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