2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1989, 1991, 1993
5 * The Regents of the University of California. All rights reserved.
6 * (c) UNIX System Laboratories, Inc.
7 * All or some portions of this file are derived from material licensed
8 * to the University of California by American Telephone and Telegraph
9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10 * the permission of UNIX System Laboratories, Inc.
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * @(#)ufs_bmap.c 8.7 (Berkeley) 3/21/95
40 #include <sys/param.h>
41 #include <sys/systm.h>
45 #include <sys/vnode.h>
46 #include <sys/mount.h>
47 #include <sys/racct.h>
48 #include <sys/resourcevar.h>
51 #include <fs/ext2fs/fs.h>
52 #include <fs/ext2fs/inode.h>
53 #include <fs/ext2fs/ext2fs.h>
54 #include <fs/ext2fs/ext2_dinode.h>
55 #include <fs/ext2fs/ext2_extern.h>
56 #include <fs/ext2fs/ext2_mount.h>
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_header *ehp;
98 struct ext4_extent *ep;
99 struct ext4_extent_path *path = NULL;
106 ehp = (struct ext4_extent_header *)ip->i_data;
107 depth = ehp->eh_depth;
115 error = ext4_ext_find_extent(ip, lbn, &path);
119 ep = path[depth].ep_ext;
121 if (lbn < ep->e_blk) {
123 *runp = ep->e_blk - lbn - 1;
124 } else if (ep->e_blk <= lbn && lbn < ep->e_blk + ep->e_len) {
125 *bnp = fsbtodb(fs, lbn - ep->e_blk +
126 (ep->e_start_lo | (daddr_t)ep->e_start_hi << 32));
128 *runp = ep->e_len - (lbn - ep->e_blk) - 1;
130 *runb = lbn - ep->e_blk;
133 *runb = ep->e_blk + lbn - ep->e_len;
137 ext4_ext_path_free(path);
143 * Indirect blocks are now on the vnode for the file. They are given negative
144 * logical block numbers. Indirect blocks are addressed by the negative
145 * address of the first data block to which they point. Double indirect blocks
146 * are addressed by one less than the address of the first indirect block to
147 * which they point. Triple indirect blocks are addressed by one less than
148 * the address of the first double indirect block to which they point.
150 * ext2_bmaparray does the bmap conversion, and if requested returns the
151 * array of logical blocks which must be traversed to get to a block.
152 * Each entry contains the offset into that block that gets you to the
153 * next block and the disk address of the block (if it is assigned).
157 ext2_bmaparray(struct vnode *vp, daddr_t bn, daddr_t *bnp, int *runp, int *runb)
161 struct ext2mount *ump;
163 struct indir a[EXT2_NIADDR + 1], *ap;
166 int error, num, maxrun = 0, bsize;
174 bsize = EXT2_BLOCK_SIZE(ump->um_e2fs);
177 maxrun = mp->mnt_iosize_max / bsize - 1;
186 error = ext2_getlbns(vp, bn, ap, nump);
192 *bnp = blkptrtodb(ump, ip->i_db[bn]);
198 for (++bn; bn < EXT2_NDADDR && *runp < maxrun &&
199 is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]);
202 if (runb && (bn > 0)) {
203 for (--bn; (bn >= 0) && (*runb < maxrun) &&
204 is_sequential(ump, ip->i_db[bn],
212 /* Get disk address out of indirect block array */
213 daddr = ip->i_ib[ap->in_off];
215 for (bp = NULL, ++ap; --num; ++ap) {
217 * Exit the loop if there is no disk address assigned yet and
218 * the indirect block isn't in the cache, or if we were
219 * looking for an indirect block and we've found it.
222 metalbn = ap->in_lbn;
223 if ((daddr == 0 && !incore(&vp->v_bufobj, metalbn)) || metalbn == bn)
226 * If we get here, we've either got the block in the cache
227 * or we have a disk address for it, go fetch it.
232 bp = getblk(vp, metalbn, bsize, 0, 0, 0);
233 if ((bp->b_flags & B_CACHE) == 0) {
236 panic("ext2_bmaparray: indirect block not in cache");
238 bp->b_blkno = blkptrtodb(ump, daddr);
239 bp->b_iocmd = BIO_READ;
240 bp->b_flags &= ~B_INVAL;
241 bp->b_ioflags &= ~BIO_ERROR;
242 vfs_busy_pages(bp, 0);
243 bp->b_iooffset = dbtob(bp->b_blkno);
248 racct_add_buf(curproc, bp, 0);
249 PROC_UNLOCK(curproc);
252 curthread->td_ru.ru_inblock++;
260 daddr = ((e2fs_daddr_t *)bp->b_data)[ap->in_off];
261 if (num == 1 && daddr && runp) {
262 for (bn = ap->in_off + 1;
263 bn < MNINDIR(ump) && *runp < maxrun &&
265 ((e2fs_daddr_t *)bp->b_data)[bn - 1],
266 ((e2fs_daddr_t *)bp->b_data)[bn]);
270 for (--bn; bn >= 0 && *runb < maxrun &&
272 ((e2fs_daddr_t *)bp->b_data)[bn],
273 ((e2fs_daddr_t *)bp->b_data)[bn + 1]);
282 * Since this is FFS independent code, we are out of scope for the
283 * definitions of BLK_NOCOPY and BLK_SNAP, but we do know that they
284 * will fall in the range 1..um_seqinc, so we use that test and
285 * return a request for a zeroed out buffer if attempts are made
286 * to read a BLK_NOCOPY or BLK_SNAP block.
288 if ((ip->i_flags & SF_SNAPSHOT) && daddr > 0 && daddr < ump->um_seqinc) {
292 *bnp = blkptrtodb(ump, daddr);
300 * Create an array of logical block number/offset pairs which represent the
301 * path of indirect blocks required to access a data block. The first "pair"
302 * contains the logical block number of the appropriate single, double or
303 * triple indirect block and the offset into the inode indirect block array.
304 * Note, the logical block number of the inode single/double/triple indirect
305 * block appears twice in the array, once with the offset into the i_ib and
306 * once with the offset into the page itself.
309 ext2_getlbns(struct vnode *vp, daddr_t bn, struct indir *ap, int *nump)
312 e2fs_lbn_t metalbn, realbn;
313 struct ext2mount *ump;
314 int i, numlevels, off;
317 ump = VFSTOEXT2(vp->v_mount);
325 /* The first EXT2_NDADDR blocks are direct blocks. */
326 if (bn < EXT2_NDADDR)
330 * Determine the number of levels of indirection. After this loop
331 * is done, blockcnt indicates the number of data blocks possible
332 * at the previous level of indirection, and EXT2_NIADDR - i is the
333 * number of levels of indirection needed to locate the requested block.
335 for (blockcnt = 1, i = EXT2_NIADDR, bn -= EXT2_NDADDR; ;
336 i--, bn -= blockcnt) {
340 * Use int64_t's here to avoid overflow for triple indirect
341 * blocks when longs have 32 bits and the block size is more
344 qblockcnt = (int64_t)blockcnt * MNINDIR(ump);
347 blockcnt = qblockcnt;
350 /* Calculate the address of the first meta-block. */
352 metalbn = -(realbn - bn + EXT2_NIADDR - i);
354 metalbn = -(-realbn - bn + EXT2_NIADDR - i);
357 * At each iteration, off is the offset into the bap array which is
358 * an array of disk addresses at the current level of indirection.
359 * The logical block number and the offset in that block are stored
360 * into the argument array.
362 ap->in_lbn = metalbn;
363 ap->in_off = off = EXT2_NIADDR - i;
365 for (++numlevels; i <= EXT2_NIADDR; i++) {
366 /* If searching for a meta-data block, quit when found. */
367 if (metalbn == realbn)
370 off = (bn / blockcnt) % MNINDIR(ump);
373 ap->in_lbn = metalbn;
377 metalbn -= -1 + off * blockcnt;
378 blockcnt /= MNINDIR(ump);