FreeBSD/Linux Kernel Cross Reference
sys/ufs/ufs/ufs_bmap.c
1 /*
2 * Copyright (c) 1989, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 *
38 * @(#)ufs_bmap.c 8.7 (Berkeley) 3/21/95
39 * $FreeBSD$
40 */
41
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/buf.h>
45 #include <sys/proc.h>
46 #include <sys/vnode.h>
47 #include <sys/mount.h>
48 #include <sys/resourcevar.h>
49 #include <sys/conf.h>
50
51 #include <ufs/ufs/quota.h>
52 #include <ufs/ufs/inode.h>
53 #include <ufs/ufs/ufsmount.h>
54 #include <ufs/ufs/ufs_extern.h>
55 #include <miscfs/specfs/specdev.h>
56
57 /*
58 * Bmap converts a the logical block number of a file to its physical block
59 * number on the disk. The conversion is done by using the logical block
60 * number to index into the array of block pointers described by the dinode.
61 */
62 int
63 ufs_bmap(ap)
64 struct vop_bmap_args /* {
65 struct vnode *a_vp;
66 ufs_daddr_t a_bn;
67 struct vnode **a_vpp;
68 ufs_daddr_t *a_bnp;
69 int *a_runp;
70 int *a_runb;
71 } */ *ap;
72 {
73 /*
74 * Check for underlying vnode requests and ensure that logical
75 * to physical mapping is requested.
76 */
77 if (ap->a_vpp != NULL)
78 *ap->a_vpp = VTOI(ap->a_vp)->i_devvp;
79 if (ap->a_bnp == NULL)
80 return (0);
81
82 return (ufs_bmaparray(ap->a_vp, ap->a_bn, ap->a_bnp, NULL, NULL,
83 ap->a_runp, ap->a_runb));
84 }
85
86 /*
87 * Indirect blocks are now on the vnode for the file. They are given negative
88 * logical block numbers. Indirect blocks are addressed by the negative
89 * address of the first data block to which they point. Double indirect blocks
90 * are addressed by one less than the address of the first indirect block to
91 * which they point. Triple indirect blocks are addressed by one less than
92 * the address of the first double indirect block to which they point.
93 *
94 * ufs_bmaparray does the bmap conversion, and if requested returns the
95 * array of logical blocks which must be traversed to get to a block.
96 * Each entry contains the offset into that block that gets you to the
97 * next block and the disk address of the block (if it is assigned).
98 */
99
100 int
101 ufs_bmaparray(vp, bn, bnp, ap, nump, runp, runb)
102 struct vnode *vp;
103 ufs_daddr_t bn;
104 ufs_daddr_t *bnp;
105 struct indir *ap;
106 int *nump;
107 int *runp;
108 int *runb;
109 {
110 register struct inode *ip;
111 struct buf *bp;
112 struct ufsmount *ump;
113 struct mount *mp;
114 struct vnode *devvp;
115 struct indir a[NIADDR+1], *xap;
116 ufs_daddr_t daddr;
117 long metalbn;
118 int error, maxrun = 0, num;
119
120 ip = VTOI(vp);
121 mp = vp->v_mount;
122 ump = VFSTOUFS(mp);
123 #ifdef DIAGNOSTIC
124 if (ap != NULL && nump == NULL || ap == NULL && nump != NULL)
125 panic("ufs_bmaparray: invalid arguments");
126 #endif
127
128 if (runp) {
129 *runp = 0;
130 }
131
132 if (runb) {
133 *runb = 0;
134 }
135
136 maxrun = 0;
137 if (runp || runb || (vp->v_maxio == 0)) {
138
139 struct vnode *devvp;
140 int blksize;
141
142 blksize = mp->mnt_stat.f_iosize;
143
144 /*
145 * XXX
146 * If MAXPHYS is the largest transfer the disks can handle,
147 * we probably want maxrun to be 1 block less so that we
148 * don't create a block larger than the device can handle.
149 */
150 devvp = ip->i_devvp;
151
152 if (devvp != NULL && devvp->v_tag != VT_MFS &&
153 devvp->v_type == VBLK) {
154 if (bdevsw[major(devvp->v_rdev)]->d_maxio > MAXPHYS) {
155 maxrun = MAXPHYS;
156 vp->v_maxio = MAXPHYS;
157 } else {
158 maxrun = bdevsw[major(devvp->v_rdev)]->d_maxio;
159 vp->v_maxio = bdevsw[major(devvp->v_rdev)]->d_maxio;
160 }
161 maxrun = maxrun / blksize;
162 maxrun -= 1;
163 }
164
165 if (maxrun <= 0) {
166 vp->v_maxio = DFLTPHYS;
167 maxrun = DFLTPHYS / blksize;
168 maxrun -= 1;
169 }
170 }
171
172 xap = ap == NULL ? a : ap;
173 if (!nump)
174 nump = #
175 error = ufs_getlbns(vp, bn, xap, nump);
176 if (error)
177 return (error);
178
179 num = *nump;
180 if (num == 0) {
181 *bnp = blkptrtodb(ump, ip->i_db[bn]);
182 if (*bnp == 0)
183 *bnp = -1;
184 else if (runp) {
185 daddr_t bnb = bn;
186 for (++bn; bn < NDADDR && *runp < maxrun &&
187 is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]);
188 ++bn, ++*runp);
189 bn = bnb;
190 if (runb && (bn > 0)) {
191 for (--bn; (bn >= 0) && (*runb < maxrun) &&
192 is_sequential(ump, ip->i_db[bn],
193 ip->i_db[bn+1]);
194 --bn, ++*runb);
195 }
196 }
197 return (0);
198 }
199
200
201 /* Get disk address out of indirect block array */
202 daddr = ip->i_ib[xap->in_off];
203
204 devvp = VFSTOUFS(vp->v_mount)->um_devvp;
205 for (bp = NULL, ++xap; --num; ++xap) {
206 /*
207 * Exit the loop if there is no disk address assigned yet and
208 * the indirect block isn't in the cache, or if we were
209 * looking for an indirect block and we've found it.
210 */
211
212 metalbn = xap->in_lbn;
213 if ((daddr == 0 && !incore(vp, metalbn)) || metalbn == bn)
214 break;
215 /*
216 * If we get here, we've either got the block in the cache
217 * or we have a disk address for it, go fetch it.
218 */
219 if (bp)
220 bqrelse(bp);
221
222 xap->in_exists = 1;
223 bp = getblk(vp, metalbn, mp->mnt_stat.f_iosize, 0, 0);
224 if ((bp->b_flags & B_CACHE) == 0) {
225 #ifdef DIAGNOSTIC
226 if (!daddr)
227 panic("ufs_bmaparray: indirect block not in cache");
228 #endif
229 bp->b_blkno = blkptrtodb(ump, daddr);
230 bp->b_flags |= B_READ;
231 vfs_busy_pages(bp, 0);
232 VOP_STRATEGY(bp->b_vp, bp);
233 curproc->p_stats->p_ru.ru_inblock++; /* XXX */
234 error = biowait(bp);
235 if (error) {
236 brelse(bp);
237 return (error);
238 }
239 }
240
241 daddr = ((ufs_daddr_t *)bp->b_data)[xap->in_off];
242 if (num == 1 && daddr && runp) {
243 for (bn = xap->in_off + 1;
244 bn < MNINDIR(ump) && *runp < maxrun &&
245 is_sequential(ump,
246 ((ufs_daddr_t *)bp->b_data)[bn - 1],
247 ((ufs_daddr_t *)bp->b_data)[bn]);
248 ++bn, ++*runp);
249 bn = xap->in_off;
250 if (runb && bn) {
251 for(--bn; bn > 0 && *runb < maxrun &&
252 is_sequential(ump, ((daddr_t *)bp->b_data)[bn],
253 ((daddr_t *)bp->b_data)[bn+1]);
254 --bn, ++*runb);
255 }
256 }
257 }
258 if (bp)
259 bqrelse(bp);
260
261 daddr = blkptrtodb(ump, daddr);
262 *bnp = daddr == 0 ? -1 : daddr;
263 return (0);
264 }
265
266 /*
267 * Create an array of logical block number/offset pairs which represent the
268 * path of indirect blocks required to access a data block. The first "pair"
269 * contains the logical block number of the appropriate single, double or
270 * triple indirect block and the offset into the inode indirect block array.
271 * Note, the logical block number of the inode single/double/triple indirect
272 * block appears twice in the array, once with the offset into the i_ib and
273 * once with the offset into the page itself.
274 */
275 int
276 ufs_getlbns(vp, bn, ap, nump)
277 struct vnode *vp;
278 ufs_daddr_t bn;
279 struct indir *ap;
280 int *nump;
281 {
282 long blockcnt, metalbn, realbn;
283 struct ufsmount *ump;
284 int i, numlevels, off;
285 int64_t qblockcnt;
286
287 ump = VFSTOUFS(vp->v_mount);
288 if (nump)
289 *nump = 0;
290 numlevels = 0;
291 realbn = bn;
292 if ((long)bn < 0)
293 bn = -(long)bn;
294
295 /* The first NDADDR blocks are direct blocks. */
296 if (bn < NDADDR)
297 return (0);
298
299 /*
300 * Determine the number of levels of indirection. After this loop
301 * is done, blockcnt indicates the number of data blocks possible
302 * at the previous level of indirection, and NIADDR - i is the number
303 * of levels of indirection needed to locate the requested block.
304 */
305 for (blockcnt = 1, i = NIADDR, bn -= NDADDR;; i--, bn -= blockcnt) {
306 if (i == 0)
307 return (EFBIG);
308 /*
309 * Use int64_t's here to avoid overflow for triple indirect
310 * blocks when longs have 32 bits and the block size is more
311 * than 4K.
312 */
313 qblockcnt = (int64_t)blockcnt * MNINDIR(ump);
314 if (bn < qblockcnt)
315 break;
316 blockcnt = qblockcnt;
317 }
318
319 /* Calculate the address of the first meta-block. */
320 if (realbn >= 0)
321 metalbn = -(realbn - bn + NIADDR - i);
322 else
323 metalbn = -(-realbn - bn + NIADDR - i);
324
325 /*
326 * At each iteration, off is the offset into the bap array which is
327 * an array of disk addresses at the current level of indirection.
328 * The logical block number and the offset in that block are stored
329 * into the argument array.
330 */
331 ap->in_lbn = metalbn;
332 ap->in_off = off = NIADDR - i;
333 ap->in_exists = 0;
334 ap++;
335 for (++numlevels; i <= NIADDR; i++) {
336 /* If searching for a meta-data block, quit when found. */
337 if (metalbn == realbn)
338 break;
339
340 off = (bn / blockcnt) % MNINDIR(ump);
341
342 ++numlevels;
343 ap->in_lbn = metalbn;
344 ap->in_off = off;
345 ap->in_exists = 0;
346 ++ap;
347
348 metalbn -= -1 + off * blockcnt;
349 blockcnt /= MNINDIR(ump);
350 }
351 if (nump)
352 *nump = numlevels;
353 return (0);
354 }
Cache object: 9a8a17c3853e4aeddf28f36cac9f564f
|