1 /* $NetBSD: ext2fs_inode.c,v 1.55 2006/06/07 22:34:18 kardel 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 1.55 2006/06/07 22:34:18 kardel 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 #include <sys/kauth.h> 82 83 #include <ufs/ufs/inode.h> 84 #include <ufs/ufs/ufsmount.h> 85 #include <ufs/ufs/ufs_extern.h> 86 87 #include <ufs/ext2fs/ext2fs.h> 88 #include <ufs/ext2fs/ext2fs_extern.h> 89 90 extern int prtactive; 91 92 static int ext2fs_indirtrunc(struct inode *, daddr_t, daddr_t, 93 daddr_t, int, long *); 94 95 /* 96 * Get the size of an inode. 97 */ 98 u_int64_t 99 ext2fs_size(struct inode *ip) 100 { 101 u_int64_t size = ip->i_e2fs_size; 102 103 if ((ip->i_e2fs_mode & IFMT) == IFREG) 104 size |= (u_int64_t)ip->i_e2fs_dacl << 32; 105 return size; 106 } 107 108 int 109 ext2fs_setsize(struct inode *ip, u_int64_t size) 110 { 111 if ((ip->i_e2fs_mode & IFMT) == IFREG || 112 ip->i_e2fs_mode == 0) { 113 ip->i_e2fs_dacl = size >> 32; 114 if (size >= 0x80000000U) { 115 struct m_ext2fs *fs = ip->i_e2fs; 116 117 if (fs->e2fs.e2fs_rev <= E2FS_REV0) { 118 /* Linux automagically upgrades to REV1 here! */ 119 return EFBIG; 120 } 121 if (!(fs->e2fs.e2fs_features_rocompat 122 & EXT2F_ROCOMPAT_LARGEFILE)) { 123 fs->e2fs.e2fs_features_rocompat |= 124 EXT2F_ROCOMPAT_LARGEFILE; 125 fs->e2fs_fmod = 1; 126 } 127 } 128 } else if (size >= 0x80000000U) 129 return EFBIG; 130 131 ip->i_e2fs_size = size; 132 133 return 0; 134 } 135 136 /* 137 * Last reference to an inode. If necessary, write or delete it. 138 */ 139 int 140 ext2fs_inactive(void *v) 141 { 142 struct vop_inactive_args /* { 143 struct vnode *a_vp; 144 struct lwp *a_l; 145 } */ *ap = v; 146 struct vnode *vp = ap->a_vp; 147 struct inode *ip = VTOI(vp); 148 struct mount *mp; 149 struct lwp *l = ap->a_l; 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 ip->i_e2fs_dtime = time_uptime; 165 ip->i_flag |= IN_CHANGE | IN_UPDATE; 166 ext2fs_vfree(vp, ip->i_number, ip->i_e2fs_mode); 167 vn_finished_write(mp, V_LOWER); 168 } 169 if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) { 170 vn_start_write(vp, &mp, V_WAIT | V_LOWER); 171 ext2fs_update(vp, NULL, NULL, 0); 172 vn_finished_write(mp, V_LOWER); 173 } 174 out: 175 VOP_UNLOCK(vp, 0); 176 /* 177 * If we are done with the inode, reclaim it 178 * so that it can be reused immediately. 179 */ 180 if (ip->i_e2fs_dtime != 0) 181 vrecycle(vp, NULL, l); 182 return (error); 183 } 184 185 186 /* 187 * Update the access, modified, and inode change times as specified by the 188 * IACCESS, IUPDATE, and ICHANGE flags respectively. The IMODIFIED flag is 189 * used to specify that the inode needs to be updated but that the times have 190 * already been set. The access and modified times are taken from the second 191 * and third parameters; the inode change time is always taken from the current 192 * time. If UPDATE_WAIT or UPDATE_DIROP is set, then wait for the disk 193 * write of the inode to complete. 194 */ 195 int 196 ext2fs_update(struct vnode *vp, const struct timespec *acc, 197 const struct timespec *mod, int updflags) 198 { 199 struct m_ext2fs *fs; 200 struct buf *bp; 201 struct inode *ip; 202 int error; 203 caddr_t cp; 204 int flags; 205 206 if (vp->v_mount->mnt_flag & MNT_RDONLY) 207 return (0); 208 ip = VTOI(vp); 209 EXT2FS_ITIMES(ip, acc, mod, NULL); 210 if (updflags & UPDATE_CLOSE) 211 flags = ip->i_flag & (IN_MODIFIED | IN_ACCESSED); 212 else 213 flags = ip->i_flag & IN_MODIFIED; 214 if (flags == 0) 215 return (0); 216 fs = ip->i_e2fs; 217 218 error = bread(ip->i_devvp, 219 fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 220 (int)fs->e2fs_bsize, NOCRED, &bp); 221 if (error) { 222 brelse(bp); 223 return (error); 224 } 225 ip->i_flag &= ~(IN_MODIFIED | IN_ACCESSED); 226 cp = (caddr_t)bp->b_data + 227 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 228 e2fs_isave(ip->i_din.e2fs_din, (struct ext2fs_dinode *)cp); 229 if ((updflags & (UPDATE_WAIT|UPDATE_DIROP)) != 0 && 230 (flags & IN_MODIFIED) != 0 && 231 (vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 232 return (bwrite(bp)); 233 else { 234 bdwrite(bp); 235 return (0); 236 } 237 } 238 239 #define SINGLE 0 /* index of single indirect block */ 240 #define DOUBLE 1 /* index of double indirect block */ 241 #define TRIPLE 2 /* index of triple indirect block */ 242 /* 243 * Truncate the inode oip to at most length size, freeing the 244 * disk blocks. 245 */ 246 int 247 ext2fs_truncate(struct vnode *ovp, off_t length, int ioflag, 248 kauth_cred_t cred, struct proc *p) 249 { 250 daddr_t lastblock; 251 struct inode *oip = VTOI(ovp); 252 daddr_t bn, lastiblock[NIADDR], indir_lbn[NIADDR]; 253 /* XXX ondisk32 */ 254 int32_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; 255 struct m_ext2fs *fs; 256 int offset, size, level; 257 long count, blocksreleased = 0; 258 int i, nblocks; 259 int error, allerror = 0; 260 off_t osize; 261 int sync; 262 struct ufsmount *ump = oip->i_ump; 263 264 if (ovp->v_type == VCHR || ovp->v_type == VBLK || 265 ovp->v_type == VFIFO || ovp->v_type == VSOCK) { 266 return 0; 267 } 268 269 if (length < 0) 270 return (EINVAL); 271 272 if (ovp->v_type == VLNK && 273 (ext2fs_size(oip) < ump->um_maxsymlinklen || 274 (ump->um_maxsymlinklen == 0 && oip->i_e2fs_nblock == 0))) { 275 KDASSERT(length == 0); 276 memset((char *)&oip->i_din.e2fs_din->e2di_shortlink, 0, 277 (u_int)ext2fs_size(oip)); 278 (void)ext2fs_setsize(oip, 0); 279 oip->i_flag |= IN_CHANGE | IN_UPDATE; 280 return (ext2fs_update(ovp, NULL, NULL, 0)); 281 } 282 if (ext2fs_size(oip) == length) { 283 oip->i_flag |= IN_CHANGE | IN_UPDATE; 284 return (ext2fs_update(ovp, NULL, NULL, 0)); 285 } 286 fs = oip->i_e2fs; 287 if (length > ump->um_maxfilesize) 288 return (EFBIG); 289 290 osize = ext2fs_size(oip); 291 292 /* 293 * Lengthen the size of the file. We must ensure that the 294 * last byte of the file is allocated. Since the smallest 295 * value of osize is 0, length will be at least 1. 296 */ 297 if (osize < length) { 298 error = ufs_balloc_range(ovp, length - 1, 1, cred, 299 ioflag & IO_SYNC ? B_SYNC : 0); 300 if (error) { 301 (void) ext2fs_truncate(ovp, osize, ioflag & IO_SYNC, 302 cred, p); 303 return (error); 304 } 305 uvm_vnp_setsize(ovp, length); 306 oip->i_flag |= IN_CHANGE | IN_UPDATE; 307 KASSERT(error || ovp->v_size == ext2fs_size(oip)); 308 return (ext2fs_update(ovp, NULL, NULL, 0)); 309 } 310 /* 311 * Shorten the size of the file. If the file is not being 312 * truncated to a block boundry, the contents of the 313 * partial block following the end of the file must be 314 * zero'ed in case it ever become accessible again because 315 * of subsequent file growth. 316 */ 317 offset = blkoff(fs, length); 318 if (offset != 0) { 319 size = fs->e2fs_bsize; 320 321 /* XXXUBC we should handle more than just VREG */ 322 uvm_vnp_zerorange(ovp, length, size - offset); 323 } 324 (void)ext2fs_setsize(oip, length); 325 uvm_vnp_setsize(ovp, length); 326 /* 327 * Calculate index into inode's block list of 328 * last direct and indirect blocks (if any) 329 * which we want to keep. Lastblock is -1 when 330 * the file is truncated to 0. 331 */ 332 lastblock = lblkno(fs, length + fs->e2fs_bsize - 1) - 1; 333 lastiblock[SINGLE] = lastblock - NDADDR; 334 lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); 335 lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); 336 nblocks = btodb(fs->e2fs_bsize); 337 /* 338 * Update file and block pointers on disk before we start freeing 339 * blocks. If we crash before free'ing blocks below, the blocks 340 * will be returned to the free list. lastiblock values are also 341 * normalized to -1 for calls to ext2fs_indirtrunc below. 342 */ 343 memcpy((caddr_t)oldblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof oldblks); 344 sync = 0; 345 for (level = TRIPLE; level >= SINGLE; level--) { 346 if (lastiblock[level] < 0 && oldblks[NDADDR + level] != 0) { 347 sync = 1; 348 oip->i_e2fs_blocks[NDADDR + level] = 0; 349 lastiblock[level] = -1; 350 } 351 } 352 for (i = 0; i < NDADDR; i++) { 353 if (i > lastblock && oldblks[i] != 0) { 354 sync = 1; 355 oip->i_e2fs_blocks[i] = 0; 356 } 357 } 358 oip->i_flag |= IN_CHANGE | IN_UPDATE; 359 if (sync) { 360 error = ext2fs_update(ovp, NULL, NULL, UPDATE_WAIT); 361 if (error && !allerror) 362 allerror = error; 363 } 364 365 /* 366 * Having written the new inode to disk, save its new configuration 367 * and put back the old block pointers long enough to process them. 368 * Note that we save the new block configuration so we can check it 369 * when we are done. 370 */ 371 memcpy((caddr_t)newblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof newblks); 372 memcpy((caddr_t)&oip->i_e2fs_blocks[0], (caddr_t)oldblks, sizeof oldblks); 373 374 (void)ext2fs_setsize(oip, osize); 375 error = vtruncbuf(ovp, lastblock + 1, 0, 0); 376 if (error && !allerror) 377 allerror = error; 378 379 /* 380 * Indirect blocks first. 381 */ 382 indir_lbn[SINGLE] = -NDADDR; 383 indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) -1; 384 indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; 385 for (level = TRIPLE; level >= SINGLE; level--) { 386 /* XXX ondisk32 */ 387 bn = fs2h32(oip->i_e2fs_blocks[NDADDR + level]); 388 if (bn != 0) { 389 error = ext2fs_indirtrunc(oip, indir_lbn[level], 390 fsbtodb(fs, bn), lastiblock[level], level, &count); 391 if (error) 392 allerror = error; 393 blocksreleased += count; 394 if (lastiblock[level] < 0) { 395 oip->i_e2fs_blocks[NDADDR + level] = 0; 396 ext2fs_blkfree(oip, bn); 397 blocksreleased += nblocks; 398 } 399 } 400 if (lastiblock[level] >= 0) 401 goto done; 402 } 403 404 /* 405 * All whole direct blocks or frags. 406 */ 407 for (i = NDADDR - 1; i > lastblock; i--) { 408 /* XXX ondisk32 */ 409 bn = fs2h32(oip->i_e2fs_blocks[i]); 410 if (bn == 0) 411 continue; 412 oip->i_e2fs_blocks[i] = 0; 413 ext2fs_blkfree(oip, bn); 414 blocksreleased += btodb(fs->e2fs_bsize); 415 } 416 417 done: 418 #ifdef DIAGNOSTIC 419 for (level = SINGLE; level <= TRIPLE; level++) 420 if (newblks[NDADDR + level] != 421 oip->i_e2fs_blocks[NDADDR + level]) 422 panic("ext2fs_truncate1"); 423 for (i = 0; i < NDADDR; i++) 424 if (newblks[i] != oip->i_e2fs_blocks[i]) 425 panic("ext2fs_truncate2"); 426 if (length == 0 && 427 (!LIST_EMPTY(&ovp->v_cleanblkhd) || 428 !LIST_EMPTY(&ovp->v_dirtyblkhd))) 429 panic("ext2fs_truncate3"); 430 #endif /* DIAGNOSTIC */ 431 /* 432 * Put back the real size. 433 */ 434 (void)ext2fs_setsize(oip, length); 435 oip->i_e2fs_nblock -= blocksreleased; 436 oip->i_flag |= IN_CHANGE; 437 KASSERT(ovp->v_type != VREG || ovp->v_size == ext2fs_size(oip)); 438 return (allerror); 439 } 440 441 /* 442 * Release blocks associated with the inode ip and stored in the indirect 443 * block bn. Blocks are free'd in LIFO order up to (but not including) 444 * lastbn. If level is greater than SINGLE, the block is an indirect block 445 * and recursive calls to indirtrunc must be used to cleanse other indirect 446 * blocks. 447 * 448 * NB: triple indirect blocks are untested. 449 */ 450 static int 451 ext2fs_indirtrunc(struct inode *ip, daddr_t lbn, daddr_t dbn, daddr_t lastbn, 452 int level, long *countp) 453 { 454 int i; 455 struct buf *bp; 456 struct m_ext2fs *fs = ip->i_e2fs; 457 int32_t *bap; /* XXX ondisk32 */ 458 struct vnode *vp; 459 daddr_t nb, nlbn, last; 460 int32_t *copy = NULL; /* XXX ondisk32 */ 461 long blkcount, factor; 462 int nblocks, blocksreleased = 0; 463 int error = 0, allerror = 0; 464 465 /* 466 * Calculate index in current block of last 467 * block to be kept. -1 indicates the entire 468 * block so we need not calculate the index. 469 */ 470 factor = 1; 471 for (i = SINGLE; i < level; i++) 472 factor *= NINDIR(fs); 473 last = lastbn; 474 if (lastbn > 0) 475 last /= factor; 476 nblocks = btodb(fs->e2fs_bsize); 477 /* 478 * Get buffer of block pointers, zero those entries corresponding 479 * to blocks to be free'd, and update on disk copy first. Since 480 * double(triple) indirect before single(double) indirect, calls 481 * to bmap on these blocks will fail. However, we already have 482 * the on disk address, so we have to set the b_blkno field 483 * explicitly instead of letting bread do everything for us. 484 */ 485 vp = ITOV(ip); 486 bp = getblk(vp, lbn, (int)fs->e2fs_bsize, 0, 0); 487 if (bp->b_flags & (B_DONE | B_DELWRI)) { 488 /* Braces must be here in case trace evaluates to nothing. */ 489 trace(TR_BREADHIT, pack(vp, fs->e2fs_bsize), lbn); 490 } else { 491 trace(TR_BREADMISS, pack(vp, fs->e2fs_bsize), lbn); 492 curproc->p_stats->p_ru.ru_inblock++; /* pay for read */ 493 bp->b_flags |= B_READ; 494 if (bp->b_bcount > bp->b_bufsize) 495 panic("ext2fs_indirtrunc: bad buffer size"); 496 bp->b_blkno = dbn; 497 VOP_STRATEGY(vp, bp); 498 error = biowait(bp); 499 } 500 if (error) { 501 brelse(bp); 502 *countp = 0; 503 return (error); 504 } 505 506 bap = (int32_t *)bp->b_data; /* XXX ondisk32 */ 507 if (lastbn >= 0) { 508 /* XXX ondisk32 */ 509 copy = malloc(fs->e2fs_bsize, M_TEMP, M_WAITOK); 510 memcpy((caddr_t)copy, (caddr_t)bap, (u_int)fs->e2fs_bsize); 511 memset((caddr_t)&bap[last + 1], 0, 512 (u_int)(NINDIR(fs) - (last + 1)) * sizeof (u_int32_t)); 513 error = bwrite(bp); 514 if (error) 515 allerror = error; 516 bap = copy; 517 } 518 519 /* 520 * Recursively free totally unused blocks. 521 */ 522 for (i = NINDIR(fs) - 1, 523 nlbn = lbn + 1 - i * factor; i > last; 524 i--, nlbn += factor) { 525 /* XXX ondisk32 */ 526 nb = fs2h32(bap[i]); 527 if (nb == 0) 528 continue; 529 if (level > SINGLE) { 530 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 531 (daddr_t)-1, level - 1, 532 &blkcount); 533 if (error) 534 allerror = error; 535 blocksreleased += blkcount; 536 } 537 ext2fs_blkfree(ip, nb); 538 blocksreleased += nblocks; 539 } 540 541 /* 542 * Recursively free last partial block. 543 */ 544 if (level > SINGLE && lastbn >= 0) { 545 last = lastbn % factor; 546 /* XXX ondisk32 */ 547 nb = fs2h32(bap[i]); 548 if (nb != 0) { 549 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 550 last, level - 1, &blkcount); 551 if (error) 552 allerror = error; 553 blocksreleased += blkcount; 554 } 555 } 556 557 if (copy != NULL) { 558 FREE(copy, M_TEMP); 559 } else { 560 bp->b_flags |= B_INVAL; 561 brelse(bp); 562 } 563 564 *countp = blocksreleased; 565 return (allerror); 566 } 567