ufs_bmap.c revision 284021
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 * 4. 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. 21 * 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 32 * SUCH DAMAGE. 33 * 34 * @(#)ufs_bmap.c 8.7 (Berkeley) 3/21/95 35 */ 36 37#include <sys/cdefs.h> 38__FBSDID("$FreeBSD: stable/10/sys/ufs/ufs/ufs_bmap.c 284021 2015-06-05 08:36:25Z kib $"); 39 40#include <sys/param.h> 41#include <sys/systm.h> 42#include <sys/bio.h> 43#include <sys/buf.h> 44#include <sys/proc.h> 45#include <sys/vnode.h> 46#include <sys/mount.h> 47#include <sys/resourcevar.h> 48#include <sys/stat.h> 49 50#include <ufs/ufs/extattr.h> 51#include <ufs/ufs/quota.h> 52#include <ufs/ufs/inode.h> 53#include <ufs/ufs/ufsmount.h> 54#include <ufs/ufs/ufs_extern.h> 55 56/* 57 * Bmap converts the logical block number of a file to its physical block 58 * number on the disk. The conversion is done by using the logical block 59 * number to index into the array of block pointers described by the dinode. 60 */ 61int 62ufs_bmap(ap) 63 struct vop_bmap_args /* { 64 struct vnode *a_vp; 65 daddr_t a_bn; 66 struct bufobj **a_bop; 67 daddr_t *a_bnp; 68 int *a_runp; 69 int *a_runb; 70 } */ *ap; 71{ 72 ufs2_daddr_t blkno; 73 int error; 74 75 /* 76 * Check for underlying vnode requests and ensure that logical 77 * to physical mapping is requested. 78 */ 79 if (ap->a_bop != NULL) 80 *ap->a_bop = &VTOI(ap->a_vp)->i_devvp->v_bufobj; 81 if (ap->a_bnp == NULL) 82 return (0); 83 84 error = ufs_bmaparray(ap->a_vp, ap->a_bn, &blkno, NULL, 85 ap->a_runp, ap->a_runb); 86 *ap->a_bnp = blkno; 87 return (error); 88} 89 90/* 91 * Indirect blocks are now on the vnode for the file. They are given negative 92 * logical block numbers. Indirect blocks are addressed by the negative 93 * address of the first data block to which they point. Double indirect blocks 94 * are addressed by one less than the address of the first indirect block to 95 * which they point. Triple indirect blocks are addressed by one less than 96 * the address of the first double indirect block to which they point. 97 * 98 * ufs_bmaparray does the bmap conversion, and if requested returns the 99 * array of logical blocks which must be traversed to get to a block. 100 * Each entry contains the offset into that block that gets you to the 101 * next block and the disk address of the block (if it is assigned). 102 */ 103 104int 105ufs_bmaparray(vp, bn, bnp, nbp, runp, runb) 106 struct vnode *vp; 107 ufs2_daddr_t bn; 108 ufs2_daddr_t *bnp; 109 struct buf *nbp; 110 int *runp; 111 int *runb; 112{ 113 struct inode *ip; 114 struct buf *bp; 115 struct ufsmount *ump; 116 struct mount *mp; 117 struct indir a[NIADDR+1], *ap; 118 ufs2_daddr_t daddr; 119 ufs_lbn_t metalbn; 120 int error, num, maxrun = 0; 121 int *nump; 122 123 ap = NULL; 124 ip = VTOI(vp); 125 mp = vp->v_mount; 126 ump = VFSTOUFS(mp); 127 128 if (runp) { 129 maxrun = mp->mnt_iosize_max / mp->mnt_stat.f_iosize - 1; 130 *runp = 0; 131 } 132 133 if (runb) { 134 *runb = 0; 135 } 136 137 138 ap = a; 139 nump = # 140 error = ufs_getlbns(vp, bn, ap, nump); 141 if (error) 142 return (error); 143 144 num = *nump; 145 if (num == 0) { 146 if (bn >= 0 && bn < NDADDR) { 147 *bnp = blkptrtodb(ump, DIP(ip, i_db[bn])); 148 } else if (bn < 0 && bn >= -NXADDR) { 149 *bnp = blkptrtodb(ump, ip->i_din2->di_extb[-1 - bn]); 150 if (*bnp == 0) 151 *bnp = -1; 152 if (nbp == NULL) 153 panic("ufs_bmaparray: mapping ext data"); 154 nbp->b_xflags |= BX_ALTDATA; 155 return (0); 156 } else { 157 panic("ufs_bmaparray: blkno out of range"); 158 } 159 /* 160 * Since this is FFS independent code, we are out of 161 * scope for the definitions of BLK_NOCOPY and 162 * BLK_SNAP, but we do know that they will fall in 163 * the range 1..um_seqinc, so we use that test and 164 * return a request for a zeroed out buffer if attempts 165 * are made to read a BLK_NOCOPY or BLK_SNAP block. 166 */ 167 if ((ip->i_flags & SF_SNAPSHOT) && DIP(ip, i_db[bn]) > 0 && 168 DIP(ip, i_db[bn]) < ump->um_seqinc) { 169 *bnp = -1; 170 } else if (*bnp == 0) { 171 if (ip->i_flags & SF_SNAPSHOT) 172 *bnp = blkptrtodb(ump, bn * ump->um_seqinc); 173 else 174 *bnp = -1; 175 } else if (runp) { 176 ufs2_daddr_t bnb = bn; 177 for (++bn; bn < NDADDR && *runp < maxrun && 178 is_sequential(ump, DIP(ip, i_db[bn - 1]), 179 DIP(ip, i_db[bn])); 180 ++bn, ++*runp); 181 bn = bnb; 182 if (runb && (bn > 0)) { 183 for (--bn; (bn >= 0) && (*runb < maxrun) && 184 is_sequential(ump, DIP(ip, i_db[bn]), 185 DIP(ip, i_db[bn+1])); 186 --bn, ++*runb); 187 } 188 } 189 return (0); 190 } 191 192 193 /* Get disk address out of indirect block array */ 194 daddr = DIP(ip, i_ib[ap->in_off]); 195 196 for (bp = NULL, ++ap; --num; ++ap) { 197 /* 198 * Exit the loop if there is no disk address assigned yet and 199 * the indirect block isn't in the cache, or if we were 200 * looking for an indirect block and we've found it. 201 */ 202 203 metalbn = ap->in_lbn; 204 if ((daddr == 0 && !incore(&vp->v_bufobj, metalbn)) || metalbn == bn) 205 break; 206 /* 207 * If we get here, we've either got the block in the cache 208 * or we have a disk address for it, go fetch it. 209 */ 210 if (bp) 211 bqrelse(bp); 212 213 bp = getblk(vp, metalbn, mp->mnt_stat.f_iosize, 0, 0, 0); 214 if ((bp->b_flags & B_CACHE) == 0) { 215#ifdef INVARIANTS 216 if (!daddr) 217 panic("ufs_bmaparray: indirect block not in cache"); 218#endif 219 bp->b_blkno = blkptrtodb(ump, daddr); 220 bp->b_iocmd = BIO_READ; 221 bp->b_flags &= ~B_INVAL; 222 bp->b_ioflags &= ~BIO_ERROR; 223 vfs_busy_pages(bp, 0); 224 bp->b_iooffset = dbtob(bp->b_blkno); 225 bstrategy(bp); 226 curthread->td_ru.ru_inblock++; 227 error = bufwait(bp); 228 if (error) { 229 brelse(bp); 230 return (error); 231 } 232 } 233 234 if (ip->i_ump->um_fstype == UFS1) { 235 daddr = ((ufs1_daddr_t *)bp->b_data)[ap->in_off]; 236 if (num == 1 && daddr && runp) { 237 for (bn = ap->in_off + 1; 238 bn < MNINDIR(ump) && *runp < maxrun && 239 is_sequential(ump, 240 ((ufs1_daddr_t *)bp->b_data)[bn - 1], 241 ((ufs1_daddr_t *)bp->b_data)[bn]); 242 ++bn, ++*runp); 243 bn = ap->in_off; 244 if (runb && bn) { 245 for (--bn; bn >= 0 && *runb < maxrun && 246 is_sequential(ump, 247 ((ufs1_daddr_t *)bp->b_data)[bn], 248 ((ufs1_daddr_t *)bp->b_data)[bn+1]); 249 --bn, ++*runb); 250 } 251 } 252 continue; 253 } 254 daddr = ((ufs2_daddr_t *)bp->b_data)[ap->in_off]; 255 if (num == 1 && daddr && runp) { 256 for (bn = ap->in_off + 1; 257 bn < MNINDIR(ump) && *runp < maxrun && 258 is_sequential(ump, 259 ((ufs2_daddr_t *)bp->b_data)[bn - 1], 260 ((ufs2_daddr_t *)bp->b_data)[bn]); 261 ++bn, ++*runp); 262 bn = ap->in_off; 263 if (runb && bn) { 264 for (--bn; bn >= 0 && *runb < maxrun && 265 is_sequential(ump, 266 ((ufs2_daddr_t *)bp->b_data)[bn], 267 ((ufs2_daddr_t *)bp->b_data)[bn + 1]); 268 --bn, ++*runb); 269 } 270 } 271 } 272 if (bp) 273 bqrelse(bp); 274 275 /* 276 * Since this is FFS independent code, we are out of scope for the 277 * definitions of BLK_NOCOPY and BLK_SNAP, but we do know that they 278 * will fall in the range 1..um_seqinc, so we use that test and 279 * return a request for a zeroed out buffer if attempts are made 280 * to read a BLK_NOCOPY or BLK_SNAP block. 281 */ 282 if ((ip->i_flags & SF_SNAPSHOT) && daddr > 0 && daddr < ump->um_seqinc){ 283 *bnp = -1; 284 return (0); 285 } 286 *bnp = blkptrtodb(ump, daddr); 287 if (*bnp == 0) { 288 if (ip->i_flags & SF_SNAPSHOT) 289 *bnp = blkptrtodb(ump, bn * ump->um_seqinc); 290 else 291 *bnp = -1; 292 } 293 return (0); 294} 295 296/* 297 * Create an array of logical block number/offset pairs which represent the 298 * path of indirect blocks required to access a data block. The first "pair" 299 * contains the logical block number of the appropriate single, double or 300 * triple indirect block and the offset into the inode indirect block array. 301 * Note, the logical block number of the inode single/double/triple indirect 302 * block appears twice in the array, once with the offset into the i_ib and 303 * once with the offset into the page itself. 304 */ 305int 306ufs_getlbns(vp, bn, ap, nump) 307 struct vnode *vp; 308 ufs2_daddr_t bn; 309 struct indir *ap; 310 int *nump; 311{ 312 ufs2_daddr_t blockcnt; 313 ufs_lbn_t metalbn, realbn; 314 struct ufsmount *ump; 315 int i, numlevels, off; 316 317 ump = VFSTOUFS(vp->v_mount); 318 if (nump) 319 *nump = 0; 320 numlevels = 0; 321 realbn = bn; 322 if (bn < 0) 323 bn = -bn; 324 325 /* The first NDADDR blocks are direct blocks. */ 326 if (bn < NDADDR) 327 return (0); 328 329 /* 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 NIADDR - i is the number 333 * of levels of indirection needed to locate the requested block. 334 */ 335 for (blockcnt = 1, i = NIADDR, bn -= NDADDR;; i--, bn -= blockcnt) { 336 if (i == 0) 337 return (EFBIG); 338 blockcnt *= MNINDIR(ump); 339 if (bn < blockcnt) 340 break; 341 } 342 343 /* Calculate the address of the first meta-block. */ 344 if (realbn >= 0) 345 metalbn = -(realbn - bn + NIADDR - i); 346 else 347 metalbn = -(-realbn - bn + NIADDR - i); 348 349 /* 350 * At each iteration, off is the offset into the bap array which is 351 * an array of disk addresses at the current level of indirection. 352 * The logical block number and the offset in that block are stored 353 * into the argument array. 354 */ 355 ap->in_lbn = metalbn; 356 ap->in_off = off = NIADDR - i; 357 ap++; 358 for (++numlevels; i <= NIADDR; i++) { 359 /* If searching for a meta-data block, quit when found. */ 360 if (metalbn == realbn) 361 break; 362 363 blockcnt /= MNINDIR(ump); 364 off = (bn / blockcnt) % MNINDIR(ump); 365 366 ++numlevels; 367 ap->in_lbn = metalbn; 368 ap->in_off = off; 369 ++ap; 370 371 metalbn -= -1 + off * blockcnt; 372 } 373 if (nump) 374 *nump = numlevels; 375 return (0); 376} 377