1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 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. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 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. 23 * 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 34 * SUCH DAMAGE. 35 * 36 * @(#)ufs_bmap.c 8.7 (Berkeley) 3/21/95 37 */ 38 39 #include <sys/cdefs.h> 40 __FBSDID("$FreeBSD$"); 41 42 #include <sys/param.h> 43 #include <sys/systm.h> 44 #include <sys/bio.h> 45 #include <sys/buf.h> 46 #include <sys/proc.h> 47 #include <sys/vnode.h> 48 #include <sys/mount.h> 49 #include <sys/racct.h> 50 #include <sys/resourcevar.h> 51 #include <sys/stat.h> 52 53 #include <ufs/ufs/extattr.h> 54 #include <ufs/ufs/quota.h> 55 #include <ufs/ufs/inode.h> 56 #include <ufs/ufs/ufsmount.h> 57 #include <ufs/ufs/ufs_extern.h> 58 59 /* 60 * Bmap converts the logical block number of a file to its physical block 61 * number on the disk. The conversion is done by using the logical block 62 * number to index into the array of block pointers described by the dinode. 63 */ 64 int 65 ufs_bmap(ap) 66 struct vop_bmap_args /* { 67 struct vnode *a_vp; 68 daddr_t a_bn; 69 struct bufobj **a_bop; 70 daddr_t *a_bnp; 71 int *a_runp; 72 int *a_runb; 73 } */ *ap; 74 { 75 ufs2_daddr_t blkno; 76 int error; 77 78 /* 79 * Check for underlying vnode requests and ensure that logical 80 * to physical mapping is requested. 81 */ 82 if (ap->a_bop != NULL) 83 *ap->a_bop = &VFSTOUFS(ap->a_vp->v_mount)->um_devvp->v_bufobj; 84 if (ap->a_bnp == NULL) 85 return (0); 86 87 error = ufs_bmaparray(ap->a_vp, ap->a_bn, &blkno, NULL, 88 ap->a_runp, ap->a_runb); 89 *ap->a_bnp = blkno; 90 return (error); 91 } 92 93 /* 94 * Indirect blocks are now on the vnode for the file. They are given negative 95 * logical block numbers. Indirect blocks are addressed by the negative 96 * address of the first data block to which they point. Double indirect blocks 97 * are addressed by one less than the address of the first indirect block to 98 * which they point. Triple indirect blocks are addressed by one less than 99 * the address of the first double indirect block to which they point. 100 * 101 * ufs_bmaparray does the bmap conversion, and if requested returns the 102 * array of logical blocks which must be traversed to get to a block. 103 * Each entry contains the offset into that block that gets you to the 104 * next block and the disk address of the block (if it is assigned). 105 */ 106 107 int 108 ufs_bmaparray(vp, bn, bnp, nbp, runp, runb) 109 struct vnode *vp; 110 ufs2_daddr_t bn; 111 ufs2_daddr_t *bnp; 112 struct buf *nbp; 113 int *runp; 114 int *runb; 115 { 116 struct inode *ip; 117 struct buf *bp; 118 struct ufsmount *ump; 119 struct mount *mp; 120 struct indir a[UFS_NIADDR+1], *ap; 121 ufs2_daddr_t daddr; 122 ufs_lbn_t metalbn; 123 int error, num, maxrun = 0; 124 int *nump; 125 126 ap = NULL; 127 ip = VTOI(vp); 128 mp = vp->v_mount; 129 ump = VFSTOUFS(mp); 130 131 if (runp) { 132 maxrun = mp->mnt_iosize_max / mp->mnt_stat.f_iosize - 1; 133 *runp = 0; 134 } 135 136 if (runb) { 137 *runb = 0; 138 } 139 140 141 ap = a; 142 nump = # 143 error = ufs_getlbns(vp, bn, ap, nump); 144 if (error) 145 return (error); 146 147 num = *nump; 148 if (num == 0) { 149 if (bn >= 0 && bn < UFS_NDADDR) { 150 *bnp = blkptrtodb(ump, DIP(ip, i_db[bn])); 151 } else if (bn < 0 && bn >= -UFS_NXADDR) { 152 *bnp = blkptrtodb(ump, ip->i_din2->di_extb[-1 - bn]); 153 if (*bnp == 0) 154 *bnp = -1; 155 if (nbp == NULL) 156 panic("ufs_bmaparray: mapping ext data"); 157 nbp->b_xflags |= BX_ALTDATA; 158 return (0); 159 } else { 160 panic("ufs_bmaparray: blkno out of range"); 161 } 162 /* 163 * Since this is FFS independent code, we are out of 164 * scope for the definitions of BLK_NOCOPY and 165 * BLK_SNAP, but we do know that they will fall in 166 * the range 1..um_seqinc, so we use that test and 167 * return a request for a zeroed out buffer if attempts 168 * are made to read a BLK_NOCOPY or BLK_SNAP block. 169 */ 170 if ((ip->i_flags & SF_SNAPSHOT) && DIP(ip, i_db[bn]) > 0 && 171 DIP(ip, i_db[bn]) < ump->um_seqinc) { 172 *bnp = -1; 173 } else if (*bnp == 0) { 174 if (ip->i_flags & SF_SNAPSHOT) 175 *bnp = blkptrtodb(ump, bn * ump->um_seqinc); 176 else 177 *bnp = -1; 178 } else if (runp) { 179 ufs2_daddr_t bnb = bn; 180 for (++bn; bn < UFS_NDADDR && *runp < maxrun && 181 is_sequential(ump, DIP(ip, i_db[bn - 1]), 182 DIP(ip, i_db[bn])); 183 ++bn, ++*runp); 184 bn = bnb; 185 if (runb && (bn > 0)) { 186 for (--bn; (bn >= 0) && (*runb < maxrun) && 187 is_sequential(ump, DIP(ip, i_db[bn]), 188 DIP(ip, i_db[bn+1])); 189 --bn, ++*runb); 190 } 191 } 192 return (0); 193 } 194 195 196 /* Get disk address out of indirect block array */ 197 daddr = DIP(ip, i_ib[ap->in_off]); 198 199 for (bp = NULL, ++ap; --num; ++ap) { 200 /* 201 * Exit the loop if there is no disk address assigned yet and 202 * the indirect block isn't in the cache, or if we were 203 * looking for an indirect block and we've found it. 204 */ 205 206 metalbn = ap->in_lbn; 207 if ((daddr == 0 && !incore(&vp->v_bufobj, metalbn)) || metalbn == bn) 208 break; 209 /* 210 * If we get here, we've either got the block in the cache 211 * or we have a disk address for it, go fetch it. 212 */ 213 if (bp) 214 bqrelse(bp); 215 216 bp = getblk(vp, metalbn, mp->mnt_stat.f_iosize, 0, 0, 0); 217 if ((bp->b_flags & B_CACHE) == 0) { 218 #ifdef INVARIANTS 219 if (!daddr) 220 panic("ufs_bmaparray: indirect block not in cache"); 221 #endif 222 bp->b_blkno = blkptrtodb(ump, daddr); 223 bp->b_iocmd = BIO_READ; 224 bp->b_flags &= ~B_INVAL; 225 bp->b_ioflags &= ~BIO_ERROR; 226 vfs_busy_pages(bp, 0); 227 bp->b_iooffset = dbtob(bp->b_blkno); 228 bstrategy(bp); 229 #ifdef RACCT 230 if (racct_enable) { 231 PROC_LOCK(curproc); 232 racct_add_buf(curproc, bp, 0); 233 PROC_UNLOCK(curproc); 234 } 235 #endif /* RACCT */ 236 curthread->td_ru.ru_inblock++; 237 error = bufwait(bp); 238 if (error) { 239 brelse(bp); 240 return (error); 241 } 242 } 243 244 if (I_IS_UFS1(ip)) { 245 daddr = ((ufs1_daddr_t *)bp->b_data)[ap->in_off]; 246 if (num == 1 && daddr && runp) { 247 for (bn = ap->in_off + 1; 248 bn < MNINDIR(ump) && *runp < maxrun && 249 is_sequential(ump, 250 ((ufs1_daddr_t *)bp->b_data)[bn - 1], 251 ((ufs1_daddr_t *)bp->b_data)[bn]); 252 ++bn, ++*runp); 253 bn = ap->in_off; 254 if (runb && bn) { 255 for (--bn; bn >= 0 && *runb < maxrun && 256 is_sequential(ump, 257 ((ufs1_daddr_t *)bp->b_data)[bn], 258 ((ufs1_daddr_t *)bp->b_data)[bn+1]); 259 --bn, ++*runb); 260 } 261 } 262 continue; 263 } 264 daddr = ((ufs2_daddr_t *)bp->b_data)[ap->in_off]; 265 if (num == 1 && daddr && runp) { 266 for (bn = ap->in_off + 1; 267 bn < MNINDIR(ump) && *runp < maxrun && 268 is_sequential(ump, 269 ((ufs2_daddr_t *)bp->b_data)[bn - 1], 270 ((ufs2_daddr_t *)bp->b_data)[bn]); 271 ++bn, ++*runp); 272 bn = ap->in_off; 273 if (runb && bn) { 274 for (--bn; bn >= 0 && *runb < maxrun && 275 is_sequential(ump, 276 ((ufs2_daddr_t *)bp->b_data)[bn], 277 ((ufs2_daddr_t *)bp->b_data)[bn + 1]); 278 --bn, ++*runb); 279 } 280 } 281 } 282 if (bp) 283 bqrelse(bp); 284 285 /* 286 * Since this is FFS independent code, we are out of scope for the 287 * definitions of BLK_NOCOPY and BLK_SNAP, but we do know that they 288 * will fall in the range 1..um_seqinc, so we use that test and 289 * return a request for a zeroed out buffer if attempts are made 290 * to read a BLK_NOCOPY or BLK_SNAP block. 291 */ 292 if ((ip->i_flags & SF_SNAPSHOT) && daddr > 0 && daddr < ump->um_seqinc){ 293 *bnp = -1; 294 return (0); 295 } 296 *bnp = blkptrtodb(ump, daddr); 297 if (*bnp == 0) { 298 if (ip->i_flags & SF_SNAPSHOT) 299 *bnp = blkptrtodb(ump, bn * ump->um_seqinc); 300 else 301 *bnp = -1; 302 } 303 return (0); 304 } 305 306 /* 307 * Create an array of logical block number/offset pairs which represent the 308 * path of indirect blocks required to access a data block. The first "pair" 309 * contains the logical block number of the appropriate single, double or 310 * triple indirect block and the offset into the inode indirect block array. 311 * Note, the logical block number of the inode single/double/triple indirect 312 * block appears twice in the array, once with the offset into the i_ib and 313 * once with the offset into the page itself. 314 */ 315 int 316 ufs_getlbns(vp, bn, ap, nump) 317 struct vnode *vp; 318 ufs2_daddr_t bn; 319 struct indir *ap; 320 int *nump; 321 { 322 ufs2_daddr_t blockcnt; 323 ufs_lbn_t metalbn, realbn; 324 struct ufsmount *ump; 325 int i, numlevels, off; 326 327 ump = VFSTOUFS(vp->v_mount); 328 if (nump) 329 *nump = 0; 330 numlevels = 0; 331 realbn = bn; 332 if (bn < 0) 333 bn = -bn; 334 335 /* The first UFS_NDADDR blocks are direct blocks. */ 336 if (bn < UFS_NDADDR) 337 return (0); 338 339 /* 340 * Determine the number of levels of indirection. After this loop 341 * is done, blockcnt indicates the number of data blocks possible 342 * at the previous level of indirection, and UFS_NIADDR - i is the 343 * number of levels of indirection needed to locate the requested block. 344 */ 345 for (blockcnt = 1, i = UFS_NIADDR, bn -= UFS_NDADDR; ; 346 i--, bn -= blockcnt) { 347 if (i == 0) 348 return (EFBIG); 349 blockcnt *= MNINDIR(ump); 350 if (bn < blockcnt) 351 break; 352 } 353 354 /* Calculate the address of the first meta-block. */ 355 if (realbn >= 0) 356 metalbn = -(realbn - bn + UFS_NIADDR - i); 357 else 358 metalbn = -(-realbn - bn + UFS_NIADDR - i); 359 360 /* 361 * At each iteration, off is the offset into the bap array which is 362 * an array of disk addresses at the current level of indirection. 363 * The logical block number and the offset in that block are stored 364 * into the argument array. 365 */ 366 ap->in_lbn = metalbn; 367 ap->in_off = off = UFS_NIADDR - i; 368 ap++; 369 for (++numlevels; i <= UFS_NIADDR; i++) { 370 /* If searching for a meta-data block, quit when found. */ 371 if (metalbn == realbn) 372 break; 373 374 blockcnt /= MNINDIR(ump); 375 off = (bn / blockcnt) % MNINDIR(ump); 376 377 ++numlevels; 378 ap->in_lbn = metalbn; 379 ap->in_off = off; 380 ++ap; 381 382 metalbn -= -1 + off * blockcnt; 383 } 384 if (nump) 385 *nump = numlevels; 386 return (0); 387 } 388