1 /* $OpenBSD: ftree.c,v 1.29 2009/10/27 23:59:22 deraadt Exp $ */ 2 /* $NetBSD: ftree.c,v 1.4 1995/03/21 09:07:21 cgd Exp $ */ 3 4 /*- 5 * Copyright (c) 1992 Keith Muller. 6 * Copyright (c) 1992, 1993 7 * The Regents of the University of California. All rights reserved. 8 * 9 * This code is derived from software contributed to Berkeley by 10 * Keith Muller of the University of California, San Diego. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 3. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 */ 36 37 #include <sys/types.h> 38 #include <sys/time.h> 39 #include <sys/stat.h> 40 #include <sys/param.h> 41 #include <unistd.h> 42 #include <string.h> 43 #include <stdio.h> 44 #include <errno.h> 45 #include <stdlib.h> 46 #include <fts.h> 47 #include "pax.h" 48 #include "ftree.h" 49 #include "extern.h" 50 51 /* 52 * routines to interface with the fts library function. 53 * 54 * file args supplied to pax are stored on a single linked list (of type FTREE) 55 * and given to fts to be processed one at a time. pax "selects" files from 56 * the expansion of each arg into the corresponding file tree (if the arg is a 57 * directory, otherwise the node itself is just passed to pax). The selection 58 * is modified by the -n and -u flags. The user is informed when a specific 59 * file arg does not generate any selected files. -n keeps expanding the file 60 * tree arg until one of its files is selected, then skips to the next file 61 * arg. when the user does not supply the file trees as command line args to 62 * pax, they are read from stdin 63 */ 64 65 static FTS *ftsp = NULL; /* current FTS handle */ 66 static int ftsopts; /* options to be used on fts_open */ 67 static char *farray[2]; /* array for passing each arg to fts */ 68 static FTREE *fthead = NULL; /* head of linked list of file args */ 69 static FTREE *fttail = NULL; /* tail of linked list of file args */ 70 static FTREE *ftcur = NULL; /* current file arg being processed */ 71 static FTSENT *ftent = NULL; /* current file tree entry */ 72 static int ftree_skip; /* when set skip to next file arg */ 73 74 static int ftree_arg(void); 75 static char *getpathname(char *, int); 76 77 /* 78 * ftree_start() 79 * initialize the options passed to fts_open() during this run of pax 80 * options are based on the selection of pax options by the user 81 * fts_start() also calls fts_arg() to open the first valid file arg. We 82 * also attempt to reset directory access times when -t (tflag) is set. 83 * Return: 84 * 0 if there is at least one valid file arg to process, -1 otherwise 85 */ 86 87 int 88 ftree_start(void) 89 { 90 /* 91 * set up the operation mode of fts, open the first file arg. We must 92 * use FTS_NOCHDIR, as the user may have to open multiple archives and 93 * if fts did a chdir off into the boondocks, we may create an archive 94 * volume in an place where the user did not expect to. 95 */ 96 ftsopts = FTS_NOCHDIR; 97 98 /* 99 * optional user flags that effect file traversal 100 * -H command line symlink follow only (half follow) 101 * -L follow sylinks (logical) 102 * -P do not follow sylinks (physical). This is the default. 103 * -X do not cross over mount points 104 * -t preserve access times on files read. 105 * -n select only the first member of a file tree when a match is found 106 * -d do not extract subtrees rooted at a directory arg. 107 */ 108 if (Lflag) 109 ftsopts |= FTS_LOGICAL; 110 else 111 ftsopts |= FTS_PHYSICAL; 112 if (Hflag) 113 ftsopts |= FTS_COMFOLLOW; 114 if (Xflag) 115 ftsopts |= FTS_XDEV; 116 117 if ((fthead == NULL) && ((farray[0] = malloc(PAXPATHLEN+2)) == NULL)) { 118 paxwarn(1, "Unable to allocate memory for file name buffer"); 119 return(-1); 120 } 121 122 if (ftree_arg() < 0) 123 return(-1); 124 if (tflag && (atdir_start() < 0)) 125 return(-1); 126 return(0); 127 } 128 129 /* 130 * ftree_add() 131 * add the arg to the linked list of files to process. Each will be 132 * processed by fts one at a time 133 * Return: 134 * 0 if added to the linked list, -1 if failed 135 */ 136 137 int 138 ftree_add(char *str, int chflg) 139 { 140 FTREE *ft; 141 int len; 142 143 /* 144 * simple check for bad args 145 */ 146 if ((str == NULL) || (*str == '\0')) { 147 paxwarn(0, "Invalid file name argument"); 148 return(-1); 149 } 150 151 /* 152 * allocate FTREE node and add to the end of the linked list (args are 153 * processed in the same order they were passed to pax). Get rid of any 154 * trailing / the user may pass us. (watch out for / by itself). 155 */ 156 if ((ft = (FTREE *)malloc(sizeof(FTREE))) == NULL) { 157 paxwarn(0, "Unable to allocate memory for filename"); 158 return(-1); 159 } 160 161 if (((len = strlen(str) - 1) > 0) && (str[len] == '/')) 162 str[len] = '\0'; 163 ft->fname = str; 164 ft->refcnt = 0; 165 ft->newercnt = 0; 166 ft->chflg = chflg; 167 ft->fow = NULL; 168 if (fthead == NULL) { 169 fttail = fthead = ft; 170 return(0); 171 } 172 fttail->fow = ft; 173 fttail = ft; 174 return(0); 175 } 176 177 /* 178 * ftree_sel() 179 * this entry has been selected by pax. bump up reference count and handle 180 * -n and -d processing. 181 */ 182 183 void 184 ftree_sel(ARCHD *arcn) 185 { 186 /* 187 * set reference bit for this pattern. This linked list is only used 188 * when file trees are supplied pax as args. The list is not used when 189 * the trees are read from stdin. 190 */ 191 if (ftcur != NULL) 192 ftcur->refcnt = 1; 193 194 /* 195 * if -n we are done with this arg, force a skip to the next arg when 196 * pax asks for the next file in next_file(). 197 * if -d we tell fts only to match the directory (if the arg is a dir) 198 * and not the entire file tree rooted at that point. 199 */ 200 if (nflag) 201 ftree_skip = 1; 202 203 if (!dflag || (arcn->type != PAX_DIR)) 204 return; 205 206 if (ftent != NULL) 207 (void)fts_set(ftsp, ftent, FTS_SKIP); 208 } 209 210 /* 211 * ftree_skipped_newer() 212 * file has been skipped because a newer file exists and -u/-D given 213 */ 214 215 void 216 ftree_skipped_newer(ARCHD *arcn) 217 { 218 /* skipped due to -u/-D, mark accordingly */ 219 if (ftcur != NULL) 220 ftcur->newercnt = 1; 221 } 222 223 /* 224 * ftree_chk() 225 * called at end on pax execution. Prints all those file args that did not 226 * have a selected member (reference count still 0) 227 */ 228 229 void 230 ftree_chk(void) 231 { 232 FTREE *ft; 233 int wban = 0; 234 235 /* 236 * make sure all dir access times were reset. 237 */ 238 if (tflag) 239 atdir_end(); 240 241 /* 242 * walk down list and check reference count. Print out those members 243 * that never had a match 244 */ 245 for (ft = fthead; ft != NULL; ft = ft->fow) { 246 if ((ft->refcnt > 0) || ft->newercnt > 0 || ft->chflg) 247 continue; 248 if (wban == 0) { 249 paxwarn(1,"WARNING! These file names were not selected:"); 250 ++wban; 251 } 252 (void)fprintf(stderr, "%s\n", ft->fname); 253 } 254 } 255 256 /* 257 * ftree_arg() 258 * Get the next file arg for fts to process. Can be from either the linked 259 * list or read from stdin when the user did not them as args to pax. Each 260 * arg is processed until the first successful fts_open(). 261 * Return: 262 * 0 when the next arg is ready to go, -1 if out of file args (or EOF on 263 * stdin). 264 */ 265 266 static int 267 ftree_arg(void) 268 { 269 270 /* 271 * close off the current file tree 272 */ 273 if (ftsp != NULL) { 274 (void)fts_close(ftsp); 275 ftsp = NULL; 276 } 277 278 /* 279 * keep looping until we get a valid file tree to process. Stop when we 280 * reach the end of the list (or get an eof on stdin) 281 */ 282 for (;;) { 283 if (fthead == NULL) { 284 /* 285 * the user didn't supply any args, get the file trees 286 * to process from stdin; 287 */ 288 if (getpathname(farray[0], PAXPATHLEN+1) == NULL) 289 return(-1); 290 } else { 291 /* 292 * the user supplied the file args as arguments to pax 293 */ 294 if (ftcur == NULL) 295 ftcur = fthead; 296 else if ((ftcur = ftcur->fow) == NULL) 297 return(-1); 298 if (ftcur->chflg) { 299 /* First fchdir() back... */ 300 if (fchdir(cwdfd) < 0) { 301 syswarn(1, errno, 302 "Can't fchdir to starting directory"); 303 return(-1); 304 } 305 if (chdir(ftcur->fname) < 0) { 306 syswarn(1, errno, "Can't chdir to %s", 307 ftcur->fname); 308 return(-1); 309 } 310 continue; 311 } else 312 farray[0] = ftcur->fname; 313 } 314 315 /* 316 * watch it, fts wants the file arg stored in a array of char 317 * ptrs, with the last one a null. we use a two element array 318 * and set farray[0] to point at the buffer with the file name 319 * in it. We cannot pass all the file args to fts at one shot 320 * as we need to keep a handle on which file arg generates what 321 * files (the -n and -d flags need this). If the open is 322 * successful, return a 0. 323 */ 324 if ((ftsp = fts_open(farray, ftsopts, NULL)) != NULL) 325 break; 326 } 327 return(0); 328 } 329 330 /* 331 * next_file() 332 * supplies the next file to process in the supplied archd structure. 333 * Return: 334 * 0 when contents of arcn have been set with the next file, -1 when done. 335 */ 336 337 int 338 next_file(ARCHD *arcn) 339 { 340 int cnt; 341 time_t atime; 342 time_t mtime; 343 344 /* 345 * ftree_sel() might have set the ftree_skip flag if the user has the 346 * -n option and a file was selected from this file arg tree. (-n says 347 * only one member is matched for each pattern) ftree_skip being 1 348 * forces us to go to the next arg now. 349 */ 350 if (ftree_skip) { 351 /* 352 * clear and go to next arg 353 */ 354 ftree_skip = 0; 355 if (ftree_arg() < 0) 356 return(-1); 357 } 358 359 /* 360 * loop until we get a valid file to process 361 */ 362 for (;;) { 363 if ((ftent = fts_read(ftsp)) == NULL) { 364 if (errno) 365 syswarn(1, errno, "next_file"); 366 /* 367 * out of files in this tree, go to next arg, if none 368 * we are done 369 */ 370 if (ftree_arg() < 0) 371 return(-1); 372 continue; 373 } 374 375 /* 376 * handle each type of fts_read() flag 377 */ 378 switch (ftent->fts_info) { 379 case FTS_D: 380 case FTS_DEFAULT: 381 case FTS_F: 382 case FTS_SL: 383 case FTS_SLNONE: 384 /* 385 * these are all ok 386 */ 387 break; 388 case FTS_DP: 389 /* 390 * already saw this directory. If the user wants file 391 * access times reset, we use this to restore the 392 * access time for this directory since this is the 393 * last time we will see it in this file subtree 394 * remember to force the time (this is -t on a read 395 * directory, not a created directory). 396 */ 397 if (!tflag || (get_atdir(ftent->fts_statp->st_dev, 398 ftent->fts_statp->st_ino, &mtime, &atime) < 0)) 399 continue; 400 set_ftime(ftent->fts_path, mtime, atime, 1); 401 continue; 402 case FTS_DC: 403 /* 404 * fts claims a file system cycle 405 */ 406 paxwarn(1,"File system cycle found at %s",ftent->fts_path); 407 continue; 408 case FTS_DNR: 409 syswarn(1, ftent->fts_errno, 410 "Unable to read directory %s", ftent->fts_path); 411 continue; 412 case FTS_ERR: 413 syswarn(1, ftent->fts_errno, 414 "File system traversal error"); 415 continue; 416 case FTS_NS: 417 case FTS_NSOK: 418 syswarn(1, ftent->fts_errno, 419 "Unable to access %s", ftent->fts_path); 420 continue; 421 } 422 423 /* 424 * ok got a file tree node to process. copy info into arcn 425 * structure (initialize as required) 426 */ 427 arcn->skip = 0; 428 arcn->pad = 0; 429 arcn->ln_nlen = 0; 430 arcn->ln_name[0] = '\0'; 431 memcpy(&arcn->sb, ftent->fts_statp, sizeof(arcn->sb)); 432 433 /* 434 * file type based set up and copy into the arcn struct 435 * SIDE NOTE: 436 * we try to reset the access time on all files and directories 437 * we may read when the -t flag is specified. files are reset 438 * when we close them after copying. we reset the directories 439 * when we are done with their file tree (we also clean up at 440 * end in case we cut short a file tree traversal). However 441 * there is no way to reset access times on symlinks. 442 */ 443 switch (S_IFMT & arcn->sb.st_mode) { 444 case S_IFDIR: 445 arcn->type = PAX_DIR; 446 if (!tflag) 447 break; 448 add_atdir(ftent->fts_path, arcn->sb.st_dev, 449 arcn->sb.st_ino, arcn->sb.st_mtime, 450 arcn->sb.st_atime); 451 break; 452 case S_IFCHR: 453 arcn->type = PAX_CHR; 454 break; 455 case S_IFBLK: 456 arcn->type = PAX_BLK; 457 break; 458 case S_IFREG: 459 /* 460 * only regular files with have data to store on the 461 * archive. all others will store a zero length skip. 462 * the skip field is used by pax for actual data it has 463 * to read (or skip over). 464 */ 465 arcn->type = PAX_REG; 466 arcn->skip = arcn->sb.st_size; 467 break; 468 case S_IFLNK: 469 arcn->type = PAX_SLK; 470 /* 471 * have to read the symlink path from the file 472 */ 473 if ((cnt = readlink(ftent->fts_path, arcn->ln_name, 474 PAXPATHLEN)) < 0) { 475 syswarn(1, errno, "Unable to read symlink %s", 476 ftent->fts_path); 477 continue; 478 } 479 /* 480 * set link name length, watch out readlink does not 481 * always NUL terminate the link path 482 */ 483 arcn->ln_name[cnt] = '\0'; 484 arcn->ln_nlen = cnt; 485 break; 486 case S_IFSOCK: 487 /* 488 * under BSD storing a socket is senseless but we will 489 * let the format specific write function make the 490 * decision of what to do with it. 491 */ 492 arcn->type = PAX_SCK; 493 break; 494 case S_IFIFO: 495 arcn->type = PAX_FIF; 496 break; 497 } 498 break; 499 } 500 501 /* 502 * copy file name, set file name length 503 */ 504 arcn->nlen = strlcpy(arcn->name, ftent->fts_path, sizeof(arcn->name)); 505 if (arcn->nlen >= sizeof(arcn->name)) 506 arcn->nlen = sizeof(arcn->name) - 1; /* XXX truncate? */ 507 arcn->org_name = ftent->fts_path; 508 return(0); 509 } 510 511 /* 512 * getpathname() 513 * Reads a pathname from stdin, handling NUL- or newline-termination. 514 * Return: 515 * NULL at end of file, otherwise the NUL-terminated buffer. 516 */ 517 518 static char * 519 getpathname(char *buf, int buflen) 520 { 521 char *bp, *ep; 522 int ch, term; 523 524 if (zeroflag) { 525 /* 526 * Read a NUL-terminated pathname, being especially 527 * paranoid about proper termination and pathname length. 528 */ 529 for (bp = buf, ep = buf + buflen; bp < ep; bp++) { 530 if ((ch = getchar()) == EOF) { 531 if (bp != buf) 532 paxwarn(1, "Ignoring unterminated " 533 "pathname at EOF"); 534 return(NULL); 535 } 536 if ((*bp = ch) == '\0') 537 return(buf); 538 } 539 /* Too long - skip this path */ 540 *--bp = '\0'; 541 term = '\0'; 542 } else { 543 if (fgets(buf, buflen, stdin) == NULL) 544 return(NULL); 545 if ((bp = strchr(buf, '\n')) != NULL || feof(stdin)) { 546 if (bp != NULL) 547 *bp = '\0'; 548 return(buf); 549 } 550 /* Too long - skip this path */ 551 term = '\n'; 552 } 553 while ((ch = getchar()) != term && ch != EOF) 554 ; 555 paxwarn(1, "Ignoring too-long pathname: %s", buf); 556 return(NULL); 557 } 558