2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1989, 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
31 * @(#)ffs_subr.c 8.5 (Berkeley) 3/21/95
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
37 #include <sys/param.h>
44 #include <sys/errno.h>
45 #include <ufs/ufs/dinode.h>
46 #include <ufs/ffs/fs.h>
48 uint32_t calculate_crc32c(uint32_t, const void *, size_t);
49 uint32_t ffs_calc_sbhash(struct fs *);
51 #define UFS_MALLOC(size, type, flags) malloc(size)
52 #define UFS_FREE(ptr, type) free(ptr)
53 #define UFS_TIME time(NULL)
55 * Request standard superblock location in ffs_sbget
57 #define STDSB -1 /* Fail if check-hash is bad */
58 #define STDSB_NOHASHFAIL -2 /* Ignore check-hash failure */
61 #include <sys/systm.h>
62 #include <sys/gsb_crc32.h>
64 #include <sys/malloc.h>
65 #include <sys/mount.h>
66 #include <sys/vnode.h>
69 #include <sys/ucred.h>
71 #include <ufs/ufs/quota.h>
72 #include <ufs/ufs/inode.h>
73 #include <ufs/ufs/extattr.h>
74 #include <ufs/ufs/ufsmount.h>
75 #include <ufs/ufs/ufs_extern.h>
76 #include <ufs/ffs/ffs_extern.h>
77 #include <ufs/ffs/fs.h>
79 #define UFS_MALLOC(size, type, flags) malloc(size, type, flags)
80 #define UFS_FREE(ptr, type) free(ptr, type)
81 #define UFS_TIME time_second
84 * Return buffer with the contents of block "offset" from the beginning of
85 * directory "ip". If "res" is non-zero, fill it in with a pointer to the
86 * remaining space in the directory.
89 ffs_blkatoff(struct vnode *vp, off_t offset, char **res, struct buf **bpp)
99 lbn = lblkno(fs, offset);
100 bsize = blksize(fs, ip, lbn);
103 error = bread(vp, lbn, bsize, NOCRED, &bp);
109 *res = (char *)bp->b_data + blkoff(fs, offset);
115 * Load up the contents of an inode and copy the appropriate pieces
116 * to the incore copy.
119 ffs_load_inode(struct buf *bp, struct inode *ip, struct fs *fs, ino_t ino)
121 struct ufs1_dinode *dip1;
122 struct ufs2_dinode *dip2;
128 *((struct ufs1_dinode *)bp->b_data + ino_to_fsbo(fs, ino));
129 ip->i_mode = dip1->di_mode;
130 ip->i_nlink = dip1->di_nlink;
131 ip->i_effnlink = dip1->di_nlink;
132 ip->i_size = dip1->di_size;
133 ip->i_flags = dip1->di_flags;
134 ip->i_gen = dip1->di_gen;
135 ip->i_uid = dip1->di_uid;
136 ip->i_gid = dip1->di_gid;
139 dip2 = ((struct ufs2_dinode *)bp->b_data + ino_to_fsbo(fs, ino));
140 if ((error = ffs_verify_dinode_ckhash(fs, dip2)) != 0) {
141 printf("%s: inode %jd: check-hash failed\n", fs->fs_fsmnt,
147 ip->i_mode = dip2->di_mode;
148 ip->i_nlink = dip2->di_nlink;
149 ip->i_effnlink = dip2->di_nlink;
150 ip->i_size = dip2->di_size;
151 ip->i_flags = dip2->di_flags;
152 ip->i_gen = dip2->di_gen;
153 ip->i_uid = dip2->di_uid;
154 ip->i_gid = dip2->di_gid;
159 * Verify that a filesystem block number is a valid data block.
160 * This routine is only called on untrusted filesystems.
163 ffs_check_blkno(struct mount *mp, ino_t inum, ufs2_daddr_t daddr, int blksize)
166 struct ufsmount *ump;
167 ufs2_daddr_t end_daddr;
170 KASSERT((mp->mnt_flag & MNT_UNTRUSTED) != 0,
171 ("ffs_check_blkno called on a trusted file system"));
174 cg = dtog(fs, daddr);
175 end_daddr = daddr + numfrags(fs, blksize);
177 * Verify that the block number is a valid data block. Also check
178 * that it does not point to an inode block or a superblock. Accept
179 * blocks that are unalloacted (0) or part of snapshot metadata
180 * (BLK_NOCOPY or BLK_SNAP).
182 * Thus, the block must be in a valid range for the filesystem and
183 * either in the space before a backup superblock (except the first
184 * cylinder group where that space is used by the bootstrap code) or
185 * after the inode blocks and before the end of the cylinder group.
187 if ((uint64_t)daddr <= BLK_SNAP ||
188 ((uint64_t)end_daddr <= fs->fs_size &&
189 ((cg > 0 && end_daddr <= cgsblock(fs, cg)) ||
190 (daddr >= cgdmin(fs, cg) &&
191 end_daddr <= cgbase(fs, cg) + fs->fs_fpg))))
193 if ((havemtx = mtx_owned(UFS_MTX(ump))) == 0)
195 if (ppsratecheck(&ump->um_last_integritymsg,
196 &ump->um_secs_integritymsg, 1)) {
198 uprintf("\n%s: inode %jd, out-of-range indirect block "
199 "number %jd\n", mp->mnt_stat.f_mntonname, inum, daddr);
209 * Verify an inode check-hash.
212 ffs_verify_dinode_ckhash(struct fs *fs, struct ufs2_dinode *dip)
214 uint32_t ckhash, save_ckhash;
217 * Return success if unallocated or we are not doing inode check-hash.
219 if (dip->di_mode == 0 || (fs->fs_metackhash & CK_INODE) == 0)
222 * Exclude di_ckhash from the crc32 calculation, e.g., always use
223 * a check-hash value of zero when calculating the check-hash.
225 save_ckhash = dip->di_ckhash;
227 ckhash = calculate_crc32c(~0L, (void *)dip, sizeof(*dip));
228 dip->di_ckhash = save_ckhash;
229 if (save_ckhash == ckhash)
235 * Update an inode check-hash.
238 ffs_update_dinode_ckhash(struct fs *fs, struct ufs2_dinode *dip)
241 if (dip->di_mode == 0 || (fs->fs_metackhash & CK_INODE) == 0)
244 * Exclude old di_ckhash from the crc32 calculation, e.g., always use
245 * a check-hash value of zero when calculating the new check-hash.
248 dip->di_ckhash = calculate_crc32c(~0L, (void *)dip, sizeof(*dip));
252 * These are the low-level functions that actually read and write
253 * the superblock and its associated data.
255 static off_t sblock_try[] = SBLOCKSEARCH;
256 static int readsuper(void *, struct fs **, off_t, int, int,
257 int (*)(void *, off_t, void **, int));
260 * Read a superblock from the devfd device.
262 * If an alternate superblock is specified, it is read. Otherwise the
263 * set of locations given in the SBLOCKSEARCH list is searched for a
264 * superblock. Memory is allocated for the superblock by the readfunc and
265 * is returned. If filltype is non-NULL, additional memory is allocated
266 * of type filltype and filled in with the superblock summary information.
267 * All memory is freed when any error is returned.
269 * If a superblock is found, zero is returned. Otherwise one of the
270 * following error values is returned:
271 * EIO: non-existent or truncated superblock.
272 * EIO: error reading summary information.
273 * ENOENT: no usable known superblock found.
274 * ENOSPC: failed to allocate space for the superblock.
275 * EINVAL: The previous newfs operation on this volume did not complete.
276 * The administrator must complete newfs before using this volume.
279 ffs_sbget(void *devfd, struct fs **fsp, off_t altsblock,
280 struct malloc_type *filltype,
281 int (*readfunc)(void *devfd, off_t loc, void **bufp, int size))
284 int i, error, size, blks;
293 if (altsblock >= 0) {
294 if ((error = readsuper(devfd, &fs, altsblock, 1, chkhash,
297 UFS_FREE(fs, filltype);
301 if (altsblock == STDSB_NOHASHFAIL)
303 for (i = 0; sblock_try[i] != -1; i++) {
304 if ((error = readsuper(devfd, &fs, sblock_try[i], 0,
305 chkhash, readfunc)) == 0)
308 UFS_FREE(fs, filltype);
315 if (sblock_try[i] == -1)
319 * Read in the superblock summary information.
321 size = fs->fs_cssize;
322 blks = howmany(size, fs->fs_fsize);
323 if (fs->fs_contigsumsize > 0)
324 size += fs->fs_ncg * sizeof(int32_t);
325 size += fs->fs_ncg * sizeof(u_int8_t);
326 /* When running in libufs or libsa, UFS_MALLOC may fail */
327 if ((space = UFS_MALLOC(size, filltype, M_WAITOK)) == NULL) {
328 UFS_FREE(fs, filltype);
331 fs->fs_csp = (struct csum *)space;
332 for (i = 0; i < blks; i += fs->fs_frag) {
334 if (i + fs->fs_frag > blks)
335 size = (blks - i) * fs->fs_fsize;
337 error = (*readfunc)(devfd,
338 dbtob(fsbtodb(fs, fs->fs_csaddr + i)), (void **)&buf, size);
341 UFS_FREE(buf, filltype);
342 UFS_FREE(fs->fs_csp, filltype);
343 UFS_FREE(fs, filltype);
346 memcpy(space, buf, size);
347 UFS_FREE(buf, filltype);
350 if (fs->fs_contigsumsize > 0) {
351 fs->fs_maxcluster = lp = (int32_t *)space;
352 for (i = 0; i < fs->fs_ncg; i++)
353 *lp++ = fs->fs_contigsumsize;
354 space = (uint8_t *)lp;
356 size = fs->fs_ncg * sizeof(u_int8_t);
357 fs->fs_contigdirs = (u_int8_t *)space;
358 bzero(fs->fs_contigdirs, size);
364 * Try to read a superblock from the location specified by sblockloc.
365 * Return zero on success or an errno on failure.
368 readsuper(void *devfd, struct fs **fsp, off_t sblockloc, int isaltsblk,
369 int chkhash, int (*readfunc)(void *devfd, off_t loc, void **bufp, int size))
375 error = (*readfunc)(devfd, sblockloc, (void **)fsp, SBLOCKSIZE);
379 if (fs->fs_magic == FS_BAD_MAGIC)
381 if (((fs->fs_magic == FS_UFS1_MAGIC && (isaltsblk ||
382 sblockloc <= SBLOCK_UFS1)) ||
383 (fs->fs_magic == FS_UFS2_MAGIC && (isaltsblk ||
384 sblockloc == fs->fs_sblockloc))) &&
386 fs->fs_bsize >= MINBSIZE &&
387 fs->fs_bsize <= MAXBSIZE &&
388 fs->fs_bsize >= roundup(sizeof(struct fs), DEV_BSIZE) &&
389 fs->fs_sbsize <= SBLOCKSIZE) {
391 * If the filesystem has been run on a kernel without
392 * metadata check hashes, disable them.
394 if ((fs->fs_flags & FS_METACKHASH) == 0)
395 fs->fs_metackhash = 0;
396 if (fs->fs_ckhash != (ckhash = ffs_calc_sbhash(fs))) {
398 res = uprintf("Superblock check-hash failed: recorded "
399 "check-hash 0x%x != computed check-hash 0x%x%s\n",
400 fs->fs_ckhash, ckhash,
401 chkhash == 0 ? " (Ignored)" : "");
406 * Print check-hash failure if no controlling terminal
407 * in kernel or always if in user-mode (libufs).
410 printf("Superblock check-hash failed: recorded "
411 "check-hash 0x%x != computed check-hash "
412 "0x%x%s\n", fs->fs_ckhash, ckhash,
413 chkhash == 0 ? " (Ignored)" : "");
415 fs->fs_flags |= FS_NEEDSFSCK;
422 /* Have to set for old filesystems that predate this field */
423 fs->fs_sblockactualloc = sblockloc;
424 /* Not yet any summary information */
432 * Write a superblock to the devfd device from the memory pointed to by fs.
433 * Write out the superblock summary information if it is present.
435 * If the write is successful, zero is returned. Otherwise one of the
436 * following error values is returned:
437 * EIO: failed to write superblock.
438 * EIO: failed to write superblock summary information.
441 ffs_sbput(void *devfd, struct fs *fs, off_t loc,
442 int (*writefunc)(void *devfd, off_t loc, void *buf, int size))
444 int i, error, blks, size;
448 * If there is summary information, write it first, so if there
449 * is an error, the superblock will not be marked as clean.
451 if (fs->fs_csp != NULL) {
452 blks = howmany(fs->fs_cssize, fs->fs_fsize);
453 space = (uint8_t *)fs->fs_csp;
454 for (i = 0; i < blks; i += fs->fs_frag) {
456 if (i + fs->fs_frag > blks)
457 size = (blks - i) * fs->fs_fsize;
458 if ((error = (*writefunc)(devfd,
459 dbtob(fsbtodb(fs, fs->fs_csaddr + i)),
466 fs->fs_time = UFS_TIME;
467 fs->fs_ckhash = ffs_calc_sbhash(fs);
468 if ((error = (*writefunc)(devfd, loc, fs, fs->fs_sbsize)) != 0)
474 * Calculate the check-hash for a superblock.
477 ffs_calc_sbhash(struct fs *fs)
479 uint32_t ckhash, save_ckhash;
482 * A filesystem that was using a superblock ckhash may be moved
483 * to an older kernel that does not support ckhashes. The
484 * older kernel will clear the FS_METACKHASH flag indicating
485 * that it does not update hashes. When the disk is moved back
486 * to a kernel capable of ckhashes it disables them on mount:
488 * if ((fs->fs_flags & FS_METACKHASH) == 0)
489 * fs->fs_metackhash = 0;
491 * This leaves (fs->fs_metackhash & CK_SUPERBLOCK) == 0) with an
492 * old stale value in the fs->fs_ckhash field. Thus the need to
493 * just accept what is there.
495 if ((fs->fs_metackhash & CK_SUPERBLOCK) == 0)
496 return (fs->fs_ckhash);
498 save_ckhash = fs->fs_ckhash;
501 * If newly read from disk, the caller is responsible for
502 * verifying that fs->fs_sbsize <= SBLOCKSIZE.
504 ckhash = calculate_crc32c(~0L, (void *)fs, fs->fs_sbsize);
505 fs->fs_ckhash = save_ckhash;
510 * Update the frsum fields to reflect addition or deletion
514 ffs_fragacct(struct fs *fs, int fragmap, int32_t fraglist[], int cnt)
520 inblk = (int)(fragtbl[fs->fs_frag][fragmap]) << 1;
522 for (siz = 1; siz < fs->fs_frag; siz++) {
523 if ((inblk & (1 << (siz + (fs->fs_frag % NBBY)))) == 0)
526 subfield = inside[siz];
527 for (pos = siz; pos <= fs->fs_frag; pos++) {
528 if ((fragmap & field) == subfield) {
529 fraglist[siz] += cnt;
543 * check if a block is available
546 ffs_isblock(struct fs *fs, unsigned char *cp, ufs1_daddr_t h)
550 switch ((int)fs->fs_frag) {
552 return (cp[h] == 0xff);
554 mask = 0x0f << ((h & 0x1) << 2);
555 return ((cp[h >> 1] & mask) == mask);
557 mask = 0x03 << ((h & 0x3) << 1);
558 return ((cp[h >> 2] & mask) == mask);
560 mask = 0x01 << (h & 0x7);
561 return ((cp[h >> 3] & mask) == mask);
564 panic("ffs_isblock");
572 * check if a block is free
575 ffs_isfreeblock(struct fs *fs, u_char *cp, ufs1_daddr_t h)
578 switch ((int)fs->fs_frag) {
582 return ((cp[h >> 1] & (0x0f << ((h & 0x1) << 2))) == 0);
584 return ((cp[h >> 2] & (0x03 << ((h & 0x3) << 1))) == 0);
586 return ((cp[h >> 3] & (0x01 << (h & 0x7))) == 0);
589 panic("ffs_isfreeblock");
597 * take a block out of the map
600 ffs_clrblock(struct fs *fs, u_char *cp, ufs1_daddr_t h)
603 switch ((int)fs->fs_frag) {
608 cp[h >> 1] &= ~(0x0f << ((h & 0x1) << 2));
611 cp[h >> 2] &= ~(0x03 << ((h & 0x3) << 1));
614 cp[h >> 3] &= ~(0x01 << (h & 0x7));
618 panic("ffs_clrblock");
625 * put a block into the map
628 ffs_setblock(struct fs *fs, unsigned char *cp, ufs1_daddr_t h)
631 switch ((int)fs->fs_frag) {
637 cp[h >> 1] |= (0x0f << ((h & 0x1) << 2));
640 cp[h >> 2] |= (0x03 << ((h & 0x3) << 1));
643 cp[h >> 3] |= (0x01 << (h & 0x7));
647 panic("ffs_setblock");
654 * Update the cluster map because of an allocation or free.
656 * Cnt == 1 means free; cnt == -1 means allocating.
659 ffs_clusteracct(struct fs *fs, struct cg *cgp, ufs1_daddr_t blkno, int cnt)
663 u_char *freemapp, *mapp;
664 int i, start, end, forw, back, map;
667 if (fs->fs_contigsumsize <= 0)
669 freemapp = cg_clustersfree(cgp);
670 sump = cg_clustersum(cgp);
672 * Allocate or clear the actual block.
675 setbit(freemapp, blkno);
677 clrbit(freemapp, blkno);
679 * Find the size of the cluster going forward.
682 end = start + fs->fs_contigsumsize;
683 if (end >= cgp->cg_nclusterblks)
684 end = cgp->cg_nclusterblks;
685 mapp = &freemapp[start / NBBY];
687 bit = 1U << (start % NBBY);
688 for (i = start; i < end; i++) {
689 if ((map & bit) == 0)
691 if ((i & (NBBY - 1)) != (NBBY - 1)) {
700 * Find the size of the cluster going backward.
703 end = start - fs->fs_contigsumsize;
706 mapp = &freemapp[start / NBBY];
708 bit = 1U << (start % NBBY);
709 for (i = start; i > end; i--) {
710 if ((map & bit) == 0)
712 if ((i & (NBBY - 1)) != 0) {
716 bit = 1U << (NBBY - 1);
721 * Account for old cluster and the possibly new forward and
725 if (i > fs->fs_contigsumsize)
726 i = fs->fs_contigsumsize;
733 * Update cluster summary information.
735 lp = &sump[fs->fs_contigsumsize];
736 for (i = fs->fs_contigsumsize; i > 0; i--)
739 fs->fs_maxcluster[cgp->cg_cgx] = i;