xref: /netbsd-src/sbin/fsck_lfs/lfs.c (revision 404fbe5fb94ca1e054339640cabb2801ce52dd30)
1 /* $NetBSD: lfs.c,v 1.29 2008/05/16 09:21:59 hannken Exp $ */
2 /*-
3  * Copyright (c) 2003 The NetBSD Foundation, Inc.
4  * All rights reserved.
5  *
6  * This code is derived from software contributed to The NetBSD Foundation
7  * by Konrad E. Schroder <perseant@hhhh.org>.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
19  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
20  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
21  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
22  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28  * POSSIBILITY OF SUCH DAMAGE.
29  */
30 /*
31  * Copyright (c) 1989, 1991, 1993
32  *	The Regents of the University of California.  All rights reserved.
33  * (c) UNIX System Laboratories, Inc.
34  * All or some portions of this file are derived from material licensed
35  * to the University of California by American Telephone and Telegraph
36  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
37  * the permission of UNIX System Laboratories, Inc.
38  *
39  * Redistribution and use in source and binary forms, with or without
40  * modification, are permitted provided that the following conditions
41  * are met:
42  * 1. Redistributions of source code must retain the above copyright
43  *    notice, this list of conditions and the following disclaimer.
44  * 2. Redistributions in binary form must reproduce the above copyright
45  *    notice, this list of conditions and the following disclaimer in the
46  *    documentation and/or other materials provided with the distribution.
47  * 3. Neither the name of the University nor the names of its contributors
48  *    may be used to endorse or promote products derived from this software
49  *    without specific prior written permission.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
52  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
53  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
54  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
55  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
56  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
57  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
59  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
60  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
61  * SUCH DAMAGE.
62  *
63  *	@(#)ufs_bmap.c	8.8 (Berkeley) 8/11/95
64  */
65 
66 
67 #include <sys/types.h>
68 #include <sys/param.h>
69 #include <sys/time.h>
70 #include <sys/buf.h>
71 #include <sys/mount.h>
72 
73 #include <ufs/ufs/inode.h>
74 #include <ufs/ufs/ufsmount.h>
75 #define vnode uvnode
76 #include <ufs/lfs/lfs.h>
77 #undef vnode
78 
79 #include <assert.h>
80 #include <err.h>
81 #include <errno.h>
82 #include <stdarg.h>
83 #include <stdio.h>
84 #include <stdlib.h>
85 #include <string.h>
86 #include <unistd.h>
87 #include <util.h>
88 
89 #include "bufcache.h"
90 #include "vnode.h"
91 #include "lfs_user.h"
92 #include "segwrite.h"
93 
94 #define panic call_panic
95 
96 extern u_int32_t cksum(void *, size_t);
97 extern u_int32_t lfs_sb_cksum(struct dlfs *);
98 extern void pwarn(const char *, ...);
99 
100 extern struct uvnodelst vnodelist;
101 extern struct uvnodelst getvnodelist[VNODE_HASH_MAX];
102 extern int nvnodes;
103 
104 static int
105 lfs_fragextend(struct uvnode *, int, int, daddr_t, struct ubuf **);
106 
107 int fsdirty = 0;
108 void (*panic_func)(int, const char *, va_list) = my_vpanic;
109 
110 /*
111  * LFS buffer and uvnode operations
112  */
113 
114 int
115 lfs_vop_strategy(struct ubuf * bp)
116 {
117 	int count;
118 
119 	if (bp->b_flags & B_READ) {
120 		count = pread(bp->b_vp->v_fd, bp->b_data, bp->b_bcount,
121 		    dbtob(bp->b_blkno));
122 		if (count == bp->b_bcount)
123 			bp->b_flags |= B_DONE;
124 	} else {
125 		count = pwrite(bp->b_vp->v_fd, bp->b_data, bp->b_bcount,
126 		    dbtob(bp->b_blkno));
127 		if (count == 0) {
128 			perror("pwrite");
129 			return -1;
130 		}
131 		bp->b_flags &= ~B_DELWRI;
132 		reassignbuf(bp, bp->b_vp);
133 	}
134 	return 0;
135 }
136 
137 int
138 lfs_vop_bwrite(struct ubuf * bp)
139 {
140 	struct lfs *fs;
141 
142 	fs = bp->b_vp->v_fs;
143 	if (!(bp->b_flags & B_DELWRI)) {
144 		fs->lfs_avail -= btofsb(fs, bp->b_bcount);
145 	}
146 	bp->b_flags |= B_DELWRI | B_LOCKED;
147 	reassignbuf(bp, bp->b_vp);
148 	brelse(bp, 0);
149 	return 0;
150 }
151 
152 /*
153  * ufs_bmaparray does the bmap conversion, and if requested returns the
154  * array of logical blocks which must be traversed to get to a block.
155  * Each entry contains the offset into that block that gets you to the
156  * next block and the disk address of the block (if it is assigned).
157  */
158 int
159 ufs_bmaparray(struct lfs * fs, struct uvnode * vp, daddr_t bn, daddr_t * bnp, struct indir * ap, int *nump)
160 {
161 	struct inode *ip;
162 	struct ubuf *bp;
163 	struct indir a[NIADDR + 1], *xap;
164 	daddr_t daddr;
165 	daddr_t metalbn;
166 	int error, num;
167 
168 	ip = VTOI(vp);
169 
170 	if (bn >= 0 && bn < NDADDR) {
171 		if (nump != NULL)
172 			*nump = 0;
173 		*bnp = fsbtodb(fs, ip->i_ffs1_db[bn]);
174 		if (*bnp == 0)
175 			*bnp = -1;
176 		return (0);
177 	}
178 	xap = ap == NULL ? a : ap;
179 	if (!nump)
180 		nump = &num;
181 	if ((error = ufs_getlbns(fs, vp, bn, xap, nump)) != 0)
182 		return (error);
183 
184 	num = *nump;
185 
186 	/* Get disk address out of indirect block array */
187 	daddr = ip->i_ffs1_ib[xap->in_off];
188 
189 	for (bp = NULL, ++xap; --num; ++xap) {
190 		/* Exit the loop if there is no disk address assigned yet and
191 		 * the indirect block isn't in the cache, or if we were
192 		 * looking for an indirect block and we've found it. */
193 
194 		metalbn = xap->in_lbn;
195 		if ((daddr == 0 && !incore(vp, metalbn)) || metalbn == bn)
196 			break;
197 		/*
198 		 * If we get here, we've either got the block in the cache
199 		 * or we have a disk address for it, go fetch it.
200 		 */
201 		if (bp)
202 			brelse(bp, 0);
203 
204 		xap->in_exists = 1;
205 		bp = getblk(vp, metalbn, fs->lfs_bsize);
206 
207 		if (!(bp->b_flags & (B_DONE | B_DELWRI))) {
208 			bp->b_blkno = fsbtodb(fs, daddr);
209 			bp->b_flags |= B_READ;
210 			VOP_STRATEGY(bp);
211 		}
212 		daddr = ((ufs_daddr_t *) bp->b_data)[xap->in_off];
213 	}
214 	if (bp)
215 		brelse(bp, 0);
216 
217 	daddr = fsbtodb(fs, (ufs_daddr_t) daddr);
218 	*bnp = daddr == 0 ? -1 : daddr;
219 	return (0);
220 }
221 
222 /*
223  * Create an array of logical block number/offset pairs which represent the
224  * path of indirect blocks required to access a data block.  The first "pair"
225  * contains the logical block number of the appropriate single, double or
226  * triple indirect block and the offset into the inode indirect block array.
227  * Note, the logical block number of the inode single/double/triple indirect
228  * block appears twice in the array, once with the offset into the i_ffs1_ib and
229  * once with the offset into the page itself.
230  */
231 int
232 ufs_getlbns(struct lfs * fs, struct uvnode * vp, daddr_t bn, struct indir * ap, int *nump)
233 {
234 	daddr_t metalbn, realbn;
235 	int64_t blockcnt;
236 	int lbc;
237 	int i, numlevels, off;
238 	int lognindir, indir;
239 
240 	metalbn = 0;    /* XXXGCC -Wuninitialized [sh3] */
241 
242 	if (nump)
243 		*nump = 0;
244 	numlevels = 0;
245 	realbn = bn;
246 	if (bn < 0)
247 		bn = -bn;
248 
249 	lognindir = -1;
250 	for (indir = fs->lfs_nindir; indir; indir >>= 1)
251 		++lognindir;
252 
253 	/* Determine the number of levels of indirection.  After this loop is
254 	 * done, blockcnt indicates the number of data blocks possible at the
255 	 * given level of indirection, and NIADDR - i is the number of levels
256 	 * of indirection needed to locate the requested block. */
257 
258 	bn -= NDADDR;
259 	for (lbc = 0, i = NIADDR;; i--, bn -= blockcnt) {
260 		if (i == 0)
261 			return (EFBIG);
262 
263 		lbc += lognindir;
264 		blockcnt = (int64_t) 1 << lbc;
265 
266 		if (bn < blockcnt)
267 			break;
268 	}
269 
270 	/* Calculate the address of the first meta-block. */
271 	metalbn = -((realbn >= 0 ? realbn : -realbn) - bn + NIADDR - i);
272 
273 	/* At each iteration, off is the offset into the bap array which is an
274 	 * array of disk addresses at the current level of indirection. The
275 	 * logical block number and the offset in that block are stored into
276 	 * the argument array. */
277 	ap->in_lbn = metalbn;
278 	ap->in_off = off = NIADDR - i;
279 	ap->in_exists = 0;
280 	ap++;
281 	for (++numlevels; i <= NIADDR; i++) {
282 		/* If searching for a meta-data block, quit when found. */
283 		if (metalbn == realbn)
284 			break;
285 
286 		lbc -= lognindir;
287 		blockcnt = (int64_t) 1 << lbc;
288 		off = (bn >> lbc) & (fs->lfs_nindir - 1);
289 
290 		++numlevels;
291 		ap->in_lbn = metalbn;
292 		ap->in_off = off;
293 		ap->in_exists = 0;
294 		++ap;
295 
296 		metalbn -= -1 + (off << lbc);
297 	}
298 	if (nump)
299 		*nump = numlevels;
300 	return (0);
301 }
302 
303 int
304 lfs_vop_bmap(struct uvnode * vp, daddr_t lbn, daddr_t * daddrp)
305 {
306 	return ufs_bmaparray(vp->v_fs, vp, lbn, daddrp, NULL, NULL);
307 }
308 
309 /* Search a block for a specific dinode. */
310 struct ufs1_dinode *
311 lfs_ifind(struct lfs * fs, ino_t ino, struct ubuf * bp)
312 {
313 	struct ufs1_dinode *dip = (struct ufs1_dinode *) bp->b_data;
314 	struct ufs1_dinode *ldip, *fin;
315 
316 	fin = dip + INOPB(fs);
317 
318 	/*
319 	 * Read the inode block backwards, since later versions of the
320 	 * inode will supercede earlier ones.  Though it is unlikely, it is
321 	 * possible that the same inode will appear in the same inode block.
322 	 */
323 	for (ldip = fin - 1; ldip >= dip; --ldip)
324 		if (ldip->di_inumber == ino)
325 			return (ldip);
326 	return NULL;
327 }
328 
329 /*
330  * lfs_raw_vget makes us a new vnode from the inode at the given disk address.
331  * XXX it currently loses atime information.
332  */
333 struct uvnode *
334 lfs_raw_vget(struct lfs * fs, ino_t ino, int fd, ufs_daddr_t daddr)
335 {
336 	struct uvnode *vp;
337 	struct inode *ip;
338 	struct ufs1_dinode *dip;
339 	struct ubuf *bp;
340 	int i, hash;
341 
342 	vp = ecalloc(1, sizeof(*vp));
343 	vp->v_fd = fd;
344 	vp->v_fs = fs;
345 	vp->v_usecount = 0;
346 	vp->v_strategy_op = lfs_vop_strategy;
347 	vp->v_bwrite_op = lfs_vop_bwrite;
348 	vp->v_bmap_op = lfs_vop_bmap;
349 	LIST_INIT(&vp->v_cleanblkhd);
350 	LIST_INIT(&vp->v_dirtyblkhd);
351 
352 	ip = ecalloc(1, sizeof(*ip));
353 
354 	ip->i_din.ffs1_din = ecalloc(1, sizeof(*ip->i_din.ffs1_din));
355 
356 	/* Initialize the inode -- from lfs_vcreate. */
357 	ip->inode_ext.lfs = ecalloc(1, sizeof(*ip->inode_ext.lfs));
358 	vp->v_data = ip;
359 	/* ip->i_vnode = vp; */
360 	ip->i_number = ino;
361 	ip->i_lockf = 0;
362 	ip->i_diroff = 0;
363 	ip->i_lfs_effnblks = 0;
364 	ip->i_flag = 0;
365 
366 	/* Load inode block and find inode */
367 	if (daddr > 0) {
368 		bread(fs->lfs_devvp, fsbtodb(fs, daddr), fs->lfs_ibsize,
369 		    NULL, 0, &bp);
370 		bp->b_flags |= B_AGE;
371 		dip = lfs_ifind(fs, ino, bp);
372 		if (dip == NULL) {
373 			brelse(bp, 0);
374 			free(ip);
375 			free(vp);
376 			return NULL;
377 		}
378 		memcpy(ip->i_din.ffs1_din, dip, sizeof(*dip));
379 		brelse(bp, 0);
380 	}
381 	ip->i_number = ino;
382 	/* ip->i_devvp = fs->lfs_devvp; */
383 	ip->i_lfs = fs;
384 
385 	ip->i_ffs_effnlink = ip->i_ffs1_nlink;
386 	ip->i_lfs_effnblks = ip->i_ffs1_blocks;
387 	ip->i_lfs_osize = ip->i_ffs1_size;
388 #if 0
389 	if (fs->lfs_version > 1) {
390 		ip->i_ffs1_atime = ts.tv_sec;
391 		ip->i_ffs1_atimensec = ts.tv_nsec;
392 	}
393 #endif
394 
395 	memset(ip->i_lfs_fragsize, 0, NDADDR * sizeof(*ip->i_lfs_fragsize));
396 	for (i = 0; i < NDADDR; i++)
397 		if (ip->i_ffs1_db[i] != 0)
398 			ip->i_lfs_fragsize[i] = blksize(fs, ip, i);
399 
400 	++nvnodes;
401 	hash = ((int)(intptr_t)fs + ino) & (VNODE_HASH_MAX - 1);
402 	LIST_INSERT_HEAD(&getvnodelist[hash], vp, v_getvnodes);
403 	LIST_INSERT_HEAD(&vnodelist, vp, v_mntvnodes);
404 
405 	return vp;
406 }
407 
408 static struct uvnode *
409 lfs_vget(void *vfs, ino_t ino)
410 {
411 	struct lfs *fs = (struct lfs *)vfs;
412 	ufs_daddr_t daddr;
413 	struct ubuf *bp;
414 	IFILE *ifp;
415 
416 	LFS_IENTRY(ifp, fs, ino, bp);
417 	daddr = ifp->if_daddr;
418 	brelse(bp, 0);
419 	if (daddr <= 0 || dtosn(fs, daddr) >= fs->lfs_nseg)
420 		return NULL;
421 	return lfs_raw_vget(fs, ino, fs->lfs_ivnode->v_fd, daddr);
422 }
423 
424 /* Check superblock magic number and checksum */
425 static int
426 check_sb(struct lfs *fs)
427 {
428 	u_int32_t checksum;
429 
430 	if (fs->lfs_magic != LFS_MAGIC) {
431 		printf("Superblock magic number (0x%lx) does not match "
432 		       "expected 0x%lx\n", (unsigned long) fs->lfs_magic,
433 		       (unsigned long) LFS_MAGIC);
434 		return 1;
435 	}
436 	/* checksum */
437 	checksum = lfs_sb_cksum(&(fs->lfs_dlfs));
438 	if (fs->lfs_cksum != checksum) {
439 		printf("Superblock checksum (%lx) does not match computed checksum (%lx)\n",
440 		    (unsigned long) fs->lfs_cksum, (unsigned long) checksum);
441 		return 1;
442 	}
443 	return 0;
444 }
445 
446 /* Initialize LFS library; load superblocks and choose which to use. */
447 struct lfs *
448 lfs_init(int devfd, daddr_t sblkno, daddr_t idaddr, int dummy_read, int debug)
449 {
450 	struct uvnode *devvp;
451 	struct ubuf *bp;
452 	int tryalt;
453 	struct lfs *fs, *altfs;
454 	int error;
455 
456 	vfs_init();
457 
458 	devvp = ecalloc(1, sizeof(*devvp));
459 	devvp->v_fs = NULL;
460 	devvp->v_fd = devfd;
461 	devvp->v_strategy_op = raw_vop_strategy;
462 	devvp->v_bwrite_op = raw_vop_bwrite;
463 	devvp->v_bmap_op = raw_vop_bmap;
464 	LIST_INIT(&devvp->v_cleanblkhd);
465 	LIST_INIT(&devvp->v_dirtyblkhd);
466 
467 	tryalt = 0;
468 	if (dummy_read) {
469 		if (sblkno == 0)
470 			sblkno = btodb(LFS_LABELPAD);
471 		fs = ecalloc(1, sizeof(*fs));
472 		fs->lfs_devvp = devvp;
473 	} else {
474 		if (sblkno == 0) {
475 			sblkno = btodb(LFS_LABELPAD);
476 			tryalt = 1;
477 		} else if (debug) {
478 			printf("No -b flag given, not attempting to verify checkpoint\n");
479 		}
480 		error = bread(devvp, sblkno, LFS_SBPAD, NOCRED, 0, &bp);
481 		fs = ecalloc(1, sizeof(*fs));
482 		fs->lfs_dlfs = *((struct dlfs *) bp->b_data);
483 		fs->lfs_devvp = devvp;
484 		bp->b_flags |= B_INVAL;
485 		brelse(bp, 0);
486 
487 		if (tryalt) {
488 			error = bread(devvp, fsbtodb(fs, fs->lfs_sboffs[1]),
489 		    	LFS_SBPAD, NOCRED, 0, &bp);
490 			altfs = ecalloc(1, sizeof(*altfs));
491 			altfs->lfs_dlfs = *((struct dlfs *) bp->b_data);
492 			altfs->lfs_devvp = devvp;
493 			bp->b_flags |= B_INVAL;
494 			brelse(bp, 0);
495 
496 			if (check_sb(fs) || fs->lfs_idaddr <= 0) {
497 				if (debug)
498 					printf("Primary superblock is no good, using first alternate\n");
499 				free(fs);
500 				fs = altfs;
501 			} else {
502 				/* If both superblocks check out, try verification */
503 				if (check_sb(altfs)) {
504 					if (debug)
505 						printf("First alternate superblock is no good, using primary\n");
506 					free(altfs);
507 				} else {
508 					if (lfs_verify(fs, altfs, devvp, debug) == fs) {
509 						free(altfs);
510 					} else {
511 						free(fs);
512 						fs = altfs;
513 					}
514 				}
515 			}
516 		}
517 		if (check_sb(fs)) {
518 			free(fs);
519 			return NULL;
520 		}
521 	}
522 
523 	/* Compatibility */
524 	if (fs->lfs_version < 2) {
525 		fs->lfs_sumsize = LFS_V1_SUMMARY_SIZE;
526 		fs->lfs_ibsize = fs->lfs_bsize;
527 		fs->lfs_start = fs->lfs_sboffs[0];
528 		fs->lfs_tstamp = fs->lfs_otstamp;
529 		fs->lfs_fsbtodb = 0;
530 	}
531 
532 	if (!dummy_read) {
533 		fs->lfs_suflags = emalloc(2 * sizeof(u_int32_t *));
534 		fs->lfs_suflags[0] = emalloc(fs->lfs_nseg * sizeof(u_int32_t));
535 		fs->lfs_suflags[1] = emalloc(fs->lfs_nseg * sizeof(u_int32_t));
536 	}
537 
538 	if (idaddr == 0)
539 		idaddr = fs->lfs_idaddr;
540 	else
541 		fs->lfs_idaddr = idaddr;
542 	/* NB: If dummy_read!=0, idaddr==0 here so we get a fake inode. */
543 	fs->lfs_ivnode = lfs_raw_vget(fs,
544 		(dummy_read ? LFS_IFILE_INUM : fs->lfs_ifile), devvp->v_fd,
545 		idaddr);
546 	if (fs->lfs_ivnode == NULL)
547 		return NULL;
548 
549 	register_vget((void *)fs, lfs_vget);
550 
551 	return fs;
552 }
553 
554 /*
555  * Check partial segment validity between fs->lfs_offset and the given goal.
556  *
557  * If goal == 0, just keep on going until the segments stop making sense,
558  * and return the address of the last valid partial segment.
559  *
560  * If goal != 0, return the address of the first partial segment that failed,
561  * or "goal" if we reached it without failure (the partial segment *at* goal
562  * need not be valid).
563  */
564 ufs_daddr_t
565 try_verify(struct lfs *osb, struct uvnode *devvp, ufs_daddr_t goal, int debug)
566 {
567 	ufs_daddr_t daddr, odaddr;
568 	SEGSUM *sp;
569 	int i, bc, hitclean;
570 	struct ubuf *bp;
571 	ufs_daddr_t nodirop_daddr;
572 	u_int64_t serial;
573 
574 	bc = 0;
575 	hitclean = 0;
576 	odaddr = -1;
577 	daddr = osb->lfs_offset;
578 	nodirop_daddr = daddr;
579 	serial = osb->lfs_serial;
580 	while (daddr != goal) {
581 		/*
582 		 * Don't mistakenly read a superblock, if there is one here.
583 		 */
584 		if (sntod(osb, dtosn(osb, daddr)) == daddr) {
585 			if (daddr == osb->lfs_start)
586 				daddr += btofsb(osb, LFS_LABELPAD);
587 			for (i = 0; i < LFS_MAXNUMSB; i++) {
588 				if (osb->lfs_sboffs[i] < daddr)
589 					break;
590 				if (osb->lfs_sboffs[i] == daddr)
591 					daddr += btofsb(osb, LFS_SBPAD);
592 			}
593 		}
594 
595 		/* Read in summary block */
596 		bread(devvp, fsbtodb(osb, daddr), osb->lfs_sumsize,
597 		    NULL, 0, &bp);
598 		sp = (SEGSUM *)bp->b_data;
599 
600 		/*
601 		 * Check for a valid segment summary belonging to our fs.
602 		 */
603 		if (sp->ss_magic != SS_MAGIC ||
604 		    sp->ss_ident != osb->lfs_ident ||
605 		    sp->ss_serial < serial ||	/* XXX strengthen this */
606 		    sp->ss_sumsum != cksum(&sp->ss_datasum, osb->lfs_sumsize -
607 			sizeof(sp->ss_sumsum))) {
608 			brelse(bp, 0);
609 			if (debug) {
610 				if (sp->ss_magic != SS_MAGIC)
611 					pwarn("pseg at 0x%x: "
612 					      "wrong magic number\n",
613 					      (int)daddr);
614 				else if (sp->ss_ident != osb->lfs_ident)
615 					pwarn("pseg at 0x%x: "
616 					      "expected ident %llx, got %llx\n",
617 					      (int)daddr,
618 					      (long long)sp->ss_ident,
619 					      (long long)osb->lfs_ident);
620 				else if (sp->ss_serial >= serial)
621 					pwarn("pseg at 0x%x: "
622 					      "serial %d < %d\n", (int)daddr,
623 					      (int)sp->ss_serial, (int)serial);
624 				else
625 					pwarn("pseg at 0x%x: "
626 					      "summary checksum wrong\n",
627 					      (int)daddr);
628 			}
629 			break;
630 		}
631 		if (debug && sp->ss_serial != serial)
632 			pwarn("warning, serial=%d ss_serial=%d\n",
633 				(int)serial, (int)sp->ss_serial);
634 		++serial;
635 		bc = check_summary(osb, sp, daddr, debug, devvp, NULL);
636 		if (bc == 0) {
637 			brelse(bp, 0);
638 			break;
639 		}
640 		if (debug)
641 			pwarn("summary good: 0x%x/%d\n", (int)daddr,
642 			      (int)sp->ss_serial);
643 		assert (bc > 0);
644 		odaddr = daddr;
645 		daddr += btofsb(osb, osb->lfs_sumsize + bc);
646 		if (dtosn(osb, odaddr) != dtosn(osb, daddr) ||
647 		    dtosn(osb, daddr) != dtosn(osb, daddr +
648 			btofsb(osb, osb->lfs_sumsize + osb->lfs_bsize) - 1)) {
649 			daddr = sp->ss_next;
650 		}
651 
652 		/*
653 		 * Check for the beginning and ending of a sequence of
654 		 * dirops.  Writes from the cleaner never involve new
655 		 * information, and are always checkpoints; so don't try
656 		 * to roll forward through them.  Likewise, psegs written
657 		 * by a previous roll-forward attempt are not interesting.
658 		 */
659 		if (sp->ss_flags & (SS_CLEAN | SS_RFW))
660 			hitclean = 1;
661 		if (hitclean == 0 && (sp->ss_flags & SS_CONT) == 0)
662 			nodirop_daddr = daddr;
663 
664 		brelse(bp, 0);
665 	}
666 
667 	if (goal == 0)
668 		return nodirop_daddr;
669 	else
670 		return daddr;
671 }
672 
673 /* Use try_verify to check whether the newer superblock is valid. */
674 struct lfs *
675 lfs_verify(struct lfs *sb0, struct lfs *sb1, struct uvnode *devvp, int debug)
676 {
677 	ufs_daddr_t daddr;
678 	struct lfs *osb, *nsb;
679 
680 	/*
681 	 * Verify the checkpoint of the newer superblock,
682 	 * if the timestamp/serial number of the two superblocks is
683 	 * different.
684 	 */
685 
686 	osb = NULL;
687 	if (debug)
688 		pwarn("sb0 %lld, sb1 %lld",
689 		      (long long) sb0->lfs_serial,
690 		      (long long) sb1->lfs_serial);
691 
692 	if ((sb0->lfs_version == 1 &&
693 		sb0->lfs_otstamp != sb1->lfs_otstamp) ||
694 	    (sb0->lfs_version > 1 &&
695 		sb0->lfs_serial != sb1->lfs_serial)) {
696 		if (sb0->lfs_version == 1) {
697 			if (sb0->lfs_otstamp > sb1->lfs_otstamp) {
698 				osb = sb1;
699 				nsb = sb0;
700 			} else {
701 				osb = sb0;
702 				nsb = sb1;
703 			}
704 		} else {
705 			if (sb0->lfs_serial > sb1->lfs_serial) {
706 				osb = sb1;
707 				nsb = sb0;
708 			} else {
709 				osb = sb0;
710 				nsb = sb1;
711 			}
712 		}
713 		if (debug) {
714 			printf("Attempting to verify newer checkpoint...");
715 			fflush(stdout);
716 		}
717 		daddr = try_verify(osb, devvp, nsb->lfs_offset, debug);
718 
719 		if (debug)
720 			printf("done.\n");
721 		if (daddr == nsb->lfs_offset) {
722 			pwarn("** Newer checkpoint verified, recovered %lld seconds of data\n",
723 			    (long long) nsb->lfs_tstamp - (long long) osb->lfs_tstamp);
724 			sbdirty();
725 		} else {
726 			pwarn("** Newer checkpoint invalid, lost %lld seconds of data\n", (long long) nsb->lfs_tstamp - (long long) osb->lfs_tstamp);
727 		}
728 		return (daddr == nsb->lfs_offset ? nsb : osb);
729 	}
730 	/* Nothing to check */
731 	return osb;
732 }
733 
734 /* Verify a partial-segment summary; return the number of bytes on disk. */
735 int
736 check_summary(struct lfs *fs, SEGSUM *sp, ufs_daddr_t pseg_addr, int debug,
737 	      struct uvnode *devvp, void (func(ufs_daddr_t, FINFO *)))
738 {
739 	FINFO *fp;
740 	int bc;			/* Bytes in partial segment */
741 	int nblocks;
742 	ufs_daddr_t seg_addr, daddr;
743 	ufs_daddr_t *dp, *idp;
744 	struct ubuf *bp;
745 	int i, j, k, datac, len;
746 	long sn;
747 	u_int32_t *datap;
748 	u_int32_t ccksum;
749 
750 	sn = dtosn(fs, pseg_addr);
751 	seg_addr = sntod(fs, sn);
752 
753 	/* We've already checked the sumsum, just do the data bounds and sum */
754 
755 	/* Count the blocks. */
756 	nblocks = howmany(sp->ss_ninos, INOPB(fs));
757 	bc = nblocks << (fs->lfs_version > 1 ? fs->lfs_ffshift : fs->lfs_bshift);
758 	assert(bc >= 0);
759 
760 	fp = (FINFO *) (sp + 1);
761 	for (i = 0; i < sp->ss_nfinfo; i++) {
762 		nblocks += fp->fi_nblocks;
763 		bc += fp->fi_lastlength + ((fp->fi_nblocks - 1)
764 					   << fs->lfs_bshift);
765 		assert(bc >= 0);
766 		fp = (FINFO *) (fp->fi_blocks + fp->fi_nblocks);
767 		if (((char *)fp) - (char *)sp > fs->lfs_sumsize)
768 			return 0;
769 	}
770 	datap = emalloc(nblocks * sizeof(*datap));
771 	datac = 0;
772 
773 	dp = (ufs_daddr_t *) sp;
774 	dp += fs->lfs_sumsize / sizeof(ufs_daddr_t);
775 	dp--;
776 
777 	idp = dp;
778 	daddr = pseg_addr + btofsb(fs, fs->lfs_sumsize);
779 	fp = (FINFO *) (sp + 1);
780 	for (i = 0, j = 0;
781 	     i < sp->ss_nfinfo || j < howmany(sp->ss_ninos, INOPB(fs)); i++) {
782 		if (i >= sp->ss_nfinfo && *idp != daddr) {
783 			pwarn("Not enough inode blocks in pseg at 0x%" PRIx32
784 			      ": found %d, wanted %d\n",
785 			      pseg_addr, j, howmany(sp->ss_ninos, INOPB(fs)));
786 			if (debug)
787 				pwarn("*idp=%x, daddr=%" PRIx32 "\n", *idp,
788 				      daddr);
789 			break;
790 		}
791 		while (j < howmany(sp->ss_ninos, INOPB(fs)) && *idp == daddr) {
792 			bread(devvp, fsbtodb(fs, daddr), fs->lfs_ibsize,
793 			    NOCRED, 0, &bp);
794 			datap[datac++] = ((u_int32_t *) (bp->b_data))[0];
795 			brelse(bp, 0);
796 
797 			++j;
798 			daddr += btofsb(fs, fs->lfs_ibsize);
799 			--idp;
800 		}
801 		if (i < sp->ss_nfinfo) {
802 			if (func)
803 				func(daddr, fp);
804 			for (k = 0; k < fp->fi_nblocks; k++) {
805 				len = (k == fp->fi_nblocks - 1 ?
806 				       fp->fi_lastlength
807 				       : fs->lfs_bsize);
808 				bread(devvp, fsbtodb(fs, daddr), len,
809 				    NOCRED, 0, &bp);
810 				datap[datac++] = ((u_int32_t *) (bp->b_data))[0];
811 				brelse(bp, 0);
812 				daddr += btofsb(fs, len);
813 			}
814 			fp = (FINFO *) (fp->fi_blocks + fp->fi_nblocks);
815 		}
816 	}
817 
818 	if (datac != nblocks) {
819 		pwarn("Partial segment at 0x%llx expected %d blocks counted %d\n",
820 		    (long long) pseg_addr, nblocks, datac);
821 	}
822 	ccksum = cksum(datap, nblocks * sizeof(u_int32_t));
823 	/* Check the data checksum */
824 	if (ccksum != sp->ss_datasum) {
825 		pwarn("Partial segment at 0x%" PRIx32 " data checksum"
826 		      " mismatch: given 0x%x, computed 0x%x\n",
827 		      pseg_addr, sp->ss_datasum, ccksum);
828 		free(datap);
829 		return 0;
830 	}
831 	free(datap);
832 	assert(bc >= 0);
833 	return bc;
834 }
835 
836 /* print message and exit */
837 void
838 my_vpanic(int fatal, const char *fmt, va_list ap)
839 {
840         (void) vprintf(fmt, ap);
841 	exit(8);
842 }
843 
844 void
845 call_panic(const char *fmt, ...)
846 {
847 	va_list ap;
848 
849 	va_start(ap, fmt);
850         panic_func(1, fmt, ap);
851 	va_end(ap);
852 }
853 
854 /* Allocate a new inode. */
855 struct uvnode *
856 lfs_valloc(struct lfs *fs, ino_t ino)
857 {
858 	struct ubuf *bp, *cbp;
859 	struct ifile *ifp;
860 	ino_t new_ino;
861 	int error;
862 	int new_gen;
863 	CLEANERINFO *cip;
864 
865 	/* Get the head of the freelist. */
866 	LFS_GET_HEADFREE(fs, cip, cbp, &new_ino);
867 
868 	/*
869 	 * Remove the inode from the free list and write the new start
870 	 * of the free list into the superblock.
871 	 */
872 	LFS_IENTRY(ifp, fs, new_ino, bp);
873 	if (ifp->if_daddr != LFS_UNUSED_DADDR)
874 		panic("lfs_valloc: inuse inode %d on the free list", new_ino);
875 	LFS_PUT_HEADFREE(fs, cip, cbp, ifp->if_nextfree);
876 
877 	new_gen = ifp->if_version; /* version was updated by vfree */
878 	brelse(bp, 0);
879 
880 	/* Extend IFILE so that the next lfs_valloc will succeed. */
881 	if (fs->lfs_freehd == LFS_UNUSED_INUM) {
882 		if ((error = extend_ifile(fs)) != 0) {
883 			LFS_PUT_HEADFREE(fs, cip, cbp, new_ino);
884 			return NULL;
885 		}
886 	}
887 
888 	/* Set superblock modified bit and increment file count. */
889         sbdirty();
890 	++fs->lfs_nfiles;
891 
892         return lfs_raw_vget(fs, ino, fs->lfs_devvp->v_fd, 0x0);
893 }
894 
895 #ifdef IN_FSCK_LFS
896 void reset_maxino(ino_t);
897 #endif
898 
899 /*
900  * Add a new block to the Ifile, to accommodate future file creations.
901  */
902 int
903 extend_ifile(struct lfs *fs)
904 {
905 	struct uvnode *vp;
906 	struct inode *ip;
907 	IFILE *ifp;
908 	IFILE_V1 *ifp_v1;
909 	struct ubuf *bp, *cbp;
910 	daddr_t i, blkno, max;
911 	ino_t oldlast;
912 	CLEANERINFO *cip;
913 
914 	vp = fs->lfs_ivnode;
915 	ip = VTOI(vp);
916 	blkno = lblkno(fs, ip->i_ffs1_size);
917 
918 	lfs_balloc(vp, ip->i_ffs1_size, fs->lfs_bsize, &bp);
919 	ip->i_ffs1_size += fs->lfs_bsize;
920 	ip->i_flag |= IN_MODIFIED;
921 
922 	i = (blkno - fs->lfs_segtabsz - fs->lfs_cleansz) *
923 		fs->lfs_ifpb;
924 	LFS_GET_HEADFREE(fs, cip, cbp, &oldlast);
925 	LFS_PUT_HEADFREE(fs, cip, cbp, i);
926 	max = i + fs->lfs_ifpb;
927 	fs->lfs_bfree -= btofsb(fs, fs->lfs_bsize);
928 
929 	if (fs->lfs_version == 1) {
930 		for (ifp_v1 = (IFILE_V1 *)bp->b_data; i < max; ++ifp_v1) {
931 			ifp_v1->if_version = 1;
932 			ifp_v1->if_daddr = LFS_UNUSED_DADDR;
933 			ifp_v1->if_nextfree = ++i;
934 		}
935 		ifp_v1--;
936 		ifp_v1->if_nextfree = oldlast;
937 	} else {
938 		for (ifp = (IFILE *)bp->b_data; i < max; ++ifp) {
939 			ifp->if_version = 1;
940 			ifp->if_daddr = LFS_UNUSED_DADDR;
941 			ifp->if_nextfree = ++i;
942 		}
943 		ifp--;
944 		ifp->if_nextfree = oldlast;
945 	}
946 	LFS_PUT_TAILFREE(fs, cip, cbp, max - 1);
947 
948 	LFS_BWRITE_LOG(bp);
949 
950 #ifdef IN_FSCK_LFS
951 	reset_maxino(((ip->i_ffs1_size >> fs->lfs_bshift) - fs->lfs_segtabsz -
952 		     fs->lfs_cleansz) * fs->lfs_ifpb);
953 #endif
954 	return 0;
955 }
956 
957 /*
958  * Allocate a block, and to inode and filesystem block accounting for it
959  * and for any indirect blocks the may need to be created in order for
960  * this block to be created.
961  *
962  * Blocks which have never been accounted for (i.e., which "do not exist")
963  * have disk address 0, which is translated by ufs_bmap to the special value
964  * UNASSIGNED == -1, as in the historical UFS.
965  *
966  * Blocks which have been accounted for but which have not yet been written
967  * to disk are given the new special disk address UNWRITTEN == -2, so that
968  * they can be differentiated from completely new blocks.
969  */
970 int
971 lfs_balloc(struct uvnode *vp, off_t startoffset, int iosize, struct ubuf **bpp)
972 {
973 	int offset;
974 	daddr_t daddr, idaddr;
975 	struct ubuf *ibp, *bp;
976 	struct inode *ip;
977 	struct lfs *fs;
978 	struct indir indirs[NIADDR+2], *idp;
979 	daddr_t	lbn, lastblock;
980 	int bb, bcount;
981 	int error, frags, i, nsize, osize, num;
982 
983 	ip = VTOI(vp);
984 	fs = ip->i_lfs;
985 	offset = blkoff(fs, startoffset);
986 	lbn = lblkno(fs, startoffset);
987 
988 	/*
989 	 * Three cases: it's a block beyond the end of file, it's a block in
990 	 * the file that may or may not have been assigned a disk address or
991 	 * we're writing an entire block.
992 	 *
993 	 * Note, if the daddr is UNWRITTEN, the block already exists in
994 	 * the cache (it was read or written earlier).	If so, make sure
995 	 * we don't count it as a new block or zero out its contents. If
996 	 * it did not, make sure we allocate any necessary indirect
997 	 * blocks.
998 	 *
999 	 * If we are writing a block beyond the end of the file, we need to
1000 	 * check if the old last block was a fragment.	If it was, we need
1001 	 * to rewrite it.
1002 	 */
1003 
1004 	if (bpp)
1005 		*bpp = NULL;
1006 
1007 	/* Check for block beyond end of file and fragment extension needed. */
1008 	lastblock = lblkno(fs, ip->i_ffs1_size);
1009 	if (lastblock < NDADDR && lastblock < lbn) {
1010 		osize = blksize(fs, ip, lastblock);
1011 		if (osize < fs->lfs_bsize && osize > 0) {
1012 			if ((error = lfs_fragextend(vp, osize, fs->lfs_bsize,
1013 						    lastblock,
1014 						    (bpp ? &bp : NULL))))
1015 				return (error);
1016 			ip->i_ffs1_size = ip->i_ffs1_size =
1017 			    (lastblock + 1) * fs->lfs_bsize;
1018 			ip->i_flag |= IN_CHANGE | IN_UPDATE;
1019 			if (bpp)
1020 				(void) VOP_BWRITE(bp);
1021 		}
1022 	}
1023 
1024 	/*
1025 	 * If the block we are writing is a direct block, it's the last
1026 	 * block in the file, and offset + iosize is less than a full
1027 	 * block, we can write one or more fragments.  There are two cases:
1028 	 * the block is brand new and we should allocate it the correct
1029 	 * size or it already exists and contains some fragments and
1030 	 * may need to extend it.
1031 	 */
1032 	if (lbn < NDADDR && lblkno(fs, ip->i_ffs1_size) <= lbn) {
1033 		osize = blksize(fs, ip, lbn);
1034 		nsize = fragroundup(fs, offset + iosize);
1035 		if (lblktosize(fs, lbn) >= ip->i_ffs1_size) {
1036 			/* Brand new block or fragment */
1037 			frags = numfrags(fs, nsize);
1038 			bb = fragstofsb(fs, frags);
1039 			if (bpp) {
1040 				*bpp = bp = getblk(vp, lbn, nsize);
1041 				bp->b_blkno = UNWRITTEN;
1042 			}
1043 			ip->i_lfs_effnblks += bb;
1044 			fs->lfs_bfree -= bb;
1045 			ip->i_ffs1_db[lbn] = UNWRITTEN;
1046 		} else {
1047 			if (nsize <= osize) {
1048 				/* No need to extend */
1049 				if (bpp && (error = bread(vp, lbn, osize,
1050 				    NOCRED, 0, &bp)))
1051 					return error;
1052 			} else {
1053 				/* Extend existing block */
1054 				if ((error =
1055 				     lfs_fragextend(vp, osize, nsize, lbn,
1056 						    (bpp ? &bp : NULL))))
1057 					return error;
1058 			}
1059 			if (bpp)
1060 				*bpp = bp;
1061 		}
1062 		return 0;
1063 	}
1064 
1065 	error = ufs_bmaparray(fs, vp, lbn, &daddr, &indirs[0], &num);
1066 	if (error)
1067 		return (error);
1068 
1069 	daddr = (daddr_t)((int32_t)daddr); /* XXX ondisk32 */
1070 
1071 	/*
1072 	 * Do byte accounting all at once, so we can gracefully fail *before*
1073 	 * we start assigning blocks.
1074 	 */
1075         bb = fsbtodb(fs, 1); /* bb = VFSTOUFS(vp->v_mount)->um_seqinc; */
1076 	bcount = 0;
1077 	if (daddr == UNASSIGNED) {
1078 		bcount = bb;
1079 	}
1080 	for (i = 1; i < num; ++i) {
1081 		if (!indirs[i].in_exists) {
1082 			bcount += bb;
1083 		}
1084 	}
1085 	fs->lfs_bfree -= bcount;
1086 	ip->i_lfs_effnblks += bcount;
1087 
1088 	if (daddr == UNASSIGNED) {
1089 		if (num > 0 && ip->i_ffs1_ib[indirs[0].in_off] == 0) {
1090 			ip->i_ffs1_ib[indirs[0].in_off] = UNWRITTEN;
1091 		}
1092 
1093 		/*
1094 		 * Create new indirect blocks if necessary
1095 		 */
1096 		if (num > 1) {
1097 			idaddr = ip->i_ffs1_ib[indirs[0].in_off];
1098 			for (i = 1; i < num; ++i) {
1099 				ibp = getblk(vp, indirs[i].in_lbn,
1100 				    fs->lfs_bsize);
1101 				if (!indirs[i].in_exists) {
1102 					memset(ibp->b_data, 0, ibp->b_bufsize);
1103 					ibp->b_blkno = UNWRITTEN;
1104 				} else if (!(ibp->b_flags & (B_DELWRI | B_DONE))) {
1105 					ibp->b_blkno = fsbtodb(fs, idaddr);
1106 					ibp->b_flags |= B_READ;
1107 					VOP_STRATEGY(ibp);
1108 				}
1109 				/*
1110 				 * This block exists, but the next one may not.
1111 				 * If that is the case mark it UNWRITTEN to
1112                                  * keep the accounting straight.
1113 				 */
1114 				/* XXX ondisk32 */
1115 				if (((int32_t *)ibp->b_data)[indirs[i].in_off] == 0)
1116 					((int32_t *)ibp->b_data)[indirs[i].in_off] =
1117 						UNWRITTEN;
1118 				/* XXX ondisk32 */
1119 				idaddr = ((int32_t *)ibp->b_data)[indirs[i].in_off];
1120 				if ((error = VOP_BWRITE(ibp)))
1121 					return error;
1122 			}
1123 		}
1124 	}
1125 
1126 
1127 	/*
1128 	 * Get the existing block from the cache, if requested.
1129 	 */
1130 	frags = fsbtofrags(fs, bb);
1131 	if (bpp)
1132 		*bpp = bp = getblk(vp, lbn, blksize(fs, ip, lbn));
1133 
1134 	/*
1135 	 * The block we are writing may be a brand new block
1136 	 * in which case we need to do accounting.
1137 	 *
1138 	 * We can tell a truly new block because ufs_bmaparray will say
1139 	 * it is UNASSIGNED.  Once we allocate it we will assign it the
1140 	 * disk address UNWRITTEN.
1141 	 */
1142 	if (daddr == UNASSIGNED) {
1143 		if (bpp) {
1144 			/* Note the new address */
1145 			bp->b_blkno = UNWRITTEN;
1146 		}
1147 
1148 		switch (num) {
1149 		    case 0:
1150 			ip->i_ffs1_db[lbn] = UNWRITTEN;
1151 			break;
1152 		    case 1:
1153 			ip->i_ffs1_ib[indirs[0].in_off] = UNWRITTEN;
1154 			break;
1155 		    default:
1156 			idp = &indirs[num - 1];
1157 			if (bread(vp, idp->in_lbn, fs->lfs_bsize, NOCRED,
1158 				  0, &ibp))
1159 				panic("lfs_balloc: bread bno %lld",
1160 				    (long long)idp->in_lbn);
1161 			/* XXX ondisk32 */
1162 			((int32_t *)ibp->b_data)[idp->in_off] = UNWRITTEN;
1163 			VOP_BWRITE(ibp);
1164 		}
1165 	} else if (bpp && !(bp->b_flags & (B_DONE|B_DELWRI))) {
1166 		/*
1167 		 * Not a brand new block, also not in the cache;
1168 		 * read it in from disk.
1169 		 */
1170 		if (iosize == fs->lfs_bsize)
1171 			/* Optimization: I/O is unnecessary. */
1172 			bp->b_blkno = daddr;
1173 		else {
1174 			/*
1175 			 * We need to read the block to preserve the
1176 			 * existing bytes.
1177 			 */
1178 			bp->b_blkno = daddr;
1179 			bp->b_flags |= B_READ;
1180 			VOP_STRATEGY(bp);
1181 			return 0;
1182 		}
1183 	}
1184 
1185 	return (0);
1186 }
1187 
1188 int
1189 lfs_fragextend(struct uvnode *vp, int osize, int nsize, daddr_t lbn,
1190                struct ubuf **bpp)
1191 {
1192 	struct inode *ip;
1193 	struct lfs *fs;
1194 	long bb;
1195 	int error;
1196 	size_t obufsize;
1197 
1198 	ip = VTOI(vp);
1199 	fs = ip->i_lfs;
1200 	bb = (long)fragstofsb(fs, numfrags(fs, nsize - osize));
1201 	error = 0;
1202 
1203 	/*
1204 	 * If we are not asked to actually return the block, all we need
1205 	 * to do is allocate space for it.  UBC will handle dirtying the
1206 	 * appropriate things and making sure it all goes to disk.
1207 	 * Don't bother to read in that case.
1208 	 */
1209 	if (bpp && (error = bread(vp, lbn, osize, NOCRED, 0, bpp))) {
1210 		brelse(*bpp, 0);
1211 		goto out;
1212 	}
1213 
1214 	fs->lfs_bfree -= bb;
1215 	ip->i_lfs_effnblks += bb;
1216 	ip->i_flag |= IN_CHANGE | IN_UPDATE;
1217 
1218 	if (bpp) {
1219 		obufsize = (*bpp)->b_bufsize;
1220 		(*bpp)->b_data = erealloc((*bpp)->b_data, nsize);
1221 		(void)memset((*bpp)->b_data + osize, 0, nsize - osize);
1222 	}
1223 
1224     out:
1225 	return (error);
1226 }
1227