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 38 #ifndef lint 39 static char copyright[] = 40 "@(#) Copyright (c) 1992, 1993\n\ 41 The Regents of the University of California. All rights reserved.\n"; 42 #endif /* not lint */ 43 44 #ifndef lint 45 /*static char sccsid[] = "from: @(#)pax.c 8.2 (Berkeley) 4/18/94";*/ 46 static char *rcsid = "$Id: pax.c,v 1.3 1994/06/14 00:43:06 mycroft Exp $"; 47 #endif /* not lint */ 48 49 #include <stdio.h> 50 #include <sys/types.h> 51 #include <sys/param.h> 52 #include <sys/stat.h> 53 #include <sys/time.h> 54 #include <sys/resource.h> 55 #include <signal.h> 56 #include <unistd.h> 57 #include <stdlib.h> 58 #include <errno.h> 59 #include "pax.h" 60 #include "extern.h" 61 static int gen_init __P((void)); 62 63 /* 64 * PAX main routines, general globals and some simple start up routines 65 */ 66 67 /* 68 * Variables that can be accessed by any routine within pax 69 */ 70 int act = DEFOP; /* read/write/append/copy */ 71 FSUB *frmt = NULL; /* archive format type */ 72 int cflag; /* match all EXCEPT pattern/file */ 73 int dflag; /* directory member match only */ 74 int iflag; /* interactive file/archive rename */ 75 int kflag; /* do not overwrite existing files */ 76 int lflag; /* use hard links when possible */ 77 int nflag; /* select first archive member match */ 78 int tflag; /* restore access time after read */ 79 int uflag; /* ignore older modification time files */ 80 int vflag; /* produce verbose output */ 81 int Dflag; /* same as uflag except inode change time */ 82 int Hflag; /* follow command line symlinks (write only) */ 83 int Lflag; /* follow symlinks when writing */ 84 int Xflag; /* archive files with same device id only */ 85 int Yflag; /* same as Dflg except after name mode */ 86 int Zflag; /* same as uflg except after name mode */ 87 int vfpart; /* is partial verbose output in progress */ 88 int patime = 1; /* preserve file access time */ 89 int pmtime = 1; /* preserve file modification times */ 90 int pmode; /* preserve file mode bits */ 91 int pids; /* preserve file uid/gid */ 92 int exit_val; /* exit value */ 93 int docrc; /* check/create file crc */ 94 char *dirptr; /* destination dir in a copy */ 95 char *ltmfrmt; /* -v locale time format (if any) */ 96 char *argv0; /* root of argv[0] */ 97 sigset_t s_mask; /* signal mask for cleanup critical sect */ 98 99 /* 100 * PAX - Portable Archive Interchange 101 * 102 * A utility to read, write, and write lists of the members of archive 103 * files and copy directory hierarchies. A variety of archive formats 104 * are supported (some are described in POSIX 1003.1 10.1): 105 * 106 * ustar - 10.1.1 extended tar interchange format 107 * cpio - 10.1.2 extended cpio interchange format 108 * tar - old BSD 4.3 tar format 109 * binary cpio - old cpio with binary header format 110 * sysVR4 cpio - with and without CRC 111 * 112 * This version is a superset of IEEE Std 1003.2b-d3 113 * 114 * Summary of Extensions to the IEEE Standard: 115 * 116 * 1 READ ENHANCEMENTS 117 * 1.1 Operations which read archives will continue to operate even when 118 * processing archives which may be damaged, truncated, or fail to meet 119 * format specs in several different ways. Damaged sections of archives 120 * are detected and avoided if possible. Attempts will be made to resync 121 * archive read operations even with badly damaged media. 122 * 1.2 Blocksize requirements are not strictly enforced on archive read. 123 * Tapes which have variable sized records can be read without errors. 124 * 1.3 The user can specify via the non-standard option flag -E if error 125 * resync operation should stop on a media error, try a specified number 126 * of times to correct, or try to correct forever. 127 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks 128 * of all zeros will be restored with holes appropriate for the target 129 * filesystem 130 * 1.5 The user is notified whenever something is found during archive 131 * read operations which violates spec (but the read will continue). 132 * 1.6 Multiple archive volumes can be read and may span over different 133 * archive devices 134 * 1.7 Rigidly restores all file attributes exactly as they are stored on the 135 * archive. 136 * 1.8 Modification change time ranges can be specified via multiple -T 137 * options. These allow a user to select files whose modification time 138 * lies within a specific time range. 139 * 1.9 Files can be selected based on owner (user name or uid) via one or more 140 * -U options. 141 * 1.10 Files can be selected based on group (group name or gid) via one o 142 * more -G options. 143 * 1.11 File modification time can be checked against exisiting file after 144 * name modification (-Z) 145 * 146 * 2 WRITE ENHANCEMENTS 147 * 2.1 Write operation will stop instead of allowing a user to create a flawed 148 * flawed archive (due to any problem). 149 * 2.2 Archives writtens by pax are forced to strictly conform to both the 150 * archive and pax the spceific format specifications. 151 * 2.3 Blocking size and format is rigidly enforced on writes. 152 * 2.4 Formats which may exhibit header overflow problems (they have fields 153 * too small for large file systems, such as inode number storage), use 154 * routines designed to repair this problem. These techniques still 155 * conform to both pax and format specifications, but no longer truncate 156 * these fields. This removes any restrictions on using these archive 157 * formats on large file systems. 158 * 2.5 Multiple archive volumes can be written and may span over different 159 * archive devices 160 * 2.6 A archive volume record limit allows the user to specify the number 161 * of bytes stored on an archive volume. When reached the user is 162 * prompted for the next archive volume. This is specified with the 163 * non-standard -B flag. THe limit is rounded up to the next blocksize. 164 * 2.7 All archive padding during write use zero filled sections. This makes 165 * it much easier to pull data out of flawed archive during read 166 * operations. 167 * 2.8 Access time reset with the -t applies to all file nodes (including 168 * directories). 169 * 2.9 Symbolic links can be followed with -L (optional in the spec). 170 * 2.10 Modification or inode change time ranges can be specified via 171 * multiple -T options. These allow a user to select files whose 172 * modification or inode change time lies within a specific time range. 173 * 2.11 Files can be selected based on owner (user name or uid) via one or more 174 * -U options. 175 * 2.12 Files can be selected based on group (group name or gid) via one o 176 * more -G options. 177 * 2.13 Symlinks which appear on the command line can be followed (without 178 * following other symlinks; -H flag) 179 * 180 * 3 COPY ENHANCEMENTS 181 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes 182 * into zero filled blocks. The file copy is created with holes which are 183 * appropriate for the target filesystem 184 * 3.2 Access time as well as modification time on copied file trees can be 185 * preserved with the appropriate -p options. 186 * 3.3 Access time reset with the -t applies to all file nodes (including 187 * directories). 188 * 3.4 Symbolic links can be followed with -L (optional in the spec). 189 * 3.5 Modification or inode change time ranges can be specified via 190 * multiple -T options. These allow a user to select files whose 191 * modification or inode change time lies within a specific time range. 192 * 3.6 Files can be selected based on owner (user name or uid) via one or more 193 * -U options. 194 * 3.7 Files can be selected based on group (group name or gid) via one o 195 * more -G options. 196 * 3.8 Symlinks which appear on the command line can be followed (without 197 * following other symlinks; -H flag) 198 * 3.9 File inode change time can be checked against exisiting file before 199 * name modification (-D) 200 * 3.10 File inode change time can be checked against exisiting file after 201 * name modification (-Y) 202 * 3.11 File modification time can be checked against exisiting file after 203 * name modification (-Z) 204 * 205 * 4 GENERAL ENHANCEMENTS 206 * 4.1 Internal structure is designed to isolate format dependent and 207 * independent functions. Formats are selected via a format driver table. 208 * This encourages the addition of new archive formats by only having to 209 * write those routines which id, read and write the archive header. 210 */ 211 212 /* 213 * main() 214 * parse options, set up and operate as specified by the user. 215 * any operational flaw will set exit_val to non-zero 216 * Return: 0 if ok, 1 otherwise 217 */ 218 219 #if __STDC__ 220 int 221 main(int argc, char **argv) 222 #else 223 int 224 main(argc, argv) 225 int argc; 226 char **argv; 227 #endif 228 { 229 /* 230 * parse options, determine operational mode, general init 231 */ 232 options(argc, argv); 233 if ((gen_init() < 0) || (tty_init() < 0)) 234 return(exit_val); 235 236 /* 237 * select a primary operation mode 238 */ 239 switch(act) { 240 case EXTRACT: 241 extract(); 242 break; 243 case ARCHIVE: 244 archive(); 245 break; 246 case APPND: 247 append(); 248 break; 249 case COPY: 250 copy(); 251 break; 252 default: 253 case LIST: 254 list(); 255 break; 256 } 257 return(exit_val); 258 } 259 260 /* 261 * sig_cleanup() 262 * when interrupted we try to do whatever delayed processing we can. 263 * This is not critical, but we really ought to limit our damage when we 264 * are aborted by the user. 265 * Return: 266 * never.... 267 */ 268 269 #if __STDC__ 270 void 271 sig_cleanup(int which_sig) 272 #else 273 void 274 sig_cleanup(which_sig) 275 int which_sig; 276 #endif 277 { 278 /* 279 * restore modes and times for any dirs we may have created 280 * or any dirs we may have read. Set vflag and vfpart so the user 281 * will clearly see the message on a line by itself. 282 */ 283 vflag = vfpart = 1; 284 if (which_sig == SIGXCPU) 285 warn(0, "Cpu time limit reached, cleaning up."); 286 else 287 warn(0, "Signal caught, cleaning up."); 288 289 ar_close(); 290 proc_dir(); 291 if (tflag) 292 atdir_end(); 293 exit(1); 294 } 295 296 /* 297 * gen_init() 298 * general setup routines. Not all are required, but they really help 299 * when dealing with a medium to large sized archives. 300 */ 301 302 #if __STDC__ 303 static int 304 gen_init(void) 305 #else 306 static int 307 gen_init() 308 #endif 309 { 310 struct rlimit reslimit; 311 struct sigaction n_hand; 312 struct sigaction o_hand; 313 314 /* 315 * Really needed to handle large archives. We can run out of memory for 316 * internal tables really fast when we have a whole lot of files... 317 */ 318 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){ 319 reslimit.rlim_cur = reslimit.rlim_max; 320 (void)setrlimit(RLIMIT_DATA , &reslimit); 321 } 322 323 /* 324 * should file size limits be waived? if the os limits us, this is 325 * needed if we want to write a large archive 326 */ 327 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){ 328 reslimit.rlim_cur = reslimit.rlim_max; 329 (void)setrlimit(RLIMIT_FSIZE , &reslimit); 330 } 331 332 /* 333 * increase the size the stack can grow to 334 */ 335 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){ 336 reslimit.rlim_cur = reslimit.rlim_max; 337 (void)setrlimit(RLIMIT_STACK , &reslimit); 338 } 339 340 /* 341 * not really needed, but doesn't hurt 342 */ 343 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){ 344 reslimit.rlim_cur = reslimit.rlim_max; 345 (void)setrlimit(RLIMIT_RSS , &reslimit); 346 } 347 348 /* 349 * Handle posix locale 350 * 351 * set user defines time printing format for -v option 352 */ 353 ltmfrmt = getenv("LC_TIME"); 354 355 /* 356 * signal handling to reset stored directory times and modes. Since 357 * we deal with broken pipes via failed writes we ignore it. We also 358 * deal with any file size limit thorugh failed writes. Cpu time 359 * limits are caught and a cleanup is forced. 360 */ 361 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) || 362 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) || 363 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) || 364 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) { 365 warn(1, "Unable to set up signal mask"); 366 return(-1); 367 } 368 n_hand.sa_mask = s_mask; 369 n_hand.sa_flags = 0; 370 n_hand.sa_handler = sig_cleanup; 371 372 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) && 373 (o_hand.sa_handler == SIG_IGN) && 374 (sigaction(SIGHUP, &o_hand, &o_hand) < 0)) 375 goto out; 376 377 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) && 378 (o_hand.sa_handler == SIG_IGN) && 379 (sigaction(SIGTERM, &o_hand, &o_hand) < 0)) 380 goto out; 381 382 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) && 383 (o_hand.sa_handler == SIG_IGN) && 384 (sigaction(SIGINT, &o_hand, &o_hand) < 0)) 385 goto out; 386 387 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) && 388 (o_hand.sa_handler == SIG_IGN) && 389 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0)) 390 goto out; 391 392 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) && 393 (o_hand.sa_handler == SIG_IGN) && 394 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0)) 395 goto out; 396 397 n_hand.sa_handler = SIG_IGN; 398 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) || 399 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0)) 400 goto out; 401 return(0); 402 403 out: 404 syswarn(1, errno, "Unable to set up signal handler"); 405 return(-1); 406 } 407