 268 * never.... 269 */ 270 271#if __STDC__ 272void 273sig_cleanup(int which_sig) 274#else 275void 276sig_cleanup(which_sig) 277 int which_sig; 278#endif 279{ 280 /* 281 * restore modes and times for any dirs we may have created 282 * or any dirs we may have read. Set vflag and vfpart so the user 283 * will clearly see the message on a line by itself. 284 */ 285 vflag = vfpart = 1; 286 if (which_sig == SIGXCPU) 287 pax_warn(0, "Cpu time limit reached, cleaning up."); 288 else 289 pax_warn(0, "Signal caught, cleaning up."); 290 291 ar_close(); 292 proc_dir(); 293 if (tflag) 294 atdir_end(); 295 exit(1); 296} 297 298/* 299 * gen_init() 300 * general setup routines. Not all are required, but they really help 301 * when dealing with a medium to large sized archives. 302 */ 303 304#if __STDC__ 305static int 306gen_init(void) 307#else 308static int 309gen_init() 310#endif 311{ 312 struct rlimit reslimit; 313 struct sigaction n_hand; 314 struct sigaction o_hand; 315 316 /* 317 * Really needed to handle large archives. We can run out of memory for 318 * internal tables really fast when we have a whole lot of files... 319 */ 320 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){ 321 reslimit.rlim_cur = reslimit.rlim_max; 322 (void)setrlimit(RLIMIT_DATA , &reslimit); 323 } 324 325 /* 326 * should file size limits be waived? if the os limits us, this is 327 * needed if we want to write a large archive 328 */ 329 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){ 330 reslimit.rlim_cur = reslimit.rlim_max; 331 (void)setrlimit(RLIMIT_FSIZE , &reslimit); 332 } 333 334 /* 335 * increase the size the stack can grow to 336 */ 337 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){ 338 reslimit.rlim_cur = reslimit.rlim_max; 339 (void)setrlimit(RLIMIT_STACK , &reslimit); 340 } 341 342 /* 343 * not really needed, but doesn't hurt 344 */ 345 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){ 346 reslimit.rlim_cur = reslimit.rlim_max; 347 (void)setrlimit(RLIMIT_RSS , &reslimit); 348 } 349 350 /* 351 * Handle posix locale 352 * 353 * set user defines time printing format for -v option 354 */ 355 ltmfrmt = getenv("LC_TIME"); 356 357 /* 358 * signal handling to reset stored directory times and modes. Since 359 * we deal with broken pipes via failed writes we ignore it. We also 360 * deal with any file size limit thorugh failed writes. Cpu time 361 * limits are caught and a cleanup is forced. 362 */ 363 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) || 364 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) || 365 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) || 366 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) { 367 pax_warn(1, "Unable to set up signal mask"); 368 return(-1); 369 } 370 n_hand.sa_mask = s_mask; 371 n_hand.sa_flags = 0; 372 n_hand.sa_handler = sig_cleanup; 373 374 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) && 375 (o_hand.sa_handler == SIG_IGN) && 376 (sigaction(SIGHUP, &o_hand, &o_hand) < 0)) 377 goto out; 378 379 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) && 380 (o_hand.sa_handler == SIG_IGN) && 381 (sigaction(SIGTERM, &o_hand, &o_hand) < 0)) 382 goto out; 383 384 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) && 385 (o_hand.sa_handler == SIG_IGN) && 386 (sigaction(SIGINT, &o_hand, &o_hand) < 0)) 387 goto out; 388 389 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) && 390 (o_hand.sa_handler == SIG_IGN) && 391 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0)) 392 goto out; 393 394 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) && 395 (o_hand.sa_handler == SIG_IGN) && 396 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0)) 397 goto out; 398 399 n_hand.sa_handler = SIG_IGN; 400 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) || 401 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0)) 402 goto out; 403 return(0); 404 405 out: 406 sys_warn(1, errno, "Unable to set up signal handler"); 407 return(-1); 408} 409