1 /* $NetBSD: ext2fs_inode.c,v 1.13 1999/03/24 05:51:30 mrg Exp $ */ 2 3 /* 4 * Copyright (c) 1997 Manuel Bouyer. 5 * Copyright (c) 1982, 1986, 1989, 1993 6 * The Regents of the University of California. All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. 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 * @(#)ffs_inode.c 8.8 (Berkeley) 10/19/94 37 * Modified for ext2fs by Manuel Bouyer. 38 */ 39 40 #include <sys/param.h> 41 #include <sys/systm.h> 42 #include <sys/mount.h> 43 #include <sys/proc.h> 44 #include <sys/file.h> 45 #include <sys/buf.h> 46 #include <sys/vnode.h> 47 #include <sys/kernel.h> 48 #include <sys/malloc.h> 49 #include <sys/trace.h> 50 #include <sys/resourcevar.h> 51 52 #include <vm/vm.h> 53 54 #include <uvm/uvm_extern.h> 55 56 #include <ufs/ufs/quota.h> 57 #include <ufs/ufs/inode.h> 58 #include <ufs/ufs/ufsmount.h> 59 #include <ufs/ufs/ufs_extern.h> 60 61 #include <ufs/ext2fs/ext2fs.h> 62 #include <ufs/ext2fs/ext2fs_extern.h> 63 64 static int ext2fs_indirtrunc __P((struct inode *, ufs_daddr_t, ufs_daddr_t, 65 ufs_daddr_t, int, long *)); 66 67 /* 68 * Last reference to an inode. If necessary, write or delete it. 69 */ 70 int 71 ext2fs_inactive(v) 72 void *v; 73 { 74 struct vop_inactive_args /* { 75 struct vnode *a_vp; 76 struct proc *a_p; 77 } */ *ap = v; 78 struct vnode *vp = ap->a_vp; 79 struct inode *ip = VTOI(vp); 80 struct proc *p = ap->a_p; 81 struct timespec ts; 82 int error = 0; 83 extern int prtactive; 84 85 if (prtactive && vp->v_usecount != 0) 86 vprint("ext2fs_inactive: pushing active", vp); 87 /* Get rid of inodes related to stale file handles. */ 88 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0) 89 goto out; 90 91 error = 0; 92 if (ip->i_e2fs_nlink == 0 && (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { 93 error = VOP_TRUNCATE(vp, (off_t)0, 0, NOCRED, NULL); 94 TIMEVAL_TO_TIMESPEC(&time, &ts); 95 ip->i_e2fs_dtime = ts.tv_sec; 96 ip->i_flag |= IN_CHANGE | IN_UPDATE; 97 VOP_VFREE(vp, ip->i_number, ip->i_e2fs_mode); 98 } 99 if (ip->i_flag & (IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) 100 VOP_UPDATE(vp, NULL, NULL, 0); 101 out: 102 VOP_UNLOCK(vp, 0); 103 /* 104 * If we are done with the inode, reclaim it 105 * so that it can be reused immediately. 106 */ 107 if (ip->i_e2fs_dtime != 0) 108 vrecycle(vp, (struct simplelock *)0, p); 109 return (error); 110 } 111 112 113 /* 114 * Update the access, modified, and inode change times as specified by the 115 * IACCESS, IUPDATE, and ICHANGE flags respectively. The IMODIFIED flag is 116 * used to specify that the inode needs to be updated but that the times have 117 * already been set. The access and modified times are taken from the second 118 * and third parameters; the inode change time is always taken from the current 119 * time. If waitfor is set, then wait for the disk write of the inode to 120 * complete. 121 */ 122 int 123 ext2fs_update(v) 124 void *v; 125 { 126 struct vop_update_args /* { 127 struct vnode *a_vp; 128 struct timespec *a_access; 129 struct timespec *a_modify; 130 int a_waitfor; 131 } */ *ap = v; 132 register struct m_ext2fs *fs; 133 struct buf *bp; 134 struct inode *ip; 135 int error; 136 struct timespec ts; 137 caddr_t cp; 138 139 if (ap->a_vp->v_mount->mnt_flag & MNT_RDONLY) 140 return (0); 141 ip = VTOI(ap->a_vp); 142 TIMEVAL_TO_TIMESPEC(&time, &ts); 143 EXT2FS_ITIMES(ip, 144 ap->a_access ? ap->a_access : &ts, 145 ap->a_modify ? ap->a_modify : &ts, &ts); 146 if ((ip->i_flag & IN_MODIFIED) == 0) 147 return (0); 148 ip->i_flag &= ~IN_MODIFIED; 149 fs = ip->i_e2fs; 150 error = bread(ip->i_devvp, 151 fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 152 (int)fs->e2fs_bsize, NOCRED, &bp); 153 if (error) { 154 brelse(bp); 155 return (error); 156 } 157 cp = (caddr_t)bp->b_data + 158 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 159 e2fs_isave(&ip->i_din.e2fs_din, (struct ext2fs_dinode *)cp); 160 if (ap->a_waitfor && (ap->a_vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 161 return (bwrite(bp)); 162 else { 163 bdwrite(bp); 164 return (0); 165 } 166 } 167 168 #define SINGLE 0 /* index of single indirect block */ 169 #define DOUBLE 1 /* index of double indirect block */ 170 #define TRIPLE 2 /* index of triple indirect block */ 171 /* 172 * Truncate the inode oip to at most length size, freeing the 173 * disk blocks. 174 */ 175 int 176 ext2fs_truncate(v) 177 void *v; 178 { 179 struct vop_truncate_args /* { 180 struct vnode *a_vp; 181 off_t a_length; 182 int a_flags; 183 struct ucred *a_cred; 184 struct proc *a_p; 185 } */ *ap = v; 186 register struct vnode *ovp = ap->a_vp; 187 register ufs_daddr_t lastblock; 188 register struct inode *oip; 189 ufs_daddr_t bn, lbn, lastiblock[NIADDR], indir_lbn[NIADDR]; 190 ufs_daddr_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR]; 191 off_t length = ap->a_length; 192 register struct m_ext2fs *fs; 193 struct buf *bp; 194 int offset, size, level; 195 long count, nblocks, vflags, blocksreleased = 0; 196 register int i; 197 int aflags, error, allerror; 198 off_t osize; 199 200 if (length < 0) 201 return (EINVAL); 202 203 oip = VTOI(ovp); 204 if (ovp->v_type == VLNK && 205 (oip->i_e2fs_size < ovp->v_mount->mnt_maxsymlinklen || 206 (ovp->v_mount->mnt_maxsymlinklen == 0 && 207 oip->i_e2fs_nblock == 0))) { 208 #ifdef DIAGNOSTIC 209 if (length != 0) 210 panic("ext2fs_truncate: partial truncate of symlink"); 211 #endif 212 memset((char *)&oip->i_din.e2fs_din.e2di_shortlink, 0, 213 (u_int)oip->i_e2fs_size); 214 oip->i_e2fs_size = 0; 215 oip->i_flag |= IN_CHANGE | IN_UPDATE; 216 return (VOP_UPDATE(ovp, NULL, NULL, 1)); 217 } 218 if (oip->i_e2fs_size == length) { 219 oip->i_flag |= IN_CHANGE | IN_UPDATE; 220 return (VOP_UPDATE(ovp, NULL, NULL, 0)); 221 } 222 fs = oip->i_e2fs; 223 osize = oip->i_e2fs_size; 224 /* 225 * Lengthen the size of the file. We must ensure that the 226 * last byte of the file is allocated. Since the smallest 227 * value of osize is 0, length will be at least 1. 228 */ 229 if (osize < length) { 230 #if 0 /* XXX */ 231 if (length > fs->fs_maxfilesize) 232 return (EFBIG); 233 #endif 234 offset = blkoff(fs, length - 1); 235 lbn = lblkno(fs, length - 1); 236 aflags = B_CLRBUF; 237 if (ap->a_flags & IO_SYNC) 238 aflags |= B_SYNC; 239 error = ext2fs_balloc(oip, lbn, offset + 1, ap->a_cred, &bp, 240 aflags); 241 if (error) 242 return (error); 243 oip->i_e2fs_size = length; 244 uvm_vnp_setsize(ovp, length); 245 (void) uvm_vnp_uncache(ovp); 246 if (aflags & B_SYNC) 247 bwrite(bp); 248 else 249 bawrite(bp); 250 oip->i_flag |= IN_CHANGE | IN_UPDATE; 251 return (VOP_UPDATE(ovp, NULL, NULL, 1)); 252 } 253 /* 254 * Shorten the size of the file. If the file is not being 255 * truncated to a block boundry, the contents of the 256 * partial block following the end of the file must be 257 * zero'ed in case it ever become accessable again because 258 * of subsequent file growth. 259 */ 260 offset = blkoff(fs, length); 261 if (offset == 0) { 262 oip->i_e2fs_size = length; 263 } else { 264 lbn = lblkno(fs, length); 265 aflags = B_CLRBUF; 266 if (ap->a_flags & IO_SYNC) 267 aflags |= B_SYNC; 268 error = ext2fs_balloc(oip, lbn, offset, ap->a_cred, &bp, aflags); 269 if (error) 270 return (error); 271 oip->i_e2fs_size = length; 272 size = fs->e2fs_bsize; 273 (void) uvm_vnp_uncache(ovp); 274 memset((char *)bp->b_data + offset, 0, (u_int)(size - offset)); 275 allocbuf(bp, size); 276 if (aflags & B_SYNC) 277 bwrite(bp); 278 else 279 bawrite(bp); 280 } 281 uvm_vnp_setsize(ovp, length); 282 283 /* 284 * Calculate index into inode's block list of 285 * last direct and indirect blocks (if any) 286 * which we want to keep. Lastblock is -1 when 287 * the file is truncated to 0. 288 */ 289 lastblock = lblkno(fs, length + fs->e2fs_bsize - 1) - 1; 290 lastiblock[SINGLE] = lastblock - NDADDR; 291 lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs); 292 lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs); 293 nblocks = btodb(fs->e2fs_bsize); 294 /* 295 * Update file and block pointers on disk before we start freeing 296 * blocks. If we crash before free'ing blocks below, the blocks 297 * will be returned to the free list. lastiblock values are also 298 * normalized to -1 for calls to ext2fs_indirtrunc below. 299 */ 300 memcpy((caddr_t)oldblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof oldblks); 301 for (level = TRIPLE; level >= SINGLE; level--) 302 if (lastiblock[level] < 0) { 303 oip->i_e2fs_blocks[NDADDR + level] = 0; 304 lastiblock[level] = -1; 305 } 306 for (i = NDADDR - 1; i > lastblock; i--) 307 oip->i_e2fs_blocks[i] = 0; 308 oip->i_flag |= IN_CHANGE | IN_UPDATE; 309 if ((error = VOP_UPDATE(ovp, NULL, NULL, 1)) != 0) 310 allerror = error; 311 /* 312 * Having written the new inode to disk, save its new configuration 313 * and put back the old block pointers long enough to process them. 314 * Note that we save the new block configuration so we can check it 315 * when we are done. 316 */ 317 memcpy((caddr_t)newblks, (caddr_t)&oip->i_e2fs_blocks[0], sizeof newblks); 318 memcpy((caddr_t)&oip->i_e2fs_blocks[0], (caddr_t)oldblks, sizeof oldblks); 319 oip->i_e2fs_size = osize; 320 vflags = ((length > 0) ? V_SAVE : 0) | V_SAVEMETA; 321 allerror = vinvalbuf(ovp, vflags, ap->a_cred, ap->a_p, 0, 0); 322 323 /* 324 * Indirect blocks first. 325 */ 326 indir_lbn[SINGLE] = -NDADDR; 327 indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) -1; 328 indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1; 329 for (level = TRIPLE; level >= SINGLE; level--) { 330 bn = fs2h32(oip->i_e2fs_blocks[NDADDR + level]); 331 if (bn != 0) { 332 error = ext2fs_indirtrunc(oip, indir_lbn[level], 333 fsbtodb(fs, bn), lastiblock[level], level, &count); 334 if (error) 335 allerror = error; 336 blocksreleased += count; 337 if (lastiblock[level] < 0) { 338 oip->i_e2fs_blocks[NDADDR + level] = 0; 339 ext2fs_blkfree(oip, bn); 340 blocksreleased += nblocks; 341 } 342 } 343 if (lastiblock[level] >= 0) 344 goto done; 345 } 346 347 /* 348 * All whole direct blocks or frags. 349 */ 350 for (i = NDADDR - 1; i > lastblock; i--) { 351 bn = fs2h32(oip->i_e2fs_blocks[i]); 352 if (bn == 0) 353 continue; 354 oip->i_e2fs_blocks[i] = 0; 355 ext2fs_blkfree(oip, bn); 356 blocksreleased += btodb(fs->e2fs_bsize); 357 } 358 if (lastblock < 0) 359 goto done; 360 361 done: 362 #ifdef DIAGNOSTIC 363 for (level = SINGLE; level <= TRIPLE; level++) 364 if (newblks[NDADDR + level] != 365 fs2h32(oip->i_e2fs_blocks[NDADDR + level])) 366 panic("itrunc1"); 367 for (i = 0; i < NDADDR; i++) 368 if (newblks[i] != fs2h32(oip->i_e2fs_blocks[i])) 369 panic("itrunc2"); 370 if (length == 0 && 371 (ovp->v_dirtyblkhd.lh_first || ovp->v_cleanblkhd.lh_first)) 372 panic("itrunc3"); 373 #endif /* DIAGNOSTIC */ 374 /* 375 * Put back the real size. 376 */ 377 oip->i_e2fs_size = length; 378 oip->i_e2fs_nblock -= blocksreleased; 379 if (oip->i_e2fs_nblock < 0) /* sanity */ 380 oip->i_e2fs_nblock = 0; 381 oip->i_flag |= IN_CHANGE; 382 return (allerror); 383 } 384 385 /* 386 * Release blocks associated with the inode ip and stored in the indirect 387 * block bn. Blocks are free'd in LIFO order up to (but not including) 388 * lastbn. If level is greater than SINGLE, the block is an indirect block 389 * and recursive calls to indirtrunc must be used to cleanse other indirect 390 * blocks. 391 * 392 * NB: triple indirect blocks are untested. 393 */ 394 static int 395 ext2fs_indirtrunc(ip, lbn, dbn, lastbn, level, countp) 396 register struct inode *ip; 397 ufs_daddr_t lbn, lastbn; 398 ufs_daddr_t dbn; 399 int level; 400 long *countp; 401 { 402 register int i; 403 struct buf *bp; 404 register struct m_ext2fs *fs = ip->i_e2fs; 405 register ufs_daddr_t *bap; 406 struct vnode *vp; 407 ufs_daddr_t *copy = NULL, nb, nlbn, last; 408 long blkcount, factor; 409 int nblocks, blocksreleased = 0; 410 int error = 0, allerror = 0; 411 412 /* 413 * Calculate index in current block of last 414 * block to be kept. -1 indicates the entire 415 * block so we need not calculate the index. 416 */ 417 factor = 1; 418 for (i = SINGLE; i < level; i++) 419 factor *= NINDIR(fs); 420 last = lastbn; 421 if (lastbn > 0) 422 last /= factor; 423 nblocks = btodb(fs->e2fs_bsize); 424 /* 425 * Get buffer of block pointers, zero those entries corresponding 426 * to blocks to be free'd, and update on disk copy first. Since 427 * double(triple) indirect before single(double) indirect, calls 428 * to bmap on these blocks will fail. However, we already have 429 * the on disk address, so we have to set the b_blkno field 430 * explicitly instead of letting bread do everything for us. 431 */ 432 vp = ITOV(ip); 433 bp = getblk(vp, lbn, (int)fs->e2fs_bsize, 0, 0); 434 if (bp->b_flags & (B_DONE | B_DELWRI)) { 435 /* Braces must be here in case trace evaluates to nothing. */ 436 trace(TR_BREADHIT, pack(vp, fs->e2fs_bsize), lbn); 437 } else { 438 trace(TR_BREADMISS, pack(vp, fs->e2fs_bsize), lbn); 439 curproc->p_stats->p_ru.ru_inblock++; /* pay for read */ 440 bp->b_flags |= B_READ; 441 if (bp->b_bcount > bp->b_bufsize) 442 panic("ext2fs_indirtrunc: bad buffer size"); 443 bp->b_blkno = dbn; 444 VOP_STRATEGY(bp); 445 error = biowait(bp); 446 } 447 if (error) { 448 brelse(bp); 449 *countp = 0; 450 return (error); 451 } 452 453 bap = (ufs_daddr_t *)bp->b_data; 454 if (lastbn != -1) { 455 MALLOC(copy, ufs_daddr_t *, fs->e2fs_bsize, M_TEMP, M_WAITOK); 456 memcpy((caddr_t)copy, (caddr_t)bap, (u_int)fs->e2fs_bsize); 457 memset((caddr_t)&bap[last + 1], 0, 458 (u_int)(NINDIR(fs) - (last + 1)) * sizeof (u_int32_t)); 459 error = bwrite(bp); 460 if (error) 461 allerror = error; 462 bap = copy; 463 } 464 465 /* 466 * Recursively free totally unused blocks. 467 */ 468 for (i = NINDIR(fs) - 1, 469 nlbn = lbn + 1 - i * factor; i > last; 470 i--, nlbn += factor) { 471 nb = fs2h32(bap[i]); 472 if (nb == 0) 473 continue; 474 if (level > SINGLE) { 475 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 476 (ufs_daddr_t)-1, level - 1, 477 &blkcount); 478 if (error) 479 allerror = error; 480 blocksreleased += blkcount; 481 } 482 ext2fs_blkfree(ip, nb); 483 blocksreleased += nblocks; 484 } 485 486 /* 487 * Recursively free last partial block. 488 */ 489 if (level > SINGLE && lastbn >= 0) { 490 last = lastbn % factor; 491 nb = fs2h32(bap[i]); 492 if (nb != 0) { 493 error = ext2fs_indirtrunc(ip, nlbn, fsbtodb(fs, nb), 494 last, level - 1, &blkcount); 495 if (error) 496 allerror = error; 497 blocksreleased += blkcount; 498 } 499 } 500 501 if (copy != NULL) { 502 FREE(copy, M_TEMP); 503 } else { 504 bp->b_flags |= B_INVAL; 505 brelse(bp); 506 } 507 508 *countp = blocksreleased; 509 return (allerror); 510 } 511