xref: /netbsd-src/sbin/dump/traverse.c (revision 3b01aba77a7a698587faaae455bbfe740923c1f5)
1 /*	$NetBSD: traverse.c,v 1.31 2001/05/28 01:09:55 lukem Exp $	*/
2 
3 /*-
4  * Copyright (c) 1980, 1988, 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. All advertising materials mentioning features or use of this software
16  *    must display the following acknowledgement:
17  *	This product includes software developed by the University of
18  *	California, Berkeley and its contributors.
19  * 4. 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/cdefs.h>
37 #ifndef lint
38 #if 0
39 static char sccsid[] = "@(#)traverse.c	8.7 (Berkeley) 6/15/95";
40 #else
41 __RCSID("$NetBSD: traverse.c,v 1.31 2001/05/28 01:09:55 lukem Exp $");
42 #endif
43 #endif /* not lint */
44 
45 #include <sys/param.h>
46 #include <sys/time.h>
47 #include <sys/stat.h>
48 #ifdef sunos
49 #include <sys/vnode.h>
50 
51 #include <ufs/fs.h>
52 #include <ufs/fsdir.h>
53 #include <ufs/inode.h>
54 #else
55 #include <ufs/ufs/dir.h>
56 #include <ufs/ufs/dinode.h>
57 #include <ufs/ffs/fs.h>
58 #include <ufs/ffs/ffs_extern.h>
59 #endif
60 
61 #include <protocols/dumprestore.h>
62 
63 #include <ctype.h>
64 #include <errno.h>
65 #include <fts.h>
66 #include <stdio.h>
67 #include <string.h>
68 #include <unistd.h>
69 
70 #include "dump.h"
71 
72 #define	HASDUMPEDFILE	0x1
73 #define	HASSUBDIRS	0x2
74 
75 #ifdef	FS_44INODEFMT
76 typedef	quad_t fsizeT;
77 #else
78 typedef	int32_t fsizeT;
79 #endif
80 
81 static	int dirindir(ino_t, daddr_t, int, long *, long *, int);
82 static	void dmpindir(ino_t, daddr_t, int, fsizeT *);
83 static	int searchdir(ino_t, daddr_t, long, long, long *, int);
84 
85 /*
86  * This is an estimation of the number of TP_BSIZE blocks in the file.
87  * It estimates the number of blocks in files with holes by assuming
88  * that all of the blocks accounted for by di_blocks are data blocks
89  * (when some of the blocks are usually used for indirect pointers);
90  * hence the estimate may be high.
91  */
92 long
93 blockest(struct dinode *dp)
94 {
95 	long blkest, sizeest;
96 
97 	/*
98 	 * dp->di_size is the size of the file in bytes.
99 	 * dp->di_blocks stores the number of sectors actually in the file.
100 	 * If there are more sectors than the size would indicate, this just
101 	 *	means that there are indirect blocks in the file or unused
102 	 *	sectors in the last file block; we can safely ignore these
103 	 *	(blkest = sizeest below).
104 	 * If the file is bigger than the number of sectors would indicate,
105 	 *	then the file has holes in it.	In this case we must use the
106 	 *	block count to estimate the number of data blocks used, but
107 	 *	we use the actual size for estimating the number of indirect
108 	 *	dump blocks (sizeest vs. blkest in the indirect block
109 	 *	calculation).
110 	 */
111 	blkest = howmany(dbtob((u_int64_t)dp->di_blocks), TP_BSIZE);
112 	sizeest = howmany(dp->di_size, TP_BSIZE);
113 	if (blkest > sizeest)
114 		blkest = sizeest;
115 	if (dp->di_size > ufsib->ufs_bsize * NDADDR) {
116 		/* calculate the number of indirect blocks on the dump tape */
117 		blkest +=
118 			howmany(sizeest - NDADDR * ufsib->ufs_bsize / TP_BSIZE,
119 			TP_NINDIR);
120 	}
121 	return (blkest + 1);
122 }
123 
124 /* Auxiliary macro to pick up files changed since previous dump. */
125 #define	CHANGEDSINCE(dp, t) \
126 	((dp)->di_mtime >= (t) || (dp)->di_ctime >= (t))
127 
128 /* The WANTTODUMP macro decides whether a file should be dumped. */
129 #ifdef UF_NODUMP
130 #define	WANTTODUMP(dp) \
131 	(CHANGEDSINCE(dp, iswap32(spcl.c_ddate)) && \
132 	 (nonodump || ((dp)->di_flags & UF_NODUMP) != UF_NODUMP))
133 #else
134 #define	WANTTODUMP(dp) CHANGEDSINCE(dp, iswap32(spcl.c_ddate))
135 #endif
136 
137 /*
138  * Determine if given inode should be dumped
139  */
140 void
141 mapfileino(ino_t ino, long *tapesize, int *dirskipped)
142 {
143 	int mode;
144 	struct dinode *dp;
145 
146 	/*
147 	 * Skip inode if we've already marked it for dumping
148 	 */
149 	if (TSTINO(ino, usedinomap))
150 		return;
151 	dp = getino(ino);
152 	if ((mode = (dp->di_mode & IFMT)) == 0)
153 		return;
154 	/*
155 	 * Put all dirs in dumpdirmap, inodes that are to be dumped in the
156 	 * used map. All inode but dirs who have the nodump attribute go
157 	 * to the usedinomap.
158 	 */
159 	SETINO(ino, usedinomap);
160 	if (mode == IFDIR)
161 		SETINO(ino, dumpdirmap);
162 	if (WANTTODUMP(dp)) {
163 		SETINO(ino, dumpinomap);
164 		if (mode != IFREG && mode != IFDIR && mode != IFLNK)
165 			*tapesize += 1;
166 		else
167 			*tapesize += blockest(dp);
168 		return;
169 	}
170 	if (mode == IFDIR) {
171 #ifdef UF_NODUMP
172 		if (!nonodump && (dp->di_flags & UF_NODUMP))
173 			CLRINO(ino, usedinomap);
174 #endif
175 		*dirskipped = 1;
176 	}
177 }
178 
179 /*
180  * Dump pass 1.
181  *
182  * Walk the inode list for a filesystem to find all allocated inodes
183  * that have been modified since the previous dump time. Also, find all
184  * the directories in the filesystem.
185  * disk may be NULL if dirv is NULL.
186  */
187 int
188 mapfiles(ino_t maxino, long *tapesize, char *disk, char * const *dirv)
189 {
190 	int anydirskipped = 0;
191 
192 	if (dirv != NULL) {
193 		char	 curdir[MAXPATHLEN];
194 		FTS	*dirh;
195 		FTSENT	*entry;
196 		int	 d;
197 
198 		if (getcwd(curdir, sizeof(curdir)) == NULL) {
199 			msg("Can't determine cwd: %s\n", strerror(errno));
200 			dumpabort(0);
201 		}
202 		if ((dirh = fts_open(dirv, FTS_PHYSICAL|FTS_SEEDOT|FTS_XDEV,
203 		    		    NULL)) == NULL) {
204 			msg("fts_open failed: %s\n", strerror(errno));
205 			dumpabort(0);
206 		}
207 		while ((entry = fts_read(dirh)) != NULL) {
208 			switch (entry->fts_info) {
209 			case FTS_DNR:		/* an error */
210 			case FTS_ERR:
211 			case FTS_NS:
212 				msg("Can't fts_read %s: %s\n", entry->fts_path,
213 				    strerror(errno));
214 			case FTS_DP:		/* already seen dir */
215 				continue;
216 			}
217 			mapfileino(entry->fts_statp->st_ino, tapesize,
218 			    &anydirskipped);
219 		}
220 		(void)fts_close(dirh);
221 
222 		/*
223 		 * Add any parent directories
224 		 */
225 		for (d = 0 ; dirv[d] != NULL ; d++) {
226 			char path[MAXPATHLEN];
227 
228 			if (dirv[d][0] != '/')
229 				(void)snprintf(path, sizeof(path), "%s/%s",
230 				    curdir, dirv[d]);
231 			else
232 				(void)snprintf(path, sizeof(path), "%s",
233 				    dirv[d]);
234 			while (strcmp(path, disk) != 0) {
235 				char *p;
236 				struct stat sb;
237 
238 				if (*path == '\0')
239 					break;
240 				if ((p = strrchr(path, '/')) == NULL)
241 					break;
242 				if (p == path)
243 					break;
244 				*p = '\0';
245 				if (stat(path, &sb) == -1) {
246 					msg("Can't stat %s: %s\n", path,
247 					    strerror(errno));
248 					break;
249 				}
250 				mapfileino(sb.st_ino, tapesize, &anydirskipped);
251 			}
252 		}
253 
254 		/*
255 		 * Ensure that the root inode actually appears in the
256 		 * file list for a subdir
257 		 */
258 		mapfileino(ROOTINO, tapesize, &anydirskipped);
259 	} else {
260 		ino_t ino;
261 
262 		for (ino = ROOTINO; ino < maxino; ino++) {
263 			mapfileino(ino, tapesize, &anydirskipped);
264 		}
265 	}
266 	/*
267 	 * Restore gets very upset if the root is not dumped,
268 	 * so ensure that it always is dumped.
269 	 */
270 	SETINO(ROOTINO, dumpinomap);
271 	return (anydirskipped);
272 }
273 
274 /*
275  * Dump pass 2.
276  *
277  * Scan each directory on the filesystem to see if it has any modified
278  * files in it. If it does, and has not already been added to the dump
279  * list (because it was itself modified), then add it. If a directory
280  * has not been modified itself, contains no modified files and has no
281  * subdirectories, then it can be deleted from the dump list and from
282  * the list of directories. By deleting it from the list of directories,
283  * its parent may now qualify for the same treatment on this or a later
284  * pass using this algorithm.
285  */
286 int
287 mapdirs(ino_t maxino, long *tapesize)
288 {
289 	struct  dinode *dp, di;
290 	int i, isdir, nodump;
291 	char *map;
292 	ino_t ino;
293 	long filesize;
294 	int ret, change = 0;
295 
296 	isdir = 0;		/* XXX just to get gcc to shut up */
297 	for (map = dumpdirmap, ino = 1; ino < maxino; ino++) {
298 		if (((ino - 1) % NBBY) == 0)	/* map is offset by 1 */
299 			isdir = *map++;
300 		else
301 			isdir >>= 1;
302 		/*
303 		 * If dir has been removed from the used map, it's either
304 		 * because it had the nodump flag, or it herited it from
305 		 * its parent. A directory can't be in dumpinomap if
306 		 * not in usedinomap, but we have to go throuh it anyway
307 		 * to propagate the nodump attribute.
308 		 */
309 		nodump = (TSTINO(ino, usedinomap) == 0);
310 		if ((isdir & 1) == 0 ||
311 		    (TSTINO(ino, dumpinomap) && nodump == 0))
312 			continue;
313 
314 		dp = getino(ino);
315 		di = *dp; /* inode buf may be changed in searchdir */
316 		filesize = di.di_size;
317 		for (ret = 0, i = 0; filesize > 0 && i < NDADDR; i++) {
318 			if (di.di_db[i] != 0)
319 				ret |= searchdir(ino, iswap32(di.di_db[i]),
320 					(long)ufs_dblksize(ufsib, &di, i),
321 					filesize, tapesize, nodump);
322 			if (ret & HASDUMPEDFILE)
323 				filesize = 0;
324 			else
325 				filesize -= ufsib->ufs_bsize;
326 		}
327 		for (i = 0; filesize > 0 && i < NIADDR; i++) {
328 			if (di.di_ib[i] == 0)
329 				continue;
330 			ret |= dirindir(ino, di.di_ib[i], i, &filesize,
331 			    tapesize, nodump);
332 		}
333 		if (ret & HASDUMPEDFILE) {
334 			SETINO(ino, dumpinomap);
335 			*tapesize += blockest(&di);
336 			change = 1;
337 			continue;
338 		}
339 		if (nodump) {
340 			if (ret & HASSUBDIRS)
341 				change = 1; /* subdirs have inherited nodump */
342 			CLRINO(ino, dumpdirmap);
343 		} else if ((ret & HASSUBDIRS) == 0) {
344 			if (!TSTINO(ino, dumpinomap)) {
345 				CLRINO(ino, dumpdirmap);
346 				change = 1;
347 			}
348 		}
349 	}
350 	return (change);
351 }
352 
353 /*
354  * Read indirect blocks, and pass the data blocks to be searched
355  * as directories. Quit as soon as any entry is found that will
356  * require the directory to be dumped.
357  */
358 static int
359 dirindir(ino_t ino, daddr_t blkno, int ind_level, long *filesize,
360 	long *tapesize, int nodump)
361 {
362 	int ret = 0;
363 	int i;
364 	daddr_t	idblk[MAXNINDIR];
365 
366 	bread(fsatoda(ufsib, iswap32(blkno)), (char *)idblk,
367 		(int)ufsib->ufs_bsize);
368 	if (ind_level <= 0) {
369 		for (i = 0; *filesize > 0 && i < ufsib->ufs_nindir; i++) {
370 			blkno = idblk[i];
371 			if (blkno != 0)
372 				ret |= searchdir(ino, iswap32(blkno),
373 				    ufsib->ufs_bsize, *filesize,
374 				    tapesize, nodump);
375 			if (ret & HASDUMPEDFILE)
376 				*filesize = 0;
377 			else
378 				*filesize -= ufsib->ufs_bsize;
379 		}
380 		return (ret);
381 	}
382 	ind_level--;
383 	for (i = 0; *filesize > 0 && i < ufsib->ufs_nindir; i++) {
384 		blkno = idblk[i];
385 		if (blkno != 0)
386 			ret |= dirindir(ino, blkno, ind_level, filesize,
387 			    tapesize, nodump);
388 	}
389 	return (ret);
390 }
391 
392 /*
393  * Scan a disk block containing directory information looking to see if
394  * any of the entries are on the dump list and to see if the directory
395  * contains any subdirectories.
396  */
397 static int
398 searchdir(ino_t dino, daddr_t blkno, long size, long filesize,
399 	long *tapesize, int nodump)
400 {
401 	struct direct *dp;
402 	struct dinode *ip;
403 	long loc, ret = 0;
404 	char dblk[MAXBSIZE];
405 	ino_t ino;
406 
407 	bread(fsatoda(ufsib, blkno), dblk, (int)size);
408 	if (filesize < size)
409 		size = filesize;
410 	for (loc = 0; loc < size; ) {
411 		dp = (struct direct *)(dblk + loc);
412 		if (dp->d_reclen == 0) {
413 			msg("corrupted directory, inumber %d\n", dino);
414 			break;
415 		}
416 		loc += iswap16(dp->d_reclen);
417 		if (dp->d_ino == 0)
418 			continue;
419 		if (dp->d_name[0] == '.') {
420 			if (dp->d_name[1] == '\0')
421 				continue;
422 			if (dp->d_name[1] == '.' && dp->d_name[2] == '\0')
423 				continue;
424 		}
425 		ino = iswap32(dp->d_ino);
426 		if (nodump) {
427 			ip = getino(ino);
428 			if (TSTINO(ino, dumpinomap)) {
429 				CLRINO(ino, dumpinomap);
430 				CLRINO(ino, usedinomap);
431 				*tapesize -= blockest(ip);
432 			}
433 			/* Add dir back to the dir map, to propagate nodump */
434 			if ((ip->di_mode & IFMT) == IFDIR) {
435 				SETINO(ino, dumpdirmap);
436 				ret |= HASSUBDIRS;
437 			}
438 		} else {
439 			if (TSTINO(ino, dumpinomap)) {
440 				ret |= HASDUMPEDFILE;
441 				if (ret & HASSUBDIRS)
442 					break;
443 			}
444 			if (TSTINO(ino, dumpdirmap)) {
445 				ret |= HASSUBDIRS;
446 				if (ret & HASDUMPEDFILE)
447 					break;
448 			}
449 		}
450 	}
451 	return (ret);
452 }
453 
454 /*
455  * Dump passes 3 and 4.
456  *
457  * Dump the contents of an inode to tape.
458  */
459 void
460 dumpino(struct dinode *dp, ino_t ino)
461 {
462 	int ind_level, cnt;
463 	fsizeT size;
464 	char buf[TP_BSIZE];
465 
466 	if (newtape) {
467 		newtape = 0;
468 		dumpmap(dumpinomap, TS_BITS, ino);
469 	}
470 	CLRINO(ino, dumpinomap);
471 	if (needswap)
472 		ffs_dinode_swap(dp, &spcl.c_dinode);
473 	else
474 		spcl.c_dinode = *dp;
475 	spcl.c_type = iswap32(TS_INODE);
476 	spcl.c_count = 0;
477 	switch (dp->di_mode & IFMT) {
478 
479 	case 0:
480 		/*
481 		 * Freed inode.
482 		 */
483 		return;
484 
485 	case IFLNK:
486 		/*
487 		 * Check for short symbolic link.
488 		 */
489 		if (dp->di_size > 0 &&
490 #ifdef FS_44INODEFMT
491 		    (dp->di_size < ufsib->ufs_maxsymlinklen ||
492 		     (ufsib->ufs_maxsymlinklen == 0 && dp->di_blocks == 0))
493 #else
494 		    dp->di_blocks == 0
495 #endif
496 			) {
497 			spcl.c_addr[0] = 1;
498 			spcl.c_count = iswap32(1);
499 			writeheader(ino);
500 			memmove(buf, dp->di_shortlink, (u_long)dp->di_size);
501 			buf[dp->di_size] = '\0';
502 			writerec(buf, 0);
503 			return;
504 		}
505 		/* fall through */
506 
507 	case IFDIR:
508 	case IFREG:
509 		if (dp->di_size > 0)
510 			break;
511 		/* fall through */
512 
513 	case IFIFO:
514 	case IFSOCK:
515 	case IFCHR:
516 	case IFBLK:
517 		writeheader(ino);
518 		return;
519 
520 	default:
521 		msg("Warning: undefined file type 0%o\n", dp->di_mode & IFMT);
522 		return;
523 	}
524 	if (dp->di_size > NDADDR * ufsib->ufs_bsize)
525 		cnt = NDADDR * ufsib->ufs_frag;
526 	else
527 		cnt = howmany(dp->di_size, ufsib->ufs_fsize);
528 	blksout(&dp->di_db[0], cnt, ino);
529 	if ((size = dp->di_size - NDADDR * ufsib->ufs_bsize) <= 0)
530 		return;
531 	for (ind_level = 0; ind_level < NIADDR; ind_level++) {
532 		dmpindir(ino, dp->di_ib[ind_level], ind_level, &size);
533 		if (size <= 0)
534 			return;
535 	}
536 }
537 
538 /*
539  * Read indirect blocks, and pass the data blocks to be dumped.
540  */
541 static void
542 dmpindir(ino_t ino, daddr_t blk, int ind_level, fsizeT *size)
543 {
544 	int i, cnt;
545 	daddr_t idblk[MAXNINDIR];
546 
547 	if (blk != 0)
548 		bread(fsatoda(ufsib, iswap32(blk)), (char *)idblk,
549 			(int) ufsib->ufs_bsize);
550 	else
551 		memset(idblk, 0, (int)ufsib->ufs_bsize);
552 	if (ind_level <= 0) {
553 		if (*size < ufsib->ufs_nindir * ufsib->ufs_bsize)
554 			cnt = howmany(*size, ufsib->ufs_fsize);
555 		else
556 			cnt = ufsib->ufs_nindir * ufsib->ufs_frag;
557 		*size -= ufsib->ufs_nindir * ufsib->ufs_bsize;
558 		blksout(&idblk[0], cnt, ino);
559 		return;
560 	}
561 	ind_level--;
562 	for (i = 0; i < ufsib->ufs_nindir; i++) {
563 		dmpindir(ino, idblk[i], ind_level, size);
564 		if (*size <= 0)
565 			return;
566 	}
567 }
568 
569 /*
570  * Collect up the data into tape record sized buffers and output them.
571  */
572 void
573 blksout(daddr_t *blkp, int frags, ino_t ino)
574 {
575 	daddr_t *bp;
576 	int i, j, count, blks, tbperdb;
577 
578 	blks = howmany(frags * ufsib->ufs_fsize, TP_BSIZE);
579 	tbperdb = ufsib->ufs_bsize >> tp_bshift;
580 	for (i = 0; i < blks; i += TP_NINDIR) {
581 		if (i + TP_NINDIR > blks)
582 			count = blks;
583 		else
584 			count = i + TP_NINDIR;
585 		for (j = i; j < count; j++)
586 			if (blkp[j / tbperdb] != 0)
587 				spcl.c_addr[j - i] = 1;
588 			else
589 				spcl.c_addr[j - i] = 0;
590 		spcl.c_count = iswap32(count - i);
591 		writeheader(ino);
592 		bp = &blkp[i / tbperdb];
593 		for (j = i; j < count; j += tbperdb, bp++)
594 			if (*bp != 0) {
595 				if (j + tbperdb <= count)
596 					dumpblock(iswap32(*bp), (int)ufsib->ufs_bsize);
597 				else
598 					dumpblock(iswap32(*bp), (count - j) * TP_BSIZE);
599 			}
600 		spcl.c_type = iswap32(TS_ADDR);
601 	}
602 }
603 
604 /*
605  * Dump a map to the tape.
606  */
607 void
608 dumpmap(char *map, int type, ino_t ino)
609 {
610 	int i;
611 	char *cp;
612 
613 	spcl.c_type = iswap32(type);
614 	spcl.c_count = iswap32(howmany(mapsize * sizeof(char), TP_BSIZE));
615 	writeheader(ino);
616 	for (i = 0, cp = map; i < iswap32(spcl.c_count); i++, cp += TP_BSIZE)
617 		writerec(cp, 0);
618 }
619 
620 /*
621  * Write a header record to the dump tape.
622  */
623 void
624 writeheader(ino_t ino)
625 {
626 	int32_t sum, cnt, *lp;
627 
628 	spcl.c_inumber = iswap32(ino);
629 	spcl.c_magic = iswap32(NFS_MAGIC);
630 	spcl.c_checksum = 0;
631 	lp = (int32_t *)&spcl;
632 	sum = 0;
633 	cnt = sizeof(union u_spcl) / (4 * sizeof(int32_t));
634 	while (--cnt >= 0) {
635 		sum += iswap32(*lp++);
636 		sum += iswap32(*lp++);
637 		sum += iswap32(*lp++);
638 		sum += iswap32(*lp++);
639 	}
640 	spcl.c_checksum = iswap32(CHECKSUM - sum);
641 	writerec((char *)&spcl, 1);
642 }
643