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 const 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 #if 0 46 static char sccsid[] = "@(#)pax.c 8.2 (Berkeley) 4/18/94"; 47 #endif 48 static const char rcsid[] = 49 "$FreeBSD$"; 50 #endif /* not lint */ 51 52 #include <sys/types.h> 53 #include <sys/stat.h> 54 #include <sys/time.h> 55 #include <sys/resource.h> 56 #include <err.h> 57 #include <errno.h> 58 #include <fcntl.h> 59 #include <locale.h> 60 #include <paths.h> 61 #include <signal.h> 62 #include <stdio.h> 63 #include <stdlib.h> 64 #include <unistd.h> 65 #include "pax.h" 66 #include "extern.h" 67 static int gen_init __P((void)); 68 69 /* 70 * PAX main routines, general globals and some simple start up routines 71 */ 72 73 /* 74 * Variables that can be accessed by any routine within pax 75 */ 76 int act = DEFOP; /* read/write/append/copy */ 77 FSUB *frmt = NULL; /* archive format type */ 78 int cflag; /* match all EXCEPT pattern/file */ 79 int cwdfd; /* starting cwd */ 80 int dflag; /* directory member match only */ 81 int iflag; /* interactive file/archive rename */ 82 int kflag; /* do not overwrite existing files */ 83 int lflag; /* use hard links when possible */ 84 int nflag; /* select first archive member match */ 85 int tflag; /* restore access time after read */ 86 int uflag; /* ignore older modification time files */ 87 int vflag; /* produce verbose output */ 88 int Dflag; /* same as uflag except inode change time */ 89 int Hflag; /* follow command line symlinks (write only) */ 90 int Lflag; /* follow symlinks when writing */ 91 int Xflag; /* archive files with same device id only */ 92 int Yflag; /* same as Dflg except after name mode */ 93 int Zflag; /* same as uflg except after name mode */ 94 int vfpart; /* is partial verbose output in progress */ 95 int patime = 1; /* preserve file access time */ 96 int pmtime = 1; /* preserve file modification times */ 97 int nodirs; /* do not create directories as needed */ 98 int pmode; /* preserve file mode bits */ 99 int pids; /* preserve file uid/gid */ 100 int rmleadslash = 0; /* remove leading '/' from pathnames */ 101 int exit_val; /* exit value */ 102 int docrc; /* check/create file crc */ 103 char *dirptr; /* destination dir in a copy */ 104 char *argv0; /* root of argv[0] */ 105 sigset_t s_mask; /* signal mask for cleanup critical sect */ 106 FILE *listf = stderr; /* file pointer to print file list to */ 107 char *tempfile; /* tempfile to use for mkstemp(3) */ 108 char *tempbase; /* basename of tempfile to use for mkstemp(3) */ 109 110 /* 111 * PAX - Portable Archive Interchange 112 * 113 * A utility to read, write, and write lists of the members of archive 114 * files and copy directory hierarchies. A variety of archive formats 115 * are supported (some are described in POSIX 1003.1 10.1): 116 * 117 * ustar - 10.1.1 extended tar interchange format 118 * cpio - 10.1.2 extended cpio interchange format 119 * tar - old BSD 4.3 tar format 120 * binary cpio - old cpio with binary header format 121 * sysVR4 cpio - with and without CRC 122 * 123 * This version is a superset of IEEE Std 1003.2b-d3 124 * 125 * Summary of Extensions to the IEEE Standard: 126 * 127 * 1 READ ENHANCEMENTS 128 * 1.1 Operations which read archives will continue to operate even when 129 * processing archives which may be damaged, truncated, or fail to meet 130 * format specs in several different ways. Damaged sections of archives 131 * are detected and avoided if possible. Attempts will be made to resync 132 * archive read operations even with badly damaged media. 133 * 1.2 Blocksize requirements are not strictly enforced on archive read. 134 * Tapes which have variable sized records can be read without errors. 135 * 1.3 The user can specify via the non-standard option flag -E if error 136 * resync operation should stop on a media error, try a specified number 137 * of times to correct, or try to correct forever. 138 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks 139 * of all zeros will be restored with holes appropriate for the target 140 * filesystem 141 * 1.5 The user is notified whenever something is found during archive 142 * read operations which violates spec (but the read will continue). 143 * 1.6 Multiple archive volumes can be read and may span over different 144 * archive devices 145 * 1.7 Rigidly restores all file attributes exactly as they are stored on the 146 * archive. 147 * 1.8 Modification change time ranges can be specified via multiple -T 148 * options. These allow a user to select files whose modification time 149 * lies within a specific time range. 150 * 1.9 Files can be selected based on owner (user name or uid) via one or more 151 * -U options. 152 * 1.10 Files can be selected based on group (group name or gid) via one o 153 * more -G options. 154 * 1.11 File modification time can be checked against existing file after 155 * name modification (-Z) 156 * 157 * 2 WRITE ENHANCEMENTS 158 * 2.1 Write operation will stop instead of allowing a user to create a flawed 159 * flawed archive (due to any problem). 160 * 2.2 Archives written by pax are forced to strictly conform to both the 161 * archive and pax the specific format specifications. 162 * 2.3 Blocking size and format is rigidly enforced on writes. 163 * 2.4 Formats which may exhibit header overflow problems (they have fields 164 * too small for large file systems, such as inode number storage), use 165 * routines designed to repair this problem. These techniques still 166 * conform to both pax and format specifications, but no longer truncate 167 * these fields. This removes any restrictions on using these archive 168 * formats on large file systems. 169 * 2.5 Multiple archive volumes can be written and may span over different 170 * archive devices 171 * 2.6 A archive volume record limit allows the user to specify the number 172 * of bytes stored on an archive volume. When reached the user is 173 * prompted for the next archive volume. This is specified with the 174 * non-standard -B flag. The limit is rounded up to the next blocksize. 175 * 2.7 All archive padding during write use zero filled sections. This makes 176 * it much easier to pull data out of flawed archive during read 177 * operations. 178 * 2.8 Access time reset with the -t applies to all file nodes (including 179 * directories). 180 * 2.9 Symbolic links can be followed with -L (optional in the spec). 181 * 2.10 Modification or inode change time ranges can be specified via 182 * multiple -T options. These allow a user to select files whose 183 * modification or inode change time lies within a specific time range. 184 * 2.11 Files can be selected based on owner (user name or uid) via one or more 185 * -U options. 186 * 2.12 Files can be selected based on group (group name or gid) via one o 187 * more -G options. 188 * 2.13 Symlinks which appear on the command line can be followed (without 189 * following other symlinks; -H flag) 190 * 191 * 3 COPY ENHANCEMENTS 192 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes 193 * into zero filled blocks. The file copy is created with holes which are 194 * appropriate for the target filesystem 195 * 3.2 Access time as well as modification time on copied file trees can be 196 * preserved with the appropriate -p options. 197 * 3.3 Access time reset with the -t applies to all file nodes (including 198 * directories). 199 * 3.4 Symbolic links can be followed with -L (optional in the spec). 200 * 3.5 Modification or inode change time ranges can be specified via 201 * multiple -T options. These allow a user to select files whose 202 * modification or inode change time lies within a specific time range. 203 * 3.6 Files can be selected based on owner (user name or uid) via one or more 204 * -U options. 205 * 3.7 Files can be selected based on group (group name or gid) via one o 206 * more -G options. 207 * 3.8 Symlinks which appear on the command line can be followed (without 208 * following other symlinks; -H flag) 209 * 3.9 File inode change time can be checked against existing file before 210 * name modification (-D) 211 * 3.10 File inode change time can be checked against existing file after 212 * name modification (-Y) 213 * 3.11 File modification time can be checked against existing file after 214 * name modification (-Z) 215 * 216 * 4 GENERAL ENHANCEMENTS 217 * 4.1 Internal structure is designed to isolate format dependent and 218 * independent functions. Formats are selected via a format driver table. 219 * This encourages the addition of new archive formats by only having to 220 * write those routines which id, read and write the archive header. 221 */ 222 223 /* 224 * main() 225 * parse options, set up and operate as specified by the user. 226 * any operational flaw will set exit_val to non-zero 227 * Return: 0 if ok, 1 otherwise 228 */ 229 230 #ifdef __STDC__ 231 int 232 main(int argc, char **argv) 233 #else 234 int 235 main(argc, argv) 236 int argc; 237 char **argv; 238 #endif 239 { 240 char *tmpdir; 241 size_t tdlen; 242 243 (void) setlocale(LC_ALL, ""); 244 /* 245 * Keep a reference to cwd, so we can always come back home. 246 */ 247 cwdfd = open(".", O_RDONLY); 248 if (cwdfd < 0) { 249 syswarn(0, errno, "Can't open current working directory."); 250 return(exit_val); 251 } 252 253 /* 254 * Where should we put temporary files? 255 */ 256 if ((tmpdir = getenv("TMPDIR")) == NULL || *tmpdir == '\0') 257 tmpdir = _PATH_TMP; 258 tdlen = strlen(tmpdir); 259 while(tdlen > 0 && tmpdir[tdlen - 1] == '/') 260 tdlen--; 261 tempfile = malloc(tdlen + 1 + sizeof(_TFILE_BASE)); 262 if (tempfile == NULL) { 263 paxwarn(1, "Cannot allocate memory for temp file name."); 264 return(exit_val); 265 } 266 if (tdlen) 267 memcpy(tempfile, tmpdir, tdlen); 268 tempbase = tempfile + tdlen; 269 *tempbase++ = '/'; 270 271 /* 272 * parse options, determine operational mode, general init 273 */ 274 options(argc, argv); 275 if ((gen_init() < 0) || (tty_init() < 0)) 276 return(exit_val); 277 278 /* 279 * select a primary operation mode 280 */ 281 switch(act) { 282 case EXTRACT: 283 extract(); 284 break; 285 case ARCHIVE: 286 archive(); 287 break; 288 case APPND: 289 if (gzip_program != NULL) 290 err(1, "can not gzip while appending"); 291 append(); 292 break; 293 case COPY: 294 copy(); 295 break; 296 default: 297 case LIST: 298 list(); 299 break; 300 } 301 return(exit_val); 302 } 303 304 /* 305 * sig_cleanup() 306 * when interrupted we try to do whatever delayed processing we can. 307 * This is not critical, but we really ought to limit our damage when we 308 * are aborted by the user. 309 * Return: 310 * never.... 311 */ 312 313 #ifdef __STDC__ 314 void 315 sig_cleanup(int which_sig) 316 #else 317 void 318 sig_cleanup(which_sig) 319 int which_sig; 320 #endif 321 { 322 /* 323 * restore modes and times for any dirs we may have created 324 * or any dirs we may have read. Set vflag and vfpart so the user 325 * will clearly see the message on a line by itself. 326 */ 327 vflag = vfpart = 1; 328 if (which_sig == SIGXCPU) 329 paxwarn(0, "Cpu time limit reached, cleaning up."); 330 else 331 paxwarn(0, "Signal caught, cleaning up."); 332 333 ar_close(); 334 proc_dir(); 335 if (tflag) 336 atdir_end(); 337 exit(1); 338 } 339 340 /* 341 * gen_init() 342 * general setup routines. Not all are required, but they really help 343 * when dealing with a medium to large sized archives. 344 */ 345 346 #ifdef __STDC__ 347 static int 348 gen_init(void) 349 #else 350 static int 351 gen_init() 352 #endif 353 { 354 struct rlimit reslimit; 355 struct sigaction n_hand; 356 struct sigaction o_hand; 357 358 /* 359 * Really needed to handle large archives. We can run out of memory for 360 * internal tables really fast when we have a whole lot of files... 361 */ 362 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){ 363 reslimit.rlim_cur = reslimit.rlim_max; 364 (void)setrlimit(RLIMIT_DATA , &reslimit); 365 } 366 367 /* 368 * should file size limits be waived? if the os limits us, this is 369 * needed if we want to write a large archive 370 */ 371 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){ 372 reslimit.rlim_cur = reslimit.rlim_max; 373 (void)setrlimit(RLIMIT_FSIZE , &reslimit); 374 } 375 376 /* 377 * increase the size the stack can grow to 378 */ 379 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){ 380 reslimit.rlim_cur = reslimit.rlim_max; 381 (void)setrlimit(RLIMIT_STACK , &reslimit); 382 } 383 384 /* 385 * not really needed, but doesn't hurt 386 */ 387 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){ 388 reslimit.rlim_cur = reslimit.rlim_max; 389 (void)setrlimit(RLIMIT_RSS , &reslimit); 390 } 391 392 /* 393 * signal handling to reset stored directory times and modes. Since 394 * we deal with broken pipes via failed writes we ignore it. We also 395 * deal with any file size limit thorugh failed writes. Cpu time 396 * limits are caught and a cleanup is forced. 397 */ 398 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) || 399 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) || 400 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) || 401 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) { 402 paxwarn(1, "Unable to set up signal mask"); 403 return(-1); 404 } 405 memset(&n_hand, 0, sizeof n_hand); 406 n_hand.sa_mask = s_mask; 407 n_hand.sa_flags = 0; 408 n_hand.sa_handler = sig_cleanup; 409 410 if ((sigaction(SIGHUP, &n_hand, &o_hand) < 0) && 411 (o_hand.sa_handler == SIG_IGN) && 412 (sigaction(SIGHUP, &o_hand, &o_hand) < 0)) 413 goto out; 414 415 if ((sigaction(SIGTERM, &n_hand, &o_hand) < 0) && 416 (o_hand.sa_handler == SIG_IGN) && 417 (sigaction(SIGTERM, &o_hand, &o_hand) < 0)) 418 goto out; 419 420 if ((sigaction(SIGINT, &n_hand, &o_hand) < 0) && 421 (o_hand.sa_handler == SIG_IGN) && 422 (sigaction(SIGINT, &o_hand, &o_hand) < 0)) 423 goto out; 424 425 if ((sigaction(SIGQUIT, &n_hand, &o_hand) < 0) && 426 (o_hand.sa_handler == SIG_IGN) && 427 (sigaction(SIGQUIT, &o_hand, &o_hand) < 0)) 428 goto out; 429 430 if ((sigaction(SIGXCPU, &n_hand, &o_hand) < 0) && 431 (o_hand.sa_handler == SIG_IGN) && 432 (sigaction(SIGXCPU, &o_hand, &o_hand) < 0)) 433 goto out; 434 435 n_hand.sa_handler = SIG_IGN; 436 if ((sigaction(SIGPIPE, &n_hand, &o_hand) < 0) || 437 (sigaction(SIGXFSZ, &n_hand, &o_hand) < 0)) 438 goto out; 439 return(0); 440 441 out: 442 syswarn(1, errno, "Unable to set up signal handler"); 443 return(-1); 444 } 445