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. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)tables.c 8.1 (Berkeley) 5/31/93 34 * $FreeBSD: src/bin/pax/tables.c,v 1.13.2.1 2001/08/01 05:03:12 obrien Exp $ 35 */ 36 37 #include <sys/types.h> 38 #include <sys/time.h> 39 #include <sys/stat.h> 40 #include <sys/fcntl.h> 41 #include <errno.h> 42 #include <stdio.h> 43 #include <stdlib.h> 44 #include <string.h> 45 #include <unistd.h> 46 #include "pax.h" 47 #include "tables.h" 48 #include "extern.h" 49 50 /* 51 * Routines for controlling the contents of all the different databases pax 52 * keeps. Tables are dynamically created only when they are needed. The 53 * goal was speed and the ability to work with HUGE archives. The databases 54 * were kept simple, but do have complex rules for when the contents change. 55 * As of this writing, the POSIX library functions were more complex than 56 * needed for this application (pax databases have very short lifetimes and 57 * do not survive after pax is finished). Pax is required to handle very 58 * large archives. These database routines carefully combine memory usage and 59 * temporary file storage in ways which will not significantly impact runtime 60 * performance while allowing the largest possible archives to be handled. 61 * Trying to force the fit to the POSIX database routines was not considered 62 * time well spent. 63 */ 64 65 static HRDLNK **ltab = NULL; /* hard link table for detecting hard links */ 66 static FTM **ftab = NULL; /* file time table for updating arch */ 67 static NAMT **ntab = NULL; /* interactive rename storage table */ 68 static DEVT **dtab = NULL; /* device/inode mapping tables */ 69 static ATDIR **atab = NULL; /* file tree directory time reset table */ 70 static int dirfd = -1; /* storage for setting created dir time/mode */ 71 static u_long dircnt; /* entries in dir time/mode storage */ 72 static int ffd = -1; /* tmp file for file time table name storage */ 73 74 static DEVT *chk_dev (dev_t, int); 75 76 /* 77 * hard link table routines 78 * 79 * The hard link table tries to detect hard links to files using the device and 80 * inode values. We do this when writing an archive, so we can tell the format 81 * write routine that this file is a hard link to another file. The format 82 * write routine then can store this file in whatever way it wants (as a hard 83 * link if the format supports that like tar, or ignore this info like cpio). 84 * (Actually a field in the format driver table tells us if the format wants 85 * hard link info. if not, we do not waste time looking for them). We also use 86 * the same table when reading an archive. In that situation, this table is 87 * used by the format read routine to detect hard links from stored dev and 88 * inode numbers (like cpio). This will allow pax to create a link when one 89 * can be detected by the archive format. 90 */ 91 92 /* 93 * lnk_start 94 * Creates the hard link table. 95 * Return: 96 * 0 if created, -1 if failure 97 */ 98 99 int 100 lnk_start(void) 101 { 102 if (ltab != NULL) 103 return(0); 104 if ((ltab = (HRDLNK **)calloc(L_TAB_SZ, sizeof(HRDLNK *))) == NULL) { 105 paxwarn(1, "Cannot allocate memory for hard link table"); 106 return(-1); 107 } 108 return(0); 109 } 110 111 /* 112 * chk_lnk() 113 * Looks up entry in hard link hash table. If found, it copies the name 114 * of the file it is linked to (we already saw that file) into ln_name. 115 * lnkcnt is decremented and if goes to 1 the node is deleted from the 116 * database. (We have seen all the links to this file). If not found, 117 * we add the file to the database if it has the potential for having 118 * hard links to other files we may process (it has a link count > 1) 119 * Return: 120 * if found returns 1; if not found returns 0; -1 on error 121 */ 122 123 int 124 chk_lnk(ARCHD *arcn) 125 { 126 HRDLNK *pt; 127 HRDLNK **ppt; 128 u_int indx; 129 130 if (ltab == NULL) 131 return(-1); 132 /* 133 * ignore those nodes that cannot have hard links 134 */ 135 if ((arcn->type == PAX_DIR) || (arcn->sb.st_nlink <= 1)) 136 return(0); 137 138 /* 139 * hash inode number and look for this file 140 */ 141 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ; 142 if ((pt = ltab[indx]) != NULL) { 143 /* 144 * it's hash chain in not empty, walk down looking for it 145 */ 146 ppt = &(ltab[indx]); 147 while (pt != NULL) { 148 if ((pt->ino == arcn->sb.st_ino) && 149 (pt->dev == arcn->sb.st_dev)) 150 break; 151 ppt = &(pt->fow); 152 pt = pt->fow; 153 } 154 155 if (pt != NULL) { 156 /* 157 * found a link. set the node type and copy in the 158 * name of the file it is to link to. we need to 159 * handle hardlinks to regular files differently than 160 * other links. 161 */ 162 arcn->ln_nlen = l_strncpy(arcn->ln_name, pt->name, 163 sizeof(arcn->ln_name) - 1); 164 arcn->ln_name[arcn->ln_nlen] = '\0'; 165 if (arcn->type == PAX_REG) 166 arcn->type = PAX_HRG; 167 else 168 arcn->type = PAX_HLK; 169 170 /* 171 * if we have found all the links to this file, remove 172 * it from the database 173 */ 174 if (--pt->nlink <= 1) { 175 *ppt = pt->fow; 176 free((char *)pt->name); 177 free((char *)pt); 178 } 179 return(1); 180 } 181 } 182 183 /* 184 * we never saw this file before. It has links so we add it to the 185 * front of this hash chain 186 */ 187 if ((pt = (HRDLNK *)malloc(sizeof(HRDLNK))) != NULL) { 188 if ((pt->name = strdup(arcn->name)) != NULL) { 189 pt->dev = arcn->sb.st_dev; 190 pt->ino = arcn->sb.st_ino; 191 pt->nlink = arcn->sb.st_nlink; 192 pt->fow = ltab[indx]; 193 ltab[indx] = pt; 194 return(0); 195 } 196 free((char *)pt); 197 } 198 199 paxwarn(1, "Hard link table out of memory"); 200 return(-1); 201 } 202 203 /* 204 * purg_lnk 205 * remove reference for a file that we may have added to the data base as 206 * a potential source for hard links. We ended up not using the file, so 207 * we do not want to accidently point another file at it later on. 208 */ 209 210 void 211 purg_lnk(ARCHD *arcn) 212 { 213 HRDLNK *pt; 214 HRDLNK **ppt; 215 u_int indx; 216 217 if (ltab == NULL) 218 return; 219 /* 220 * do not bother to look if it could not be in the database 221 */ 222 if ((arcn->sb.st_nlink <= 1) || (arcn->type == PAX_DIR) || 223 (arcn->type == PAX_HLK) || (arcn->type == PAX_HRG)) 224 return; 225 226 /* 227 * find the hash chain for this inode value, if empty return 228 */ 229 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ; 230 if ((pt = ltab[indx]) == NULL) 231 return; 232 233 /* 234 * walk down the list looking for the inode/dev pair, unlink and 235 * free if found 236 */ 237 ppt = &(ltab[indx]); 238 while (pt != NULL) { 239 if ((pt->ino == arcn->sb.st_ino) && 240 (pt->dev == arcn->sb.st_dev)) 241 break; 242 ppt = &(pt->fow); 243 pt = pt->fow; 244 } 245 if (pt == NULL) 246 return; 247 248 /* 249 * remove and free it 250 */ 251 *ppt = pt->fow; 252 free((char *)pt->name); 253 free((char *)pt); 254 } 255 256 /* 257 * lnk_end() 258 * pull apart a existing link table so we can reuse it. We do this between 259 * read and write phases of append with update. (The format may have 260 * used the link table, and we need to start with a fresh table for the 261 * write phase 262 */ 263 264 void 265 lnk_end(void) 266 { 267 int i; 268 HRDLNK *pt; 269 HRDLNK *ppt; 270 271 if (ltab == NULL) 272 return; 273 274 for (i = 0; i < L_TAB_SZ; ++i) { 275 if (ltab[i] == NULL) 276 continue; 277 pt = ltab[i]; 278 ltab[i] = NULL; 279 280 /* 281 * free up each entry on this chain 282 */ 283 while (pt != NULL) { 284 ppt = pt; 285 pt = ppt->fow; 286 free((char *)ppt->name); 287 free((char *)ppt); 288 } 289 } 290 return; 291 } 292 293 /* 294 * modification time table routines 295 * 296 * The modification time table keeps track of last modification times for all 297 * files stored in an archive during a write phase when -u is set. We only 298 * add a file to the archive if it is newer than a file with the same name 299 * already stored on the archive (if there is no other file with the same 300 * name on the archive it is added). This applies to writes and appends. 301 * An append with an -u must read the archive and store the modification time 302 * for every file on that archive before starting the write phase. It is clear 303 * that this is one HUGE database. To save memory space, the actual file names 304 * are stored in a scratch file and indexed by an in memory hash table. The 305 * hash table is indexed by hashing the file path. The nodes in the table store 306 * the length of the filename and the lseek offset within the scratch file 307 * where the actual name is stored. Since there are never any deletions from 308 * this table, fragmentation of the scratch file is never a issue. Lookups 309 * seem to not exhibit any locality at all (files in the database are rarely 310 * looked up more than once...), so caching is just a waste of memory. The 311 * only limitation is the amount of scratch file space available to store the 312 * path names. 313 */ 314 315 /* 316 * ftime_start() 317 * create the file time hash table and open for read/write the scratch 318 * file. (after created it is unlinked, so when we exit we leave 319 * no witnesses). 320 * Return: 321 * 0 if the table and file was created ok, -1 otherwise 322 */ 323 324 int 325 ftime_start(void) 326 { 327 328 if (ftab != NULL) 329 return(0); 330 if ((ftab = (FTM **)calloc(F_TAB_SZ, sizeof(FTM *))) == NULL) { 331 paxwarn(1, "Cannot allocate memory for file time table"); 332 return(-1); 333 } 334 335 /* 336 * get random name and create temporary scratch file, unlink name 337 * so it will get removed on exit 338 */ 339 memcpy(tempbase, _TFILE_BASE, sizeof(_TFILE_BASE)); 340 if ((ffd = mkstemp(tempfile)) < 0) { 341 syswarn(1, errno, "Unable to create temporary file: %s", 342 tempfile); 343 return(-1); 344 } 345 unlink(tempfile); 346 347 return(0); 348 } 349 350 /* 351 * chk_ftime() 352 * looks up entry in file time hash table. If not found, the file is 353 * added to the hash table and the file named stored in the scratch file. 354 * If a file with the same name is found, the file times are compared and 355 * the most recent file time is retained. If the new file was younger (or 356 * was not in the database) the new file is selected for storage. 357 * Return: 358 * 0 if file should be added to the archive, 1 if it should be skipped, 359 * -1 on error 360 */ 361 362 int 363 chk_ftime(ARCHD *arcn) 364 { 365 FTM *pt; 366 int namelen; 367 u_int indx; 368 char ckname[PAXPATHLEN+1]; 369 370 /* 371 * no info, go ahead and add to archive 372 */ 373 if (ftab == NULL) 374 return(0); 375 376 /* 377 * hash the pathname and look up in table 378 */ 379 namelen = arcn->nlen; 380 indx = st_hash(arcn->name, namelen, F_TAB_SZ); 381 if ((pt = ftab[indx]) != NULL) { 382 /* 383 * the hash chain is not empty, walk down looking for match 384 * only read up the path names if the lengths match, speeds 385 * up the search a lot 386 */ 387 while (pt != NULL) { 388 if (pt->namelen == namelen) { 389 /* 390 * potential match, have to read the name 391 * from the scratch file. 392 */ 393 if (lseek(ffd,pt->seek,SEEK_SET) != pt->seek) { 394 syswarn(1, errno, 395 "Failed ftime table seek"); 396 return(-1); 397 } 398 if (read(ffd, ckname, namelen) != namelen) { 399 syswarn(1, errno, 400 "Failed ftime table read"); 401 return(-1); 402 } 403 404 /* 405 * if the names match, we are done 406 */ 407 if (!strncmp(ckname, arcn->name, namelen)) 408 break; 409 } 410 411 /* 412 * try the next entry on the chain 413 */ 414 pt = pt->fow; 415 } 416 417 if (pt != NULL) { 418 /* 419 * found the file, compare the times, save the newer 420 */ 421 if (arcn->sb.st_mtime > pt->mtime) { 422 /* 423 * file is newer 424 */ 425 pt->mtime = arcn->sb.st_mtime; 426 return(0); 427 } 428 /* 429 * file is older 430 */ 431 return(1); 432 } 433 } 434 435 /* 436 * not in table, add it 437 */ 438 if ((pt = (FTM *)malloc(sizeof(FTM))) != NULL) { 439 /* 440 * add the name at the end of the scratch file, saving the 441 * offset. add the file to the head of the hash chain 442 */ 443 if ((pt->seek = lseek(ffd, (off_t)0, SEEK_END)) >= 0) { 444 if (write(ffd, arcn->name, namelen) == namelen) { 445 pt->mtime = arcn->sb.st_mtime; 446 pt->namelen = namelen; 447 pt->fow = ftab[indx]; 448 ftab[indx] = pt; 449 return(0); 450 } 451 syswarn(1, errno, "Failed write to file time table"); 452 } else 453 syswarn(1, errno, "Failed seek on file time table"); 454 } else 455 paxwarn(1, "File time table ran out of memory"); 456 457 if (pt != NULL) 458 free((char *)pt); 459 return(-1); 460 } 461 462 /* 463 * Interactive rename table routines 464 * 465 * The interactive rename table keeps track of the new names that the user 466 * assigns to files from tty input. Since this map is unique for each file 467 * we must store it in case there is a reference to the file later in archive 468 * (a link). Otherwise we will be unable to find the file we know was 469 * extracted. The remapping of these files is stored in a memory based hash 470 * table (it is assumed since input must come from /dev/tty, it is unlikely to 471 * be a very large table). 472 */ 473 474 /* 475 * name_start() 476 * create the interactive rename table 477 * Return: 478 * 0 if successful, -1 otherwise 479 */ 480 481 int 482 name_start(void) 483 { 484 if (ntab != NULL) 485 return(0); 486 if ((ntab = (NAMT **)calloc(N_TAB_SZ, sizeof(NAMT *))) == NULL) { 487 paxwarn(1, "Cannot allocate memory for interactive rename table"); 488 return(-1); 489 } 490 return(0); 491 } 492 493 /* 494 * add_name() 495 * add the new name to old name mapping just created by the user. 496 * If an old name mapping is found (there may be duplicate names on an 497 * archive) only the most recent is kept. 498 * Return: 499 * 0 if added, -1 otherwise 500 */ 501 502 int 503 add_name(char *oname, int onamelen, char *nname) 504 { 505 NAMT *pt; 506 u_int indx; 507 508 if (ntab == NULL) { 509 /* 510 * should never happen 511 */ 512 paxwarn(0, "No interactive rename table, links may fail\n"); 513 return(0); 514 } 515 516 /* 517 * look to see if we have already mapped this file, if so we 518 * will update it 519 */ 520 indx = st_hash(oname, onamelen, N_TAB_SZ); 521 if ((pt = ntab[indx]) != NULL) { 522 /* 523 * look down the has chain for the file 524 */ 525 while ((pt != NULL) && (strcmp(oname, pt->oname) != 0)) 526 pt = pt->fow; 527 528 if (pt != NULL) { 529 /* 530 * found an old mapping, replace it with the new one 531 * the user just input (if it is different) 532 */ 533 if (strcmp(nname, pt->nname) == 0) 534 return(0); 535 536 free((char *)pt->nname); 537 if ((pt->nname = strdup(nname)) == NULL) { 538 paxwarn(1, "Cannot update rename table"); 539 return(-1); 540 } 541 return(0); 542 } 543 } 544 545 /* 546 * this is a new mapping, add it to the table 547 */ 548 if ((pt = (NAMT *)malloc(sizeof(NAMT))) != NULL) { 549 if ((pt->oname = strdup(oname)) != NULL) { 550 if ((pt->nname = strdup(nname)) != NULL) { 551 pt->fow = ntab[indx]; 552 ntab[indx] = pt; 553 return(0); 554 } 555 free((char *)pt->oname); 556 } 557 free((char *)pt); 558 } 559 paxwarn(1, "Interactive rename table out of memory"); 560 return(-1); 561 } 562 563 /* 564 * sub_name() 565 * look up a link name to see if it points at a file that has been 566 * remapped by the user. If found, the link is adjusted to contain the 567 * new name (oname is the link to name) 568 */ 569 570 void 571 sub_name(char *oname, int *onamelen, size_t onamesize) 572 { 573 NAMT *pt; 574 u_int indx; 575 576 if (ntab == NULL) 577 return; 578 /* 579 * look the name up in the hash table 580 */ 581 indx = st_hash(oname, *onamelen, N_TAB_SZ); 582 if ((pt = ntab[indx]) == NULL) 583 return; 584 585 while (pt != NULL) { 586 /* 587 * walk down the hash chain looking for a match 588 */ 589 if (strcmp(oname, pt->oname) == 0) { 590 /* 591 * found it, replace it with the new name 592 * and return (we know that oname has enough space) 593 */ 594 *onamelen = l_strncpy(oname, pt->nname, onamesize - 1); 595 oname[*onamelen] = '\0'; 596 return; 597 } 598 pt = pt->fow; 599 } 600 601 /* 602 * no match, just return 603 */ 604 return; 605 } 606 607 /* 608 * device/inode mapping table routines 609 * (used with formats that store device and inodes fields) 610 * 611 * device/inode mapping tables remap the device field in a archive header. The 612 * device/inode fields are used to determine when files are hard links to each 613 * other. However these values have very little meaning outside of that. This 614 * database is used to solve one of two different problems. 615 * 616 * 1) when files are appended to an archive, while the new files may have hard 617 * links to each other, you cannot determine if they have hard links to any 618 * file already stored on the archive from a prior run of pax. We must assume 619 * that these inode/device pairs are unique only within a SINGLE run of pax 620 * (which adds a set of files to an archive). So we have to make sure the 621 * inode/dev pairs we add each time are always unique. We do this by observing 622 * while the inode field is very dense, the use of the dev field is fairly 623 * sparse. Within each run of pax, we remap any device number of a new archive 624 * member that has a device number used in a prior run and already stored in a 625 * file on the archive. During the read phase of the append, we store the 626 * device numbers used and mark them to not be used by any file during the 627 * write phase. If during write we go to use one of those old device numbers, 628 * we remap it to a new value. 629 * 630 * 2) Often the fields in the archive header used to store these values are 631 * too small to store the entire value. The result is an inode or device value 632 * which can be truncated. This really can foul up an archive. With truncation 633 * we end up creating links between files that are really not links (after 634 * truncation the inodes are the same value). We address that by detecting 635 * truncation and forcing a remap of the device field to split truncated 636 * inodes away from each other. Each truncation creates a pattern of bits that 637 * are removed. We use this pattern of truncated bits to partition the inodes 638 * on a single device to many different devices (each one represented by the 639 * truncated bit pattern). All inodes on the same device that have the same 640 * truncation pattern are mapped to the same new device. Two inodes that 641 * truncate to the same value clearly will always have different truncation 642 * bit patterns, so they will be split from away each other. When we spot 643 * device truncation we remap the device number to a non truncated value. 644 * (for more info see table.h for the data structures involved). 645 */ 646 647 /* 648 * dev_start() 649 * create the device mapping table 650 * Return: 651 * 0 if successful, -1 otherwise 652 */ 653 654 int 655 dev_start(void) 656 { 657 if (dtab != NULL) 658 return(0); 659 if ((dtab = (DEVT **)calloc(D_TAB_SZ, sizeof(DEVT *))) == NULL) { 660 paxwarn(1, "Cannot allocate memory for device mapping table"); 661 return(-1); 662 } 663 return(0); 664 } 665 666 /* 667 * add_dev() 668 * add a device number to the table. this will force the device to be 669 * remapped to a new value if it be used during a write phase. This 670 * function is called during the read phase of an append to prohibit the 671 * use of any device number already in the archive. 672 * Return: 673 * 0 if added ok, -1 otherwise 674 */ 675 676 int 677 add_dev(ARCHD *arcn) 678 { 679 if (chk_dev(arcn->sb.st_dev, 1) == NULL) 680 return(-1); 681 return(0); 682 } 683 684 /* 685 * chk_dev() 686 * check for a device value in the device table. If not found and the add 687 * flag is set, it is added. This does NOT assign any mapping values, just 688 * adds the device number as one that need to be remapped. If this device 689 * is already mapped, just return with a pointer to that entry. 690 * Return: 691 * pointer to the entry for this device in the device map table. Null 692 * if the add flag is not set and the device is not in the table (it is 693 * not been seen yet). If add is set and the device cannot be added, null 694 * is returned (indicates an error). 695 */ 696 697 static DEVT * 698 chk_dev(dev_t dev, int add) 699 { 700 DEVT *pt; 701 u_int indx; 702 703 if (dtab == NULL) 704 return(NULL); 705 /* 706 * look to see if this device is already in the table 707 */ 708 indx = ((unsigned)dev) % D_TAB_SZ; 709 if ((pt = dtab[indx]) != NULL) { 710 while ((pt != NULL) && (pt->dev != dev)) 711 pt = pt->fow; 712 713 /* 714 * found it, return a pointer to it 715 */ 716 if (pt != NULL) 717 return(pt); 718 } 719 720 /* 721 * not in table, we add it only if told to as this may just be a check 722 * to see if a device number is being used. 723 */ 724 if (add == 0) 725 return(NULL); 726 727 /* 728 * allocate a node for this device and add it to the front of the hash 729 * chain. Note we do not assign remaps values here, so the pt->list 730 * list must be NULL. 731 */ 732 if ((pt = (DEVT *)malloc(sizeof(DEVT))) == NULL) { 733 paxwarn(1, "Device map table out of memory"); 734 return(NULL); 735 } 736 pt->dev = dev; 737 pt->list = NULL; 738 pt->fow = dtab[indx]; 739 dtab[indx] = pt; 740 return(pt); 741 } 742 /* 743 * map_dev() 744 * given an inode and device storage mask (the mask has a 1 for each bit 745 * the archive format is able to store in a header), we check for inode 746 * and device truncation and remap the device as required. Device mapping 747 * can also occur when during the read phase of append a device number was 748 * seen (and was marked as do not use during the write phase). WE ASSUME 749 * that unsigned longs are the same size or bigger than the fields used 750 * for ino_t and dev_t. If not the types will have to be changed. 751 * Return: 752 * 0 if all ok, -1 otherwise. 753 */ 754 755 int 756 map_dev(ARCHD *arcn, u_long dev_mask, u_long ino_mask) 757 { 758 DEVT *pt; 759 DLIST *dpt; 760 static dev_t lastdev = 0; /* next device number to try */ 761 int trc_ino = 0; 762 int trc_dev = 0; 763 ino_t trunc_bits = 0; 764 ino_t nino; 765 766 if (dtab == NULL) 767 return(0); 768 /* 769 * check for device and inode truncation, and extract the truncated 770 * bit pattern. 771 */ 772 if ((arcn->sb.st_dev & (dev_t)dev_mask) != arcn->sb.st_dev) 773 ++trc_dev; 774 if ((nino = arcn->sb.st_ino & (ino_t)ino_mask) != arcn->sb.st_ino) { 775 ++trc_ino; 776 trunc_bits = arcn->sb.st_ino & (ino_t)(~ino_mask); 777 } 778 779 /* 780 * see if this device is already being mapped, look up the device 781 * then find the truncation bit pattern which applies 782 */ 783 if ((pt = chk_dev(arcn->sb.st_dev, 0)) != NULL) { 784 /* 785 * this device is already marked to be remapped 786 */ 787 for (dpt = pt->list; dpt != NULL; dpt = dpt->fow) 788 if (dpt->trunc_bits == trunc_bits) 789 break; 790 791 if (dpt != NULL) { 792 /* 793 * we are being remapped for this device and pattern 794 * change the device number to be stored and return 795 */ 796 arcn->sb.st_dev = dpt->dev; 797 arcn->sb.st_ino = nino; 798 return(0); 799 } 800 } else { 801 /* 802 * this device is not being remapped YET. if we do not have any 803 * form of truncation, we do not need a remap 804 */ 805 if (!trc_ino && !trc_dev) 806 return(0); 807 808 /* 809 * we have truncation, have to add this as a device to remap 810 */ 811 if ((pt = chk_dev(arcn->sb.st_dev, 1)) == NULL) 812 goto bad; 813 814 /* 815 * if we just have a truncated inode, we have to make sure that 816 * all future inodes that do not truncate (they have the 817 * truncation pattern of all 0's) continue to map to the same 818 * device number. We probably have already written inodes with 819 * this device number to the archive with the truncation 820 * pattern of all 0's. So we add the mapping for all 0's to the 821 * same device number. 822 */ 823 if (!trc_dev && (trunc_bits != 0)) { 824 if ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL) 825 goto bad; 826 dpt->trunc_bits = 0; 827 dpt->dev = arcn->sb.st_dev; 828 dpt->fow = pt->list; 829 pt->list = dpt; 830 } 831 } 832 833 /* 834 * look for a device number not being used. We must watch for wrap 835 * around on lastdev (so we do not get stuck looking forever!) 836 */ 837 while (++lastdev > 0) { 838 if (chk_dev(lastdev, 0) != NULL) 839 continue; 840 /* 841 * found an unused value. If we have reached truncation point 842 * for this format we are hosed, so we give up. Otherwise we 843 * mark it as being used. 844 */ 845 if (((lastdev & ((dev_t)dev_mask)) != lastdev) || 846 (chk_dev(lastdev, 1) == NULL)) 847 goto bad; 848 break; 849 } 850 851 if ((lastdev <= 0) || ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL)) 852 goto bad; 853 854 /* 855 * got a new device number, store it under this truncation pattern. 856 * change the device number this file is being stored with. 857 */ 858 dpt->trunc_bits = trunc_bits; 859 dpt->dev = lastdev; 860 dpt->fow = pt->list; 861 pt->list = dpt; 862 arcn->sb.st_dev = lastdev; 863 arcn->sb.st_ino = nino; 864 return(0); 865 866 bad: 867 paxwarn(1, "Unable to fix truncated inode/device field when storing %s", 868 arcn->name); 869 paxwarn(0, "Archive may create improper hard links when extracted"); 870 return(0); 871 } 872 873 /* 874 * directory access/mod time reset table routines (for directories READ by pax) 875 * 876 * The pax -t flag requires that access times of archive files be the same 877 * before being read by pax. For regular files, access time is restored after 878 * the file has been copied. This database provides the same functionality for 879 * directories read during file tree traversal. Restoring directory access time 880 * is more complex than files since directories may be read several times until 881 * all the descendants in their subtree are visited by fts. Directory access 882 * and modification times are stored during the fts pre-order visit (done 883 * before any descendants in the subtree are visited) and restored after the 884 * fts post-order visit (after all the descendants have been visited). In the 885 * case of premature exit from a subtree (like from the effects of -n), any 886 * directory entries left in this database are reset during final cleanup 887 * operations of pax. Entries are hashed by inode number for fast lookup. 888 */ 889 890 /* 891 * atdir_start() 892 * create the directory access time database for directories READ by pax. 893 * Return: 894 * 0 is created ok, -1 otherwise. 895 */ 896 897 int 898 atdir_start(void) 899 { 900 if (atab != NULL) 901 return(0); 902 if ((atab = (ATDIR **)calloc(A_TAB_SZ, sizeof(ATDIR *))) == NULL) { 903 paxwarn(1,"Cannot allocate space for directory access time table"); 904 return(-1); 905 } 906 return(0); 907 } 908 909 910 /* 911 * atdir_end() 912 * walk through the directory access time table and reset the access time 913 * of any directory who still has an entry left in the database. These 914 * entries are for directories READ by pax 915 */ 916 917 void 918 atdir_end(void) 919 { 920 ATDIR *pt; 921 int i; 922 923 if (atab == NULL) 924 return; 925 /* 926 * for each non-empty hash table entry reset all the directories 927 * chained there. 928 */ 929 for (i = 0; i < A_TAB_SZ; ++i) { 930 if ((pt = atab[i]) == NULL) 931 continue; 932 /* 933 * remember to force the times, set_ftime() looks at pmtime 934 * and patime, which only applies to things CREATED by pax, 935 * not read by pax. Read time reset is controlled by -t. 936 */ 937 for (; pt != NULL; pt = pt->fow) 938 set_ftime(pt->name, pt->mtime, pt->atime, 1); 939 } 940 } 941 942 /* 943 * add_atdir() 944 * add a directory to the directory access time table. Table is hashed 945 * and chained by inode number. This is for directories READ by pax 946 */ 947 948 void 949 add_atdir(char *fname, dev_t dev, ino_t ino, time_t mtime, time_t atime) 950 { 951 ATDIR *pt; 952 u_int indx; 953 954 if (atab == NULL) 955 return; 956 957 /* 958 * make sure this directory is not already in the table, if so just 959 * return (the older entry always has the correct time). The only 960 * way this will happen is when the same subtree can be traversed by 961 * different args to pax and the -n option is aborting fts out of a 962 * subtree before all the post-order visits have been made. 963 */ 964 indx = ((unsigned)ino) % A_TAB_SZ; 965 if ((pt = atab[indx]) != NULL) { 966 while (pt != NULL) { 967 if ((pt->ino == ino) && (pt->dev == dev)) 968 break; 969 pt = pt->fow; 970 } 971 972 /* 973 * oops, already there. Leave it alone. 974 */ 975 if (pt != NULL) 976 return; 977 } 978 979 /* 980 * add it to the front of the hash chain 981 */ 982 if ((pt = (ATDIR *)malloc(sizeof(ATDIR))) != NULL) { 983 if ((pt->name = strdup(fname)) != NULL) { 984 pt->dev = dev; 985 pt->ino = ino; 986 pt->mtime = mtime; 987 pt->atime = atime; 988 pt->fow = atab[indx]; 989 atab[indx] = pt; 990 return; 991 } 992 free((char *)pt); 993 } 994 995 paxwarn(1, "Directory access time reset table ran out of memory"); 996 return; 997 } 998 999 /* 1000 * get_atdir() 1001 * look up a directory by inode and device number to obtain the access 1002 * and modification time you want to set to. If found, the modification 1003 * and access time parameters are set and the entry is removed from the 1004 * table (as it is no longer needed). These are for directories READ by 1005 * pax 1006 * Return: 1007 * 0 if found, -1 if not found. 1008 */ 1009 1010 int 1011 get_atdir(dev_t dev, ino_t ino, time_t *mtime, time_t *atime) 1012 { 1013 ATDIR *pt; 1014 ATDIR **ppt; 1015 u_int indx; 1016 1017 if (atab == NULL) 1018 return(-1); 1019 /* 1020 * hash by inode and search the chain for an inode and device match 1021 */ 1022 indx = ((unsigned)ino) % A_TAB_SZ; 1023 if ((pt = atab[indx]) == NULL) 1024 return(-1); 1025 1026 ppt = &(atab[indx]); 1027 while (pt != NULL) { 1028 if ((pt->ino == ino) && (pt->dev == dev)) 1029 break; 1030 /* 1031 * no match, go to next one 1032 */ 1033 ppt = &(pt->fow); 1034 pt = pt->fow; 1035 } 1036 1037 /* 1038 * return if we did not find it. 1039 */ 1040 if (pt == NULL) 1041 return(-1); 1042 1043 /* 1044 * found it. return the times and remove the entry from the table. 1045 */ 1046 *ppt = pt->fow; 1047 *mtime = pt->mtime; 1048 *atime = pt->atime; 1049 free((char *)pt->name); 1050 free((char *)pt); 1051 return(0); 1052 } 1053 1054 /* 1055 * directory access mode and time storage routines (for directories CREATED 1056 * by pax). 1057 * 1058 * Pax requires that extracted directories, by default, have their access/mod 1059 * times and permissions set to the values specified in the archive. During the 1060 * actions of extracting (and creating the destination subtree during -rw copy) 1061 * directories extracted may be modified after being created. Even worse is 1062 * that these directories may have been created with file permissions which 1063 * prohibits any descendants of these directories from being extracted. When 1064 * directories are created by pax, access rights may be added to permit the 1065 * creation of files in their subtree. Every time pax creates a directory, the 1066 * times and file permissions specified by the archive are stored. After all 1067 * files have been extracted (or copied), these directories have their times 1068 * and file modes reset to the stored values. The directory info is restored in 1069 * reverse order as entries were added to the data file from root to leaf. To 1070 * restore atime properly, we must go backwards. The data file consists of 1071 * records with two parts, the file name followed by a DIRDATA trailer. The 1072 * fixed sized trailer contains the size of the name plus the off_t location in 1073 * the file. To restore we work backwards through the file reading the trailer 1074 * then the file name. 1075 */ 1076 1077 /* 1078 * dir_start() 1079 * set up the directory time and file mode storage for directories CREATED 1080 * by pax. 1081 * Return: 1082 * 0 if ok, -1 otherwise 1083 */ 1084 1085 int 1086 dir_start(void) 1087 { 1088 1089 if (dirfd != -1) 1090 return(0); 1091 1092 /* 1093 * unlink the file so it goes away at termination by itself 1094 */ 1095 memcpy(tempbase, _TFILE_BASE, sizeof(_TFILE_BASE)); 1096 if ((dirfd = mkstemp(tempfile)) >= 0) { 1097 unlink(tempfile); 1098 return(0); 1099 } 1100 paxwarn(1, "Unable to create temporary file for directory times: %s", 1101 tempfile); 1102 return(-1); 1103 } 1104 1105 /* 1106 * add_dir() 1107 * add the mode and times for a newly CREATED directory 1108 * name is name of the directory, psb the stat buffer with the data in it, 1109 * frc_mode is a flag that says whether to force the setting of the mode 1110 * (ignoring the user set values for preserving file mode). Frc_mode is 1111 * for the case where we created a file and found that the resulting 1112 * directory was not writeable and the user asked for file modes to NOT 1113 * be preserved. (we have to preserve what was created by default, so we 1114 * have to force the setting at the end. this is stated explicitly in the 1115 * pax spec) 1116 */ 1117 1118 void 1119 add_dir(char *name, int nlen, struct stat *psb, int frc_mode) 1120 { 1121 DIRDATA dblk; 1122 1123 if (dirfd < 0) 1124 return; 1125 1126 /* 1127 * get current position (where file name will start) so we can store it 1128 * in the trailer 1129 */ 1130 if ((dblk.npos = lseek(dirfd, 0L, SEEK_CUR)) < 0) { 1131 paxwarn(1,"Unable to store mode and times for directory: %s",name); 1132 return; 1133 } 1134 1135 /* 1136 * write the file name followed by the trailer 1137 */ 1138 dblk.nlen = nlen + 1; 1139 dblk.mode = psb->st_mode & 0xffff; 1140 dblk.mtime = psb->st_mtime; 1141 dblk.atime = psb->st_atime; 1142 dblk.frc_mode = frc_mode; 1143 if ((write(dirfd, name, dblk.nlen) == dblk.nlen) && 1144 (write(dirfd, (char *)&dblk, sizeof(dblk)) == sizeof(dblk))) { 1145 ++dircnt; 1146 return; 1147 } 1148 1149 paxwarn(1,"Unable to store mode and times for created directory: %s",name); 1150 return; 1151 } 1152 1153 /* 1154 * proc_dir() 1155 * process all file modes and times stored for directories CREATED 1156 * by pax 1157 */ 1158 1159 void 1160 proc_dir(void) 1161 { 1162 char name[PAXPATHLEN+1]; 1163 DIRDATA dblk; 1164 u_long cnt; 1165 1166 if (dirfd < 0) 1167 return; 1168 /* 1169 * read backwards through the file and process each directory 1170 */ 1171 for (cnt = 0; cnt < dircnt; ++cnt) { 1172 /* 1173 * read the trailer, then the file name, if this fails 1174 * just give up. 1175 */ 1176 if (lseek(dirfd, -((off_t)sizeof(dblk)), SEEK_CUR) < 0) 1177 break; 1178 if (read(dirfd,(char *)&dblk, sizeof(dblk)) != sizeof(dblk)) 1179 break; 1180 if (lseek(dirfd, dblk.npos, SEEK_SET) < 0) 1181 break; 1182 if (read(dirfd, name, dblk.nlen) != dblk.nlen) 1183 break; 1184 if (lseek(dirfd, dblk.npos, SEEK_SET) < 0) 1185 break; 1186 1187 /* 1188 * frc_mode set, make sure we set the file modes even if 1189 * the user didn't ask for it (see file_subs.c for more info) 1190 */ 1191 if (pmode || dblk.frc_mode) 1192 set_pmode(name, dblk.mode); 1193 if (patime || pmtime) 1194 set_ftime(name, dblk.mtime, dblk.atime, 0); 1195 } 1196 1197 close(dirfd); 1198 dirfd = -1; 1199 if (cnt != dircnt) 1200 paxwarn(1,"Unable to set mode and times for created directories"); 1201 return; 1202 } 1203 1204 /* 1205 * database independent routines 1206 */ 1207 1208 /* 1209 * st_hash() 1210 * hashes filenames to a u_int for hashing into a table. Looks at the tail 1211 * end of file, as this provides far better distribution than any other 1212 * part of the name. For performance reasons we only care about the last 1213 * MAXKEYLEN chars (should be at LEAST large enough to pick off the file 1214 * name). Was tested on 500,000 name file tree traversal from the root 1215 * and gave almost a perfectly uniform distribution of keys when used with 1216 * prime sized tables (MAXKEYLEN was 128 in test). Hashes (sizeof int) 1217 * chars at a time and pads with 0 for last addition. 1218 * Return: 1219 * the hash value of the string MOD (%) the table size. 1220 */ 1221 1222 u_int 1223 st_hash(char *name, int len, int tabsz) 1224 { 1225 char *pt; 1226 char *dest; 1227 char *end; 1228 int i; 1229 u_int key = 0; 1230 int steps; 1231 int res; 1232 u_int val; 1233 1234 /* 1235 * only look at the tail up to MAXKEYLEN, we do not need to waste 1236 * time here (remember these are pathnames, the tail is what will 1237 * spread out the keys) 1238 */ 1239 if (len > MAXKEYLEN) { 1240 pt = &(name[len - MAXKEYLEN]); 1241 len = MAXKEYLEN; 1242 } else 1243 pt = name; 1244 1245 /* 1246 * calculate the number of u_int size steps in the string and if 1247 * there is a runt to deal with 1248 */ 1249 steps = len/sizeof(u_int); 1250 res = len % sizeof(u_int); 1251 1252 /* 1253 * add up the value of the string in unsigned integer sized pieces 1254 * too bad we cannot have unsigned int aligned strings, then we 1255 * could avoid the expensive copy. 1256 */ 1257 for (i = 0; i < steps; ++i) { 1258 end = pt + sizeof(u_int); 1259 dest = (char *)&val; 1260 while (pt < end) 1261 *dest++ = *pt++; 1262 key += val; 1263 } 1264 1265 /* 1266 * add in the runt padded with zero to the right 1267 */ 1268 if (res) { 1269 val = 0; 1270 end = pt + res; 1271 dest = (char *)&val; 1272 while (pt < end) 1273 *dest++ = *pt++; 1274 key += val; 1275 } 1276 1277 /* 1278 * return the result mod the table size 1279 */ 1280 return(key % tabsz); 1281 } 1282