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/ufs/ufs_bmap.c,v 1.3 2003/06/26 20:27:53 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 <ufs/ufs/quota.h> 53 #include <ufs/ufs/inode.h> 54 #include <ufs/ufs/ufsmount.h> 55 #include <ufs/ufs/ufs_extern.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, num; 119 120 ip = VTOI(vp); 121 mp = vp->v_mount; 122 ump = VFSTOUFS(mp); 123 devvp = ump->um_devvp; 124 #ifdef DIAGNOSTIC 125 if ((ap != NULL && nump == NULL) || (ap == NULL && nump != NULL)) 126 panic("ufs_bmaparray: invalid arguments"); 127 #endif 128 129 if (runp) { 130 *runp = 0; 131 } 132 133 if (runb) { 134 *runb = 0; 135 } 136 137 maxrun = mp->mnt_iosize_max / mp->mnt_stat.f_iosize - 1; 138 139 xap = ap == NULL ? a : ap; 140 if (!nump) 141 nump = # 142 error = ufs_getlbns(vp, bn, xap, nump); 143 if (error) 144 return (error); 145 146 num = *nump; 147 if (num == 0) { 148 *bnp = blkptrtodb(ump, ip->i_db[bn]); 149 if (*bnp == 0) 150 *bnp = -1; 151 else if (runp) { 152 daddr_t bnb = bn; 153 for (++bn; bn < NDADDR && *runp < maxrun && 154 is_sequential(ump, ip->i_db[bn - 1], ip->i_db[bn]); 155 ++bn, ++*runp); 156 bn = bnb; 157 if (runb && (bn > 0)) { 158 for (--bn; (bn >= 0) && (*runb < maxrun) && 159 is_sequential(ump, ip->i_db[bn], 160 ip->i_db[bn+1]); 161 --bn, ++*runb); 162 } 163 } 164 return (0); 165 } 166 167 168 /* Get disk address out of indirect block array */ 169 daddr = ip->i_ib[xap->in_off]; 170 171 for (bp = NULL, ++xap; --num; ++xap) { 172 /* 173 * Exit the loop if there is no disk address assigned yet and 174 * the indirect block isn't in the cache, or if we were 175 * looking for an indirect block and we've found it. 176 */ 177 178 metalbn = xap->in_lbn; 179 if ((daddr == 0 && !incore(vp, metalbn)) || metalbn == bn) 180 break; 181 /* 182 * If we get here, we've either got the block in the cache 183 * or we have a disk address for it, go fetch it. 184 */ 185 if (bp) 186 bqrelse(bp); 187 188 xap->in_exists = 1; 189 bp = getblk(vp, metalbn, mp->mnt_stat.f_iosize, 0, 0); 190 if ((bp->b_flags & B_CACHE) == 0) { 191 #ifdef DIAGNOSTIC 192 if (!daddr) 193 panic("ufs_bmaparray: indirect block not in cache"); 194 #endif 195 bp->b_blkno = blkptrtodb(ump, daddr); 196 bp->b_flags |= B_READ; 197 bp->b_flags &= ~(B_INVAL|B_ERROR); 198 vfs_busy_pages(bp, 0); 199 VOP_STRATEGY(bp->b_vp, bp); 200 error = biowait(bp); 201 if (error) { 202 brelse(bp); 203 return (error); 204 } 205 } 206 207 daddr = ((ufs_daddr_t *)bp->b_data)[xap->in_off]; 208 if (num == 1 && daddr && runp) { 209 for (bn = xap->in_off + 1; 210 bn < MNINDIR(ump) && *runp < maxrun && 211 is_sequential(ump, 212 ((ufs_daddr_t *)bp->b_data)[bn - 1], 213 ((ufs_daddr_t *)bp->b_data)[bn]); 214 ++bn, ++*runp); 215 bn = xap->in_off; 216 if (runb && bn) { 217 for(--bn; bn >= 0 && *runb < maxrun && 218 is_sequential(ump, ((daddr_t *)bp->b_data)[bn], 219 ((daddr_t *)bp->b_data)[bn+1]); 220 --bn, ++*runb); 221 } 222 } 223 } 224 if (bp) 225 bqrelse(bp); 226 227 daddr = blkptrtodb(ump, daddr); 228 *bnp = daddr == 0 ? -1 : daddr; 229 return (0); 230 } 231 232 /* 233 * Create an array of logical block number/offset pairs which represent the 234 * path of indirect blocks required to access a data block. The first "pair" 235 * contains the logical block number of the appropriate single, double or 236 * triple indirect block and the offset into the inode indirect block array. 237 * Note, the logical block number of the inode single/double/triple indirect 238 * block appears twice in the array, once with the offset into the i_ib and 239 * once with the offset into the page itself. 240 */ 241 int 242 ufs_getlbns(vp, bn, ap, nump) 243 struct vnode *vp; 244 ufs_daddr_t bn; 245 struct indir *ap; 246 int *nump; 247 { 248 long blockcnt, metalbn, realbn; 249 struct ufsmount *ump; 250 int i, numlevels, off; 251 int64_t qblockcnt; 252 253 ump = VFSTOUFS(vp->v_mount); 254 if (nump) 255 *nump = 0; 256 numlevels = 0; 257 realbn = bn; 258 if ((long)bn < 0) 259 bn = -(long)bn; 260 261 /* The first NDADDR blocks are direct blocks. */ 262 if (bn < NDADDR) 263 return (0); 264 265 /* 266 * Determine the number of levels of indirection. After this loop 267 * is done, blockcnt indicates the number of data blocks possible 268 * at the previous level of indirection, and NIADDR - i is the number 269 * of levels of indirection needed to locate the requested block. 270 */ 271 for (blockcnt = 1, i = NIADDR, bn -= NDADDR;; i--, bn -= blockcnt) { 272 if (i == 0) 273 return (EFBIG); 274 /* 275 * Use int64_t's here to avoid overflow for triple indirect 276 * blocks when longs have 32 bits and the block size is more 277 * than 4K. 278 */ 279 qblockcnt = (int64_t)blockcnt * MNINDIR(ump); 280 if (bn < qblockcnt) 281 break; 282 blockcnt = qblockcnt; 283 } 284 285 /* Calculate the address of the first meta-block. */ 286 if (realbn >= 0) 287 metalbn = -(realbn - bn + NIADDR - i); 288 else 289 metalbn = -(-realbn - bn + NIADDR - i); 290 291 /* 292 * At each iteration, off is the offset into the bap array which is 293 * an array of disk addresses at the current level of indirection. 294 * The logical block number and the offset in that block are stored 295 * into the argument array. 296 */ 297 ap->in_lbn = metalbn; 298 ap->in_off = off = NIADDR - i; 299 ap->in_exists = 0; 300 ap++; 301 for (++numlevels; i <= NIADDR; i++) { 302 /* If searching for a meta-data block, quit when found. */ 303 if (metalbn == realbn) 304 break; 305 306 off = (bn / blockcnt) % MNINDIR(ump); 307 308 ++numlevels; 309 ap->in_lbn = metalbn; 310 ap->in_off = off; 311 ap->in_exists = 0; 312 ++ap; 313 314 metalbn -= -1 + off * blockcnt; 315 blockcnt /= MNINDIR(ump); 316 } 317 if (nump) 318 *nump = numlevels; 319 return (0); 320 } 321