file_cmds-116.9.tar.gz
[apple/file_cmds.git] / pax / tar.c
1 /* $OpenBSD: tar.c,v 1.12 1997/09/01 18:30:03 deraadt Exp $ */
2 /* $NetBSD: tar.c,v 1.5 1995/03/21 09:07:49 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. All advertising materials mentioning features or use of this software
21 * must display the following acknowledgement:
22 * This product includes software developed by the University of
23 * California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 * may be used to endorse or promote products derived from this software
26 * without specific prior written permission.
27 *
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 * SUCH DAMAGE.
39 */
40
41 #ifndef lint
42 #if 0
43 static char sccsid[] = "@(#)tar.c 8.2 (Berkeley) 4/18/94";
44 #else
45 static char rcsid[] __attribute__((__unused__)) = "$OpenBSD: tar.c,v 1.12 1997/09/01 18:30:03 deraadt Exp $";
46 #endif
47 #endif /* not lint */
48
49 #include <sys/types.h>
50 #include <sys/time.h>
51 #include <sys/stat.h>
52 #include <sys/param.h>
53 #include <string.h>
54 #include <stdio.h>
55 #include <unistd.h>
56 #include <stdlib.h>
57 #include "pax.h"
58 #include "extern.h"
59 #include "tar.h"
60
61 /*
62 * Routines for reading, writing and header identify of various versions of tar
63 */
64
65 static u_long tar_chksm __P((register char *, register int));
66 static char *name_split __P((register char *, register int));
67 static int ul_oct __P((u_long, register char *, register int, int));
68 #ifndef NET2_STAT
69 static int uqd_oct __P((u_quad_t, register char *, register int, int));
70 #endif
71
72 /*
73 * Routines common to all versions of tar
74 */
75
76 static int tar_nodir; /* do not write dirs under old tar */
77
78 /*
79 * tar_endwr()
80 * add the tar trailer of two null blocks
81 * Return:
82 * 0 if ok, -1 otherwise (what wr_skip returns)
83 */
84
85 #ifdef __STDC__
86 int
87 tar_endwr(void)
88 #else
89 int
90 tar_endwr()
91 #endif
92 {
93 return(wr_skip((off_t)(NULLCNT*BLKMULT)));
94 }
95
96 /*
97 * tar_endrd()
98 * no cleanup needed here, just return size of trailer (for append)
99 * Return:
100 * size of trailer (2 * BLKMULT)
101 */
102
103 #ifdef __STDC__
104 off_t
105 tar_endrd(void)
106 #else
107 off_t
108 tar_endrd()
109 #endif
110 {
111 return((off_t)(NULLCNT*BLKMULT));
112 }
113
114 /*
115 * tar_trail()
116 * Called to determine if a header block is a valid trailer. We are passed
117 * the block, the in_sync flag (which tells us we are in resync mode;
118 * looking for a valid header), and cnt (which starts at zero) which is
119 * used to count the number of empty blocks we have seen so far.
120 * Return:
121 * 0 if a valid trailer, -1 if not a valid trailer, or 1 if the block
122 * could never contain a header.
123 */
124
125 #ifdef __STDC__
126 int
127 tar_trail(register char *buf, register int in_resync, register int *cnt)
128 #else
129 int
130 tar_trail(buf, in_resync, cnt)
131 register char *buf;
132 register int in_resync;
133 register int *cnt;
134 #endif
135 {
136 register int i;
137
138 /*
139 * look for all zero, trailer is two consecutive blocks of zero
140 */
141 for (i = 0; i < BLKMULT; ++i) {
142 if (buf[i] != '\0')
143 break;
144 }
145
146 /*
147 * if not all zero it is not a trailer, but MIGHT be a header.
148 */
149 if (i != BLKMULT)
150 return(-1);
151
152 /*
153 * When given a zero block, we must be careful!
154 * If we are not in resync mode, check for the trailer. Have to watch
155 * out that we do not mis-identify file data as the trailer, so we do
156 * NOT try to id a trailer during resync mode. During resync mode we
157 * might as well throw this block out since a valid header can NEVER be
158 * a block of all 0 (we must have a valid file name).
159 */
160 if (!in_resync && (++*cnt >= NULLCNT))
161 return(0);
162 return(1);
163 }
164
165 /*
166 * ul_oct()
167 * convert an unsigned long to an octal string. many oddball field
168 * termination characters are used by the various versions of tar in the
169 * different fields. term selects which kind to use. str is '0' padded
170 * at the front to len. we are unable to use only one format as many old
171 * tar readers are very cranky about this.
172 * Return:
173 * 0 if the number fit into the string, -1 otherwise
174 */
175
176 #ifdef __STDC__
177 static int
178 ul_oct(u_long val, register char *str, register int len, int term)
179 #else
180 static int
181 ul_oct(val, str, len, term)
182 u_long val;
183 register char *str;
184 register int len;
185 int term;
186 #endif
187 {
188 register char *pt;
189
190 /*
191 * term selects the appropriate character(s) for the end of the string
192 */
193 pt = str + len - 1;
194 switch(term) {
195 case 3:
196 *pt-- = '\0';
197 break;
198 case 2:
199 *pt-- = ' ';
200 *pt-- = '\0';
201 break;
202 case 1:
203 *pt-- = ' ';
204 break;
205 case 0:
206 default:
207 *pt-- = '\0';
208 *pt-- = ' ';
209 break;
210 }
211
212 /*
213 * convert and blank pad if there is space
214 */
215 while (pt >= str) {
216 *pt-- = '0' + (char)(val & 0x7);
217 if ((val = val >> 3) == (u_long)0)
218 break;
219 }
220
221 while (pt >= str)
222 *pt-- = '0';
223 if (val != (u_long)0)
224 return(-1);
225 return(0);
226 }
227
228 #ifndef NET2_STAT
229 /*
230 * uqd_oct()
231 * convert an u_quad_t to an octal string. one of many oddball field
232 * termination characters are used by the various versions of tar in the
233 * different fields. term selects which kind to use. str is '0' padded
234 * at the front to len. we are unable to use only one format as many old
235 * tar readers are very cranky about this.
236 * Return:
237 * 0 if the number fit into the string, -1 otherwise
238 */
239
240 #ifdef __STDC__
241 static int
242 uqd_oct(u_quad_t val, register char *str, register int len, int term)
243 #else
244 static int
245 uqd_oct(val, str, len, term)
246 u_quad_t val;
247 register char *str;
248 register int len;
249 int term;
250 #endif
251 {
252 register char *pt;
253
254 /*
255 * term selects the appropriate character(s) for the end of the string
256 */
257 pt = str + len - 1;
258 switch(term) {
259 case 3:
260 *pt-- = '\0';
261 break;
262 case 2:
263 *pt-- = ' ';
264 *pt-- = '\0';
265 break;
266 case 1:
267 *pt-- = ' ';
268 break;
269 case 0:
270 default:
271 *pt-- = '\0';
272 *pt-- = ' ';
273 break;
274 }
275
276 /*
277 * convert and blank pad if there is space
278 */
279 while (pt >= str) {
280 *pt-- = '0' + (char)(val & 0x7);
281 if ((val = val >> 3) == 0)
282 break;
283 }
284
285 while (pt >= str)
286 *pt-- = '0';
287 if (val != (u_quad_t)0)
288 return(-1);
289 return(0);
290 }
291 #endif
292
293 /*
294 * tar_chksm()
295 * calculate the checksum for a tar block counting the checksum field as
296 * all blanks (BLNKSUM is that value pre-calculated, the sume of 8 blanks).
297 * NOTE: we use len to short circuit summing 0's on write since we ALWAYS
298 * pad headers with 0.
299 * Return:
300 * unsigned long checksum
301 */
302
303 #ifdef __STDC__
304 static u_long
305 tar_chksm(register char *blk, register int len)
306 #else
307 static u_long
308 tar_chksm(blk, len)
309 register char *blk;
310 register int len;
311 #endif
312 {
313 register char *stop;
314 register char *pt;
315 u_long chksm = BLNKSUM; /* inital value is checksum field sum */
316
317 /*
318 * add the part of the block before the checksum field
319 */
320 pt = blk;
321 stop = blk + CHK_OFFSET;
322 while (pt < stop)
323 chksm += (u_long)(*pt++ & 0xff);
324 /*
325 * move past the checksum field and keep going, spec counts the
326 * checksum field as the sum of 8 blanks (which is pre-computed as
327 * BLNKSUM).
328 * ASSUMED: len is greater than CHK_OFFSET. (len is where our 0 padding
329 * starts, no point in summing zero's)
330 */
331 pt += CHK_LEN;
332 stop = blk + len;
333 while (pt < stop)
334 chksm += (u_long)(*pt++ & 0xff);
335 return(chksm);
336 }
337
338 /*
339 * Routines for old BSD style tar (also made portable to sysV tar)
340 */
341
342 /*
343 * tar_id()
344 * determine if a block given to us is a valid tar header (and not a USTAR
345 * header). We have to be on the lookout for those pesky blocks of all
346 * zero's.
347 * Return:
348 * 0 if a tar header, -1 otherwise
349 */
350
351 #ifdef __STDC__
352 int
353 tar_id(register char *blk, int size)
354 #else
355 int
356 tar_id(blk, size)
357 register char *blk;
358 int size;
359 #endif
360 {
361 register HD_TAR *hd;
362 register HD_USTAR *uhd;
363
364 if (size < BLKMULT)
365 return(-1);
366 hd = (HD_TAR *)blk;
367 uhd = (HD_USTAR *)blk;
368
369 /*
370 * check for block of zero's first, a simple and fast test, then make
371 * sure this is not a ustar header by looking for the ustar magic
372 * cookie. We should use TMAGLEN, but some USTAR archive programs are
373 * wrong and create archives missing the \0. Last we check the
374 * checksum. If this is ok we have to assume it is a valid header.
375 */
376 if (hd->name[0] == '\0')
377 return(-1);
378 if (strncmp(uhd->magic, TMAGIC, TMAGLEN - 1) == 0)
379 return(-1);
380 if (asc_ul(hd->chksum,sizeof(hd->chksum),OCT) != tar_chksm(blk,BLKMULT))
381 return(-1);
382 return(0);
383 }
384
385 /*
386 * tar_opt()
387 * handle tar format specific -o options
388 * Return:
389 * 0 if ok -1 otherwise
390 */
391
392 #ifdef __STDC__
393 int
394 tar_opt(void)
395 #else
396 int
397 tar_opt()
398 #endif
399 {
400 OPLIST *opt;
401
402 while ((opt = opt_next()) != NULL) {
403 if (strcmp(opt->name, TAR_OPTION) ||
404 strcmp(opt->value, TAR_NODIR)) {
405 paxwarn(1, "Unknown tar format -o option/value pair %s=%s",
406 opt->name, opt->value);
407 paxwarn(1,"%s=%s is the only supported tar format option",
408 TAR_OPTION, TAR_NODIR);
409 return(-1);
410 }
411
412 /*
413 * we only support one option, and only when writing
414 */
415 if ((act != APPND) && (act != ARCHIVE)) {
416 paxwarn(1, "%s=%s is only supported when writing.",
417 opt->name, opt->value);
418 return(-1);
419 }
420 tar_nodir = 1;
421 }
422 return(0);
423 }
424
425
426 /*
427 * tar_rd()
428 * extract the values out of block already determined to be a tar header.
429 * store the values in the ARCHD parameter.
430 * Return:
431 * 0
432 */
433
434 #ifdef __STDC__
435 int
436 tar_rd(register ARCHD *arcn, register char *buf)
437 #else
438 int
439 tar_rd(arcn, buf)
440 register ARCHD *arcn;
441 register char *buf;
442 #endif
443 {
444 register HD_TAR *hd;
445 register char *pt;
446
447 /*
448 * we only get proper sized buffers passed to us
449 */
450 if (tar_id(buf, BLKMULT) < 0)
451 return(-1);
452 arcn->org_name = arcn->name;
453 arcn->sb.st_nlink = 1;
454 arcn->pat = NULL;
455
456 /*
457 * copy out the name and values in the stat buffer
458 */
459 hd = (HD_TAR *)buf;
460 arcn->nlen = l_strncpy(arcn->name, hd->name, sizeof(arcn->name) - 1);
461 arcn->name[arcn->nlen] = '\0';
462 arcn->sb.st_mode = (mode_t)(asc_ul(hd->mode,sizeof(hd->mode),OCT) &
463 0xfff);
464 arcn->sb.st_uid = (uid_t)asc_ul(hd->uid, sizeof(hd->uid), OCT);
465 arcn->sb.st_gid = (gid_t)asc_ul(hd->gid, sizeof(hd->gid), OCT);
466 arcn->sb.st_size = (size_t)asc_ul(hd->size, sizeof(hd->size), OCT);
467 arcn->sb.st_mtime = (time_t)asc_ul(hd->mtime, sizeof(hd->mtime), OCT);
468 arcn->sb.st_ctime = arcn->sb.st_atime = arcn->sb.st_mtime;
469
470 /*
471 * have to look at the last character, it may be a '/' and that is used
472 * to encode this as a directory
473 */
474 pt = &(arcn->name[arcn->nlen - 1]);
475 arcn->pad = 0;
476 arcn->skip = 0;
477 switch(hd->linkflag) {
478 case SYMTYPE:
479 /*
480 * symbolic link, need to get the link name and set the type in
481 * the st_mode so -v printing will look correct.
482 */
483 arcn->type = PAX_SLK;
484 arcn->ln_nlen = l_strncpy(arcn->ln_name, hd->linkname,
485 sizeof(arcn->ln_name) - 1);
486 arcn->ln_name[arcn->ln_nlen] = '\0';
487 arcn->sb.st_mode |= S_IFLNK;
488 break;
489 case LNKTYPE:
490 /*
491 * hard link, need to get the link name, set the type in the
492 * st_mode and st_nlink so -v printing will look better.
493 */
494 arcn->type = PAX_HLK;
495 arcn->sb.st_nlink = 2;
496 arcn->ln_nlen = l_strncpy(arcn->ln_name, hd->linkname,
497 sizeof(arcn->ln_name) - 1);
498 arcn->ln_name[arcn->ln_nlen] = '\0';
499
500 /*
501 * no idea of what type this thing really points at, but
502 * we set something for printing only.
503 */
504 arcn->sb.st_mode |= S_IFREG;
505 break;
506 case DIRTYPE:
507 /*
508 * It is a directory, set the mode for -v printing
509 */
510 arcn->type = PAX_DIR;
511 arcn->sb.st_mode |= S_IFDIR;
512 arcn->sb.st_nlink = 2;
513 arcn->ln_name[0] = '\0';
514 arcn->ln_nlen = 0;
515 break;
516 case AREGTYPE:
517 case REGTYPE:
518 default:
519 /*
520 * If we have a trailing / this is a directory and NOT a file.
521 */
522 arcn->ln_name[0] = '\0';
523 arcn->ln_nlen = 0;
524 if (*pt == '/') {
525 /*
526 * it is a directory, set the mode for -v printing
527 */
528 arcn->type = PAX_DIR;
529 arcn->sb.st_mode |= S_IFDIR;
530 arcn->sb.st_nlink = 2;
531 } else {
532 /*
533 * have a file that will be followed by data. Set the
534 * skip value to the size field and caluculate the size
535 * of the padding.
536 */
537 arcn->type = PAX_REG;
538 arcn->sb.st_mode |= S_IFREG;
539 arcn->pad = TAR_PAD(arcn->sb.st_size);
540 arcn->skip = arcn->sb.st_size;
541 }
542 break;
543 }
544
545 /*
546 * strip off any trailing slash.
547 */
548 if (*pt == '/') {
549 *pt = '\0';
550 --arcn->nlen;
551 }
552 return(0);
553 }
554
555 /*
556 * tar_wr()
557 * write a tar header for the file specified in the ARCHD to the archive.
558 * Have to check for file types that cannot be stored and file names that
559 * are too long. Be careful of the term (last arg) to ul_oct, each field
560 * of tar has it own spec for the termination character(s).
561 * ASSUMED: space after header in header block is zero filled
562 * Return:
563 * 0 if file has data to be written after the header, 1 if file has NO
564 * data to write after the header, -1 if archive write failed
565 */
566
567 #ifdef __STDC__
568 int
569 tar_wr(register ARCHD *arcn)
570 #else
571 int
572 tar_wr(arcn)
573 register ARCHD *arcn;
574 #endif
575 {
576 register HD_TAR *hd;
577 int len;
578 char hdblk[sizeof(HD_TAR)];
579
580 /*
581 * check for those file system types which tar cannot store
582 */
583 switch(arcn->type) {
584 case PAX_DIR:
585 /*
586 * user asked that dirs not be written to the archive
587 */
588 if (tar_nodir)
589 return(1);
590 break;
591 case PAX_CHR:
592 paxwarn(1, "Tar cannot archive a character device %s",
593 arcn->org_name);
594 return(1);
595 case PAX_BLK:
596 paxwarn(1, "Tar cannot archive a block device %s", arcn->org_name);
597 return(1);
598 case PAX_SCK:
599 paxwarn(1, "Tar cannot archive a socket %s", arcn->org_name);
600 return(1);
601 case PAX_FIF:
602 paxwarn(1, "Tar cannot archive a fifo %s", arcn->org_name);
603 return(1);
604 case PAX_SLK:
605 case PAX_HLK:
606 case PAX_HRG:
607 if (arcn->ln_nlen > sizeof(hd->linkname)) {
608 paxwarn(1,"Link name too long for tar %s", arcn->ln_name);
609 return(1);
610 }
611 break;
612 case PAX_REG:
613 case PAX_CTG:
614 default:
615 break;
616 }
617
618 /*
619 * check file name len, remember extra char for dirs (the / at the end)
620 */
621 len = arcn->nlen;
622 if (arcn->type == PAX_DIR)
623 ++len;
624 if (len >= sizeof(hd->name)) {
625 paxwarn(1, "File name too long for tar %s", arcn->name);
626 return(1);
627 }
628
629 /*
630 * copy the data out of the ARCHD into the tar header based on the type
631 * of the file. Remember many tar readers want the unused fields to be
632 * padded with zero. We set the linkflag field (type), the linkname
633 * (or zero if not used),the size, and set the padding (if any) to be
634 * added after the file data (0 for all other types, as they only have
635 * a header)
636 */
637 hd = (HD_TAR *)hdblk;
638 l_strncpy(hd->name, arcn->name, sizeof(hd->name) - 1);
639 hd->name[sizeof(hd->name) - 1] = '\0';
640 arcn->pad = 0;
641
642 if (arcn->type == PAX_DIR) {
643 /*
644 * directories are the same as files, except have a filename
645 * that ends with a /, we add the slash here. No data follows,
646 * dirs, so no pad.
647 */
648 hd->linkflag = AREGTYPE;
649 memset(hd->linkname, 0, sizeof(hd->linkname));
650 hd->name[len-1] = '/';
651 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 1))
652 goto out;
653 } else if (arcn->type == PAX_SLK) {
654 /*
655 * no data follows this file, so no pad
656 */
657 hd->linkflag = SYMTYPE;
658 l_strncpy(hd->linkname,arcn->ln_name, sizeof(hd->linkname) - 1);
659 hd->linkname[sizeof(hd->linkname) - 1] = '\0';
660 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 1))
661 goto out;
662 } else if ((arcn->type == PAX_HLK) || (arcn->type == PAX_HRG)) {
663 /*
664 * no data follows this file, so no pad
665 */
666 hd->linkflag = LNKTYPE;
667 l_strncpy(hd->linkname,arcn->ln_name, sizeof(hd->linkname) - 1);
668 hd->linkname[sizeof(hd->linkname) - 1] = '\0';
669 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 1))
670 goto out;
671 } else {
672 /*
673 * data follows this file, so set the pad
674 */
675 hd->linkflag = AREGTYPE;
676 memset(hd->linkname, 0, sizeof(hd->linkname));
677 # ifdef NET2_STAT
678 if (ul_oct((u_long)arcn->sb.st_size, hd->size,
679 sizeof(hd->size), 1)) {
680 # else
681 if (uqd_oct((u_quad_t)arcn->sb.st_size, hd->size,
682 sizeof(hd->size), 1)) {
683 # endif
684 paxwarn(1,"File is too large for tar %s", arcn->org_name);
685 return(1);
686 }
687 arcn->pad = TAR_PAD(arcn->sb.st_size);
688 }
689
690 /*
691 * copy those fields that are independent of the type
692 */
693 if (ul_oct((u_long)arcn->sb.st_mode, hd->mode, sizeof(hd->mode), 0) ||
694 ul_oct((u_long)arcn->sb.st_uid, hd->uid, sizeof(hd->uid), 0) ||
695 ul_oct((u_long)arcn->sb.st_gid, hd->gid, sizeof(hd->gid), 0) ||
696 ul_oct((u_long)arcn->sb.st_mtime, hd->mtime, sizeof(hd->mtime), 1))
697 goto out;
698
699 /*
700 * calculate and add the checksum, then write the header. A return of
701 * 0 tells the caller to now write the file data, 1 says no data needs
702 * to be written
703 */
704 if (ul_oct(tar_chksm(hdblk, sizeof(HD_TAR)), hd->chksum,
705 sizeof(hd->chksum), 3))
706 goto out;
707 if (wr_rdbuf(hdblk, sizeof(HD_TAR)) < 0)
708 return(-1);
709 if (wr_skip((off_t)(BLKMULT - sizeof(HD_TAR))) < 0)
710 return(-1);
711 if ((arcn->type == PAX_CTG) || (arcn->type == PAX_REG))
712 return(0);
713 return(1);
714
715 out:
716 /*
717 * header field is out of range
718 */
719 paxwarn(1, "Tar header field is too small for %s", arcn->org_name);
720 return(1);
721 }
722
723 /*
724 * Routines for POSIX ustar
725 */
726
727 /*
728 * ustar_strd()
729 * initialization for ustar read
730 * Return:
731 * 0 if ok, -1 otherwise
732 */
733
734 #ifdef __STDC__
735 int
736 ustar_strd(void)
737 #else
738 int
739 ustar_strd()
740 #endif
741 {
742 if ((usrtb_start() < 0) || (grptb_start() < 0))
743 return(-1);
744 return(0);
745 }
746
747 /*
748 * ustar_stwr()
749 * initialization for ustar write
750 * Return:
751 * 0 if ok, -1 otherwise
752 */
753
754 #ifdef __STDC__
755 int
756 ustar_stwr(void)
757 #else
758 int
759 ustar_stwr()
760 #endif
761 {
762 if ((uidtb_start() < 0) || (gidtb_start() < 0))
763 return(-1);
764 return(0);
765 }
766
767 /*
768 * ustar_id()
769 * determine if a block given to us is a valid ustar header. We have to
770 * be on the lookout for those pesky blocks of all zero's
771 * Return:
772 * 0 if a ustar header, -1 otherwise
773 */
774
775 #ifdef __STDC__
776 int
777 ustar_id(char *blk, int size)
778 #else
779 int
780 ustar_id(blk, size)
781 char *blk;
782 int size;
783 #endif
784 {
785 register HD_USTAR *hd;
786
787 if (size < BLKMULT)
788 return(-1);
789 hd = (HD_USTAR *)blk;
790
791 /*
792 * check for block of zero's first, a simple and fast test then check
793 * ustar magic cookie. We should use TMAGLEN, but some USTAR archive
794 * programs are fouled up and create archives missing the \0. Last we
795 * check the checksum. If ok we have to assume it is a valid header.
796 */
797 if (hd->name[0] == '\0')
798 return(-1);
799 if (strncmp(hd->magic, TMAGIC, TMAGLEN - 1) != 0)
800 return(-1);
801 if (asc_ul(hd->chksum,sizeof(hd->chksum),OCT) != tar_chksm(blk,BLKMULT))
802 return(-1);
803 return(0);
804 }
805
806 /*
807 * ustar_rd()
808 * extract the values out of block already determined to be a ustar header.
809 * store the values in the ARCHD parameter.
810 * Return:
811 * 0
812 */
813
814 #ifdef __STDC__
815 int
816 ustar_rd(register ARCHD *arcn, register char *buf)
817 #else
818 int
819 ustar_rd(arcn, buf)
820 register ARCHD *arcn;
821 register char *buf;
822 #endif
823 {
824 register HD_USTAR *hd;
825 register char *dest;
826 register int cnt = 0;
827 dev_t devmajor;
828 dev_t devminor;
829
830 /*
831 * we only get proper sized buffers
832 */
833 if (ustar_id(buf, BLKMULT) < 0)
834 return(-1);
835 arcn->org_name = arcn->name;
836 arcn->sb.st_nlink = 1;
837 arcn->pat = NULL;
838 arcn->nlen = 0;
839 hd = (HD_USTAR *)buf;
840
841 /*
842 * see if the filename is split into two parts. if, so joint the parts.
843 * we copy the prefix first and add a / between the prefix and name.
844 */
845 dest = arcn->name;
846 if (*(hd->prefix) != '\0') {
847 cnt = l_strncpy(dest, hd->prefix, sizeof(arcn->name) - 2);
848 dest += cnt;
849 *dest++ = '/';
850 cnt++;
851 }
852 arcn->nlen = cnt + l_strncpy(dest, hd->name, sizeof(arcn->name) - cnt);
853 arcn->name[arcn->nlen] = '\0';
854
855 /*
856 * follow the spec to the letter. we should only have mode bits, strip
857 * off all other crud we may be passed.
858 */
859 arcn->sb.st_mode = (mode_t)(asc_ul(hd->mode, sizeof(hd->mode), OCT) &
860 0xfff);
861 arcn->sb.st_size = (size_t)asc_ul(hd->size, sizeof(hd->size), OCT);
862 arcn->sb.st_mtime = (time_t)asc_ul(hd->mtime, sizeof(hd->mtime), OCT);
863 arcn->sb.st_ctime = arcn->sb.st_atime = arcn->sb.st_mtime;
864
865 /*
866 * If we can find the ascii names for gname and uname in the password
867 * and group files we will use the uid's and gid they bind. Otherwise
868 * we use the uid and gid values stored in the header. (This is what
869 * the posix spec wants).
870 */
871 hd->gname[sizeof(hd->gname) - 1] = '\0';
872 if (gid_name(hd->gname, &(arcn->sb.st_gid)) < 0)
873 arcn->sb.st_gid = (gid_t)asc_ul(hd->gid, sizeof(hd->gid), OCT);
874 hd->uname[sizeof(hd->uname) - 1] = '\0';
875 if (uid_name(hd->uname, &(arcn->sb.st_uid)) < 0)
876 arcn->sb.st_uid = (uid_t)asc_ul(hd->uid, sizeof(hd->uid), OCT);
877
878 /*
879 * set the defaults, these may be changed depending on the file type
880 */
881 arcn->ln_name[0] = '\0';
882 arcn->ln_nlen = 0;
883 arcn->pad = 0;
884 arcn->skip = 0;
885 arcn->sb.st_rdev = (dev_t)0;
886
887 /*
888 * set the mode and PAX type according to the typeflag in the header
889 */
890 switch(hd->typeflag) {
891 case FIFOTYPE:
892 arcn->type = PAX_FIF;
893 arcn->sb.st_mode |= S_IFIFO;
894 break;
895 case DIRTYPE:
896 arcn->type = PAX_DIR;
897 arcn->sb.st_mode |= S_IFDIR;
898 arcn->sb.st_nlink = 2;
899
900 /*
901 * Some programs that create ustar archives append a '/'
902 * to the pathname for directories. This clearly violates
903 * ustar specs, but we will silently strip it off anyway.
904 */
905 if (arcn->name[arcn->nlen - 1] == '/')
906 arcn->name[--arcn->nlen] = '\0';
907 break;
908 case BLKTYPE:
909 case CHRTYPE:
910 /*
911 * this type requires the rdev field to be set.
912 */
913 if (hd->typeflag == BLKTYPE) {
914 arcn->type = PAX_BLK;
915 arcn->sb.st_mode |= S_IFBLK;
916 } else {
917 arcn->type = PAX_CHR;
918 arcn->sb.st_mode |= S_IFCHR;
919 }
920 devmajor = (dev_t)asc_ul(hd->devmajor,sizeof(hd->devmajor),OCT);
921 devminor = (dev_t)asc_ul(hd->devminor,sizeof(hd->devminor),OCT);
922 arcn->sb.st_rdev = TODEV(devmajor, devminor);
923 break;
924 case SYMTYPE:
925 case LNKTYPE:
926 if (hd->typeflag == SYMTYPE) {
927 arcn->type = PAX_SLK;
928 arcn->sb.st_mode |= S_IFLNK;
929 } else {
930 arcn->type = PAX_HLK;
931 /*
932 * so printing looks better
933 */
934 arcn->sb.st_mode |= S_IFREG;
935 arcn->sb.st_nlink = 2;
936 }
937 /*
938 * copy the link name
939 */
940 arcn->ln_nlen = l_strncpy(arcn->ln_name, hd->linkname,
941 sizeof(arcn->ln_name) - 1);
942 arcn->ln_name[arcn->ln_nlen] = '\0';
943 break;
944 case CONTTYPE:
945 case AREGTYPE:
946 case REGTYPE:
947 default:
948 /*
949 * these types have file data that follows. Set the skip and
950 * pad fields.
951 */
952 arcn->type = PAX_REG;
953 arcn->pad = TAR_PAD(arcn->sb.st_size);
954 arcn->skip = arcn->sb.st_size;
955 arcn->sb.st_mode |= S_IFREG;
956 break;
957 }
958 return(0);
959 }
960
961 /*
962 * ustar_wr()
963 * write a ustar header for the file specified in the ARCHD to the archive
964 * Have to check for file types that cannot be stored and file names that
965 * are too long. Be careful of the term (last arg) to ul_oct, we only use
966 * '\0' for the termination character (this is different than picky tar)
967 * ASSUMED: space after header in header block is zero filled
968 * Return:
969 * 0 if file has data to be written after the header, 1 if file has NO
970 * data to write after the header, -1 if archive write failed
971 */
972
973 #ifdef __STDC__
974 int
975 ustar_wr(register ARCHD *arcn)
976 #else
977 int
978 ustar_wr(arcn)
979 register ARCHD *arcn;
980 #endif
981 {
982 register HD_USTAR *hd;
983 register char *pt;
984 char hdblk[sizeof(HD_USTAR)];
985
986 /*
987 * check for those file system types ustar cannot store
988 */
989 if (arcn->type == PAX_SCK) {
990 paxwarn(1, "Ustar cannot archive a socket %s", arcn->org_name);
991 return(1);
992 }
993
994 /*
995 * check the length of the linkname
996 */
997 if (((arcn->type == PAX_SLK) || (arcn->type == PAX_HLK) ||
998 (arcn->type == PAX_HRG)) && (arcn->ln_nlen >= sizeof(hd->linkname))){
999 paxwarn(1, "Link name too long for ustar %s", arcn->ln_name);
1000 return(1);
1001 }
1002
1003 /*
1004 * split the path name into prefix and name fields (if needed). if
1005 * pt != arcn->name, the name has to be split
1006 */
1007 if ((pt = name_split(arcn->name, arcn->nlen)) == NULL) {
1008 paxwarn(1, "File name too long for ustar %s", arcn->name);
1009 return(1);
1010 }
1011 hd = (HD_USTAR *)hdblk;
1012 arcn->pad = 0L;
1013
1014 /*
1015 * split the name, or zero out the prefix
1016 */
1017 if (pt != arcn->name) {
1018 /*
1019 * name was split, pt points at the / where the split is to
1020 * occur, we remove the / and copy the first part to the prefix
1021 */
1022 *pt = '\0';
1023 l_strncpy(hd->prefix, arcn->name, sizeof(hd->prefix) - 1);
1024 *pt++ = '/';
1025 } else
1026 memset(hd->prefix, 0, sizeof(hd->prefix));
1027
1028 /*
1029 * copy the name part. this may be the whole path or the part after
1030 * the prefix
1031 */
1032 l_strncpy(hd->name, pt, sizeof(hd->name) - 1);
1033 hd->name[sizeof(hd->name) - 1] = '\0';
1034
1035 /*
1036 * set the fields in the header that are type dependent
1037 */
1038 switch(arcn->type) {
1039 case PAX_DIR:
1040 hd->typeflag = DIRTYPE;
1041 memset(hd->linkname, 0, sizeof(hd->linkname));
1042 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 3))
1043 goto out;
1044 break;
1045 case PAX_CHR:
1046 case PAX_BLK:
1047 if (arcn->type == PAX_CHR)
1048 hd->typeflag = CHRTYPE;
1049 else
1050 hd->typeflag = BLKTYPE;
1051 memset(hd->linkname, 0, sizeof(hd->linkname));
1052 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 3))
1053 goto out;
1054
1055 break;
1056 case PAX_FIF:
1057 hd->typeflag = FIFOTYPE;
1058 memset(hd->linkname, 0, sizeof(hd->linkname));
1059 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 3))
1060 goto out;
1061 break;
1062 case PAX_SLK:
1063 case PAX_HLK:
1064 case PAX_HRG:
1065 if (arcn->type == PAX_SLK)
1066 hd->typeflag = SYMTYPE;
1067 else
1068 hd->typeflag = LNKTYPE;
1069 l_strncpy(hd->linkname,arcn->ln_name, sizeof(hd->linkname) - 1);
1070 hd->linkname[sizeof(hd->linkname) - 1] = '\0';
1071 if (ul_oct((u_long)0L, hd->size, sizeof(hd->size), 3))
1072 goto out;
1073 break;
1074 case PAX_REG:
1075 case PAX_CTG:
1076 default:
1077 /*
1078 * file data with this type, set the padding
1079 */
1080 if (arcn->type == PAX_CTG)
1081 hd->typeflag = CONTTYPE;
1082 else
1083 hd->typeflag = REGTYPE;
1084 memset(hd->linkname, 0, sizeof(hd->linkname));
1085 arcn->pad = TAR_PAD(arcn->sb.st_size);
1086 # ifdef NET2_STAT
1087 if (ul_oct((u_long)arcn->sb.st_size, hd->size,
1088 sizeof(hd->size), 3)) {
1089 # else
1090 if (uqd_oct((u_quad_t)arcn->sb.st_size, hd->size,
1091 sizeof(hd->size), 3)) {
1092 # endif
1093 paxwarn(1,"File is too long for ustar %s",arcn->org_name);
1094 return(1);
1095 }
1096 break;
1097 }
1098
1099 /* set devmajor and devminor for all types as per spec */
1100 if (ul_oct((u_long)MAJOR(arcn->sb.st_rdev), hd->devmajor,
1101 sizeof(hd->devmajor), 3) ||
1102 ul_oct((u_long)MINOR(arcn->sb.st_rdev), hd->devminor,
1103 sizeof(hd->devminor), 3))
1104 goto out;
1105
1106 l_strncpy(hd->magic, TMAGIC, TMAGLEN);
1107 l_strncpy(hd->version, TVERSION, TVERSLEN);
1108
1109 /*
1110 * set the remaining fields. Some versions want all 16 bits of mode
1111 * we better humor them (they really do not meet spec though)....
1112 */
1113 if (ul_oct((u_long)arcn->sb.st_mode, hd->mode, sizeof(hd->mode), 3) ||
1114 ul_oct((u_long)arcn->sb.st_uid, hd->uid, sizeof(hd->uid), 3) ||
1115 ul_oct((u_long)arcn->sb.st_gid, hd->gid, sizeof(hd->gid), 3) ||
1116 ul_oct((u_long)arcn->sb.st_mtime,hd->mtime,sizeof(hd->mtime),3))
1117 goto out;
1118 l_strncpy(hd->uname,name_uid(arcn->sb.st_uid, 0),sizeof(hd->uname));
1119 l_strncpy(hd->gname,name_gid(arcn->sb.st_gid, 0),sizeof(hd->gname));
1120
1121 /*
1122 * calculate and store the checksum write the header to the archive
1123 * return 0 tells the caller to now write the file data, 1 says no data
1124 * needs to be written
1125 */
1126 if (ul_oct(tar_chksm(hdblk, sizeof(HD_USTAR)), hd->chksum,
1127 sizeof(hd->chksum), 3))
1128 goto out;
1129 if (wr_rdbuf(hdblk, sizeof(HD_USTAR)) < 0)
1130 return(-1);
1131 if (wr_skip((off_t)(BLKMULT - sizeof(HD_USTAR))) < 0)
1132 return(-1);
1133 if ((arcn->type == PAX_CTG) || (arcn->type == PAX_REG))
1134 return(0);
1135 return(1);
1136
1137 out:
1138 /*
1139 * header field is out of range
1140 */
1141 paxwarn(1, "Ustar header field is too small for %s", arcn->org_name);
1142 return(1);
1143 }
1144
1145 /*
1146 * name_split()
1147 * see if the name has to be split for storage in a ustar header. We try
1148 * to fit the entire name in the name field without splitting if we can.
1149 * The split point is always at a /
1150 * Return
1151 * character pointer to split point (always the / that is to be removed
1152 * if the split is not needed, the points is set to the start of the file
1153 * name (it would violate the spec to split there). A NULL is returned if
1154 * the file name is too long
1155 */
1156
1157 #ifdef __STDC__
1158 static char *
1159 name_split(register char *name, register int len)
1160 #else
1161 static char *
1162 name_split(name, len)
1163 register char *name;
1164 register int len;
1165 #endif
1166 {
1167 register char *start;
1168
1169 /*
1170 * check to see if the file name is small enough to fit in the name
1171 * field. if so just return a pointer to the name.
1172 */
1173 if (len < TNMSZ)
1174 return(name);
1175 if (len > (TPFSZ + TNMSZ))
1176 return(NULL);
1177
1178 /*
1179 * we start looking at the biggest sized piece that fits in the name
1180 * field. We walk foward looking for a slash to split at. The idea is
1181 * to find the biggest piece to fit in the name field (or the smallest
1182 * prefix we can find)
1183 */
1184 start = name + len - TNMSZ;
1185 while ((*start != '\0') && (*start != '/'))
1186 ++start;
1187
1188 /*
1189 * if we hit the end of the string, this name cannot be split, so we
1190 * cannot store this file.
1191 */
1192 if (*start == '\0')
1193 return(NULL);
1194 len = start - name;
1195
1196 /*
1197 * NOTE: /str where the length of str == TNMSZ can not be stored under
1198 * the p1003.1-1990 spec for ustar. We could force a prefix of / and
1199 * the file would then expand on extract to //str. The len == 0 below
1200 * makes this special case follow the spec to the letter.
1201 */
1202 if ((len >= TPFSZ) || (len == 0))
1203 return(NULL);
1204
1205 /*
1206 * ok have a split point, return it to the caller
1207 */
1208 return(start);
1209 }