Bug Summary

File:src/bin/pax/tables.c
Warning:line 1766, column 7
Assigned value is garbage or undefined

Annotated Source Code

Press '?' to see keyboard shortcuts

clang -cc1 -cc1 -triple amd64-unknown-openbsd7.0 -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name tables.c -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -mrelocation-model pic -pic-level 1 -pic-is-pie -mframe-pointer=all -relaxed-aliasing -fno-rounding-math -mconstructor-aliases -munwind-tables -target-cpu x86-64 -target-feature +retpoline-indirect-calls -target-feature +retpoline-indirect-branches -tune-cpu generic -debugger-tuning=gdb -fcoverage-compilation-dir=/usr/src/bin/pax/obj -resource-dir /usr/local/lib/clang/13.0.0 -internal-isystem /usr/local/lib/clang/13.0.0/include -internal-externc-isystem /usr/include -O2 -fdebug-compilation-dir=/usr/src/bin/pax/obj -ferror-limit 19 -fwrapv -D_RET_PROTECTOR -ret-protector -fgnuc-version=4.2.1 -vectorize-loops -vectorize-slp -fno-builtin-malloc -fno-builtin-calloc -fno-builtin-realloc -fno-builtin-valloc -fno-builtin-free -fno-builtin-strdup -fno-builtin-strndup -analyzer-output=html -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /home/ben/Projects/vmm/scan-build/2022-01-12-194120-40624-1 -x c /usr/src/bin/pax/tables.c
1/* $OpenBSD: tables.c,v 1.54 2019/06/28 05:35:34 deraadt Exp $ */
2/* $NetBSD: tables.c,v 1.4 1995/03/21 09:07:45 cgd Exp $ */
3
4/*-
5 * Copyright (c) 1992 Keith Muller.
6 * Copyright (c) 1992, 1993
7 * The Regents of the University of California. All rights reserved.
8 *
9 * This code is derived from software contributed to Berkeley by
10 * Keith Muller of the University of California, San Diego.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 */
36
37#include <sys/types.h>
38#include <sys/stat.h>
39#include <errno(*__errno()).h>
40#include <fcntl.h>
41#include <limits.h>
42#include <signal.h>
43#include <stdio.h>
44#include <stdlib.h>
45#include <string.h>
46#include <unistd.h>
47
48#include "pax.h"
49#include "extern.h"
50
51/*
52 * Routines for controlling the contents of all the different databases pax
53 * keeps. Tables are dynamically created only when they are needed. The
54 * goal was speed and the ability to work with HUGE archives. The databases
55 * were kept simple, but do have complex rules for when the contents change.
56 * As of this writing, the posix library functions were more complex than
57 * needed for this application (pax databases have very short lifetimes and
58 * do not survive after pax is finished). Pax is required to handle very
59 * large archives. These database routines carefully combine memory usage and
60 * temporary file storage in ways which will not significantly impact runtime
61 * performance while allowing the largest possible archives to be handled.
62 * Trying to force the fit to the posix database routines was not considered
63 * time well spent.
64 */
65
66/*
67 * data structures and constants used by the different databases kept by pax
68 */
69
70/*
71 * Hash Table Sizes MUST BE PRIME, if set too small performance suffers.
72 * Probably safe to expect 500000 inodes per tape. Assuming good key
73 * distribution (inodes) chains of under 50 long (worst case) is ok.
74 */
75#define L_TAB_SZ2503 2503 /* hard link hash table size */
76#define F_TAB_SZ50503 50503 /* file time hash table size */
77#define N_TAB_SZ541 541 /* interactive rename hash table */
78#define D_TAB_SZ317 317 /* unique device mapping table */
79#define A_TAB_SZ317 317 /* ftree dir access time reset table */
80#define SL_TAB_SZ317 317 /* escape symlink tables */
81#define MAXKEYLEN64 64 /* max number of chars for hash */
82#define DIRP_SIZE64 64 /* initial size of created dir table */
83
84/*
85 * file hard link structure (hashed by dev/ino and chained) used to find the
86 * hard links in a file system or with some archive formats (cpio)
87 */
88typedef struct hrdlnk {
89 ino_t ino; /* files inode number */
90 char *name; /* name of first file seen with this ino/dev */
91 dev_t dev; /* files device number */
92 u_long nlink; /* expected link count */
93 struct hrdlnk *fow;
94} HRDLNK;
95
96/*
97 * Archive write update file time table (the -u, -C flag), hashed by filename.
98 * Filenames are stored in a scratch file at seek offset into the file. The
99 * file time (mod time) and the file name length (for a quick check) are
100 * stored in a hash table node. We were forced to use a scratch file because
101 * with -u, the mtime for every node in the archive must always be available
102 * to compare against (and this data can get REALLY large with big archives).
103 * By being careful to read only when we have a good chance of a match, the
104 * performance loss is not measurable (and the size of the archive we can
105 * handle is greatly increased).
106 */
107typedef struct ftm {
108 off_t seek; /* location in scratch file */
109 struct timespec mtim; /* files last modification time */
110 struct ftm *fow;
111 int namelen; /* file name length */
112} FTM;
113
114/*
115 * Interactive rename table (-i flag), hashed by orig filename.
116 * We assume this will not be a large table as this mapping data can only be
117 * obtained through interactive input by the user. Nobody is going to type in
118 * changes for 500000 files? We use chaining to resolve collisions.
119 */
120
121typedef struct namt {
122 char *oname; /* old name */
123 char *nname; /* new name typed in by the user */
124 struct namt *fow;
125} NAMT;
126
127/*
128 * Unique device mapping tables. Some protocols (e.g. cpio) require that the
129 * <c_dev,c_ino> pair will uniquely identify a file in an archive unless they
130 * are links to the same file. Appending to archives can break this. For those
131 * protocols that have this requirement we map c_dev to a unique value not seen
132 * in the archive when we append. We also try to handle inode truncation with
133 * this table. (When the inode field in the archive header are too small, we
134 * remap the dev on writes to remove accidental collisions).
135 *
136 * The list is hashed by device number using chain collision resolution. Off of
137 * each DEVT are linked the various remaps for this device based on those bits
138 * in the inode which were truncated. For example if we are just remapping to
139 * avoid a device number during an update append, off the DEVT we would have
140 * only a single DLIST that has a truncation id of 0 (no inode bits were
141 * stripped for this device so far). When we spot inode truncation we create
142 * a new mapping based on the set of bits in the inode which were stripped off.
143 * so if the top four bits of the inode are stripped and they have a pattern of
144 * 0110...... (where . are those bits not truncated) we would have a mapping
145 * assigned for all inodes that has the same 0110.... pattern (with this dev
146 * number of course). This keeps the mapping sparse and should be able to store
147 * close to the limit of files which can be represented by the optimal
148 * combination of dev and inode bits, and without creating a fouled up archive.
149 * Note we also remap truncated devs in the same way (an exercise for the
150 * dedicated reader; always wanted to say that...:)
151 */
152
153typedef struct devt {
154 dev_t dev; /* the orig device number we now have to map */
155 struct devt *fow; /* new device map list */
156 struct dlist *list; /* map list based on inode truncation bits */
157} DEVT;
158
159typedef struct dlist {
160 ino_t trunc_bits; /* truncation pattern for a specific map */
161 dev_t dev; /* the new device id we use */
162 struct dlist *fow;
163} DLIST;
164
165/*
166 * ftree directory access time reset table. When we are done with a
167 * subtree we reset the access and mod time of the directory when the tflag is
168 * set. Not really explicitly specified in the pax spec, but easy and fast to
169 * do (and this may have even been intended in the spec, it is not clear).
170 * table is hashed by inode with chaining.
171 */
172
173typedef struct atdir {
174 struct file_times ft;
175 struct atdir *fow;
176} ATDIR;
177
178/*
179 * created directory time and mode storage entry. After pax is finished during
180 * extraction or copy, we must reset directory access modes and times that
181 * may have been modified after creation (they no longer have the specified
182 * times and/or modes). We must reset time in the reverse order of creation,
183 * because entries are added from the top of the file tree to the bottom.
184 * We MUST reset times from leaf to root (it will not work the other
185 * direction).
186 */
187
188typedef struct dirdata {
189 struct file_times ft;
190 u_int16_t mode; /* file mode to restore */
191 u_int16_t frc_mode; /* do we force mode settings? */
192} DIRDATA;
193
194static HRDLNK **ltab = NULL((void *)0); /* hard link table for detecting hard links */
195static FTM **ftab = NULL((void *)0); /* file time table for updating arch */
196static NAMT **ntab = NULL((void *)0); /* interactive rename storage table */
197#ifndef NOCPIO
198static DEVT **dtab = NULL((void *)0); /* device/inode mapping tables */
199#endif
200static ATDIR **atab = NULL((void *)0); /* file tree directory time reset table */
201static DIRDATA *dirp = NULL((void *)0); /* storage for setting created dir time/mode */
202static size_t dirsize; /* size of dirp table */
203static size_t dircnt = 0; /* entries in dir time/mode storage */
204static int ffd = -1; /* tmp file for file time table name storage */
205
206/*
207 * hard link table routines
208 *
209 * The hard link table tries to detect hard links to files using the device and
210 * inode values. We do this when writing an archive, so we can tell the format
211 * write routine that this file is a hard link to another file. The format
212 * write routine then can store this file in whatever way it wants (as a hard
213 * link if the format supports that like tar, or ignore this info like cpio).
214 * (Actually a field in the format driver table tells us if the format wants
215 * hard link info. if not, we do not waste time looking for them). We also use
216 * the same table when reading an archive. In that situation, this table is
217 * used by the format read routine to detect hard links from stored dev and
218 * inode numbers (like cpio). This will allow pax to create a link when one
219 * can be detected by the archive format.
220 */
221
222/*
223 * lnk_start
224 * Creates the hard link table.
225 * Return:
226 * 0 if created, -1 if failure
227 */
228
229int
230lnk_start(void)
231{
232 if (ltab != NULL((void *)0))
233 return(0);
234 if ((ltab = calloc(L_TAB_SZ2503, sizeof(HRDLNK *))) == NULL((void *)0)) {
235 paxwarn(1, "Cannot allocate memory for hard link table");
236 return(-1);
237 }
238 return(0);
239}
240
241/*
242 * chk_lnk()
243 * Looks up entry in hard link hash table. If found, it copies the name
244 * of the file it is linked to (we already saw that file) into ln_name.
245 * lnkcnt is decremented and if goes to 1 the node is deleted from the
246 * database. (We have seen all the links to this file). If not found,
247 * we add the file to the database if it has the potential for having
248 * hard links to other files we may process (it has a link count > 1)
249 * Return:
250 * if found returns 1; if not found returns 0; -1 on error
251 */
252
253int
254chk_lnk(ARCHD *arcn)
255{
256 HRDLNK *pt;
257 HRDLNK **ppt;
258 u_int indx;
259
260 if (ltab == NULL((void *)0))
261 return(-1);
262 /*
263 * ignore those nodes that cannot have hard links
264 */
265 if ((arcn->type == PAX_DIR1) || (arcn->sb.st_nlink <= 1))
266 return(0);
267
268 /*
269 * hash inode number and look for this file
270 */
271 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ2503;
272 if ((pt = ltab[indx]) != NULL((void *)0)) {
273 /*
274 * its hash chain in not empty, walk down looking for it
275 */
276 ppt = &(ltab[indx]);
277 while (pt != NULL((void *)0)) {
278 if ((pt->ino == arcn->sb.st_ino) &&
279 (pt->dev == arcn->sb.st_dev))
280 break;
281 ppt = &(pt->fow);
282 pt = pt->fow;
283 }
284
285 if (pt != NULL((void *)0)) {
286 /*
287 * found a link. set the node type and copy in the
288 * name of the file it is to link to. we need to
289 * handle hardlinks to regular files differently than
290 * other links.
291 */
292 arcn->ln_nlen = strlcpy(arcn->ln_name, pt->name,
293 sizeof(arcn->ln_name));
294 /* XXX truncate? */
295 if ((size_t)arcn->nlen >= sizeof(arcn->name))
296 arcn->nlen = sizeof(arcn->name) - 1;
297 if (arcn->type == PAX_REG4)
298 arcn->type = PAX_HRG9;
299 else
300 arcn->type = PAX_HLK8;
301
302 /*
303 * if we have found all the links to this file, remove
304 * it from the database
305 */
306 if (--pt->nlink <= 1) {
307 *ppt = pt->fow;
308 free(pt->name);
309 free(pt);
310 }
311 return(1);
312 }
313 }
314
315 /*
316 * we never saw this file before. It has links so we add it to the
317 * front of this hash chain
318 */
319 if ((pt = malloc(sizeof(HRDLNK))) != NULL((void *)0)) {
320 if ((pt->name = strdup(arcn->name)) != NULL((void *)0)) {
321 pt->dev = arcn->sb.st_dev;
322 pt->ino = arcn->sb.st_ino;
323 pt->nlink = arcn->sb.st_nlink;
324 pt->fow = ltab[indx];
325 ltab[indx] = pt;
326 return(0);
327 }
328 free(pt);
329 }
330
331 paxwarn(1, "Hard link table out of memory");
332 return(-1);
333}
334
335/*
336 * purg_lnk
337 * remove reference for a file that we may have added to the data base as
338 * a potential source for hard links. We ended up not using the file, so
339 * we do not want to accidently point another file at it later on.
340 */
341
342void
343purg_lnk(ARCHD *arcn)
344{
345 HRDLNK *pt;
346 HRDLNK **ppt;
347 u_int indx;
348
349 if (ltab == NULL((void *)0))
350 return;
351 /*
352 * do not bother to look if it could not be in the database
353 */
354 if ((arcn->sb.st_nlink <= 1) || (arcn->type == PAX_DIR1) ||
355 PAX_IS_HARDLINK(arcn->type)((arcn->type) == 8 || (arcn->type) == 9))
356 return;
357
358 /*
359 * find the hash chain for this inode value, if empty return
360 */
361 indx = ((unsigned)arcn->sb.st_ino) % L_TAB_SZ2503;
362 if ((pt = ltab[indx]) == NULL((void *)0))
363 return;
364
365 /*
366 * walk down the list looking for the inode/dev pair, unlink and
367 * free if found
368 */
369 ppt = &(ltab[indx]);
370 while (pt != NULL((void *)0)) {
371 if ((pt->ino == arcn->sb.st_ino) &&
372 (pt->dev == arcn->sb.st_dev))
373 break;
374 ppt = &(pt->fow);
375 pt = pt->fow;
376 }
377 if (pt == NULL((void *)0))
378 return;
379
380 /*
381 * remove and free it
382 */
383 *ppt = pt->fow;
384 free(pt->name);
385 free(pt);
386}
387
388/*
389 * lnk_end()
390 * pull apart a existing link table so we can reuse it. We do this between
391 * read and write phases of append with update. (The format may have
392 * used the link table, and we need to start with a fresh table for the
393 * write phase
394 */
395
396void
397lnk_end(void)
398{
399 int i;
400 HRDLNK *pt;
401 HRDLNK *ppt;
402
403 if (ltab == NULL((void *)0))
404 return;
405
406 for (i = 0; i < L_TAB_SZ2503; ++i) {
407 if (ltab[i] == NULL((void *)0))
408 continue;
409 pt = ltab[i];
410 ltab[i] = NULL((void *)0);
411
412 /*
413 * free up each entry on this chain
414 */
415 while (pt != NULL((void *)0)) {
416 ppt = pt;
417 pt = ppt->fow;
418 free(ppt->name);
419 free(ppt);
420 }
421 }
422}
423
424/*
425 * modification time table routines
426 *
427 * The modification time table keeps track of last modification times for all
428 * files stored in an archive during a write phase when -u is set. We only
429 * add a file to the archive if it is newer than a file with the same name
430 * already stored on the archive (if there is no other file with the same
431 * name on the archive it is added). This applies to writes and appends.
432 * An append with an -u must read the archive and store the modification time
433 * for every file on that archive before starting the write phase. It is clear
434 * that this is one HUGE database. To save memory space, the actual file names
435 * are stored in a scratch file and indexed by an in-memory hash table. The
436 * hash table is indexed by hashing the file path. The nodes in the table store
437 * the length of the filename and the lseek offset within the scratch file
438 * where the actual name is stored. Since there are never any deletions from
439 * this table, fragmentation of the scratch file is never a issue. Lookups
440 * seem to not exhibit any locality at all (files in the database are rarely
441 * looked up more than once...), so caching is just a waste of memory. The
442 * only limitation is the amount of scratch file space available to store the
443 * path names.
444 */
445
446/*
447 * ftime_start()
448 * create the file time hash table and open for read/write the scratch
449 * file. (after created it is unlinked, so when we exit we leave
450 * no witnesses).
451 * Return:
452 * 0 if the table and file was created ok, -1 otherwise
453 */
454
455int
456ftime_start(void)
457{
458
459 if (ftab != NULL((void *)0))
460 return(0);
461 if ((ftab = calloc(F_TAB_SZ50503, sizeof(FTM *))) == NULL((void *)0)) {
462 paxwarn(1, "Cannot allocate memory for file time table");
463 return(-1);
464 }
465
466 /*
467 * get random name and create temporary scratch file, unlink name
468 * so it will get removed on exit
469 */
470 memcpy(tempbase, _TFILE_BASE"paxXXXXXXXXXX", sizeof(_TFILE_BASE"paxXXXXXXXXXX"));
471 if ((ffd = mkstemp(tempfile)) == -1) {
472 syswarn(1, errno(*__errno()), "Unable to create temporary file: %s",
473 tempfile);
474 return(-1);
475 }
476 (void)unlink(tempfile);
477
478 return(0);
479}
480
481/*
482 * chk_ftime()
483 * looks up entry in file time hash table. If not found, the file is
484 * added to the hash table and the file named stored in the scratch file.
485 * If a file with the same name is found, the file times are compared and
486 * the most recent file time is retained. If the new file was younger (or
487 * was not in the database) the new file is selected for storage.
488 * Return:
489 * 0 if file should be added to the archive, 1 if it should be skipped,
490 * -1 on error
491 */
492
493int
494chk_ftime(ARCHD *arcn)
495{
496 FTM *pt;
497 int namelen;
498 u_int indx;
499 char ckname[PAXPATHLEN3072+1];
500
501 /*
502 * no info, go ahead and add to archive
503 */
504 if (ftab == NULL((void *)0))
505 return(0);
506
507 /*
508 * hash the pathname and look up in table
509 */
510 namelen = arcn->nlen;
511 indx = st_hash(arcn->name, namelen, F_TAB_SZ50503);
512 if ((pt = ftab[indx]) != NULL((void *)0)) {
513 /*
514 * the hash chain is not empty, walk down looking for match
515 * only read up the path names if the lengths match, speeds
516 * up the search a lot
517 */
518 while (pt != NULL((void *)0)) {
519 if (pt->namelen == namelen) {
520 /*
521 * potential match, have to read the name
522 * from the scratch file.
523 */
524 if (lseek(ffd,pt->seek,SEEK_SET0) != pt->seek) {
525 syswarn(1, errno(*__errno()),
526 "Failed ftime table seek");
527 return(-1);
528 }
529 if (read(ffd, ckname, namelen) != namelen) {
530 syswarn(1, errno(*__errno()),
531 "Failed ftime table read");
532 return(-1);
533 }
534
535 /*
536 * if the names match, we are done
537 */
538 if (!strncmp(ckname, arcn->name, namelen))
539 break;
540 }
541
542 /*
543 * try the next entry on the chain
544 */
545 pt = pt->fow;
546 }
547
548 if (pt != NULL((void *)0)) {
549 /*
550 * found the file, compare the times, save the newer
551 */
552 if (timespeccmp(&arcn->sb.st_mtim, &pt->mtim, >)(((&arcn->sb.st_mtim)->tv_sec == (&pt->mtim)
->tv_sec) ? ((&arcn->sb.st_mtim)->tv_nsec > (
&pt->mtim)->tv_nsec) : ((&arcn->sb.st_mtim)->
tv_sec > (&pt->mtim)->tv_sec))
) {
553 /*
554 * file is newer
555 */
556 pt->mtim = arcn->sb.st_mtim;
557 return(0);
558 }
559 /*
560 * file is older
561 */
562 return(1);
563 }
564 }
565
566 /*
567 * not in table, add it
568 */
569 if ((pt = malloc(sizeof(FTM))) != NULL((void *)0)) {
570 /*
571 * add the name at the end of the scratch file, saving the
572 * offset. add the file to the head of the hash chain
573 */
574 if ((pt->seek = lseek(ffd, 0, SEEK_END2)) >= 0) {
575 if (write(ffd, arcn->name, namelen) == namelen) {
576 pt->mtim = arcn->sb.st_mtim;
577 pt->namelen = namelen;
578 pt->fow = ftab[indx];
579 ftab[indx] = pt;
580 return(0);
581 }
582 syswarn(1, errno(*__errno()), "Failed write to file time table");
583 } else
584 syswarn(1, errno(*__errno()), "Failed seek on file time table");
585 } else
586 paxwarn(1, "File time table ran out of memory");
587
588 if (pt != NULL((void *)0))
589 free(pt);
590 return(-1);
591}
592
593/*
594 * escaping (absolute or w/"..") symlink table routines
595 *
596 * By default, an archive shouldn't be able extract to outside of the
597 * current directory. What should we do if the archive contains a symlink
598 * whose value is either absolute or contains ".." components? What we'll
599 * do is initially create the path as an empty file (to block attempts to
600 * reference _through_ it) and instead record its path and desired
601 * final value and mode. Then once all the other archive
602 * members are created (but before the pass to set timestamps on
603 * directories) we'll process those records, replacing the placeholder with
604 * the correct symlink and setting them to the correct mode, owner, group,
605 * and timestamps.
606 *
607 * Note: we also need to handle hardlinks to symlinks (barf) as well as
608 * hardlinks whose target is replaced by a later entry in the archive (barf^2).
609 *
610 * So we track things by dev+ino of the placeholder file, associating with
611 * that the value and mode of the final symlink and a list of paths that
612 * should all be hardlinks of that. We'll 'store' the symlink's desired
613 * timestamps, owner, and group by setting them on the placeholder file.
614 *
615 * The operations are:
616 * a) create an escaping symlink: create the placeholder file and add an entry
617 * for the new link
618 * b) create a hardlink: do the link. If the target turns out to be a
619 * zero-length file whose dev+ino are in the symlink table, then add this
620 * path to the list of names for that link
621 * c) perform deferred processing: for each entry, check each associated path:
622 * if it's a zero-length file with the correct dev+ino then recreate it as
623 * the specified symlink or hardlink to the first such
624 */
625
626struct slpath {
627 char *sp_path;
628 struct slpath *sp_next;
629};
630struct slinode {
631 ino_t sli_ino;
632 char *sli_value;
633 struct slpath sli_paths;
634 struct slinode *sli_fow; /* hash table chain */
635 dev_t sli_dev;
636 mode_t sli_mode;
637};
638
639static struct slinode **slitab = NULL((void *)0);
640
641/*
642 * sltab_start()
643 * create the hash table
644 * Return:
645 * 0 if the table and file was created ok, -1 otherwise
646 */
647
648int
649sltab_start(void)
650{
651
652 if ((slitab = calloc(SL_TAB_SZ317, sizeof *slitab)) == NULL((void *)0)) {
653 syswarn(1, errno(*__errno()), "symlink table");
654 return(-1);
655 }
656
657 return(0);
658}
659
660/*
661 * sltab_add_sym()
662 * Create the placeholder and tracking info for an escaping symlink.
663 * Return:
664 * 0 on success, -1 otherwise
665 */
666
667int
668sltab_add_sym(const char *path0, const char *value0, mode_t mode)
669{
670 struct stat sb;
671 struct slinode *s;
672 struct slpath *p;
673 char *path, *value;
674 u_int indx;
675 int fd;
676
677 /* create the placeholder */
678 fd = open(path0, O_WRONLY0x0001 | O_CREAT0x0200 | O_EXCL0x0800 | O_CLOEXEC0x10000, 0600);
679 if (fd == -1)
680 return (-1);
681 if (fstat(fd, &sb) == -1) {
682 unlink(path0);
683 close(fd);
684 return (-1);
685 }
686 close(fd);
687
688 if (havechd && *path0 != '/') {
689 if ((path = realpath(path0, NULL((void *)0))) == NULL((void *)0)) {
690 syswarn(1, errno(*__errno()), "Cannot canonicalize %s", path0);
691 unlink(path0);
692 return (-1);
693 }
694 } else if ((path = strdup(path0)) == NULL((void *)0)) {
695 syswarn(1, errno(*__errno()), "defered symlink path");
696 unlink(path0);
697 return (-1);
698 }
699 if ((value = strdup(value0)) == NULL((void *)0)) {
700 syswarn(1, errno(*__errno()), "defered symlink value");
701 unlink(path);
702 free(path);
703 return (-1);
704 }
705
706 /* now check the hash table for conflicting entry */
707 indx = (sb.st_ino ^ sb.st_dev) % SL_TAB_SZ317;
708 for (s = slitab[indx]; s != NULL((void *)0); s = s->sli_fow) {
709 if (s->sli_ino != sb.st_ino || s->sli_dev != sb.st_dev)
710 continue;
711
712 /*
713 * One of our placeholders got removed behind our back and
714 * we've reused the inode. Weird, but clean up the mess.
715 */
716 free(s->sli_value);
717 free(s->sli_paths.sp_path);
718 p = s->sli_paths.sp_next;
719 while (p != NULL((void *)0)) {
720 struct slpath *next_p = p->sp_next;
721
722 free(p->sp_path);
723 free(p);
724 p = next_p;
725 }
726 goto set_value;
727 }
728
729 /* Normal case: create a new node */
730 if ((s = malloc(sizeof *s)) == NULL((void *)0)) {
731 syswarn(1, errno(*__errno()), "defered symlink");
732 unlink(path);
733 free(path);
734 free(value);
735 return (-1);
736 }
737 s->sli_ino = sb.st_ino;
738 s->sli_dev = sb.st_dev;
739 s->sli_fow = slitab[indx];
740 slitab[indx] = s;
741
742set_value:
743 s->sli_paths.sp_path = path;
744 s->sli_paths.sp_next = NULL((void *)0);
745 s->sli_value = value;
746 s->sli_mode = mode;
747 return (0);
748}
749
750/*
751 * sltab_add_link()
752 * A hardlink was created; if it looks like a placeholder, handle the
753 * tracking.
754 * Return:
755 * 0 if things are ok, -1 if something went wrong
756 */
757
758int
759sltab_add_link(const char *path, const struct stat *sb)
760{
761 struct slinode *s;
762 struct slpath *p;
763 u_int indx;
764
765 if (!S_ISREG(sb->st_mode)((sb->st_mode & 0170000) == 0100000) || sb->st_size != 0)
766 return (1);
767
768 /* find the hash table entry for this hardlink */
769 indx = (sb->st_ino ^ sb->st_dev) % SL_TAB_SZ317;
770 for (s = slitab[indx]; s != NULL((void *)0); s = s->sli_fow) {
771 if (s->sli_ino != sb->st_ino || s->sli_dev != sb->st_dev)
772 continue;
773
774 if ((p = malloc(sizeof *p)) == NULL((void *)0)) {
775 syswarn(1, errno(*__errno()), "deferred symlink hardlink");
776 return (-1);
777 }
778 if (havechd && *path != '/') {
779 if ((p->sp_path = realpath(path, NULL((void *)0))) == NULL((void *)0)) {
780 syswarn(1, errno(*__errno()), "Cannot canonicalize %s",
781 path);
782 free(p);
783 return (-1);
784 }
785 } else if ((p->sp_path = strdup(path)) == NULL((void *)0)) {
786 syswarn(1, errno(*__errno()), "defered symlink hardlink path");
787 free(p);
788 return (-1);
789 }
790
791 /* link it in */
792 p->sp_next = s->sli_paths.sp_next;
793 s->sli_paths.sp_next = p;
794 return (0);
795 }
796
797 /* not found */
798 return (1);
799}
800
801
802static int
803sltab_process_one(struct slinode *s, struct slpath *p, const char *first,
804 int in_sig)
805{
806 struct stat sb;
807 char *path = p->sp_path;
808 mode_t mode;
809 int err;
810
811 /*
812 * is it the expected placeholder? This can fail legimately
813 * if the archive overwrote the link with another, later entry,
814 * so don't warn.
815 */
816 if (stat(path, &sb) != 0 || !S_ISREG(sb.st_mode)((sb.st_mode & 0170000) == 0100000) || sb.st_size != 0 ||
817 sb.st_ino != s->sli_ino || sb.st_dev != s->sli_dev)
818 return (0);
819
820 if (unlink(path) && errno(*__errno()) != ENOENT2) {
821 if (!in_sig)
822 syswarn(1, errno(*__errno()), "deferred symlink removal");
823 return (0);
824 }
825
826 err = 0;
827 if (first != NULL((void *)0)) {
828 /* add another hardlink to the existing symlink */
829 if (linkat(AT_FDCWD-100, first, AT_FDCWD-100, path, 0) == 0)
830 return (0);
831
832 /*
833 * Couldn't hardlink the symlink for some reason, so we'll
834 * try creating it as its own symlink, but save the error
835 * for reporting if that fails.
836 */
837 err = errno(*__errno());
838 }
839
840 if (symlink(s->sli_value, path)) {
841 if (!in_sig) {
842 const char *qualifier = "";
843 if (err)
844 qualifier = " hardlink";
845 else
846 err = errno(*__errno());
847
848 syswarn(1, err, "deferred symlink%s: %s",
849 qualifier, path);
850 }
851 return (0);
852 }
853
854 /* success, so set the id, mode, and times */
855 mode = s->sli_mode;
856 if (pids) {
857 /* if can't set the ids, force the set[ug]id bits off */
858 if (set_ids(path, sb.st_uid, sb.st_gid))
859 mode &= ~(SETBITS(0004000 | 0002000));
860 }
861
862 if (pmode)
863 set_pmode(path, mode);
864
865 if (patime || pmtime)
866 set_ftime(path, &sb.st_mtim, &sb.st_atim, 0);
867
868 /*
869 * If we tried to link to first but failed, then this new symlink
870 * might be a better one to try in the future. Guess from the errno.
871 */
872 if (err == 0 || err == ENOENT2 || err == EMLINK31 || err == EOPNOTSUPP45)
873 return (1);
874 return (0);
875}
876
877/*
878 * sltab_process()
879 * Do all the delayed process for escape symlinks
880 */
881
882void
883sltab_process(int in_sig)
884{
885 struct slinode *s;
886 struct slpath *p;
887 char *first;
888 u_int indx;
889
890 if (slitab == NULL((void *)0))
891 return;
892
893 /* walk across the entire hash table */
894 for (indx = 0; indx < SL_TAB_SZ317; indx++) {
895 while ((s = slitab[indx]) != NULL((void *)0)) {
896 /* pop this entry */
897 slitab[indx] = s->sli_fow;
898
899 first = NULL((void *)0);
900 p = &s->sli_paths;
901 while (1) {
902 struct slpath *next_p;
903
904 if (sltab_process_one(s, p, first, in_sig)) {
905 if (!in_sig)
906 free(first);
907 first = p->sp_path;
908 } else if (!in_sig)
909 free(p->sp_path);
910
911 if ((next_p = p->sp_next) == NULL((void *)0))
912 break;
913 *p = *next_p;
914 if (!in_sig)
915 free(next_p);
916 }
917 if (!in_sig) {
918 free(first);
919 free(s->sli_value);
920 free(s);
921 }
922 }
923 }
924 if (!in_sig)
925 free(slitab);
926 slitab = NULL((void *)0);
927}
928
929
930/*
931 * Interactive rename table routines
932 *
933 * The interactive rename table keeps track of the new names that the user
934 * assigns to files from tty input. Since this map is unique for each file
935 * we must store it in case there is a reference to the file later in archive
936 * (a link). Otherwise we will be unable to find the file we know was
937 * extracted. The remapping of these files is stored in a memory based hash
938 * table (it is assumed since input must come from /dev/tty, it is unlikely to
939 * be a very large table).
940 */
941
942/*
943 * name_start()
944 * create the interactive rename table
945 * Return:
946 * 0 if successful, -1 otherwise
947 */
948
949int
950name_start(void)
951{
952 if (ntab != NULL((void *)0))
953 return(0);
954 if ((ntab = calloc(N_TAB_SZ541, sizeof(NAMT *))) == NULL((void *)0)) {
955 paxwarn(1, "Cannot allocate memory for interactive rename table");
956 return(-1);
957 }
958 return(0);
959}
960
961/*
962 * add_name()
963 * add the new name to old name mapping just created by the user.
964 * If an old name mapping is found (there may be duplicate names on an
965 * archive) only the most recent is kept.
966 * Return:
967 * 0 if added, -1 otherwise
968 */
969
970int
971add_name(char *oname, int onamelen, char *nname)
972{
973 NAMT *pt;
974 u_int indx;
975
976 if (ntab == NULL((void *)0)) {
977 /*
978 * should never happen
979 */
980 paxwarn(0, "No interactive rename table, links may fail");
981 return(0);
982 }
983
984 /*
985 * look to see if we have already mapped this file, if so we
986 * will update it
987 */
988 indx = st_hash(oname, onamelen, N_TAB_SZ541);
989 if ((pt = ntab[indx]) != NULL((void *)0)) {
990 /*
991 * look down the has chain for the file
992 */
993 while ((pt != NULL((void *)0)) && (strcmp(oname, pt->oname) != 0))
994 pt = pt->fow;
995
996 if (pt != NULL((void *)0)) {
997 /*
998 * found an old mapping, replace it with the new one
999 * the user just input (if it is different)
1000 */
1001 if (strcmp(nname, pt->nname) == 0)
1002 return(0);
1003
1004 free(pt->nname);
1005 if ((pt->nname = strdup(nname)) == NULL((void *)0)) {
1006 paxwarn(1, "Cannot update rename table");
1007 return(-1);
1008 }
1009 return(0);
1010 }
1011 }
1012
1013 /*
1014 * this is a new mapping, add it to the table
1015 */
1016 if ((pt = malloc(sizeof(NAMT))) != NULL((void *)0)) {
1017 if ((pt->oname = strdup(oname)) != NULL((void *)0)) {
1018 if ((pt->nname = strdup(nname)) != NULL((void *)0)) {
1019 pt->fow = ntab[indx];
1020 ntab[indx] = pt;
1021 return(0);
1022 }
1023 free(pt->oname);
1024 }
1025 free(pt);
1026 }
1027 paxwarn(1, "Interactive rename table out of memory");
1028 return(-1);
1029}
1030
1031/*
1032 * sub_name()
1033 * look up a link name to see if it points at a file that has been
1034 * remapped by the user. If found, the link is adjusted to contain the
1035 * new name (oname is the link to name)
1036 */
1037
1038void
1039sub_name(char *oname, int *onamelen, int onamesize)
1040{
1041 NAMT *pt;
1042 u_int indx;
1043
1044 if (ntab == NULL((void *)0))
1
Assuming 'ntab' is not equal to NULL
2
Taking false branch
1045 return;
1046 /*
1047 * look the name up in the hash table
1048 */
1049 indx = st_hash(oname, *onamelen, N_TAB_SZ541);
3
Calling 'st_hash'
1050 if ((pt = ntab[indx]) == NULL((void *)0))
1051 return;
1052
1053 while (pt != NULL((void *)0)) {
1054 /*
1055 * walk down the hash chain looking for a match
1056 */
1057 if (strcmp(oname, pt->oname) == 0) {
1058 /*
1059 * found it, replace it with the new name
1060 * and return (we know that oname has enough space)
1061 */
1062 *onamelen = strlcpy(oname, pt->nname, onamesize);
1063 if (*onamelen >= onamesize)
1064 *onamelen = onamesize - 1; /* XXX truncate? */
1065 return;
1066 }
1067 pt = pt->fow;
1068 }
1069
1070 /*
1071 * no match, just return
1072 */
1073}
1074
1075#ifndef NOCPIO
1076/*
1077 * device/inode mapping table routines
1078 * (used with formats that store device and inodes fields)
1079 *
1080 * device/inode mapping tables remap the device field in a archive header. The
1081 * device/inode fields are used to determine when files are hard links to each
1082 * other. However these values have very little meaning outside of that. This
1083 * database is used to solve one of two different problems.
1084 *
1085 * 1) when files are appended to an archive, while the new files may have hard
1086 * links to each other, you cannot determine if they have hard links to any
1087 * file already stored on the archive from a prior run of pax. We must assume
1088 * that these inode/device pairs are unique only within a SINGLE run of pax
1089 * (which adds a set of files to an archive). So we have to make sure the
1090 * inode/dev pairs we add each time are always unique. We do this by observing
1091 * while the inode field is very dense, the use of the dev field is fairly
1092 * sparse. Within each run of pax, we remap any device number of a new archive
1093 * member that has a device number used in a prior run and already stored in a
1094 * file on the archive. During the read phase of the append, we store the
1095 * device numbers used and mark them to not be used by any file during the
1096 * write phase. If during write we go to use one of those old device numbers,
1097 * we remap it to a new value.
1098 *
1099 * 2) Often the fields in the archive header used to store these values are
1100 * too small to store the entire value. The result is an inode or device value
1101 * which can be truncated. This really can foul up an archive. With truncation
1102 * we end up creating links between files that are really not links (after
1103 * truncation the inodes are the same value). We address that by detecting
1104 * truncation and forcing a remap of the device field to split truncated
1105 * inodes away from each other. Each truncation creates a pattern of bits that
1106 * are removed. We use this pattern of truncated bits to partition the inodes
1107 * on a single device to many different devices (each one represented by the
1108 * truncated bit pattern). All inodes on the same device that have the same
1109 * truncation pattern are mapped to the same new device. Two inodes that
1110 * truncate to the same value clearly will always have different truncation
1111 * bit patterns, so they will be split from away each other. When we spot
1112 * device truncation we remap the device number to a non truncated value.
1113 * (for more info see table.h for the data structures involved).
1114 */
1115
1116static DEVT *chk_dev(dev_t, int);
1117
1118/*
1119 * dev_start()
1120 * create the device mapping table
1121 * Return:
1122 * 0 if successful, -1 otherwise
1123 */
1124
1125int
1126dev_start(void)
1127{
1128 if (dtab != NULL((void *)0))
1129 return(0);
1130 if ((dtab = calloc(D_TAB_SZ317, sizeof(DEVT *))) == NULL((void *)0)) {
1131 paxwarn(1, "Cannot allocate memory for device mapping table");
1132 return(-1);
1133 }
1134 return(0);
1135}
1136
1137/*
1138 * add_dev()
1139 * add a device number to the table. this will force the device to be
1140 * remapped to a new value if it be used during a write phase. This
1141 * function is called during the read phase of an append to prohibit the
1142 * use of any device number already in the archive.
1143 * Return:
1144 * 0 if added ok, -1 otherwise
1145 */
1146
1147int
1148add_dev(ARCHD *arcn)
1149{
1150 if (chk_dev(arcn->sb.st_dev, 1) == NULL((void *)0))
1151 return(-1);
1152 return(0);
1153}
1154
1155/*
1156 * chk_dev()
1157 * check for a device value in the device table. If not found and the add
1158 * flag is set, it is added. This does NOT assign any mapping values, just
1159 * adds the device number as one that need to be remapped. If this device
1160 * is already mapped, just return with a pointer to that entry.
1161 * Return:
1162 * pointer to the entry for this device in the device map table. Null
1163 * if the add flag is not set and the device is not in the table (it is
1164 * not been seen yet). If add is set and the device cannot be added, null
1165 * is returned (indicates an error).
1166 */
1167
1168static DEVT *
1169chk_dev(dev_t dev, int add)
1170{
1171 DEVT *pt;
1172 u_int indx;
1173
1174 if (dtab == NULL((void *)0))
1175 return(NULL((void *)0));
1176 /*
1177 * look to see if this device is already in the table
1178 */
1179 indx = ((unsigned)dev) % D_TAB_SZ317;
1180 if ((pt = dtab[indx]) != NULL((void *)0)) {
1181 while ((pt != NULL((void *)0)) && (pt->dev != dev))
1182 pt = pt->fow;
1183
1184 /*
1185 * found it, return a pointer to it
1186 */
1187 if (pt != NULL((void *)0))
1188 return(pt);
1189 }
1190
1191 /*
1192 * not in table, we add it only if told to as this may just be a check
1193 * to see if a device number is being used.
1194 */
1195 if (add == 0)
1196 return(NULL((void *)0));
1197
1198 /*
1199 * allocate a node for this device and add it to the front of the hash
1200 * chain. Note we do not assign remaps values here, so the pt->list
1201 * list must be NULL.
1202 */
1203 if ((pt = malloc(sizeof(DEVT))) == NULL((void *)0)) {
1204 paxwarn(1, "Device map table out of memory");
1205 return(NULL((void *)0));
1206 }
1207 pt->dev = dev;
1208 pt->list = NULL((void *)0);
1209 pt->fow = dtab[indx];
1210 dtab[indx] = pt;
1211 return(pt);
1212}
1213/*
1214 * map_dev()
1215 * given an inode and device storage mask (the mask has a 1 for each bit
1216 * the archive format is able to store in a header), we check for inode
1217 * and device truncation and remap the device as required. Device mapping
1218 * can also occur when during the read phase of append a device number was
1219 * seen (and was marked as do not use during the write phase). WE ASSUME
1220 * that unsigned longs are the same size or bigger than the fields used
1221 * for ino_t and dev_t. If not the types will have to be changed.
1222 * Return:
1223 * 0 if all ok, -1 otherwise.
1224 */
1225
1226int
1227map_dev(ARCHD *arcn, u_long dev_mask, u_long ino_mask)
1228{
1229 DEVT *pt;
1230 DLIST *dpt;
1231 static dev_t lastdev = 0; /* next device number to try */
1232 int trc_ino = 0;
1233 int trc_dev = 0;
1234 ino_t trunc_bits = 0;
1235 ino_t nino;
1236
1237 if (dtab == NULL((void *)0))
1238 return(0);
1239 /*
1240 * check for device and inode truncation, and extract the truncated
1241 * bit pattern.
1242 */
1243 if ((arcn->sb.st_dev & (dev_t)dev_mask) != arcn->sb.st_dev)
1244 ++trc_dev;
1245 if ((nino = arcn->sb.st_ino & (ino_t)ino_mask) != arcn->sb.st_ino) {
1246 ++trc_ino;
1247 trunc_bits = arcn->sb.st_ino & (ino_t)(~ino_mask);
1248 }
1249
1250 /*
1251 * see if this device is already being mapped, look up the device
1252 * then find the truncation bit pattern which applies
1253 */
1254 if ((pt = chk_dev(arcn->sb.st_dev, 0)) != NULL((void *)0)) {
1255 /*
1256 * this device is already marked to be remapped
1257 */
1258 for (dpt = pt->list; dpt != NULL((void *)0); dpt = dpt->fow)
1259 if (dpt->trunc_bits == trunc_bits)
1260 break;
1261
1262 if (dpt != NULL((void *)0)) {
1263 /*
1264 * we are being remapped for this device and pattern
1265 * change the device number to be stored and return
1266 */
1267 arcn->sb.st_dev = dpt->dev;
1268 arcn->sb.st_ino = nino;
1269 return(0);
1270 }
1271 } else {
1272 /*
1273 * this device is not being remapped YET. if we do not have any
1274 * form of truncation, we do not need a remap
1275 */
1276 if (!trc_ino && !trc_dev)
1277 return(0);
1278
1279 /*
1280 * we have truncation, have to add this as a device to remap
1281 */
1282 if ((pt = chk_dev(arcn->sb.st_dev, 1)) == NULL((void *)0))
1283 goto bad;
1284
1285 /*
1286 * if we just have a truncated inode, we have to make sure that
1287 * all future inodes that do not truncate (they have the
1288 * truncation pattern of all 0's) continue to map to the same
1289 * device number. We probably have already written inodes with
1290 * this device number to the archive with the truncation
1291 * pattern of all 0's. So we add the mapping for all 0's to the
1292 * same device number.
1293 */
1294 if (!trc_dev && (trunc_bits != 0)) {
1295 if ((dpt = malloc(sizeof(DLIST))) == NULL((void *)0))
1296 goto bad;
1297 dpt->trunc_bits = 0;
1298 dpt->dev = arcn->sb.st_dev;
1299 dpt->fow = pt->list;
1300 pt->list = dpt;
1301 }
1302 }
1303
1304 /*
1305 * look for a device number not being used. We must watch for wrap
1306 * around on lastdev (so we do not get stuck looking forever!)
1307 */
1308 while (++lastdev > 0) {
1309 if (chk_dev(lastdev, 0) != NULL((void *)0))
1310 continue;
1311 /*
1312 * found an unused value. If we have reached truncation point
1313 * for this format we are hosed, so we give up. Otherwise we
1314 * mark it as being used.
1315 */
1316 if (((lastdev & ((dev_t)dev_mask)) != lastdev) ||
1317 (chk_dev(lastdev, 1) == NULL((void *)0)))
1318 goto bad;
1319 break;
1320 }
1321
1322 if ((lastdev <= 0) || ((dpt = malloc(sizeof(DLIST))) == NULL((void *)0)))
1323 goto bad;
1324
1325 /*
1326 * got a new device number, store it under this truncation pattern.
1327 * change the device number this file is being stored with.
1328 */
1329 dpt->trunc_bits = trunc_bits;
1330 dpt->dev = lastdev;
1331 dpt->fow = pt->list;
1332 pt->list = dpt;
1333 arcn->sb.st_dev = lastdev;
1334 arcn->sb.st_ino = nino;
1335 return(0);
1336
1337 bad:
1338 paxwarn(1, "Unable to fix truncated inode/device field when storing %s",
1339 arcn->name);
1340 paxwarn(0, "Archive may create improper hard links when extracted");
1341 return(0);
1342}
1343#endif /* NOCPIO */
1344
1345/*
1346 * directory access/mod time reset table routines (for directories READ by pax)
1347 *
1348 * The pax -t flag requires that access times of archive files be the same
1349 * before being read by pax. For regular files, access time is restored after
1350 * the file has been copied. This database provides the same functionality for
1351 * directories read during file tree traversal. Restoring directory access time
1352 * is more complex than files since directories may be read several times until
1353 * all the descendants in their subtree are visited by fts. Directory access
1354 * and modification times are stored during the fts pre-order visit (done
1355 * before any descendants in the subtree are visited) and restored after the
1356 * fts post-order visit (after all the descendants have been visited). In the
1357 * case of premature exit from a subtree (like from the effects of -n), any
1358 * directory entries left in this database are reset during final cleanup
1359 * operations of pax. Entries are hashed by inode number for fast lookup.
1360 */
1361
1362/*
1363 * atdir_start()
1364 * create the directory access time database for directories READ by pax.
1365 * Return:
1366 * 0 is created ok, -1 otherwise.
1367 */
1368
1369int
1370atdir_start(void)
1371{
1372 if (atab != NULL((void *)0))
1373 return(0);
1374 if ((atab = calloc(A_TAB_SZ317, sizeof(ATDIR *))) == NULL((void *)0)) {
1375 paxwarn(1,"Cannot allocate space for directory access time table");
1376 return(-1);
1377 }
1378 return(0);
1379}
1380
1381
1382/*
1383 * atdir_end()
1384 * walk through the directory access time table and reset the access time
1385 * of any directory who still has an entry left in the database. These
1386 * entries are for directories READ by pax
1387 */
1388
1389void
1390atdir_end(void)
1391{
1392 ATDIR *pt;
1393 int i;
1394
1395 if (atab == NULL((void *)0))
1396 return;
1397 /*
1398 * for each non-empty hash table entry reset all the directories
1399 * chained there.
1400 */
1401 for (i = 0; i < A_TAB_SZ317; ++i) {
1402 if ((pt = atab[i]) == NULL((void *)0))
1403 continue;
1404 /*
1405 * remember to force the times, set_ftime() looks at pmtime
1406 * and patime, which only applies to things CREATED by pax,
1407 * not read by pax. Read time reset is controlled by -t.
1408 */
1409 for (; pt != NULL((void *)0); pt = pt->fow)
1410 set_attr(&pt->ft, 1, 0, 0, 0);
1411 }
1412}
1413
1414/*
1415 * add_atdir()
1416 * add a directory to the directory access time table. Table is hashed
1417 * and chained by inode number. This is for directories READ by pax
1418 */
1419
1420void
1421add_atdir(char *fname, dev_t dev, ino_t ino, const struct timespec *mtimp,
1422 const struct timespec *atimp)
1423{
1424 ATDIR *pt;
1425 sigset_t allsigs, savedsigs;
1426 u_int indx;
1427
1428 if (atab == NULL((void *)0))
1429 return;
1430
1431 /*
1432 * make sure this directory is not already in the table, if so just
1433 * return (the older entry always has the correct time). The only
1434 * way this will happen is when the same subtree can be traversed by
1435 * different args to pax and the -n option is aborting fts out of a
1436 * subtree before all the post-order visits have been made.
1437 */
1438 indx = ((unsigned)ino) % A_TAB_SZ317;
1439 if ((pt = atab[indx]) != NULL((void *)0)) {
1440 while (pt != NULL((void *)0)) {
1441 if ((pt->ft.ft_ino == ino) && (pt->ft.ft_dev == dev))
1442 break;
1443 pt = pt->fow;
1444 }
1445
1446 /*
1447 * oops, already there. Leave it alone.
1448 */
1449 if (pt != NULL((void *)0))
1450 return;
1451 }
1452
1453 /*
1454 * add it to the front of the hash chain
1455 */
1456 sigfillset(&allsigs);
1457 sigprocmask(SIG_BLOCK1, &allsigs, &savedsigs);
1458 if ((pt = malloc(sizeof *pt)) != NULL((void *)0)) {
1459 if ((pt->ft.ft_name = strdup(fname)) != NULL((void *)0)) {
1460 pt->ft.ft_dev = dev;
1461 pt->ft.ft_ino = ino;
1462 pt->ft.ft_mtim = *mtimp;
1463 pt->ft.ft_atim = *atimp;
1464 pt->fow = atab[indx];
1465 atab[indx] = pt;
1466 sigprocmask(SIG_SETMASK3, &savedsigs, NULL((void *)0));
1467 return;
1468 }
1469 free(pt);
1470 }
1471
1472 sigprocmask(SIG_SETMASK3, &savedsigs, NULL((void *)0));
1473 paxwarn(1, "Directory access time reset table ran out of memory");
1474}
1475
1476/*
1477 * get_atdir()
1478 * look up a directory by inode and device number to obtain the access
1479 * and modification time you want to set to. If found, the modification
1480 * and access time parameters are set and the entry is removed from the
1481 * table (as it is no longer needed). These are for directories READ by
1482 * pax
1483 * Return:
1484 * 0 if found, -1 if not found.
1485 */
1486
1487int
1488do_atdir(const char *name, dev_t dev, ino_t ino)
1489{
1490 ATDIR *pt;
1491 ATDIR **ppt;
1492 sigset_t allsigs, savedsigs;
1493 u_int indx;
1494
1495 if (atab == NULL((void *)0))
1496 return(-1);
1497 /*
1498 * hash by inode and search the chain for an inode and device match
1499 */
1500 indx = ((unsigned)ino) % A_TAB_SZ317;
1501 if ((pt = atab[indx]) == NULL((void *)0))
1502 return(-1);
1503
1504 ppt = &(atab[indx]);
1505 while (pt != NULL((void *)0)) {
1506 if ((pt->ft.ft_ino == ino) && (pt->ft.ft_dev == dev))
1507 break;
1508 /*
1509 * no match, go to next one
1510 */
1511 ppt = &(pt->fow);
1512 pt = pt->fow;
1513 }
1514
1515 /*
1516 * return if we did not find it.
1517 */
1518 if (pt == NULL((void *)0) || pt->ft.ft_name == NULL((void *)0) ||
1519 strcmp(name, pt->ft.ft_name) == 0)
1520 return(-1);
1521
1522 /*
1523 * found it. set the times and remove the entry from the table.
1524 */
1525 set_attr(&pt->ft, 1, 0, 0, 0);
1526 sigfillset(&allsigs);
1527 sigprocmask(SIG_BLOCK1, &allsigs, &savedsigs);
1528 *ppt = pt->fow;
1529 sigprocmask(SIG_SETMASK3, &savedsigs, NULL((void *)0));
1530 free(pt->ft.ft_name);
1531 free(pt);
1532 return(0);
1533}
1534
1535/*
1536 * directory access mode and time storage routines (for directories CREATED
1537 * by pax).
1538 *
1539 * Pax requires that extracted directories, by default, have their access/mod
1540 * times and permissions set to the values specified in the archive. During the
1541 * actions of extracting (and creating the destination subtree during -rw copy)
1542 * directories extracted may be modified after being created. Even worse is
1543 * that these directories may have been created with file permissions which
1544 * prohibits any descendants of these directories from being extracted. When
1545 * directories are created by pax, access rights may be added to permit the
1546 * creation of files in their subtree. Every time pax creates a directory, the
1547 * times and file permissions specified by the archive are stored. After all
1548 * files have been extracted (or copied), these directories have their times
1549 * and file modes reset to the stored values. The directory info is restored in
1550 * reverse order as entries were added from root to leaf: to restore atime
1551 * properly, we must go backwards.
1552 */
1553
1554/*
1555 * dir_start()
1556 * set up the directory time and file mode storage for directories CREATED
1557 * by pax.
1558 * Return:
1559 * 0 if ok, -1 otherwise
1560 */
1561
1562int
1563dir_start(void)
1564{
1565 if (dirp != NULL((void *)0))
1566 return(0);
1567
1568 dirsize = DIRP_SIZE64;
1569 if ((dirp = reallocarray(NULL((void *)0), dirsize, sizeof(DIRDATA))) == NULL((void *)0)) {
1570 paxwarn(1, "Unable to allocate memory for directory times");
1571 return(-1);
1572 }
1573 return(0);
1574}
1575
1576/*
1577 * add_dir()
1578 * add the mode and times for a newly CREATED directory
1579 * name is name of the directory, psb the stat buffer with the data in it,
1580 * frc_mode is a flag that says whether to force the setting of the mode
1581 * (ignoring the user set values for preserving file mode). Frc_mode is
1582 * for the case where we created a file and found that the resulting
1583 * directory was not writeable and the user asked for file modes to NOT
1584 * be preserved. (we have to preserve what was created by default, so we
1585 * have to force the setting at the end. this is stated explicitly in the
1586 * pax spec)
1587 */
1588
1589void
1590add_dir(char *name, struct stat *psb, int frc_mode)
1591{
1592 DIRDATA *dblk;
1593 sigset_t allsigs, savedsigs;
1594 char realname[PATH_MAX1024], *rp;
1595
1596 if (dirp == NULL((void *)0))
1597 return;
1598
1599 if (havechd && *name != '/') {
1600 if ((rp = realpath(name, realname)) == NULL((void *)0)) {
1601 paxwarn(1, "Cannot canonicalize %s", name);
1602 return;
1603 }
1604 name = rp;
1605 }
1606 if (dircnt == dirsize) {
1607 dblk = reallocarray(dirp, dirsize * 2, sizeof(DIRDATA));
1608 if (dblk == NULL((void *)0)) {
1609 paxwarn(1, "Unable to store mode and times for created"
1610 " directory: %s", name);
1611 return;
1612 }
1613 sigprocmask(SIG_BLOCK1, &allsigs, &savedsigs);
1614 dirp = dblk;
1615 dirsize *= 2;
1616 sigprocmask(SIG_SETMASK3, &savedsigs, NULL((void *)0));
1617 }
1618 dblk = &dirp[dircnt];
1619 if ((dblk->ft.ft_name = strdup(name)) == NULL((void *)0)) {
1620 paxwarn(1, "Unable to store mode and times for created"
1621 " directory: %s", name);
1622 return;
1623 }
1624 dblk->ft.ft_mtim = psb->st_mtim;
1625 dblk->ft.ft_atim = psb->st_atim;
1626 dblk->ft.ft_ino = psb->st_ino;
1627 dblk->ft.ft_dev = psb->st_dev;
1628 dblk->mode = psb->st_mode & ABITS((0001000 | 0000700 | 0000070 | 0000007) | (0004000 | 0002000
))
;
1629 dblk->frc_mode = frc_mode;
1630 sigprocmask(SIG_BLOCK1, &allsigs, &savedsigs);
1631 ++dircnt;
1632 sigprocmask(SIG_SETMASK3, &savedsigs, NULL((void *)0));
1633}
1634
1635/*
1636 * delete_dir()
1637 * When we rmdir a directory, we may want to make sure we don't
1638 * later warn about being unable to set its mode and times.
1639 */
1640
1641void
1642delete_dir(dev_t dev, ino_t ino)
1643{
1644 DIRDATA *dblk;
1645 char *name;
1646 size_t i;
1647
1648 if (dirp == NULL((void *)0))
1649 return;
1650 for (i = 0; i < dircnt; i++) {
1651 dblk = &dirp[i];
1652
1653 if (dblk->ft.ft_name == NULL((void *)0))
1654 continue;
1655 if (dblk->ft.ft_dev == dev && dblk->ft.ft_ino == ino) {
1656 name = dblk->ft.ft_name;
1657 dblk->ft.ft_name = NULL((void *)0);
1658 free(name);
1659 break;
1660 }
1661 }
1662}
1663
1664/*
1665 * proc_dir(int in_sig)
1666 * process all file modes and times stored for directories CREATED
1667 * by pax. If in_sig is set, we're in a signal handler and can't
1668 * free stuff.
1669 */
1670
1671void
1672proc_dir(int in_sig)
1673{
1674 DIRDATA *dblk;
1675 size_t cnt;
1676
1677 if (dirp == NULL((void *)0))
1678 return;
1679 /*
1680 * read backwards through the file and process each directory
1681 */
1682 cnt = dircnt;
1683 while (cnt-- > 0) {
1684 dblk = &dirp[cnt];
1685 /*
1686 * If we remove a directory we created, we replace the
1687 * ft_name with NULL. Ignore those.
1688 */
1689 if (dblk->ft.ft_name == NULL((void *)0))
1690 continue;
1691
1692 /*
1693 * frc_mode set, make sure we set the file modes even if
1694 * the user didn't ask for it (see file_subs.c for more info)
1695 */
1696 set_attr(&dblk->ft, 0, dblk->mode, pmode || dblk->frc_mode,
1697 in_sig);
1698 if (!in_sig)
1699 free(dblk->ft.ft_name);
1700 }
1701
1702 if (!in_sig)
1703 free(dirp);
1704 dirp = NULL((void *)0);
1705 dircnt = 0;
1706}
1707
1708/*
1709 * database independent routines
1710 */
1711
1712/*
1713 * st_hash()
1714 * hashes filenames to a u_int for hashing into a table. Looks at the tail
1715 * end of file, as this provides far better distribution than any other
1716 * part of the name. For performance reasons we only care about the last
1717 * MAXKEYLEN chars (should be at LEAST large enough to pick off the file
1718 * name). Was tested on 500,000 name file tree traversal from the root
1719 * and gave almost a perfectly uniform distribution of keys when used with
1720 * prime sized tables (MAXKEYLEN was 128 in test). Hashes (sizeof int)
1721 * chars at a time and pads with 0 for last addition.
1722 * Return:
1723 * the hash value of the string MOD (%) the table size.
1724 */
1725
1726u_int
1727st_hash(const char *name, int len, int tabsz)
1728{
1729 const char *pt;
1730 char *dest;
1731 const char *end;
1732 int i;
1733 u_int key = 0;
1734 int steps;
1735 int res;
1736 u_int val;
4
'val' declared without an initial value
1737
1738 /*
1739 * only look at the tail up to MAXKEYLEN, we do not need to waste
1740 * time here (remember these are pathnames, the tail is what will
1741 * spread out the keys)
1742 */
1743 if (len > MAXKEYLEN64) {
5
Assuming 'len' is > MAXKEYLEN
6
Taking true branch
1744 pt = &(name[len - MAXKEYLEN64]);
1745 len = MAXKEYLEN64;
1746 } else
1747 pt = name;
1748
1749 /*
1750 * calculate the number of u_int size steps in the string and if
1751 * there is a runt to deal with
1752 */
1753 steps = len/sizeof(u_int);
1754 res = len % sizeof(u_int);
1755
1756 /*
1757 * add up the value of the string in unsigned integer sized pieces
1758 * too bad we cannot have unsigned int aligned strings, then we
1759 * could avoid the expensive copy.
1760 */
1761 for (i = 0; i < steps; ++i) {
7
Loop condition is true. Entering loop body
1762 end = pt + sizeof(u_int);
1763 dest = (char *)&val;
1764 while (pt < end)
8
Assuming 'pt' is >= 'end'
9
Loop condition is false. Execution continues on line 1766
1765 *dest++ = *pt++;
1766 key += val;
10
Assigned value is garbage or undefined
1767 }
1768
1769 /*
1770 * add in the runt padded with zero to the right
1771 */
1772 if (res) {
1773 val = 0;
1774 end = pt + res;
1775 dest = (char *)&val;
1776 while (pt < end)
1777 *dest++ = *pt++;
1778 key += val;
1779 }
1780
1781 /*
1782 * return the result mod the table size
1783 */
1784 return(key % tabsz);
1785}