2 * Copyright (c) 1982, 1986, 1989, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
33 * @(#)ffs_balloc.c 8.8 (Berkeley) 6/16/95
34 * $FreeBSD: src/sys/ufs/ffs/ffs_balloc.c,v 1.26.2.1 2002/10/10 19:48:20 dillon Exp $
35 * $DragonFly: src/sys/vfs/ufs/ffs_balloc.c,v 1.19 2008/05/21 18:49:49 dillon Exp $
38 #include <sys/param.h>
39 #include <sys/systm.h>
43 #include <sys/mount.h>
44 #include <sys/vnode.h>
50 #include "ufs_extern.h"
53 #include "ffs_extern.h"
56 * ffs_balloc(struct vnode *a_vp, ufs_daddr_t a_lbn, int a_size,
57 * struct ucred *a_cred, int a_flags, struct buf *a_bpp)
59 * Balloc defines the structure of filesystem storage by allocating
60 * the physical blocks on a device given the inode and the logical
61 * block number in a file.
63 * NOTE: B_CLRBUF - this flag tells balloc to clear invalid portions
64 * of the buffer. However, any dirty bits will override missing
65 * valid bits. This case occurs when writable mmaps are truncated
69 ffs_balloc(struct vop_balloc_args *ap)
78 struct buf *bp, *nbp, *dbp;
80 struct indir indirs[NIADDR + 2];
81 ufs_daddr_t newb, *bap, pref;
82 int deallocated, osize, nsize, num, i, error;
83 ufs_daddr_t *allocib, *blkp, *allocblk, allociblk[NIADDR + 1];
84 ufs_daddr_t *lbns_remfree, lbns[NIADDR + 1];
91 lbn = lblkno(fs, ap->a_startoffset);
92 size = blkoff(fs, ap->a_startoffset) + ap->a_size;
93 if (size > fs->fs_bsize)
94 panic("ffs_balloc: blk too big");
102 * The vnode must be locked for us to be able to safely mess
103 * around with the inode.
105 if (vn_islocked(vp) != LK_EXCLUSIVE) {
106 panic("ffs_balloc: vnode %p not exclusively locked!", vp);
110 * If the next write will extend the file into a new block,
111 * and the file is currently composed of a fragment
112 * this fragment has to be extended to be a full block.
114 nb = lblkno(fs, ip->i_size);
115 if (nb < NDADDR && nb < lbn) {
117 * The filesize prior to this write can fit in direct
118 * blocks (ex. fragmentation is possibly done)
119 * we are now extending the file write beyond
120 * the block which has end of the file prior to this write.
122 osize = blksize(fs, ip, nb);
124 * osize gives disk allocated size in the last block. It is
125 * either in fragments or a file system block size.
127 if (osize < fs->fs_bsize && osize > 0) {
128 /* A few fragments are already allocated, since the
129 * current extends beyond this block allocated the
130 * complete block as fragments are on in last block.
132 error = ffs_realloccg(ip, nb,
133 ffs_blkpref(ip, nb, (int)nb, &ip->i_db[0]),
134 osize, (int)fs->fs_bsize, cred, &bp);
137 if (DOINGSOFTDEP(vp))
138 softdep_setup_allocdirect(ip, nb,
139 dofftofsb(fs, bp->b_bio2.bio_offset),
140 ip->i_db[nb], fs->fs_bsize, osize, bp);
141 /* adjust the inode size, we just grew */
142 ip->i_size = smalllblktosize(fs, nb + 1);
143 ip->i_db[nb] = dofftofsb(fs, bp->b_bio2.bio_offset);
144 ip->i_flag |= IN_CHANGE | IN_UPDATE;
149 /* bp is already released here */
153 * The first NDADDR blocks are direct blocks
157 if (nb != 0 && ip->i_size >= smalllblktosize(fs, lbn + 1)) {
158 error = bread(vp, lblktodoff(fs, lbn), fs->fs_bsize, &bp);
163 bp->b_bio2.bio_offset = fsbtodoff(fs, nb);
169 * Consider need to reallocate a fragment.
171 osize = fragroundup(fs, blkoff(fs, ip->i_size));
172 nsize = fragroundup(fs, size);
173 if (nsize <= osize) {
174 error = bread(vp, lblktodoff(fs, lbn),
180 bp->b_bio2.bio_offset = fsbtodoff(fs, nb);
183 * NOTE: ffs_realloccg() issues a bread().
185 error = ffs_realloccg(ip, lbn,
186 ffs_blkpref(ip, lbn, (int)lbn,
187 &ip->i_db[0]), osize, nsize, cred, &bp);
190 if (DOINGSOFTDEP(vp))
191 softdep_setup_allocdirect(ip, lbn,
192 dofftofsb(fs, bp->b_bio2.bio_offset),
193 nb, nsize, osize, bp);
196 if (ip->i_size < smalllblktosize(fs, lbn + 1))
197 nsize = fragroundup(fs, size);
199 nsize = fs->fs_bsize;
200 error = ffs_alloc(ip, lbn,
201 ffs_blkpref(ip, lbn, (int)lbn, &ip->i_db[0]),
205 bp = getblk(vp, lblktodoff(fs, lbn), nsize, 0, 0);
206 bp->b_bio2.bio_offset = fsbtodoff(fs, newb);
207 if (flags & B_CLRBUF)
209 if (DOINGSOFTDEP(vp))
210 softdep_setup_allocdirect(ip, lbn, newb, 0,
213 ip->i_db[lbn] = dofftofsb(fs, bp->b_bio2.bio_offset);
214 ip->i_flag |= IN_CHANGE | IN_UPDATE;
219 * Determine the number of levels of indirection.
222 if ((error = ufs_getlbns(vp, lbn, indirs, &num)) != 0)
226 panic ("ffs_balloc: ufs_bmaparray returned indirect block");
229 * Get a handle on the data block buffer before working through
230 * indirect blocks to avoid a deadlock between the VM system holding
231 * a locked VM page and issuing a BMAP (which tries to lock the
232 * indirect blocks), and the filesystem holding a locked indirect
233 * block and then trying to read a data block (which tries to lock
234 * the underlying VM pages).
236 dbp = getblk(vp, lblktodoff(fs, lbn), fs->fs_bsize, 0, 0);
242 allocblk = allociblk;
248 * Fetch the first indirect block directly from the inode, allocating
252 nb = ip->i_ib[indirs[0].in_off];
254 pref = ffs_blkpref(ip, lbn, 0, NULL);
256 * If the filesystem has run out of space we can skip the
257 * full fsync/undo of the main [fail] case since no undo
258 * history has been built yet. Hence the goto fail2.
260 if ((error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize,
265 *lbns_remfree++ = indirs[1].in_lbn;
266 bp = getblk(vp, lblktodoff(fs, indirs[1].in_lbn),
268 bp->b_bio2.bio_offset = fsbtodoff(fs, nb);
270 if (DOINGSOFTDEP(vp)) {
271 softdep_setup_allocdirect(ip, NDADDR + indirs[0].in_off,
272 newb, 0, fs->fs_bsize, 0, bp);
276 * Write synchronously so that indirect blocks
277 * never point at garbage.
281 else if ((error = bwrite(bp)) != 0)
284 allocib = &ip->i_ib[indirs[0].in_off];
286 ip->i_flag |= IN_CHANGE | IN_UPDATE;
290 * Fetch through the indirect blocks, allocating as necessary.
293 error = bread(vp, lblktodoff(fs, indirs[i].in_lbn), (int)fs->fs_bsize, &bp);
298 bap = (ufs_daddr_t *)bp->b_data;
299 nb = bap[indirs[i].in_off];
308 pref = ffs_blkpref(ip, lbn, 0, NULL);
310 ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, cred, &newb)) != 0) {
316 *lbns_remfree++ = indirs[i].in_lbn;
317 nbp = getblk(vp, lblktodoff(fs, indirs[i].in_lbn),
319 nbp->b_bio2.bio_offset = fsbtodoff(fs, nb);
321 if (DOINGSOFTDEP(vp)) {
322 softdep_setup_allocindir_meta(nbp, ip, bp,
323 indirs[i - 1].in_off, nb);
327 * Write synchronously so that indirect blocks
328 * never point at garbage.
330 if ((error = bwrite(nbp)) != 0) {
335 bap[indirs[i - 1].in_off] = nb;
336 if (allocib == NULL && unwindidx < 0)
339 * If required, write synchronously, otherwise use
342 if (flags & B_SYNC) {
345 if (bp->b_bufsize == fs->fs_bsize)
346 bp->b_flags |= B_CLUSTEROK;
352 * Get the data block, allocating if necessary. We have already
353 * called getblk() on the data block buffer, dbp. If we have to
354 * allocate it and B_CLRBUF has been set the inference is an intention
355 * to zero out the related disk blocks, so we do not have to issue
356 * a read. Instead we simply call vfs_bio_clrbuf(). If B_CLRBUF is
357 * not set the caller intends to overwrite the entire contents of the
358 * buffer and we don't waste time trying to clean up the contents.
360 * bp references the current indirect block. When allocating,
361 * the block must be updated.
364 pref = ffs_blkpref(ip, lbn, indirs[i].in_off, &bap[0]);
365 error = ffs_alloc(ip,
366 lbn, pref, (int)fs->fs_bsize, cred, &newb);
373 *lbns_remfree++ = lbn;
374 dbp->b_bio2.bio_offset = fsbtodoff(fs, nb);
375 if (flags & B_CLRBUF)
377 if (DOINGSOFTDEP(vp))
378 softdep_setup_allocindir_page(ip, lbn, bp,
379 indirs[i].in_off, nb, 0, dbp);
380 bap[indirs[i].in_off] = nb;
382 * If required, write synchronously, otherwise use
385 if (flags & B_SYNC) {
388 if (bp->b_bufsize == fs->fs_bsize)
389 bp->b_flags |= B_CLUSTEROK;
398 * At this point all related indirect blocks have been allocated
399 * if necessary and released. bp is no longer valid. dbp holds
400 * our getblk()'d data block.
402 * XXX we previously performed a cluster_read operation here.
404 if (flags & B_CLRBUF) {
406 * If B_CLRBUF is set we must validate the invalid portions
407 * of the buffer. This typically requires a read-before-
408 * write. The strategy call will fill in bio_offset in that
411 * If we hit this case we do a cluster read if possible
412 * since nearby data blocks are likely to be accessed soon
415 if ((dbp->b_flags & B_CACHE) == 0) {
417 seqcount = (flags & B_SEQMASK) >> B_SEQSHIFT;
419 (vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0) {
420 error = cluster_read(vp, (off_t)ip->i_size,
427 error = bread(vp, lblktodoff(fs, lbn),
428 (int)fs->fs_bsize, &dbp);
433 dbp->b_bio2.bio_offset = fsbtodoff(fs, nb);
437 * If B_CLRBUF is not set the caller intends to overwrite
438 * the entire contents of the buffer. We can simply set
439 * bio_offset and we are done.
441 dbp->b_bio2.bio_offset = fsbtodoff(fs, nb);
447 * If we have failed part way through block allocation, we
448 * have to deallocate any indirect blocks that we have allocated.
449 * We have to fsync the file before we start to get rid of all
450 * of its dependencies so that we do not leave them dangling.
451 * We have to sync it at the end so that the soft updates code
452 * does not find any untracked changes. Although this is really
453 * slow, running out of disk space is not expected to be a common
454 * occurence. The error return from fsync is ignored as we already
455 * have an error to return to the user.
457 VOP_FSYNC(vp, MNT_WAIT, 0);
458 for (deallocated = 0, blkp = allociblk, lbns_remfree = lbns;
459 blkp < allocblk; blkp++, lbns_remfree++) {
461 * We shall not leave the freed blocks on the vnode
462 * buffer object lists.
464 bp = getblk(vp, *lbns_remfree, fs->fs_bsize, 0, 0);
465 bp->b_flags |= (B_INVAL | B_RELBUF);
467 deallocated += fs->fs_bsize;
470 if (allocib != NULL) {
472 } else if (unwindidx >= 0) {
475 r = bread(vp, lblktodoff(fs, indirs[unwindidx].in_lbn), (int)fs->fs_bsize, &bp);
477 panic("Could not unwind indirect block, error %d", r);
480 bap = (ufs_daddr_t *)bp->b_data;
481 bap[indirs[unwindidx].in_off] = 0;
482 if (flags & B_SYNC) {
485 if (bp->b_bufsize == fs->fs_bsize)
486 bp->b_flags |= B_CLUSTEROK;
494 * Restore user's disk quota because allocation failed.
496 (void) ufs_chkdq(ip, (long)-btodb(deallocated), cred, FORCE);
498 ip->i_blocks -= btodb(deallocated);
499 ip->i_flag |= IN_CHANGE | IN_UPDATE;
501 VOP_FSYNC(vp, MNT_WAIT, 0);
504 * After the buffers are invalidated and on-disk pointers are
505 * cleared, free the blocks.
507 for (blkp = allociblk; blkp < allocblk; blkp++) {
508 ffs_blkfree(ip, *blkp, fs->fs_bsize);
512 * Cleanup the data block we getblk()'d before returning.