xref: /freebsd/bin/pax/tables.c (revision e043f37205ffbde5627ff299ad25cd532f2956f0)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
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. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  */
35 
36 #include <sys/types.h>
37 #include <sys/time.h>
38 #include <sys/stat.h>
39 #include <sys/fcntl.h>
40 #include <errno.h>
41 #include <stdio.h>
42 #include <stdlib.h>
43 #include <string.h>
44 #include <unistd.h>
45 #include "pax.h"
46 #include "tables.h"
47 #include "extern.h"
48 
49 /*
50  * Routines for controlling the contents of all the different databases pax
51  * keeps. Tables are dynamically created only when they are needed. The
52  * goal was speed and the ability to work with HUGE archives. The databases
53  * were kept simple, but do have complex rules for when the contents change.
54  * As of this writing, the POSIX library functions were more complex than
55  * needed for this application (pax databases have very short lifetimes and
56  * do not survive after pax is finished). Pax is required to handle very
57  * large archives. These database routines carefully combine memory usage and
58  * temporary file storage in ways which will not significantly impact runtime
59  * performance while allowing the largest possible archives to be handled.
60  * Trying to force the fit to the POSIX database routines was not considered
61  * time well spent.
62  */
63 
64 static HRDLNK **ltab = NULL;	/* hard link table for detecting hard links */
65 static FTM **ftab = NULL;	/* file time table for updating arch */
66 static NAMT **ntab = NULL;	/* interactive rename storage table */
67 static DEVT **dtab = NULL;	/* device/inode mapping tables */
68 static ATDIR **atab = NULL;	/* file tree directory time reset table */
69 static int dirfd = -1;		/* storage for setting created dir time/mode */
70 static u_long dircnt;		/* entries in dir time/mode storage */
71 static int ffd = -1;		/* tmp file for file time table name storage */
72 
73 static DEVT *chk_dev(dev_t, int);
74 
75 /*
76  * hard link table routines
77  *
78  * The hard link table tries to detect hard links to files using the device and
79  * inode values. We do this when writing an archive, so we can tell the format
80  * write routine that this file is a hard link to another file. The format
81  * write routine then can store this file in whatever way it wants (as a hard
82  * link if the format supports that like tar, or ignore this info like cpio).
83  * (Actually a field in the format driver table tells us if the format wants
84  * hard link info. if not, we do not waste time looking for them). We also use
85  * the same table when reading an archive. In that situation, this table is
86  * used by the format read routine to detect hard links from stored dev and
87  * inode numbers (like cpio). This will allow pax to create a link when one
88  * can be detected by the archive format.
89  */
90 
91 /*
92  * lnk_start
93  *	Creates the hard link table.
94  * Return:
95  *	0 if created, -1 if failure
96  */
97 
98 int
lnk_start(void)99 lnk_start(void)
100 {
101 	if (ltab != NULL)
102 		return(0);
103  	if ((ltab = (HRDLNK **)calloc(L_TAB_SZ, sizeof(HRDLNK *))) == NULL) {
104 		paxwarn(1, "Cannot allocate memory for hard link table");
105 		return(-1);
106 	}
107 	return(0);
108 }
109 
110 /*
111  * chk_lnk()
112  *	Looks up entry in hard link hash table. If found, it copies the name
113  *	of the file it is linked to (we already saw that file) into ln_name.
114  *	lnkcnt is decremented and if goes to 1 the node is deleted from the
115  *	database. (We have seen all the links to this file). If not found,
116  *	we add the file to the database if it has the potential for having
117  *	hard links to other files we may process (it has a link count > 1)
118  * Return:
119  *	if found returns 1; if not found returns 0; -1 on error
120  */
121 
122 int
chk_lnk(ARCHD * arcn)123 chk_lnk(ARCHD *arcn)
124 {
125 	HRDLNK *pt;
126 	HRDLNK **ppt;
127 	u_int indx;
128 
129 	if (ltab == NULL)
130 		return(-1);
131 	/*
132 	 * ignore those nodes that cannot have hard links
133 	 */
134 	if ((arcn->type == PAX_DIR) || (arcn->sb.st_nlink <= 1))
135 		return(0);
136 
137 	/*
138 	 * hash inode number and look for this file
139 	 */
140 	indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ;
141 	if ((pt = ltab[indx]) != NULL) {
142 		/*
143 		 * it's hash chain in not empty, walk down looking for it
144 		 */
145 		ppt = &(ltab[indx]);
146 		while (pt != NULL) {
147 			if ((pt->ino == arcn->sb.st_ino) &&
148 			    (pt->dev == arcn->sb.st_dev))
149 				break;
150 			ppt = &(pt->fow);
151 			pt = pt->fow;
152 		}
153 
154 		if (pt != NULL) {
155 			/*
156 			 * found a link. set the node type and copy in the
157 			 * name of the file it is to link to. we need to
158 			 * handle hardlinks to regular files differently than
159 			 * other links.
160 			 */
161 			arcn->ln_nlen = l_strncpy(arcn->ln_name, pt->name,
162 				sizeof(arcn->ln_name) - 1);
163 			arcn->ln_name[arcn->ln_nlen] = '\0';
164 			if (arcn->type == PAX_REG)
165 				arcn->type = PAX_HRG;
166 			else
167 				arcn->type = PAX_HLK;
168 
169 			/*
170 			 * if we have found all the links to this file, remove
171 			 * it from the database
172 			 */
173 			if (--pt->nlink <= 1) {
174 				*ppt = pt->fow;
175 				free(pt->name);
176 				free(pt);
177 			}
178 			return(1);
179 		}
180 	}
181 
182 	/*
183 	 * we never saw this file before. It has links so we add it to the
184 	 * front of this hash chain
185 	 */
186 	if ((pt = (HRDLNK *)malloc(sizeof(HRDLNK))) != NULL) {
187 		if ((pt->name = strdup(arcn->name)) != NULL) {
188 			pt->dev = arcn->sb.st_dev;
189 			pt->ino = arcn->sb.st_ino;
190 			pt->nlink = arcn->sb.st_nlink;
191 			pt->fow = ltab[indx];
192 			ltab[indx] = pt;
193 			return(0);
194 		}
195 		free(pt);
196 	}
197 
198 	paxwarn(1, "Hard link table out of memory");
199 	return(-1);
200 }
201 
202 /*
203  * purg_lnk
204  *	remove reference for a file that we may have added to the data base as
205  *	a potential source for hard links. We ended up not using the file, so
206  *	we do not want to accidentally point another file at it later on.
207  */
208 
209 void
purg_lnk(ARCHD * arcn)210 purg_lnk(ARCHD *arcn)
211 {
212 	HRDLNK *pt;
213 	HRDLNK **ppt;
214 	u_int indx;
215 
216 	if (ltab == NULL)
217 		return;
218 	/*
219 	 * do not bother to look if it could not be in the database
220 	 */
221 	if ((arcn->sb.st_nlink <= 1) || (arcn->type == PAX_DIR) ||
222 	    (arcn->type == PAX_HLK) || (arcn->type == PAX_HRG))
223 		return;
224 
225 	/*
226 	 * find the hash chain for this inode value, if empty return
227 	 */
228 	indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ;
229 	if ((pt = ltab[indx]) == NULL)
230 		return;
231 
232 	/*
233 	 * walk down the list looking for the inode/dev pair, unlink and
234 	 * free if found
235 	 */
236 	ppt = &(ltab[indx]);
237 	while (pt != NULL) {
238 		if ((pt->ino == arcn->sb.st_ino) &&
239 		    (pt->dev == arcn->sb.st_dev))
240 			break;
241 		ppt = &(pt->fow);
242 		pt = pt->fow;
243 	}
244 	if (pt == NULL)
245 		return;
246 
247 	/*
248 	 * remove and free it
249 	 */
250 	*ppt = pt->fow;
251 	free(pt->name);
252 	free(pt);
253 }
254 
255 /*
256  * lnk_end()
257  *	Pull apart an existing link table so we can reuse it. We do this between
258  *	read and write phases of append with update. (The format may have
259  *	used the link table, and we need to start with a fresh table for the
260  *	write phase).
261  */
262 
263 void
lnk_end(void)264 lnk_end(void)
265 {
266 	int i;
267 	HRDLNK *pt;
268 	HRDLNK *ppt;
269 
270 	if (ltab == NULL)
271 		return;
272 
273 	for (i = 0; i < L_TAB_SZ; ++i) {
274 		if (ltab[i] == NULL)
275 			continue;
276 		pt = ltab[i];
277 		ltab[i] = NULL;
278 
279 		/*
280 		 * free up each entry on this chain
281 		 */
282 		while (pt != NULL) {
283 			ppt = pt;
284 			pt = ppt->fow;
285 			free(ppt->name);
286 			free(ppt);
287 		}
288 	}
289 	return;
290 }
291 
292 /*
293  * modification time table routines
294  *
295  * The modification time table keeps track of last modification times for all
296  * files stored in an archive during a write phase when -u is set. We only
297  * add a file to the archive if it is newer than a file with the same name
298  * already stored on the archive (if there is no other file with the same
299  * name on the archive it is added). This applies to writes and appends.
300  * An append with an -u must read the archive and store the modification time
301  * for every file on that archive before starting the write phase. It is clear
302  * that this is one HUGE database. To save memory space, the actual file names
303  * are stored in a scratch file and indexed by an in memory hash table. The
304  * hash table is indexed by hashing the file path. The nodes in the table store
305  * the length of the filename and the lseek offset within the scratch file
306  * where the actual name is stored. Since there are never any deletions from
307  * this table, fragmentation of the scratch file is never an issue. Lookups
308  * seem to not exhibit any locality at all (files in the database are rarely
309  * looked up more than once...). So caching is just a waste of memory. The
310  * only limitation is the amount of scratch file space available to store the
311  * path names.
312  */
313 
314 /*
315  * ftime_start()
316  *	create the file time hash table and open for read/write the scratch
317  *	file. (after created it is unlinked, so when we exit we leave
318  *	no witnesses).
319  * Return:
320  *	0 if the table and file was created ok, -1 otherwise
321  */
322 
323 int
ftime_start(void)324 ftime_start(void)
325 {
326 
327 	if (ftab != NULL)
328 		return(0);
329  	if ((ftab = (FTM **)calloc(F_TAB_SZ, sizeof(FTM *))) == NULL) {
330 		paxwarn(1, "Cannot allocate memory for file time table");
331 		return(-1);
332 	}
333 
334 	/*
335 	 * get random name and create temporary scratch file, unlink name
336 	 * so it will get removed on exit
337 	 */
338 	memcpy(tempbase, _TFILE_BASE, sizeof(_TFILE_BASE));
339 	if ((ffd = mkstemp(tempfile)) < 0) {
340 		syswarn(1, errno, "Unable to create temporary file: %s",
341 		    tempfile);
342 		return(-1);
343 	}
344 	(void)unlink(tempfile);
345 
346 	return(0);
347 }
348 
349 /*
350  * chk_ftime()
351  *	looks up entry in file time hash table. If not found, the file is
352  *	added to the hash table and the file named stored in the scratch file.
353  *	If a file with the same name is found, the file times are compared and
354  *	the most recent file time is retained. If the new file was younger (or
355  *	was not in the database) the new file is selected for storage.
356  * Return:
357  *	0 if file should be added to the archive, 1 if it should be skipped,
358  *	-1 on error
359  */
360 
361 int
chk_ftime(ARCHD * arcn)362 chk_ftime(ARCHD *arcn)
363 {
364 	FTM *pt;
365 	int namelen;
366 	u_int indx;
367 	char ckname[PAXPATHLEN+1];
368 
369 	/*
370 	 * no info, go ahead and add to archive
371 	 */
372 	if (ftab == NULL)
373 		return(0);
374 
375 	/*
376 	 * hash the pathname and look up in table
377 	 */
378 	namelen = arcn->nlen;
379 	indx = st_hash(arcn->name, namelen, F_TAB_SZ);
380 	if ((pt = ftab[indx]) != NULL) {
381 		/*
382 		 * the hash chain is not empty, walk down looking for match
383 		 * only read up the path names if the lengths match, speeds
384 		 * up the search a lot
385 		 */
386 		while (pt != NULL) {
387 			if (pt->namelen == namelen) {
388 				/*
389 				 * potential match, have to read the name
390 				 * from the scratch file.
391 				 */
392 				if (lseek(ffd,pt->seek,SEEK_SET) != pt->seek) {
393 					syswarn(1, errno,
394 					    "Failed ftime table seek");
395 					return(-1);
396 				}
397 				if (read(ffd, ckname, namelen) != namelen) {
398 					syswarn(1, errno,
399 					    "Failed ftime table read");
400 					return(-1);
401 				}
402 
403 				/*
404 				 * if the names match, we are done
405 				 */
406 				if (!strncmp(ckname, arcn->name, namelen))
407 					break;
408 			}
409 
410 			/*
411 			 * try the next entry on the chain
412 			 */
413 			pt = pt->fow;
414 		}
415 
416 		if (pt != NULL) {
417 			/*
418 			 * found the file, compare the times, save the newer
419 			 */
420 			if (arcn->sb.st_mtime > pt->mtime) {
421 				/*
422 				 * file is newer
423 				 */
424 				pt->mtime = arcn->sb.st_mtime;
425 				return(0);
426 			}
427 			/*
428 			 * file is older
429 			 */
430 			return(1);
431 		}
432 	}
433 
434 	/*
435 	 * not in table, add it
436 	 */
437 	if ((pt = (FTM *)malloc(sizeof(FTM))) != NULL) {
438 		/*
439 		 * add the name at the end of the scratch file, saving the
440 		 * offset. add the file to the head of the hash chain
441 		 */
442 		if ((pt->seek = lseek(ffd, (off_t)0, SEEK_END)) >= 0) {
443 			if (write(ffd, arcn->name, namelen) == namelen) {
444 				pt->mtime = arcn->sb.st_mtime;
445 				pt->namelen = namelen;
446 				pt->fow = ftab[indx];
447 				ftab[indx] = pt;
448 				return(0);
449 			}
450 			syswarn(1, errno, "Failed write to file time table");
451 		} else
452 			syswarn(1, errno, "Failed seek on file time table");
453 	} else
454 		paxwarn(1, "File time table ran out of memory");
455 
456 	if (pt != NULL)
457 		free(pt);
458 	return(-1);
459 }
460 
461 /*
462  * Interactive rename table routines
463  *
464  * The interactive rename table keeps track of the new names that the user
465  * assigns to files from tty input. Since this map is unique for each file
466  * we must store it in case there is a reference to the file later in archive
467  * (a link). Otherwise we will be unable to find the file we know was
468  * extracted. The remapping of these files is stored in a memory based hash
469  * table (it is assumed since input must come from /dev/tty, it is unlikely to
470  * be a very large table).
471  */
472 
473 /*
474  * name_start()
475  *	create the interactive rename table
476  * Return:
477  *	0 if successful, -1 otherwise
478  */
479 
480 int
name_start(void)481 name_start(void)
482 {
483 	if (ntab != NULL)
484 		return(0);
485  	if ((ntab = (NAMT **)calloc(N_TAB_SZ, sizeof(NAMT *))) == NULL) {
486 		paxwarn(1, "Cannot allocate memory for interactive rename table");
487 		return(-1);
488 	}
489 	return(0);
490 }
491 
492 /*
493  * add_name()
494  *	add the new name to old name mapping just created by the user.
495  *	If an old name mapping is found (there may be duplicate names on an
496  *	archive) only the most recent is kept.
497  * Return:
498  *	0 if added, -1 otherwise
499  */
500 
501 int
add_name(char * oname,int onamelen,char * nname)502 add_name(char *oname, int onamelen, char *nname)
503 {
504 	NAMT *pt;
505 	u_int indx;
506 
507 	if (ntab == NULL) {
508 		/*
509 		 * should never happen
510 		 */
511 		paxwarn(0, "No interactive rename table, links may fail\n");
512 		return(0);
513 	}
514 
515 	/*
516 	 * look to see if we have already mapped this file, if so we
517 	 * will update it
518 	 */
519 	indx = st_hash(oname, onamelen, N_TAB_SZ);
520 	if ((pt = ntab[indx]) != NULL) {
521 		/*
522 		 * look down the has chain for the file
523 		 */
524 		while ((pt != NULL) && (strcmp(oname, pt->oname) != 0))
525 			pt = pt->fow;
526 
527 		if (pt != NULL) {
528 			/*
529 			 * found an old mapping, replace it with the new one
530 			 * the user just input (if it is different)
531 			 */
532 			if (strcmp(nname, pt->nname) == 0)
533 				return(0);
534 
535 			free(pt->nname);
536 			if ((pt->nname = strdup(nname)) == NULL) {
537 				paxwarn(1, "Cannot update rename table");
538 				return(-1);
539 			}
540 			return(0);
541 		}
542 	}
543 
544 	/*
545 	 * this is a new mapping, add it to the table
546 	 */
547 	if ((pt = (NAMT *)malloc(sizeof(NAMT))) != NULL) {
548 		if ((pt->oname = strdup(oname)) != NULL) {
549 			if ((pt->nname = strdup(nname)) != NULL) {
550 				pt->fow = ntab[indx];
551 				ntab[indx] = pt;
552 				return(0);
553 			}
554 			free(pt->oname);
555 		}
556 		free(pt);
557 	}
558 	paxwarn(1, "Interactive rename table out of memory");
559 	return(-1);
560 }
561 
562 /*
563  * sub_name()
564  *	look up a link name to see if it points at a file that has been
565  *	remapped by the user. If found, the link is adjusted to contain the
566  *	new name (oname is the link to name)
567  */
568 
569 void
sub_name(char * oname,int * onamelen,size_t onamesize)570 sub_name(char *oname, int *onamelen, size_t onamesize)
571 {
572 	NAMT *pt;
573 	u_int indx;
574 
575 	if (ntab == NULL)
576 		return;
577 	/*
578 	 * look the name up in the hash table
579 	 */
580 	indx = st_hash(oname, *onamelen, N_TAB_SZ);
581 	if ((pt = ntab[indx]) == NULL)
582 		return;
583 
584 	while (pt != NULL) {
585 		/*
586 		 * walk down the hash chain looking for a match
587 		 */
588 		if (strcmp(oname, pt->oname) == 0) {
589 			/*
590 			 * found it, replace it with the new name
591 			 * and return (we know that oname has enough space)
592 			 */
593 			*onamelen = l_strncpy(oname, pt->nname, onamesize - 1);
594 			oname[*onamelen] = '\0';
595 			return;
596 		}
597 		pt = pt->fow;
598 	}
599 
600 	/*
601 	 * no match, just return
602 	 */
603 	return;
604 }
605 
606 /*
607  * device/inode mapping table routines
608  * (used with formats that store device and inodes fields)
609  *
610  * device/inode mapping tables remap the device field in an archive header. The
611  * device/inode fields are used to determine when files are hard links to each
612  * other. However these values have very little meaning outside of that. This
613  * database is used to solve one of two different problems.
614  *
615  * 1) when files are appended to an archive, while the new files may have hard
616  * links to each other, you cannot determine if they have hard links to any
617  * file already stored on the archive from a prior run of pax. We must assume
618  * that these inode/device pairs are unique only within a SINGLE run of pax
619  * (which adds a set of files to an archive). So we have to make sure the
620  * inode/dev pairs we add each time are always unique. We do this by observing
621  * while the inode field is very dense, the use of the dev field is fairly
622  * sparse. Within each run of pax, we remap any device number of a new archive
623  * member that has a device number used in a prior run and already stored in a
624  * file on the archive. During the read phase of the append, we store the
625  * device numbers used and mark them to not be used by any file during the
626  * write phase. If during write we go to use one of those old device numbers,
627  * we remap it to a new value.
628  *
629  * 2) Often the fields in the archive header used to store these values are
630  * too small to store the entire value. The result is an inode or device value
631  * which can be truncated. This really can foul up an archive. With truncation
632  * we end up creating links between files that are really not links (after
633  * truncation the inodes are the same value). We address that by detecting
634  * truncation and forcing a remap of the device field to split truncated
635  * inodes away from each other. Each truncation creates a pattern of bits that
636  * are removed. We use this pattern of truncated bits to partition the inodes
637  * on a single device to many different devices (each one represented by the
638  * truncated bit pattern). All inodes on the same device that have the same
639  * truncation pattern are mapped to the same new device. Two inodes that
640  * truncate to the same value clearly will always have different truncation
641  * bit patterns, so they will be split from away each other. When we spot
642  * device truncation we remap the device number to a non truncated value.
643  * (for more info see table.h for the data structures involved).
644  */
645 
646 /*
647  * dev_start()
648  *	create the device mapping table
649  * Return:
650  *	0 if successful, -1 otherwise
651  */
652 
653 int
dev_start(void)654 dev_start(void)
655 {
656 	if (dtab != NULL)
657 		return(0);
658  	if ((dtab = (DEVT **)calloc(D_TAB_SZ, sizeof(DEVT *))) == NULL) {
659 		paxwarn(1, "Cannot allocate memory for device mapping table");
660 		return(-1);
661 	}
662 	return(0);
663 }
664 
665 /*
666  * add_dev()
667  *	add a device number to the table. this will force the device to be
668  *	remapped to a new value if it be used during a write phase. This
669  *	function is called during the read phase of an append to prohibit the
670  *	use of any device number already in the archive.
671  * Return:
672  *	0 if added ok, -1 otherwise
673  */
674 
675 int
add_dev(ARCHD * arcn)676 add_dev(ARCHD *arcn)
677 {
678 	if (chk_dev(arcn->sb.st_dev, 1) == NULL)
679 		return(-1);
680 	return(0);
681 }
682 
683 /*
684  * chk_dev()
685  *	check for a device value in the device table. If not found and the add
686  *	flag is set, it is added. This does NOT assign any mapping values, just
687  *	adds the device number as one that need to be remapped. If this device
688  *	is already mapped, just return with a pointer to that entry.
689  * Return:
690  *	pointer to the entry for this device in the device map table. Null
691  *	if the add flag is not set and the device is not in the table (it is
692  *	not been seen yet). If add is set and the device cannot be added, null
693  *	is returned (indicates an error).
694  */
695 
696 static DEVT *
chk_dev(dev_t dev,int add)697 chk_dev(dev_t dev, int add)
698 {
699 	DEVT *pt;
700 	u_int indx;
701 
702 	if (dtab == NULL)
703 		return(NULL);
704 	/*
705 	 * look to see if this device is already in the table
706 	 */
707 	indx = ((unsigned)dev) % D_TAB_SZ;
708 	if ((pt = dtab[indx]) != NULL) {
709 		while ((pt != NULL) && (pt->dev != dev))
710 			pt = pt->fow;
711 
712 		/*
713 		 * found it, return a pointer to it
714 		 */
715 		if (pt != NULL)
716 			return(pt);
717 	}
718 
719 	/*
720 	 * not in table, we add it only if told to as this may just be a check
721 	 * to see if a device number is being used.
722 	 */
723 	if (add == 0)
724 		return(NULL);
725 
726 	/*
727 	 * allocate a node for this device and add it to the front of the hash
728 	 * chain. Note we do not assign remaps values here, so the pt->list
729 	 * list must be NULL.
730 	 */
731 	if ((pt = (DEVT *)malloc(sizeof(DEVT))) == NULL) {
732 		paxwarn(1, "Device map table out of memory");
733 		return(NULL);
734 	}
735 	pt->dev = dev;
736 	pt->list = NULL;
737 	pt->fow = dtab[indx];
738 	dtab[indx] = pt;
739 	return(pt);
740 }
741 /*
742  * map_dev()
743  *	given an inode and device storage mask (the mask has a 1 for each bit
744  *	the archive format is able to store in a header), we check for inode
745  *	and device truncation and remap the device as required. Device mapping
746  *	can also occur when during the read phase of append a device number was
747  *	seen (and was marked as do not use during the write phase). WE ASSUME
748  *	that unsigned longs are the same size or bigger than the fields used
749  *	for ino_t and dev_t. If not the types will have to be changed.
750  * Return:
751  *	0 if all ok, -1 otherwise.
752  */
753 
754 int
map_dev(ARCHD * arcn,u_long dev_mask,u_long ino_mask)755 map_dev(ARCHD *arcn, u_long dev_mask, u_long ino_mask)
756 {
757 	DEVT *pt;
758 	DLIST *dpt;
759 	static dev_t lastdev = 0;	/* next device number to try */
760 	int trc_ino = 0;
761 	int trc_dev = 0;
762 	ino_t trunc_bits = 0;
763 	ino_t nino;
764 
765 	if (dtab == NULL)
766 		return(0);
767 	/*
768 	 * check for device and inode truncation, and extract the truncated
769 	 * bit pattern.
770 	 */
771 	if ((arcn->sb.st_dev & (dev_t)dev_mask) != arcn->sb.st_dev)
772 		++trc_dev;
773 	if ((nino = arcn->sb.st_ino & (ino_t)ino_mask) != arcn->sb.st_ino) {
774 		++trc_ino;
775 		trunc_bits = arcn->sb.st_ino & (ino_t)(~ino_mask);
776 	}
777 
778 	/*
779 	 * see if this device is already being mapped, look up the device
780 	 * then find the truncation bit pattern which applies
781 	 */
782 	if ((pt = chk_dev(arcn->sb.st_dev, 0)) != NULL) {
783 		/*
784 		 * this device is already marked to be remapped
785 		 */
786 		for (dpt = pt->list; dpt != NULL; dpt = dpt->fow)
787 			if (dpt->trunc_bits == trunc_bits)
788 				break;
789 
790 		if (dpt != NULL) {
791 			/*
792 			 * we are being remapped for this device and pattern
793 			 * change the device number to be stored and return
794 			 */
795 			arcn->sb.st_dev = dpt->dev;
796 			arcn->sb.st_ino = nino;
797 			return(0);
798 		}
799 	} else {
800 		/*
801 		 * this device is not being remapped YET. if we do not have any
802 		 * form of truncation, we do not need a remap
803 		 */
804 		if (!trc_ino && !trc_dev)
805 			return(0);
806 
807 		/*
808 		 * we have truncation, have to add this as a device to remap
809 		 */
810 		if ((pt = chk_dev(arcn->sb.st_dev, 1)) == NULL)
811 			goto bad;
812 
813 		/*
814 		 * if we just have a truncated inode, we have to make sure that
815 		 * all future inodes that do not truncate (they have the
816 		 * truncation pattern of all 0's) continue to map to the same
817 		 * device number. We probably have already written inodes with
818 		 * this device number to the archive with the truncation
819 		 * pattern of all 0's. So we add the mapping for all 0's to the
820 		 * same device number.
821 		 */
822 		if (!trc_dev && (trunc_bits != 0)) {
823 			if ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL)
824 				goto bad;
825 			dpt->trunc_bits = 0;
826 			dpt->dev = arcn->sb.st_dev;
827 			dpt->fow = pt->list;
828 			pt->list = dpt;
829 		}
830 	}
831 
832 	/*
833 	 * look for a device number not being used. We must watch for wrap
834 	 * around on lastdev (so we do not get stuck looking forever!)
835 	 */
836 	while (++lastdev > 0) {
837 		if (chk_dev(lastdev, 0) != NULL)
838 			continue;
839 		/*
840 		 * found an unused value. If we have reached truncation point
841 		 * for this format we are hosed, so we give up. Otherwise we
842 		 * mark it as being used.
843 		 */
844 		if (((lastdev & ((dev_t)dev_mask)) != lastdev) ||
845 		    (chk_dev(lastdev, 1) == NULL))
846 			goto bad;
847 		break;
848 	}
849 
850 	if ((lastdev <= 0) || ((dpt = (DLIST *)malloc(sizeof(DLIST))) == NULL))
851 		goto bad;
852 
853 	/*
854 	 * got a new device number, store it under this truncation pattern.
855 	 * change the device number this file is being stored with.
856 	 */
857 	dpt->trunc_bits = trunc_bits;
858 	dpt->dev = lastdev;
859 	dpt->fow = pt->list;
860 	pt->list = dpt;
861 	arcn->sb.st_dev = lastdev;
862 	arcn->sb.st_ino = nino;
863 	return(0);
864 
865     bad:
866 	paxwarn(1, "Unable to fix truncated inode/device field when storing %s",
867 	    arcn->name);
868 	paxwarn(0, "Archive may create improper hard links when extracted");
869 	return(0);
870 }
871 
872 /*
873  * directory access/mod time reset table routines (for directories READ by pax)
874  *
875  * The pax -t flag requires that access times of archive files be the same
876  * before being read by pax. For regular files, access time is restored after
877  * the file has been copied. This database provides the same functionality for
878  * directories read during file tree traversal. Restoring directory access time
879  * is more complex than files since directories may be read several times until
880  * all the descendants in their subtree are visited by fts. Directory access
881  * and modification times are stored during the fts pre-order visit (done
882  * before any descendants in the subtree are visited) and restored after the
883  * fts post-order visit (after all the descendants have been visited). In the
884  * case of premature exit from a subtree (like from the effects of -n), any
885  * directory entries left in this database are reset during final cleanup
886  * operations of pax. Entries are hashed by inode number for fast lookup.
887  */
888 
889 /*
890  * atdir_start()
891  *	create the directory access time database for directories READ by pax.
892  * Return:
893  *	0 is created ok, -1 otherwise.
894  */
895 
896 int
atdir_start(void)897 atdir_start(void)
898 {
899 	if (atab != NULL)
900 		return(0);
901  	if ((atab = (ATDIR **)calloc(A_TAB_SZ, sizeof(ATDIR *))) == NULL) {
902 		paxwarn(1,"Cannot allocate space for directory access time table");
903 		return(-1);
904 	}
905 	return(0);
906 }
907 
908 
909 /*
910  * atdir_end()
911  *	walk through the directory access time table and reset the access time
912  *	of any directory who still has an entry left in the database. These
913  *	entries are for directories READ by pax
914  */
915 
916 void
atdir_end(void)917 atdir_end(void)
918 {
919 	ATDIR *pt;
920 	int i;
921 
922 	if (atab == NULL)
923 		return;
924 	/*
925 	 * for each non-empty hash table entry reset all the directories
926 	 * chained there.
927 	 */
928 	for (i = 0; i < A_TAB_SZ; ++i) {
929 		if ((pt = atab[i]) == NULL)
930 			continue;
931 		/*
932 		 * remember to force the times, set_ftime() looks at pmtime
933 		 * and patime, which only applies to things CREATED by pax,
934 		 * not read by pax. Read time reset is controlled by -t.
935 		 */
936 		for (; pt != NULL; pt = pt->fow)
937 			set_ftime(pt->name, pt->mtime, pt->atime, 1);
938 	}
939 }
940 
941 /*
942  * add_atdir()
943  *	add a directory to the directory access time table. Table is hashed
944  *	and chained by inode number. This is for directories READ by pax
945  */
946 
947 void
add_atdir(char * fname,dev_t dev,ino_t ino,time_t mtime,time_t atime)948 add_atdir(char *fname, dev_t dev, ino_t ino, time_t mtime, time_t atime)
949 {
950 	ATDIR *pt;
951 	u_int indx;
952 
953 	if (atab == NULL)
954 		return;
955 
956 	/*
957 	 * make sure this directory is not already in the table, if so just
958 	 * return (the older entry always has the correct time). The only
959 	 * way this will happen is when the same subtree can be traversed by
960 	 * different args to pax and the -n option is aborting fts out of a
961 	 * subtree before all the post-order visits have been made.
962 	 */
963 	indx = ((unsigned)ino) % A_TAB_SZ;
964 	if ((pt = atab[indx]) != NULL) {
965 		while (pt != NULL) {
966 			if ((pt->ino == ino) && (pt->dev == dev))
967 				break;
968 			pt = pt->fow;
969 		}
970 
971 		/*
972 		 * oops, already there. Leave it alone.
973 		 */
974 		if (pt != NULL)
975 			return;
976 	}
977 
978 	/*
979 	 * add it to the front of the hash chain
980 	 */
981 	if ((pt = (ATDIR *)malloc(sizeof(ATDIR))) != NULL) {
982 		if ((pt->name = strdup(fname)) != NULL) {
983 			pt->dev = dev;
984 			pt->ino = ino;
985 			pt->mtime = mtime;
986 			pt->atime = atime;
987 			pt->fow = atab[indx];
988 			atab[indx] = pt;
989 			return;
990 		}
991 		free(pt);
992 	}
993 
994 	paxwarn(1, "Directory access time reset table ran out of memory");
995 	return;
996 }
997 
998 /*
999  * get_atdir()
1000  *	look up a directory by inode and device number to obtain the access
1001  *	and modification time you want to set to. If found, the modification
1002  *	and access time parameters are set and the entry is removed from the
1003  *	table (as it is no longer needed). These are for directories READ by
1004  *	pax
1005  * Return:
1006  *	0 if found, -1 if not found.
1007  */
1008 
1009 int
get_atdir(dev_t dev,ino_t ino,time_t * mtime,time_t * atime)1010 get_atdir(dev_t dev, ino_t ino, time_t *mtime, time_t *atime)
1011 {
1012 	ATDIR *pt;
1013 	ATDIR **ppt;
1014 	u_int indx;
1015 
1016 	if (atab == NULL)
1017 		return(-1);
1018 	/*
1019 	 * hash by inode and search the chain for an inode and device match
1020 	 */
1021 	indx = ((unsigned)ino) % A_TAB_SZ;
1022 	if ((pt = atab[indx]) == NULL)
1023 		return(-1);
1024 
1025 	ppt = &(atab[indx]);
1026 	while (pt != NULL) {
1027 		if ((pt->ino == ino) && (pt->dev == dev))
1028 			break;
1029 		/*
1030 		 * no match, go to next one
1031 		 */
1032 		ppt = &(pt->fow);
1033 		pt = pt->fow;
1034 	}
1035 
1036 	/*
1037 	 * return if we did not find it.
1038 	 */
1039 	if (pt == NULL)
1040 		return(-1);
1041 
1042 	/*
1043 	 * found it. return the times and remove the entry from the table.
1044 	 */
1045 	*ppt = pt->fow;
1046 	*mtime = pt->mtime;
1047 	*atime = pt->atime;
1048 	free(pt->name);
1049 	free(pt);
1050 	return(0);
1051 }
1052 
1053 /*
1054  * directory access mode and time storage routines (for directories CREATED
1055  * by pax).
1056  *
1057  * Pax requires that extracted directories, by default, have their access/mod
1058  * times and permissions set to the values specified in the archive. During the
1059  * actions of extracting (and creating the destination subtree during -rw copy)
1060  * directories extracted may be modified after being created. Even worse is
1061  * that these directories may have been created with file permissions which
1062  * prohibits any descendants of these directories from being extracted. When
1063  * directories are created by pax, access rights may be added to permit the
1064  * creation of files in their subtree. Every time pax creates a directory, the
1065  * times and file permissions specified by the archive are stored. After all
1066  * files have been extracted (or copied), these directories have their times
1067  * and file modes reset to the stored values. The directory info is restored in
1068  * reverse order as entries were added to the data file from root to leaf. To
1069  * restore atime properly, we must go backwards. The data file consists of
1070  * records with two parts, the file name followed by a DIRDATA trailer. The
1071  * fixed sized trailer contains the size of the name plus the off_t location in
1072  * the file. To restore we work backwards through the file reading the trailer
1073  * then the file name.
1074  */
1075 
1076 /*
1077  * dir_start()
1078  *	set up the directory time and file mode storage for directories CREATED
1079  *	by pax.
1080  * Return:
1081  *	0 if ok, -1 otherwise
1082  */
1083 
1084 int
dir_start(void)1085 dir_start(void)
1086 {
1087 
1088 	if (dirfd != -1)
1089 		return(0);
1090 
1091 	/*
1092 	 * unlink the file so it goes away at termination by itself
1093 	 */
1094 	memcpy(tempbase, _TFILE_BASE, sizeof(_TFILE_BASE));
1095 	if ((dirfd = mkstemp(tempfile)) >= 0) {
1096 		(void)unlink(tempfile);
1097 		return(0);
1098 	}
1099 	paxwarn(1, "Unable to create temporary file for directory times: %s",
1100 	    tempfile);
1101 	return(-1);
1102 }
1103 
1104 /*
1105  * add_dir()
1106  *	add the mode and times for a newly CREATED directory
1107  *	name is name of the directory, psb the stat buffer with the data in it,
1108  *	frc_mode is a flag that says whether to force the setting of the mode
1109  *	(ignoring the user set values for preserving file mode). Frc_mode is
1110  *	for the case where we created a file and found that the resulting
1111  *	directory was not writeable and the user asked for file modes to NOT
1112  *	be preserved. (we have to preserve what was created by default, so we
1113  *	have to force the setting at the end. this is stated explicitly in the
1114  *	pax spec)
1115  */
1116 
1117 void
add_dir(char * name,int nlen,struct stat * psb,int frc_mode)1118 add_dir(char *name, int nlen, struct stat *psb, int frc_mode)
1119 {
1120 	DIRDATA dblk;
1121 
1122 	if (dirfd < 0)
1123 		return;
1124 
1125 	/*
1126 	 * get current position (where file name will start) so we can store it
1127 	 * in the trailer
1128 	 */
1129 	if ((dblk.npos = lseek(dirfd, 0L, SEEK_CUR)) < 0) {
1130 		paxwarn(1,"Unable to store mode and times for directory: %s",name);
1131 		return;
1132 	}
1133 
1134 	/*
1135 	 * write the file name followed by the trailer
1136 	 */
1137 	dblk.nlen = nlen + 1;
1138 	dblk.mode = psb->st_mode & 0xffff;
1139 	dblk.mtime = psb->st_mtime;
1140 	dblk.atime = psb->st_atime;
1141 	dblk.frc_mode = frc_mode;
1142 	if ((write(dirfd, name, dblk.nlen) == dblk.nlen) &&
1143 	    (write(dirfd, (char *)&dblk, sizeof(dblk)) == sizeof(dblk))) {
1144 		++dircnt;
1145 		return;
1146 	}
1147 
1148 	paxwarn(1,"Unable to store mode and times for created directory: %s",name);
1149 	return;
1150 }
1151 
1152 /*
1153  * proc_dir()
1154  *	process all file modes and times stored for directories CREATED
1155  *	by pax
1156  */
1157 
1158 void
proc_dir(void)1159 proc_dir(void)
1160 {
1161 	char name[PAXPATHLEN+1];
1162 	DIRDATA dblk;
1163 	u_long cnt;
1164 
1165 	if (dirfd < 0)
1166 		return;
1167 	/*
1168 	 * read backwards through the file and process each directory
1169 	 */
1170 	for (cnt = 0; cnt < dircnt; ++cnt) {
1171 		/*
1172 		 * read the trailer, then the file name, if this fails
1173 		 * just give up.
1174 		 */
1175 		if (lseek(dirfd, -((off_t)sizeof(dblk)), SEEK_CUR) < 0)
1176 			break;
1177 		if (read(dirfd,(char *)&dblk, sizeof(dblk)) != sizeof(dblk))
1178 			break;
1179 		if (lseek(dirfd, dblk.npos, SEEK_SET) < 0)
1180 			break;
1181 		if (read(dirfd, name, dblk.nlen) != dblk.nlen)
1182 			break;
1183 		if (lseek(dirfd, dblk.npos, SEEK_SET) < 0)
1184 			break;
1185 
1186 		/*
1187 		 * frc_mode set, make sure we set the file modes even if
1188 		 * the user didn't ask for it (see file_subs.c for more info)
1189 		 */
1190 		if (pmode || dblk.frc_mode)
1191 			set_pmode(name, dblk.mode);
1192 		if (patime || pmtime)
1193 			set_ftime(name, dblk.mtime, dblk.atime, 0);
1194 	}
1195 
1196 	(void)close(dirfd);
1197 	dirfd = -1;
1198 	if (cnt != dircnt)
1199 		paxwarn(1,"Unable to set mode and times for created directories");
1200 	return;
1201 }
1202 
1203 /*
1204  * database independent routines
1205  */
1206 
1207 /*
1208  * st_hash()
1209  *	hashes filenames to a u_int for hashing into a table. Looks at the tail
1210  *	end of file, as this provides far better distribution than any other
1211  *	part of the name. For performance reasons we only care about the last
1212  *	MAXKEYLEN chars (should be at LEAST large enough to pick off the file
1213  *	name). Was tested on 500,000 name file tree traversal from the root
1214  *	and gave almost a perfectly uniform distribution of keys when used with
1215  *	prime sized tables (MAXKEYLEN was 128 in test). Hashes (sizeof int)
1216  *	chars at a time and pads with 0 for last addition.
1217  * Return:
1218  *	the hash value of the string MOD (%) the table size.
1219  */
1220 
1221 u_int
st_hash(char * name,int len,int tabsz)1222 st_hash(char *name, int len, int tabsz)
1223 {
1224 	char *pt;
1225 	char *dest;
1226 	char *end;
1227 	int i;
1228 	u_int key = 0;
1229 	int steps;
1230 	int res;
1231 	u_int val;
1232 
1233 	/*
1234 	 * only look at the tail up to MAXKEYLEN, we do not need to waste
1235 	 * time here (remember these are pathnames, the tail is what will
1236 	 * spread out the keys)
1237 	 */
1238 	if (len > MAXKEYLEN) {
1239 		pt = &(name[len - MAXKEYLEN]);
1240 		len = MAXKEYLEN;
1241 	} else
1242 		pt = name;
1243 
1244 	/*
1245 	 * calculate the number of u_int size steps in the string and if
1246 	 * there is a runt to deal with
1247 	 */
1248 	steps = len/sizeof(u_int);
1249 	res = len % sizeof(u_int);
1250 
1251 	/*
1252 	 * add up the value of the string in unsigned integer sized pieces
1253 	 * too bad we cannot have unsigned int aligned strings, then we
1254 	 * could avoid the expensive copy.
1255 	 */
1256 	for (i = 0; i < steps; ++i) {
1257 		end = pt + sizeof(u_int);
1258 		dest = (char *)&val;
1259 		while (pt < end)
1260 			*dest++ = *pt++;
1261 		key += val;
1262 	}
1263 
1264 	/*
1265 	 * add in the runt padded with zero to the right
1266 	 */
1267 	if (res) {
1268 		val = 0;
1269 		end = pt + res;
1270 		dest = (char *)&val;
1271 		while (pt < end)
1272 			*dest++ = *pt++;
1273 		key += val;
1274 	}
1275 
1276 	/*
1277 	 * return the result mod the table size
1278 	 */
1279 	return(key % tabsz);
1280 }
1281