source: freewrt/tools/paxmirabilis/src/ar_subs.c@ 3784d08

freewrt_1_0 freewrt_2_0
Last change on this file since 3784d08 was c41ecbb, checked in by Thorsten Glaser <tg@…>, 17 years ago

FreeWRT trunk:

  • merge new MirCPIO upstream, changes: – allow multiple -v options not only for tar but also for cpio, pax – add code to work around broken archives such as the CPIO archive inside

Fedora Core 4 glibc-common-2.3.6-3.i386.rpm which carry the actual data
of hardlinks not in the first but a later (here, the last) occurence of
the file in question: iff hardlinking succeeds (no cross-device!), size
of the linked files is 0, size of the archive member is greater than 0,
we are extracting, but not to stdout, proceed writing out the data.

  • add mircpio, mirpax, mirtar links in order to prevent interfering with native tools when both are in $PATH (this has worked since the last update, can be used now)

FreeWRT 1.0-stable:

  • merge paxmirabilis complete from trunk

note: untested

git-svn-id: ​svn://www.freewrt.org/branches/freewrt_1_0@3816 afb5a338-a214-0410-bd46-81f09a774fd1
  • Property mode set to 100644
File size: 31.8 KB
Line 
1/* $OpenBSD: ar_subs.c,v 1.31 2006/11/17 08:38:04 otto Exp $ */
2/* $NetBSD: ar_subs.c,v 1.5 1995/03/21 09:07:06 cgd Exp $ */
3
4/*-
5 * Copyright (c) 2008
6 * Thorsten Glaser <tg@mirbsd.de>
7 * Copyright (c) 1992 Keith Muller.
8 * Copyright (c) 1992, 1993
9 * The Regents of the University of California. All rights reserved.
10 *
11 * This code is derived from software contributed to Berkeley by
12 * Keith Muller of the University of California, San Diego.
13 *
14 * Redistribution and use in source and binary forms, with or without
15 * modification, are permitted provided that the following conditions
16 * are met:
17 * 1. Redistributions of source code must retain the above copyright
18 * notice, this list of conditions and the following disclaimer.
19 * 2. Redistributions in binary form must reproduce the above copyright
20 * notice, this list of conditions and the following disclaimer in the
21 * documentation and/or other materials provided with the distribution.
22 * 3. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 */
38
39#include <sys/param.h>
40#include <sys/time.h>
41#include <sys/stat.h>
42#include <signal.h>
43#include <string.h>
44#include <stdio.h>
45#include <fcntl.h>
46#include <errno.h>
47#include <unistd.h>
48#include <stdlib.h>
49#include "pax.h"
50#include "extern.h"
51#include "options.h"
52
53__SCCSID("@(#)ar_subs.c 8.2 (Berkeley) 4/18/94");
54__RCSID("$MirOS: src/bin/pax/ar_subs.c,v 1.7 2008/10/29 17:34:48 tg Exp $");
55
56static void wr_archive(ARCHD *, int is_app);
57static int get_arc(void);
58static int next_head(ARCHD *);
59extern sigset_t s_mask;
60
61/*
62 * Routines which control the overall operation modes of pax as specified by
63 * the user: list, append, read ...
64 */
65
66static char hdbuf[BLKMULT]; /* space for archive header on read */
67u_long flcnt; /* number of files processed */
68
69/*
70 * list()
71 * list the contents of an archive which match user supplied pattern(s)
72 * (no pattern matches all).
73 */
74
75void
76list(void)
77{
78 ARCHD *arcn;
79 int res;
80 ARCHD archd;
81 time_t now;
82
83 arcn = &archd;
84 /*
85 * figure out archive type; pass any format specific options to the
86 * archive option processing routine; call the format init routine. We
87 * also save current time for ls_list() so we do not make a system
88 * call for each file we need to print. If verbose (vflag) start up
89 * the name and group caches.
90 */
91 if ((get_arc() < 0) || ((*frmt->options)() < 0) ||
92 ((*frmt->st_rd)() < 0))
93 return;
94
95 if (vflag && ((uidtb_start() < 0) || (gidtb_start() < 0)))
96 return;
97
98 now = time(NULL);
99
100 /*
101 * step through the archive until the format says it is done
102 */
103 while (next_head(arcn) == 0) {
104 if (arcn->type == PAX_GLL || arcn->type == PAX_GLF) {
105 /*
106 * we need to read, to get the real filename
107 */
108 off_t cnt;
109 if (!(*frmt->rd_data)(arcn, arcn->type == PAX_GLF
110 ? -1 : -2, &cnt))
111 (void)rd_skip(cnt + arcn->pad);
112 continue;
113 }
114
115 /*
116 * check for pattern, and user specified options match.
117 * When all patterns are matched we are done.
118 */
119 if ((res = pat_match(arcn)) < 0)
120 break;
121
122 if ((res == 0) && (sel_chk(arcn) == 0)) {
123 /*
124 * pattern resulted in a selected file
125 */
126 if (pat_sel(arcn) < 0)
127 break;
128
129 /*
130 * modify the name as requested by the user if name
131 * survives modification, do a listing of the file
132 */
133 if ((res = mod_name(arcn)) < 0)
134 break;
135 if (res == 0)
136 ls_list(arcn, now, stdout);
137 }
138
139 /*
140 * skip to next archive format header using values calculated
141 * by the format header read routine
142 */
143 if (rd_skip(arcn->skip + arcn->pad) == 1)
144 break;
145 }
146
147 /*
148 * all done, let format have a chance to cleanup, and make sure that
149 * the patterns supplied by the user were all matched
150 */
151 (void)(*frmt->end_rd)();
152 (void)sigprocmask(SIG_BLOCK, &s_mask, NULL);
153 ar_close();
154 pat_chk();
155}
156
157/*
158 * extract()
159 * extract the member(s) of an archive as specified by user supplied
160 * pattern(s) (no patterns extracts all members)
161 */
162
163void
164extract(void)
165{
166 ARCHD *arcn;
167 int res;
168 off_t cnt;
169 ARCHD archd;
170 struct stat sb;
171 int fd;
172 time_t now;
173
174 arcn = &archd;
175 /*
176 * figure out archive type; pass any format specific options to the
177 * archive option processing routine; call the format init routine;
178 * start up the directory modification time and access mode database
179 */
180 if ((get_arc() < 0) || ((*frmt->options)() < 0) ||
181 ((*frmt->st_rd)() < 0) || (dir_start() < 0))
182 return;
183
184 /*
185 * When we are doing interactive rename, we store the mapping of names
186 * so we can fix up hard links files later in the archive.
187 */
188 if (iflag && (name_start() < 0))
189 return;
190
191 now = time(NULL);
192
193 /*
194 * step through each entry on the archive until the format read routine
195 * says it is done
196 */
197 while (next_head(arcn) == 0) {
198 if (arcn->type == PAX_GLL || arcn->type == PAX_GLF) {
199 /*
200 * we need to read, to get the real filename
201 */
202 if (!(*frmt->rd_data)(arcn, arcn->type == PAX_GLF
203 ? -1 : -2, &cnt))
204 (void)rd_skip(cnt + arcn->pad);
205 continue;
206 }
207
208 /*
209 * check for pattern, and user specified options match. When
210 * all the patterns are matched we are done
211 */
212 if ((res = pat_match(arcn)) < 0)
213 break;
214
215 if ((res > 0) || (sel_chk(arcn) != 0)) {
216 /*
217 * file is not selected. skip past any file data and
218 * padding and go back for the next archive member
219 */
220 (void)rd_skip(arcn->skip + arcn->pad);
221 continue;
222 }
223
224 /*
225 * with -u or -D only extract when the archive member is newer
226 * than the file with the same name in the file system (no
227 * test of being the same type is required).
228 * NOTE: this test is done BEFORE name modifications as
229 * specified by pax. this operation can be confusing to the
230 * user who might expect the test to be done on an existing
231 * file AFTER the name mod. In honesty the pax spec is probably
232 * flawed in this respect.
233 */
234 if ((uflag || Dflag) && ((lstat(arcn->name, &sb) == 0))) {
235 if (uflag && Dflag) {
236 if ((arcn->sb.st_mtime <= sb.st_mtime) &&
237 (arcn->sb.st_ctime <= sb.st_ctime)) {
238 (void)rd_skip(arcn->skip + arcn->pad);
239 continue;
240 }
241 } else if (Dflag) {
242 if (arcn->sb.st_ctime <= sb.st_ctime) {
243 (void)rd_skip(arcn->skip + arcn->pad);
244 continue;
245 }
246 } else if (arcn->sb.st_mtime <= sb.st_mtime) {
247 (void)rd_skip(arcn->skip + arcn->pad);
248 continue;
249 }
250 }
251
252 /*
253 * this archive member is now been selected. modify the name.
254 */
255 if ((pat_sel(arcn) < 0) || ((res = mod_name(arcn)) < 0))
256 break;
257 if (res > 0) {
258 /*
259 * a bad name mod, skip and purge name from link table
260 */
261 purg_lnk(arcn);
262 (void)rd_skip(arcn->skip + arcn->pad);
263 continue;
264 }
265
266 /*
267 * Non standard -Y and -Z flag. When the existing file is
268 * same age or newer skip
269 */
270 if ((Yflag || Zflag) && ((lstat(arcn->name, &sb) == 0))) {
271 if (Yflag && Zflag) {
272 if ((arcn->sb.st_mtime <= sb.st_mtime) &&
273 (arcn->sb.st_ctime <= sb.st_ctime)) {
274 (void)rd_skip(arcn->skip + arcn->pad);
275 continue;
276 }
277 } else if (Yflag) {
278 if (arcn->sb.st_ctime <= sb.st_ctime) {
279 (void)rd_skip(arcn->skip + arcn->pad);
280 continue;
281 }
282 } else if (arcn->sb.st_mtime <= sb.st_mtime) {
283 (void)rd_skip(arcn->skip + arcn->pad);
284 continue;
285 }
286 }
287
288 if (vflag) {
289 if (vflag > 1)
290 ls_list(arcn, now, listf);
291 else {
292 (void)safe_print(arcn->name, listf);
293 vfpart = 1;
294 }
295 }
296
297 /*
298 * if required, chdir around.
299 */
300 if ((arcn->pat != NULL) && (arcn->pat->chdname != NULL))
301 if (!to_stdout && chdir(arcn->pat->chdname) != 0)
302 syswarn(1, errno, "Cannot chdir to %s",
303 arcn->pat->chdname);
304
305 /*
306 * all ok, extract this member based on type
307 */
308 if ((arcn->type != PAX_REG) && (arcn->type != PAX_CTG)) {
309 /*
310 * process archive members that are not regular files.
311 * throw out padding and any data that might follow the
312 * header (as determined by the format).
313 */
314 if (!to_stdout) {
315 if ((arcn->type == PAX_HLK) || (arcn->type == PAX_HRG)) {
316 res = lnk_creat(arcn, &fd);
317 if (fd != -1)
318 goto extdata;
319 } else
320 res = node_creat(arcn);
321 }
322
323 (void)rd_skip(arcn->skip + arcn->pad);
324 if (res < 0)
325 purg_lnk(arcn);
326
327 if (vflag && vfpart) {
328 (void)putc('\n', listf);
329 vfpart = 0;
330 }
331 goto popd;
332 }
333 /*
334 * we have a file with data here. If we can not create it, skip
335 * over the data and purge the name from hard link table
336 */
337 if (to_stdout)
338 fd = STDOUT_FILENO;
339 else if ((fd = file_creat(arcn)) < 0) {
340 (void)rd_skip(arcn->skip + arcn->pad);
341 purg_lnk(arcn);
342 goto popd;
343 }
344 /*
345 * extract the file from the archive and skip over padding and
346 * any unprocessed data
347 */
348 extdata:
349 res = (*frmt->rd_data)(arcn, fd, &cnt);
350 if (fd != STDOUT_FILENO)
351 file_close(arcn, fd);
352 if (vflag && vfpart) {
353 (void)putc('\n', listf);
354 vfpart = 0;
355 }
356 if (!res)
357 (void)rd_skip(cnt + arcn->pad);
358
359 popd:
360 /*
361 * if required, chdir around.
362 */
363 if ((arcn->pat != NULL) && (arcn->pat->chdname != NULL))
364 if (!to_stdout && fchdir(cwdfd) != 0)
365 syswarn(1, errno,
366 "Can't fchdir to starting directory");
367 }
368
369 /*
370 * all done, restore directory modes and times as required; make sure
371 * all patterns supplied by the user were matched; block off signals
372 * to avoid chance for multiple entry into the cleanup code.
373 */
374 (void)(*frmt->end_rd)();
375 (void)sigprocmask(SIG_BLOCK, &s_mask, NULL);
376 ar_close();
377 proc_dir();
378 pat_chk();
379}
380
381/*
382 * wr_archive()
383 * Write an archive. used in both creating a new archive and appends on
384 * previously written archive.
385 */
386
387static void
388wr_archive(ARCHD *arcn, int is_app)
389{
390 int res;
391 int hlk;
392 int wr_one;
393 off_t cnt;
394 int (*wrf)(ARCHD *);
395 int fd = -1;
396 time_t now;
397
398 /*
399 * if this format supports hard link storage, start up the database
400 * that detects them.
401 */
402 if (((hlk = frmt->hlk) == 1) && (lnk_start() < 0))
403 return;
404
405 /*
406 * if this is not append, and there are no files, we do not write a
407 * trailer
408 */
409 wr_one = is_app;
410
411 /*
412 * start up the file traversal code and format specific write
413 */
414 if (ftree_start() < 0) {
415 if (is_app)
416 goto trailer;
417 return;
418 } else if (((*frmt->st_wr)() < 0))
419 return;
420
421 wrf = frmt->wr;
422
423 /*
424 * When we are doing interactive rename, we store the mapping of names
425 * so we can fix up hard links files later in the archive.
426 */
427 if (iflag && (name_start() < 0))
428 return;
429
430 now = time(NULL);
431
432 /*
433 * while there are files to archive, process them one at at time
434 */
435 while (next_file(arcn) == 0) {
436 /*
437 * check if this file meets user specified options match.
438 */
439 if (sel_chk(arcn) != 0)
440 continue;
441 fd = -1;
442 if (uflag) {
443 /*
444 * only archive if this file is newer than a file with
445 * the same name that is already stored on the archive
446 */
447 if ((res = chk_ftime(arcn)) < 0)
448 break;
449 if (res > 0)
450 continue;
451 }
452
453 /*
454 * this file is considered selected now. see if this is a hard
455 * link to a file already stored
456 */
457 ftree_sel(arcn);
458 if (hlk && (chk_lnk(arcn) < 0))
459 break;
460
461 if ((arcn->type == PAX_REG) || (arcn->type == PAX_HRG) ||
462 (arcn->type == PAX_CTG)) {
463 /*
464 * we will have to read this file. by opening it now we
465 * can avoid writing a header to the archive for a file
466 * we were later unable to read (we also purge it from
467 * the link table).
468 */
469 if ((fd = open(arcn->org_name, O_RDONLY, 0)) < 0) {
470 syswarn(1,errno, "Unable to open %s to read",
471 arcn->org_name);
472 purg_lnk(arcn);
473 continue;
474 }
475 }
476
477 /*
478 * Now modify the name as requested by the user
479 */
480 if ((res = mod_name(arcn)) < 0) {
481 /*
482 * name modification says to skip this file, close the
483 * file and purge link table entry
484 */
485 rdfile_close(arcn, &fd);
486 purg_lnk(arcn);
487 break;
488 }
489
490 if ((res > 0) || (docrc && (set_crc(arcn, fd) < 0))) {
491 /*
492 * unable to obtain the crc we need, close the file,
493 * purge link table entry
494 */
495 rdfile_close(arcn, &fd);
496 purg_lnk(arcn);
497 continue;
498 }
499
500 if (vflag) {
501 if (vflag > 1)
502 ls_list(arcn, now, listf);
503 else {
504 (void)safe_print(arcn->name, listf);
505 vfpart = 1;
506 }
507 }
508 ++flcnt;
509
510 /*
511 * looks safe to store the file, have the format specific
512 * routine write routine store the file header on the archive
513 */
514 if ((res = (*wrf)(arcn)) < 0) {
515 rdfile_close(arcn, &fd);
516 break;
517 }
518 wr_one = 1;
519 if (res > 0) {
520 /*
521 * format write says no file data needs to be stored
522 * so we are done messing with this file
523 */
524 if (vflag && vfpart) {
525 (void)putc('\n', listf);
526 vfpart = 0;
527 }
528 rdfile_close(arcn, &fd);
529 continue;
530 }
531
532 /*
533 * Add file data to the archive, quit on write error. if we
534 * cannot write the entire file contents to the archive we
535 * must pad the archive to replace the missing file data
536 * (otherwise during an extract the file header for the file
537 * which FOLLOWS this one will not be where we expect it to
538 * be).
539 */
540 res = (*frmt->wr_data)(arcn, fd, &cnt);
541 rdfile_close(arcn, &fd);
542 if (vflag && vfpart) {
543 (void)putc('\n', listf);
544 vfpart = 0;
545 }
546 if (res < 0)
547 break;
548
549 /*
550 * pad as required, cnt is number of bytes not written
551 */
552 if (((cnt > 0) && (wr_skip(cnt) < 0)) ||
553 ((arcn->pad > 0) && (wr_skip(arcn->pad) < 0)))
554 break;
555 }
556
557trailer:
558 /*
559 * tell format to write trailer; pad to block boundary; reset directory
560 * mode/access times, and check if all patterns supplied by the user
561 * were matched. block off signals to avoid chance for multiple entry
562 * into the cleanup code
563 */
564 if (wr_one) {
565 (*frmt->end_wr)();
566 wr_fin();
567 }
568 (void)sigprocmask(SIG_BLOCK, &s_mask, NULL);
569 ar_close();
570 if (tflag)
571 proc_dir();
572 ftree_chk();
573}
574
575/*
576 * append()
577 * Add file to previously written archive. Archive format specified by the
578 * user must agree with archive. The archive is read first to collect
579 * modification times (if -u) and locate the archive trailer. The archive
580 * is positioned in front of the record with the trailer and wr_archive()
581 * is called to add the new members.
582 * PAX IMPLEMENTATION DETAIL NOTE:
583 * -u is implemented by adding the new members to the end of the archive.
584 * Care is taken so that these do not end up as links to the older
585 * version of the same file already stored in the archive. It is expected
586 * when extraction occurs these newer versions will over-write the older
587 * ones stored "earlier" in the archive (this may be a bad assumption as
588 * it depends on the implementation of the program doing the extraction).
589 * It is really difficult to splice in members without either re-writing
590 * the entire archive (from the point were the old version was), or having
591 * assistance of the format specification in terms of a special update
592 * header that invalidates a previous archive record. The posix spec left
593 * the method used to implement -u unspecified. This pax is able to
594 * over write existing files that it creates.
595 */
596
597void
598append(void)
599{
600 ARCHD *arcn;
601 int res;
602 ARCHD archd;
603 FSUB *orgfrmt;
604 int udev;
605 off_t tlen;
606
607 arcn = &archd;
608 orgfrmt = frmt;
609
610 /*
611 * Do not allow an append operation if the actual archive is of a
612 * different format than the user specified format.
613 */
614 if (get_arc() < 0)
615 return;
616 if ((orgfrmt != NULL) && (orgfrmt != frmt)) {
617 paxwarn(1, "Cannot mix current archive format %s with %s",
618 frmt->name, orgfrmt->name);
619 return;
620 }
621
622 /*
623 * pass the format any options and start up format
624 */
625 if (((*frmt->options)() < 0) || ((*frmt->st_rd)() < 0))
626 return;
627
628 /*
629 * if we only are adding members that are newer, we need to save the
630 * mod times for all files we see.
631 */
632 if (uflag && (ftime_start() < 0))
633 return;
634
635 /*
636 * some archive formats encode hard links by recording the device and
637 * file serial number (inode) but copy the file anyway (multiple times)
638 * to the archive. When we append, we run the risk that newly added
639 * files may have the same device and inode numbers as those recorded
640 * on the archive but during a previous run. If this happens, when the
641 * archive is extracted we get INCORRECT hard links. We avoid this by
642 * remapping the device numbers so that newly added files will never
643 * use the same device number as one found on the archive. remapping
644 * allows new members to safely have links among themselves. remapping
645 * also avoids problems with file inode (serial number) truncations
646 * when the inode number is larger than storage space in the archive
647 * header. See the remap routines for more details.
648 */
649 if ((udev = frmt->udev) && (dev_start() < 0))
650 return;
651
652 /*
653 * reading the archive may take a long time. If verbose tell the user
654 */
655 if (vflag) {
656 (void)fprintf(listf,
657 "%s: Reading archive to position at the end...", argv0);
658 vfpart = 1;
659 }
660
661 /*
662 * step through the archive until the format says it is done
663 */
664 while (next_head(arcn) == 0) {
665 /*
666 * check if this file meets user specified options.
667 */
668 if (sel_chk(arcn) != 0) {
669 if (rd_skip(arcn->skip + arcn->pad) == 1)
670 break;
671 continue;
672 }
673
674 if (uflag) {
675 /*
676 * see if this is the newest version of this file has
677 * already been seen, if so skip.
678 */
679 if ((res = chk_ftime(arcn)) < 0)
680 break;
681 if (res > 0) {
682 if (rd_skip(arcn->skip + arcn->pad) == 1)
683 break;
684 continue;
685 }
686 }
687
688 /*
689 * Store this device number. Device numbers seen during the
690 * read phase of append will cause newly appended files with a
691 * device number seen in the old part of the archive to be
692 * remapped to an unused device number.
693 */
694 if ((udev && (add_dev(arcn) < 0)) ||
695 (rd_skip(arcn->skip + arcn->pad) == 1))
696 break;
697 }
698
699 /*
700 * done, finish up read and get the number of bytes to back up so we
701 * can add new members. The format might have used the hard link table,
702 * purge it.
703 */
704 tlen = (*frmt->end_rd)();
705 lnk_end();
706
707 /*
708 * try to position for write, if this fails quit. if any error occurs,
709 * we will refuse to write
710 */
711 if (appnd_start(tlen) < 0)
712 return;
713
714 /*
715 * tell the user we are done reading.
716 */
717 if (vflag && vfpart) {
718 (void)fputs("done.\n", listf);
719 vfpart = 0;
720 }
721
722 /*
723 * go to the writing phase to add the new members
724 */
725 wr_archive(arcn, 1);
726}
727
728/*
729 * archive()
730 * write a new archive
731 */
732
733void
734archive(void)
735{
736 ARCHD archd;
737
738 /*
739 * if we only are adding members that are newer, we need to save the
740 * mod times for all files; set up for writing; pass the format any
741 * options write the archive
742 */
743 if ((uflag && (ftime_start() < 0)) || (wr_start() < 0))
744 return;
745 if ((*frmt->options)() < 0)
746 return;
747
748 wr_archive(&archd, 0);
749}
750
751/*
752 * copy()
753 * copy files from one part of the file system to another. this does not
754 * use any archive storage. The EFFECT OF THE COPY IS THE SAME as if an
755 * archive was written and then extracted in the destination directory
756 * (except the files are forced to be under the destination directory).
757 */
758
759void
760copy(void)
761{
762 ARCHD *arcn;
763 int res;
764 int fddest;
765 char *dest_pt;
766 size_t dlen;
767 size_t drem;
768 int fdsrc = -1;
769 struct stat sb;
770 ARCHD archd;
771 char dirbuf[PAXPATHLEN+1];
772
773 arcn = &archd;
774 /*
775 * set up the destination dir path and make sure it is a directory. We
776 * make sure we have a trailing / on the destination
777 */
778 dlen = strlcpy(dirbuf, dirptr, sizeof(dirbuf));
779 if (dlen >= sizeof(dirbuf) ||
780 (dlen == sizeof(dirbuf) - 1 && dirbuf[dlen - 1] != '/')) {
781 paxwarn(1, "directory name is too long %s", dirptr);
782 return;
783 }
784 dest_pt = dirbuf + dlen;
785 if (*(dest_pt-1) != '/') {
786 *dest_pt++ = '/';
787 *dest_pt = '\0';
788 ++dlen;
789 }
790 drem = PAXPATHLEN - dlen;
791
792 if (stat(dirptr, &sb) < 0) {
793 syswarn(1, errno, "Cannot access destination directory %s",
794 dirptr);
795 return;
796 }
797 if (!S_ISDIR(sb.st_mode)) {
798 paxwarn(1, "Destination is not a directory %s", dirptr);
799 return;
800 }
801
802 /*
803 * start up the hard link table; file traversal routines and the
804 * modification time and access mode database
805 */
806 if ((lnk_start() < 0) || (ftree_start() < 0) || (dir_start() < 0))
807 return;
808
809 /*
810 * When we are doing interactive rename, we store the mapping of names
811 * so we can fix up hard links files later in the archive.
812 */
813 if (iflag && (name_start() < 0))
814 return;
815
816 /*
817 * set up to cp file trees
818 */
819 cp_start();
820
821 /*
822 * while there are files to archive, process them
823 */
824 while (next_file(arcn) == 0) {
825 fdsrc = -1;
826
827 /*
828 * check if this file meets user specified options
829 */
830 if (sel_chk(arcn) != 0)
831 continue;
832
833 /*
834 * if there is already a file in the destination directory with
835 * the same name and it is newer, skip the one stored on the
836 * archive.
837 * NOTE: this test is done BEFORE name modifications as
838 * specified by pax. this can be confusing to the user who
839 * might expect the test to be done on an existing file AFTER
840 * the name mod. In honesty the pax spec is probably flawed in
841 * this respect
842 */
843 if (uflag || Dflag) {
844 /*
845 * create the destination name
846 */
847 if (strlcpy(dest_pt, arcn->name + (*arcn->name == '/'),
848 drem + 1) > drem) {
849 paxwarn(1, "Destination pathname too long %s",
850 arcn->name);
851 continue;
852 }
853
854 /*
855 * if existing file is same age or newer skip
856 */
857 res = lstat(dirbuf, &sb);
858 *dest_pt = '\0';
859
860 if (res == 0) {
861 if (uflag && Dflag) {
862 if ((arcn->sb.st_mtime<=sb.st_mtime) &&
863 (arcn->sb.st_ctime<=sb.st_ctime))
864 continue;
865 } else if (Dflag) {
866 if (arcn->sb.st_ctime <= sb.st_ctime)
867 continue;
868 } else if (arcn->sb.st_mtime <= sb.st_mtime)
869 continue;
870 }
871 }
872
873 /*
874 * this file is considered selected. See if this is a hard link
875 * to a previous file; modify the name as requested by the
876 * user; set the final destination.
877 */
878 ftree_sel(arcn);
879 if ((chk_lnk(arcn) < 0) || ((res = mod_name(arcn)) < 0))
880 break;
881 if ((res > 0) || (set_dest(arcn, dirbuf, dlen) < 0)) {
882 /*
883 * skip file, purge from link table
884 */
885 purg_lnk(arcn);
886 continue;
887 }
888
889 /*
890 * Non standard -Y and -Z flag. When the existing file is
891 * same age or newer skip
892 */
893 if ((Yflag || Zflag) && ((lstat(arcn->name, &sb) == 0))) {
894 if (Yflag && Zflag) {
895 if ((arcn->sb.st_mtime <= sb.st_mtime) &&
896 (arcn->sb.st_ctime <= sb.st_ctime))
897 continue;
898 } else if (Yflag) {
899 if (arcn->sb.st_ctime <= sb.st_ctime)
900 continue;
901 } else if (arcn->sb.st_mtime <= sb.st_mtime)
902 continue;
903 }
904
905 if (vflag) {
906 (void)safe_print(arcn->name, listf);
907 vfpart = 1;
908 }
909 ++flcnt;
910
911 /*
912 * try to create a hard link to the src file if requested
913 * but make sure we are not trying to overwrite ourselves.
914 */
915 if (lflag)
916 res = cross_lnk(arcn);
917 else
918 res = chk_same(arcn);
919 if (res <= 0) {
920 if (vflag && vfpart) {
921 (void)putc('\n', listf);
922 vfpart = 0;
923 }
924 continue;
925 }
926
927 /*
928 * have to create a new file
929 */
930 if ((arcn->type != PAX_REG) && (arcn->type != PAX_CTG)) {
931 /*
932 * create a link or special file
933 */
934 if ((arcn->type == PAX_HLK) || (arcn->type == PAX_HRG))
935 res = lnk_creat(arcn, NULL);
936 else
937 res = node_creat(arcn);
938 if (res < 0)
939 purg_lnk(arcn);
940 if (vflag && vfpart) {
941 (void)putc('\n', listf);
942 vfpart = 0;
943 }
944 continue;
945 }
946
947 /*
948 * have to copy a regular file to the destination directory.
949 * first open source file and then create the destination file
950 */
951 if ((fdsrc = open(arcn->org_name, O_RDONLY, 0)) < 0) {
952 syswarn(1, errno, "Unable to open %s to read",
953 arcn->org_name);
954 purg_lnk(arcn);
955 continue;
956 }
957 if ((fddest = file_creat(arcn)) < 0) {
958 rdfile_close(arcn, &fdsrc);
959 purg_lnk(arcn);
960 continue;
961 }
962
963 /*
964 * copy source file data to the destination file
965 */
966 cp_file(arcn, fdsrc, fddest);
967 file_close(arcn, fddest);
968 rdfile_close(arcn, &fdsrc);
969
970 if (vflag && vfpart) {
971 (void)putc('\n', listf);
972 vfpart = 0;
973 }
974 }
975
976 /*
977 * restore directory modes and times as required; make sure all
978 * patterns were selected block off signals to avoid chance for
979 * multiple entry into the cleanup code.
980 */
981 (void)sigprocmask(SIG_BLOCK, &s_mask, NULL);
982 ar_close();
983 proc_dir();
984 ftree_chk();
985}
986
987/*
988 * next_head()
989 * try to find a valid header in the archive. Uses format specific
990 * routines to extract the header and id the trailer. Trailers may be
991 * located within a valid header or in an invalid header (the location
992 * is format specific. The inhead field from the option table tells us
993 * where to look for the trailer).
994 * We keep reading (and resyncing) until we get enough contiguous data
995 * to check for a header. If we cannot find one, we shift by a byte
996 * add a new byte from the archive to the end of the buffer and try again.
997 * If we get a read error, we throw out what we have (as we must have
998 * contiguous data) and start over again.
999 * ASSUMED: headers fit within a BLKMULT header.
1000 * Return:
1001 * 0 if we got a header, -1 if we are unable to ever find another one
1002 * (we reached the end of input, or we reached the limit on retries. see
1003 * the specs for rd_wrbuf() for more details)
1004 */
1005
1006static int
1007next_head(ARCHD *arcn)
1008{
1009 int ret;
1010 char *hdend;
1011 int res;
1012 int shftsz;
1013 int hsz;
1014 int in_resync = 0; /* set when we are in resync mode */
1015 int cnt = 0; /* counter for trailer function */
1016 int first = 1; /* on 1st read, EOF isn't premature. */
1017
1018 /*
1019 * set up initial conditions, we want a whole frmt->hsz block as we
1020 * have no data yet.
1021 */
1022 res = hsz = frmt->hsz;
1023 hdend = hdbuf;
1024 shftsz = hsz - 1;
1025 for (;;) {
1026 /*
1027 * keep looping until we get a contiguous FULL buffer
1028 * (frmt->hsz is the proper size)
1029 */
1030 for (;;) {
1031 if ((ret = rd_wrbuf(hdend, res)) == res)
1032 break;
1033
1034 /*
1035 * If we read 0 bytes (EOF) from an archive when we
1036 * expect to find a header, we have stepped upon
1037 * an archive without the customary block of zeroes
1038 * end marker. It's just stupid to error out on
1039 * them, so exit gracefully.
1040 */
1041 if (first && ret == 0)
1042 return(-1);
1043 first = 0;
1044
1045 /*
1046 * some kind of archive read problem, try to resync the
1047 * storage device, better give the user the bad news.
1048 */
1049 if ((ret == 0) || (rd_sync() < 0)) {
1050 paxwarn(1,"Premature end of file on archive read");
1051 return(-1);
1052 }
1053 if (!in_resync) {
1054 if (act == APPND) {
1055 paxwarn(1,
1056 "Archive I/O error, cannot continue");
1057 return(-1);
1058 }
1059 paxwarn(1,"Archive I/O error. Trying to recover.");
1060 ++in_resync;
1061 }
1062
1063 /*
1064 * oh well, throw it all out and start over
1065 */
1066 res = hsz;
1067 hdend = hdbuf;
1068 }
1069
1070 /*
1071 * ok we have a contiguous buffer of the right size. Call the
1072 * format read routine. If this was not a valid header and this
1073 * format stores trailers outside of the header, call the
1074 * format specific trailer routine to check for a trailer. We
1075 * have to watch out that we do not mis-identify file data or
1076 * block padding as a header or trailer. Format specific
1077 * trailer functions must NOT check for the trailer while we
1078 * are running in resync mode. Some trailer functions may tell
1079 * us that this block cannot contain a valid header either, so
1080 * we then throw out the entire block and start over.
1081 */
1082 if ((*frmt->rd)(arcn, hdbuf) == 0)
1083 break;
1084
1085 if (!frmt->inhead) {
1086 /*
1087 * this format has trailers outside of valid headers
1088 */
1089 if ((ret = (*frmt->trail)(arcn,hdbuf,in_resync,&cnt)) == 0){
1090 /*
1091 * valid trailer found, drain input as required
1092 */
1093 ar_drain();
1094 return(-1);
1095 }
1096
1097 if (ret == 1) {
1098 /*
1099 * we are in resync and we were told to throw
1100 * the whole block out because none of the
1101 * bytes in this block can be used to form a
1102 * valid header
1103 */
1104 res = hsz;
1105 hdend = hdbuf;
1106 continue;
1107 }
1108 }
1109
1110 /*
1111 * Brute force section.
1112 * not a valid header. We may be able to find a header yet. So
1113 * we shift over by one byte, and set up to read one byte at a
1114 * time from the archive and place it at the end of the buffer.
1115 * We will keep moving byte at a time until we find a header or
1116 * get a read error and have to start over.
1117 */
1118 if (!in_resync) {
1119 if (act == APPND) {
1120 paxwarn(1,"Unable to append, archive header flaw");
1121 return(-1);
1122 }
1123 paxwarn(1,"Invalid header, starting valid header search.");
1124 ++in_resync;
1125 }
1126 memmove(hdbuf, hdbuf+1, shftsz);
1127 res = 1;
1128 hdend = hdbuf + shftsz;
1129 }
1130
1131 /*
1132 * ok got a valid header, check for trailer if format encodes it in the
1133 * the header. NOTE: the parameters are different than trailer routines
1134 * which encode trailers outside of the header!
1135 */
1136 if (frmt->inhead && ((*frmt->trail)(arcn,NULL,0,NULL) == 0)) {
1137 /*
1138 * valid trailer found, drain input as required
1139 */
1140 ar_drain();
1141 return(-1);
1142 }
1143
1144 ++flcnt;
1145 return(0);
1146}
1147
1148/*
1149 * get_arc()
1150 * Figure out what format an archive is. Handles archive with flaws by
1151 * brute force searches for a legal header in any supported format. The
1152 * format id routines have to be careful to NOT mis-identify a format.
1153 * ASSUMED: headers fit within a BLKMULT header.
1154 * Return:
1155 * 0 if archive found -1 otherwise
1156 */
1157
1158static int
1159get_arc(void)
1160{
1161 int i;
1162 int hdsz = 0;
1163 int res;
1164 int minhd = BLKMULT;
1165 char *hdend;
1166 int notice = 0;
1167
1168 /*
1169 * find the smallest header size in all archive formats and then set up
1170 * to read the archive.
1171 */
1172 for (i = 0; ford[i] >= 0; ++i) {
1173 if (fsub[ford[i]].hsz < minhd)
1174 minhd = fsub[ford[i]].hsz;
1175 }
1176 if (rd_start() < 0)
1177 return(-1);
1178 res = BLKMULT;
1179 hdsz = 0;
1180 hdend = hdbuf;
1181 for (;;) {
1182 for (;;) {
1183 /*
1184 * fill the buffer with at least the smallest header
1185 */
1186 i = rd_wrbuf(hdend, res);
1187 if (i > 0)
1188 hdsz += i;
1189 if (hdsz >= minhd)
1190 break;
1191
1192 /*
1193 * if we cannot recover from a read error quit
1194 */
1195 if ((i == 0) || (rd_sync() < 0))
1196 goto out;
1197
1198 /*
1199 * when we get an error none of the data we already
1200 * have can be used to create a legal header (we just
1201 * got an error in the middle), so we throw it all out
1202 * and refill the buffer with fresh data.
1203 */
1204 res = BLKMULT;
1205 hdsz = 0;
1206 hdend = hdbuf;
1207 if (!notice) {
1208 if (act == APPND)
1209 return(-1);
1210 paxwarn(1,"Cannot identify format. Searching...");
1211 ++notice;
1212 }
1213 }
1214
1215 /*
1216 * we have at least the size of the smallest header in any
1217 * archive format. Look to see if we have a match. The array
1218 * ford[] is used to specify the header id order to reduce the
1219 * chance of incorrectly id'ing a valid header (some formats
1220 * may be subsets of each other and the order would then be
1221 * important).
1222 */
1223 for (i = 0; ford[i] >= 0; ++i) {
1224 if ((*fsub[ford[i]].id)(hdbuf, hdsz) < 0)
1225 continue;
1226 frmt = &(fsub[ford[i]]);
1227 /*
1228 * yuck, to avoid slow special case code in the extract
1229 * routines, just push this header back as if it was
1230 * not seen. We have left extra space at start of the
1231 * buffer for this purpose. This is a bit ugly, but
1232 * adding all the special case code is far worse.
1233 */
1234 pback(hdbuf, hdsz);
1235 return(0);
1236 }
1237
1238 /*
1239 * We have a flawed archive, no match. we start searching, but
1240 * we never allow additions to flawed archives
1241 */
1242 if (!notice) {
1243 if (act == APPND)
1244 return(-1);
1245 paxwarn(1, "Cannot identify format. Searching...");
1246 ++notice;
1247 }
1248
1249 /*
1250 * brute force search for a header that we can id.
1251 * we shift through byte at a time. this is slow, but we cannot
1252 * determine the nature of the flaw in the archive in a
1253 * portable manner
1254 */
1255 if (--hdsz > 0) {
1256 memmove(hdbuf, hdbuf+1, hdsz);
1257 res = BLKMULT - hdsz;
1258 hdend = hdbuf + hdsz;
1259 } else {
1260 res = BLKMULT;
1261 hdend = hdbuf;
1262 hdsz = 0;
1263 }
1264 }
1265
1266 out:
1267 /*
1268 * we cannot find a header, bow, apologize and quit
1269 */
1270 paxwarn(1, "Sorry, unable to determine archive format.");
1271 return(-1);
1272}
Note: See TracBrowser for help on using the repository browser.