1 /* $NetBSD: ext2fs_inode.c,v 1.52 2005/12/11 12:25:25 christos 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 1.52 2005/12/11 12:25:25 christos 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(struct inode *, daddr_t, daddr_t, 92 daddr_t, int, long *); 93 94 /* 95 * Get the size of an inode. 96 */ 97 u_int64_t 98 ext2fs_size(struct inode *ip) 99 { 100 u_int64_t size = ip->i_e2fs_size; 101 102 if ((ip->i_e2fs_mode & IFMT) == IFREG) 103 size |= (u_int64_t)ip->i_e2fs_dacl << 32; 104 return size; 105 } 106 107 int 108 ext2fs_setsize(struct inode *ip, u_int64_t size) 109 { 110 if ((ip->i_e2fs_mode & IFMT) == IFREG || 111 ip->i_e2fs_mode == 0) { 112 ip->i_e2fs_dacl = size >> 32; 113 if (size >= 0x80000000U) { 114 struct m_ext2fs *fs = ip->i_e2fs; 115 116 if (fs->e2fs.e2fs_rev <= E2FS_REV0) { 117 /* Linux automagically upgrades to REV1 here! */ 118 return EFBIG; 119 } 120 if (!(fs->e2fs.e2fs_features_rocompat 121 & EXT2F_ROCOMPAT_LARGEFILE)) { 122 fs->e2fs.e2fs_features_rocompat |= 123 EXT2F_ROCOMPAT_LARGEFILE; 124 fs->e2fs_fmod = 1; 125 } 126 } 127 } else if (size >= 0x80000000U) 128 return EFBIG; 129 130 ip->i_e2fs_size = size; 131 132 return 0; 133 } 134 135 /* 136 * Last reference to an inode. If necessary, write or delete it. 137 */ 138 int 139 ext2fs_inactive(void *v) 140 { 141 struct vop_inactive_args /* { 142 struct vnode *a_vp; 143 struct lwp *a_l; 144 } */ *ap = v; 145 struct vnode *vp = ap->a_vp; 146 struct inode *ip = VTOI(vp); 147 struct mount *mp; 148 struct lwp *l = ap->a_l; 149 struct timespec ts; 150 int error = 0; 151 152 if (prtactive && vp->v_usecount != 0) 153 vprint("ext2fs_inactive: pushing active", vp); 154 /* Get rid of inodes related to stale file handles. */ 155 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0) 156 goto out; 157 158 error = 0; 159 if (ip->i_e2fs_nlink == 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { 160 vn_start_write(vp, &mp, V_WAIT | V_LOWER); 161 if (ext2fs_size(ip) != 0) { 162 error = ext2fs_truncate(vp, (off_t)0, 0, NOCRED, NULL); 163 } 164 nanotime(&ts); 165 ip->i_e2fs_dtime = ts.tv_sec; 166 ip->i_flag |= IN_CHANGE | IN_UPDATE; 167 ext2fs_vfree(vp, ip->i_number, ip->i_e2fs_mode); 168 vn_finished_write(mp, V_LOWER); 169 } 170 if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) { 171 vn_start_write(vp, &mp, V_WAIT | V_LOWER); 172 ext2fs_update(vp, NULL, NULL, 0); 173 vn_finished_write(mp, V_LOWER); 174 } 175 out: 176 VOP_UNLOCK(vp, 0); 177 /* 178 * If we are done with the inode, reclaim it 179 * so that it can be reused immediately. 180 */ 181 if (ip->i_e2fs_dtime != 0) 182 vrecycle(vp, NULL, l); 183 return (error); 184 } 185 186 187 /* 188 * Update the access, modified, and inode change times as specified by the 189 * IACCESS, IUPDATE, and ICHANGE flags respectively. The IMODIFIED flag is 190 * used to specify that the inode needs to be updated but that the times have 191 * already been set. The access and modified times are taken from the second 192 * and third parameters; the inode change time is always taken from the current 193 * time. If UPDATE_WAIT or UPDATE_DIROP is set, then wait for the disk 194 * write of the inode to complete. 195 */ 196 int 197 ext2fs_update(struct vnode *vp, const struct timespec *acc, 198 const struct timespec *mod, int updflags) 199 { 200 struct m_ext2fs *fs; 201 struct buf *bp; 202 struct inode *ip; 203 int error; 204 caddr_t cp; 205 int flags; 206 207 if (vp->v_mount->mnt_flag & MNT_RDONLY) 208 return (0); 209 ip = VTOI(vp); 210 EXT2FS_ITIMES(ip, acc, mod, NULL); 211 if (updflags & UPDATE_CLOSE) 212 flags = ip->i_flag & (IN_MODIFIED | IN_ACCESSED); 213 else 214 flags = ip->i_flag & IN_MODIFIED; 215 if (flags == 0) 216 return (0); 217 fs = ip->i_e2fs; 218 219 error = bread(ip->i_devvp, 220 fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 221 (int)fs->e2fs_bsize, NOCRED, &bp); 222 if (error) { 223 brelse(bp); 224 return (error); 225 } 226 ip->i_flag &= ~(IN_MODIFIED | IN_ACCESSED); 227 cp = (caddr_t)bp->b_data + 228 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 229 e2fs_isave(ip->i_din.e2fs_din, (struct ext2fs_dinode *)cp); 230 if ((updflags & (UPDATE_WAIT|UPDATE_DIROP)) != 0 && 231 (flags & IN_MODIFIED) != 0 && 232 (vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 233 return (bwrite(bp)); 234 else { 235 bdwrite(bp); 236 return (0); 237 } 238 } 239 240 #define SINGLE 0 /* index of single indirect block */ 241 #define DOUBLE 1 /* index of double indirect block */ 242 #define TRIPLE 2 /* index of triple indirect block */ 243 /* 244 * Truncate the inode oip to at most length size, freeing the 245 * disk blocks. 246 */ 247 int 248 ext2fs_truncate(struct vnode *ovp, off_t length, int ioflag, 249 struct ucred *cred, struct proc *p) 250 { 251 daddr_t lastblock; 252 struct inode *oip = VTOI(ovp); 253 daddr_t bn, lastiblock[NIADDR], indir_lbn[NIADDR]; 254 /* XXX ondisk32 */ 255 int32_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; 256 struct m_ext2fs *fs; 257 int offset, size, level; 258 long count, blocksreleased = 0; 259 int i, nblocks; 260 int error, allerror = 0; 261 off_t osize; 262 int sync; 263 struct ufsmount *ump = oip->i_ump; 264 265 if (ovp->v_type == VCHR || ovp->v_type == VBLK || 266 ovp->v_type == VFIFO || ovp->v_type == VSOCK) { 267 return 0; 268 } 269 270 if (length < 0) 271 return (EINVAL); 272 273 if (ovp->v_type == VLNK && 274 (ext2fs_size(oip) < ump->um_maxsymlinklen || 275 (ump->um_maxsymlinklen == 0 && oip->i_e2fs_nblock == 0))) { 276 KDASSERT(length == 0); 277 memset((char *)&oip->i_din.e2fs_din->e2di_shortlink, 0, 278 (u_int)ext2fs_size(oip)); 279 (void)ext2fs_setsize(oip, 0); 280 oip->i_flag |= IN_CHANGE | IN_UPDATE; 281 return (ext2fs_update(ovp, NULL, NULL, 0)); 282 } 283 if (ext2fs_size(oip) == length) { 284 oip->i_flag |= IN_CHANGE | IN_UPDATE; 285 return (ext2fs_update(ovp, NULL, NULL, 0)); 286 } 287 fs = oip->i_e2fs; 288 if (length > ump->um_maxfilesize) 289 return (EFBIG); 290 291 osize = ext2fs_size(oip); 292 293 /* 294 * Lengthen the size of the file. We must ensure that the 295 * last byte of the file is allocated. Since the smallest 296 * value of osize is 0, length will be at least 1. 297 */ 298 if (osize < length) { 299 error = ufs_balloc_range(ovp, length - 1, 1, cred, 300 ioflag & IO_SYNC ? B_SYNC : 0); 301 if (error) { 302 (void) ext2fs_truncate(ovp, osize, ioflag & IO_SYNC, 303 cred, p); 304 return (error); 305 } 306 uvm_vnp_setsize(ovp, length); 307 oip->i_flag |= IN_CHANGE | IN_UPDATE; 308 KASSERT(error || ovp->v_size == ext2fs_size(oip)); 309 return (ext2fs_update(ovp, NULL, NULL, 0)); 310 } 311 /* 312 * Shorten the size of the file. If the file is not being 313 * truncated to a block boundry, the contents of the 314 * partial block following the end of the file must be 315 * zero'ed in case it ever become accessible again because 316 * of subsequent file growth. 317 */ 318 offset = blkoff(fs, length); 319 if (offset != 0) { 320 size = fs->e2fs_bsize; 321 322 /* XXXUBC we should handle more than just VREG */ 323 uvm_vnp_zerorange(ovp, length, size - offset); 324 } 325 (void)ext2fs_setsize(oip, length); 326 uvm_vnp_setsize(ovp, length); 327 /* 328 * Calculate index into inode's block list of 329 * last direct and indirect blocks (if any) 330 * which we want to keep. Lastblock is -1 when 331 * the file is truncated to 0. 332 */ 333 lastblock = lblkno(fs, length + fs->e2fs_bsize - 1) - 1; 334 lastiblock[SINGLE] = lastblock - NDADDR; 335 lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); 336 lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); 337 nblocks = btodb(fs->e2fs_bsize); 338 /* 339 * Update file and block pointers on disk before we start freeing 340 * blocks. If we crash before free'ing blocks below, the blocks 341 * will be returned to the free list. lastiblock values are also 342 * normalized to -1 for calls to ext2fs_indirtrunc below. 343 */ 344 memcpy((caddr_t)oldblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof oldblks); 345 sync = 0; 346 for (level = TRIPLE; level >= SINGLE; level--) { 347 if (lastiblock[level] < 0 && oldblks[NDADDR + level] != 0) { 348 sync = 1; 349 oip->i_e2fs_blocks[NDADDR + level] = 0; 350 lastiblock[level] = -1; 351 } 352 } 353 for (i = 0; i < NDADDR; i++) { 354 if (i > lastblock && oldblks[i] != 0) { 355 sync = 1; 356 oip->i_e2fs_blocks[i] = 0; 357 } 358 } 359 oip->i_flag |= IN_CHANGE | IN_UPDATE; 360 if (sync) { 361 error = ext2fs_update(ovp, NULL, NULL, UPDATE_WAIT); 362 if (error && !allerror) 363 allerror = error; 364 } 365 366 /* 367 * Having written the new inode to disk, save its new configuration 368 * and put back the old block pointers long enough to process them. 369 * Note that we save the new block configuration so we can check it 370 * when we are done. 371 */ 372 memcpy((caddr_t)newblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof newblks); 373 memcpy((caddr_t)&oip->i_e2fs_blocks[0], (caddr_t)oldblks, sizeof oldblks); 374 375 (void)ext2fs_setsize(oip, osize); 376 error = vtruncbuf(ovp, lastblock + 1, 0, 0); 377 if (error && !allerror) 378 allerror = error; 379 380 /* 381 * Indirect blocks first. 382 */ 383 indir_lbn[SINGLE] = -NDADDR; 384 indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) -1; 385 indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; 386 for (level = TRIPLE; level >= SINGLE; level--) { 387 /* XXX ondisk32 */ 388 bn = fs2h32(oip->i_e2fs_blocks[NDADDR + level]); 389 if (bn != 0) { 390 error = ext2fs_indirtrunc(oip, indir_lbn[level], 391 fsbtodb(fs, bn), lastiblock[level], level, &count); 392 if (error) 393 allerror = error; 394 blocksreleased += count; 395 if (lastiblock[level] < 0) { 396 oip->i_e2fs_blocks[NDADDR + level] = 0; 397 ext2fs_blkfree(oip, bn); 398 blocksreleased += nblocks; 399 } 400 } 401 if (lastiblock[level] >= 0) 402 goto done; 403 } 404 405 /* 406 * All whole direct blocks or frags. 407 */ 408 for (i = NDADDR - 1; i > lastblock; i--) { 409 /* XXX ondisk32 */ 410 bn = fs2h32(oip->i_e2fs_blocks[i]); 411 if (bn == 0) 412 continue; 413 oip->i_e2fs_blocks[i] = 0; 414 ext2fs_blkfree(oip, bn); 415 blocksreleased += btodb(fs->e2fs_bsize); 416 } 417 418 done: 419 #ifdef DIAGNOSTIC 420 for (level = SINGLE; level <= TRIPLE; level++) 421 if (newblks[NDADDR + level] != 422 oip->i_e2fs_blocks[NDADDR + level]) 423 panic("ext2fs_truncate1"); 424 for (i = 0; i < NDADDR; i++) 425 if (newblks[i] != oip->i_e2fs_blocks[i]) 426 panic("ext2fs_truncate2"); 427 if (length == 0 && 428 (!LIST_EMPTY(&ovp->v_cleanblkhd) || 429 !LIST_EMPTY(&ovp->v_dirtyblkhd))) 430 panic("ext2fs_truncate3"); 431 #endif /* DIAGNOSTIC */ 432 /* 433 * Put back the real size. 434 */ 435 (void)ext2fs_setsize(oip, length); 436 oip->i_e2fs_nblock -= blocksreleased; 437 oip->i_flag |= IN_CHANGE; 438 KASSERT(ovp->v_type != VREG || ovp->v_size == ext2fs_size(oip)); 439 return (allerror); 440 } 441 442 /* 443 * Release blocks associated with the inode ip and stored in the indirect 444 * block bn. Blocks are free'd in LIFO order up to (but not including) 445 * lastbn. If level is greater than SINGLE, the block is an indirect block 446 * and recursive calls to indirtrunc must be used to cleanse other indirect 447 * blocks. 448 * 449 * NB: triple indirect blocks are untested. 450 */ 451 static int 452 ext2fs_indirtrunc(struct inode *ip, daddr_t lbn, daddr_t dbn, daddr_t lastbn, 453 int level, long *countp) 454 { 455 int i; 456 struct buf *bp; 457 struct m_ext2fs *fs = ip->i_e2fs; 458 int32_t *bap; /* XXX ondisk32 */ 459 struct vnode *vp; 460 daddr_t nb, nlbn, last; 461 int32_t *copy = NULL; /* XXX ondisk32 */ 462 long blkcount, factor; 463 int nblocks, blocksreleased = 0; 464 int error = 0, allerror = 0; 465 466 /* 467 * Calculate index in current block of last 468 * block to be kept. -1 indicates the entire 469 * block so we need not calculate the index. 470 */ 471 factor = 1; 472 for (i = SINGLE; i < level; i++) 473 factor *= NINDIR(fs); 474 last = lastbn; 475 if (lastbn > 0) 476 last /= factor; 477 nblocks = btodb(fs->e2fs_bsize); 478 /* 479 * Get buffer of block pointers, zero those entries corresponding 480 * to blocks to be free'd, and update on disk copy first. Since 481 * double(triple) indirect before single(double) indirect, calls 482 * to bmap on these blocks will fail. However, we already have 483 * the on disk address, so we have to set the b_blkno field 484 * explicitly instead of letting bread do everything for us. 485 */ 486 vp = ITOV(ip); 487 bp = getblk(vp, lbn, (int)fs->e2fs_bsize, 0, 0); 488 if (bp->b_flags & (B_DONE | B_DELWRI)) { 489 /* Braces must be here in case trace evaluates to nothing. */ 490 trace(TR_BREADHIT, pack(vp, fs->e2fs_bsize), lbn); 491 } else { 492 trace(TR_BREADMISS, pack(vp, fs->e2fs_bsize), lbn); 493 curproc->p_stats->p_ru.ru_inblock++; /* pay for read */ 494 bp->b_flags |= B_READ; 495 if (bp->b_bcount > bp->b_bufsize) 496 panic("ext2fs_indirtrunc: bad buffer size"); 497 bp->b_blkno = dbn; 498 VOP_STRATEGY(vp, bp); 499 error = biowait(bp); 500 } 501 if (error) { 502 brelse(bp); 503 *countp = 0; 504 return (error); 505 } 506 507 bap = (int32_t *)bp->b_data; /* XXX ondisk32 */ 508 if (lastbn >= 0) { 509 /* XXX ondisk32 */ 510 MALLOC(copy, int32_t *, fs->e2fs_bsize, M_TEMP, M_WAITOK); 511 memcpy((caddr_t)copy, (caddr_t)bap, (u_int)fs->e2fs_bsize); 512 memset((caddr_t)&bap[last + 1], 0, 513 (u_int)(NINDIR(fs) - (last + 1)) * sizeof (u_int32_t)); 514 error = bwrite(bp); 515 if (error) 516 allerror = error; 517 bap = copy; 518 } 519 520 /* 521 * Recursively free totally unused blocks. 522 */ 523 for (i = NINDIR(fs) - 1, 524 nlbn = lbn + 1 - i * factor; i > last; 525 i--, nlbn += factor) { 526 /* XXX ondisk32 */ 527 nb = fs2h32(bap[i]); 528 if (nb == 0) 529 continue; 530 if (level > SINGLE) { 531 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 532 (daddr_t)-1, level - 1, 533 &blkcount); 534 if (error) 535 allerror = error; 536 blocksreleased += blkcount; 537 } 538 ext2fs_blkfree(ip, nb); 539 blocksreleased += nblocks; 540 } 541 542 /* 543 * Recursively free last partial block. 544 */ 545 if (level > SINGLE && lastbn >= 0) { 546 last = lastbn % factor; 547 /* XXX ondisk32 */ 548 nb = fs2h32(bap[i]); 549 if (nb != 0) { 550 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 551 last, level - 1, &blkcount); 552 if (error) 553 allerror = error; 554 blocksreleased += blkcount; 555 } 556 } 557 558 if (copy != NULL) { 559 FREE(copy, M_TEMP); 560 } else { 561 bp->b_flags |= B_INVAL; 562 brelse(bp); 563 } 564 565 *countp = blocksreleased; 566 return (allerror); 567 } 568