1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1992 Keith Muller.
5 * Copyright (c) 1992, 1993
6 * The Regents of the University of California. All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Keith Muller of the University of California, San Diego.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36 #include <sys/types.h>
37 #include <sys/time.h>
38 #include <sys/stat.h>
39 #include <unistd.h>
40 #include <string.h>
41 #include <stdio.h>
42 #include <errno.h>
43 #include <stdlib.h>
44 #include <fts.h>
45 #include "pax.h"
46 #include "ftree.h"
47 #include "extern.h"
48
49 /*
50 * routines to interface with the fts library function.
51 *
52 * file args supplied to pax are stored on a single linked list (of type FTREE)
53 * and given to fts to be processed one at a time. pax "selects" files from
54 * the expansion of each arg into the corresponding file tree (if the arg is a
55 * directory, otherwise the node itself is just passed to pax). The selection
56 * is modified by the -n and -u flags. The user is informed when a specific
57 * file arg does not generate any selected files. -n keeps expanding the file
58 * tree arg until one of its files is selected, then skips to the next file
59 * arg. when the user does not supply the file trees as command line args to
60 * pax, they are read from stdin
61 */
62
63 static FTS *ftsp = NULL; /* current FTS handle */
64 static int ftsopts; /* options to be used on fts_open */
65 static char *farray[2]; /* array for passing each arg to fts */
66 static FTREE *fthead = NULL; /* head of linked list of file args */
67 static FTREE *fttail = NULL; /* tail of linked list of file args */
68 static FTREE *ftcur = NULL; /* current file arg being processed */
69 static FTSENT *ftent = NULL; /* current file tree entry */
70 static int ftree_skip; /* when set skip to next file arg */
71
72 static int ftree_arg(void);
73
74 /*
75 * ftree_start()
76 * initialize the options passed to fts_open() during this run of pax
77 * options are based on the selection of pax options by the user
78 * fts_start() also calls fts_arg() to open the first valid file arg. We
79 * also attempt to reset directory access times when -t (tflag) is set.
80 * Return:
81 * 0 if there is at least one valid file arg to process, -1 otherwise
82 */
83
84 int
ftree_start(void)85 ftree_start(void)
86 {
87 /*
88 * Set up the operation mode of fts, open the first file arg. We must
89 * use FTS_NOCHDIR, as the user may have to open multiple archives and
90 * if fts did a chdir off into the boondocks, we may create an archive
91 * volume in a place where the user did not expect to.
92 */
93 ftsopts = FTS_NOCHDIR;
94
95 /*
96 * optional user flags that effect file traversal
97 * -H command line symlink follow only (half follow)
98 * -L follow symlinks (logical)
99 * -P do not follow symlinks (physical). This is the default.
100 * -X do not cross over mount points
101 * -t preserve access times on files read.
102 * -n select only the first member of a file tree when a match is found
103 * -d do not extract subtrees rooted at a directory arg.
104 */
105 if (Lflag)
106 ftsopts |= FTS_LOGICAL;
107 else
108 ftsopts |= FTS_PHYSICAL;
109 if (Hflag)
110 ftsopts |= FTS_COMFOLLOW;
111 if (Xflag)
112 ftsopts |= FTS_XDEV;
113
114 if ((fthead == NULL) && ((farray[0] = malloc(PAXPATHLEN+2)) == NULL)) {
115 paxwarn(1, "Unable to allocate memory for file name buffer");
116 return(-1);
117 }
118
119 if (ftree_arg() < 0)
120 return(-1);
121 if (tflag && (atdir_start() < 0))
122 return(-1);
123 return(0);
124 }
125
126 /*
127 * ftree_add()
128 * add the arg to the linked list of files to process. Each will be
129 * processed by fts one at a time
130 * Return:
131 * 0 if added to the linked list, -1 if failed
132 */
133
134 int
ftree_add(char * str,int chflg)135 ftree_add(char *str, int chflg)
136 {
137 FTREE *ft;
138 int len;
139
140 /*
141 * simple check for bad args
142 */
143 if ((str == NULL) || (*str == '\0')) {
144 paxwarn(0, "Invalid file name argument");
145 return(-1);
146 }
147
148 /*
149 * allocate FTREE node and add to the end of the linked list (args are
150 * processed in the same order they were passed to pax). Get rid of any
151 * trailing / the user may pass us. (watch out for / by itself).
152 */
153 if ((ft = (FTREE *)malloc(sizeof(FTREE))) == NULL) {
154 paxwarn(0, "Unable to allocate memory for filename");
155 return(-1);
156 }
157
158 if (((len = strlen(str) - 1) > 0) && (str[len] == '/'))
159 str[len] = '\0';
160 ft->fname = str;
161 ft->refcnt = 0;
162 ft->chflg = chflg;
163 ft->fow = NULL;
164 if (fthead == NULL) {
165 fttail = fthead = ft;
166 return(0);
167 }
168 fttail->fow = ft;
169 fttail = ft;
170 return(0);
171 }
172
173 /*
174 * ftree_sel()
175 * this entry has been selected by pax. bump up reference count and handle
176 * -n and -d processing.
177 */
178
179 void
ftree_sel(ARCHD * arcn)180 ftree_sel(ARCHD *arcn)
181 {
182 /*
183 * set reference bit for this pattern. This linked list is only used
184 * when file trees are supplied pax as args. The list is not used when
185 * the trees are read from stdin.
186 */
187 if (ftcur != NULL)
188 ftcur->refcnt = 1;
189
190 /*
191 * if -n we are done with this arg, force a skip to the next arg when
192 * pax asks for the next file in next_file().
193 * if -d we tell fts only to match the directory (if the arg is a dir)
194 * and not the entire file tree rooted at that point.
195 */
196 if (nflag)
197 ftree_skip = 1;
198
199 if (!dflag || (arcn->type != PAX_DIR))
200 return;
201
202 if (ftent != NULL)
203 (void)fts_set(ftsp, ftent, FTS_SKIP);
204 }
205
206 /*
207 * ftree_notsel()
208 * this entry has not been selected by pax.
209 */
210
211 void
ftree_notsel(void)212 ftree_notsel(void)
213 {
214 if (ftent != NULL)
215 (void)fts_set(ftsp, ftent, FTS_SKIP);
216 }
217
218 /*
219 * ftree_chk()
220 * called at end on pax execution. Prints all those file args that did not
221 * have a selected member (reference count still 0)
222 */
223
224 void
ftree_chk(void)225 ftree_chk(void)
226 {
227 FTREE *ft;
228 int wban = 0;
229
230 /*
231 * make sure all dir access times were reset.
232 */
233 if (tflag)
234 atdir_end();
235
236 /*
237 * walk down list and check reference count. Print out those members
238 * that never had a match
239 */
240 for (ft = fthead; ft != NULL; ft = ft->fow) {
241 if ((ft->refcnt > 0) || ft->chflg)
242 continue;
243 if (wban == 0) {
244 paxwarn(1,"WARNING! These file names were not selected:");
245 ++wban;
246 }
247 (void)fprintf(stderr, "%s\n", ft->fname);
248 }
249 }
250
251 /*
252 * ftree_arg()
253 * Get the next file arg for fts to process. Can be from either the linked
254 * list or read from stdin when the user did not them as args to pax. Each
255 * arg is processed until the first successful fts_open().
256 * Return:
257 * 0 when the next arg is ready to go, -1 if out of file args (or EOF on
258 * stdin).
259 */
260
261 static int
ftree_arg(void)262 ftree_arg(void)
263 {
264 char *pt;
265
266 /*
267 * close off the current file tree
268 */
269 if (ftsp != NULL) {
270 (void)fts_close(ftsp);
271 ftsp = NULL;
272 }
273
274 /*
275 * keep looping until we get a valid file tree to process. Stop when we
276 * reach the end of the list (or get an eof on stdin)
277 */
278 for(;;) {
279 if (fthead == NULL) {
280 /*
281 * the user didn't supply any args, get the file trees
282 * to process from stdin;
283 */
284 if (fgets(farray[0], PAXPATHLEN+1, stdin) == NULL)
285 return(-1);
286 if ((pt = strchr(farray[0], '\n')) != NULL)
287 *pt = '\0';
288 } else {
289 /*
290 * the user supplied the file args as arguments to pax
291 */
292 if (ftcur == NULL)
293 ftcur = fthead;
294 else if ((ftcur = ftcur->fow) == NULL)
295 return(-1);
296 if (ftcur->chflg) {
297 /* First fchdir() back... */
298 if (fchdir(cwdfd) < 0) {
299 syswarn(1, errno,
300 "Can't fchdir to starting directory");
301 return(-1);
302 }
303 if (chdir(ftcur->fname) < 0) {
304 syswarn(1, errno, "Can't chdir to %s",
305 ftcur->fname);
306 return(-1);
307 }
308 continue;
309 } else
310 farray[0] = ftcur->fname;
311 }
312
313 /*
314 * Watch it, fts wants the file arg stored in an array of char
315 * ptrs, with the last one a null. We use a two element array
316 * and set farray[0] to point at the buffer with the file name
317 * in it. We cannot pass all the file args to fts at one shot
318 * as we need to keep a handle on which file arg generates what
319 * files (the -n and -d flags need this). If the open is
320 * successful, return a 0.
321 */
322 if ((ftsp = fts_open(farray, ftsopts, NULL)) != NULL)
323 break;
324 }
325 return(0);
326 }
327
328 /*
329 * next_file()
330 * supplies the next file to process in the supplied archd structure.
331 * Return:
332 * 0 when contents of arcn have been set with the next file, -1 when done.
333 */
334
335 int
next_file(ARCHD * arcn)336 next_file(ARCHD *arcn)
337 {
338 int cnt;
339 time_t atime;
340 time_t mtime;
341
342 /*
343 * ftree_sel() might have set the ftree_skip flag if the user has the
344 * -n option and a file was selected from this file arg tree. (-n says
345 * only one member is matched for each pattern) ftree_skip being 1
346 * forces us to go to the next arg now.
347 */
348 if (ftree_skip) {
349 /*
350 * clear and go to next arg
351 */
352 ftree_skip = 0;
353 if (ftree_arg() < 0)
354 return(-1);
355 }
356
357 /*
358 * loop until we get a valid file to process
359 */
360 for(;;) {
361 if ((ftent = fts_read(ftsp)) == NULL) {
362 /*
363 * out of files in this tree, go to next arg, if none
364 * we are done
365 */
366 if (ftree_arg() < 0)
367 return(-1);
368 continue;
369 }
370
371 /*
372 * handle each type of fts_read() flag
373 */
374 switch(ftent->fts_info) {
375 case FTS_D:
376 case FTS_DEFAULT:
377 case FTS_F:
378 case FTS_SL:
379 case FTS_SLNONE:
380 /*
381 * these are all ok
382 */
383 break;
384 case FTS_DP:
385 /*
386 * already saw this directory. If the user wants file
387 * access times reset, we use this to restore the
388 * access time for this directory since this is the
389 * last time we will see it in this file subtree
390 * remember to force the time (this is -t on a read
391 * directory, not a created directory).
392 */
393 if (!tflag || (get_atdir(ftent->fts_statp->st_dev,
394 ftent->fts_statp->st_ino, &mtime, &atime) < 0))
395 continue;
396 set_ftime(ftent->fts_path, mtime, atime, 1);
397 continue;
398 case FTS_DC:
399 /*
400 * fts claims a file system cycle
401 */
402 paxwarn(1,"File system cycle found at %s",ftent->fts_path);
403 continue;
404 case FTS_DNR:
405 syswarn(1, ftent->fts_errno,
406 "Unable to read directory %s", ftent->fts_path);
407 continue;
408 case FTS_ERR:
409 syswarn(1, ftent->fts_errno,
410 "File system traversal error");
411 continue;
412 case FTS_NS:
413 case FTS_NSOK:
414 syswarn(1, ftent->fts_errno,
415 "Unable to access %s", ftent->fts_path);
416 continue;
417 }
418
419 /*
420 * ok got a file tree node to process. copy info into arcn
421 * structure (initialize as required)
422 */
423 arcn->skip = 0;
424 arcn->pad = 0;
425 arcn->ln_nlen = 0;
426 arcn->ln_name[0] = '\0';
427 arcn->sb = *(ftent->fts_statp);
428
429 /*
430 * file type based set up and copy into the arcn struct
431 * SIDE NOTE:
432 * we try to reset the access time on all files and directories
433 * we may read when the -t flag is specified. files are reset
434 * when we close them after copying. we reset the directories
435 * when we are done with their file tree (we also clean up at
436 * end in case we cut short a file tree traversal). However
437 * there is no way to reset access times on symlinks.
438 */
439 switch(S_IFMT & arcn->sb.st_mode) {
440 case S_IFDIR:
441 arcn->type = PAX_DIR;
442 if (!tflag)
443 break;
444 add_atdir(ftent->fts_path, arcn->sb.st_dev,
445 arcn->sb.st_ino, arcn->sb.st_mtime,
446 arcn->sb.st_atime);
447 break;
448 case S_IFCHR:
449 arcn->type = PAX_CHR;
450 break;
451 case S_IFBLK:
452 arcn->type = PAX_BLK;
453 break;
454 case S_IFREG:
455 /*
456 * only regular files with have data to store on the
457 * archive. all others will store a zero length skip.
458 * the skip field is used by pax for actual data it has
459 * to read (or skip over).
460 */
461 arcn->type = PAX_REG;
462 arcn->skip = arcn->sb.st_size;
463 break;
464 case S_IFLNK:
465 arcn->type = PAX_SLK;
466 /*
467 * have to read the symlink path from the file
468 */
469 if ((cnt = readlink(ftent->fts_path, arcn->ln_name,
470 PAXPATHLEN - 1)) < 0) {
471 syswarn(1, errno, "Unable to read symlink %s",
472 ftent->fts_path);
473 continue;
474 }
475 /*
476 * set link name length, watch out readlink does not
477 * always NUL terminate the link path
478 */
479 arcn->ln_name[cnt] = '\0';
480 arcn->ln_nlen = cnt;
481 break;
482 case S_IFSOCK:
483 /*
484 * under BSD storing a socket is senseless but we will
485 * let the format specific write function make the
486 * decision of what to do with it.
487 */
488 arcn->type = PAX_SCK;
489 break;
490 case S_IFIFO:
491 arcn->type = PAX_FIF;
492 break;
493 }
494 break;
495 }
496
497 /*
498 * copy file name, set file name length
499 */
500 arcn->nlen = l_strncpy(arcn->name, ftent->fts_path, sizeof(arcn->name) - 1);
501 arcn->name[arcn->nlen] = '\0';
502 arcn->org_name = ftent->fts_path;
503 return(0);
504 }
505