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 * $Id$ 38 */ 39 40 #ifndef lint 41 static 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 47 static char const sccsid[] = "@(#)pax.c 8.2 (Berkeley) 4/18/94"; 48 #endif /* not lint */ 49 50 #include <stdio.h> 51 #include <sys/types.h> 52 #include <sys/param.h> 53 #include <sys/stat.h> 54 #include <sys/time.h> 55 #include <sys/resource.h> 56 #include <signal.h> 57 #include <unistd.h> 58 #include <stdlib.h> 59 #include <errno.h> 60 #include <locale.h> 61 #include "pax.h" 62 #include "extern.h" 63 static int gen_init __P((void)); 64 65 /* 66 * PAX main routines, general globals and some simple start up routines 67 */ 68 69 /* 70 * Variables that can be accessed by any routine within pax 71 */ 72 int act = DEFOP; /* read/write/append/copy */ 73 FSUB *frmt = NULL; /* archive format type */ 74 int cflag; /* match all EXCEPT pattern/file */ 75 int dflag; /* directory member match only */ 76 int iflag; /* interactive file/archive rename */ 77 int kflag; /* do not overwrite existing files */ 78 int lflag; /* use hard links when possible */ 79 int nflag; /* select first archive member match */ 80 int tflag; /* restore access time after read */ 81 int uflag; /* ignore older modification time files */ 82 int vflag; /* produce verbose output */ 83 int Dflag; /* same as uflag except inode change time */ 84 int Hflag; /* follow command line symlinks (write only) */ 85 int Lflag; /* follow symlinks when writing */ 86 int Xflag; /* archive files with same device id only */ 87 int Yflag; /* same as Dflg except after name mode */ 88 int Zflag; /* same as uflg except after name mode */ 89 int vfpart; /* is partial verbose output in progress */ 90 int patime = 1; /* preserve file access time */ 91 int pmtime = 1; /* preserve file modification times */ 92 int pmode; /* preserve file mode bits */ 93 int pids; /* preserve file uid/gid */ 94 int exit_val; /* exit value */ 95 int docrc; /* check/create file crc */ 96 char *dirptr; /* destination dir in a copy */ 97 char *ltmfrmt; /* -v locale time format (if any) */ 98 char *argv0; /* root of argv[0] */ 99 sigset_t s_mask; /* signal mask for cleanup critical sect */ 100 101 /* 102 * PAX - Portable Archive Interchange 103 * 104 * A utility to read, write, and write lists of the members of archive 105 * files and copy directory hierarchies. A variety of archive formats 106 * are supported (some are described in POSIX 1003.1 10.1): 107 * 108 * ustar - 10.1.1 extended tar interchange format 109 * cpio - 10.1.2 extended cpio interchange format 110 * tar - old BSD 4.3 tar format 111 * binary cpio - old cpio with binary header format 112 * sysVR4 cpio - with and without CRC 113 * 114 * This version is a superset of IEEE Std 1003.2b-d3 115 * 116 * Summary of Extensions to the IEEE Standard: 117 * 118 * 1 READ ENHANCEMENTS 119 * 1.1 Operations which read archives will continue to operate even when 120 * processing archives which may be damaged, truncated, or fail to meet 121 * format specs in several different ways. Damaged sections of archives 122 * are detected and avoided if possible. Attempts will be made to resync 123 * archive read operations even with badly damaged media. 124 * 1.2 Blocksize requirements are not strictly enforced on archive read. 125 * Tapes which have variable sized records can be read without errors. 126 * 1.3 The user can specify via the non-standard option flag -E if error 127 * resync operation should stop on a media error, try a specified number 128 * of times to correct, or try to correct forever. 129 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks 130 * of all zeros will be restored with holes appropriate for the target 131 * filesystem 132 * 1.5 The user is notified whenever something is found during archive 133 * read operations which violates spec (but the read will continue). 134 * 1.6 Multiple archive volumes can be read and may span over different 135 * archive devices 136 * 1.7 Rigidly restores all file attributes exactly as they are stored on the 137 * archive. 138 * 1.8 Modification change time ranges can be specified via multiple -T 139 * options. These allow a user to select files whose modification time 140 * lies within a specific time range. 141 * 1.9 Files can be selected based on owner (user name or uid) via one or more 142 * -U options. 143 * 1.10 Files can be selected based on group (group name or gid) via one o 144 * more -G options. 145 * 1.11 File modification time can be checked against exisiting file after 146 * name modification (-Z) 147 * 148 * 2 WRITE ENHANCEMENTS 149 * 2.1 Write operation will stop instead of allowing a user to create a flawed 150 * flawed archive (due to any problem). 151 * 2.2 Archives writtens by pax are forced to strictly conform to both the 152 * archive and pax the spceific format specifications. 153 * 2.3 Blocking size and format is rigidly enforced on writes. 154 * 2.4 Formats which may exhibit header overflow problems (they have fields 155 * too small for large file systems, such as inode number storage), use 156 * routines designed to repair this problem. These techniques still 157 * conform to both pax and format specifications, but no longer truncate 158 * these fields. This removes any restrictions on using these archive 159 * formats on large file systems. 160 * 2.5 Multiple archive volumes can be written and may span over different 161 * archive devices 162 * 2.6 A archive volume record limit allows the user to specify the number 163 * of bytes stored on an archive volume. When reached the user is 164 * prompted for the next archive volume. This is specified with the 165 * non-standard -B flag. THe limit is rounded up to the next blocksize. 166 * 2.7 All archive padding during write use zero filled sections. This makes 167 * it much easier to pull data out of flawed archive during read 168 * operations. 169 * 2.8 Access time reset with the -t applies to all file nodes (including 170 * directories). 171 * 2.9 Symbolic links can be followed with -L (optional in the spec). 172 * 2.10 Modification or inode change time ranges can be specified via 173 * multiple -T options. These allow a user to select files whose 174 * modification or inode change time lies within a specific time range. 175 * 2.11 Files can be selected based on owner (user name or uid) via one or more 176 * -U options. 177 * 2.12 Files can be selected based on group (group name or gid) via one o 178 * more -G options. 179 * 2.13 Symlinks which appear on the command line can be followed (without 180 * following other symlinks; -H flag) 181 * 182 * 3 COPY ENHANCEMENTS 183 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes 184 * into zero filled blocks. The file copy is created with holes which are 185 * appropriate for the target filesystem 186 * 3.2 Access time as well as modification time on copied file trees can be 187 * preserved with the appropriate -p options. 188 * 3.3 Access time reset with the -t applies to all file nodes (including 189 * directories). 190 * 3.4 Symbolic links can be followed with -L (optional in the spec). 191 * 3.5 Modification or inode change time ranges can be specified via 192 * multiple -T options. These allow a user to select files whose 193 * modification or inode change time lies within a specific time range. 194 * 3.6 Files can be selected based on owner (user name or uid) via one or more 195 * -U options. 196 * 3.7 Files can be selected based on group (group name or gid) via one o 197 * more -G options. 198 * 3.8 Symlinks which appear on the command line can be followed (without 199 * following other symlinks; -H flag) 200 * 3.9 File inode change time can be checked against exisiting file before 201 * name modification (-D) 202 * 3.10 File inode change time can be checked against exisiting file after 203 * name modification (-Y) 204 * 3.11 File modification time can be checked against exisiting file after 205 * name modification (-Z) 206 * 207 * 4 GENERAL ENHANCEMENTS 208 * 4.1 Internal structure is designed to isolate format dependent and 209 * independent functions. Formats are selected via a format driver table. 210 * This encourages the addition of new archive formats by only having to 211 * write those routines which id, read and write the archive header. 212 */ 213 214 /* 215 * main() 216 * parse options, set up and operate as specified by the user. 217 * any operational flaw will set exit_val to non-zero 218 * Return: 0 if ok, 1 otherwise 219 */ 220 221 #if __STDC__ 222 int 223 main(int argc, char **argv) 224 #else 225 int 226 main(argc, argv) 227 int argc; 228 char **argv; 229 #endif 230 { 231 (void) setlocale(LC_ALL, ""); 232 /* 233 * parse options, determine operational mode, general init 234 */ 235 options(argc, argv); 236 if ((gen_init() < 0) || (tty_init() < 0)) 237 return(exit_val); 238 239 /* 240 * select a primary operation mode 241 */ 242 switch(act) { 243 case EXTRACT: 244 extract(); 245 break; 246 case ARCHIVE: 247 archive(); 248 break; 249 case APPND: 250 append(); 251 break; 252 case COPY: 253 copy(); 254 break; 255 default: 256 case LIST: 257 list(); 258 break; 259 } 260 return(exit_val); 261 } 262 263 /* 264 * sig_cleanup() 265 * when interrupted we try to do whatever delayed processing we can. 266 * This is not critical, but we really ought to limit our damage when we 267 * are aborted by the user. 268 * Return: 269 * never.... 270 */ 271 272 #if __STDC__ 273 void 274 sig_cleanup(int which_sig) 275 #else 276 void 277 sig_cleanup(which_sig) 278 int which_sig; 279 #endif 280 { 281 /* 282 * restore modes and times for any dirs we may have created 283 * or any dirs we may have read. Set vflag and vfpart so the user 284 * will clearly see the message on a line by itself. 285 */ 286 vflag = vfpart = 1; 287 if (which_sig == SIGXCPU) 288 warn(0, "Cpu time limit reached, cleaning up."); 289 else 290 warn(0, "Signal caught, cleaning up."); 291 292 ar_close(); 293 proc_dir(); 294 if (tflag) 295 atdir_end(); 296 exit(1); 297 } 298 299 /* 300 * gen_init() 301 * general setup routines. Not all are required, but they really help 302 * when dealing with a medium to large sized archives. 303 */ 304 305 #if __STDC__ 306 static int 307 gen_init(void) 308 #else 309 static int 310 gen_init() 311 #endif 312 { 313 struct rlimit reslimit; 314 struct sigaction n_hand; 315 struct sigaction o_hand; 316 317 /* 318 * Really needed to handle large archives. We can run out of memory for 319 * internal tables really fast when we have a whole lot of files... 320 */ 321 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){ 322 reslimit.rlim_cur = reslimit.rlim_max; 323 (void)setrlimit(RLIMIT_DATA , &reslimit); 324 } 325 326 /* 327 * should file size limits be waived? if the os limits us, this is 328 * needed if we want to write a large archive 329 */ 330 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){ 331 reslimit.rlim_cur = reslimit.rlim_max; 332 (void)setrlimit(RLIMIT_FSIZE , &reslimit); 333 } 334 335 /* 336 * increase the size the stack can grow to 337 */ 338 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){ 339 reslimit.rlim_cur = reslimit.rlim_max; 340 (void)setrlimit(RLIMIT_STACK , &reslimit); 341 } 342 343 /* 344 * not really needed, but doesn't hurt 345 */ 346 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){ 347 reslimit.rlim_cur = reslimit.rlim_max; 348 (void)setrlimit(RLIMIT_RSS , &reslimit); 349 } 350 351 /* 352 * Handle posix locale 353 * 354 * set user defines time printing format for -v option 355 */ 356 ltmfrmt = getenv("LC_TIME"); 357 358 /* 359 * signal handling to reset stored directory times and modes. Since 360 * we deal with broken pipes via failed writes we ignore it. We also 361 * deal with any file size limit thorugh failed writes. Cpu time 362 * limits are caught and a cleanup is forced. 363 */ 364 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) || 365 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) || 366 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) || 367 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) { 368 warn(1, "Unable to set up signal mask"); 369 return(-1); 370 } 371 n_hand.sa_mask = s_mask; 372 n_hand.sa_flags = 0; 373 n_hand.sa_handler = sig_cleanup; 374 375 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) && 376 (o_hand.sa_handler == SIG_IGN) && 377 (sigaction(SIGHUP, &o_hand, &o_hand) < 0)) 378 goto out; 379 380 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) && 381 (o_hand.sa_handler == SIG_IGN) && 382 (sigaction(SIGTERM, &o_hand, &o_hand) < 0)) 383 goto out; 384 385 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) && 386 (o_hand.sa_handler == SIG_IGN) && 387 (sigaction(SIGINT, &o_hand, &o_hand) < 0)) 388 goto out; 389 390 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) && 391 (o_hand.sa_handler == SIG_IGN) && 392 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0)) 393 goto out; 394 395 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) && 396 (o_hand.sa_handler == SIG_IGN) && 397 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0)) 398 goto out; 399 400 n_hand.sa_handler = SIG_IGN; 401 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) || 402 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0)) 403 goto out; 404 return(0); 405 406 out: 407 syswarn(1, errno, "Unable to set up signal handler"); 408 return(-1); 409 } 410