1 /* $NetBSD: tables.c,v 1.7 1997/07/20 20:32:45 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 1992 Keith Muller. 5 * Copyright (c) 1992, 1993 6 * The Regents of the University of California. All rights reserved. 7 * 8 * This code is derived from software contributed to Berkeley by 9 * Keith Muller of the University of California, San Diego. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the University of 22 * California, Berkeley and its contributors. 23 * 4. Neither the name of the University nor the names of its contributors 24 * may be used to endorse or promote products derived from this software 25 * without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 37 * SUCH DAMAGE. 38 */ 39 40 #include <sys/cdefs.h> 41 #ifndef lint 42 #if 0 43 static char sccsid[] = "@(#)tables.c 8.1 (Berkeley) 5/31/93"; 44 #else 45 __RCSID("$NetBSD: tables.c,v 1.7 1997/07/20 20:32:45 christos Exp $"); 46 #endif 47 #endif /* not lint */ 48 49 #include <sys/types.h> 50 #include <sys/time.h> 51 #include <sys/stat.h> 52 #include <sys/param.h> 53 #include <sys/fcntl.h> 54 #include <stdio.h> 55 #include <ctype.h> 56 #include <string.h> 57 #include <unistd.h> 58 #include <errno.h> 59 #include <stdlib.h> 60 #include "pax.h" 61 #include "tables.h" 62 #include "extern.h" 63 64 /* 65 * Routines for controlling the contents of all the different databases pax 66 * keeps. Tables are dynamically created only when they are needed. The 67 * goal was speed and the ability to work with HUGE archives. The databases 68 * were kept simple, but do have complex rules for when the contents change. 69 * As of this writing, the posix library functions were more complex than 70 * needed for this application (pax databases have very short lifetimes and 71 * do not survive after pax is finished). Pax is required to handle very 72 * large archives. These database routines carefully combine memory usage and 73 * temporary file storage in ways which will not significantly impact runtime 74 * performance while allowing the largest possible archives to be handled. 75 * Trying to force the fit to the posix databases routines was not considered 76 * time well spent. 77 */ 78 79 static HRDLNK **ltab = NULL; /* hard link table for detecting hard links */ 80 static FTM **ftab = NULL; /* file time table for updating arch */ 81 static NAMT **ntab = NULL; /* interactive rename storage table */ 82 static DEVT **dtab = NULL; /* device/inode mapping tables */ 83 static ATDIR **atab = NULL; /* file tree directory time reset table */ 84 static int dirfd = -1; /* storage for setting created dir time/mode */ 85 static u_long dircnt; /* entries in dir time/mode storage */ 86 static int ffd = -1; /* tmp file for file time table name storage */ 87 88 static DEVT *chk_dev __P((dev_t, int)); 89 90 /* 91 * hard link table routines 92 * 93 * The hard link table tries to detect hard links to files using the device and 94 * inode values. We do this when writing an archive, so we can tell the format 95 * write routine that this file is a hard link to another file. The format 96 * write routine then can store this file in whatever way it wants (as a hard 97 * link if the format supports that like tar, or ignore this info like cpio). 98 * (Actually a field in the format driver table tells us if the format wants 99 * hard link info. if not, we do not waste time looking for them). We also use 100 * the same table when reading an archive. In that situation, this table is 101 * used by the format read routine to detect hard links from stored dev and 102 * inode numbers (like cpio). This will allow pax to create a link when one 103 * can be detected by the archive format. 104 */ 105 106 /* 107 * lnk_start 108 * Creates the hard link table. 109 * Return: 110 * 0 if created, -1 if failure 111 */ 112 113 #if __STDC__ 114 int 115 lnk_start(void) 116 #else 117 int 118 lnk_start() 119 #endif 120 { 121 if (ltab != NULL) 122 return(0); 123 if ((ltab = (HRDLNK **)calloc(L_TAB_SZ, sizeof(HRDLNK *))) == NULL) { 124 tty_warn(1, "Cannot allocate memory for hard link table"); 125 return(-1); 126 } 127 return(0); 128 } 129 130 /* 131 * chk_lnk() 132 * Looks up entry in hard link hash table. If found, it copies the name 133 * of the file it is linked to (we already saw that file) into ln_name. 134 * lnkcnt is decremented and if goes to 1 the node is deleted from the 135 * database. (We have seen all the links to this file). If not found, 136 * we add the file to the database if it has the potential for having 137 * hard links to other files we may process (it has a link count > 1) 138 * Return: 139 * if found returns 1; if not found returns 0; -1 on error 140 */ 141 142 #if __STDC__ 143 int 144 chk_lnk(ARCHD *arcn) 145 #else 146 int 147 chk_lnk(arcn) 148 ARCHD *arcn; 149 #endif 150 { 151 HRDLNK *pt; 152 HRDLNK **ppt; 153 u_int indx; 154 155 if (ltab == NULL) 156 return(-1); 157 /* 158 * ignore those nodes that cannot have hard links 159 */ 160 if ((arcn->type == PAX_DIR) || (arcn->sb.st_nlink <= 1)) 161 return(0); 162 163 /* 164 * hash inode number and look for this file 165 */ 166 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ; 167 if ((pt = ltab[indx]) != NULL) { 168 /* 169 * it's hash chain in not empty, walk down looking for it 170 */ 171 ppt = &(ltab[indx]); 172 while (pt != NULL) { 173 if ((pt->ino == arcn->sb.st_ino) && 174 (pt->dev == arcn->sb.st_dev)) 175 break; 176 ppt = &(pt->fow); 177 pt = pt->fow; 178 } 179 180 if (pt != NULL) { 181 /* 182 * found a link. set the node type and copy in the 183 * name of the file it is to link to. we need to 184 * handle hardlinks to regular files differently than 185 * other links. 186 */ 187 arcn->ln_nlen = l_strncpy(arcn->ln_name, pt->name, 188 PAXPATHLEN+1); 189 if (arcn->type == PAX_REG) 190 arcn->type = PAX_HRG; 191 else 192 arcn->type = PAX_HLK; 193 194 /* 195 * if we have found all the links to this file, remove 196 * it from the database 197 */ 198 if (--pt->nlink <= 1) { 199 *ppt = pt->fow; 200 (void)free((char *)pt->name); 201 (void)free((char *)pt); 202 } 203 return(1); 204 } 205 } 206 207 /* 208 * we never saw this file before. It has links so we add it to the 209 * front of this hash chain 210 */ 211 if ((pt = (HRDLNK *)malloc(sizeof(HRDLNK))) != NULL) { 212 if ((pt->name = strdup(arcn->name)) != NULL) { 213 pt->dev = arcn->sb.st_dev; 214 pt->ino = arcn->sb.st_ino; 215 pt->nlink = arcn->sb.st_nlink; 216 pt->fow = ltab[indx]; 217 ltab[indx] = pt; 218 return(0); 219 } 220 (void)free((char *)pt); 221 } 222 223 tty_warn(1, "Hard link table out of memory"); 224 return(-1); 225 } 226 227 /* 228 * purg_lnk 229 * remove reference for a file that we may have added to the data base as 230 * a potential source for hard links. We ended up not using the file, so 231 * we do not want to accidently point another file at it later on. 232 */ 233 234 #if __STDC__ 235 void 236 purg_lnk(ARCHD *arcn) 237 #else 238 void 239 purg_lnk(arcn) 240 ARCHD *arcn; 241 #endif 242 { 243 HRDLNK *pt; 244 HRDLNK **ppt; 245 u_int indx; 246 247 if (ltab == NULL) 248 return; 249 /* 250 * do not bother to look if it could not be in the database 251 */ 252 if ((arcn->sb.st_nlink <= 1) || (arcn->type == PAX_DIR) || 253 (arcn->type == PAX_HLK) || (arcn->type == PAX_HRG)) 254 return; 255 256 /* 257 * find the hash chain for this inode value, if empty return 258 */ 259 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ; 260 if ((pt = ltab[indx]) == NULL) 261 return; 262 263 /* 264 * walk down the list looking for the inode/dev pair, unlink and 265 * free if found 266 */ 267 ppt = &(ltab[indx]); 268 while (pt != NULL) { 269 if ((pt->ino == arcn->sb.st_ino) && 270 (pt->dev == arcn->sb.st_dev)) 271 break; 272 ppt = &(pt->fow); 273 pt = pt->fow; 274 } 275 if (pt == NULL) 276 return; 277 278 /* 279 * remove and free it 280 */ 281 *ppt = pt->fow; 282 (void)free((char *)pt->name); 283 (void)free((char *)pt); 284 } 285 286 /* 287 * lnk_end() 288 * pull apart a existing link table so we can reuse it. We do this between 289 * read and write phases of append with update. (The format may have 290 * used the link table, and we need to start with a fresh table for the 291 * write phase 292 */ 293 294 #if __STDC__ 295 void 296 lnk_end(void) 297 #else 298 void 299 lnk_end() 300 #endif 301 { 302 int i; 303 HRDLNK *pt; 304 HRDLNK *ppt; 305 306 if (ltab == NULL) 307 return; 308 309 for (i = 0; i < L_TAB_SZ; ++i) { 310 if (ltab[i] == NULL) 311 continue; 312 pt = ltab[i]; 313 ltab[i] = NULL; 314 315 /* 316 * free up each entry on this chain 317 */ 318 while (pt != NULL) { 319 ppt = pt; 320 pt = ppt->fow; 321 (void)free((char *)ppt->name); 322 (void)free((char *)ppt); 323 } 324 } 325 return; 326 } 327 328 /* 329 * modification time table routines 330 * 331 * The modification time table keeps track of last modification times for all 332 * files stored in an archive during a write phase when -u is set. We only 333 * add a file to the archive if it is newer than a file with the same name 334 * already stored on the archive (if there is no other file with the same 335 * name on the archive it is added). This applies to writes and appends. 336 * An append with an -u must read the archive and store the modification time 337 * for every file on that archive before starting the write phase. It is clear 338 * that this is one HUGE database. To save memory space, the actual file names 339 * are stored in a scatch file and indexed by an in memory hash table. The 340 * hash table is indexed by hashing the file path. The nodes in the table store 341 * the length of the filename and the lseek offset within the scratch file 342 * where the actual name is stored. Since there are never any deletions to this 343 * table, fragmentation of the scratch file is never a issue. Lookups seem to 344 * not exhibit any locality at all (files in the database are rarely 345 * looked up more than once...). So caching is just a waste of memory. The 346 * only limitation is the amount of scatch file space available to store the 347 * path names. 348 */ 349 350 /* 351 * ftime_start() 352 * create the file time hash table and open for read/write the scratch 353 * file. (after created it is unlinked, so when we exit we leave 354 * no witnesses). 355 * Return: 356 * 0 if the table and file was created ok, -1 otherwise 357 */ 358 359 #if __STDC__ 360 int 361 ftime_start(void) 362 #else 363 int 364 ftime_start() 365 #endif 366 { 367 char *pt; 368 369 if (ftab != NULL) 370 return(0); 371 if ((ftab = (FTM **)calloc(F_TAB_SZ, sizeof(FTM *))) == NULL) { 372 tty_warn(1, "Cannot allocate memory for file time table"); 373 return(-1); 374 } 375 376 /* 377 * get random name and create temporary scratch file, unlink name 378 * so it will get removed on exit 379 */ 380 pt = strdup("/tmp/paxXXXXXX"); 381 if (pt == NULL) { 382 tty_warn(1, "Unable to allocate memory"); 383 return(-1); 384 } 385 if ((ffd = mkstemp(pt)) == -1) { 386 syswarn(1, errno, "Unable to create temporary file: %s", pt); 387 free(pt); 388 return(-1); 389 } 390 391 (void)unlink(pt); 392 free(pt); 393 return(0); 394 } 395 396 /* 397 * chk_ftime() 398 * looks up entry in file time hash table. If not found, the file is 399 * added to the hash table and the file named stored in the scratch file. 400 * If a file with the same name is found, the file times are compared and 401 * the most recent file time is retained. If the new file was younger (or 402 * was not in the database) the new file is selected for storage. 403 * Return: 404 * 0 if file should be added to the archive, 1 if it should be skipped, 405 * -1 on error 406 */ 407 408 #if __STDC__ 409 int 410 chk_ftime(ARCHD *arcn) 411 #else 412 int 413 chk_ftime(arcn) 414 ARCHD *arcn; 415 #endif 416 { 417 FTM *pt; 418 int namelen; 419 u_int indx; 420 char ckname[PAXPATHLEN+1]; 421 422 /* 423 * no info, go ahead and add to archive 424 */ 425 if (ftab == NULL) 426 return(0); 427 428 /* 429 * hash the pathname and look up in table 430 */ 431 namelen = arcn->nlen; 432 indx = st_hash(arcn->name, namelen, F_TAB_SZ); 433 if ((pt = ftab[indx]) != NULL) { 434 /* 435 * the hash chain is not empty, walk down looking for match 436 * only read up the path names if the lengths match, speeds 437 * up the search a lot 438 */ 439 while (pt != NULL) { 440 if (pt->namelen == namelen) { 441 /* 442 * potential match, have to read the name 443 * from the scratch file. 444 */ 445 if (lseek(ffd,pt->seek,SEEK_SET) != pt->seek) { 446 syswarn(1, errno, 447 "Failed ftime table seek"); 448 return(-1); 449 } 450 if (read(ffd, ckname, namelen) != namelen) { 451 syswarn(1, errno, 452 "Failed ftime table read"); 453 return(-1); 454 } 455 456 /* 457 * if the names match, we are done 458 */ 459 if (!strncmp(ckname, arcn->name, namelen)) 460 break; 461 } 462 463 /* 464 * try the next entry on the chain 465 */ 466 pt = pt->fow; 467 } 468 469 if (pt != NULL) { 470 /* 471 * found the file, compare the times, save the newer 472 */ 473 if (arcn->sb.st_mtime > pt->mtime) { 474 /* 475 * file is newer 476 */ 477 pt->mtime = arcn->sb.st_mtime; 478 return(0); 479 } 480 /* 481 * file is older 482 */ 483 return(1); 484 } 485 } 486 487 /* 488 * not in table, add it 489 */ 490 if ((pt = (FTM *)malloc(sizeof(FTM))) != NULL) { 491 /* 492 * add the name at the end of the scratch file, saving the 493 * offset. add the file to the head of the hash chain 494 */ 495 if ((pt->seek = lseek(ffd, (off_t)0, SEEK_END)) >= 0) { 496 if (write(ffd, arcn->name, namelen) == namelen) { 497 pt->mtime = arcn->sb.st_mtime; 498 pt->namelen = namelen; 499 pt->fow = ftab[indx]; 500 ftab[indx] = pt; 501 return(0); 502 } 503 syswarn(1, errno, "Failed write to file time table"); 504 } else 505 syswarn(1, errno, "Failed seek on file time table"); 506 } else 507 tty_warn(1, "File time table ran out of memory"); 508 509 if (pt != NULL) 510 (void)free((char *)pt); 511 return(-1); 512 } 513 514 /* 515 * Interactive rename table routines 516 * 517 * The interactive rename table keeps track of the new names that the user 518 * assignes to files from tty input. Since this map is unique for each file 519 * we must store it in case there is a reference to the file later in archive 520 * (a link). Otherwise we will be unable to find the file we know was 521 * extracted. The remapping of these files is stored in a memory based hash 522 * table (it is assumed since input must come from /dev/tty, it is unlikely to 523 * be a very large table). 524 */ 525 526 /* 527 * name_start() 528 * create the interactive rename table 529 * Return: 530 * 0 if successful, -1 otherwise 531 */ 532 533 #if __STDC__ 534 int 535 name_start(void) 536 #else 537 int 538 name_start() 539 #endif 540 { 541 if (ntab != NULL) 542 return(0); 543 if ((ntab = (NAMT **)calloc(N_TAB_SZ, sizeof(NAMT *))) == NULL) { 544 tty_warn(1, 545 "Cannot allocate memory for interactive rename table"); 546 return(-1); 547 } 548 return(0); 549 } 550 551 /* 552 * add_name() 553 * add the new name to old name mapping just created by the user. 554 * If an old name mapping is found (there may be duplicate names on an 555 * archive) only the most recent is kept. 556 * Return: 557 * 0 if added, -1 otherwise 558 */ 559 560 #if __STDC__ 561 int 562 add_name(char *oname, int onamelen, char *nname) 563 #else 564 int 565 add_name(oname, onamelen, nname) 566 char *oname; 567 int onamelen; 568 char *nname; 569 #endif 570 { 571 NAMT *pt; 572 u_int indx; 573 574 if (ntab == NULL) { 575 /* 576 * should never happen 577 */ 578 tty_warn(0, "No interactive rename table, links may fail\n"); 579 return(0); 580 } 581 582 /* 583 * look to see if we have already mapped this file, if so we 584 * will update it 585 */ 586 indx = st_hash(oname, onamelen, N_TAB_SZ); 587 if ((pt = ntab[indx]) != NULL) { 588 /* 589 * look down the has chain for the file 590 */ 591 while ((pt != NULL) && (strcmp(oname, pt->oname) != 0)) 592 pt = pt->fow; 593 594 if (pt != NULL) { 595 /* 596 * found an old mapping, replace it with the new one 597 * the user just input (if it is different) 598 */ 599 if (strcmp(nname, pt->nname) == 0) 600 return(0); 601 602 (void)free((char *)pt->nname); 603 if ((pt->nname = strdup(nname)) == NULL) { 604 tty_warn(1, "Cannot update rename table"); 605 return(-1); 606 } 607 return(0); 608 } 609 } 610 611 /* 612 * this is a new mapping, add it to the table 613 */ 614 if ((pt = (NAMT *)malloc(sizeof(NAMT))) != NULL) { 615 if ((pt->oname = strdup(oname)) != NULL) { 616 if ((pt->nname = strdup(nname)) != NULL) { 617 pt->fow = ntab[indx]; 618 ntab[indx] = pt; 619 return(0); 620 } 621 (void)free((char *)pt->oname); 622 } 623 (void)free((char *)pt); 624 } 625 tty_warn(1, "Interactive rename table out of memory"); 626 return(-1); 627 } 628 629 /* 630 * sub_name() 631 * look up a link name to see if it points at a file that has been 632 * remapped by the user. If found, the link is adjusted to contain the 633 * new name (oname is the link to name) 634 */ 635 636 #if __STDC__ 637 void 638 sub_name(char *oname, int *onamelen) 639 #else 640 void 641 sub_name(oname, onamelen) 642 char *oname; 643 int *onamelen; 644 #endif 645 { 646 NAMT *pt; 647 u_int indx; 648 649 if (ntab == NULL) 650 return; 651 /* 652 * look the name up in the hash table 653 */ 654 indx = st_hash(oname, *onamelen, N_TAB_SZ); 655 if ((pt = ntab[indx]) == NULL) 656 return; 657 658 while (pt != NULL) { 659 /* 660 * walk down the hash cahin looking for a match 661 */ 662 if (strcmp(oname, pt->oname) == 0) { 663 /* 664 * found it, replace it with the new name 665 * and return (we know that oname has enough space) 666 */ 667 *onamelen = l_strncpy(oname, pt->nname, PAXPATHLEN+1); 668 return; 669 } 670 pt = pt->fow; 671 } 672 673 /* 674 * no match, just return 675 */ 676 return; 677 } 678 679 /* 680 * device/inode mapping table routines 681 * (used with formats that store device and inodes fields) 682 * 683 * device/inode mapping tables remap the device field in a archive header. The 684 * device/inode fields are used to determine when files are hard links to each 685 * other. However these values have very little meaning outside of that. This 686 * database is used to solve one of two different problems. 687 * 688 * 1) when files are appended to an archive, while the new files may have hard 689 * links to each other, you cannot determine if they have hard links to any 690 * file already stored on the archive from a prior run of pax. We must assume 691 * that these inode/device pairs are unique only within a SINGLE run of pax 692 * (which adds a set of files to an archive). So we have to make sure the 693 * inode/dev pairs we add each time are always unique. We do this by observing 694 * while the inode field is very dense, the use of the dev field is fairly 695 * sparse. Within each run of pax, we remap any device number of a new archive 696 * member that has a device number used in a prior run and already stored in a 697 * file on the archive. During the read phase of the append, we store the 698 * device numbers used and mark them to not be used by any file during the 699 * write phase. If during write we go to use one of those old device numbers, 700 * we remap it to a new value. 701 * 702 * 2) Often the fields in the archive header used to store these values are 703 * too small to store the entire value. The result is an inode or device value 704 * which can be truncated. This really can foul up an archive. With truncation 705 * we end up creating links between files that are really not links (after 706 * truncation the inodes are the same value). We address that by detecting 707 * truncation and forcing a remap of the device field to split truncated 708 * inodes away from each other. Each truncation creates a pattern of bits that 709 * are removed. We use this pattern of truncated bits to partition the inodes 710 * on a single device to many different devices (each one represented by the 711 * truncated bit pattern). All inodes on the same device that have the same 712 * truncation pattern are mapped to the same new device. Two inodes that 713 * truncate to the same value clearly will always have different truncation 714 * bit patterns, so they will be split from away each other. When we spot 715 * device truncation we remap the device number to a non truncated value. 716 * (for more info see table.h for the data structures involved). 717 */ 718 719 /* 720 * dev_start() 721 * create the device mapping table 722 * Return: 723 * 0 if successful, -1 otherwise 724 */ 725 726 #if __STDC__ 727 int 728 dev_start(void) 729 #else 730 int 731 dev_start() 732 #endif 733 { 734 if (dtab != NULL) 735 return(0); 736 if ((dtab = (DEVT **)calloc(D_TAB_SZ, sizeof(DEVT *))) == NULL) { 737 tty_warn(1, "Cannot allocate memory for device mapping table"); 738 return(-1); 739 } 740 return(0); 741 } 742 743 /* 744 * add_dev() 745 * add a device number to the table. this will force the device to be 746 * remapped to a new value if it be used during a write phase. This 747 * function is called during the read phase of an append to prohibit the 748 * use of any device number already in the archive. 749 * Return: 750 * 0 if added ok, -1 otherwise 751 */ 752 753 #if __STDC__ 754 int 755 add_dev(ARCHD *arcn) 756 #else 757 int 758 add_dev(arcn) 759 ARCHD *arcn; 760 #endif 761 { 762 if (chk_dev(arcn->sb.st_dev, 1) == NULL) 763 return(-1); 764 return(0); 765 } 766 767 /* 768 * chk_dev() 769 * check for a device value in the device table. If not found and the add 770 * flag is set, it is added. This does NOT assign any mapping values, just 771 * adds the device number as one that need to be remapped. If this device 772 * is alread mapped, just return with a pointer to that entry. 773 * Return: 774 * pointer to the entry for this device in the device map table. Null 775 * if the add flag is not set and the device is not in the table (it is 776 * not been seen yet). If add is set and the device cannot be added, null 777 * is returned (indicates an error). 778 */ 779 780 #if __STDC__ 781 static DEVT * 782 chk_dev(dev_t dev, int add) 783 #else 784 static DEVT * 785 chk_dev(dev, add) 786 dev_t dev; 787 int add; 788 #endif 789 { 790 DEVT *pt; 791 u_int indx; 792 793 if (dtab == NULL) 794 return(NULL); 795 /* 796 * look to see if this device is already in the table 797 */ 798 indx = ((unsigned)dev) % D_TAB_SZ; 799 if ((pt = dtab[indx]) != NULL) { 800 while ((pt != NULL) && (pt->dev != dev)) 801 pt = pt->fow; 802 803 /* 804 * found it, return a pointer to it 805 */ 806 if (pt != NULL) 807 return(pt); 808 } 809 810 /* 811 * not in table, we add it only if told to as this may just be a check 812 * to see if a device number is being used. 813 */ 814 if (add == 0) 815 return(NULL); 816 817 /* 818 * allocate a node for this device and add it to the front of the hash 819 * chain. Note we do not assign remaps values here, so the pt->list 820 * list must be NULL. 821 */ 822 if ((pt = (DEVT *)malloc(sizeof(DEVT))) == NULL) { 823 tty_warn(1, "Device map table out of memory"); 824 return(NULL); 825 } 826 pt->dev = dev; 827 pt->list = NULL; 828 pt->fow = dtab[indx]; 829 dtab[indx] = pt; 830 return(pt); 831 } 832 /* 833 * map_dev() 834 * given an inode and device storage mask (the mask has a 1 for each bit 835 * the archive format is able to store in a header), we check for inode 836 * and device truncation and remap the device as required. Device mapping 837 * can also occur when during the read phase of append a device number was 838 * seen (and was marked as do not use during the write phase). WE ASSUME 839 * that unsigned longs are the same size or bigger than the fields used 840 * for ino_t and dev_t. If not the types will have to be changed. 841 * Return: 842 * 0 if all ok, -1 otherwise. 843 */ 844 845 #if __STDC__ 846 int 847 map_dev(ARCHD *arcn, u_long dev_mask, u_long ino_mask) 848 #else 849 int 850 map_dev(arcn, dev_mask, ino_mask) 851 ARCHD *arcn; 852 u_long dev_mask; 853 u_long ino_mask; 854 #endif 855 { 856 DEVT *pt; 857 DLIST *dpt; 858 static dev_t lastdev = 0; /* next device number to try */ 859 int trc_ino = 0; 860 int trc_dev = 0; 861 ino_t trunc_bits = 0; 862 ino_t nino; 863 864 if (dtab == NULL) 865 return(0); 866 /* 867 * check for device and inode truncation, and extract the truncated 868 * bit pattern. 869 */ 870 if ((arcn->sb.st_dev & (dev_t)dev_mask) != arcn->sb.st_dev) 871 ++trc_dev; 872 if ((nino = arcn->sb.st_ino & (ino_t)ino_mask) != arcn->sb.st_ino) { 873 ++trc_ino; 874 trunc_bits = arcn->sb.st_ino & (ino_t)(~ino_mask); 875 } 876 877 /* 878 * see if this device is already being mapped, look up the device 879 * then find the truncation bit pattern which applies 880 */ 881 if ((pt = chk_dev(arcn->sb.st_dev, 0)) != NULL) { 882 /* 883 * this device is already marked to be remapped 884 */ 885 for (dpt = pt->list; dpt != NULL; dpt = dpt->fow) 886 if (dpt->trunc_bits == trunc_bits) 887 break; 888 889 if (dpt != NULL) { 890 /* 891 * we are being remapped for this device and pattern 892 * change the device number to be stored and return 893 */ 894 arcn->sb.st_dev = dpt->dev; 895 arcn->sb.st_ino = nino; 896 return(0); 897 } 898 } else { 899 /* 900 * this device is not being remapped YET. if we do not have any 901 * form of truncation, we do not need a remap 902 */ 903 if (!trc_ino && !trc_dev) 904 return(0); 905 906 /* 907 * we have truncation, have to add this as a device to remap 908 */ 909 if ((pt = chk_dev(arcn->sb.st_dev, 1)) == NULL) 910 goto bad; 911 912 /* 913 * if we just have a truncated inode, we have to make sure that 914 * all future inodes that do not truncate (they have the 915 * truncation pattern of all 0's) continue to map to the same 916 * device number. We probably have already written inodes with 917 * this device number to the archive with the truncation 918 * pattern of all 0's. So we add the mapping for all 0's to the 919 * same device number. 920 */ 921 if (!trc_dev && (trunc_bits != 0)) { 922 if ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL) 923 goto bad; 924 dpt->trunc_bits = 0; 925 dpt->dev = arcn->sb.st_dev; 926 dpt->fow = pt->list; 927 pt->list = dpt; 928 } 929 } 930 931 /* 932 * look for a device number not being used. We must watch for wrap 933 * around on lastdev (so we do not get stuck looking forever!) 934 */ 935 while (++lastdev > 0) { 936 if (chk_dev(lastdev, 0) != NULL) 937 continue; 938 /* 939 * found an unused value. If we have reached truncation point 940 * for this format we are hosed, so we give up. Otherwise we 941 * mark it as being used. 942 */ 943 if (((lastdev & ((dev_t)dev_mask)) != lastdev) || 944 (chk_dev(lastdev, 1) == NULL)) 945 goto bad; 946 break; 947 } 948 949 if ((lastdev <= 0) || ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL)) 950 goto bad; 951 952 /* 953 * got a new device number, store it under this truncation pattern. 954 * change the device number this file is being stored with. 955 */ 956 dpt->trunc_bits = trunc_bits; 957 dpt->dev = lastdev; 958 dpt->fow = pt->list; 959 pt->list = dpt; 960 arcn->sb.st_dev = lastdev; 961 arcn->sb.st_ino = nino; 962 return(0); 963 964 bad: 965 tty_warn(1, 966 "Unable to fix truncated inode/device field when storing %s", 967 arcn->name); 968 tty_warn(0, "Archive may create improper hard links when extracted"); 969 return(0); 970 } 971 972 /* 973 * directory access/mod time reset table routines (for directories READ by pax) 974 * 975 * The pax -t flag requires that access times of archive files to be the same 976 * before being read by pax. For regular files, access time is restored after 977 * the file has been copied. This database provides the same functionality for 978 * directories read during file tree traversal. Restoring directory access time 979 * is more complex than files since directories may be read several times until 980 * all the descendants in their subtree are visited by fts. Directory access 981 * and modification times are stored during the fts pre-order visit (done 982 * before any descendants in the subtree is visited) and restored after the 983 * fts post-order visit (after all the descendants have been visited). In the 984 * case of premature exit from a subtree (like from the effects of -n), any 985 * directory entries left in this database are reset during final cleanup 986 * operations of pax. Entries are hashed by inode number for fast lookup. 987 */ 988 989 /* 990 * atdir_start() 991 * create the directory access time database for directories READ by pax. 992 * Return: 993 * 0 is created ok, -1 otherwise. 994 */ 995 996 #if __STDC__ 997 int 998 atdir_start(void) 999 #else 1000 int 1001 atdir_start() 1002 #endif 1003 { 1004 if (atab != NULL) 1005 return(0); 1006 if ((atab = (ATDIR **)calloc(A_TAB_SZ, sizeof(ATDIR *))) == NULL) { 1007 tty_warn(1, 1008 "Cannot allocate space for directory access time table"); 1009 return(-1); 1010 } 1011 return(0); 1012 } 1013 1014 1015 /* 1016 * atdir_end() 1017 * walk through the directory access time table and reset the access time 1018 * of any directory who still has an entry left in the database. These 1019 * entries are for directories READ by pax 1020 */ 1021 1022 #if __STDC__ 1023 void 1024 atdir_end(void) 1025 #else 1026 void 1027 atdir_end() 1028 #endif 1029 { 1030 ATDIR *pt; 1031 int i; 1032 1033 if (atab == NULL) 1034 return; 1035 /* 1036 * for each non-empty hash table entry reset all the directories 1037 * chained there. 1038 */ 1039 for (i = 0; i < A_TAB_SZ; ++i) { 1040 if ((pt = atab[i]) == NULL) 1041 continue; 1042 /* 1043 * remember to force the times, set_ftime() looks at pmtime 1044 * and patime, which only applies to things CREATED by pax, 1045 * not read by pax. Read time reset is controlled by -t. 1046 */ 1047 for (; pt != NULL; pt = pt->fow) 1048 set_ftime(pt->name, pt->mtime, pt->atime, 1); 1049 } 1050 } 1051 1052 /* 1053 * add_atdir() 1054 * add a directory to the directory access time table. Table is hashed 1055 * and chained by inode number. This is for directories READ by pax 1056 */ 1057 1058 #if __STDC__ 1059 void 1060 add_atdir(char *fname, dev_t dev, ino_t ino, time_t mtime, time_t atime) 1061 #else 1062 void 1063 add_atdir(fname, dev, ino, mtime, atime) 1064 char *fname; 1065 dev_t dev; 1066 ino_t ino; 1067 time_t mtime; 1068 time_t atime; 1069 #endif 1070 { 1071 ATDIR *pt; 1072 u_int indx; 1073 1074 if (atab == NULL) 1075 return; 1076 1077 /* 1078 * make sure this directory is not already in the table, if so just 1079 * return (the older entry always has the correct time). The only 1080 * way this will happen is when the same subtree can be traversed by 1081 * different args to pax and the -n option is aborting fts out of a 1082 * subtree before all the post-order visits have been made). 1083 */ 1084 indx = ((unsigned)ino) % A_TAB_SZ; 1085 if ((pt = atab[indx]) != NULL) { 1086 while (pt != NULL) { 1087 if ((pt->ino == ino) && (pt->dev == dev)) 1088 break; 1089 pt = pt->fow; 1090 } 1091 1092 /* 1093 * oops, already there. Leave it alone. 1094 */ 1095 if (pt != NULL) 1096 return; 1097 } 1098 1099 /* 1100 * add it to the front of the hash chain 1101 */ 1102 if ((pt = (ATDIR *)malloc(sizeof(ATDIR))) != NULL) { 1103 if ((pt->name = strdup(fname)) != NULL) { 1104 pt->dev = dev; 1105 pt->ino = ino; 1106 pt->mtime = mtime; 1107 pt->atime = atime; 1108 pt->fow = atab[indx]; 1109 atab[indx] = pt; 1110 return; 1111 } 1112 (void)free((char *)pt); 1113 } 1114 1115 tty_warn(1, "Directory access time reset table ran out of memory"); 1116 return; 1117 } 1118 1119 /* 1120 * get_atdir() 1121 * look up a directory by inode and device number to obtain the access 1122 * and modification time you want to set to. If found, the modification 1123 * and access time parameters are set and the entry is removed from the 1124 * table (as it is no longer needed). These are for directories READ by 1125 * pax 1126 * Return: 1127 * 0 if found, -1 if not found. 1128 */ 1129 1130 #if __STDC__ 1131 int 1132 get_atdir(dev_t dev, ino_t ino, time_t *mtime, time_t *atime) 1133 #else 1134 int 1135 get_atdir(dev, ino, mtime, atime) 1136 dev_t dev; 1137 ino_t ino; 1138 time_t *mtime; 1139 time_t *atime; 1140 #endif 1141 { 1142 ATDIR *pt; 1143 ATDIR **ppt; 1144 u_int indx; 1145 1146 if (atab == NULL) 1147 return(-1); 1148 /* 1149 * hash by inode and search the chain for an inode and device match 1150 */ 1151 indx = ((unsigned)ino) % A_TAB_SZ; 1152 if ((pt = atab[indx]) == NULL) 1153 return(-1); 1154 1155 ppt = &(atab[indx]); 1156 while (pt != NULL) { 1157 if ((pt->ino == ino) && (pt->dev == dev)) 1158 break; 1159 /* 1160 * no match, go to next one 1161 */ 1162 ppt = &(pt->fow); 1163 pt = pt->fow; 1164 } 1165 1166 /* 1167 * return if we did not find it. 1168 */ 1169 if (pt == NULL) 1170 return(-1); 1171 1172 /* 1173 * found it. return the times and remove the entry from the table. 1174 */ 1175 *ppt = pt->fow; 1176 *mtime = pt->mtime; 1177 *atime = pt->atime; 1178 (void)free((char *)pt->name); 1179 (void)free((char *)pt); 1180 return(0); 1181 } 1182 1183 /* 1184 * directory access mode and time storage routines (for directories CREATED 1185 * by pax). 1186 * 1187 * Pax requires that extracted directories, by default, have their access/mod 1188 * times and permissions set to the values specified in the archive. During the 1189 * actions of extracting (and creating the destination subtree during -rw copy) 1190 * directories extracted may be modified after being created. Even worse is 1191 * that these directories may have been created with file permissions which 1192 * prohibits any descendants of these directories from being extracted. When 1193 * directories are created by pax, access rights may be added to permit the 1194 * creation of files in their subtree. Every time pax creates a directory, the 1195 * times and file permissions specified by the archive are stored. After all 1196 * files have been extracted (or copied), these directories have their times 1197 * and file modes reset to the stored values. The directory info is restored in 1198 * reverse order as entries were added to the data file from root to leaf. To 1199 * restore atime properly, we must go backwards. The data file consists of 1200 * records with two parts, the file name followed by a DIRDATA trailer. The 1201 * fixed sized trailer contains the size of the name plus the off_t location in 1202 * the file. To restore we work backwards through the file reading the trailer 1203 * then the file name. 1204 */ 1205 1206 /* 1207 * dir_start() 1208 * set up the directory time and file mode storage for directories CREATED 1209 * by pax. 1210 * Return: 1211 * 0 if ok, -1 otherwise 1212 */ 1213 1214 #if __STDC__ 1215 int 1216 dir_start(void) 1217 #else 1218 int 1219 dir_start() 1220 #endif 1221 { 1222 char *pt; 1223 1224 if (dirfd != -1) 1225 return(0); 1226 1227 /* 1228 * unlink the file so it goes away at termination by itself 1229 */ 1230 pt = strdup("/tmp/paxXXXXXX"); 1231 if (pt == NULL) { 1232 tty_warn(1, "Unable to allocate memory"); 1233 return(-1); 1234 } 1235 if ((dirfd = mkstemp(pt)) >= 0) { 1236 (void)unlink(pt); 1237 free(pt); 1238 return(0); 1239 } 1240 tty_warn(1, "Unable to create temporary file for directory times: %s", 1241 pt); 1242 free(pt); 1243 return(-1); 1244 } 1245 1246 /* 1247 * add_dir() 1248 * add the mode and times for a newly CREATED directory 1249 * name is name of the directory, psb the stat buffer with the data in it, 1250 * frc_mode is a flag that says whether to force the setting of the mode 1251 * (ignoring the user set values for preserving file mode). Frc_mode is 1252 * for the case where we created a file and found that the resulting 1253 * directory was not writeable and the user asked for file modes to NOT 1254 * be preserved. (we have to preserve what was created by default, so we 1255 * have to force the setting at the end. this is stated explicitly in the 1256 * pax spec) 1257 */ 1258 1259 #if __STDC__ 1260 void 1261 add_dir(char *name, int nlen, struct stat *psb, int frc_mode) 1262 #else 1263 void 1264 add_dir(name, nlen, psb, frc_mode) 1265 char *name; 1266 int nlen; 1267 struct stat *psb; 1268 int frc_mode; 1269 #endif 1270 { 1271 DIRDATA dblk; 1272 1273 if (dirfd < 0) 1274 return; 1275 1276 /* 1277 * get current position (where file name will start) so we can store it 1278 * in the trailer 1279 */ 1280 if ((dblk.npos = lseek(dirfd, 0L, SEEK_CUR)) < 0) { 1281 tty_warn(1, 1282 "Unable to store mode and times for directory: %s",name); 1283 return; 1284 } 1285 1286 /* 1287 * write the file name followed by the trailer 1288 */ 1289 dblk.nlen = nlen + 1; 1290 dblk.mode = psb->st_mode & 0xffff; 1291 dblk.mtime = psb->st_mtime; 1292 dblk.atime = psb->st_atime; 1293 dblk.frc_mode = frc_mode; 1294 if ((write(dirfd, name, dblk.nlen) == dblk.nlen) && 1295 (write(dirfd, (char *)&dblk, sizeof(dblk)) == sizeof(dblk))) { 1296 ++dircnt; 1297 return; 1298 } 1299 1300 tty_warn(1, 1301 "Unable to store mode and times for created directory: %s",name); 1302 return; 1303 } 1304 1305 /* 1306 * proc_dir() 1307 * process all file modes and times stored for directories CREATED 1308 * by pax 1309 */ 1310 1311 #if __STDC__ 1312 void 1313 proc_dir(void) 1314 #else 1315 void 1316 proc_dir() 1317 #endif 1318 { 1319 char name[PAXPATHLEN+1]; 1320 DIRDATA dblk; 1321 u_long cnt; 1322 1323 if (dirfd < 0) 1324 return; 1325 /* 1326 * read backwards through the file and process each directory 1327 */ 1328 for (cnt = 0; cnt < dircnt; ++cnt) { 1329 /* 1330 * read the trailer, then the file name, if this fails 1331 * just give up. 1332 */ 1333 if (lseek(dirfd, -((off_t)sizeof(dblk)), SEEK_CUR) < 0) 1334 break; 1335 if (read(dirfd,(char *)&dblk, sizeof(dblk)) != sizeof(dblk)) 1336 break; 1337 if (lseek(dirfd, dblk.npos, SEEK_SET) < 0) 1338 break; 1339 if (read(dirfd, name, dblk.nlen) != dblk.nlen) 1340 break; 1341 if (lseek(dirfd, dblk.npos, SEEK_SET) < 0) 1342 break; 1343 1344 /* 1345 * frc_mode set, make sure we set the file modes even if 1346 * the user didn't ask for it (see file_subs.c for more info) 1347 */ 1348 if (pmode || dblk.frc_mode) 1349 set_pmode(name, dblk.mode); 1350 if (patime || pmtime) 1351 set_ftime(name, dblk.mtime, dblk.atime, 0); 1352 } 1353 1354 (void)close(dirfd); 1355 dirfd = -1; 1356 if (cnt != dircnt) 1357 tty_warn(1, 1358 "Unable to set mode and times for created directories"); 1359 return; 1360 } 1361 1362 /* 1363 * database independent routines 1364 */ 1365 1366 /* 1367 * st_hash() 1368 * hashes filenames to a u_int for hashing into a table. Looks at the tail 1369 * end of file, as this provides far better distribution than any other 1370 * part of the name. For performance reasons we only care about the last 1371 * MAXKEYLEN chars (should be at LEAST large enough to pick off the file 1372 * name). Was tested on 500,000 name file tree traversal from the root 1373 * and gave almost a perfectly uniform distribution of keys when used with 1374 * prime sized tables (MAXKEYLEN was 128 in test). Hashes (sizeof int) 1375 * chars at a time and pads with 0 for last addition. 1376 * Return: 1377 * the hash value of the string MOD (%) the table size. 1378 */ 1379 1380 #if __STDC__ 1381 u_int 1382 st_hash(char *name, int len, int tabsz) 1383 #else 1384 u_int 1385 st_hash(name, len, tabsz) 1386 char *name; 1387 int len; 1388 int tabsz; 1389 #endif 1390 { 1391 char *pt; 1392 char *dest; 1393 char *end; 1394 int i; 1395 u_int key = 0; 1396 int steps; 1397 int res; 1398 u_int val; 1399 1400 /* 1401 * only look at the tail up to MAXKEYLEN, we do not need to waste 1402 * time here (remember these are pathnames, the tail is what will 1403 * spread out the keys) 1404 */ 1405 if (len > MAXKEYLEN) { 1406 pt = &(name[len - MAXKEYLEN]); 1407 len = MAXKEYLEN; 1408 } else 1409 pt = name; 1410 1411 /* 1412 * calculate the number of u_int size steps in the string and if 1413 * there is a runt to deal with 1414 */ 1415 steps = len/sizeof(u_int); 1416 res = len % sizeof(u_int); 1417 1418 /* 1419 * add up the value of the string in unsigned integer sized pieces 1420 * too bad we cannot have unsigned int aligned strings, then we 1421 * could avoid the expensive copy. 1422 */ 1423 for (i = 0; i < steps; ++i) { 1424 end = pt + sizeof(u_int); 1425 dest = (char *)&val; 1426 while (pt < end) 1427 *dest++ = *pt++; 1428 key += val; 1429 } 1430 1431 /* 1432 * add in the runt padded with zero to the right 1433 */ 1434 if (res) { 1435 val = 0; 1436 end = pt + res; 1437 dest = (char *)&val; 1438 while (pt < end) 1439 *dest++ = *pt++; 1440 key += val; 1441 } 1442 1443 /* 1444 * return the result mod the table size 1445 */ 1446 return(key % tabsz); 1447 } 1448