inode.c revision 57573 
1/*
2 * Copyright (c) 1980, 1986, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 *    must display the following acknowledgement:
15 *	This product includes software developed by the University of
16 *	California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 *    may be used to endorse or promote products derived from this software
19 *    without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 */
33
34#ifndef lint
35#if 0
36static const char sccsid[] = "@(#)inode.c	8.8 (Berkeley) 4/28/95";
37#endif
38static const char rcsid[] =
39  "$FreeBSD: head/sbin/fsck_ffs/inode.c 57573 2000-02-28 20:02:41Z mckusick $";
40#endif /* not lint */
41
42#include <sys/param.h>
43#include <sys/time.h>
44
45#include <ufs/ufs/dinode.h>
46#include <ufs/ufs/dir.h>
47#include <ufs/ffs/fs.h>
48
49#include <err.h>
50#include <pwd.h>
51#include <string.h>
52
53#include "fsck.h"
54
55static ino_t startinum;
56
57static int iblock __P((struct inodesc *, long ilevel, quad_t isize));
58
59int
60ckinode(dp, idesc)
61	struct dinode *dp;
62	register struct inodesc *idesc;
63{
64	ufs_daddr_t *ap;
65	int ret;
66	long n, ndb, offset;
67	struct dinode dino;
68	quad_t remsize, sizepb;
69	mode_t mode;
70	char pathbuf[MAXPATHLEN + 1];
71
72	if (idesc->id_fix != IGNORE)
73		idesc->id_fix = DONTKNOW;
74	idesc->id_entryno = 0;
75	idesc->id_filesize = dp->di_size;
76	mode = dp->di_mode & IFMT;
77	if (mode == IFBLK || mode == IFCHR || (mode == IFLNK &&
78	    dp->di_size < (unsigned)sblock.fs_maxsymlinklen))
79		return (KEEPON);
80	dino = *dp;
81	ndb = howmany(dino.di_size, sblock.fs_bsize);
82	for (ap = &dino.di_db[0]; ap < &dino.di_db[NDADDR]; ap++) {
83		if (--ndb == 0 && (offset = blkoff(&sblock, dino.di_size)) != 0)
84			idesc->id_numfrags =
85				numfrags(&sblock, fragroundup(&sblock, offset));
86		else
87			idesc->id_numfrags = sblock.fs_frag;
88		if (*ap == 0) {
89			if (idesc->id_type == DATA && ndb >= 0) {
90				/* An empty block in a directory XXX */
91				getpathname(pathbuf, idesc->id_number,
92						idesc->id_number);
93                        	pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
94					pathbuf);
95                        	if (reply("ADJUST LENGTH") == 1) {
96					dp = ginode(idesc->id_number);
97                                	dp->di_size = (ap - &dino.di_db[0]) *
98					    sblock.fs_bsize;
99					printf(
100					    "YOU MUST RERUN FSCK AFTERWARDS\n");
101					rerun = 1;
102                                	inodirty();
103
104                        	}
105			}
106			continue;
107		}
108		idesc->id_blkno = *ap;
109		if (idesc->id_type == ADDR)
110			ret = (*idesc->id_func)(idesc);
111		else
112			ret = dirscan(idesc);
113		if (ret & STOP)
114			return (ret);
115	}
116	idesc->id_numfrags = sblock.fs_frag;
117	remsize = dino.di_size - sblock.fs_bsize * NDADDR;
118	sizepb = sblock.fs_bsize;
119	for (ap = &dino.di_ib[0], n = 1; n <= NIADDR; ap++, n++) {
120		if (*ap) {
121			idesc->id_blkno = *ap;
122			ret = iblock(idesc, n, remsize);
123			if (ret & STOP)
124				return (ret);
125		} else {
126			if (idesc->id_type == DATA && remsize > 0) {
127				/* An empty block in a directory XXX */
128				getpathname(pathbuf, idesc->id_number,
129						idesc->id_number);
130                        	pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
131					pathbuf);
132                        	if (reply("ADJUST LENGTH") == 1) {
133					dp = ginode(idesc->id_number);
134                                	dp->di_size -= remsize;
135					remsize = 0;
136					printf(
137					    "YOU MUST RERUN FSCK AFTERWARDS\n");
138					rerun = 1;
139                                	inodirty();
140					break;
141                        	}
142			}
143		}
144		sizepb *= NINDIR(&sblock);
145		remsize -= sizepb;
146	}
147	return (KEEPON);
148}
149
150static int
151iblock(idesc, ilevel, isize)
152	struct inodesc *idesc;
153	long ilevel;
154	quad_t isize;
155{
156	ufs_daddr_t *ap;
157	ufs_daddr_t *aplim;
158	struct bufarea *bp;
159	int i, n, (*func)(), nif;
160	quad_t sizepb;
161	char buf[BUFSIZ];
162	char pathbuf[MAXPATHLEN + 1];
163	struct dinode *dp;
164
165	if (idesc->id_type == ADDR) {
166		func = idesc->id_func;
167		if (((n = (*func)(idesc)) & KEEPON) == 0)
168			return (n);
169	} else
170		func = dirscan;
171	if (chkrange(idesc->id_blkno, idesc->id_numfrags))
172		return (SKIP);
173	bp = getdatablk(idesc->id_blkno, sblock.fs_bsize);
174	ilevel--;
175	for (sizepb = sblock.fs_bsize, i = 0; i < ilevel; i++)
176		sizepb *= NINDIR(&sblock);
177	nif = howmany(isize , sizepb);
178	if (nif > NINDIR(&sblock))
179		nif = NINDIR(&sblock);
180	if (idesc->id_func == pass1check && nif < NINDIR(&sblock)) {
181		aplim = &bp->b_un.b_indir[NINDIR(&sblock)];
182		for (ap = &bp->b_un.b_indir[nif]; ap < aplim; ap++) {
183			if (*ap == 0)
184				continue;
185			(void)sprintf(buf, "PARTIALLY TRUNCATED INODE I=%lu",
186			    (u_long)idesc->id_number);
187			if (dofix(idesc, buf)) {
188				*ap = 0;
189				dirty(bp);
190			}
191		}
192		flush(fswritefd, bp);
193	}
194	aplim = &bp->b_un.b_indir[nif];
195	for (ap = bp->b_un.b_indir; ap < aplim; ap++) {
196		if (*ap) {
197			idesc->id_blkno = *ap;
198			if (ilevel == 0)
199				n = (*func)(idesc);
200			else
201				n = iblock(idesc, ilevel, isize);
202			if (n & STOP) {
203				bp->b_flags &= ~B_INUSE;
204				return (n);
205			}
206		} else {
207			if (idesc->id_type == DATA && isize > 0) {
208				/* An empty block in a directory XXX */
209				getpathname(pathbuf, idesc->id_number,
210						idesc->id_number);
211                        	pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
212					pathbuf);
213                        	if (reply("ADJUST LENGTH") == 1) {
214					dp = ginode(idesc->id_number);
215                                	dp->di_size -= isize;
216					isize = 0;
217					printf(
218					    "YOU MUST RERUN FSCK AFTERWARDS\n");
219					rerun = 1;
220                                	inodirty();
221					bp->b_flags &= ~B_INUSE;
222					return(STOP);
223                        	}
224			}
225		}
226		isize -= sizepb;
227	}
228	bp->b_flags &= ~B_INUSE;
229	return (KEEPON);
230}
231
232/*
233 * Check that a block in a legal block number.
234 * Return 0 if in range, 1 if out of range.
235 */
236int
237chkrange(blk, cnt)
238	ufs_daddr_t blk;
239	int cnt;
240{
241	register int c;
242
243	if (cnt <= 0 || blk <= 0 || blk > maxfsblock ||
244	    cnt - 1 > maxfsblock - blk)
245		return (1);
246	if (cnt > sblock.fs_frag ||
247	    fragnum(&sblock, blk) + cnt > sblock.fs_frag) {
248		if (debug)
249			printf("bad size: blk %ld, offset %ld, size %ld\n",
250				blk, fragnum(&sblock, blk), cnt);
251		return (1);
252	}
253	c = dtog(&sblock, blk);
254	if (blk < cgdmin(&sblock, c)) {
255		if ((blk + cnt) > cgsblock(&sblock, c)) {
256			if (debug) {
257				printf("blk %ld < cgdmin %ld;",
258				    (long)blk, (long)cgdmin(&sblock, c));
259				printf(" blk + cnt %ld > cgsbase %ld\n",
260				    (long)(blk + cnt),
261				    (long)cgsblock(&sblock, c));
262			}
263			return (1);
264		}
265	} else {
266		if ((blk + cnt) > cgbase(&sblock, c+1)) {
267			if (debug)  {
268				printf("blk %ld >= cgdmin %ld;",
269				    (long)blk, (long)cgdmin(&sblock, c));
270				printf(" blk + cnt %ld > sblock.fs_fpg %ld\n",
271				    (long)(blk + cnt), (long)sblock.fs_fpg);
272			}
273			return (1);
274		}
275	}
276	return (0);
277}
278
279/*
280 * General purpose interface for reading inodes.
281 */
282struct dinode *
283ginode(inumber)
284	ino_t inumber;
285{
286	ufs_daddr_t iblk;
287
288	if (inumber < ROOTINO || inumber > maxino)
289		errx(EEXIT, "bad inode number %d to ginode", inumber);
290	if (startinum == 0 ||
291	    inumber < startinum || inumber >= startinum + INOPB(&sblock)) {
292		iblk = ino_to_fsba(&sblock, inumber);
293		if (pbp != 0)
294			pbp->b_flags &= ~B_INUSE;
295		pbp = getdatablk(iblk, sblock.fs_bsize);
296		startinum = (inumber / INOPB(&sblock)) * INOPB(&sblock);
297	}
298	return (&pbp->b_un.b_dinode[inumber % INOPB(&sblock)]);
299}
300
301/*
302 * Special purpose version of ginode used to optimize first pass
303 * over all the inodes in numerical order.
304 */
305ino_t nextino, lastinum;
306long readcnt, readpercg, fullcnt, inobufsize, partialcnt, partialsize;
307struct dinode *inodebuf;
308
309struct dinode *
310getnextinode(inumber)
311	ino_t inumber;
312{
313	long size;
314	ufs_daddr_t dblk;
315	static struct dinode *dp;
316
317	if (inumber != nextino++ || inumber > maxino)
318		errx(EEXIT, "bad inode number %d to nextinode", inumber);
319	if (inumber >= lastinum) {
320		readcnt++;
321		dblk = fsbtodb(&sblock, ino_to_fsba(&sblock, lastinum));
322		if (readcnt % readpercg == 0) {
323			size = partialsize;
324			lastinum += partialcnt;
325		} else {
326			size = inobufsize;
327			lastinum += fullcnt;
328		}
329		/*
330		 * If bread returns an error, it will already have zeroed
331		 * out the buffer, so we do not need to do so here.
332		 */
333		(void)bread(fsreadfd, (char *)inodebuf, dblk, size);
334		dp = inodebuf;
335	}
336	return (dp++);
337}
338
339void
340setinodebuf(inum)
341	ino_t inum;
342{
343
344	if (inum % sblock.fs_ipg != 0)
345		errx(EEXIT, "bad inode number %d to setinodebuf", inum);
346	startinum = 0;
347	nextino = inum;
348	lastinum = inum;
349	readcnt = 0;
350	if (inodebuf != NULL)
351		return;
352	inobufsize = blkroundup(&sblock, INOBUFSIZE);
353	fullcnt = inobufsize / sizeof(struct dinode);
354	readpercg = sblock.fs_ipg / fullcnt;
355	partialcnt = sblock.fs_ipg % fullcnt;
356	partialsize = partialcnt * sizeof(struct dinode);
357	if (partialcnt != 0) {
358		readpercg++;
359	} else {
360		partialcnt = fullcnt;
361		partialsize = inobufsize;
362	}
363	if ((inodebuf = (struct dinode *)malloc((unsigned)inobufsize)) == NULL)
364		errx(EEXIT, "cannot allocate space for inode buffer");
365}
366
367void
368freeinodebuf()
369{
370
371	if (inodebuf != NULL)
372		free((char *)inodebuf);
373	inodebuf = NULL;
374}
375
376/*
377 * Routines to maintain information about directory inodes.
378 * This is built during the first pass and used during the
379 * second and third passes.
380 *
381 * Enter inodes into the cache.
382 */
383void
384cacheino(dp, inumber)
385	register struct dinode *dp;
386	ino_t inumber;
387{
388	register struct inoinfo *inp;
389	struct inoinfo **inpp;
390	int blks;
391
392	blks = howmany(dp->di_size, sblock.fs_bsize);
393	if (blks > NDADDR)
394		blks = NDADDR + NIADDR;
395	inp = (struct inoinfo *)
396		malloc(sizeof(*inp) + (blks - 1) * sizeof(ufs_daddr_t));
397	if (inp == NULL)
398		errx(EEXIT, "cannot increase directory list");
399	inpp = &inphead[inumber % dirhash];
400	inp->i_nexthash = *inpp;
401	*inpp = inp;
402	inp->i_parent = inumber == ROOTINO ? ROOTINO : (ino_t)0;
403	inp->i_dotdot = (ino_t)0;
404	inp->i_number = inumber;
405	inp->i_isize = dp->di_size;
406	inp->i_numblks = blks * sizeof(ufs_daddr_t);
407	memmove(&inp->i_blks[0], &dp->di_db[0], (size_t)inp->i_numblks);
408	if (inplast == listmax) {
409		listmax += 100;
410		inpsort = (struct inoinfo **)realloc((char *)inpsort,
411		    (unsigned)listmax * sizeof(struct inoinfo *));
412		if (inpsort == NULL)
413			errx(EEXIT, "cannot increase directory list");
414	}
415	inpsort[inplast++] = inp;
416}
417
418/*
419 * Look up an inode cache structure.
420 */
421struct inoinfo *
422getinoinfo(inumber)
423	ino_t inumber;
424{
425	register struct inoinfo *inp;
426
427	for (inp = inphead[inumber % dirhash]; inp; inp = inp->i_nexthash) {
428		if (inp->i_number != inumber)
429			continue;
430		return (inp);
431	}
432	errx(EEXIT, "cannot find inode %d", inumber);
433	return ((struct inoinfo *)0);
434}
435
436/*
437 * Clean up all the inode cache structure.
438 */
439void
440inocleanup()
441{
442	register struct inoinfo **inpp;
443
444	if (inphead == NULL)
445		return;
446	for (inpp = &inpsort[inplast - 1]; inpp >= inpsort; inpp--)
447		free((char *)(*inpp));
448	free((char *)inphead);
449	free((char *)inpsort);
450	inphead = inpsort = NULL;
451}
452
453void
454inodirty()
455{
456
457	dirty(pbp);
458}
459
460void
461clri(idesc, type, flag)
462	register struct inodesc *idesc;
463	char *type;
464	int flag;
465{
466	register struct dinode *dp;
467
468	dp = ginode(idesc->id_number);
469	if (flag == 1) {
470		pwarn("%s %s", type,
471		    (dp->di_mode & IFMT) == IFDIR ? "DIR" : "FILE");
472		pinode(idesc->id_number);
473	}
474	if (preen || reply("CLEAR") == 1) {
475		if (preen)
476			printf(" (CLEARED)\n");
477		n_files--;
478		(void)ckinode(dp, idesc);
479		clearinode(dp);
480		inoinfo(idesc->id_number)->ino_state = USTATE;
481		inodirty();
482	}
483}
484
485int
486findname(idesc)
487	struct inodesc *idesc;
488{
489	register struct direct *dirp = idesc->id_dirp;
490
491	if (dirp->d_ino != idesc->id_parent || idesc->id_entryno < 2) {
492		idesc->id_entryno++;
493		return (KEEPON);
494	}
495	memmove(idesc->id_name, dirp->d_name, (size_t)dirp->d_namlen + 1);
496	return (STOP|FOUND);
497}
498
499int
500findino(idesc)
501	struct inodesc *idesc;
502{
503	register struct direct *dirp = idesc->id_dirp;
504
505	if (dirp->d_ino == 0)
506		return (KEEPON);
507	if (strcmp(dirp->d_name, idesc->id_name) == 0 &&
508	    dirp->d_ino >= ROOTINO && dirp->d_ino <= maxino) {
509		idesc->id_parent = dirp->d_ino;
510		return (STOP|FOUND);
511	}
512	return (KEEPON);
513}
514
515int
516clearentry(idesc)
517	struct inodesc *idesc;
518{
519	register struct direct *dirp = idesc->id_dirp;
520
521	if (dirp->d_ino != idesc->id_parent || idesc->id_entryno < 2) {
522		idesc->id_entryno++;
523		return (KEEPON);
524	}
525	dirp->d_ino = 0;
526	return (STOP|FOUND|ALTERED);
527}
528
529void
530pinode(ino)
531	ino_t ino;
532{
533	register struct dinode *dp;
534	register char *p;
535	struct passwd *pw;
536	time_t t;
537
538	printf(" I=%lu ", (u_long)ino);
539	if (ino < ROOTINO || ino > maxino)
540		return;
541	dp = ginode(ino);
542	printf(" OWNER=");
543	if ((pw = getpwuid((int)dp->di_uid)) != 0)
544		printf("%s ", pw->pw_name);
545	else
546		printf("%u ", (unsigned)dp->di_uid);
547	printf("MODE=%o\n", dp->di_mode);
548	if (preen)
549		printf("%s: ", cdevname);
550	printf("SIZE=%qu ", dp->di_size);
551	t = dp->di_mtime;
552	p = ctime(&t);
553	printf("MTIME=%12.12s %4.4s ", &p[4], &p[20]);
554}
555
556void
557blkerror(ino, type, blk)
558	ino_t ino;
559	char *type;
560	ufs_daddr_t blk;
561{
562
563	pfatal("%ld %s I=%lu", blk, type, ino);
564	printf("\n");
565	switch (inoinfo(ino)->ino_state) {
566
567	case FSTATE:
568		inoinfo(ino)->ino_state = FCLEAR;
569		return;
570
571	case DSTATE:
572		inoinfo(ino)->ino_state = DCLEAR;
573		return;
574
575	case FCLEAR:
576	case DCLEAR:
577		return;
578
579	default:
580		errx(EEXIT, "BAD STATE %d TO BLKERR", inoinfo(ino)->ino_state);
581		/* NOTREACHED */
582	}
583}
584
585/*
586 * allocate an unused inode
587 */
588ino_t
589allocino(request, type)
590	ino_t request;
591	int type;
592{
593	register ino_t ino;
594	register struct dinode *dp;
595	struct cg *cgp = &cgrp;
596	int cg;
597
598	if (request == 0)
599		request = ROOTINO;
600	else if (inoinfo(request)->ino_state != USTATE)
601		return (0);
602	for (ino = request; ino < maxino; ino++)
603		if (inoinfo(ino)->ino_state == USTATE)
604			break;
605	if (ino == maxino)
606		return (0);
607	cg = ino_to_cg(&sblock, ino);
608	getblk(&cgblk, cgtod(&sblock, cg), sblock.fs_cgsize);
609	if (!cg_chkmagic(cgp))
610		pfatal("CG %d: BAD MAGIC NUMBER\n", cg);
611	setbit(cg_inosused(cgp), ino % sblock.fs_ipg);
612	cgp->cg_cs.cs_nifree--;
613	switch (type & IFMT) {
614	case IFDIR:
615		inoinfo(ino)->ino_state = DSTATE;
616		cgp->cg_cs.cs_ndir++;
617		break;
618	case IFREG:
619	case IFLNK:
620		inoinfo(ino)->ino_state = FSTATE;
621		break;
622	default:
623		return (0);
624	}
625	cgdirty();
626	dp = ginode(ino);
627	dp->di_db[0] = allocblk((long)1);
628	if (dp->di_db[0] == 0) {
629		inoinfo(ino)->ino_state = USTATE;
630		return (0);
631	}
632	dp->di_mode = type;
633	dp->di_flags = 0;
634	dp->di_atime = time(NULL);
635	dp->di_mtime = dp->di_ctime = dp->di_atime;
636	dp->di_mtimensec = dp->di_ctimensec = dp->di_atimensec = 0;
637	dp->di_size = sblock.fs_fsize;
638	dp->di_blocks = btodb(sblock.fs_fsize);
639	n_files++;
640	inodirty();
641	if (newinofmt)
642		inoinfo(ino)->ino_type = IFTODT(type);
643	return (ino);
644}
645
646/*
647 * deallocate an inode
648 */
649void
650freeino(ino)
651	ino_t ino;
652{
653	struct inodesc idesc;
654	struct dinode *dp;
655
656	memset(&idesc, 0, sizeof(struct inodesc));
657	idesc.id_type = ADDR;
658	idesc.id_func = pass4check;
659	idesc.id_number = ino;
660	dp = ginode(ino);
661	(void)ckinode(dp, &idesc);
662	clearinode(dp);
663	inodirty();
664	inoinfo(ino)->ino_state = USTATE;
665	n_files--;
666}
667