2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1980, 1986, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 static const char sccsid[] = "@(#)inode.c 8.8 (Berkeley) 4/28/95";
37 #include <sys/cdefs.h>
38 __FBSDID("$FreeBSD$");
40 #include <sys/param.h>
41 #include <sys/stdint.h>
42 #include <sys/sysctl.h>
44 #include <ufs/ufs/dinode.h>
45 #include <ufs/ufs/dir.h>
46 #include <ufs/ffs/fs.h>
55 static ino_t startinum;
57 static int iblock(struct inodesc *, off_t isize, int type);
60 ckinode(union dinode *dp, struct inodesc *idesc)
62 off_t remsize, sizepb;
67 char pathbuf[MAXPATHLEN + 1];
69 if (idesc->id_fix != IGNORE)
70 idesc->id_fix = DONTKNOW;
72 idesc->id_lballoc = -1;
74 idesc->id_entryno = 0;
75 idesc->id_filesize = DIP(dp, di_size);
76 mode = DIP(dp, di_mode) & IFMT;
77 if (mode == IFBLK || mode == IFCHR || (mode == IFLNK &&
78 DIP(dp, di_size) < (unsigned)sblock.fs_maxsymlinklen))
80 if (sblock.fs_magic == FS_UFS1_MAGIC)
84 ndb = howmany(DIP(&dino, di_size), sblock.fs_bsize);
85 for (i = 0; i < UFS_NDADDR; i++) {
88 (offset = blkoff(&sblock, DIP(&dino, di_size))) != 0)
90 numfrags(&sblock, fragroundup(&sblock, offset));
92 idesc->id_numfrags = sblock.fs_frag;
93 if (DIP(&dino, di_db[i]) == 0) {
94 if (idesc->id_type == DATA && ndb >= 0) {
95 /* An empty block in a directory XXX */
96 getpathname(pathbuf, idesc->id_number,
98 pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
100 if (reply("ADJUST LENGTH") == 1) {
101 dp = ginode(idesc->id_number);
103 i * sblock.fs_bsize);
105 "YOU MUST RERUN FSCK AFTERWARDS\n");
113 idesc->id_blkno = DIP(&dino, di_db[i]);
114 if (idesc->id_type != DATA)
115 ret = (*idesc->id_func)(idesc);
117 ret = dirscan(idesc);
121 idesc->id_numfrags = sblock.fs_frag;
122 remsize = DIP(&dino, di_size) - sblock.fs_bsize * UFS_NDADDR;
123 sizepb = sblock.fs_bsize;
124 for (i = 0; i < UFS_NIADDR; i++) {
125 sizepb *= NINDIR(&sblock);
126 idesc->id_level = i + 1;
127 if (DIP(&dino, di_ib[i])) {
128 idesc->id_blkno = DIP(&dino, di_ib[i]);
129 ret = iblock(idesc, remsize, BT_LEVEL1 + i);
132 } else if (remsize > 0) {
133 idesc->id_lbn += sizepb / sblock.fs_bsize;
134 if (idesc->id_type == DATA) {
135 /* An empty block in a directory XXX */
136 getpathname(pathbuf, idesc->id_number,
138 pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
140 if (reply("ADJUST LENGTH") == 1) {
141 dp = ginode(idesc->id_number);
143 DIP(dp, di_size) - remsize);
146 "YOU MUST RERUN FSCK AFTERWARDS\n");
159 iblock(struct inodesc *idesc, off_t isize, int type)
162 int i, n, (*func)(struct inodesc *), nif;
165 char pathbuf[MAXPATHLEN + 1];
168 if (idesc->id_type != DATA) {
169 func = idesc->id_func;
170 if (((n = (*func)(idesc)) & KEEPON) == 0)
174 if (chkrange(idesc->id_blkno, idesc->id_numfrags))
176 bp = getdatablk(idesc->id_blkno, sblock.fs_bsize, type);
178 for (sizepb = sblock.fs_bsize, i = 0; i < idesc->id_level; i++)
179 sizepb *= NINDIR(&sblock);
180 if (howmany(isize, sizepb) > NINDIR(&sblock))
181 nif = NINDIR(&sblock);
183 nif = howmany(isize, sizepb);
184 if (idesc->id_func == pass1check && nif < NINDIR(&sblock)) {
185 for (i = nif; i < NINDIR(&sblock); i++) {
186 if (IBLK(bp, i) == 0)
188 (void)sprintf(buf, "PARTIALLY TRUNCATED INODE I=%lu",
189 (u_long)idesc->id_number);
192 } else if (dofix(idesc, buf)) {
197 flush(fswritefd, bp);
199 for (i = 0; i < nif; i++) {
201 idesc->id_blkno = IBLK(bp, i);
202 if (idesc->id_level == 0) {
206 n = iblock(idesc, isize, type);
210 bp->b_flags &= ~B_INUSE;
214 idesc->id_lbn += sizepb / sblock.fs_bsize;
215 if (idesc->id_type == DATA && isize > 0) {
216 /* An empty block in a directory XXX */
217 getpathname(pathbuf, idesc->id_number,
219 pfatal("DIRECTORY %s: CONTAINS EMPTY BLOCKS",
221 if (reply("ADJUST LENGTH") == 1) {
222 dp = ginode(idesc->id_number);
224 DIP(dp, di_size) - isize);
227 "YOU MUST RERUN FSCK AFTERWARDS\n");
230 bp->b_flags &= ~B_INUSE;
237 bp->b_flags &= ~B_INUSE;
242 * Check that a block in a legal block number.
243 * Return 0 if in range, 1 if out of range.
246 chkrange(ufs2_daddr_t blk, int cnt)
250 if (cnt <= 0 || blk <= 0 || blk > maxfsblock ||
251 cnt - 1 > maxfsblock - blk)
253 if (cnt > sblock.fs_frag ||
254 fragnum(&sblock, blk) + cnt > sblock.fs_frag) {
256 printf("bad size: blk %ld, offset %i, size %d\n",
257 (long)blk, (int)fragnum(&sblock, blk), cnt);
260 c = dtog(&sblock, blk);
261 if (blk < cgdmin(&sblock, c)) {
262 if ((blk + cnt) > cgsblock(&sblock, c)) {
264 printf("blk %ld < cgdmin %ld;",
265 (long)blk, (long)cgdmin(&sblock, c));
266 printf(" blk + cnt %ld > cgsbase %ld\n",
268 (long)cgsblock(&sblock, c));
273 if ((blk + cnt) > cgbase(&sblock, c+1)) {
275 printf("blk %ld >= cgdmin %ld;",
276 (long)blk, (long)cgdmin(&sblock, c));
277 printf(" blk + cnt %ld > sblock.fs_fpg %ld\n",
278 (long)(blk + cnt), (long)sblock.fs_fpg);
287 * General purpose interface for reading inodes.
290 ginode(ino_t inumber)
294 if (inumber < UFS_ROOTINO || inumber > maxino)
295 errx(EEXIT, "bad inode number %ju to ginode",
297 if (startinum == 0 ||
298 inumber < startinum || inumber >= startinum + INOPB(&sblock)) {
299 iblk = ino_to_fsba(&sblock, inumber);
301 pbp->b_flags &= ~B_INUSE;
302 pbp = getdatablk(iblk, sblock.fs_bsize, BT_INODES);
303 startinum = rounddown(inumber, INOPB(&sblock));
305 if (sblock.fs_magic == FS_UFS1_MAGIC)
306 return ((union dinode *)
307 &pbp->b_un.b_dinode1[inumber % INOPB(&sblock)]);
308 return ((union dinode *)&pbp->b_un.b_dinode2[inumber % INOPB(&sblock)]);
312 * Special purpose version of ginode used to optimize first pass
313 * over all the inodes in numerical order.
315 static ino_t nextino, lastinum, lastvalidinum;
316 static long readcount, readpercg, fullcnt, inobufsize, partialcnt, partialsize;
317 static struct bufarea inobuf;
320 getnextinode(ino_t inumber, int rebuildcg)
325 ufs2_daddr_t ndb, blk;
327 static caddr_t nextinop;
329 if (inumber != nextino++ || inumber > lastvalidinum)
330 errx(EEXIT, "bad inode number %ju to nextinode",
332 if (inumber >= lastinum) {
334 blk = ino_to_fsba(&sblock, lastinum);
335 if (readcount % readpercg == 0) {
337 lastinum += partialcnt;
343 * If getblk encounters an error, it will already have zeroed
344 * out the buffer, so we do not need to do so here.
346 getblk(&inobuf, blk, size);
347 nextinop = inobuf.b_un.b_buf;
349 dp = (union dinode *)nextinop;
350 if (rebuildcg && nextinop == inobuf.b_un.b_buf) {
352 * Try to determine if we have reached the end of the
355 mode = DIP(dp, di_mode) & IFMT;
357 if (memcmp(dp->dp2.di_db, ufs2_zino.di_db,
358 UFS_NDADDR * sizeof(ufs2_daddr_t)) ||
359 memcmp(dp->dp2.di_ib, ufs2_zino.di_ib,
360 UFS_NIADDR * sizeof(ufs2_daddr_t)) ||
361 dp->dp2.di_mode || dp->dp2.di_size)
367 ndb = howmany(DIP(dp, di_size), sblock.fs_bsize);
370 if (mode == IFBLK || mode == IFCHR)
374 * Fake ndb value so direct/indirect block checks below
375 * will detect any garbage after symlink string.
377 if (DIP(dp, di_size) < (off_t)sblock.fs_maxsymlinklen) {
378 ndb = howmany(DIP(dp, di_size),
379 sizeof(ufs2_daddr_t));
380 if (ndb > UFS_NDADDR) {
381 j = ndb - UFS_NDADDR;
382 for (ndb = 1; j > 1; j--)
383 ndb *= NINDIR(&sblock);
388 for (j = ndb; ndb < UFS_NDADDR && j < UFS_NDADDR; j++)
389 if (DIP(dp, di_db[j]) != 0)
391 for (j = 0, ndb -= UFS_NDADDR; ndb > 0; j++)
392 ndb /= NINDIR(&sblock);
393 for (; j < UFS_NIADDR; j++)
394 if (DIP(dp, di_ib[j]) != 0)
398 if (sblock.fs_magic == FS_UFS1_MAGIC)
399 nextinop += sizeof(struct ufs1_dinode);
401 nextinop += sizeof(struct ufs2_dinode);
406 setinodebuf(ino_t inum)
409 if (inum % sblock.fs_ipg != 0)
410 errx(EEXIT, "bad inode number %ju to setinodebuf",
412 lastvalidinum = inum + sblock.fs_ipg - 1;
417 if (inobuf.b_un.b_buf != NULL)
419 inobufsize = blkroundup(&sblock, INOBUFSIZE);
420 fullcnt = inobufsize / ((sblock.fs_magic == FS_UFS1_MAGIC) ?
421 sizeof(struct ufs1_dinode) : sizeof(struct ufs2_dinode));
422 readpercg = sblock.fs_ipg / fullcnt;
423 partialcnt = sblock.fs_ipg % fullcnt;
424 partialsize = partialcnt * ((sblock.fs_magic == FS_UFS1_MAGIC) ?
425 sizeof(struct ufs1_dinode) : sizeof(struct ufs2_dinode));
426 if (partialcnt != 0) {
429 partialcnt = fullcnt;
430 partialsize = inobufsize;
432 initbarea(&inobuf, BT_INODES);
433 if ((inobuf.b_un.b_buf = Malloc((unsigned)inobufsize)) == NULL)
434 errx(EEXIT, "cannot allocate space for inode buffer");
441 if (inobuf.b_un.b_buf != NULL)
442 free((char *)inobuf.b_un.b_buf);
443 inobuf.b_un.b_buf = NULL;
447 * Routines to maintain information about directory inodes.
448 * This is built during the first pass and used during the
449 * second and third passes.
451 * Enter inodes into the cache.
454 cacheino(union dinode *dp, ino_t inumber)
456 struct inoinfo *inp, **inpp;
459 if (howmany(DIP(dp, di_size), sblock.fs_bsize) > UFS_NDADDR)
460 blks = UFS_NDADDR + UFS_NIADDR;
461 else if (DIP(dp, di_size) > 0)
462 blks = howmany(DIP(dp, di_size), sblock.fs_bsize);
465 inp = (struct inoinfo *)
466 Malloc(sizeof(*inp) + (blks - 1) * sizeof(ufs2_daddr_t));
468 errx(EEXIT, "cannot increase directory list");
469 inpp = &inphead[inumber % dirhash];
470 inp->i_nexthash = *inpp;
472 inp->i_parent = inumber == UFS_ROOTINO ? UFS_ROOTINO : (ino_t)0;
473 inp->i_dotdot = (ino_t)0;
474 inp->i_number = inumber;
475 inp->i_isize = DIP(dp, di_size);
476 inp->i_numblks = blks;
477 for (i = 0; i < MIN(blks, UFS_NDADDR); i++)
478 inp->i_blks[i] = DIP(dp, di_db[i]);
479 if (blks > UFS_NDADDR)
480 for (i = 0; i < UFS_NIADDR; i++)
481 inp->i_blks[UFS_NDADDR + i] = DIP(dp, di_ib[i]);
482 if (inplast == listmax) {
484 inpsort = (struct inoinfo **)reallocarray((char *)inpsort,
485 listmax, sizeof(struct inoinfo *));
487 errx(EEXIT, "cannot increase directory list");
489 inpsort[inplast++] = inp;
493 * Look up an inode cache structure.
496 getinoinfo(ino_t inumber)
500 for (inp = inphead[inumber % dirhash]; inp; inp = inp->i_nexthash) {
501 if (inp->i_number != inumber)
505 errx(EEXIT, "cannot find inode %ju", (uintmax_t)inumber);
506 return ((struct inoinfo *)0);
510 * Clean up all the inode cache structure.
515 struct inoinfo **inpp;
519 for (inpp = &inpsort[inplast - 1]; inpp >= inpsort; inpp--)
520 free((char *)(*inpp));
521 free((char *)inphead);
522 free((char *)inpsort);
523 inphead = inpsort = NULL;
527 inodirty(union dinode *dp)
534 clri(struct inodesc *idesc, const char *type, int flag)
538 dp = ginode(idesc->id_number);
541 (DIP(dp, di_mode) & IFMT) == IFDIR ? "DIR" : "FILE");
542 pinode(idesc->id_number);
544 if (preen || reply("CLEAR") == 1) {
546 printf(" (CLEARED)\n");
548 if (bkgrdflag == 0) {
549 (void)ckinode(dp, idesc);
550 inoinfo(idesc->id_number)->ino_state = USTATE;
554 cmd.value = idesc->id_number;
555 cmd.size = -DIP(dp, di_nlink);
557 printf("adjrefcnt ino %ld amt %lld\n",
558 (long)cmd.value, (long long)cmd.size);
559 if (sysctl(adjrefcnt, MIBSIZE, 0, 0,
560 &cmd, sizeof cmd) == -1)
561 rwerror("ADJUST INODE", cmd.value);
567 findname(struct inodesc *idesc)
569 struct direct *dirp = idesc->id_dirp;
571 if (dirp->d_ino != idesc->id_parent || idesc->id_entryno < 2) {
575 memmove(idesc->id_name, dirp->d_name, (size_t)dirp->d_namlen + 1);
580 findino(struct inodesc *idesc)
582 struct direct *dirp = idesc->id_dirp;
584 if (dirp->d_ino == 0)
586 if (strcmp(dirp->d_name, idesc->id_name) == 0 &&
587 dirp->d_ino >= UFS_ROOTINO && dirp->d_ino <= maxino) {
588 idesc->id_parent = dirp->d_ino;
595 clearentry(struct inodesc *idesc)
597 struct direct *dirp = idesc->id_dirp;
599 if (dirp->d_ino != idesc->id_parent || idesc->id_entryno < 2) {
604 return (STOP|FOUND|ALTERED);
615 printf(" I=%lu ", (u_long)ino);
616 if (ino < UFS_ROOTINO || ino > maxino)
620 if ((pw = getpwuid((int)DIP(dp, di_uid))) != NULL)
621 printf("%s ", pw->pw_name);
623 printf("%u ", (unsigned)DIP(dp, di_uid));
624 printf("MODE=%o\n", DIP(dp, di_mode));
626 printf("%s: ", cdevname);
627 printf("SIZE=%ju ", (uintmax_t)DIP(dp, di_size));
628 t = DIP(dp, di_mtime);
630 printf("MTIME=%12.12s %4.4s ", &p[4], &p[20]);
634 blkerror(ino_t ino, const char *type, ufs2_daddr_t blk)
637 pfatal("%jd %s I=%ju", (intmax_t)blk, type, (uintmax_t)ino);
639 switch (inoinfo(ino)->ino_state) {
643 inoinfo(ino)->ino_state = FCLEAR;
648 inoinfo(ino)->ino_state = DCLEAR;
656 errx(EEXIT, "BAD STATE %d TO BLKERR", inoinfo(ino)->ino_state);
662 * allocate an unused inode
665 allocino(ino_t request, int type)
669 struct bufarea *cgbp;
674 request = UFS_ROOTINO;
675 else if (inoinfo(request)->ino_state != USTATE)
677 for (ino = request; ino < maxino; ino++)
678 if (inoinfo(ino)->ino_state == USTATE)
682 cg = ino_to_cg(&sblock, ino);
684 cgp = cgbp->b_un.b_cg;
685 if (!check_cgmagic(cg, cgbp))
687 setbit(cg_inosused(cgp), ino % sblock.fs_ipg);
688 cgp->cg_cs.cs_nifree--;
689 switch (type & IFMT) {
691 inoinfo(ino)->ino_state = DSTATE;
692 cgp->cg_cs.cs_ndir++;
696 inoinfo(ino)->ino_state = FSTATE;
703 DIP_SET(dp, di_db[0], allocblk((long)1));
704 if (DIP(dp, di_db[0]) == 0) {
705 inoinfo(ino)->ino_state = USTATE;
708 DIP_SET(dp, di_mode, type);
709 DIP_SET(dp, di_flags, 0);
710 DIP_SET(dp, di_atime, time(NULL));
711 DIP_SET(dp, di_ctime, DIP(dp, di_atime));
712 DIP_SET(dp, di_mtime, DIP(dp, di_ctime));
713 DIP_SET(dp, di_mtimensec, 0);
714 DIP_SET(dp, di_ctimensec, 0);
715 DIP_SET(dp, di_atimensec, 0);
716 DIP_SET(dp, di_size, sblock.fs_fsize);
717 DIP_SET(dp, di_blocks, btodb(sblock.fs_fsize));
720 inoinfo(ino)->ino_type = IFTODT(type);
725 * deallocate an inode
730 struct inodesc idesc;
733 memset(&idesc, 0, sizeof(struct inodesc));
734 idesc.id_type = ADDR;
735 idesc.id_func = pass4check;
736 idesc.id_number = ino;
738 (void)ckinode(dp, &idesc);
741 inoinfo(ino)->ino_state = USTATE;