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.
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
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. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * @(#)ufs_bmap.c 8.7 (Berkeley) 3/21/95
38 #include <sys/param.h>
39 #include <sys/systm.h>
43 #include <sys/vnode.h>
44 #include <sys/mount.h>
45 #include <sys/racct.h>
46 #include <sys/resourcevar.h>
49 #include <fs/ext2fs/inode.h>
50 #include <fs/ext2fs/fs.h>
51 #include <fs/ext2fs/ext2fs.h>
52 #include <fs/ext2fs/ext2_dinode.h>
53 #include <fs/ext2fs/ext2_extern.h>
54 #include <fs/ext2fs/ext2_mount.h>
56 static int ext4_bmapext(struct vnode *, int32_t, int64_t *, int *, int *);
59 * Bmap converts the logical block number of a file to its physical block
60 * number on the disk. The conversion is done by using the logical block
61 * number to index into the array of block pointers described by the dinode.
64 ext2_bmap(struct vop_bmap_args *ap)
70 * Check for underlying vnode requests and ensure that logical
71 * to physical mapping is requested.
73 if (ap->a_bop != NULL)
74 *ap->a_bop = &VTOI(ap->a_vp)->i_devvp->v_bufobj;
75 if (ap->a_bnp == NULL)
78 if (VTOI(ap->a_vp)->i_flag & IN_E4EXTENTS)
79 error = ext4_bmapext(ap->a_vp, ap->a_bn, &blkno,
80 ap->a_runp, ap->a_runb);
82 error = ext2_bmaparray(ap->a_vp, ap->a_bn, &blkno,
83 ap->a_runp, ap->a_runb);
89 * Convert the logical block number of a file to its physical block number
90 * on the disk within ext4 extents.
93 ext4_bmapext(struct vnode *vp, int32_t bn, int64_t *bnp, int *runp, int *runb)
97 struct ext4_extent *ep;
98 struct ext4_extent_path path = {.ep_bp = NULL};
112 ext4_ext_find_extent(fs, ip, lbn, &path);
113 if (path.ep_is_sparse) {
116 *runp = path.ep_sparse_ext.e_len -
117 (lbn - path.ep_sparse_ext.e_blk) - 1;
119 *runb = lbn - path.ep_sparse_ext.e_blk;
121 if (path.ep_ext == NULL) {
126 *bnp = fsbtodb(fs, lbn - ep->e_blk +
127 (ep->e_start_lo | (daddr_t)ep->e_start_hi << 32));
133 *runp = ep->e_len - (lbn - ep->e_blk) - 1;
135 *runb = lbn - ep->e_blk;
139 if (path.ep_bp != NULL)
146 * Indirect blocks are now on the vnode for the file. They are given negative
147 * logical block numbers. Indirect blocks are addressed by the negative
148 * address of the first data block to which they point. Double indirect blocks
149 * are addressed by one less than the address of the first indirect block to
150 * which they point. Triple indirect blocks are addressed by one less than
151 * the address of the first double indirect block to which they point.
153 * ext2_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).
160 ext2_bmaparray(struct vnode *vp, daddr_t bn, daddr_t *bnp, int *runp, int *runb)
164 struct ext2mount *ump;
166 struct indir a[EXT2_NIADDR + 1], *ap;
169 int error, num, maxrun = 0, bsize;
177 bsize = EXT2_BLOCK_SIZE(ump->um_e2fs);
180 maxrun = mp->mnt_iosize_max / bsize - 1;
189 error = ext2_getlbns(vp, bn, ap, nump);
195 *bnp = blkptrtodb(ump, ip->i_db[bn]);
201 for (++bn; bn < EXT2_NDADDR && *runp < maxrun &&
202 is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]);
205 if (runb && (bn > 0)) {
206 for (--bn; (bn >= 0) && (*runb < maxrun) &&
207 is_sequential(ump, ip->i_db[bn],
215 /* Get disk address out of indirect block array */
216 daddr = ip->i_ib[ap->in_off];
218 for (bp = NULL, ++ap; --num; ++ap) {
220 * Exit the loop if there is no disk address assigned yet and
221 * the indirect block isn't in the cache, or if we were
222 * looking for an indirect block and we've found it.
225 metalbn = ap->in_lbn;
226 if ((daddr == 0 && !incore(&vp->v_bufobj, metalbn)) || metalbn == bn)
229 * If we get here, we've either got the block in the cache
230 * or we have a disk address for it, go fetch it.
235 bp = getblk(vp, metalbn, bsize, 0, 0, 0);
236 if ((bp->b_flags & B_CACHE) == 0) {
239 panic("ext2_bmaparray: indirect block not in cache");
241 bp->b_blkno = blkptrtodb(ump, daddr);
242 bp->b_iocmd = BIO_READ;
243 bp->b_flags &= ~B_INVAL;
244 bp->b_ioflags &= ~BIO_ERROR;
245 vfs_busy_pages(bp, 0);
246 bp->b_iooffset = dbtob(bp->b_blkno);
251 racct_add_buf(curproc, bp, 0);
252 PROC_UNLOCK(curproc);
255 curthread->td_ru.ru_inblock++;
263 daddr = ((e2fs_daddr_t *)bp->b_data)[ap->in_off];
264 if (num == 1 && daddr && runp) {
265 for (bn = ap->in_off + 1;
266 bn < MNINDIR(ump) && *runp < maxrun &&
268 ((e2fs_daddr_t *)bp->b_data)[bn - 1],
269 ((e2fs_daddr_t *)bp->b_data)[bn]);
273 for (--bn; bn >= 0 && *runb < maxrun &&
275 ((e2fs_daddr_t *)bp->b_data)[bn],
276 ((e2fs_daddr_t *)bp->b_data)[bn + 1]);
285 * Since this is FFS independent code, we are out of scope for the
286 * definitions of BLK_NOCOPY and BLK_SNAP, but we do know that they
287 * will fall in the range 1..um_seqinc, so we use that test and
288 * return a request for a zeroed out buffer if attempts are made
289 * to read a BLK_NOCOPY or BLK_SNAP block.
291 if ((ip->i_flags & SF_SNAPSHOT) && daddr > 0 && daddr < ump->um_seqinc) {
295 *bnp = blkptrtodb(ump, daddr);
303 * Create an array of logical block number/offset pairs which represent the
304 * path of indirect blocks required to access a data block. The first "pair"
305 * contains the logical block number of the appropriate single, double or
306 * triple indirect block and the offset into the inode indirect block array.
307 * Note, the logical block number of the inode single/double/triple indirect
308 * block appears twice in the array, once with the offset into the i_ib and
309 * once with the offset into the page itself.
312 ext2_getlbns(struct vnode *vp, daddr_t bn, struct indir *ap, int *nump)
315 e2fs_lbn_t metalbn, realbn;
316 struct ext2mount *ump;
317 int i, numlevels, off;
320 ump = VFSTOEXT2(vp->v_mount);
328 /* The first EXT2_NDADDR blocks are direct blocks. */
329 if (bn < EXT2_NDADDR)
333 * Determine the number of levels of indirection. After this loop
334 * is done, blockcnt indicates the number of data blocks possible
335 * at the previous level of indirection, and EXT2_NIADDR - i is the
336 * number of levels of indirection needed to locate the requested block.
338 for (blockcnt = 1, i = EXT2_NIADDR, bn -= EXT2_NDADDR; ;
339 i--, bn -= blockcnt) {
343 * Use int64_t's here to avoid overflow for triple indirect
344 * blocks when longs have 32 bits and the block size is more
347 qblockcnt = (int64_t)blockcnt * MNINDIR(ump);
350 blockcnt = qblockcnt;
353 /* Calculate the address of the first meta-block. */
355 metalbn = -(realbn - bn + EXT2_NIADDR - i);
357 metalbn = -(-realbn - bn + EXT2_NIADDR - i);
360 * At each iteration, off is the offset into the bap array which is
361 * an array of disk addresses at the current level of indirection.
362 * The logical block number and the offset in that block are stored
363 * into the argument array.
365 ap->in_lbn = metalbn;
366 ap->in_off = off = EXT2_NIADDR - i;
368 for (++numlevels; i <= EXT2_NIADDR; i++) {
369 /* If searching for a meta-data block, quit when found. */
370 if (metalbn == realbn)
373 off = (bn / blockcnt) % MNINDIR(ump);
376 ap->in_lbn = metalbn;
380 metalbn -= -1 + off * blockcnt;
381 blockcnt /= MNINDIR(ump);