1 /* $NetBSD: ext2fs_inode.c,v 1.43 2004/08/15 07:19:56 mycroft Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1989, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#)ffs_inode.c 8.8 (Berkeley) 10/19/94 32 * Modified for ext2fs by Manuel Bouyer. 33 */ 34 35 /* 36 * Copyright (c) 1997 Manuel Bouyer. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. All advertising materials mentioning features or use of this software 47 * must display the following acknowledgement: 48 * This product includes software developed by Manuel Bouyer. 49 * 4. The name of the author may not be used to endorse or promote products 50 * derived from this software without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 53 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 54 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 55 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 56 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 57 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 58 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 59 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 60 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 61 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 62 * 63 * @(#)ffs_inode.c 8.8 (Berkeley) 10/19/94 64 * Modified for ext2fs by Manuel Bouyer. 65 */ 66 67 #include <sys/cdefs.h> 68 __KERNEL_RCSID(0, "$NetBSD: ext2fs_inode.c,v 1.43 2004/08/15 07:19:56 mycroft Exp $"); 69 70 #include <sys/param.h> 71 #include <sys/systm.h> 72 #include <sys/mount.h> 73 #include <sys/proc.h> 74 #include <sys/file.h> 75 #include <sys/buf.h> 76 #include <sys/vnode.h> 77 #include <sys/kernel.h> 78 #include <sys/malloc.h> 79 #include <sys/trace.h> 80 #include <sys/resourcevar.h> 81 82 #include <ufs/ufs/inode.h> 83 #include <ufs/ufs/ufsmount.h> 84 #include <ufs/ufs/ufs_extern.h> 85 86 #include <ufs/ext2fs/ext2fs.h> 87 #include <ufs/ext2fs/ext2fs_extern.h> 88 89 extern int prtactive; 90 91 static int ext2fs_indirtrunc __P((struct inode *, daddr_t, daddr_t, 92 daddr_t, int, long *)); 93 94 /* 95 * Last reference to an inode. If necessary, write or delete it. 96 */ 97 int 98 ext2fs_inactive(v) 99 void *v; 100 { 101 struct vop_inactive_args /* { 102 struct vnode *a_vp; 103 struct proc *a_p; 104 } */ *ap = v; 105 struct vnode *vp = ap->a_vp; 106 struct inode *ip = VTOI(vp); 107 struct mount *mp; 108 struct proc *p = ap->a_p; 109 struct timespec ts; 110 int error = 0; 111 112 if (prtactive && vp->v_usecount != 0) 113 vprint("ext2fs_inactive: pushing active", vp); 114 /* Get rid of inodes related to stale file handles. */ 115 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0) 116 goto out; 117 118 error = 0; 119 if (ip->i_e2fs_nlink == 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { 120 vn_start_write(vp, &mp, V_WAIT | V_LOWER); 121 if (ip->i_e2fs_size != 0) { 122 error = VOP_TRUNCATE(vp, (off_t)0, 0, NOCRED, NULL); 123 } 124 TIMEVAL_TO_TIMESPEC(&time, &ts); 125 ip->i_e2fs_dtime = ts.tv_sec; 126 ip->i_flag |= IN_CHANGE | IN_UPDATE; 127 VOP_VFREE(vp, ip->i_number, ip->i_e2fs_mode); 128 vn_finished_write(mp, V_LOWER); 129 } 130 if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) { 131 vn_start_write(vp, &mp, V_WAIT | V_LOWER); 132 VOP_UPDATE(vp, NULL, NULL, 0); 133 vn_finished_write(mp, V_LOWER); 134 } 135 out: 136 VOP_UNLOCK(vp, 0); 137 /* 138 * If we are done with the inode, reclaim it 139 * so that it can be reused immediately. 140 */ 141 if (ip->i_e2fs_dtime != 0) 142 vrecycle(vp, NULL, p); 143 return (error); 144 } 145 146 147 /* 148 * Update the access, modified, and inode change times as specified by the 149 * IACCESS, IUPDATE, and ICHANGE flags respectively. The IMODIFIED flag is 150 * used to specify that the inode needs to be updated but that the times have 151 * already been set. The access and modified times are taken from the second 152 * and third parameters; the inode change time is always taken from the current 153 * time. If UPDATE_WAIT or UPDATE_DIROP is set, then wait for the disk 154 * write of the inode to complete. 155 */ 156 int 157 ext2fs_update(v) 158 void *v; 159 { 160 struct vop_update_args /* { 161 struct vnode *a_vp; 162 struct timespec *a_access; 163 struct timespec *a_modify; 164 int a_flags; 165 } */ *ap = v; 166 struct m_ext2fs *fs; 167 struct buf *bp; 168 struct inode *ip; 169 int error; 170 struct timespec ts; 171 caddr_t cp; 172 int flags; 173 174 if (ap->a_vp->v_mount->mnt_flag & MNT_RDONLY) 175 return (0); 176 ip = VTOI(ap->a_vp); 177 TIMEVAL_TO_TIMESPEC(&time, &ts); 178 EXT2FS_ITIMES(ip, 179 ap->a_access ? ap->a_access : &ts, 180 ap->a_modify ? ap->a_modify : &ts, &ts); 181 if (ap->a_flags & UPDATE_CLOSE) 182 flags = ip->i_flag & (IN_MODIFIED | IN_ACCESSED); 183 else 184 flags = ip->i_flag & IN_MODIFIED; 185 if (flags == 0) 186 return (0); 187 fs = ip->i_e2fs; 188 189 error = bread(ip->i_devvp, 190 fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 191 (int)fs->e2fs_bsize, NOCRED, &bp); 192 if (error) { 193 brelse(bp); 194 return (error); 195 } 196 ip->i_flag &= ~(IN_MODIFIED | IN_ACCESSED); 197 cp = (caddr_t)bp->b_data + 198 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 199 e2fs_isave(ip->i_din.e2fs_din, (struct ext2fs_dinode *)cp); 200 if ((ap->a_flags & (UPDATE_WAIT|UPDATE_DIROP)) != 0 && 201 (flags & IN_MODIFIED) != 0 && 202 (ap->a_vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 203 return (bwrite(bp)); 204 else { 205 bdwrite(bp); 206 return (0); 207 } 208 } 209 210 #define SINGLE 0 /* index of single indirect block */ 211 #define DOUBLE 1 /* index of double indirect block */ 212 #define TRIPLE 2 /* index of triple indirect block */ 213 /* 214 * Truncate the inode oip to at most length size, freeing the 215 * disk blocks. 216 */ 217 int 218 ext2fs_truncate(v) 219 void *v; 220 { 221 struct vop_truncate_args /* { 222 struct vnode *a_vp; 223 off_t a_length; 224 int a_flags; 225 struct ucred *a_cred; 226 struct proc *a_p; 227 } */ *ap = v; 228 struct vnode *ovp = ap->a_vp; 229 daddr_t lastblock; 230 struct inode *oip = VTOI(ovp); 231 daddr_t bn, lastiblock[NIADDR], indir_lbn[NIADDR]; 232 /* XXX ondisk32 */ 233 int32_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; 234 off_t length = ap->a_length; 235 struct m_ext2fs *fs; 236 int offset, size, level; 237 long count, blocksreleased = 0; 238 int i, ioflag, nblocks; 239 int error, allerror = 0; 240 off_t osize; 241 int sync; 242 struct ufsmount *ump = oip->i_ump; 243 244 if (length < 0) 245 return (EINVAL); 246 247 if (ovp->v_type == VLNK && 248 (oip->i_e2fs_size < ump->um_maxsymlinklen || 249 (ump->um_maxsymlinklen == 0 && oip->i_e2fs_nblock == 0))) { 250 KDASSERT(length == 0); 251 memset((char *)&oip->i_din.e2fs_din->e2di_shortlink, 0, 252 (u_int)oip->i_e2fs_size); 253 oip->i_e2fs_size = 0; 254 oip->i_flag |= IN_CHANGE | IN_UPDATE; 255 return (VOP_UPDATE(ovp, NULL, NULL, 0)); 256 } 257 if (oip->i_e2fs_size == length) { 258 oip->i_flag |= IN_CHANGE | IN_UPDATE; 259 return (VOP_UPDATE(ovp, NULL, NULL, 0)); 260 } 261 fs = oip->i_e2fs; 262 if (length > ump->um_maxfilesize) 263 return (EFBIG); 264 265 osize = oip->i_e2fs_size; 266 ioflag = ap->a_flags; 267 268 /* 269 * Lengthen the size of the file. We must ensure that the 270 * last byte of the file is allocated. Since the smallest 271 * value of osize is 0, length will be at least 1. 272 */ 273 if (osize < length) { 274 error = ufs_balloc_range(ovp, length - 1, 1, ap->a_cred, 275 ioflag & IO_SYNC ? B_SYNC : 0); 276 if (error) { 277 (void) VOP_TRUNCATE(ovp, osize, ioflag & IO_SYNC, 278 ap->a_cred, ap->a_p); 279 return (error); 280 } 281 uvm_vnp_setsize(ovp, length); 282 oip->i_flag |= IN_CHANGE | IN_UPDATE; 283 KASSERT(ovp->v_size == oip->i_size); 284 return (VOP_UPDATE(ovp, NULL, NULL, 0)); 285 } 286 /* 287 * Shorten the size of the file. If the file is not being 288 * truncated to a block boundry, the contents of the 289 * partial block following the end of the file must be 290 * zero'ed in case it ever become accessible again because 291 * of subsequent file growth. 292 */ 293 offset = blkoff(fs, length); 294 if (offset != 0) { 295 size = fs->e2fs_bsize; 296 297 /* XXXUBC we should handle more than just VREG */ 298 uvm_vnp_zerorange(ovp, length, size - offset); 299 } 300 oip->i_e2fs_size = length; 301 uvm_vnp_setsize(ovp, length); 302 /* 303 * Calculate index into inode's block list of 304 * last direct and indirect blocks (if any) 305 * which we want to keep. Lastblock is -1 when 306 * the file is truncated to 0. 307 */ 308 lastblock = lblkno(fs, length + fs->e2fs_bsize - 1) - 1; 309 lastiblock[SINGLE] = lastblock - NDADDR; 310 lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); 311 lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); 312 nblocks = btodb(fs->e2fs_bsize); 313 /* 314 * Update file and block pointers on disk before we start freeing 315 * blocks. If we crash before free'ing blocks below, the blocks 316 * will be returned to the free list. lastiblock values are also 317 * normalized to -1 for calls to ext2fs_indirtrunc below. 318 */ 319 memcpy((caddr_t)oldblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof oldblks); 320 sync = 0; 321 for (level = TRIPLE; level >= SINGLE; level--) { 322 if (lastiblock[level] < 0 && oldblks[NDADDR + level] != 0) { 323 sync = 1; 324 oip->i_e2fs_blocks[NDADDR + level] = 0; 325 lastiblock[level] = -1; 326 } 327 } 328 for (i = 0; i < NDADDR; i++) { 329 if (i > lastblock && oldblks[i] != 0) { 330 sync = 1; 331 oip->i_e2fs_blocks[i] = 0; 332 } 333 } 334 oip->i_flag |= IN_CHANGE | IN_UPDATE; 335 if (sync) { 336 error = VOP_UPDATE(ovp, NULL, NULL, UPDATE_WAIT); 337 if (error && !allerror) 338 allerror = error; 339 } 340 341 /* 342 * Having written the new inode to disk, save its new configuration 343 * and put back the old block pointers long enough to process them. 344 * Note that we save the new block configuration so we can check it 345 * when we are done. 346 */ 347 memcpy((caddr_t)newblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof newblks); 348 memcpy((caddr_t)&oip->i_e2fs_blocks[0], (caddr_t)oldblks, sizeof oldblks); 349 350 oip->i_e2fs_size = osize; 351 error = vtruncbuf(ovp, lastblock + 1, 0, 0); 352 if (error && !allerror) 353 allerror = error; 354 355 /* 356 * Indirect blocks first. 357 */ 358 indir_lbn[SINGLE] = -NDADDR; 359 indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) -1; 360 indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; 361 for (level = TRIPLE; level >= SINGLE; level--) { 362 /* XXX ondisk32 */ 363 bn = fs2h32(oip->i_e2fs_blocks[NDADDR + level]); 364 if (bn != 0) { 365 error = ext2fs_indirtrunc(oip, indir_lbn[level], 366 fsbtodb(fs, bn), lastiblock[level], level, &count); 367 if (error) 368 allerror = error; 369 blocksreleased += count; 370 if (lastiblock[level] < 0) { 371 oip->i_e2fs_blocks[NDADDR + level] = 0; 372 ext2fs_blkfree(oip, bn); 373 blocksreleased += nblocks; 374 } 375 } 376 if (lastiblock[level] >= 0) 377 goto done; 378 } 379 380 /* 381 * All whole direct blocks or frags. 382 */ 383 for (i = NDADDR - 1; i > lastblock; i--) { 384 /* XXX ondisk32 */ 385 bn = fs2h32(oip->i_e2fs_blocks[i]); 386 if (bn == 0) 387 continue; 388 oip->i_e2fs_blocks[i] = 0; 389 ext2fs_blkfree(oip, bn); 390 blocksreleased += btodb(fs->e2fs_bsize); 391 } 392 393 done: 394 #ifdef DIAGNOSTIC 395 for (level = SINGLE; level <= TRIPLE; level++) 396 if (newblks[NDADDR + level] != 397 oip->i_e2fs_blocks[NDADDR + level]) 398 panic("ext2fs_truncate1"); 399 for (i = 0; i < NDADDR; i++) 400 if (newblks[i] != oip->i_e2fs_blocks[i]) 401 panic("ext2fs_truncate2"); 402 if (length == 0 && 403 (!LIST_EMPTY(&ovp->v_cleanblkhd) || 404 !LIST_EMPTY(&ovp->v_dirtyblkhd))) 405 panic("ext2fs_truncate3"); 406 #endif /* DIAGNOSTIC */ 407 /* 408 * Put back the real size. 409 */ 410 oip->i_e2fs_size = length; 411 oip->i_e2fs_nblock -= blocksreleased; 412 oip->i_flag |= IN_CHANGE; 413 KASSERT(ovp->v_type != VREG || ovp->v_size == oip->i_size); 414 return (allerror); 415 } 416 417 /* 418 * Release blocks associated with the inode ip and stored in the indirect 419 * block bn. Blocks are free'd in LIFO order up to (but not including) 420 * lastbn. If level is greater than SINGLE, the block is an indirect block 421 * and recursive calls to indirtrunc must be used to cleanse other indirect 422 * blocks. 423 * 424 * NB: triple indirect blocks are untested. 425 */ 426 static int 427 ext2fs_indirtrunc(ip, lbn, dbn, lastbn, level, countp) 428 struct inode *ip; 429 daddr_t lbn, lastbn; 430 daddr_t dbn; 431 int level; 432 long *countp; 433 { 434 int i; 435 struct buf *bp; 436 struct m_ext2fs *fs = ip->i_e2fs; 437 int32_t *bap; /* XXX ondisk32 */ 438 struct vnode *vp; 439 daddr_t nb, nlbn, last; 440 int32_t *copy = NULL; /* XXX ondisk32 */ 441 long blkcount, factor; 442 int nblocks, blocksreleased = 0; 443 int error = 0, allerror = 0; 444 445 /* 446 * Calculate index in current block of last 447 * block to be kept. -1 indicates the entire 448 * block so we need not calculate the index. 449 */ 450 factor = 1; 451 for (i = SINGLE; i < level; i++) 452 factor *= NINDIR(fs); 453 last = lastbn; 454 if (lastbn > 0) 455 last /= factor; 456 nblocks = btodb(fs->e2fs_bsize); 457 /* 458 * Get buffer of block pointers, zero those entries corresponding 459 * to blocks to be free'd, and update on disk copy first. Since 460 * double(triple) indirect before single(double) indirect, calls 461 * to bmap on these blocks will fail. However, we already have 462 * the on disk address, so we have to set the b_blkno field 463 * explicitly instead of letting bread do everything for us. 464 */ 465 vp = ITOV(ip); 466 bp = getblk(vp, lbn, (int)fs->e2fs_bsize, 0, 0); 467 if (bp->b_flags & (B_DONE | B_DELWRI)) { 468 /* Braces must be here in case trace evaluates to nothing. */ 469 trace(TR_BREADHIT, pack(vp, fs->e2fs_bsize), lbn); 470 } else { 471 trace(TR_BREADMISS, pack(vp, fs->e2fs_bsize), lbn); 472 curproc->p_stats->p_ru.ru_inblock++; /* pay for read */ 473 bp->b_flags |= B_READ; 474 if (bp->b_bcount > bp->b_bufsize) 475 panic("ext2fs_indirtrunc: bad buffer size"); 476 bp->b_blkno = dbn; 477 VOP_STRATEGY(vp, bp); 478 error = biowait(bp); 479 } 480 if (error) { 481 brelse(bp); 482 *countp = 0; 483 return (error); 484 } 485 486 bap = (int32_t *)bp->b_data; /* XXX ondisk32 */ 487 if (lastbn >= 0) { 488 /* XXX ondisk32 */ 489 MALLOC(copy, int32_t *, fs->e2fs_bsize, M_TEMP, M_WAITOK); 490 memcpy((caddr_t)copy, (caddr_t)bap, (u_int)fs->e2fs_bsize); 491 memset((caddr_t)&bap[last + 1], 0, 492 (u_int)(NINDIR(fs) - (last + 1)) * sizeof (u_int32_t)); 493 error = bwrite(bp); 494 if (error) 495 allerror = error; 496 bap = copy; 497 } 498 499 /* 500 * Recursively free totally unused blocks. 501 */ 502 for (i = NINDIR(fs) - 1, 503 nlbn = lbn + 1 - i * factor; i > last; 504 i--, nlbn += factor) { 505 /* XXX ondisk32 */ 506 nb = fs2h32(bap[i]); 507 if (nb == 0) 508 continue; 509 if (level > SINGLE) { 510 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 511 (daddr_t)-1, level - 1, 512 &blkcount); 513 if (error) 514 allerror = error; 515 blocksreleased += blkcount; 516 } 517 ext2fs_blkfree(ip, nb); 518 blocksreleased += nblocks; 519 } 520 521 /* 522 * Recursively free last partial block. 523 */ 524 if (level > SINGLE && lastbn >= 0) { 525 last = lastbn % factor; 526 /* XXX ondisk32 */ 527 nb = fs2h32(bap[i]); 528 if (nb != 0) { 529 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 530 last, level - 1, &blkcount); 531 if (error) 532 allerror = error; 533 blocksreleased += blkcount; 534 } 535 } 536 537 if (copy != NULL) { 538 FREE(copy, M_TEMP); 539 } else { 540 bp->b_flags |= B_INVAL; 541 brelse(bp); 542 } 543 544 *countp = blocksreleased; 545 return (allerror); 546 } 547