xref: /freebsd/sys/fs/ext2fs/ext2_balloc.c (revision 12a88a3d637e7e7a3726e8dedbcf82cb96cea529)
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  * 3. 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  * $FreeBSD$
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 static int
55 ext2_ext_balloc(struct inode *ip, uint32_t lbn, int size,
56     struct ucred *cred, struct buf **bpp, int flags)
57 {
58 	struct m_ext2fs *fs;
59 	struct buf *bp = NULL;
60 	struct vnode *vp = ITOV(ip);
61 	uint32_t nb;
62 	int osize, nsize, blks, error, allocated;
63 
64 	fs = ip->i_e2fs;
65 	blks = howmany(size, fs->e2fs_bsize);
66 
67 	error = ext4_ext_get_blocks(ip, lbn, blks, cred, NULL, &allocated, &nb);
68 	if (error)
69 		return (error);
70 
71 	if (allocated) {
72 		if (ip->i_size < (lbn + 1) * fs->e2fs_bsize)
73 			nsize = fragroundup(fs, size);
74 		else
75 			nsize = fs->e2fs_bsize;
76 
77 		bp = getblk(vp, lbn, nsize, 0, 0, 0);
78 		if(!bp)
79 			return (EIO);
80 
81 		bp->b_blkno = fsbtodb(fs, nb);
82 		if (flags & BA_CLRBUF)
83 			vfs_bio_clrbuf(bp);
84 	} else {
85 		if (ip->i_size >= (lbn + 1) * fs->e2fs_bsize) {
86 
87 			error = bread(vp, lbn, fs->e2fs_bsize, NOCRED, &bp);
88 			if (error) {
89 				brelse(bp);
90 				return (error);
91 			}
92 			bp->b_blkno = fsbtodb(fs, nb);
93 			*bpp = bp;
94 			return (0);
95 		}
96 
97 		/*
98 		 * Consider need to reallocate a fragment.
99 		 */
100 		osize = fragroundup(fs, blkoff(fs, ip->i_size));
101 		nsize = fragroundup(fs, size);
102 		if (nsize <= osize) {
103 			error = bread(vp, lbn, osize, NOCRED, &bp);
104 			if (error) {
105 				brelse(bp);
106 				return (error);
107 			}
108 			bp->b_blkno = fsbtodb(fs, nb);
109 		} else {
110 			error = bread(vp, lbn, fs->e2fs_bsize, NOCRED, &bp);
111 			if (error) {
112 				brelse(bp);
113 				return (error);
114 			}
115 			bp->b_blkno = fsbtodb(fs, nb);
116 		}
117 	}
118 
119 	*bpp = bp;
120 
121 	return (error);
122 }
123 
124 /*
125  * Balloc defines the structure of filesystem storage
126  * by allocating the physical blocks on a device given
127  * the inode and the logical block number in a file.
128  */
129 int
130 ext2_balloc(struct inode *ip, e2fs_lbn_t lbn, int size, struct ucred *cred,
131     struct buf **bpp, int flags)
132 {
133 	struct m_ext2fs *fs;
134 	struct ext2mount *ump;
135 	struct buf *bp, *nbp;
136 	struct vnode *vp = ITOV(ip);
137 	struct indir indirs[EXT2_NIADDR + 2];
138 	e4fs_daddr_t nb, newb;
139 	e2fs_daddr_t *bap, pref;
140 	int osize, nsize, num, i, error;
141 
142 	*bpp = NULL;
143 	if (lbn < 0)
144 		return (EFBIG);
145 	fs = ip->i_e2fs;
146 	ump = ip->i_ump;
147 
148 	/*
149 	 * check if this is a sequential block allocation.
150 	 * If so, increment next_alloc fields to allow ext2_blkpref
151 	 * to make a good guess
152 	 */
153 	if (lbn == ip->i_next_alloc_block + 1) {
154 		ip->i_next_alloc_block++;
155 		ip->i_next_alloc_goal++;
156 	}
157 
158 	if (ip->i_flag & IN_E4EXTENTS)
159 		return (ext2_ext_balloc(ip, lbn, size, cred, bpp, flags));
160 
161 	/*
162 	 * The first EXT2_NDADDR blocks are direct blocks
163 	 */
164 	if (lbn < EXT2_NDADDR) {
165 		nb = ip->i_db[lbn];
166 		/*
167 		 * no new block is to be allocated, and no need to expand
168 		 * the file
169 		 */
170 		if (nb != 0 && ip->i_size >= (lbn + 1) * fs->e2fs_bsize) {
171 			error = bread(vp, lbn, fs->e2fs_bsize, NOCRED, &bp);
172 			if (error) {
173 				brelse(bp);
174 				return (error);
175 			}
176 			bp->b_blkno = fsbtodb(fs, nb);
177 			*bpp = bp;
178 			return (0);
179 		}
180 		if (nb != 0) {
181 			/*
182 			 * Consider need to reallocate a fragment.
183 			 */
184 			osize = fragroundup(fs, blkoff(fs, ip->i_size));
185 			nsize = fragroundup(fs, size);
186 			if (nsize <= osize) {
187 				error = bread(vp, lbn, osize, NOCRED, &bp);
188 				if (error) {
189 					brelse(bp);
190 					return (error);
191 				}
192 				bp->b_blkno = fsbtodb(fs, nb);
193 			} else {
194 				/*
195 				 * Godmar thinks: this shouldn't happen w/o
196 				 * fragments
197 				 */
198 				printf("nsize %d(%d) > osize %d(%d) nb %d\n",
199 				    (int)nsize, (int)size, (int)osize,
200 				    (int)ip->i_size, (int)nb);
201 				panic(
202 				    "ext2_balloc: Something is terribly wrong");
203 /*
204  * please note there haven't been any changes from here on -
205  * FFS seems to work.
206  */
207 			}
208 		} else {
209 			if (ip->i_size < (lbn + 1) * fs->e2fs_bsize)
210 				nsize = fragroundup(fs, size);
211 			else
212 				nsize = fs->e2fs_bsize;
213 			EXT2_LOCK(ump);
214 			error = ext2_alloc(ip, lbn,
215 			    ext2_blkpref(ip, lbn, (int)lbn, &ip->i_db[0], 0),
216 			    nsize, cred, &newb);
217 			if (error)
218 				return (error);
219 			bp = getblk(vp, lbn, nsize, 0, 0, 0);
220 			bp->b_blkno = fsbtodb(fs, newb);
221 			if (flags & BA_CLRBUF)
222 				vfs_bio_clrbuf(bp);
223 		}
224 		ip->i_db[lbn] = dbtofsb(fs, bp->b_blkno);
225 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
226 		*bpp = bp;
227 		return (0);
228 	}
229 	/*
230 	 * Determine the number of levels of indirection.
231 	 */
232 	pref = 0;
233 	if ((error = ext2_getlbns(vp, lbn, indirs, &num)) != 0)
234 		return (error);
235 #ifdef INVARIANTS
236 	if (num < 1)
237 		panic("ext2_balloc: ext2_getlbns returned indirect block");
238 #endif
239 	/*
240 	 * Fetch the first indirect block allocating if necessary.
241 	 */
242 	--num;
243 	nb = ip->i_ib[indirs[0].in_off];
244 	if (nb == 0) {
245 		EXT2_LOCK(ump);
246 		pref = ext2_blkpref(ip, lbn, indirs[0].in_off +
247 		    EXT2_NDIR_BLOCKS, &ip->i_db[0], 0);
248 		if ((error = ext2_alloc(ip, lbn, pref, fs->e2fs_bsize, cred,
249 		    &newb)))
250 			return (error);
251 		nb = newb;
252 		bp = getblk(vp, indirs[1].in_lbn, fs->e2fs_bsize, 0, 0, 0);
253 		bp->b_blkno = fsbtodb(fs, newb);
254 		vfs_bio_clrbuf(bp);
255 		/*
256 		 * Write synchronously so that indirect blocks
257 		 * never point at garbage.
258 		 */
259 		if ((error = bwrite(bp)) != 0) {
260 			ext2_blkfree(ip, nb, fs->e2fs_bsize);
261 			return (error);
262 		}
263 		ip->i_ib[indirs[0].in_off] = newb;
264 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
265 	}
266 	/*
267 	 * Fetch through the indirect blocks, allocating as necessary.
268 	 */
269 	for (i = 1;;) {
270 		error = bread(vp,
271 		    indirs[i].in_lbn, (int)fs->e2fs_bsize, NOCRED, &bp);
272 		if (error) {
273 			brelse(bp);
274 			return (error);
275 		}
276 		bap = (e2fs_daddr_t *)bp->b_data;
277 		nb = bap[indirs[i].in_off];
278 		if (i == num)
279 			break;
280 		i += 1;
281 		if (nb != 0) {
282 			bqrelse(bp);
283 			continue;
284 		}
285 		EXT2_LOCK(ump);
286 		if (pref == 0)
287 			pref = ext2_blkpref(ip, lbn, indirs[i].in_off, bap,
288 			    bp->b_lblkno);
289 		error = ext2_alloc(ip, lbn, pref, (int)fs->e2fs_bsize, cred, &newb);
290 		if (error) {
291 			brelse(bp);
292 			return (error);
293 		}
294 		nb = newb;
295 		nbp = getblk(vp, indirs[i].in_lbn, fs->e2fs_bsize, 0, 0, 0);
296 		nbp->b_blkno = fsbtodb(fs, nb);
297 		vfs_bio_clrbuf(nbp);
298 		/*
299 		 * Write synchronously so that indirect blocks
300 		 * never point at garbage.
301 		 */
302 		if ((error = bwrite(nbp)) != 0) {
303 			ext2_blkfree(ip, nb, fs->e2fs_bsize);
304 			EXT2_UNLOCK(ump);
305 			brelse(bp);
306 			return (error);
307 		}
308 		bap[indirs[i - 1].in_off] = nb;
309 		/*
310 		 * If required, write synchronously, otherwise use
311 		 * delayed write.
312 		 */
313 		if (flags & IO_SYNC) {
314 			bwrite(bp);
315 		} else {
316 			if (bp->b_bufsize == fs->e2fs_bsize)
317 				bp->b_flags |= B_CLUSTEROK;
318 			bdwrite(bp);
319 		}
320 	}
321 	/*
322 	 * Get the data block, allocating if necessary.
323 	 */
324 	if (nb == 0) {
325 		EXT2_LOCK(ump);
326 		pref = ext2_blkpref(ip, lbn, indirs[i].in_off, &bap[0],
327 		    bp->b_lblkno);
328 		if ((error = ext2_alloc(ip,
329 		    lbn, pref, (int)fs->e2fs_bsize, cred, &newb)) != 0) {
330 			brelse(bp);
331 			return (error);
332 		}
333 		nb = newb;
334 		nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0, 0);
335 		nbp->b_blkno = fsbtodb(fs, nb);
336 		if (flags & BA_CLRBUF)
337 			vfs_bio_clrbuf(nbp);
338 		bap[indirs[i].in_off] = nb;
339 		/*
340 		 * If required, write synchronously, otherwise use
341 		 * delayed write.
342 		 */
343 		if (flags & IO_SYNC) {
344 			bwrite(bp);
345 		} else {
346 			if (bp->b_bufsize == fs->e2fs_bsize)
347 				bp->b_flags |= B_CLUSTEROK;
348 			bdwrite(bp);
349 		}
350 		*bpp = nbp;
351 		return (0);
352 	}
353 	brelse(bp);
354 	if (flags & BA_CLRBUF) {
355 		int seqcount = (flags & BA_SEQMASK) >> BA_SEQSHIFT;
356 
357 		if (seqcount && (vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0) {
358 			error = cluster_read(vp, ip->i_size, lbn,
359 			    (int)fs->e2fs_bsize, NOCRED,
360 			    MAXBSIZE, seqcount, 0, &nbp);
361 		} else {
362 			error = bread(vp, lbn, (int)fs->e2fs_bsize, NOCRED, &nbp);
363 		}
364 		if (error) {
365 			brelse(nbp);
366 			return (error);
367 		}
368 	} else {
369 		nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0, 0);
370 		nbp->b_blkno = fsbtodb(fs, nb);
371 	}
372 	*bpp = nbp;
373 	return (0);
374 }
375