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 vnode *devvp;
 132         struct ext2_sb_info *fs;
 133         struct indir a[NIADDR+1], *xap;
 134         ext2_daddr_t daddr;
 135         long metalbn;
 136         int error, maxrun, num;
 137
 138         ip = VTOI(vp);
 139         mp = vp->v_mount;
 140         ump = VFSTOEXT2(mp);
 141         devvp = ump->um_devvp;
 142         fs = ip->i_e2fs;
 143 #ifdef DIAGNOSTIC
 144         if ((ap != NULL && nump == NULL) || (ap == NULL && nump != NULL))
 145                 panic("ext2_bmaparray: invalid arguments");
 146 #endif
 147
 148         if (runp) {
 149                 *runp = 0;
 150         }
 151
 152         if (runb) {
 153                 *runb = 0;
 154         }
 155
 156         maxrun = mp->mnt_iosize_max / mp->mnt_stat.f_iosize - 1;
 157
 158         xap = ap == NULL ? a : ap;
 159         if (!nump)
 160                 nump = &num;
 161         error = ext2_getlbns(vp, bn, xap, nump);
 162         if (error)
 163                 return (error);
 164
 165         num = *nump;
 166         if (num == 0) {
 167                 *bnp = blkptrtodb(ump, ip->i_db[bn]);
 168                 if (*bnp == 0)
 169                         *bnp = -1;
 170                 else if (runp) {
 171                         daddr_t bnb = bn;
 172                         for (++bn; bn < NDADDR && *runp < maxrun &&
 173                             is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]);
 174                             ++bn, ++*runp);
 175                         bn = bnb;
 176                         if (runb && (bn > 0)) {
 177                                 for (--bn; (bn >= 0) && (*runb < maxrun) &&
 178                                         is_sequential(ump, ip->i_db[bn],
 179                                                 ip->i_db[bn+1]);
 180                                                 --bn, ++*runb);
 181                         }
 182                 }
 183                 return (0);
 184         }
 185
 186
 187         /* Get disk address out of indirect block array */
 188         daddr = ip->i_ib[xap->in_off];
 189
 190         for (bp = NULL, ++xap; --num; ++xap) {
 191                 /*
 192                  * Exit the loop if there is no disk address assigned yet and
 193                  * the indirect block isn't in the cache, or if we were
 194                  * looking for an indirect block and we've found it.
 195                  */
 196
 197                 metalbn = xap->in_lbn;
 198                 if ((daddr == 0 &&
 199                      !findblk(vp, dbtodoff(fs, metalbn), FINDBLK_TEST)) ||
 200                     metalbn == bn) {
 201                         break;
 202                 }
 203                 /*
 204                  * If we get here, we've either got the block in the cache
 205                  * or we have a disk address for it, go fetch it.
 206                  */
 207                 if (bp)
 208                         bqrelse(bp);
 209
 210                 xap->in_exists = 1;
 211                 bp = getblk(vp, lblktodoff(fs, metalbn),
 212                             mp->mnt_stat.f_iosize, 0, 0);
 213                 if ((bp->b_flags & B_CACHE) == 0) {
 214 #ifdef DIAGNOSTIC
 215                         if (!daddr)
 216                                 panic("ext2_bmaparray: indirect block not in cache");
 217 #endif
 218                         /*
 219                          * This runs through ext2_strategy using bio2 to
 220                          * cache the disk offset, then comes back through
 221                          * bio1.  So we want to wait on bio1
 222                          */
 223                         bp->b_bio1.bio_done = biodone_sync;
 224                         bp->b_bio1.bio_flags |= BIO_SYNC;
 225                         bp->b_bio2.bio_offset = fsbtodoff(fs, daddr);
 226                         bp->b_flags &= ~(B_INVAL|B_ERROR);
 227                         bp->b_cmd = BUF_CMD_READ;
 228                         vfs_busy_pages(bp->b_vp, bp);
 229                         vn_strategy(bp->b_vp, &bp->b_bio1);
 230                         error = biowait(&bp->b_bio1, "biord");
 231                         if (error) {
 232                                 brelse(bp);
 233                                 return (error);
 234                         }
 235                 }
 236
 237                 daddr = ((ext2_daddr_t *)bp->b_data)[xap->in_off];
 238                 if (num == 1 && daddr && runp) {
 239                         for (bn = xap->in_off + 1;
 240                             bn < MNINDIR(ump) && *runp < maxrun &&
 241                             is_sequential(ump,
 242                             ((ext2_daddr_t *)bp->b_data)[bn - 1],
 243                             ((ext2_daddr_t *)bp->b_data)[bn]);
 244                             ++bn, ++*runp);
 245                         bn = xap->in_off;
 246                         if (runb && bn) {
 247                                 for(--bn; bn >= 0 && *runb < maxrun &&
 248                                         is_sequential(ump, ((daddr_t *)bp->b_data)[bn],
 249                                             ((daddr_t *)bp->b_data)[bn+1]);
 250                                         --bn, ++*runb);
 251                         }
 252                 }
 253         }
 254         if (bp)
 255                 bqrelse(bp);
 256
 257         daddr = blkptrtodb(ump, daddr);
 258         *bnp = daddr == 0 ? -1 : daddr;
 259         return (0);
 260 }
 261
 262 /*
 263  * Create an array of logical block number/offset pairs which represent the
 264  * path of indirect blocks required to access a data block.  The first "pair"
 265  * contains the logical block number of the appropriate single, double or
 266  * triple indirect block and the offset into the inode indirect block array.
 267  * Note, the logical block number of the inode single/double/triple indirect
 268  * block appears twice in the array, once with the offset into the i_ib and
 269  * once with the offset into the page itself.
 270  */
 271 int
 272 ext2_getlbns(struct vnode *vp, ext2_daddr_t bn, struct indir *ap, int *nump)
 273 {
 274         long blockcnt, metalbn, realbn;
 275         struct ext2mount *ump;
 276         int i, numlevels, off;
 277         int64_t qblockcnt;
 278
 279         ump = VFSTOEXT2(vp->v_mount);
 280         if (nump)
 281                 *nump = 0;
 282         numlevels = 0;
 283         realbn = bn;
 284         if ((long)bn < 0)
 285                 bn = -(long)bn;
 286
 287         /* The first NDADDR blocks are direct blocks. */
 288         if (bn < NDADDR)
 289                 return (0);
 290
 291         /*
 292          * Determine the number of levels of indirection.  After this loop
 293          * is done, blockcnt indicates the number of data blocks possible
 294          * at the previous level of indirection, and NIADDR - i is the number
 295          * of levels of indirection needed to locate the requested block.
 296          */
 297         for (blockcnt = 1, i = NIADDR, bn -= NDADDR;; i--, bn -= blockcnt) {
 298                 if (i == 0)
 299                         return (EFBIG);
 300                 /*
 301                  * Use int64_t's here to avoid overflow for triple indirect
 302                  * blocks when longs have 32 bits and the block size is more
 303                  * than 4K.
 304                  */
 305                 qblockcnt = (int64_t)blockcnt * MNINDIR(ump);
 306                 if (bn < qblockcnt)
 307                         break;
 308                 blockcnt = qblockcnt;
 309         }
 310
 311         /* Calculate the address of the first meta-block. */
 312         if (realbn >= 0)
 313                 metalbn = -(realbn - bn + NIADDR - i);
 314         else
 315                 metalbn = -(-realbn - bn + NIADDR - i);
 316
 317         /*
 318          * At each iteration, off is the offset into the bap array which is
 319          * an array of disk addresses at the current level of indirection.
 320          * The logical block number and the offset in that block are stored
 321          * into the argument array.
 322          */
 323         ap->in_lbn = metalbn;
 324         ap->in_off = off = NIADDR - i;
 325         ap->in_exists = 0;
 326         ap++;
 327         for (++numlevels; i <= NIADDR; i++) {
 328                 /* If searching for a meta-data block, quit when found. */
 329                 if (metalbn == realbn)
 330                         break;
 331
 332                 off = (bn / blockcnt) % MNINDIR(ump);
 333
 334                 ++numlevels;
 335                 ap->in_lbn = metalbn;
 336                 ap->in_off = off;
 337                 ap->in_exists = 0;
 338                 ++ap;
 339
 340                 metalbn -= -1 + off * blockcnt;
 341                 blockcnt /= MNINDIR(ump);
 342         }
 343         if (nump)
 344                 *nump = numlevels;
 345         return (0);
 346 }