1 /*
2 * Copyright (c) 1982, 1986, 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * %sccs.include.redist.c%
6 *
7 * @(#)ffs_alloc.c 8.18 (Berkeley) 05/26/95
8 */
9
10 #include <sys/param.h>
11 #include <sys/systm.h>
12 #include <sys/buf.h>
13 #include <sys/proc.h>
14 #include <sys/vnode.h>
15 #include <sys/mount.h>
16 #include <sys/kernel.h>
17 #include <sys/syslog.h>
18
19 #include <vm/vm.h>
20
21 #include <ufs/ufs/quota.h>
22 #include <ufs/ufs/inode.h>
23
24 #include <ufs/ffs/fs.h>
25 #include <ufs/ffs/ffs_extern.h>
26
27 extern u_long nextgennumber;
28
29 static ufs_daddr_t ffs_alloccg __P((struct inode *, int, ufs_daddr_t, int));
30 static ufs_daddr_t ffs_alloccgblk __P((struct fs *, struct cg *, ufs_daddr_t));
31 static ufs_daddr_t ffs_clusteralloc __P((struct inode *, int, ufs_daddr_t,
32 int));
33 static ino_t ffs_dirpref __P((struct fs *));
34 static ufs_daddr_t ffs_fragextend __P((struct inode *, int, long, int, int));
35 static void ffs_fserr __P((struct fs *, u_int, char *));
36 static u_long ffs_hashalloc
37 __P((struct inode *, int, long, int, u_int32_t (*)()));
38 static ino_t ffs_nodealloccg __P((struct inode *, int, ufs_daddr_t, int));
39 static ufs_daddr_t ffs_mapsearch __P((struct fs *, struct cg *, ufs_daddr_t,
40 int));
41
42 /*
43 * Allocate a block in the file system.
44 *
45 * The size of the requested block is given, which must be some
46 * multiple of fs_fsize and <= fs_bsize.
47 * A preference may be optionally specified. If a preference is given
48 * the following hierarchy is used to allocate a block:
49 * 1) allocate the requested block.
50 * 2) allocate a rotationally optimal block in the same cylinder.
51 * 3) allocate a block in the same cylinder group.
52 * 4) quadradically rehash into other cylinder groups, until an
53 * available block is located.
54 * If no block preference is given the following heirarchy is used
55 * to allocate a block:
56 * 1) allocate a block in the cylinder group that contains the
57 * inode for the file.
58 * 2) quadradically rehash into other cylinder groups, until an
59 * available block is located.
60 */
ffs_alloc(ip,lbn,bpref,size,cred,bnp)61 ffs_alloc(ip, lbn, bpref, size, cred, bnp)
62 register struct inode *ip;
63 ufs_daddr_t lbn, bpref;
64 int size;
65 struct ucred *cred;
66 ufs_daddr_t *bnp;
67 {
68 register struct fs *fs;
69 ufs_daddr_t bno;
70 int cg, error;
71
72 *bnp = 0;
73 fs = ip->i_fs;
74 #ifdef DIAGNOSTIC
75 if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) {
76 printf("dev = 0x%x, bsize = %d, size = %d, fs = %s\n",
77 ip->i_dev, fs->fs_bsize, size, fs->fs_fsmnt);
78 panic("ffs_alloc: bad size");
79 }
80 if (cred == NOCRED)
81 panic("ffs_alloc: missing credential\n");
82 #endif /* DIAGNOSTIC */
83 if (size == fs->fs_bsize && fs->fs_cstotal.cs_nbfree == 0)
84 goto nospace;
85 if (cred->cr_uid != 0 && freespace(fs, fs->fs_minfree) <= 0)
86 goto nospace;
87 #ifdef QUOTA
88 if (error = chkdq(ip, (long)btodb(size), cred, 0))
89 return (error);
90 #endif
91 if (bpref >= fs->fs_size)
92 bpref = 0;
93 if (bpref == 0)
94 cg = ino_to_cg(fs, ip->i_number);
95 else
96 cg = dtog(fs, bpref);
97 bno = (ufs_daddr_t)ffs_hashalloc(ip, cg, (long)bpref, size,
98 (u_int32_t (*)())ffs_alloccg);
99 if (bno > 0) {
100 ip->i_blocks += btodb(size);
101 ip->i_flag |= IN_CHANGE | IN_UPDATE;
102 *bnp = bno;
103 return (0);
104 }
105 #ifdef QUOTA
106 /*
107 * Restore user's disk quota because allocation failed.
108 */
109 (void) chkdq(ip, (long)-btodb(size), cred, FORCE);
110 #endif
111 nospace:
112 ffs_fserr(fs, cred->cr_uid, "file system full");
113 uprintf("\n%s: write failed, file system is full\n", fs->fs_fsmnt);
114 return (ENOSPC);
115 }
116
117 /*
118 * Reallocate a fragment to a bigger size
119 *
120 * The number and size of the old block is given, and a preference
121 * and new size is also specified. The allocator attempts to extend
122 * the original block. Failing that, the regular block allocator is
123 * invoked to get an appropriate block.
124 */
ffs_realloccg(ip,lbprev,bpref,osize,nsize,cred,bpp)125 ffs_realloccg(ip, lbprev, bpref, osize, nsize, cred, bpp)
126 register struct inode *ip;
127 ufs_daddr_t lbprev;
128 ufs_daddr_t bpref;
129 int osize, nsize;
130 struct ucred *cred;
131 struct buf **bpp;
132 {
133 register struct fs *fs;
134 struct buf *bp;
135 int cg, request, error;
136 ufs_daddr_t bprev, bno;
137
138 *bpp = 0;
139 fs = ip->i_fs;
140 #ifdef DIAGNOSTIC
141 if ((u_int)osize > fs->fs_bsize || fragoff(fs, osize) != 0 ||
142 (u_int)nsize > fs->fs_bsize || fragoff(fs, nsize) != 0) {
143 printf(
144 "dev = 0x%x, bsize = %d, osize = %d, nsize = %d, fs = %s\n",
145 ip->i_dev, fs->fs_bsize, osize, nsize, fs->fs_fsmnt);
146 panic("ffs_realloccg: bad size");
147 }
148 if (cred == NOCRED)
149 panic("ffs_realloccg: missing credential\n");
150 #endif /* DIAGNOSTIC */
151 if (cred->cr_uid != 0 && freespace(fs, fs->fs_minfree) <= 0)
152 goto nospace;
153 if ((bprev = ip->i_db[lbprev]) == 0) {
154 printf("dev = 0x%x, bsize = %d, bprev = %d, fs = %s\n",
155 ip->i_dev, fs->fs_bsize, bprev, fs->fs_fsmnt);
156 panic("ffs_realloccg: bad bprev");
157 }
158 /*
159 * Allocate the extra space in the buffer.
160 */
161 if (error = bread(ITOV(ip), lbprev, osize, NOCRED, &bp)) {
162 brelse(bp);
163 return (error);
164 }
165 #ifdef QUOTA
166 if (error = chkdq(ip, (long)btodb(nsize - osize), cred, 0)) {
167 brelse(bp);
168 return (error);
169 }
170 #endif
171 /*
172 * Check for extension in the existing location.
173 */
174 cg = dtog(fs, bprev);
175 if (bno = ffs_fragextend(ip, cg, (long)bprev, osize, nsize)) {
176 if (bp->b_blkno != fsbtodb(fs, bno))
177 panic("bad blockno");
178 ip->i_blocks += btodb(nsize - osize);
179 ip->i_flag |= IN_CHANGE | IN_UPDATE;
180 allocbuf(bp, nsize);
181 bp->b_flags |= B_DONE;
182 bzero((char *)bp->b_data + osize, (u_int)nsize - osize);
183 *bpp = bp;
184 return (0);
185 }
186 /*
187 * Allocate a new disk location.
188 */
189 if (bpref >= fs->fs_size)
190 bpref = 0;
191 switch ((int)fs->fs_optim) {
192 case FS_OPTSPACE:
193 /*
194 * Allocate an exact sized fragment. Although this makes
195 * best use of space, we will waste time relocating it if
196 * the file continues to grow. If the fragmentation is
197 * less than half of the minimum free reserve, we choose
198 * to begin optimizing for time.
199 */
200 request = nsize;
201 if (fs->fs_minfree < 5 ||
202 fs->fs_cstotal.cs_nffree >
203 fs->fs_dsize * fs->fs_minfree / (2 * 100))
204 break;
205 log(LOG_NOTICE, "%s: optimization changed from SPACE to TIME\n",
206 fs->fs_fsmnt);
207 fs->fs_optim = FS_OPTTIME;
208 break;
209 case FS_OPTTIME:
210 /*
211 * At this point we have discovered a file that is trying to
212 * grow a small fragment to a larger fragment. To save time,
213 * we allocate a full sized block, then free the unused portion.
214 * If the file continues to grow, the `ffs_fragextend' call
215 * above will be able to grow it in place without further
216 * copying. If aberrant programs cause disk fragmentation to
217 * grow within 2% of the free reserve, we choose to begin
218 * optimizing for space.
219 */
220 request = fs->fs_bsize;
221 if (fs->fs_cstotal.cs_nffree <
222 fs->fs_dsize * (fs->fs_minfree - 2) / 100)
223 break;
224 log(LOG_NOTICE, "%s: optimization changed from TIME to SPACE\n",
225 fs->fs_fsmnt);
226 fs->fs_optim = FS_OPTSPACE;
227 break;
228 default:
229 printf("dev = 0x%x, optim = %d, fs = %s\n",
230 ip->i_dev, fs->fs_optim, fs->fs_fsmnt);
231 panic("ffs_realloccg: bad optim");
232 /* NOTREACHED */
233 }
234 bno = (ufs_daddr_t)ffs_hashalloc(ip, cg, (long)bpref, request,
235 (u_int32_t (*)())ffs_alloccg);
236 if (bno > 0) {
237 bp->b_blkno = fsbtodb(fs, bno);
238 (void) vnode_pager_uncache(ITOV(ip));
239 ffs_blkfree(ip, bprev, (long)osize);
240 if (nsize < request)
241 ffs_blkfree(ip, bno + numfrags(fs, nsize),
242 (long)(request - nsize));
243 ip->i_blocks += btodb(nsize - osize);
244 ip->i_flag |= IN_CHANGE | IN_UPDATE;
245 allocbuf(bp, nsize);
246 bp->b_flags |= B_DONE;
247 bzero((char *)bp->b_data + osize, (u_int)nsize - osize);
248 *bpp = bp;
249 return (0);
250 }
251 #ifdef QUOTA
252 /*
253 * Restore user's disk quota because allocation failed.
254 */
255 (void) chkdq(ip, (long)-btodb(nsize - osize), cred, FORCE);
256 #endif
257 brelse(bp);
258 nospace:
259 /*
260 * no space available
261 */
262 ffs_fserr(fs, cred->cr_uid, "file system full");
263 uprintf("\n%s: write failed, file system is full\n", fs->fs_fsmnt);
264 return (ENOSPC);
265 }
266
267 /*
268 * Reallocate a sequence of blocks into a contiguous sequence of blocks.
269 *
270 * The vnode and an array of buffer pointers for a range of sequential
271 * logical blocks to be made contiguous is given. The allocator attempts
272 * to find a range of sequential blocks starting as close as possible to
273 * an fs_rotdelay offset from the end of the allocation for the logical
274 * block immediately preceeding the current range. If successful, the
275 * physical block numbers in the buffer pointers and in the inode are
276 * changed to reflect the new allocation. If unsuccessful, the allocation
277 * is left unchanged. The success in doing the reallocation is returned.
278 * Note that the error return is not reflected back to the user. Rather
279 * the previous block allocation will be used.
280 */
281 int doasyncfree = 1;
282 int doreallocblks = 1;
283 int prtrealloc = 0;
284
285 int
ffs_reallocblks(ap)286 ffs_reallocblks(ap)
287 struct vop_reallocblks_args /* {
288 struct vnode *a_vp;
289 struct cluster_save *a_buflist;
290 } */ *ap;
291 {
292 struct fs *fs;
293 struct inode *ip;
294 struct vnode *vp;
295 struct buf *sbp, *ebp;
296 ufs_daddr_t *bap, *sbap, *ebap;
297 struct cluster_save *buflist;
298 ufs_daddr_t start_lbn, end_lbn, soff, eoff, newblk, blkno;
299 struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp;
300 int i, len, start_lvl, end_lvl, pref, ssize;
301
302 if (doreallocblks == 0)
303 return (ENOSPC);
304 vp = ap->a_vp;
305 ip = VTOI(vp);
306 fs = ip->i_fs;
307 if (fs->fs_contigsumsize <= 0)
308 return (ENOSPC);
309 buflist = ap->a_buflist;
310 len = buflist->bs_nchildren;
311 start_lbn = buflist->bs_children[0]->b_lblkno;
312 end_lbn = start_lbn + len - 1;
313 #ifdef DIAGNOSTIC
314 for (i = 0; i < len; i++)
315 if (!ffs_checkblk(ip,
316 dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
317 panic("ffs_reallocblks: unallocated block 1");
318 for (i = 1; i < len; i++)
319 if (buflist->bs_children[i]->b_lblkno != start_lbn + i)
320 panic("ffs_reallocblks: non-logical cluster");
321 blkno = buflist->bs_children[0]->b_blkno;
322 ssize = fsbtodb(fs, fs->fs_frag);
323 for (i = 1; i < len - 1; i++)
324 if (buflist->bs_children[i]->b_blkno != blkno + (i * ssize))
325 panic("ffs_reallocblks: non-physical cluster %d", i);
326 #endif
327 /*
328 * If the latest allocation is in a new cylinder group, assume that
329 * the filesystem has decided to move and do not force it back to
330 * the previous cylinder group.
331 */
332 if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) !=
333 dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno)))
334 return (ENOSPC);
335 if (ufs_getlbns(vp, start_lbn, start_ap, &start_lvl) ||
336 ufs_getlbns(vp, end_lbn, end_ap, &end_lvl))
337 return (ENOSPC);
338 /*
339 * Get the starting offset and block map for the first block.
340 */
341 if (start_lvl == 0) {
342 sbap = &ip->i_db[0];
343 soff = start_lbn;
344 } else {
345 idp = &start_ap[start_lvl - 1];
346 if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &sbp)) {
347 brelse(sbp);
348 return (ENOSPC);
349 }
350 sbap = (ufs_daddr_t *)sbp->b_data;
351 soff = idp->in_off;
352 }
353 /*
354 * Find the preferred location for the cluster.
355 */
356 pref = ffs_blkpref(ip, start_lbn, soff, sbap);
357 /*
358 * If the block range spans two block maps, get the second map.
359 */
360 if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) {
361 ssize = len;
362 } else {
363 #ifdef DIAGNOSTIC
364 if (start_ap[start_lvl-1].in_lbn == idp->in_lbn)
365 panic("ffs_reallocblk: start == end");
366 #endif
367 ssize = len - (idp->in_off + 1);
368 if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &ebp))
369 goto fail;
370 ebap = (ufs_daddr_t *)ebp->b_data;
371 }
372 /*
373 * Search the block map looking for an allocation of the desired size.
374 */
375 if ((newblk = (ufs_daddr_t)ffs_hashalloc(ip, dtog(fs, pref), (long)pref,
376 len, (u_int32_t (*)())ffs_clusteralloc)) == 0)
377 goto fail;
378 /*
379 * We have found a new contiguous block.
380 *
381 * First we have to replace the old block pointers with the new
382 * block pointers in the inode and indirect blocks associated
383 * with the file.
384 */
385 #ifdef DEBUG
386 if (prtrealloc)
387 printf("realloc: ino %d, lbns %d-%d\n\told:", ip->i_number,
388 start_lbn, end_lbn);
389 #endif
390 blkno = newblk;
391 for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->fs_frag) {
392 if (i == ssize)
393 bap = ebap;
394 #ifdef DIAGNOSTIC
395 if (!ffs_checkblk(ip,
396 dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
397 panic("ffs_reallocblks: unallocated block 2");
398 if (dbtofsb(fs, buflist->bs_children[i]->b_blkno) != *bap)
399 panic("ffs_reallocblks: alloc mismatch");
400 #endif
401 #ifdef DEBUG
402 if (prtrealloc)
403 printf(" %d,", *bap);
404 #endif
405 *bap++ = blkno;
406 }
407 /*
408 * Next we must write out the modified inode and indirect blocks.
409 * For strict correctness, the writes should be synchronous since
410 * the old block values may have been written to disk. In practise
411 * they are almost never written, but if we are concerned about
412 * strict correctness, the `doasyncfree' flag should be set to zero.
413 *
414 * The test on `doasyncfree' should be changed to test a flag
415 * that shows whether the associated buffers and inodes have
416 * been written. The flag should be set when the cluster is
417 * started and cleared whenever the buffer or inode is flushed.
418 * We can then check below to see if it is set, and do the
419 * synchronous write only when it has been cleared.
420 */
421 if (sbap != &ip->i_db[0]) {
422 if (doasyncfree)
423 bdwrite(sbp);
424 else
425 bwrite(sbp);
426 } else {
427 ip->i_flag |= IN_CHANGE | IN_UPDATE;
428 if (!doasyncfree)
429 VOP_UPDATE(vp, &time, &time, MNT_WAIT);
430 }
431 if (ssize < len)
432 if (doasyncfree)
433 bdwrite(ebp);
434 else
435 bwrite(ebp);
436 /*
437 * Last, free the old blocks and assign the new blocks to the buffers.
438 */
439 #ifdef DEBUG
440 if (prtrealloc)
441 printf("\n\tnew:");
442 #endif
443 for (blkno = newblk, i = 0; i < len; i++, blkno += fs->fs_frag) {
444 ffs_blkfree(ip, dbtofsb(fs, buflist->bs_children[i]->b_blkno),
445 fs->fs_bsize);
446 buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno);
447 #ifdef DEBUG
448 if (!ffs_checkblk(ip,
449 dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
450 panic("ffs_reallocblks: unallocated block 3");
451 if (prtrealloc)
452 printf(" %d,", blkno);
453 #endif
454 }
455 #ifdef DEBUG
456 if (prtrealloc) {
457 prtrealloc--;
458 printf("\n");
459 }
460 #endif
461 return (0);
462
463 fail:
464 if (ssize < len)
465 brelse(ebp);
466 if (sbap != &ip->i_db[0])
467 brelse(sbp);
468 return (ENOSPC);
469 }
470
471 /*
472 * Allocate an inode in the file system.
473 *
474 * If allocating a directory, use ffs_dirpref to select the inode.
475 * If allocating in a directory, the following hierarchy is followed:
476 * 1) allocate the preferred inode.
477 * 2) allocate an inode in the same cylinder group.
478 * 3) quadradically rehash into other cylinder groups, until an
479 * available inode is located.
480 * If no inode preference is given the following heirarchy is used
481 * to allocate an inode:
482 * 1) allocate an inode in cylinder group 0.
483 * 2) quadradically rehash into other cylinder groups, until an
484 * available inode is located.
485 */
486 ffs_valloc(ap)
487 struct vop_valloc_args /* {
488 struct vnode *a_pvp;
489 int a_mode;
490 struct ucred *a_cred;
491 struct vnode **a_vpp;
492 } */ *ap;
493 {
494 register struct vnode *pvp = ap->a_pvp;
495 register struct inode *pip;
496 register struct fs *fs;
497 register struct inode *ip;
498 mode_t mode = ap->a_mode;
499 ino_t ino, ipref;
500 int cg, error;
501
502 *ap->a_vpp = NULL;
503 pip = VTOI(pvp);
504 fs = pip->i_fs;
505 if (fs->fs_cstotal.cs_nifree == 0)
506 goto noinodes;
507
508 if ((mode & IFMT) == IFDIR)
509 ipref = ffs_dirpref(fs);
510 else
511 ipref = pip->i_number;
512 if (ipref >= fs->fs_ncg * fs->fs_ipg)
513 ipref = 0;
514 cg = ino_to_cg(fs, ipref);
515 ino = (ino_t)ffs_hashalloc(pip, cg, (long)ipref, mode, ffs_nodealloccg);
516 if (ino == 0)
517 goto noinodes;
518 error = VFS_VGET(pvp->v_mount, ino, ap->a_vpp);
519 if (error) {
520 VOP_VFREE(pvp, ino, mode);
521 return (error);
522 }
523 ip = VTOI(*ap->a_vpp);
524 if (ip->i_mode) {
525 printf("mode = 0%o, inum = %d, fs = %s\n",
526 ip->i_mode, ip->i_number, fs->fs_fsmnt);
527 panic("ffs_valloc: dup alloc");
528 }
529 if (ip->i_blocks) { /* XXX */
530 printf("free inode %s/%d had %d blocks\n",
531 fs->fs_fsmnt, ino, ip->i_blocks);
532 ip->i_blocks = 0;
533 }
534 ip->i_flags = 0;
535 /*
536 * Set up a new generation number for this inode.
537 */
538 if (++nextgennumber < (u_long)time.tv_sec)
539 nextgennumber = time.tv_sec;
540 ip->i_gen = nextgennumber;
541 return (0);
542 noinodes:
543 ffs_fserr(fs, ap->a_cred->cr_uid, "out of inodes");
544 uprintf("\n%s: create/symlink failed, no inodes free\n", fs->fs_fsmnt);
545 return (ENOSPC);
546 }
547
548 /*
549 * Find a cylinder to place a directory.
550 *
551 * The policy implemented by this algorithm is to select from
552 * among those cylinder groups with above the average number of
553 * free inodes, the one with the smallest number of directories.
554 */
555 static ino_t
ffs_dirpref(fs)556 ffs_dirpref(fs)
557 register struct fs *fs;
558 {
559 int cg, minndir, mincg, avgifree;
560
561 avgifree = fs->fs_cstotal.cs_nifree / fs->fs_ncg;
562 minndir = fs->fs_ipg;
563 mincg = 0;
564 for (cg = 0; cg < fs->fs_ncg; cg++)
565 if (fs->fs_cs(fs, cg).cs_ndir < minndir &&
566 fs->fs_cs(fs, cg).cs_nifree >= avgifree) {
567 mincg = cg;
568 minndir = fs->fs_cs(fs, cg).cs_ndir;
569 }
570 return ((ino_t)(fs->fs_ipg * mincg));
571 }
572
573 /*
574 * Select the desired position for the next block in a file. The file is
575 * logically divided into sections. The first section is composed of the
576 * direct blocks. Each additional section contains fs_maxbpg blocks.
577 *
578 * If no blocks have been allocated in the first section, the policy is to
579 * request a block in the same cylinder group as the inode that describes
580 * the file. If no blocks have been allocated in any other section, the
581 * policy is to place the section in a cylinder group with a greater than
582 * average number of free blocks. An appropriate cylinder group is found
583 * by using a rotor that sweeps the cylinder groups. When a new group of
584 * blocks is needed, the sweep begins in the cylinder group following the
585 * cylinder group from which the previous allocation was made. The sweep
586 * continues until a cylinder group with greater than the average number
587 * of free blocks is found. If the allocation is for the first block in an
588 * indirect block, the information on the previous allocation is unavailable;
589 * here a best guess is made based upon the logical block number being
590 * allocated.
591 *
592 * If a section is already partially allocated, the policy is to
593 * contiguously allocate fs_maxcontig blocks. The end of one of these
594 * contiguous blocks and the beginning of the next is physically separated
595 * so that the disk head will be in transit between them for at least
596 * fs_rotdelay milliseconds. This is to allow time for the processor to
597 * schedule another I/O transfer.
598 */
599 ufs_daddr_t
ffs_blkpref(ip,lbn,indx,bap)600 ffs_blkpref(ip, lbn, indx, bap)
601 struct inode *ip;
602 ufs_daddr_t lbn;
603 int indx;
604 ufs_daddr_t *bap;
605 {
606 register struct fs *fs;
607 register int cg;
608 int avgbfree, startcg;
609 ufs_daddr_t nextblk;
610
611 fs = ip->i_fs;
612 if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
613 if (lbn < NDADDR) {
614 cg = ino_to_cg(fs, ip->i_number);
615 return (fs->fs_fpg * cg + fs->fs_frag);
616 }
617 /*
618 * Find a cylinder with greater than average number of
619 * unused data blocks.
620 */
621 if (indx == 0 || bap[indx - 1] == 0)
622 startcg =
623 ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
624 else
625 startcg = dtog(fs, bap[indx - 1]) + 1;
626 startcg %= fs->fs_ncg;
627 avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
628 for (cg = startcg; cg < fs->fs_ncg; cg++)
629 if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
630 fs->fs_cgrotor = cg;
631 return (fs->fs_fpg * cg + fs->fs_frag);
632 }
633 for (cg = 0; cg <= startcg; cg++)
634 if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
635 fs->fs_cgrotor = cg;
636 return (fs->fs_fpg * cg + fs->fs_frag);
637 }
638 return (NULL);
639 }
640 /*
641 * One or more previous blocks have been laid out. If less
642 * than fs_maxcontig previous blocks are contiguous, the
643 * next block is requested contiguously, otherwise it is
644 * requested rotationally delayed by fs_rotdelay milliseconds.
645 */
646 nextblk = bap[indx - 1] + fs->fs_frag;
647 if (indx < fs->fs_maxcontig || bap[indx - fs->fs_maxcontig] +
648 blkstofrags(fs, fs->fs_maxcontig) != nextblk)
649 return (nextblk);
650 if (fs->fs_rotdelay != 0)
651 /*
652 * Here we convert ms of delay to frags as:
653 * (frags) = (ms) * (rev/sec) * (sect/rev) /
654 * ((sect/frag) * (ms/sec))
655 * then round up to the next block.
656 */
657 nextblk += roundup(fs->fs_rotdelay * fs->fs_rps * fs->fs_nsect /
658 (NSPF(fs) * 1000), fs->fs_frag);
659 return (nextblk);
660 }
661
662 /*
663 * Implement the cylinder overflow algorithm.
664 *
665 * The policy implemented by this algorithm is:
666 * 1) allocate the block in its requested cylinder group.
667 * 2) quadradically rehash on the cylinder group number.
668 * 3) brute force search for a free block.
669 */
670 /*VARARGS5*/
671 static u_long
ffs_hashalloc(ip,cg,pref,size,allocator)672 ffs_hashalloc(ip, cg, pref, size, allocator)
673 struct inode *ip;
674 int cg;
675 long pref;
676 int size; /* size for data blocks, mode for inodes */
677 u_int32_t (*allocator)();
678 {
679 register struct fs *fs;
680 long result;
681 int i, icg = cg;
682
683 fs = ip->i_fs;
684 /*
685 * 1: preferred cylinder group
686 */
687 result = (*allocator)(ip, cg, pref, size);
688 if (result)
689 return (result);
690 /*
691 * 2: quadratic rehash
692 */
693 for (i = 1; i < fs->fs_ncg; i *= 2) {
694 cg += i;
695 if (cg >= fs->fs_ncg)
696 cg -= fs->fs_ncg;
697 result = (*allocator)(ip, cg, 0, size);
698 if (result)
699 return (result);
700 }
701 /*
702 * 3: brute force search
703 * Note that we start at i == 2, since 0 was checked initially,
704 * and 1 is always checked in the quadratic rehash.
705 */
706 cg = (icg + 2) % fs->fs_ncg;
707 for (i = 2; i < fs->fs_ncg; i++) {
708 result = (*allocator)(ip, cg, 0, size);
709 if (result)
710 return (result);
711 cg++;
712 if (cg == fs->fs_ncg)
713 cg = 0;
714 }
715 return (NULL);
716 }
717
718 /*
719 * Determine whether a fragment can be extended.
720 *
721 * Check to see if the necessary fragments are available, and
722 * if they are, allocate them.
723 */
724 static ufs_daddr_t
ffs_fragextend(ip,cg,bprev,osize,nsize)725 ffs_fragextend(ip, cg, bprev, osize, nsize)
726 struct inode *ip;
727 int cg;
728 long bprev;
729 int osize, nsize;
730 {
731 register struct fs *fs;
732 register struct cg *cgp;
733 struct buf *bp;
734 long bno;
735 int frags, bbase;
736 int i, error;
737
738 fs = ip->i_fs;
739 if (fs->fs_cs(fs, cg).cs_nffree < numfrags(fs, nsize - osize))
740 return (NULL);
741 frags = numfrags(fs, nsize);
742 bbase = fragnum(fs, bprev);
743 if (bbase > fragnum(fs, (bprev + frags - 1))) {
744 /* cannot extend across a block boundary */
745 return (NULL);
746 }
747 error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
748 (int)fs->fs_cgsize, NOCRED, &bp);
749 if (error) {
750 brelse(bp);
751 return (NULL);
752 }
753 cgp = (struct cg *)bp->b_data;
754 if (!cg_chkmagic(cgp)) {
755 brelse(bp);
756 return (NULL);
757 }
758 cgp->cg_time = time.tv_sec;
759 bno = dtogd(fs, bprev);
760 for (i = numfrags(fs, osize); i < frags; i++)
761 if (isclr(cg_blksfree(cgp), bno + i)) {
762 brelse(bp);
763 return (NULL);
764 }
765 /*
766 * the current fragment can be extended
767 * deduct the count on fragment being extended into
768 * increase the count on the remaining fragment (if any)
769 * allocate the extended piece
770 */
771 for (i = frags; i < fs->fs_frag - bbase; i++)
772 if (isclr(cg_blksfree(cgp), bno + i))
773 break;
774 cgp->cg_frsum[i - numfrags(fs, osize)]--;
775 if (i != frags)
776 cgp->cg_frsum[i - frags]++;
777 for (i = numfrags(fs, osize); i < frags; i++) {
778 clrbit(cg_blksfree(cgp), bno + i);
779 cgp->cg_cs.cs_nffree--;
780 fs->fs_cstotal.cs_nffree--;
781 fs->fs_cs(fs, cg).cs_nffree--;
782 }
783 fs->fs_fmod = 1;
784 bdwrite(bp);
785 return (bprev);
786 }
787
788 /*
789 * Determine whether a block can be allocated.
790 *
791 * Check to see if a block of the appropriate size is available,
792 * and if it is, allocate it.
793 */
794 static ufs_daddr_t
ffs_alloccg(ip,cg,bpref,size)795 ffs_alloccg(ip, cg, bpref, size)
796 struct inode *ip;
797 int cg;
798 ufs_daddr_t bpref;
799 int size;
800 {
801 register struct fs *fs;
802 register struct cg *cgp;
803 struct buf *bp;
804 register int i;
805 int error, bno, frags, allocsiz;
806
807 fs = ip->i_fs;
808 if (fs->fs_cs(fs, cg).cs_nbfree == 0 && size == fs->fs_bsize)
809 return (NULL);
810 error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
811 (int)fs->fs_cgsize, NOCRED, &bp);
812 if (error) {
813 brelse(bp);
814 return (NULL);
815 }
816 cgp = (struct cg *)bp->b_data;
817 if (!cg_chkmagic(cgp) ||
818 (cgp->cg_cs.cs_nbfree == 0 && size == fs->fs_bsize)) {
819 brelse(bp);
820 return (NULL);
821 }
822 cgp->cg_time = time.tv_sec;
823 if (size == fs->fs_bsize) {
824 bno = ffs_alloccgblk(fs, cgp, bpref);
825 bdwrite(bp);
826 return (bno);
827 }
828 /*
829 * check to see if any fragments are already available
830 * allocsiz is the size which will be allocated, hacking
831 * it down to a smaller size if necessary
832 */
833 frags = numfrags(fs, size);
834 for (allocsiz = frags; allocsiz < fs->fs_frag; allocsiz++)
835 if (cgp->cg_frsum[allocsiz] != 0)
836 break;
837 if (allocsiz == fs->fs_frag) {
838 /*
839 * no fragments were available, so a block will be
840 * allocated, and hacked up
841 */
842 if (cgp->cg_cs.cs_nbfree == 0) {
843 brelse(bp);
844 return (NULL);
845 }
846 bno = ffs_alloccgblk(fs, cgp, bpref);
847 bpref = dtogd(fs, bno);
848 for (i = frags; i < fs->fs_frag; i++)
849 setbit(cg_blksfree(cgp), bpref + i);
850 i = fs->fs_frag - frags;
851 cgp->cg_cs.cs_nffree += i;
852 fs->fs_cstotal.cs_nffree += i;
853 fs->fs_cs(fs, cg).cs_nffree += i;
854 fs->fs_fmod = 1;
855 cgp->cg_frsum[i]++;
856 bdwrite(bp);
857 return (bno);
858 }
859 bno = ffs_mapsearch(fs, cgp, bpref, allocsiz);
860 if (bno < 0) {
861 brelse(bp);
862 return (NULL);
863 }
864 for (i = 0; i < frags; i++)
865 clrbit(cg_blksfree(cgp), bno + i);
866 cgp->cg_cs.cs_nffree -= frags;
867 fs->fs_cstotal.cs_nffree -= frags;
868 fs->fs_cs(fs, cg).cs_nffree -= frags;
869 fs->fs_fmod = 1;
870 cgp->cg_frsum[allocsiz]--;
871 if (frags != allocsiz)
872 cgp->cg_frsum[allocsiz - frags]++;
873 bdwrite(bp);
874 return (cg * fs->fs_fpg + bno);
875 }
876
877 /*
878 * Allocate a block in a cylinder group.
879 *
880 * This algorithm implements the following policy:
881 * 1) allocate the requested block.
882 * 2) allocate a rotationally optimal block in the same cylinder.
883 * 3) allocate the next available block on the block rotor for the
884 * specified cylinder group.
885 * Note that this routine only allocates fs_bsize blocks; these
886 * blocks may be fragmented by the routine that allocates them.
887 */
888 static ufs_daddr_t
ffs_alloccgblk(fs,cgp,bpref)889 ffs_alloccgblk(fs, cgp, bpref)
890 register struct fs *fs;
891 register struct cg *cgp;
892 ufs_daddr_t bpref;
893 {
894 ufs_daddr_t bno, blkno;
895 int cylno, pos, delta;
896 short *cylbp;
897 register int i;
898
899 if (bpref == 0 || dtog(fs, bpref) != cgp->cg_cgx) {
900 bpref = cgp->cg_rotor;
901 goto norot;
902 }
903 bpref = blknum(fs, bpref);
904 bpref = dtogd(fs, bpref);
905 /*
906 * if the requested block is available, use it
907 */
908 if (ffs_isblock(fs, cg_blksfree(cgp), fragstoblks(fs, bpref))) {
909 bno = bpref;
910 goto gotit;
911 }
912 if (fs->fs_nrpos <= 1 || fs->fs_cpc == 0) {
913 /*
914 * Block layout information is not available.
915 * Leaving bpref unchanged means we take the
916 * next available free block following the one
917 * we just allocated. Hopefully this will at
918 * least hit a track cache on drives of unknown
919 * geometry (e.g. SCSI).
920 */
921 goto norot;
922 }
923 /*
924 * check for a block available on the same cylinder
925 */
926 cylno = cbtocylno(fs, bpref);
927 if (cg_blktot(cgp)[cylno] == 0)
928 goto norot;
929 /*
930 * check the summary information to see if a block is
931 * available in the requested cylinder starting at the
932 * requested rotational position and proceeding around.
933 */
934 cylbp = cg_blks(fs, cgp, cylno);
935 pos = cbtorpos(fs, bpref);
936 for (i = pos; i < fs->fs_nrpos; i++)
937 if (cylbp[i] > 0)
938 break;
939 if (i == fs->fs_nrpos)
940 for (i = 0; i < pos; i++)
941 if (cylbp[i] > 0)
942 break;
943 if (cylbp[i] > 0) {
944 /*
945 * found a rotational position, now find the actual
946 * block. A panic if none is actually there.
947 */
948 pos = cylno % fs->fs_cpc;
949 bno = (cylno - pos) * fs->fs_spc / NSPB(fs);
950 if (fs_postbl(fs, pos)[i] == -1) {
951 printf("pos = %d, i = %d, fs = %s\n",
952 pos, i, fs->fs_fsmnt);
953 panic("ffs_alloccgblk: cyl groups corrupted");
954 }
955 for (i = fs_postbl(fs, pos)[i];; ) {
956 if (ffs_isblock(fs, cg_blksfree(cgp), bno + i)) {
957 bno = blkstofrags(fs, (bno + i));
958 goto gotit;
959 }
960 delta = fs_rotbl(fs)[i];
961 if (delta <= 0 ||
962 delta + i > fragstoblks(fs, fs->fs_fpg))
963 break;
964 i += delta;
965 }
966 printf("pos = %d, i = %d, fs = %s\n", pos, i, fs->fs_fsmnt);
967 panic("ffs_alloccgblk: can't find blk in cyl");
968 }
969 norot:
970 /*
971 * no blocks in the requested cylinder, so take next
972 * available one in this cylinder group.
973 */
974 bno = ffs_mapsearch(fs, cgp, bpref, (int)fs->fs_frag);
975 if (bno < 0)
976 return (NULL);
977 cgp->cg_rotor = bno;
978 gotit:
979 blkno = fragstoblks(fs, bno);
980 ffs_clrblock(fs, cg_blksfree(cgp), (long)blkno);
981 ffs_clusteracct(fs, cgp, blkno, -1);
982 cgp->cg_cs.cs_nbfree--;
983 fs->fs_cstotal.cs_nbfree--;
984 fs->fs_cs(fs, cgp->cg_cgx).cs_nbfree--;
985 cylno = cbtocylno(fs, bno);
986 cg_blks(fs, cgp, cylno)[cbtorpos(fs, bno)]--;
987 cg_blktot(cgp)[cylno]--;
988 fs->fs_fmod = 1;
989 return (cgp->cg_cgx * fs->fs_fpg + bno);
990 }
991
992 /*
993 * Determine whether a cluster can be allocated.
994 *
995 * We do not currently check for optimal rotational layout if there
996 * are multiple choices in the same cylinder group. Instead we just
997 * take the first one that we find following bpref.
998 */
999 static ufs_daddr_t
ffs_clusteralloc(ip,cg,bpref,len)1000 ffs_clusteralloc(ip, cg, bpref, len)
1001 struct inode *ip;
1002 int cg;
1003 ufs_daddr_t bpref;
1004 int len;
1005 {
1006 register struct fs *fs;
1007 register struct cg *cgp;
1008 struct buf *bp;
1009 int i, got, run, bno, bit, map;
1010 u_char *mapp;
1011 int32_t *lp;
1012
1013 fs = ip->i_fs;
1014 if (fs->fs_maxcluster[cg] < len)
1015 return (NULL);
1016 if (bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize,
1017 NOCRED, &bp))
1018 goto fail;
1019 cgp = (struct cg *)bp->b_data;
1020 if (!cg_chkmagic(cgp))
1021 goto fail;
1022 /*
1023 * Check to see if a cluster of the needed size (or bigger) is
1024 * available in this cylinder group.
1025 */
1026 lp = &cg_clustersum(cgp)[len];
1027 for (i = len; i <= fs->fs_contigsumsize; i++)
1028 if (*lp++ > 0)
1029 break;
1030 if (i > fs->fs_contigsumsize) {
1031 /*
1032 * This is the first time looking for a cluster in this
1033 * cylinder group. Update the cluster summary information
1034 * to reflect the true maximum sized cluster so that
1035 * future cluster allocation requests can avoid reading
1036 * the cylinder group map only to find no clusters.
1037 */
1038 lp = &cg_clustersum(cgp)[len - 1];
1039 for (i = len - 1; i > 0; i--)
1040 if (*lp-- > 0)
1041 break;
1042 fs->fs_maxcluster[cg] = i;
1043 goto fail;
1044 }
1045 /*
1046 * Search the cluster map to find a big enough cluster.
1047 * We take the first one that we find, even if it is larger
1048 * than we need as we prefer to get one close to the previous
1049 * block allocation. We do not search before the current
1050 * preference point as we do not want to allocate a block
1051 * that is allocated before the previous one (as we will
1052 * then have to wait for another pass of the elevator
1053 * algorithm before it will be read). We prefer to fail and
1054 * be recalled to try an allocation in the next cylinder group.
1055 */
1056 if (dtog(fs, bpref) != cg)
1057 bpref = 0;
1058 else
1059 bpref = fragstoblks(fs, dtogd(fs, blknum(fs, bpref)));
1060 mapp = &cg_clustersfree(cgp)[bpref / NBBY];
1061 map = *mapp++;
1062 bit = 1 << (bpref % NBBY);
1063 for (run = 0, got = bpref; got < cgp->cg_nclusterblks; got++) {
1064 if ((map & bit) == 0) {
1065 run = 0;
1066 } else {
1067 run++;
1068 if (run == len)
1069 break;
1070 }
1071 if ((got & (NBBY - 1)) != (NBBY - 1)) {
1072 bit <<= 1;
1073 } else {
1074 map = *mapp++;
1075 bit = 1;
1076 }
1077 }
1078 if (got == cgp->cg_nclusterblks)
1079 goto fail;
1080 /*
1081 * Allocate the cluster that we have found.
1082 */
1083 for (i = 1; i <= len; i++)
1084 if (!ffs_isblock(fs, cg_blksfree(cgp), got - run + i))
1085 panic("ffs_clusteralloc: map mismatch");
1086 bno = cg * fs->fs_fpg + blkstofrags(fs, got - run + 1);
1087 if (dtog(fs, bno) != cg)
1088 panic("ffs_clusteralloc: allocated out of group");
1089 len = blkstofrags(fs, len);
1090 for (i = 0; i < len; i += fs->fs_frag)
1091 if ((got = ffs_alloccgblk(fs, cgp, bno + i)) != bno + i)
1092 panic("ffs_clusteralloc: lost block");
1093 brelse(bp);
1094 return (bno);
1095
1096 fail:
1097 brelse(bp);
1098 return (0);
1099 }
1100
1101 /*
1102 * Determine whether an inode can be allocated.
1103 *
1104 * Check to see if an inode is available, and if it is,
1105 * allocate it using the following policy:
1106 * 1) allocate the requested inode.
1107 * 2) allocate the next available inode after the requested
1108 * inode in the specified cylinder group.
1109 */
1110 static ino_t
ffs_nodealloccg(ip,cg,ipref,mode)1111 ffs_nodealloccg(ip, cg, ipref, mode)
1112 struct inode *ip;
1113 int cg;
1114 ufs_daddr_t ipref;
1115 int mode;
1116 {
1117 register struct fs *fs;
1118 register struct cg *cgp;
1119 struct buf *bp;
1120 int error, start, len, loc, map, i;
1121
1122 fs = ip->i_fs;
1123 if (fs->fs_cs(fs, cg).cs_nifree == 0)
1124 return (NULL);
1125 error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1126 (int)fs->fs_cgsize, NOCRED, &bp);
1127 if (error) {
1128 brelse(bp);
1129 return (NULL);
1130 }
1131 cgp = (struct cg *)bp->b_data;
1132 if (!cg_chkmagic(cgp) || cgp->cg_cs.cs_nifree == 0) {
1133 brelse(bp);
1134 return (NULL);
1135 }
1136 cgp->cg_time = time.tv_sec;
1137 if (ipref) {
1138 ipref %= fs->fs_ipg;
1139 if (isclr(cg_inosused(cgp), ipref))
1140 goto gotit;
1141 }
1142 start = cgp->cg_irotor / NBBY;
1143 len = howmany(fs->fs_ipg - cgp->cg_irotor, NBBY);
1144 loc = skpc(0xff, len, &cg_inosused(cgp)[start]);
1145 if (loc == 0) {
1146 len = start + 1;
1147 start = 0;
1148 loc = skpc(0xff, len, &cg_inosused(cgp)[0]);
1149 if (loc == 0) {
1150 printf("cg = %d, irotor = %d, fs = %s\n",
1151 cg, cgp->cg_irotor, fs->fs_fsmnt);
1152 panic("ffs_nodealloccg: map corrupted");
1153 /* NOTREACHED */
1154 }
1155 }
1156 i = start + len - loc;
1157 map = cg_inosused(cgp)[i];
1158 ipref = i * NBBY;
1159 for (i = 1; i < (1 << NBBY); i <<= 1, ipref++) {
1160 if ((map & i) == 0) {
1161 cgp->cg_irotor = ipref;
1162 goto gotit;
1163 }
1164 }
1165 printf("fs = %s\n", fs->fs_fsmnt);
1166 panic("ffs_nodealloccg: block not in map");
1167 /* NOTREACHED */
1168 gotit:
1169 setbit(cg_inosused(cgp), ipref);
1170 cgp->cg_cs.cs_nifree--;
1171 fs->fs_cstotal.cs_nifree--;
1172 fs->fs_cs(fs, cg).cs_nifree--;
1173 fs->fs_fmod = 1;
1174 if ((mode & IFMT) == IFDIR) {
1175 cgp->cg_cs.cs_ndir++;
1176 fs->fs_cstotal.cs_ndir++;
1177 fs->fs_cs(fs, cg).cs_ndir++;
1178 }
1179 bdwrite(bp);
1180 return (cg * fs->fs_ipg + ipref);
1181 }
1182
1183 /*
1184 * Free a block or fragment.
1185 *
1186 * The specified block or fragment is placed back in the
1187 * free map. If a fragment is deallocated, a possible
1188 * block reassembly is checked.
1189 */
ffs_blkfree(ip,bno,size)1190 ffs_blkfree(ip, bno, size)
1191 register struct inode *ip;
1192 ufs_daddr_t bno;
1193 long size;
1194 {
1195 register struct fs *fs;
1196 register struct cg *cgp;
1197 struct buf *bp;
1198 ufs_daddr_t blkno;
1199 int i, error, cg, blk, frags, bbase;
1200
1201 fs = ip->i_fs;
1202 if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) {
1203 printf("dev = 0x%x, bsize = %d, size = %d, fs = %s\n",
1204 ip->i_dev, fs->fs_bsize, size, fs->fs_fsmnt);
1205 panic("blkfree: bad size");
1206 }
1207 cg = dtog(fs, bno);
1208 if ((u_int)bno >= fs->fs_size) {
1209 printf("bad block %d, ino %d\n", bno, ip->i_number);
1210 ffs_fserr(fs, ip->i_uid, "bad block");
1211 return;
1212 }
1213 error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1214 (int)fs->fs_cgsize, NOCRED, &bp);
1215 if (error) {
1216 brelse(bp);
1217 return;
1218 }
1219 cgp = (struct cg *)bp->b_data;
1220 if (!cg_chkmagic(cgp)) {
1221 brelse(bp);
1222 return;
1223 }
1224 cgp->cg_time = time.tv_sec;
1225 bno = dtogd(fs, bno);
1226 if (size == fs->fs_bsize) {
1227 blkno = fragstoblks(fs, bno);
1228 if (ffs_isblock(fs, cg_blksfree(cgp), blkno)) {
1229 printf("dev = 0x%x, block = %d, fs = %s\n",
1230 ip->i_dev, bno, fs->fs_fsmnt);
1231 panic("blkfree: freeing free block");
1232 }
1233 ffs_setblock(fs, cg_blksfree(cgp), blkno);
1234 ffs_clusteracct(fs, cgp, blkno, 1);
1235 cgp->cg_cs.cs_nbfree++;
1236 fs->fs_cstotal.cs_nbfree++;
1237 fs->fs_cs(fs, cg).cs_nbfree++;
1238 i = cbtocylno(fs, bno);
1239 cg_blks(fs, cgp, i)[cbtorpos(fs, bno)]++;
1240 cg_blktot(cgp)[i]++;
1241 } else {
1242 bbase = bno - fragnum(fs, bno);
1243 /*
1244 * decrement the counts associated with the old frags
1245 */
1246 blk = blkmap(fs, cg_blksfree(cgp), bbase);
1247 ffs_fragacct(fs, blk, cgp->cg_frsum, -1);
1248 /*
1249 * deallocate the fragment
1250 */
1251 frags = numfrags(fs, size);
1252 for (i = 0; i < frags; i++) {
1253 if (isset(cg_blksfree(cgp), bno + i)) {
1254 printf("dev = 0x%x, block = %d, fs = %s\n",
1255 ip->i_dev, bno + i, fs->fs_fsmnt);
1256 panic("blkfree: freeing free frag");
1257 }
1258 setbit(cg_blksfree(cgp), bno + i);
1259 }
1260 cgp->cg_cs.cs_nffree += i;
1261 fs->fs_cstotal.cs_nffree += i;
1262 fs->fs_cs(fs, cg).cs_nffree += i;
1263 /*
1264 * add back in counts associated with the new frags
1265 */
1266 blk = blkmap(fs, cg_blksfree(cgp), bbase);
1267 ffs_fragacct(fs, blk, cgp->cg_frsum, 1);
1268 /*
1269 * if a complete block has been reassembled, account for it
1270 */
1271 blkno = fragstoblks(fs, bbase);
1272 if (ffs_isblock(fs, cg_blksfree(cgp), blkno)) {
1273 cgp->cg_cs.cs_nffree -= fs->fs_frag;
1274 fs->fs_cstotal.cs_nffree -= fs->fs_frag;
1275 fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag;
1276 ffs_clusteracct(fs, cgp, blkno, 1);
1277 cgp->cg_cs.cs_nbfree++;
1278 fs->fs_cstotal.cs_nbfree++;
1279 fs->fs_cs(fs, cg).cs_nbfree++;
1280 i = cbtocylno(fs, bbase);
1281 cg_blks(fs, cgp, i)[cbtorpos(fs, bbase)]++;
1282 cg_blktot(cgp)[i]++;
1283 }
1284 }
1285 fs->fs_fmod = 1;
1286 bdwrite(bp);
1287 }
1288
1289 #ifdef DIAGNOSTIC
1290 /*
1291 * Verify allocation of a block or fragment. Returns true if block or
1292 * fragment is allocated, false if it is free.
1293 */
1294 ffs_checkblk(ip, bno, size)
1295 struct inode *ip;
1296 ufs_daddr_t bno;
1297 long size;
1298 {
1299 struct fs *fs;
1300 struct cg *cgp;
1301 struct buf *bp;
1302 int i, error, frags, free;
1303
1304 fs = ip->i_fs;
1305 if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) {
1306 printf("bsize = %d, size = %d, fs = %s\n",
1307 fs->fs_bsize, size, fs->fs_fsmnt);
1308 panic("checkblk: bad size");
1309 }
1310 if ((u_int)bno >= fs->fs_size)
1311 panic("checkblk: bad block %d", bno);
1312 error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, dtog(fs, bno))),
1313 (int)fs->fs_cgsize, NOCRED, &bp);
1314 if (error) {
1315 brelse(bp);
1316 return;
1317 }
1318 cgp = (struct cg *)bp->b_data;
1319 if (!cg_chkmagic(cgp)) {
1320 brelse(bp);
1321 return;
1322 }
1323 bno = dtogd(fs, bno);
1324 if (size == fs->fs_bsize) {
1325 free = ffs_isblock(fs, cg_blksfree(cgp), fragstoblks(fs, bno));
1326 } else {
1327 frags = numfrags(fs, size);
1328 for (free = 0, i = 0; i < frags; i++)
1329 if (isset(cg_blksfree(cgp), bno + i))
1330 free++;
1331 if (free != 0 && free != frags)
1332 panic("checkblk: partially free fragment");
1333 }
1334 brelse(bp);
1335 return (!free);
1336 }
1337 #endif /* DIAGNOSTIC */
1338
1339 /*
1340 * Free an inode.
1341 *
1342 * The specified inode is placed back in the free map.
1343 */
1344 int
ffs_vfree(ap)1345 ffs_vfree(ap)
1346 struct vop_vfree_args /* {
1347 struct vnode *a_pvp;
1348 ino_t a_ino;
1349 int a_mode;
1350 } */ *ap;
1351 {
1352 register struct fs *fs;
1353 register struct cg *cgp;
1354 register struct inode *pip;
1355 ino_t ino = ap->a_ino;
1356 struct buf *bp;
1357 int error, cg;
1358
1359 pip = VTOI(ap->a_pvp);
1360 fs = pip->i_fs;
1361 if ((u_int)ino >= fs->fs_ipg * fs->fs_ncg)
1362 panic("ifree: range: dev = 0x%x, ino = %d, fs = %s\n",
1363 pip->i_dev, ino, fs->fs_fsmnt);
1364 cg = ino_to_cg(fs, ino);
1365 error = bread(pip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1366 (int)fs->fs_cgsize, NOCRED, &bp);
1367 if (error) {
1368 brelse(bp);
1369 return (0);
1370 }
1371 cgp = (struct cg *)bp->b_data;
1372 if (!cg_chkmagic(cgp)) {
1373 brelse(bp);
1374 return (0);
1375 }
1376 cgp->cg_time = time.tv_sec;
1377 ino %= fs->fs_ipg;
1378 if (isclr(cg_inosused(cgp), ino)) {
1379 printf("dev = 0x%x, ino = %d, fs = %s\n",
1380 pip->i_dev, ino, fs->fs_fsmnt);
1381 if (fs->fs_ronly == 0)
1382 panic("ifree: freeing free inode");
1383 }
1384 clrbit(cg_inosused(cgp), ino);
1385 if (ino < cgp->cg_irotor)
1386 cgp->cg_irotor = ino;
1387 cgp->cg_cs.cs_nifree++;
1388 fs->fs_cstotal.cs_nifree++;
1389 fs->fs_cs(fs, cg).cs_nifree++;
1390 if ((ap->a_mode & IFMT) == IFDIR) {
1391 cgp->cg_cs.cs_ndir--;
1392 fs->fs_cstotal.cs_ndir--;
1393 fs->fs_cs(fs, cg).cs_ndir--;
1394 }
1395 fs->fs_fmod = 1;
1396 bdwrite(bp);
1397 return (0);
1398 }
1399
1400 /*
1401 * Find a block of the specified size in the specified cylinder group.
1402 *
1403 * It is a panic if a request is made to find a block if none are
1404 * available.
1405 */
1406 static ufs_daddr_t
ffs_mapsearch(fs,cgp,bpref,allocsiz)1407 ffs_mapsearch(fs, cgp, bpref, allocsiz)
1408 register struct fs *fs;
1409 register struct cg *cgp;
1410 ufs_daddr_t bpref;
1411 int allocsiz;
1412 {
1413 ufs_daddr_t bno;
1414 int start, len, loc, i;
1415 int blk, field, subfield, pos;
1416
1417 /*
1418 * find the fragment by searching through the free block
1419 * map for an appropriate bit pattern
1420 */
1421 if (bpref)
1422 start = dtogd(fs, bpref) / NBBY;
1423 else
1424 start = cgp->cg_frotor / NBBY;
1425 len = howmany(fs->fs_fpg, NBBY) - start;
1426 loc = scanc((u_int)len, (u_char *)&cg_blksfree(cgp)[start],
1427 (u_char *)fragtbl[fs->fs_frag],
1428 (u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
1429 if (loc == 0) {
1430 len = start + 1;
1431 start = 0;
1432 loc = scanc((u_int)len, (u_char *)&cg_blksfree(cgp)[0],
1433 (u_char *)fragtbl[fs->fs_frag],
1434 (u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
1435 if (loc == 0) {
1436 printf("start = %d, len = %d, fs = %s\n",
1437 start, len, fs->fs_fsmnt);
1438 panic("ffs_alloccg: map corrupted");
1439 /* NOTREACHED */
1440 }
1441 }
1442 bno = (start + len - loc) * NBBY;
1443 cgp->cg_frotor = bno;
1444 /*
1445 * found the byte in the map
1446 * sift through the bits to find the selected frag
1447 */
1448 for (i = bno + NBBY; bno < i; bno += fs->fs_frag) {
1449 blk = blkmap(fs, cg_blksfree(cgp), bno);
1450 blk <<= 1;
1451 field = around[allocsiz];
1452 subfield = inside[allocsiz];
1453 for (pos = 0; pos <= fs->fs_frag - allocsiz; pos++) {
1454 if ((blk & field) == subfield)
1455 return (bno + pos);
1456 field <<= 1;
1457 subfield <<= 1;
1458 }
1459 }
1460 printf("bno = %d, fs = %s\n", bno, fs->fs_fsmnt);
1461 panic("ffs_alloccg: block not in map");
1462 return (-1);
1463 }
1464
1465 /*
1466 * Update the cluster map because of an allocation or free.
1467 *
1468 * Cnt == 1 means free; cnt == -1 means allocating.
1469 */
1470 ffs_clusteracct(fs, cgp, blkno, cnt)
1471 struct fs *fs;
1472 struct cg *cgp;
1473 ufs_daddr_t blkno;
1474 int cnt;
1475 {
1476 int32_t *sump;
1477 int32_t *lp;
1478 u_char *freemapp, *mapp;
1479 int i, start, end, forw, back, map, bit;
1480
1481 if (fs->fs_contigsumsize <= 0)
1482 return;
1483 freemapp = cg_clustersfree(cgp);
1484 sump = cg_clustersum(cgp);
1485 /*
1486 * Allocate or clear the actual block.
1487 */
1488 if (cnt > 0)
1489 setbit(freemapp, blkno);
1490 else
1491 clrbit(freemapp, blkno);
1492 /*
1493 * Find the size of the cluster going forward.
1494 */
1495 start = blkno + 1;
1496 end = start + fs->fs_contigsumsize;
1497 if (end >= cgp->cg_nclusterblks)
1498 end = cgp->cg_nclusterblks;
1499 mapp = &freemapp[start / NBBY];
1500 map = *mapp++;
1501 bit = 1 << (start % NBBY);
1502 for (i = start; i < end; i++) {
1503 if ((map & bit) == 0)
1504 break;
1505 if ((i & (NBBY - 1)) != (NBBY - 1)) {
1506 bit <<= 1;
1507 } else {
1508 map = *mapp++;
1509 bit = 1;
1510 }
1511 }
1512 forw = i - start;
1513 /*
1514 * Find the size of the cluster going backward.
1515 */
1516 start = blkno - 1;
1517 end = start - fs->fs_contigsumsize;
1518 if (end < 0)
1519 end = -1;
1520 mapp = &freemapp[start / NBBY];
1521 map = *mapp--;
1522 bit = 1 << (start % NBBY);
1523 for (i = start; i > end; i--) {
1524 if ((map & bit) == 0)
1525 break;
1526 if ((i & (NBBY - 1)) != 0) {
1527 bit >>= 1;
1528 } else {
1529 map = *mapp--;
1530 bit = 1 << (NBBY - 1);
1531 }
1532 }
1533 back = start - i;
1534 /*
1535 * Account for old cluster and the possibly new forward and
1536 * back clusters.
1537 */
1538 i = back + forw + 1;
1539 if (i > fs->fs_contigsumsize)
1540 i = fs->fs_contigsumsize;
1541 sump[i] += cnt;
1542 if (back > 0)
1543 sump[back] -= cnt;
1544 if (forw > 0)
1545 sump[forw] -= cnt;
1546 /*
1547 * Update cluster summary information.
1548 */
1549 lp = &sump[fs->fs_contigsumsize];
1550 for (i = fs->fs_contigsumsize; i > 0; i--)
1551 if (*lp-- > 0)
1552 break;
1553 fs->fs_maxcluster[cgp->cg_cgx] = i;
1554 }
1555
1556 /*
1557 * Fserr prints the name of a file system with an error diagnostic.
1558 *
1559 * The form of the error message is:
1560 * fs: error message
1561 */
1562 static void
ffs_fserr(fs,uid,cp)1563 ffs_fserr(fs, uid, cp)
1564 struct fs *fs;
1565 u_int uid;
1566 char *cp;
1567 {
1568
1569 log(LOG_ERR, "uid %d on %s: %s\n", uid, fs->fs_fsmnt, cp);
1570 }
1571