1 /*-
2 * modified for Lites 1.1
3 *
4 * Aug 1995, Godmar Back (gback@cs.utah.edu)
5 * University of Utah, Department of Computer Science
6 */
7 /*-
8 * Copyright (c) 1982, 1986, 1989, 1993
9 * The Regents of the University of California. All rights reserved.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)ffs_alloc.c 8.8 (Berkeley) 2/21/94
36 * $FreeBSD: releng/9.0/sys/fs/ext2fs/ext2_alloc.c 221126 2011-04-27 18:15:34Z jhb $
37 */
38
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/conf.h>
42 #include <sys/vnode.h>
43 #include <sys/stat.h>
44 #include <sys/mount.h>
45 #include <sys/syslog.h>
46 #include <sys/buf.h>
47
48 #include <fs/ext2fs/inode.h>
49 #include <fs/ext2fs/ext2_mount.h>
50 #include <fs/ext2fs/ext2fs.h>
51 #include <fs/ext2fs/fs.h>
52 #include <fs/ext2fs/ext2_extern.h>
53
54 static daddr_t ext2_alloccg(struct inode *, int, daddr_t, int);
55 static u_long ext2_dirpref(struct inode *);
56 static void ext2_fserr(struct m_ext2fs *, uid_t, char *);
57 static u_long ext2_hashalloc(struct inode *, int, long, int,
58 daddr_t (*)(struct inode *, int, daddr_t,
59 int));
60 static daddr_t ext2_nodealloccg(struct inode *, int, daddr_t, int);
61 static daddr_t ext2_mapsearch(struct m_ext2fs *, char *, daddr_t);
62 #ifdef FANCY_REALLOC
63 static int ext2_reallocblks(struct vop_reallocblks_args *);
64 #endif
65
66 /*
67 * Allocate a block in the file system.
68 *
69 * A preference may be optionally specified. If a preference is given
70 * the following hierarchy is used to allocate a block:
71 * 1) allocate the requested block.
72 * 2) allocate a rotationally optimal block in the same cylinder.
73 * 3) allocate a block in the same cylinder group.
74 * 4) quadradically rehash into other cylinder groups, until an
75 * available block is located.
76 * If no block preference is given the following hierarchy is used
77 * to allocate a block:
78 * 1) allocate a block in the cylinder group that contains the
79 * inode for the file.
80 * 2) quadradically rehash into other cylinder groups, until an
81 * available block is located.
82 */
83 int
84 ext2_alloc(ip, lbn, bpref, size, cred, bnp)
85 struct inode *ip;
86 int32_t lbn, bpref;
87 int size;
88 struct ucred *cred;
89 int32_t *bnp;
90 {
91 struct m_ext2fs *fs;
92 struct ext2mount *ump;
93 int32_t bno;
94 int cg;
95 *bnp = 0;
96 fs = ip->i_e2fs;
97 ump = ip->i_ump;
98 mtx_assert(EXT2_MTX(ump), MA_OWNED);
99 #ifdef DIAGNOSTIC
100 if ((u_int)size > fs->e2fs_bsize || blkoff(fs, size) != 0) {
101 vn_printf(ip->i_devvp, "bsize = %lu, size = %d, fs = %s\n",
102 (long unsigned int)fs->e2fs_bsize, size, fs->e2fs_fsmnt);
103 panic("ext2_alloc: bad size");
104 }
105 if (cred == NOCRED)
106 panic("ext2_alloc: missing credential");
107 #endif /* DIAGNOSTIC */
108 if (size == fs->e2fs_bsize && fs->e2fs->e2fs_fbcount == 0)
109 goto nospace;
110 if (cred->cr_uid != 0 &&
111 fs->e2fs->e2fs_fbcount < fs->e2fs->e2fs_rbcount)
112 goto nospace;
113 if (bpref >= fs->e2fs->e2fs_bcount)
114 bpref = 0;
115 if (bpref == 0)
116 cg = ino_to_cg(fs, ip->i_number);
117 else
118 cg = dtog(fs, bpref);
119 bno = (daddr_t)ext2_hashalloc(ip, cg, bpref, fs->e2fs_bsize,
120 ext2_alloccg);
121 if (bno > 0) {
122 /* set next_alloc fields as done in block_getblk */
123 ip->i_next_alloc_block = lbn;
124 ip->i_next_alloc_goal = bno;
125
126 ip->i_blocks += btodb(fs->e2fs_bsize);
127 ip->i_flag |= IN_CHANGE | IN_UPDATE;
128 *bnp = bno;
129 return (0);
130 }
131 nospace:
132 EXT2_UNLOCK(ump);
133 ext2_fserr(fs, cred->cr_uid, "file system full");
134 uprintf("\n%s: write failed, file system is full\n", fs->e2fs_fsmnt);
135 return (ENOSPC);
136 }
137
138 /*
139 * Reallocate a sequence of blocks into a contiguous sequence of blocks.
140 *
141 * The vnode and an array of buffer pointers for a range of sequential
142 * logical blocks to be made contiguous is given. The allocator attempts
143 * to find a range of sequential blocks starting as close as possible to
144 * an fs_rotdelay offset from the end of the allocation for the logical
145 * block immediately preceding the current range. If successful, the
146 * physical block numbers in the buffer pointers and in the inode are
147 * changed to reflect the new allocation. If unsuccessful, the allocation
148 * is left unchanged. The success in doing the reallocation is returned.
149 * Note that the error return is not reflected back to the user. Rather
150 * the previous block allocation will be used.
151 */
152
153 #ifdef FANCY_REALLOC
154 SYSCTL_NODE(_vfs, OID_AUTO, ext2fs, CTLFLAG_RW, 0, "EXT2FS filesystem");
155
156 static int doasyncfree = 1;
157 SYSCTL_INT(_vfs_ext2fs, OID_AUTO, doasyncfree, CTLFLAG_RW, &doasyncfree, 0,
158 "Use asychronous writes to update block pointers when freeing blocks");
159
160 static int doreallocblks = 1;
161 SYSCTL_INT(_vfs_ext2fs, OID_AUTO, doreallocblks, CTLFLAG_RW, &doreallocblks, 0, "");
162 #endif
163
164 int
165 ext2_reallocblks(ap)
166 struct vop_reallocblks_args /* {
167 struct vnode *a_vp;
168 struct cluster_save *a_buflist;
169 } */ *ap;
170 {
171 #ifndef FANCY_REALLOC
172 /* printf("ext2_reallocblks not implemented\n"); */
173 return ENOSPC;
174 #else
175
176 struct m_ext2fs *fs;
177 struct inode *ip;
178 struct vnode *vp;
179 struct buf *sbp, *ebp;
180 int32_t *bap, *sbap, *ebap = 0;
181 struct ext2mount *ump;
182 struct cluster_save *buflist;
183 struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp;
184 int32_t start_lbn, end_lbn, soff, newblk, blkno;
185 int i, len, start_lvl, end_lvl, pref, ssize;
186
187 vp = ap->a_vp;
188 ip = VTOI(vp);
189 fs = ip->i_e2fs;
190 ump = ip->i_ump;
191 #ifdef UNKLAR
192 if (fs->fs_contigsumsize <= 0)
193 return (ENOSPC);
194 #endif
195 buflist = ap->a_buflist;
196 len = buflist->bs_nchildren;
197 start_lbn = buflist->bs_children[0]->b_lblkno;
198 end_lbn = start_lbn + len - 1;
199 #ifdef DIAGNOSTIC
200 for (i = 1; i < len; i++)
201 if (buflist->bs_children[i]->b_lblkno != start_lbn + i)
202 panic("ext2_reallocblks: non-cluster");
203 #endif
204 /*
205 * If the latest allocation is in a new cylinder group, assume that
206 * the filesystem has decided to move and do not force it back to
207 * the previous cylinder group.
208 */
209 if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) !=
210 dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno)))
211 return (ENOSPC);
212 if (ext2_getlbns(vp, start_lbn, start_ap, &start_lvl) ||
213 ext2_getlbns(vp, end_lbn, end_ap, &end_lvl))
214 return (ENOSPC);
215 /*
216 * Get the starting offset and block map for the first block.
217 */
218 if (start_lvl == 0) {
219 sbap = &ip->i_db[0];
220 soff = start_lbn;
221 } else {
222 idp = &start_ap[start_lvl - 1];
223 if (bread(vp, idp->in_lbn, (int)fs->e2fs_bsize, NOCRED, &sbp)) {
224 brelse(sbp);
225 return (ENOSPC);
226 }
227 sbap = (int32_t *)sbp->b_data;
228 soff = idp->in_off;
229 }
230 /*
231 * Find the preferred location for the cluster.
232 */
233 EXT2_LOCK(ump);
234 pref = ext2_blkpref(ip, start_lbn, soff, sbap, 0);
235 /*
236 * If the block range spans two block maps, get the second map.
237 */
238 if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) {
239 ssize = len;
240 } else {
241 #ifdef DIAGNOSTIC
242 if (start_ap[start_lvl-1].in_lbn == idp->in_lbn)
243 panic("ext2_reallocblk: start == end");
244 #endif
245 ssize = len - (idp->in_off + 1);
246 if (bread(vp, idp->in_lbn, (int)fs->e2fs_bsize, NOCRED, &ebp)){
247 EXT2_UNLOCK(ump);
248 goto fail;
249 }
250 ebap = (int32_t *)ebp->b_data;
251 }
252 /*
253 * Search the block map looking for an allocation of the desired size.
254 */
255 if ((newblk = (int32_t)ext2_hashalloc(ip, dtog(fs, pref), pref,
256 len, ext2_clusteralloc)) == 0){
257 EXT2_UNLOCK(ump);
258 goto fail;
259 }
260 /*
261 * We have found a new contiguous block.
262 *
263 * First we have to replace the old block pointers with the new
264 * block pointers in the inode and indirect blocks associated
265 * with the file.
266 */
267 blkno = newblk;
268 for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->e2fs_fpb) {
269 if (i == ssize)
270 bap = ebap;
271 soff = -i;
272 #ifdef DIAGNOSTIC
273 if (buflist->bs_children[i]->b_blkno != fsbtodb(fs, *bap))
274 panic("ext2_reallocblks: alloc mismatch");
275 #endif
276 *bap++ = blkno;
277 }
278 /*
279 * Next we must write out the modified inode and indirect blocks.
280 * For strict correctness, the writes should be synchronous since
281 * the old block values may have been written to disk. In practise
282 * they are almost never written, but if we are concerned about
283 * strict correctness, the `doasyncfree' flag should be set to zero.
284 *
285 * The test on `doasyncfree' should be changed to test a flag
286 * that shows whether the associated buffers and inodes have
287 * been written. The flag should be set when the cluster is
288 * started and cleared whenever the buffer or inode is flushed.
289 * We can then check below to see if it is set, and do the
290 * synchronous write only when it has been cleared.
291 */
292 if (sbap != &ip->i_db[0]) {
293 if (doasyncfree)
294 bdwrite(sbp);
295 else
296 bwrite(sbp);
297 } else {
298 ip->i_flag |= IN_CHANGE | IN_UPDATE;
299 if (!doasyncfree)
300 ext2_update(vp, 1);
301 }
302 if (ssize < len) {
303 if (doasyncfree)
304 bdwrite(ebp);
305 else
306 bwrite(ebp);
307 }
308 /*
309 * Last, free the old blocks and assign the new blocks to the buffers.
310 */
311 for (blkno = newblk, i = 0; i < len; i++, blkno += fs->e2fs_fpb) {
312 ext2_blkfree(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno),
313 fs->e2fs_bsize);
314 buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno);
315 }
316 return (0);
317
318 fail:
319 if (ssize < len)
320 brelse(ebp);
321 if (sbap != &ip->i_db[0])
322 brelse(sbp);
323 return (ENOSPC);
324
325 #endif /* FANCY_REALLOC */
326 }
327
328 /*
329 * Allocate an inode in the file system.
330 *
331 */
332 int
333 ext2_valloc(pvp, mode, cred, vpp)
334 struct vnode *pvp;
335 int mode;
336 struct ucred *cred;
337 struct vnode **vpp;
338 {
339 struct inode *pip;
340 struct m_ext2fs *fs;
341 struct inode *ip;
342 struct ext2mount *ump;
343 ino_t ino, ipref;
344 int i, error, cg;
345
346 *vpp = NULL;
347 pip = VTOI(pvp);
348 fs = pip->i_e2fs;
349 ump = pip->i_ump;
350
351 EXT2_LOCK(ump);
352 if (fs->e2fs->e2fs_ficount == 0)
353 goto noinodes;
354 /*
355 * If it is a directory then obtain a cylinder group based on
356 * ext2_dirpref else obtain it using ino_to_cg. The preferred inode is
357 * always the next inode.
358 */
359 if((mode & IFMT) == IFDIR) {
360 cg = ext2_dirpref(pip);
361 if (fs->e2fs_contigdirs[cg] < 255)
362 fs->e2fs_contigdirs[cg]++;
363 } else {
364 cg = ino_to_cg(fs, pip->i_number);
365 if (fs->e2fs_contigdirs[cg] > 0)
366 fs->e2fs_contigdirs[cg]--;
367 }
368 ipref = cg * fs->e2fs->e2fs_ipg + 1;
369 ino = (ino_t)ext2_hashalloc(pip, cg, (long)ipref, mode, ext2_nodealloccg);
370
371 if (ino == 0)
372 goto noinodes;
373 error = VFS_VGET(pvp->v_mount, ino, LK_EXCLUSIVE, vpp);
374 if (error) {
375 ext2_vfree(pvp, ino, mode);
376 return (error);
377 }
378 ip = VTOI(*vpp);
379
380 /*
381 the question is whether using VGET was such good idea at all -
382 Linux doesn't read the old inode in when it's allocating a
383 new one. I will set at least i_size & i_blocks the zero.
384 */
385 ip->i_mode = 0;
386 ip->i_size = 0;
387 ip->i_blocks = 0;
388 ip->i_flags = 0;
389 /* now we want to make sure that the block pointers are zeroed out */
390 for (i = 0; i < NDADDR; i++)
391 ip->i_db[i] = 0;
392 for (i = 0; i < NIADDR; i++)
393 ip->i_ib[i] = 0;
394
395 /*
396 * Set up a new generation number for this inode.
397 * XXX check if this makes sense in ext2
398 */
399 if (ip->i_gen == 0 || ++ip->i_gen == 0)
400 ip->i_gen = random() / 2 + 1;
401 /*
402 printf("ext2_valloc: allocated inode %d\n", ino);
403 */
404 return (0);
405 noinodes:
406 EXT2_UNLOCK(ump);
407 ext2_fserr(fs, cred->cr_uid, "out of inodes");
408 uprintf("\n%s: create/symlink failed, no inodes free\n", fs->e2fs_fsmnt);
409 return (ENOSPC);
410 }
411
412 /*
413 * Find a cylinder to place a directory.
414 *
415 * The policy implemented by this algorithm is to allocate a
416 * directory inode in the same cylinder group as its parent
417 * directory, but also to reserve space for its files inodes
418 * and data. Restrict the number of directories which may be
419 * allocated one after another in the same cylinder group
420 * without intervening allocation of files.
421 *
422 * If we allocate a first level directory then force allocation
423 * in another cylinder group.
424 *
425 */
426 static u_long
427 ext2_dirpref(struct inode *pip)
428 {
429 struct m_ext2fs *fs;
430 int cg, prefcg, dirsize, cgsize;
431 int avgifree, avgbfree, avgndir, curdirsize;
432 int minifree, minbfree, maxndir;
433 int mincg, minndir;
434 int maxcontigdirs;
435
436 mtx_assert(EXT2_MTX(pip->i_ump), MA_OWNED);
437 fs = pip->i_e2fs;
438
439 avgifree = fs->e2fs->e2fs_ficount / fs->e2fs_gcount;
440 avgbfree = fs->e2fs->e2fs_fbcount / fs->e2fs_gcount;
441 avgndir = fs->e2fs_total_dir / fs->e2fs_gcount;
442
443 /*
444 * Force allocation in another cg if creating a first level dir.
445 */
446 ASSERT_VOP_LOCKED(ITOV(pip), "ext2fs_dirpref");
447 if (ITOV(pip)->v_vflag & VV_ROOT) {
448 prefcg = arc4random() % fs->e2fs_gcount;
449 mincg = prefcg;
450 minndir = fs->e2fs_ipg;
451 for (cg = prefcg; cg < fs->e2fs_gcount; cg++)
452 if (fs->e2fs_gd[cg].ext2bgd_ndirs < minndir &&
453 fs->e2fs_gd[cg].ext2bgd_nifree >= avgifree &&
454 fs->e2fs_gd[cg].ext2bgd_nbfree >= avgbfree) {
455 mincg = cg;
456 minndir = fs->e2fs_gd[cg].ext2bgd_ndirs;
457 }
458 for (cg = 0; cg < prefcg; cg++)
459 if (fs->e2fs_gd[cg].ext2bgd_ndirs < minndir &&
460 fs->e2fs_gd[cg].ext2bgd_nifree >= avgifree &&
461 fs->e2fs_gd[cg].ext2bgd_nbfree >= avgbfree) {
462 mincg = cg;
463 minndir = fs->e2fs_gd[cg].ext2bgd_ndirs;
464 }
465
466 return (mincg);
467 }
468
469 /*
470 * Count various limits which used for
471 * optimal allocation of a directory inode.
472 */
473 maxndir = min(avgndir + fs->e2fs_ipg / 16, fs->e2fs_ipg);
474 minifree = avgifree - avgifree / 4;
475 if (minifree < 1)
476 minifree = 1;
477 minbfree = avgbfree - avgbfree / 4;
478 if (minbfree < 1)
479 minbfree = 1;
480 cgsize = fs->e2fs_fsize * fs->e2fs_fpg;
481 dirsize = AVGDIRSIZE;
482 curdirsize = avgndir ? (cgsize - avgbfree * fs->e2fs_bsize) / avgndir : 0;
483 if (dirsize < curdirsize)
484 dirsize = curdirsize;
485 if (dirsize <= 0)
486 maxcontigdirs = 0; /* dirsize overflowed */
487 else
488 maxcontigdirs = min((avgbfree * fs->e2fs_bsize) / dirsize, 255);
489 maxcontigdirs = min(maxcontigdirs, fs->e2fs_ipg / AFPDIR);
490 if (maxcontigdirs == 0)
491 maxcontigdirs = 1;
492
493 /*
494 * Limit number of dirs in one cg and reserve space for
495 * regular files, but only if we have no deficit in
496 * inodes or space.
497 */
498 prefcg = ino_to_cg(fs, pip->i_number);
499 for (cg = prefcg; cg < fs->e2fs_gcount; cg++)
500 if (fs->e2fs_gd[cg].ext2bgd_ndirs < maxndir &&
501 fs->e2fs_gd[cg].ext2bgd_nifree >= minifree &&
502 fs->e2fs_gd[cg].ext2bgd_nbfree >= minbfree) {
503 if (fs->e2fs_contigdirs[cg] < maxcontigdirs)
504 return (cg);
505 }
506 for (cg = 0; cg < prefcg; cg++)
507 if (fs->e2fs_gd[cg].ext2bgd_ndirs < maxndir &&
508 fs->e2fs_gd[cg].ext2bgd_nifree >= minifree &&
509 fs->e2fs_gd[cg].ext2bgd_nbfree >= minbfree) {
510 if (fs->e2fs_contigdirs[cg] < maxcontigdirs)
511 return (cg);
512 }
513 /*
514 * This is a backstop when we have deficit in space.
515 */
516 for (cg = prefcg; cg < fs->e2fs_gcount; cg++)
517 if (fs->e2fs_gd[cg].ext2bgd_nifree >= avgifree)
518 return (cg);
519 for (cg = 0; cg < prefcg; cg++)
520 if (fs->e2fs_gd[cg].ext2bgd_nifree >= avgifree)
521 break;
522 return (cg);
523 }
524
525 /*
526 * Select the desired position for the next block in a file.
527 *
528 * we try to mimic what Remy does in inode_getblk/block_getblk
529 *
530 * we note: blocknr == 0 means that we're about to allocate either
531 * a direct block or a pointer block at the first level of indirection
532 * (In other words, stuff that will go in i_db[] or i_ib[])
533 *
534 * blocknr != 0 means that we're allocating a block that is none
535 * of the above. Then, blocknr tells us the number of the block
536 * that will hold the pointer
537 */
538 int32_t
539 ext2_blkpref(ip, lbn, indx, bap, blocknr)
540 struct inode *ip;
541 int32_t lbn;
542 int indx;
543 int32_t *bap;
544 int32_t blocknr;
545 {
546 int tmp;
547 mtx_assert(EXT2_MTX(ip->i_ump), MA_OWNED);
548
549 /* if the next block is actually what we thought it is,
550 then set the goal to what we thought it should be
551 */
552 if(ip->i_next_alloc_block == lbn && ip->i_next_alloc_goal != 0)
553 return ip->i_next_alloc_goal;
554
555 /* now check whether we were provided with an array that basically
556 tells us previous blocks to which we want to stay closeby
557 */
558 if(bap)
559 for (tmp = indx - 1; tmp >= 0; tmp--)
560 if (bap[tmp])
561 return bap[tmp];
562
563 /* else let's fall back to the blocknr, or, if there is none,
564 follow the rule that a block should be allocated near its inode
565 */
566 return blocknr ? blocknr :
567 (int32_t)(ip->i_block_group *
568 EXT2_BLOCKS_PER_GROUP(ip->i_e2fs)) +
569 ip->i_e2fs->e2fs->e2fs_first_dblock;
570 }
571
572 /*
573 * Implement the cylinder overflow algorithm.
574 *
575 * The policy implemented by this algorithm is:
576 * 1) allocate the block in its requested cylinder group.
577 * 2) quadradically rehash on the cylinder group number.
578 * 3) brute force search for a free block.
579 */
580 static u_long
581 ext2_hashalloc(struct inode *ip, int cg, long pref, int size,
582 daddr_t (*allocator)(struct inode *, int, daddr_t, int))
583 {
584 struct m_ext2fs *fs;
585 ino_t result;
586 int i, icg = cg;
587
588 mtx_assert(EXT2_MTX(ip->i_ump), MA_OWNED);
589 fs = ip->i_e2fs;
590 /*
591 * 1: preferred cylinder group
592 */
593 result = (*allocator)(ip, cg, pref, size);
594 if (result)
595 return (result);
596 /*
597 * 2: quadratic rehash
598 */
599 for (i = 1; i < fs->e2fs_gcount; i *= 2) {
600 cg += i;
601 if (cg >= fs->e2fs_gcount)
602 cg -= fs->e2fs_gcount;
603 result = (*allocator)(ip, cg, 0, size);
604 if (result)
605 return (result);
606 }
607 /*
608 * 3: brute force search
609 * Note that we start at i == 2, since 0 was checked initially,
610 * and 1 is always checked in the quadratic rehash.
611 */
612 cg = (icg + 2) % fs->e2fs_gcount;
613 for (i = 2; i < fs->e2fs_gcount; i++) {
614 result = (*allocator)(ip, cg, 0, size);
615 if (result)
616 return (result);
617 cg++;
618 if (cg == fs->e2fs_gcount)
619 cg = 0;
620 }
621 return (0);
622 }
623
624 /*
625 * Determine whether a block can be allocated.
626 *
627 * Check to see if a block of the appropriate size is available,
628 * and if it is, allocate it.
629 */
630 static daddr_t
631 ext2_alloccg(struct inode *ip, int cg, daddr_t bpref, int size)
632 {
633 struct m_ext2fs *fs;
634 struct buf *bp;
635 struct ext2mount *ump;
636 daddr_t bno, runstart, runlen;
637 int bit, loc, end, error, start;
638 char *bbp;
639 /* XXX ondisk32 */
640 fs = ip->i_e2fs;
641 ump = ip->i_ump;
642 if (fs->e2fs_gd[cg].ext2bgd_nbfree == 0)
643 return (0);
644 EXT2_UNLOCK(ump);
645 error = bread(ip->i_devvp, fsbtodb(fs,
646 fs->e2fs_gd[cg].ext2bgd_b_bitmap),
647 (int)fs->e2fs_bsize, NOCRED, &bp);
648 if (error) {
649 brelse(bp);
650 EXT2_LOCK(ump);
651 return (0);
652 }
653 if (fs->e2fs_gd[cg].ext2bgd_nbfree == 0) {
654 /*
655 * Another thread allocated the last block in this
656 * group while we were waiting for the buffer.
657 */
658 brelse(bp);
659 EXT2_LOCK(ump);
660 return (0);
661 }
662 bbp = (char *)bp->b_data;
663
664 if (dtog(fs, bpref) != cg)
665 bpref = 0;
666 if (bpref != 0) {
667 bpref = dtogd(fs, bpref);
668 /*
669 * if the requested block is available, use it
670 */
671 if (isclr(bbp, bpref)) {
672 bno = bpref;
673 goto gotit;
674 }
675 }
676 /*
677 * no blocks in the requested cylinder, so take next
678 * available one in this cylinder group.
679 * first try to get 8 contigous blocks, then fall back to a single
680 * block.
681 */
682 if (bpref)
683 start = dtogd(fs, bpref) / NBBY;
684 else
685 start = 0;
686 end = howmany(fs->e2fs->e2fs_fpg, NBBY) - start;
687 retry:
688 runlen = 0;
689 runstart = 0;
690 for (loc = start; loc < end; loc++) {
691 if (bbp[loc] == (char)0xff) {
692 runlen = 0;
693 continue;
694 }
695
696 /* Start of a run, find the number of high clear bits. */
697 if (runlen == 0) {
698 bit = fls(bbp[loc]);
699 runlen = NBBY - bit;
700 runstart = loc * NBBY + bit;
701 } else if (bbp[loc] == 0) {
702 /* Continue a run. */
703 runlen += NBBY;
704 } else {
705 /*
706 * Finish the current run. If it isn't long
707 * enough, start a new one.
708 */
709 bit = ffs(bbp[loc]) - 1;
710 runlen += bit;
711 if (runlen >= 8) {
712 bno = runstart;
713 goto gotit;
714 }
715
716 /* Run was too short, start a new one. */
717 bit = fls(bbp[loc]);
718 runlen = NBBY - bit;
719 runstart = loc * NBBY + bit;
720 }
721
722 /* If the current run is long enough, use it. */
723 if (runlen >= 8) {
724 bno = runstart;
725 goto gotit;
726 }
727 }
728 if (start != 0) {
729 end = start;
730 start = 0;
731 goto retry;
732 }
733
734 bno = ext2_mapsearch(fs, bbp, bpref);
735 if (bno < 0){
736 brelse(bp);
737 EXT2_LOCK(ump);
738 return (0);
739 }
740 gotit:
741 #ifdef DIAGNOSTIC
742 if (isset(bbp, bno)) {
743 printf("ext2fs_alloccgblk: cg=%d bno=%jd fs=%s\n",
744 cg, (intmax_t)bno, fs->e2fs_fsmnt);
745 panic("ext2fs_alloccg: dup alloc");
746 }
747 #endif
748 setbit(bbp, bno);
749 EXT2_LOCK(ump);
750 fs->e2fs->e2fs_fbcount--;
751 fs->e2fs_gd[cg].ext2bgd_nbfree--;
752 fs->e2fs_fmod = 1;
753 EXT2_UNLOCK(ump);
754 bdwrite(bp);
755 return (cg * fs->e2fs->e2fs_fpg + fs->e2fs->e2fs_first_dblock + bno);
756 }
757
758 /*
759 * Determine whether an inode can be allocated.
760 *
761 * Check to see if an inode is available, and if it is,
762 * allocate it using tode in the specified cylinder group.
763 */
764 static daddr_t
765 ext2_nodealloccg(struct inode *ip, int cg, daddr_t ipref, int mode)
766 {
767 struct m_ext2fs *fs;
768 struct buf *bp;
769 struct ext2mount *ump;
770 int error, start, len, loc, map, i;
771 char *ibp;
772 ipref--; /* to avoid a lot of (ipref -1) */
773 if (ipref == -1)
774 ipref = 0;
775 fs = ip->i_e2fs;
776 ump = ip->i_ump;
777 if (fs->e2fs_gd[cg].ext2bgd_nifree == 0)
778 return (0);
779 EXT2_UNLOCK(ump);
780 error = bread(ip->i_devvp, fsbtodb(fs,
781 fs->e2fs_gd[cg].ext2bgd_i_bitmap),
782 (int)fs->e2fs_bsize, NOCRED, &bp);
783 if (error) {
784 brelse(bp);
785 EXT2_LOCK(ump);
786 return (0);
787 }
788 if (fs->e2fs_gd[cg].ext2bgd_nifree == 0) {
789 /*
790 * Another thread allocated the last i-node in this
791 * group while we were waiting for the buffer.
792 */
793 brelse(bp);
794 EXT2_LOCK(ump);
795 return (0);
796 }
797 ibp = (char *)bp->b_data;
798 if (ipref) {
799 ipref %= fs->e2fs->e2fs_ipg;
800 if (isclr(ibp, ipref))
801 goto gotit;
802 }
803 start = ipref / NBBY;
804 len = howmany(fs->e2fs->e2fs_ipg - ipref, NBBY);
805 loc = skpc(0xff, len, &ibp[start]);
806 if (loc == 0) {
807 len = start + 1;
808 start = 0;
809 loc = skpc(0xff, len, &ibp[0]);
810 if (loc == 0) {
811 printf("cg = %d, ipref = %lld, fs = %s\n",
812 cg, (long long)ipref, fs->e2fs_fsmnt);
813 panic("ext2fs_nodealloccg: map corrupted");
814 /* NOTREACHED */
815 }
816 }
817 i = start + len - loc;
818 map = ibp[i] ^ 0xff;
819 if (map == 0) {
820 printf("fs = %s\n", fs->e2fs_fsmnt);
821 panic("ext2fs_nodealloccg: block not in map");
822 }
823 ipref = i * NBBY + ffs(map) - 1;
824 gotit:
825 setbit(ibp, ipref);
826 EXT2_LOCK(ump);
827 fs->e2fs_gd[cg].ext2bgd_nifree--;
828 fs->e2fs->e2fs_ficount--;
829 fs->e2fs_fmod = 1;
830 if ((mode & IFMT) == IFDIR) {
831 fs->e2fs_gd[cg].ext2bgd_ndirs++;
832 fs->e2fs_total_dir++;
833 }
834 EXT2_UNLOCK(ump);
835 bdwrite(bp);
836 return (cg * fs->e2fs->e2fs_ipg + ipref +1);
837 }
838
839 /*
840 * Free a block or fragment.
841 *
842 */
843 void
844 ext2_blkfree(ip, bno, size)
845 struct inode *ip;
846 int32_t bno;
847 long size;
848 {
849 struct m_ext2fs *fs;
850 struct buf *bp;
851 struct ext2mount *ump;
852 int cg, error;
853 char *bbp;
854
855 fs = ip->i_e2fs;
856 ump = ip->i_ump;
857 cg = dtog(fs, bno);
858 if ((u_int)bno >= fs->e2fs->e2fs_bcount) {
859 printf("bad block %lld, ino %llu\n", (long long)bno,
860 (unsigned long long)ip->i_number);
861 ext2_fserr(fs, ip->i_uid, "bad block");
862 return;
863 }
864 error = bread(ip->i_devvp,
865 fsbtodb(fs, fs->e2fs_gd[cg].ext2bgd_b_bitmap),
866 (int)fs->e2fs_bsize, NOCRED, &bp);
867 if (error) {
868 brelse(bp);
869 return;
870 }
871 bbp = (char *)bp->b_data;
872 bno = dtogd(fs, bno);
873 if (isclr(bbp, bno)) {
874 printf("block = %lld, fs = %s\n",
875 (long long)bno, fs->e2fs_fsmnt);
876 panic("blkfree: freeing free block");
877 }
878 clrbit(bbp, bno);
879 EXT2_LOCK(ump);
880 fs->e2fs->e2fs_fbcount++;
881 fs->e2fs_gd[cg].ext2bgd_nbfree++;
882 fs->e2fs_fmod = 1;
883 EXT2_UNLOCK(ump);
884 bdwrite(bp);
885 }
886
887 /*
888 * Free an inode.
889 *
890 */
891 int
892 ext2_vfree(pvp, ino, mode)
893 struct vnode *pvp;
894 ino_t ino;
895 int mode;
896 {
897 struct m_ext2fs *fs;
898 struct inode *pip;
899 struct buf *bp;
900 struct ext2mount *ump;
901 int error, cg;
902 char * ibp;
903 /* mode_t save_i_mode; */
904
905 pip = VTOI(pvp);
906 fs = pip->i_e2fs;
907 ump = pip->i_ump;
908 if ((u_int)ino > fs->e2fs_ipg * fs->e2fs_gcount)
909 panic("ext2_vfree: range: devvp = %p, ino = %d, fs = %s",
910 pip->i_devvp, ino, fs->e2fs_fsmnt);
911
912 cg = ino_to_cg(fs, ino);
913 error = bread(pip->i_devvp,
914 fsbtodb(fs, fs->e2fs_gd[cg].ext2bgd_i_bitmap),
915 (int)fs->e2fs_bsize, NOCRED, &bp);
916 if (error) {
917 brelse(bp);
918 return (0);
919 }
920 ibp = (char *)bp->b_data;
921 ino = (ino - 1) % fs->e2fs->e2fs_ipg;
922 if (isclr(ibp, ino)) {
923 printf("ino = %llu, fs = %s\n",
924 (unsigned long long)ino, fs->e2fs_fsmnt);
925 if (fs->e2fs_ronly == 0)
926 panic("ifree: freeing free inode");
927 }
928 clrbit(ibp, ino);
929 EXT2_LOCK(ump);
930 fs->e2fs->e2fs_ficount++;
931 fs->e2fs_gd[cg].ext2bgd_nifree++;
932 if ((mode & IFMT) == IFDIR) {
933 fs->e2fs_gd[cg].ext2bgd_ndirs--;
934 fs->e2fs_total_dir--;
935 }
936 fs->e2fs_fmod = 1;
937 EXT2_UNLOCK(ump);
938 bdwrite(bp);
939 return (0);
940 }
941
942 /*
943 * Find a block in the specified cylinder group.
944 *
945 * It is a panic if a request is made to find a block if none are
946 * available.
947 */
948 static daddr_t
949 ext2_mapsearch(struct m_ext2fs *fs, char *bbp, daddr_t bpref)
950 {
951 int start, len, loc, i, map;
952
953 /*
954 * find the fragment by searching through the free block
955 * map for an appropriate bit pattern
956 */
957 if (bpref)
958 start = dtogd(fs, bpref) / NBBY;
959 else
960 start = 0;
961 len = howmany(fs->e2fs->e2fs_fpg, NBBY) - start;
962 loc = skpc(0xff, len, &bbp[start]);
963 if (loc == 0) {
964 len = start + 1;
965 start = 0;
966 loc = skpc(0xff, len, &bbp[start]);
967 if (loc == 0) {
968 printf("start = %d, len = %d, fs = %s\n",
969 start, len, fs->e2fs_fsmnt);
970 panic("ext2fs_alloccg: map corrupted");
971 /* NOTREACHED */
972 }
973 }
974 i = start + len - loc;
975 map = bbp[i] ^ 0xff;
976 if (map == 0) {
977 printf("fs = %s\n", fs->e2fs_fsmnt);
978 panic("ext2fs_mapsearch: block not in map");
979 }
980 return (i * NBBY + ffs(map) - 1);
981 }
982
983 /*
984 * Fserr prints the name of a file system with an error diagnostic.
985 *
986 * The form of the error message is:
987 * fs: error message
988 */
989 static void
990 ext2_fserr(fs, uid, cp)
991 struct m_ext2fs *fs;
992 uid_t uid;
993 char *cp;
994 {
995
996 log(LOG_ERR, "uid %u on %s: %s\n", uid, fs->e2fs_fsmnt, cp);
997 }
998
999 int
1000 cg_has_sb(int i)
1001 {
1002 int a3, a5, a7;
1003
1004 if (i == 0 || i == 1)
1005 return 1;
1006 for (a3 = 3, a5 = 5, a7 = 7;
1007 a3 <= i || a5 <= i || a7 <= i;
1008 a3 *= 3, a5 *= 5, a7 *= 7)
1009 if (i == a3 || i == a5 || i == a7)
1010 return 1;
1011 return 0;
1012 }
Cache object: 41d6774ec28080146b887dc33c7d25b5
|