sys/vfs/gnu/ext2fs/ext2_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: src/sys/ufs/ufs/ufs_bmap.c,v 1.34.2.1 2000/03/17 10:12:14 ps Exp $
  40  * $DragonFly: src/sys/vfs/gnu/ext2fs/ext2_bmap.c,v 1.4 2007/08/13 17:31:56 dillon Exp $
  41  */
  42
  43 #include <sys/param.h>
  44 #include <sys/systm.h>
  45 #include <sys/buf.h>
  46 #include <sys/proc.h>
  47 #include <sys/vnode.h>
  48 #include <sys/mount.h>
  49 #include <sys/resourcevar.h>
  50 #include <sys/conf.h>
  51
  52 #include "quota.h"
  53 #include "dinode.h"
  54 #include "inode.h"
  55 #include "ext2_fs_sb.h"
  56 #include "ext2mount.h"
  57 #include "ext2_extern.h"
  58 #include "fs.h"
  59
  60 static int ext2_bmaparray(struct vnode *vp, ext2_daddr_t bn,
  61                           ext2_daddr_t *bnp, struct indir *ap, int *nump,
  62                           int *runp, int *runb);
  63
  64 /*
  65  * Bmap converts the logical block number of a file to its physical block
  66  * number on the disk. The conversion is done by using the logical block
  67  * number to index into the array of block pointers described by the dinode.
  68  *
  69  * BMAP must return the contiguous before and after run in bytes, inclusive
  70  * of the returned block.
  71  *
  72  * ext2_bmap(struct vnode *a_vp, off_t a_loffset,
  73  *          off_t *a_doffsetp, int *a_runp, int *a_runb)
  74  */
  75 int
  76 ext2_bmap(struct vop_bmap_args *ap)
  77 {
  78         struct ext2_sb_info *fs;
  79         ext2_daddr_t lbn;
  80         ext2_daddr_t dbn;
  81         int error;
  82
  83         /*
  84          * Check for underlying vnode requests and ensure that logical
  85          * to physical mapping is requested.
  86          */
  87         if (ap->a_doffsetp == NULL)
  88                 return (0);
  89
  90         fs = VTOI(ap->a_vp)->i_e2fs;
  91         KKASSERT(((int)ap->a_loffset & ((1 << fs->s_bshift) - 1)) == 0);
  92         lbn = ap->a_loffset >> fs->s_bshift;
  93
  94         error = ext2_bmaparray(ap->a_vp, lbn, &dbn, NULL, NULL,
  95                               ap->a_runp, ap->a_runb);
  96
  97         if (error || dbn == (ext2_daddr_t)-1) {
  98                 *ap->a_doffsetp = NOOFFSET;
  99         } else {
 100                 *ap->a_doffsetp = dbtodoff(fs, dbn);
 101                 if (ap->a_runp)
 102                         *ap->a_runp = (*ap->a_runp + 1) << fs->s_bshift;
 103                 if (ap->a_runb)
 104                         *ap->a_runb = *ap->a_runb << fs->s_bshift;
 105         }
 106         return (error);
 107 }
 108
 109 /*
 110  * Indirect blocks are now on the vnode for the file.  They are given negative
 111  * logical block numbers.  Indirect blocks are addressed by the negative
 112  * address of the first data block to which they point.  Double indirect blocks
 113  * are addressed by one less than the address of the first indirect block to
 114  * which they point.  Triple indirect blocks are addressed by one less than
 115  * the address of the first double indirect block to which they point.
 116  *
 117  * ext2_bmaparray does the bmap conversion, and if requested returns the
 118  * array of logical blocks which must be traversed to get to a block.
 119  * Each entry contains the offset into that block that gets you to the
 120  * next block and the disk address of the block (if it is assigned).
 121  */
 122 static
 123 int
 124 ext2_bmaparray(struct vnode *vp, ext2_daddr_t bn, ext2_daddr_t *bnp,
 125               struct indir *ap, int *nump, int *runp, int *runb)
 126 {
 127         struct inode *ip;
 128         struct buf *bp;
 129         struct ext2mount *ump;
 130         struct mount *mp;
 131         struct ext2_sb_info *fs;
 132         struct indir a[NIADDR+1], *xap;
 133         ext2_daddr_t daddr;
 134         long metalbn;
 135         int error, maxrun, num;
 136
 137         ip = VTOI(vp);
 138         mp = vp->v_mount;
 139         ump = VFSTOEXT2(mp);
 140         fs = ip->i_e2fs;
 141 #ifdef DIAGNOSTIC
 142         if ((ap != NULL && nump == NULL) || (ap == NULL && nump != NULL))
 143                 panic("ext2_bmaparray: invalid arguments");
 144 #endif
 145
 146         if (runp) {
 147                 *runp = 0;
 148         }
 149
 150         if (runb) {
 151                 *runb = 0;
 152         }
 153
 154         maxrun = mp->mnt_iosize_max / mp->mnt_stat.f_iosize - 1;
 155
 156         xap = ap == NULL ? a : ap;
 157         if (!nump)
 158                 nump = &num;
 159         error = ext2_getlbns(vp, bn, xap, nump);
 160         if (error)
 161                 return (error);
 162
 163         num = *nump;
 164         if (num == 0) {
 165                 *bnp = blkptrtodb(ump, ip->i_db[bn]);
 166                 if (*bnp == 0)
 167                         *bnp = -1;
 168                 else if (runp) {
 169                         daddr_t bnb = bn;
 170                         for (++bn; bn < NDADDR && *runp < maxrun &&
 171                             is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]);
 172                             ++bn, ++*runp);
 173                         bn = bnb;
 174                         if (runb && (bn > 0)) {
 175                                 for (--bn; (bn >= 0) && (*runb < maxrun) &&
 176                                         is_sequential(ump, ip->i_db[bn],
 177                                                 ip->i_db[bn+1]);
 178                                                 --bn, ++*runb);
 179                         }
 180                 }
 181                 return (0);
 182         }
 183
 184
 185         /* Get disk address out of indirect block array */
 186         daddr = ip->i_ib[xap->in_off];
 187
 188         for (bp = NULL, ++xap; --num; ++xap) {
 189                 /*
 190                  * Exit the loop if there is no disk address assigned yet and
 191                  * the indirect block isn't in the cache, or if we were
 192                  * looking for an indirect block and we've found it.
 193                  */
 194
 195                 metalbn = xap->in_lbn;
 196                 if ((daddr == 0 &&
 197                      !findblk(vp, dbtodoff(fs, metalbn), FINDBLK_TEST)) ||
 198                     metalbn == bn) {
 199                         break;
 200                 }
 201                 /*
 202                  * If we get here, we've either got the block in the cache
 203                  * or we have a disk address for it, go fetch it.
 204                  */
 205                 if (bp)
 206                         bqrelse(bp);
 207
 208                 xap->in_exists = 1;
 209                 bp = getblk(vp, lblktodoff(fs, metalbn),
 210                             mp->mnt_stat.f_iosize, 0, 0);
 211                 if ((bp->b_flags & B_CACHE) == 0) {
 212 #ifdef DIAGNOSTIC
 213                         if (!daddr)
 214                                 panic("ext2_bmaparray: indirect block not in cache");
 215 #endif
 216                         /*
 217                          * This runs through ext2_strategy using bio2 to
 218                          * cache the disk offset, then comes back through
 219                          * bio1.  So we want to wait on bio1
 220                          */
 221                         bp->b_bio1.bio_done = biodone_sync;
 222                         bp->b_bio1.bio_flags |= BIO_SYNC;
 223                         bp->b_bio2.bio_offset = fsbtodoff(fs, daddr);
 224                         bp->b_flags &= ~(B_INVAL|B_ERROR);
 225                         bp->b_cmd = BUF_CMD_READ;
 226                         vfs_busy_pages(bp->b_vp, bp);
 227                         vn_strategy(bp->b_vp, &bp->b_bio1);
 228                         error = biowait(&bp->b_bio1, "biord");
 229                         if (error) {
 230                                 brelse(bp);
 231                                 return (error);
 232                         }
 233                 }
 234
 235                 daddr = ((ext2_daddr_t *)bp->b_data)[xap->in_off];
 236                 if (num == 1 && daddr && runp) {
 237                         for (bn = xap->in_off + 1;
 238                             bn < MNINDIR(ump) && *runp < maxrun &&
 239                             is_sequential(ump,
 240                             ((ext2_daddr_t *)bp->b_data)[bn - 1],
 241                             ((ext2_daddr_t *)bp->b_data)[bn]);
 242                             ++bn, ++*runp);
 243                         bn = xap->in_off;
 244                         if (runb && bn) {
 245                                 for(--bn; bn >= 0 && *runb < maxrun &&
 246                                         is_sequential(ump, ((daddr_t *)bp->b_data)[bn],
 247                                             ((daddr_t *)bp->b_data)[bn+1]);
 248                                         --bn, ++*runb);
 249                         }
 250                 }
 251         }
 252         if (bp)
 253                 bqrelse(bp);
 254
 255         daddr = blkptrtodb(ump, daddr);
 256         *bnp = daddr == 0 ? -1 : daddr;
 257         return (0);
 258 }
 259
 260 /*
 261  * Create an array of logical block number/offset pairs which represent the
 262  * path of indirect blocks required to access a data block.  The first "pair"
 263  * contains the logical block number of the appropriate single, double or
 264  * triple indirect block and the offset into the inode indirect block array.
 265  * Note, the logical block number of the inode single/double/triple indirect
 266  * block appears twice in the array, once with the offset into the i_ib and
 267  * once with the offset into the page itself.
 268  */
 269 int
 270 ext2_getlbns(struct vnode *vp, ext2_daddr_t bn, struct indir *ap, int *nump)
 271 {
 272         long blockcnt, metalbn, realbn;
 273         struct ext2mount *ump;
 274         int i, numlevels, off;
 275         int64_t qblockcnt;
 276
 277         ump = VFSTOEXT2(vp->v_mount);
 278         if (nump)
 279                 *nump = 0;
 280         numlevels = 0;
 281         realbn = bn;
 282         if ((long)bn < 0)
 283                 bn = -(long)bn;
 284
 285         /* The first NDADDR blocks are direct blocks. */
 286         if (bn < NDADDR)
 287                 return (0);
 288
 289         /*
 290          * Determine the number of levels of indirection.  After this loop
 291          * is done, blockcnt indicates the number of data blocks possible
 292          * at the previous level of indirection, and NIADDR - i is the number
 293          * of levels of indirection needed to locate the requested block.
 294          */
 295         for (blockcnt = 1, i = NIADDR, bn -= NDADDR;; i--, bn -= blockcnt) {
 296                 if (i == 0)
 297                         return (EFBIG);
 298                 /*
 299                  * Use int64_t's here to avoid overflow for triple indirect
 300                  * blocks when longs have 32 bits and the block size is more
 301                  * than 4K.
 302                  */
 303                 qblockcnt = (int64_t)blockcnt * MNINDIR(ump);
 304                 if (bn < qblockcnt)
 305                         break;
 306                 blockcnt = qblockcnt;
 307         }
 308
 309         /* Calculate the address of the first meta-block. */
 310         if (realbn >= 0)
 311                 metalbn = -(realbn - bn + NIADDR - i);
 312         else
 313                 metalbn = -(-realbn - bn + NIADDR - i);
 314
 315         /*
 316          * At each iteration, off is the offset into the bap array which is
 317          * an array of disk addresses at the current level of indirection.
 318          * The logical block number and the offset in that block are stored
 319          * into the argument array.
 320          */
 321         ap->in_lbn = metalbn;
 322         ap->in_off = off = NIADDR - i;
 323         ap->in_exists = 0;
 324         ap++;
 325         for (++numlevels; i <= NIADDR; i++) {
 326                 /* If searching for a meta-data block, quit when found. */
 327                 if (metalbn == realbn)
 328                         break;
 329
 330                 off = (bn / blockcnt) % MNINDIR(ump);
 331
 332                 ++numlevels;
 333                 ap->in_lbn = metalbn;
 334                 ap->in_off = off;
 335                 ap->in_exists = 0;
 336                 ++ap;
 337
 338                 metalbn -= -1 + off * blockcnt;
 339                 blockcnt /= MNINDIR(ump);
 340         }
 341         if (nump)
 342                 *nump = numlevels;
 343         return (0);
 344 }