ext2_balloc.c revision 311232
1/*-
2 *  modified for Lites 1.1
3 *
4 *  Aug 1995, Godmar Back (gback@cs.utah.edu)
5 *  University of Utah, Department of Computer Science
6 */
7/*-
8 * Copyright (c) 1982, 1986, 1989, 1993
9 *	The Regents of the University of California.  All rights reserved.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 *    notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 *    notice, this list of conditions and the following disclaimer in the
18 *    documentation and/or other materials provided with the distribution.
19 * 4. Neither the name of the University nor the names of its contributors
20 *    may be used to endorse or promote products derived from this software
21 *    without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 *	@(#)ffs_balloc.c	8.4 (Berkeley) 9/23/93
36 * $FreeBSD: stable/10/sys/fs/ext2fs/ext2_balloc.c 311232 2017-01-04 02:43:33Z pfg $
37 */
38
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/bio.h>
42#include <sys/buf.h>
43#include <sys/lock.h>
44#include <sys/mount.h>
45#include <sys/vnode.h>
46
47#include <fs/ext2fs/fs.h>
48#include <fs/ext2fs/inode.h>
49#include <fs/ext2fs/ext2fs.h>
50#include <fs/ext2fs/ext2_dinode.h>
51#include <fs/ext2fs/ext2_extern.h>
52#include <fs/ext2fs/ext2_mount.h>
53
54/*
55 * Balloc defines the structure of filesystem storage
56 * by allocating the physical blocks on a device given
57 * the inode and the logical block number in a file.
58 */
59int
60ext2_balloc(struct inode *ip, e2fs_lbn_t lbn, int size, struct ucred *cred,
61    struct buf **bpp, int flags)
62{
63	struct m_ext2fs *fs;
64	struct ext2mount *ump;
65	struct buf *bp, *nbp;
66	struct vnode *vp = ITOV(ip);
67	struct indir indirs[NIADDR + 2];
68	e4fs_daddr_t nb, newb;
69	e2fs_daddr_t *bap, pref;
70	int osize, nsize, num, i, error;
71
72	*bpp = NULL;
73	if (lbn < 0)
74		return (EFBIG);
75	fs = ip->i_e2fs;
76	ump = ip->i_ump;
77
78	/*
79	 * check if this is a sequential block allocation.
80	 * If so, increment next_alloc fields to allow ext2_blkpref
81	 * to make a good guess
82	 */
83	if (lbn == ip->i_next_alloc_block + 1) {
84		ip->i_next_alloc_block++;
85		ip->i_next_alloc_goal++;
86	}
87	/*
88	 * The first NDADDR blocks are direct blocks
89	 */
90	if (lbn < NDADDR) {
91		nb = ip->i_db[lbn];
92		/*
93		 * no new block is to be allocated, and no need to expand
94		 * the file
95		 */
96		if (nb != 0 && ip->i_size >= (lbn + 1) * fs->e2fs_bsize) {
97			error = bread(vp, lbn, fs->e2fs_bsize, NOCRED, &bp);
98			if (error) {
99				brelse(bp);
100				return (error);
101			}
102			bp->b_blkno = fsbtodb(fs, nb);
103			*bpp = bp;
104			return (0);
105		}
106		if (nb != 0) {
107			/*
108			 * Consider need to reallocate a fragment.
109			 */
110			osize = fragroundup(fs, blkoff(fs, ip->i_size));
111			nsize = fragroundup(fs, size);
112			if (nsize <= osize) {
113				error = bread(vp, lbn, osize, NOCRED, &bp);
114				if (error) {
115					brelse(bp);
116					return (error);
117				}
118				bp->b_blkno = fsbtodb(fs, nb);
119			} else {
120				/*
121				 * Godmar thinks: this shouldn't happen w/o
122				 * fragments
123				 */
124				printf("nsize %d(%d) > osize %d(%d) nb %d\n",
125				    (int)nsize, (int)size, (int)osize,
126				    (int)ip->i_size, (int)nb);
127				panic(
128				    "ext2_balloc: Something is terribly wrong");
129/*
130 * please note there haven't been any changes from here on -
131 * FFS seems to work.
132 */
133			}
134		} else {
135			if (ip->i_size < (lbn + 1) * fs->e2fs_bsize)
136				nsize = fragroundup(fs, size);
137			else
138				nsize = fs->e2fs_bsize;
139			EXT2_LOCK(ump);
140			error = ext2_alloc(ip, lbn,
141			    ext2_blkpref(ip, lbn, (int)lbn, &ip->i_db[0], 0),
142			    nsize, cred, &newb);
143			if (error)
144				return (error);
145			bp = getblk(vp, lbn, nsize, 0, 0, 0);
146			bp->b_blkno = fsbtodb(fs, newb);
147			if (flags & BA_CLRBUF)
148				vfs_bio_clrbuf(bp);
149		}
150		ip->i_db[lbn] = dbtofsb(fs, bp->b_blkno);
151		ip->i_flag |= IN_CHANGE | IN_UPDATE;
152		*bpp = bp;
153		return (0);
154	}
155	/*
156	 * Determine the number of levels of indirection.
157	 */
158	pref = 0;
159	if ((error = ext2_getlbns(vp, lbn, indirs, &num)) != 0)
160		return (error);
161#ifdef INVARIANTS
162	if (num < 1)
163		panic("ext2_balloc: ext2_getlbns returned indirect block");
164#endif
165	/*
166	 * Fetch the first indirect block allocating if necessary.
167	 */
168	--num;
169	nb = ip->i_ib[indirs[0].in_off];
170	if (nb == 0) {
171		EXT2_LOCK(ump);
172		pref = ext2_blkpref(ip, lbn, indirs[0].in_off +
173		    EXT2_NDIR_BLOCKS, &ip->i_db[0], 0);
174		if ((error = ext2_alloc(ip, lbn, pref, fs->e2fs_bsize, cred,
175		    &newb)))
176			return (error);
177		nb = newb;
178		bp = getblk(vp, indirs[1].in_lbn, fs->e2fs_bsize, 0, 0, 0);
179		bp->b_blkno = fsbtodb(fs, newb);
180		vfs_bio_clrbuf(bp);
181		/*
182		 * Write synchronously so that indirect blocks
183		 * never point at garbage.
184		 */
185		if ((error = bwrite(bp)) != 0) {
186			ext2_blkfree(ip, nb, fs->e2fs_bsize);
187			return (error);
188		}
189		ip->i_ib[indirs[0].in_off] = newb;
190		ip->i_flag |= IN_CHANGE | IN_UPDATE;
191	}
192	/*
193	 * Fetch through the indirect blocks, allocating as necessary.
194	 */
195	for (i = 1;;) {
196		error = bread(vp,
197		    indirs[i].in_lbn, (int)fs->e2fs_bsize, NOCRED, &bp);
198		if (error) {
199			brelse(bp);
200			return (error);
201		}
202		bap = (e2fs_daddr_t *)bp->b_data;
203		nb = bap[indirs[i].in_off];
204		if (i == num)
205			break;
206		i += 1;
207		if (nb != 0) {
208			bqrelse(bp);
209			continue;
210		}
211		EXT2_LOCK(ump);
212		if (pref == 0)
213			pref = ext2_blkpref(ip, lbn, indirs[i].in_off, bap,
214			    bp->b_lblkno);
215		error = ext2_alloc(ip, lbn, pref, (int)fs->e2fs_bsize, cred, &newb);
216		if (error) {
217			brelse(bp);
218			return (error);
219		}
220		nb = newb;
221		nbp = getblk(vp, indirs[i].in_lbn, fs->e2fs_bsize, 0, 0, 0);
222		nbp->b_blkno = fsbtodb(fs, nb);
223		vfs_bio_clrbuf(nbp);
224		/*
225		 * Write synchronously so that indirect blocks
226		 * never point at garbage.
227		 */
228		if ((error = bwrite(nbp)) != 0) {
229			ext2_blkfree(ip, nb, fs->e2fs_bsize);
230			EXT2_UNLOCK(ump);
231			brelse(bp);
232			return (error);
233		}
234		bap[indirs[i - 1].in_off] = nb;
235		/*
236		 * If required, write synchronously, otherwise use
237		 * delayed write.
238		 */
239		if (flags & IO_SYNC) {
240			bwrite(bp);
241		} else {
242			if (bp->b_bufsize == fs->e2fs_bsize)
243				bp->b_flags |= B_CLUSTEROK;
244			bdwrite(bp);
245		}
246	}
247	/*
248	 * Get the data block, allocating if necessary.
249	 */
250	if (nb == 0) {
251		EXT2_LOCK(ump);
252		pref = ext2_blkpref(ip, lbn, indirs[i].in_off, &bap[0],
253		    bp->b_lblkno);
254		if ((error = ext2_alloc(ip,
255		    lbn, pref, (int)fs->e2fs_bsize, cred, &newb)) != 0) {
256			brelse(bp);
257			return (error);
258		}
259		nb = newb;
260		nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0, 0);
261		nbp->b_blkno = fsbtodb(fs, nb);
262		if (flags & BA_CLRBUF)
263			vfs_bio_clrbuf(nbp);
264		bap[indirs[i].in_off] = nb;
265		/*
266		 * If required, write synchronously, otherwise use
267		 * delayed write.
268		 */
269		if (flags & IO_SYNC) {
270			bwrite(bp);
271		} else {
272			if (bp->b_bufsize == fs->e2fs_bsize)
273				bp->b_flags |= B_CLUSTEROK;
274			bdwrite(bp);
275		}
276		*bpp = nbp;
277		return (0);
278	}
279	brelse(bp);
280	if (flags & BA_CLRBUF) {
281		int seqcount = (flags & BA_SEQMASK) >> BA_SEQSHIFT;
282
283		if (seqcount && (vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0) {
284			error = cluster_read(vp, ip->i_size, lbn,
285			    (int)fs->e2fs_bsize, NOCRED,
286			    MAXBSIZE, seqcount, 0, &nbp);
287		} else {
288			error = bread(vp, lbn, (int)fs->e2fs_bsize, NOCRED, &nbp);
289		}
290		if (error) {
291			brelse(nbp);
292			return (error);
293		}
294	} else {
295		nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0, 0);
296		nbp->b_blkno = fsbtodb(fs, nb);
297	}
298	*bpp = nbp;
299	return (0);
300}
301