1 /* $NetBSD: ext2fs_inode.c,v 1.60 2007/10/08 18:01:27 ad 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 1.60 2007/10/08 18:01:27 ad 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 lwp *l = ap->a_l; 149 int error = 0; 150 151 if (prtactive && vp->v_usecount != 0) 152 vprint("ext2fs_inactive: pushing active", vp); 153 /* Get rid of inodes related to stale file handles. */ 154 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0) 155 goto out; 156 157 error = 0; 158 if (ip->i_e2fs_nlink == 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { 159 if (ext2fs_size(ip) != 0) { 160 error = ext2fs_truncate(vp, (off_t)0, 0, NOCRED, NULL); 161 } 162 ip->i_e2fs_dtime = time_second; 163 ip->i_flag |= IN_CHANGE | IN_UPDATE; 164 ext2fs_vfree(vp, ip->i_number, ip->i_e2fs_mode); 165 } 166 if (ip->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) { 167 ext2fs_update(vp, NULL, NULL, 0); 168 } 169 out: 170 VOP_UNLOCK(vp, 0); 171 /* 172 * If we are done with the inode, reclaim it 173 * so that it can be reused immediately. 174 */ 175 if (ip->i_e2fs_dtime != 0) 176 vrecycle(vp, NULL, l); 177 return (error); 178 } 179 180 181 /* 182 * Update the access, modified, and inode change times as specified by the 183 * IACCESS, IUPDATE, and ICHANGE flags respectively. The IMODIFIED flag is 184 * used to specify that the inode needs to be updated but that the times have 185 * already been set. The access and modified times are taken from the second 186 * and third parameters; the inode change time is always taken from the current 187 * time. If UPDATE_WAIT or UPDATE_DIROP is set, then wait for the disk 188 * write of the inode to complete. 189 */ 190 int 191 ext2fs_update(struct vnode *vp, const struct timespec *acc, 192 const struct timespec *mod, int updflags) 193 { 194 struct m_ext2fs *fs; 195 struct buf *bp; 196 struct inode *ip; 197 int error; 198 void *cp; 199 int flags; 200 201 if (vp->v_mount->mnt_flag & MNT_RDONLY) 202 return (0); 203 ip = VTOI(vp); 204 EXT2FS_ITIMES(ip, acc, mod, NULL); 205 if (updflags & UPDATE_CLOSE) 206 flags = ip->i_flag & (IN_MODIFIED | IN_ACCESSED); 207 else 208 flags = ip->i_flag & IN_MODIFIED; 209 if (flags == 0) 210 return (0); 211 fs = ip->i_e2fs; 212 213 error = bread(ip->i_devvp, 214 fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 215 (int)fs->e2fs_bsize, NOCRED, &bp); 216 if (error) { 217 brelse(bp, 0); 218 return (error); 219 } 220 ip->i_flag &= ~(IN_MODIFIED | IN_ACCESSED); 221 cp = (char *)bp->b_data + 222 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 223 e2fs_isave(ip->i_din.e2fs_din, (struct ext2fs_dinode *)cp); 224 if ((updflags & (UPDATE_WAIT|UPDATE_DIROP)) != 0 && 225 (flags & IN_MODIFIED) != 0 && 226 (vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 227 return (bwrite(bp)); 228 else { 229 bdwrite(bp); 230 return (0); 231 } 232 } 233 234 #define SINGLE 0 /* index of single indirect block */ 235 #define DOUBLE 1 /* index of double indirect block */ 236 #define TRIPLE 2 /* index of triple indirect block */ 237 /* 238 * Truncate the inode oip to at most length size, freeing the 239 * disk blocks. 240 */ 241 int 242 ext2fs_truncate(struct vnode *ovp, off_t length, int ioflag, 243 kauth_cred_t cred, struct proc *p) 244 { 245 daddr_t lastblock; 246 struct inode *oip = VTOI(ovp); 247 daddr_t bn, lastiblock[NIADDR], indir_lbn[NIADDR]; 248 /* XXX ondisk32 */ 249 int32_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; 250 struct m_ext2fs *fs; 251 int offset, size, level; 252 long count, blocksreleased = 0; 253 int i, nblocks; 254 int error, allerror = 0; 255 off_t osize; 256 int sync; 257 struct ufsmount *ump = oip->i_ump; 258 259 if (ovp->v_type == VCHR || ovp->v_type == VBLK || 260 ovp->v_type == VFIFO || ovp->v_type == VSOCK) { 261 return 0; 262 } 263 264 if (length < 0) 265 return (EINVAL); 266 267 if (ovp->v_type == VLNK && 268 (ext2fs_size(oip) < ump->um_maxsymlinklen || 269 (ump->um_maxsymlinklen == 0 && oip->i_e2fs_nblock == 0))) { 270 KDASSERT(length == 0); 271 memset((char *)&oip->i_din.e2fs_din->e2di_shortlink, 0, 272 (u_int)ext2fs_size(oip)); 273 (void)ext2fs_setsize(oip, 0); 274 oip->i_flag |= IN_CHANGE | IN_UPDATE; 275 return (ext2fs_update(ovp, NULL, NULL, 0)); 276 } 277 if (ext2fs_size(oip) == length) { 278 oip->i_flag |= IN_CHANGE | IN_UPDATE; 279 return (ext2fs_update(ovp, NULL, NULL, 0)); 280 } 281 fs = oip->i_e2fs; 282 if (length > ump->um_maxfilesize) 283 return (EFBIG); 284 285 osize = ext2fs_size(oip); 286 287 /* 288 * Lengthen the size of the file. We must ensure that the 289 * last byte of the file is allocated. Since the smallest 290 * value of osize is 0, length will be at least 1. 291 */ 292 if (osize < length) { 293 uvm_vnp_setwritesize(ovp, length); 294 error = ufs_balloc_range(ovp, length - 1, 1, cred, 295 ioflag & IO_SYNC ? B_SYNC : 0); 296 if (error) { 297 (void) ext2fs_truncate(ovp, osize, ioflag & IO_SYNC, 298 cred, p); 299 return (error); 300 } 301 uvm_vnp_setsize(ovp, length); 302 oip->i_flag |= IN_CHANGE | IN_UPDATE; 303 KASSERT(error || ovp->v_size == ext2fs_size(oip)); 304 return (ext2fs_update(ovp, NULL, NULL, 0)); 305 } 306 /* 307 * Shorten the size of the file. If the file is not being 308 * truncated to a block boundry, the contents of the 309 * partial block following the end of the file must be 310 * zero'ed in case it ever become accessible again because 311 * of subsequent file growth. 312 */ 313 offset = blkoff(fs, length); 314 if (offset != 0) { 315 size = fs->e2fs_bsize; 316 317 /* XXXUBC we should handle more than just VREG */ 318 uvm_vnp_zerorange(ovp, length, size - offset); 319 } 320 (void)ext2fs_setsize(oip, length); 321 uvm_vnp_setsize(ovp, length); 322 /* 323 * Calculate index into inode's block list of 324 * last direct and indirect blocks (if any) 325 * which we want to keep. Lastblock is -1 when 326 * the file is truncated to 0. 327 */ 328 lastblock = lblkno(fs, length + fs->e2fs_bsize - 1) - 1; 329 lastiblock[SINGLE] = lastblock - NDADDR; 330 lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); 331 lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); 332 nblocks = btodb(fs->e2fs_bsize); 333 /* 334 * Update file and block pointers on disk before we start freeing 335 * blocks. If we crash before free'ing blocks below, the blocks 336 * will be returned to the free list. lastiblock values are also 337 * normalized to -1 for calls to ext2fs_indirtrunc below. 338 */ 339 memcpy((void *)oldblks, (void *)&oip->i_e2fs_blocks[0], sizeof oldblks); 340 sync = 0; 341 for (level = TRIPLE; level >= SINGLE; level--) { 342 if (lastiblock[level] < 0 && oldblks[NDADDR + level] != 0) { 343 sync = 1; 344 oip->i_e2fs_blocks[NDADDR + level] = 0; 345 lastiblock[level] = -1; 346 } 347 } 348 for (i = 0; i < NDADDR; i++) { 349 if (i > lastblock && oldblks[i] != 0) { 350 sync = 1; 351 oip->i_e2fs_blocks[i] = 0; 352 } 353 } 354 oip->i_flag |= IN_CHANGE | IN_UPDATE; 355 if (sync) { 356 error = ext2fs_update(ovp, NULL, NULL, UPDATE_WAIT); 357 if (error && !allerror) 358 allerror = error; 359 } 360 361 /* 362 * Having written the new inode to disk, save its new configuration 363 * and put back the old block pointers long enough to process them. 364 * Note that we save the new block configuration so we can check it 365 * when we are done. 366 */ 367 memcpy((void *)newblks, (void *)&oip->i_e2fs_blocks[0], sizeof newblks); 368 memcpy((void *)&oip->i_e2fs_blocks[0], (void *)oldblks, sizeof oldblks); 369 370 (void)ext2fs_setsize(oip, osize); 371 error = vtruncbuf(ovp, lastblock + 1, 0, 0); 372 if (error && !allerror) 373 allerror = error; 374 375 /* 376 * Indirect blocks first. 377 */ 378 indir_lbn[SINGLE] = -NDADDR; 379 indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) -1; 380 indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; 381 for (level = TRIPLE; level >= SINGLE; level--) { 382 /* XXX ondisk32 */ 383 bn = fs2h32(oip->i_e2fs_blocks[NDADDR + level]); 384 if (bn != 0) { 385 error = ext2fs_indirtrunc(oip, indir_lbn[level], 386 fsbtodb(fs, bn), lastiblock[level], level, &count); 387 if (error) 388 allerror = error; 389 blocksreleased += count; 390 if (lastiblock[level] < 0) { 391 oip->i_e2fs_blocks[NDADDR + level] = 0; 392 ext2fs_blkfree(oip, bn); 393 blocksreleased += nblocks; 394 } 395 } 396 if (lastiblock[level] >= 0) 397 goto done; 398 } 399 400 /* 401 * All whole direct blocks or frags. 402 */ 403 for (i = NDADDR - 1; i > lastblock; i--) { 404 /* XXX ondisk32 */ 405 bn = fs2h32(oip->i_e2fs_blocks[i]); 406 if (bn == 0) 407 continue; 408 oip->i_e2fs_blocks[i] = 0; 409 ext2fs_blkfree(oip, bn); 410 blocksreleased += btodb(fs->e2fs_bsize); 411 } 412 413 done: 414 #ifdef DIAGNOSTIC 415 for (level = SINGLE; level <= TRIPLE; level++) 416 if (newblks[NDADDR + level] != 417 oip->i_e2fs_blocks[NDADDR + level]) 418 panic("ext2fs_truncate1"); 419 for (i = 0; i < NDADDR; i++) 420 if (newblks[i] != oip->i_e2fs_blocks[i]) 421 panic("ext2fs_truncate2"); 422 if (length == 0 && 423 (!LIST_EMPTY(&ovp->v_cleanblkhd) || 424 !LIST_EMPTY(&ovp->v_dirtyblkhd))) 425 panic("ext2fs_truncate3"); 426 #endif /* DIAGNOSTIC */ 427 /* 428 * Put back the real size. 429 */ 430 (void)ext2fs_setsize(oip, length); 431 oip->i_e2fs_nblock -= blocksreleased; 432 oip->i_flag |= IN_CHANGE; 433 KASSERT(ovp->v_type != VREG || ovp->v_size == ext2fs_size(oip)); 434 return (allerror); 435 } 436 437 /* 438 * Release blocks associated with the inode ip and stored in the indirect 439 * block bn. Blocks are free'd in LIFO order up to (but not including) 440 * lastbn. If level is greater than SINGLE, the block is an indirect block 441 * and recursive calls to indirtrunc must be used to cleanse other indirect 442 * blocks. 443 * 444 * NB: triple indirect blocks are untested. 445 */ 446 static int 447 ext2fs_indirtrunc(struct inode *ip, daddr_t lbn, daddr_t dbn, daddr_t lastbn, 448 int level, long *countp) 449 { 450 int i; 451 struct buf *bp; 452 struct m_ext2fs *fs = ip->i_e2fs; 453 int32_t *bap; /* XXX ondisk32 */ 454 struct vnode *vp; 455 daddr_t nb, nlbn, last; 456 int32_t *copy = NULL; /* XXX ondisk32 */ 457 long blkcount, factor; 458 int nblocks, blocksreleased = 0; 459 int error = 0, allerror = 0; 460 461 /* 462 * Calculate index in current block of last 463 * block to be kept. -1 indicates the entire 464 * block so we need not calculate the index. 465 */ 466 factor = 1; 467 for (i = SINGLE; i < level; i++) 468 factor *= NINDIR(fs); 469 last = lastbn; 470 if (lastbn > 0) 471 last /= factor; 472 nblocks = btodb(fs->e2fs_bsize); 473 /* 474 * Get buffer of block pointers, zero those entries corresponding 475 * to blocks to be free'd, and update on disk copy first. Since 476 * double(triple) indirect before single(double) indirect, calls 477 * to bmap on these blocks will fail. However, we already have 478 * the on disk address, so we have to set the b_blkno field 479 * explicitly instead of letting bread do everything for us. 480 */ 481 vp = ITOV(ip); 482 bp = getblk(vp, lbn, (int)fs->e2fs_bsize, 0, 0); 483 if (bp->b_flags & (B_DONE | B_DELWRI)) { 484 /* Braces must be here in case trace evaluates to nothing. */ 485 trace(TR_BREADHIT, pack(vp, fs->e2fs_bsize), lbn); 486 } else { 487 trace(TR_BREADMISS, pack(vp, fs->e2fs_bsize), lbn); 488 curproc->p_stats->p_ru.ru_inblock++; /* pay for read */ 489 bp->b_flags |= B_READ; 490 if (bp->b_bcount > bp->b_bufsize) 491 panic("ext2fs_indirtrunc: bad buffer size"); 492 bp->b_blkno = dbn; 493 VOP_STRATEGY(vp, bp); 494 error = biowait(bp); 495 } 496 if (error) { 497 brelse(bp, 0); 498 *countp = 0; 499 return (error); 500 } 501 502 bap = (int32_t *)bp->b_data; /* XXX ondisk32 */ 503 if (lastbn >= 0) { 504 /* XXX ondisk32 */ 505 copy = malloc(fs->e2fs_bsize, M_TEMP, M_WAITOK); 506 memcpy((void *)copy, (void *)bap, (u_int)fs->e2fs_bsize); 507 memset((void *)&bap[last + 1], 0, 508 (u_int)(NINDIR(fs) - (last + 1)) * sizeof (u_int32_t)); 509 error = bwrite(bp); 510 if (error) 511 allerror = error; 512 bap = copy; 513 } 514 515 /* 516 * Recursively free totally unused blocks. 517 */ 518 for (i = NINDIR(fs) - 1, 519 nlbn = lbn + 1 - i * factor; i > last; 520 i--, nlbn += factor) { 521 /* XXX ondisk32 */ 522 nb = fs2h32(bap[i]); 523 if (nb == 0) 524 continue; 525 if (level > SINGLE) { 526 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 527 (daddr_t)-1, level - 1, 528 &blkcount); 529 if (error) 530 allerror = error; 531 blocksreleased += blkcount; 532 } 533 ext2fs_blkfree(ip, nb); 534 blocksreleased += nblocks; 535 } 536 537 /* 538 * Recursively free last partial block. 539 */ 540 if (level > SINGLE && lastbn >= 0) { 541 last = lastbn % factor; 542 /* XXX ondisk32 */ 543 nb = fs2h32(bap[i]); 544 if (nb != 0) { 545 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 546 last, level - 1, &blkcount); 547 if (error) 548 allerror = error; 549 blocksreleased += blkcount; 550 } 551 } 552 553 if (copy != NULL) { 554 FREE(copy, M_TEMP); 555 } else { 556 brelse(bp, BC_INVAL); 557 } 558 559 *countp = blocksreleased; 560 return (allerror); 561 } 562