xref: /freebsd/sys/ufs/ffs/ffs_alloc.c (revision a8445737e740901f5f2c8d24c12ef7fc8b00134e)
1 /*
2  * Copyright (c) 1982, 1986, 1989, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)ffs_alloc.c	8.18 (Berkeley) 5/26/95
34  * $Id: ffs_alloc.c,v 1.52 1998/09/05 14:13:12 phk Exp $
35  */
36 
37 #include "opt_quota.h"
38 
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/buf.h>
42 #include <sys/proc.h>
43 #include <sys/vnode.h>
44 #include <sys/mount.h>
45 #ifdef notyet
46 #include <sys/sysctl.h>
47 #endif
48 #include <sys/syslog.h>
49 
50 #include <ufs/ufs/quota.h>
51 #include <ufs/ufs/inode.h>
52 #include <ufs/ufs/ufsmount.h>
53 
54 #include <ufs/ffs/fs.h>
55 #include <ufs/ffs/ffs_extern.h>
56 
57 typedef ufs_daddr_t allocfcn_t __P((struct inode *ip, int cg, ufs_daddr_t bpref,
58 				  int size));
59 
60 static ufs_daddr_t ffs_alloccg __P((struct inode *, int, ufs_daddr_t, int));
61 static ufs_daddr_t
62 	      ffs_alloccgblk __P((struct inode *, struct buf *, ufs_daddr_t));
63 #ifdef DIAGNOSTIC
64 static int	ffs_checkblk __P((struct inode *, ufs_daddr_t, long));
65 #endif
66 static void	ffs_clusteracct	__P((struct fs *, struct cg *, ufs_daddr_t,
67 				     int));
68 #ifdef notyet
69 static ufs_daddr_t ffs_clusteralloc __P((struct inode *, int, ufs_daddr_t,
70 	    int));
71 #endif
72 static ino_t	ffs_dirpref __P((struct fs *));
73 static ufs_daddr_t ffs_fragextend __P((struct inode *, int, long, int, int));
74 static void	ffs_fserr __P((struct fs *, u_int, char *));
75 static u_long	ffs_hashalloc
76 		    __P((struct inode *, int, long, int, allocfcn_t *));
77 static ino_t	ffs_nodealloccg __P((struct inode *, int, ufs_daddr_t, int));
78 static ufs_daddr_t ffs_mapsearch __P((struct fs *, struct cg *, ufs_daddr_t,
79 	    int));
80 
81 /*
82  * Allocate a block in the file system.
83  *
84  * The size of the requested block is given, which must be some
85  * multiple of fs_fsize and <= fs_bsize.
86  * A preference may be optionally specified. If a preference is given
87  * the following hierarchy is used to allocate a block:
88  *   1) allocate the requested block.
89  *   2) allocate a rotationally optimal block in the same cylinder.
90  *   3) allocate a block in the same cylinder group.
91  *   4) quadradically rehash into other cylinder groups, until an
92  *      available block is located.
93  * If no block preference is given the following heirarchy is used
94  * to allocate a block:
95  *   1) allocate a block in the cylinder group that contains the
96  *      inode for the file.
97  *   2) quadradically rehash into other cylinder groups, until an
98  *      available block is located.
99  */
100 int
101 ffs_alloc(ip, lbn, bpref, size, cred, bnp)
102 	register struct inode *ip;
103 	ufs_daddr_t lbn, bpref;
104 	int size;
105 	struct ucred *cred;
106 	ufs_daddr_t *bnp;
107 {
108 	register struct fs *fs;
109 	ufs_daddr_t bno;
110 	int cg;
111 #ifdef QUOTA
112 	int error;
113 #endif
114 
115 	*bnp = 0;
116 	fs = ip->i_fs;
117 #ifdef DIAGNOSTIC
118 	if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) {
119 		printf("dev = 0x%lx, bsize = %ld, size = %d, fs = %s\n",
120 		    (u_long)ip->i_dev, (long)fs->fs_bsize, size, fs->fs_fsmnt);
121 		panic("ffs_alloc: bad size");
122 	}
123 	if (cred == NOCRED)
124 		panic("ffs_alloc: missing credential");
125 #endif /* DIAGNOSTIC */
126 	if (size == fs->fs_bsize && fs->fs_cstotal.cs_nbfree == 0)
127 		goto nospace;
128 	if (cred->cr_uid != 0 &&
129 	    freespace(fs, fs->fs_minfree) - numfrags(fs, size) < 0)
130 		goto nospace;
131 #ifdef QUOTA
132 	error = chkdq(ip, (long)btodb(size), cred, 0);
133 	if (error)
134 		return (error);
135 #endif
136 	if (bpref >= fs->fs_size)
137 		bpref = 0;
138 	if (bpref == 0)
139 		cg = ino_to_cg(fs, ip->i_number);
140 	else
141 		cg = dtog(fs, bpref);
142 	bno = (ufs_daddr_t)ffs_hashalloc(ip, cg, (long)bpref, size,
143 					 ffs_alloccg);
144 	if (bno > 0) {
145 		ip->i_blocks += btodb(size);
146 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
147 		*bnp = bno;
148 		return (0);
149 	}
150 #ifdef QUOTA
151 	/*
152 	 * Restore user's disk quota because allocation failed.
153 	 */
154 	(void) chkdq(ip, (long)-btodb(size), cred, FORCE);
155 #endif
156 nospace:
157 	ffs_fserr(fs, cred->cr_uid, "file system full");
158 	uprintf("\n%s: write failed, file system is full\n", fs->fs_fsmnt);
159 	return (ENOSPC);
160 }
161 
162 /*
163  * Reallocate a fragment to a bigger size
164  *
165  * The number and size of the old block is given, and a preference
166  * and new size is also specified. The allocator attempts to extend
167  * the original block. Failing that, the regular block allocator is
168  * invoked to get an appropriate block.
169  */
170 int
171 ffs_realloccg(ip, lbprev, bpref, osize, nsize, cred, bpp)
172 	register struct inode *ip;
173 	ufs_daddr_t lbprev;
174 	ufs_daddr_t bpref;
175 	int osize, nsize;
176 	struct ucred *cred;
177 	struct buf **bpp;
178 {
179 	register struct fs *fs;
180 	struct buf *bp;
181 	int cg, request, error;
182 	ufs_daddr_t bprev, bno;
183 
184 	*bpp = 0;
185 	fs = ip->i_fs;
186 #ifdef DIAGNOSTIC
187 	if ((u_int)osize > fs->fs_bsize || fragoff(fs, osize) != 0 ||
188 	    (u_int)nsize > fs->fs_bsize || fragoff(fs, nsize) != 0) {
189 		printf(
190 		"dev = 0x%lx, bsize = %ld, osize = %d, nsize = %d, fs = %s\n",
191 		    (u_long)ip->i_dev, (long)fs->fs_bsize, osize,
192 		    nsize, fs->fs_fsmnt);
193 		panic("ffs_realloccg: bad size");
194 	}
195 	if (cred == NOCRED)
196 		panic("ffs_realloccg: missing credential");
197 #endif /* DIAGNOSTIC */
198 	if (cred->cr_uid != 0 &&
199 	    freespace(fs, fs->fs_minfree) -  numfrags(fs, nsize - osize) < 0)
200 		goto nospace;
201 	if ((bprev = ip->i_db[lbprev]) == 0) {
202 		printf("dev = 0x%lx, bsize = %ld, bprev = %ld, fs = %s\n",
203 		    (u_long)ip->i_dev, (long)fs->fs_bsize, (long)bprev,
204 		    fs->fs_fsmnt);
205 		panic("ffs_realloccg: bad bprev");
206 	}
207 	/*
208 	 * Allocate the extra space in the buffer.
209 	 */
210 	error = bread(ITOV(ip), lbprev, osize, NOCRED, &bp);
211 	if (error) {
212 		brelse(bp);
213 		return (error);
214 	}
215 
216 	if( bp->b_blkno == bp->b_lblkno) {
217 		if( lbprev >= NDADDR)
218 			panic("ffs_realloccg: lbprev out of range");
219 		bp->b_blkno = fsbtodb(fs, bprev);
220 	}
221 
222 #ifdef QUOTA
223 	error = chkdq(ip, (long)btodb(nsize - osize), cred, 0);
224 	if (error) {
225 		brelse(bp);
226 		return (error);
227 	}
228 #endif
229 	/*
230 	 * Check for extension in the existing location.
231 	 */
232 	cg = dtog(fs, bprev);
233 	bno = ffs_fragextend(ip, cg, (long)bprev, osize, nsize);
234 	if (bno) {
235 		if (bp->b_blkno != fsbtodb(fs, bno))
236 			panic("ffs_realloccg: bad blockno");
237 		ip->i_blocks += btodb(nsize - osize);
238 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
239 		allocbuf(bp, nsize);
240 		bp->b_flags |= B_DONE;
241 		bzero((char *)bp->b_data + osize, (u_int)nsize - osize);
242 		*bpp = bp;
243 		return (0);
244 	}
245 	/*
246 	 * Allocate a new disk location.
247 	 */
248 	if (bpref >= fs->fs_size)
249 		bpref = 0;
250 	switch ((int)fs->fs_optim) {
251 	case FS_OPTSPACE:
252 		/*
253 		 * Allocate an exact sized fragment. Although this makes
254 		 * best use of space, we will waste time relocating it if
255 		 * the file continues to grow. If the fragmentation is
256 		 * less than half of the minimum free reserve, we choose
257 		 * to begin optimizing for time.
258 		 */
259 		request = nsize;
260 		if (fs->fs_minfree <= 5 ||
261 		    fs->fs_cstotal.cs_nffree >
262 		    fs->fs_dsize * fs->fs_minfree / (2 * 100))
263 			break;
264 		log(LOG_NOTICE, "%s: optimization changed from SPACE to TIME\n",
265 			fs->fs_fsmnt);
266 		fs->fs_optim = FS_OPTTIME;
267 		break;
268 	case FS_OPTTIME:
269 		/*
270 		 * At this point we have discovered a file that is trying to
271 		 * grow a small fragment to a larger fragment. To save time,
272 		 * we allocate a full sized block, then free the unused portion.
273 		 * If the file continues to grow, the `ffs_fragextend' call
274 		 * above will be able to grow it in place without further
275 		 * copying. If aberrant programs cause disk fragmentation to
276 		 * grow within 2% of the free reserve, we choose to begin
277 		 * optimizing for space.
278 		 */
279 		request = fs->fs_bsize;
280 		if (fs->fs_cstotal.cs_nffree <
281 		    fs->fs_dsize * (fs->fs_minfree - 2) / 100)
282 			break;
283 		log(LOG_NOTICE, "%s: optimization changed from TIME to SPACE\n",
284 			fs->fs_fsmnt);
285 		fs->fs_optim = FS_OPTSPACE;
286 		break;
287 	default:
288 		printf("dev = 0x%lx, optim = %ld, fs = %s\n",
289 		    (u_long)ip->i_dev, (long)fs->fs_optim, fs->fs_fsmnt);
290 		panic("ffs_realloccg: bad optim");
291 		/* NOTREACHED */
292 	}
293 	bno = (ufs_daddr_t)ffs_hashalloc(ip, cg, (long)bpref, request,
294 					 ffs_alloccg);
295 	if (bno > 0) {
296 		bp->b_blkno = fsbtodb(fs, bno);
297 		if (!DOINGSOFTDEP(ITOV(ip)))
298 			ffs_blkfree(ip, bprev, (long)osize);
299 		if (nsize < request)
300 			ffs_blkfree(ip, bno + numfrags(fs, nsize),
301 			    (long)(request - nsize));
302 		ip->i_blocks += btodb(nsize - osize);
303 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
304 		allocbuf(bp, nsize);
305 		bp->b_flags |= B_DONE;
306 		bzero((char *)bp->b_data + osize, (u_int)nsize - osize);
307 		*bpp = bp;
308 		return (0);
309 	}
310 #ifdef QUOTA
311 	/*
312 	 * Restore user's disk quota because allocation failed.
313 	 */
314 	(void) chkdq(ip, (long)-btodb(nsize - osize), cred, FORCE);
315 #endif
316 	brelse(bp);
317 nospace:
318 	/*
319 	 * no space available
320 	 */
321 	ffs_fserr(fs, cred->cr_uid, "file system full");
322 	uprintf("\n%s: write failed, file system is full\n", fs->fs_fsmnt);
323 	return (ENOSPC);
324 }
325 
326 #ifdef notyet
327 SYSCTL_NODE(_vfs, OID_AUTO, ffs, CTLFLAG_RW, 0, "FFS filesystem");
328 
329 /*
330  * Reallocate a sequence of blocks into a contiguous sequence of blocks.
331  *
332  * The vnode and an array of buffer pointers for a range of sequential
333  * logical blocks to be made contiguous is given. The allocator attempts
334  * to find a range of sequential blocks starting as close as possible to
335  * an fs_rotdelay offset from the end of the allocation for the logical
336  * block immediately preceeding the current range. If successful, the
337  * physical block numbers in the buffer pointers and in the inode are
338  * changed to reflect the new allocation. If unsuccessful, the allocation
339  * is left unchanged. The success in doing the reallocation is returned.
340  * Note that the error return is not reflected back to the user. Rather
341  * the previous block allocation will be used.
342  */
343 static int doasyncfree = 1;
344 SYSCTL_INT(_vfs_ffs, FFS_ASYNCFREE, doasyncfree, CTLFLAG_RW, &doasyncfree, 0, "");
345 
346 static int doreallocblks = 1;
347 SYSCTL_INT(_vfs_ffs, FFS_REALLOCBLKS, doreallocblks, CTLFLAG_RW, &doreallocblks, 0, "");
348 
349 static int prtrealloc = 0;
350 #endif
351 
352 int
353 ffs_reallocblks(ap)
354 	struct vop_reallocblks_args /* {
355 		struct vnode *a_vp;
356 		struct cluster_save *a_buflist;
357 	} */ *ap;
358 {
359 #if !defined (not_yes)
360 	return (ENOSPC);
361 #else
362 	struct fs *fs;
363 	struct inode *ip;
364 	struct vnode *vp;
365 	struct buf *sbp, *ebp;
366 	ufs_daddr_t *bap, *sbap, *ebap = 0;
367 	struct cluster_save *buflist;
368 	ufs_daddr_t start_lbn, end_lbn, soff, newblk, blkno;
369 	struct indir start_ap[NIADDR + 1], end_ap[NIADDR + 1], *idp;
370 	int i, len, start_lvl, end_lvl, pref, ssize;
371 	struct timeval tv;
372 
373 	if (doreallocblks == 0)
374 		return (ENOSPC);
375 	vp = ap->a_vp;
376 	ip = VTOI(vp);
377 	fs = ip->i_fs;
378 	if (fs->fs_contigsumsize <= 0)
379 		return (ENOSPC);
380 	buflist = ap->a_buflist;
381 	len = buflist->bs_nchildren;
382 	start_lbn = buflist->bs_children[0]->b_lblkno;
383 	end_lbn = start_lbn + len - 1;
384 #ifdef DIAGNOSTIC
385 	for (i = 0; i < len; i++)
386 		if (!ffs_checkblk(ip,
387 		   dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
388 			panic("ffs_reallocblks: unallocated block 1");
389 	for (i = 1; i < len; i++)
390 		if (buflist->bs_children[i]->b_lblkno != start_lbn + i)
391 			panic("ffs_reallocblks: non-logical cluster");
392 	blkno = buflist->bs_children[0]->b_blkno;
393 	ssize = fsbtodb(fs, fs->fs_frag);
394 	for (i = 1; i < len - 1; i++)
395 		if (buflist->bs_children[i]->b_blkno != blkno + (i * ssize))
396 			panic("ffs_reallocblks: non-physical cluster %d", i);
397 #endif
398 	/*
399 	 * If the latest allocation is in a new cylinder group, assume that
400 	 * the filesystem has decided to move and do not force it back to
401 	 * the previous cylinder group.
402 	 */
403 	if (dtog(fs, dbtofsb(fs, buflist->bs_children[0]->b_blkno)) !=
404 	    dtog(fs, dbtofsb(fs, buflist->bs_children[len - 1]->b_blkno)))
405 		return (ENOSPC);
406 	if (ufs_getlbns(vp, start_lbn, start_ap, &start_lvl) ||
407 	    ufs_getlbns(vp, end_lbn, end_ap, &end_lvl))
408 		return (ENOSPC);
409 	/*
410 	 * Get the starting offset and block map for the first block.
411 	 */
412 	if (start_lvl == 0) {
413 		sbap = &ip->i_db[0];
414 		soff = start_lbn;
415 	} else {
416 		idp = &start_ap[start_lvl - 1];
417 		if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &sbp)) {
418 			brelse(sbp);
419 			return (ENOSPC);
420 		}
421 		sbap = (ufs_daddr_t *)sbp->b_data;
422 		soff = idp->in_off;
423 	}
424 	/*
425 	 * Find the preferred location for the cluster.
426 	 */
427 	pref = ffs_blkpref(ip, start_lbn, soff, sbap);
428 	/*
429 	 * If the block range spans two block maps, get the second map.
430 	 */
431 	if (end_lvl == 0 || (idp = &end_ap[end_lvl - 1])->in_off + 1 >= len) {
432 		ssize = len;
433 	} else {
434 #ifdef DIAGNOSTIC
435 		if (start_ap[start_lvl-1].in_lbn == idp->in_lbn)
436 			panic("ffs_reallocblk: start == end");
437 #endif
438 		ssize = len - (idp->in_off + 1);
439 		if (bread(vp, idp->in_lbn, (int)fs->fs_bsize, NOCRED, &ebp))
440 			goto fail;
441 		ebap = (ufs_daddr_t *)ebp->b_data;
442 	}
443 	/*
444 	 * Search the block map looking for an allocation of the desired size.
445 	 */
446 	if ((newblk = (ufs_daddr_t)ffs_hashalloc(ip, dtog(fs, pref), (long)pref,
447 	    len, ffs_clusteralloc)) == 0)
448 		goto fail;
449 	/*
450 	 * We have found a new contiguous block.
451 	 *
452 	 * First we have to replace the old block pointers with the new
453 	 * block pointers in the inode and indirect blocks associated
454 	 * with the file.
455 	 */
456 #ifdef DEBUG
457 	if (prtrealloc)
458 		printf("realloc: ino %d, lbns %d-%d\n\told:", ip->i_number,
459 		    start_lbn, end_lbn);
460 #endif
461 	blkno = newblk;
462 	for (bap = &sbap[soff], i = 0; i < len; i++, blkno += fs->fs_frag) {
463 		if (i == ssize) {
464 			bap = ebap;
465 			soff = -i;
466 		}
467 #ifdef DIAGNOSTIC
468 		if (!ffs_checkblk(ip,
469 		   dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
470 			panic("ffs_reallocblks: unallocated block 2");
471 		if (dbtofsb(fs, buflist->bs_children[i]->b_blkno) != *bap)
472 			panic("ffs_reallocblks: alloc mismatch");
473 #endif
474 #ifdef DEBUG
475 		if (prtrealloc)
476 			printf(" %d,", *bap);
477 #endif
478 		if (DOINGSOFTDEP(vp)) {
479 			if (sbap == &ip->i_db[0] && i < ssize)
480 				softdep_setup_allocdirect(ip, start_lbn + i,
481 				    blkno, *bap, fs->fs_bsize, fs->fs_bsize,
482 				    buflist->bs_children[i]);
483 			else
484 				softdep_setup_allocindir_page(ip, start_lbn + i,
485 				    i < ssize ? sbp : ebp, soff + i, blkno,
486 				    *bap, buflist->bs_children[i]);
487 		}
488 		*bap++ = blkno;
489 	}
490 	/*
491 	 * Next we must write out the modified inode and indirect blocks.
492 	 * For strict correctness, the writes should be synchronous since
493 	 * the old block values may have been written to disk. In practise
494 	 * they are almost never written, but if we are concerned about
495 	 * strict correctness, the `doasyncfree' flag should be set to zero.
496 	 *
497 	 * The test on `doasyncfree' should be changed to test a flag
498 	 * that shows whether the associated buffers and inodes have
499 	 * been written. The flag should be set when the cluster is
500 	 * started and cleared whenever the buffer or inode is flushed.
501 	 * We can then check below to see if it is set, and do the
502 	 * synchronous write only when it has been cleared.
503 	 */
504 	if (sbap != &ip->i_db[0]) {
505 		if (doasyncfree)
506 			bdwrite(sbp);
507 		else
508 			bwrite(sbp);
509 	} else {
510 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
511 		if (!doasyncfree) {
512 			gettime(&tv);
513 			UFS_UPDATE(vp, &tv, &tv, 1);
514 		}
515 	}
516 	if (ssize < len)
517 		if (doasyncfree)
518 			bdwrite(ebp);
519 		else
520 			bwrite(ebp);
521 	/*
522 	 * Last, free the old blocks and assign the new blocks to the buffers.
523 	 */
524 #ifdef DEBUG
525 	if (prtrealloc)
526 		printf("\n\tnew:");
527 #endif
528 	for (blkno = newblk, i = 0; i < len; i++, blkno += fs->fs_frag) {
529 		if (!DOINGSOFTDEP(vp))
530 			ffs_blkfree(ip,
531 			    dbtofsb(fs, buflist->bs_children[i]->b_blkno),
532 			    fs->fs_bsize);
533 		buflist->bs_children[i]->b_blkno = fsbtodb(fs, blkno);
534 #ifdef DEBUG
535 		if (!ffs_checkblk(ip,
536 		   dbtofsb(fs, buflist->bs_children[i]->b_blkno), fs->fs_bsize))
537 			panic("ffs_reallocblks: unallocated block 3");
538 		if (prtrealloc)
539 			printf(" %d,", blkno);
540 #endif
541 	}
542 #ifdef DEBUG
543 	if (prtrealloc) {
544 		prtrealloc--;
545 		printf("\n");
546 	}
547 #endif
548 	return (0);
549 
550 fail:
551 	if (ssize < len)
552 		brelse(ebp);
553 	if (sbap != &ip->i_db[0])
554 		brelse(sbp);
555 	return (ENOSPC);
556 #endif
557 }
558 
559 /*
560  * Allocate an inode in the file system.
561  *
562  * If allocating a directory, use ffs_dirpref to select the inode.
563  * If allocating in a directory, the following hierarchy is followed:
564  *   1) allocate the preferred inode.
565  *   2) allocate an inode in the same cylinder group.
566  *   3) quadradically rehash into other cylinder groups, until an
567  *      available inode is located.
568  * If no inode preference is given the following heirarchy is used
569  * to allocate an inode:
570  *   1) allocate an inode in cylinder group 0.
571  *   2) quadradically rehash into other cylinder groups, until an
572  *      available inode is located.
573  */
574 int
575 ffs_valloc(pvp, mode, cred, vpp)
576 	struct vnode *pvp;
577 	int mode;
578 	struct ucred *cred;
579 	struct vnode **vpp;
580 {
581 	register struct inode *pip;
582 	register struct fs *fs;
583 	register struct inode *ip;
584 	ino_t ino, ipref;
585 	int cg, error;
586 
587 	*vpp = NULL;
588 	pip = VTOI(pvp);
589 	fs = pip->i_fs;
590 	if (fs->fs_cstotal.cs_nifree == 0)
591 		goto noinodes;
592 
593 	if ((mode & IFMT) == IFDIR)
594 		ipref = ffs_dirpref(fs);
595 	else
596 		ipref = pip->i_number;
597 	if (ipref >= fs->fs_ncg * fs->fs_ipg)
598 		ipref = 0;
599 	cg = ino_to_cg(fs, ipref);
600 	ino = (ino_t)ffs_hashalloc(pip, cg, (long)ipref, mode,
601 					(allocfcn_t *)ffs_nodealloccg);
602 	if (ino == 0)
603 		goto noinodes;
604 	error = VFS_VGET(pvp->v_mount, ino, vpp);
605 	if (error) {
606 		UFS_VFREE(pvp, ino, mode);
607 		return (error);
608 	}
609 	ip = VTOI(*vpp);
610 	if (ip->i_mode) {
611 		printf("mode = 0%o, inum = %lu, fs = %s\n",
612 		    ip->i_mode, (u_long)ip->i_number, fs->fs_fsmnt);
613 		panic("ffs_valloc: dup alloc");
614 	}
615 	if (ip->i_blocks) {				/* XXX */
616 		printf("free inode %s/%lu had %ld blocks\n",
617 		    fs->fs_fsmnt, (u_long)ino, (long)ip->i_blocks);
618 		ip->i_blocks = 0;
619 	}
620 	ip->i_flags = 0;
621 	/*
622 	 * Set up a new generation number for this inode.
623 	 */
624 	if (ip->i_gen == 0 || ++ip->i_gen == 0)
625 		ip->i_gen = random() / 2 + 1;
626 	return (0);
627 noinodes:
628 	ffs_fserr(fs, cred->cr_uid, "out of inodes");
629 	uprintf("\n%s: create/symlink failed, no inodes free\n", fs->fs_fsmnt);
630 	return (ENOSPC);
631 }
632 
633 /*
634  * Find a cylinder to place a directory.
635  *
636  * The policy implemented by this algorithm is to select from
637  * among those cylinder groups with above the average number of
638  * free inodes, the one with the smallest number of directories.
639  */
640 static ino_t
641 ffs_dirpref(fs)
642 	register struct fs *fs;
643 {
644 	int cg, minndir, mincg, avgifree;
645 
646 	avgifree = fs->fs_cstotal.cs_nifree / fs->fs_ncg;
647 	minndir = fs->fs_ipg;
648 	mincg = 0;
649 	for (cg = 0; cg < fs->fs_ncg; cg++)
650 		if (fs->fs_cs(fs, cg).cs_ndir < minndir &&
651 		    fs->fs_cs(fs, cg).cs_nifree >= avgifree) {
652 			mincg = cg;
653 			minndir = fs->fs_cs(fs, cg).cs_ndir;
654 		}
655 	return ((ino_t)(fs->fs_ipg * mincg));
656 }
657 
658 /*
659  * Select the desired position for the next block in a file.  The file is
660  * logically divided into sections. The first section is composed of the
661  * direct blocks. Each additional section contains fs_maxbpg blocks.
662  *
663  * If no blocks have been allocated in the first section, the policy is to
664  * request a block in the same cylinder group as the inode that describes
665  * the file. If no blocks have been allocated in any other section, the
666  * policy is to place the section in a cylinder group with a greater than
667  * average number of free blocks.  An appropriate cylinder group is found
668  * by using a rotor that sweeps the cylinder groups. When a new group of
669  * blocks is needed, the sweep begins in the cylinder group following the
670  * cylinder group from which the previous allocation was made. The sweep
671  * continues until a cylinder group with greater than the average number
672  * of free blocks is found. If the allocation is for the first block in an
673  * indirect block, the information on the previous allocation is unavailable;
674  * here a best guess is made based upon the logical block number being
675  * allocated.
676  *
677  * If a section is already partially allocated, the policy is to
678  * contiguously allocate fs_maxcontig blocks.  The end of one of these
679  * contiguous blocks and the beginning of the next is physically separated
680  * so that the disk head will be in transit between them for at least
681  * fs_rotdelay milliseconds.  This is to allow time for the processor to
682  * schedule another I/O transfer.
683  */
684 ufs_daddr_t
685 ffs_blkpref(ip, lbn, indx, bap)
686 	struct inode *ip;
687 	ufs_daddr_t lbn;
688 	int indx;
689 	ufs_daddr_t *bap;
690 {
691 	register struct fs *fs;
692 	register int cg;
693 	int avgbfree, startcg;
694 	ufs_daddr_t nextblk;
695 
696 	fs = ip->i_fs;
697 	if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
698 		if (lbn < NDADDR) {
699 			cg = ino_to_cg(fs, ip->i_number);
700 			return (fs->fs_fpg * cg + fs->fs_frag);
701 		}
702 		/*
703 		 * Find a cylinder with greater than average number of
704 		 * unused data blocks.
705 		 */
706 		if (indx == 0 || bap[indx - 1] == 0)
707 			startcg =
708 			    ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
709 		else
710 			startcg = dtog(fs, bap[indx - 1]) + 1;
711 		startcg %= fs->fs_ncg;
712 		avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
713 		for (cg = startcg; cg < fs->fs_ncg; cg++)
714 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
715 				fs->fs_cgrotor = cg;
716 				return (fs->fs_fpg * cg + fs->fs_frag);
717 			}
718 		for (cg = 0; cg <= startcg; cg++)
719 			if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
720 				fs->fs_cgrotor = cg;
721 				return (fs->fs_fpg * cg + fs->fs_frag);
722 			}
723 		return (0);
724 	}
725 	/*
726 	 * One or more previous blocks have been laid out. If less
727 	 * than fs_maxcontig previous blocks are contiguous, the
728 	 * next block is requested contiguously, otherwise it is
729 	 * requested rotationally delayed by fs_rotdelay milliseconds.
730 	 */
731 	nextblk = bap[indx - 1] + fs->fs_frag;
732 	if (fs->fs_rotdelay == 0 || indx < fs->fs_maxcontig ||
733 	    bap[indx - fs->fs_maxcontig] +
734 	    blkstofrags(fs, fs->fs_maxcontig) != nextblk)
735 		return (nextblk);
736 	/*
737 	 * Here we convert ms of delay to frags as:
738 	 * (frags) = (ms) * (rev/sec) * (sect/rev) /
739 	 *	((sect/frag) * (ms/sec))
740 	 * then round up to the next block.
741 	 */
742 	nextblk += roundup(fs->fs_rotdelay * fs->fs_rps * fs->fs_nsect /
743 	    (NSPF(fs) * 1000), fs->fs_frag);
744 	return (nextblk);
745 }
746 
747 /*
748  * Implement the cylinder overflow algorithm.
749  *
750  * The policy implemented by this algorithm is:
751  *   1) allocate the block in its requested cylinder group.
752  *   2) quadradically rehash on the cylinder group number.
753  *   3) brute force search for a free block.
754  */
755 /*VARARGS5*/
756 static u_long
757 ffs_hashalloc(ip, cg, pref, size, allocator)
758 	struct inode *ip;
759 	int cg;
760 	long pref;
761 	int size;	/* size for data blocks, mode for inodes */
762 	allocfcn_t *allocator;
763 {
764 	register struct fs *fs;
765 	long result;	/* XXX why not same type as we return? */
766 	int i, icg = cg;
767 
768 	fs = ip->i_fs;
769 	/*
770 	 * 1: preferred cylinder group
771 	 */
772 	result = (*allocator)(ip, cg, pref, size);
773 	if (result)
774 		return (result);
775 	/*
776 	 * 2: quadratic rehash
777 	 */
778 	for (i = 1; i < fs->fs_ncg; i *= 2) {
779 		cg += i;
780 		if (cg >= fs->fs_ncg)
781 			cg -= fs->fs_ncg;
782 		result = (*allocator)(ip, cg, 0, size);
783 		if (result)
784 			return (result);
785 	}
786 	/*
787 	 * 3: brute force search
788 	 * Note that we start at i == 2, since 0 was checked initially,
789 	 * and 1 is always checked in the quadratic rehash.
790 	 */
791 	cg = (icg + 2) % fs->fs_ncg;
792 	for (i = 2; i < fs->fs_ncg; i++) {
793 		result = (*allocator)(ip, cg, 0, size);
794 		if (result)
795 			return (result);
796 		cg++;
797 		if (cg == fs->fs_ncg)
798 			cg = 0;
799 	}
800 	return (0);
801 }
802 
803 /*
804  * Determine whether a fragment can be extended.
805  *
806  * Check to see if the necessary fragments are available, and
807  * if they are, allocate them.
808  */
809 static ufs_daddr_t
810 ffs_fragextend(ip, cg, bprev, osize, nsize)
811 	struct inode *ip;
812 	int cg;
813 	long bprev;
814 	int osize, nsize;
815 {
816 	register struct fs *fs;
817 	register struct cg *cgp;
818 	struct buf *bp;
819 	long bno;
820 	int frags, bbase;
821 	int i, error;
822 
823 	fs = ip->i_fs;
824 	if (fs->fs_cs(fs, cg).cs_nffree < numfrags(fs, nsize - osize))
825 		return (0);
826 	frags = numfrags(fs, nsize);
827 	bbase = fragnum(fs, bprev);
828 	if (bbase > fragnum(fs, (bprev + frags - 1))) {
829 		/* cannot extend across a block boundary */
830 		return (0);
831 	}
832 	error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
833 		(int)fs->fs_cgsize, NOCRED, &bp);
834 	if (error) {
835 		brelse(bp);
836 		return (0);
837 	}
838 	cgp = (struct cg *)bp->b_data;
839 	if (!cg_chkmagic(cgp)) {
840 		brelse(bp);
841 		return (0);
842 	}
843 	cgp->cg_time = time_second;
844 	bno = dtogd(fs, bprev);
845 	for (i = numfrags(fs, osize); i < frags; i++)
846 		if (isclr(cg_blksfree(cgp), bno + i)) {
847 			brelse(bp);
848 			return (0);
849 		}
850 	/*
851 	 * the current fragment can be extended
852 	 * deduct the count on fragment being extended into
853 	 * increase the count on the remaining fragment (if any)
854 	 * allocate the extended piece
855 	 */
856 	for (i = frags; i < fs->fs_frag - bbase; i++)
857 		if (isclr(cg_blksfree(cgp), bno + i))
858 			break;
859 	cgp->cg_frsum[i - numfrags(fs, osize)]--;
860 	if (i != frags)
861 		cgp->cg_frsum[i - frags]++;
862 	for (i = numfrags(fs, osize); i < frags; i++) {
863 		clrbit(cg_blksfree(cgp), bno + i);
864 		cgp->cg_cs.cs_nffree--;
865 		fs->fs_cstotal.cs_nffree--;
866 		fs->fs_cs(fs, cg).cs_nffree--;
867 	}
868 	fs->fs_fmod = 1;
869 	if (DOINGSOFTDEP(ITOV(ip)))
870 		softdep_setup_blkmapdep(bp, fs, bprev);
871 	bdwrite(bp);
872 	return (bprev);
873 }
874 
875 /*
876  * Determine whether a block can be allocated.
877  *
878  * Check to see if a block of the appropriate size is available,
879  * and if it is, allocate it.
880  */
881 static ufs_daddr_t
882 ffs_alloccg(ip, cg, bpref, size)
883 	struct inode *ip;
884 	int cg;
885 	ufs_daddr_t bpref;
886 	int size;
887 {
888 	register struct fs *fs;
889 	register struct cg *cgp;
890 	struct buf *bp;
891 	register int i;
892 	ufs_daddr_t bno, blkno;
893 	int allocsiz, error, frags;
894 
895 	fs = ip->i_fs;
896 	if (fs->fs_cs(fs, cg).cs_nbfree == 0 && size == fs->fs_bsize)
897 		return (0);
898 	error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
899 		(int)fs->fs_cgsize, NOCRED, &bp);
900 	if (error) {
901 		brelse(bp);
902 		return (0);
903 	}
904 	cgp = (struct cg *)bp->b_data;
905 	if (!cg_chkmagic(cgp) ||
906 	    (cgp->cg_cs.cs_nbfree == 0 && size == fs->fs_bsize)) {
907 		brelse(bp);
908 		return (0);
909 	}
910 	cgp->cg_time = time_second;
911 	if (size == fs->fs_bsize) {
912 		bno = ffs_alloccgblk(ip, bp, bpref);
913 		bdwrite(bp);
914 		return (bno);
915 	}
916 	/*
917 	 * check to see if any fragments are already available
918 	 * allocsiz is the size which will be allocated, hacking
919 	 * it down to a smaller size if necessary
920 	 */
921 	frags = numfrags(fs, size);
922 	for (allocsiz = frags; allocsiz < fs->fs_frag; allocsiz++)
923 		if (cgp->cg_frsum[allocsiz] != 0)
924 			break;
925 	if (allocsiz == fs->fs_frag) {
926 		/*
927 		 * no fragments were available, so a block will be
928 		 * allocated, and hacked up
929 		 */
930 		if (cgp->cg_cs.cs_nbfree == 0) {
931 			brelse(bp);
932 			return (0);
933 		}
934 		bno = ffs_alloccgblk(ip, bp, bpref);
935 		bpref = dtogd(fs, bno);
936 		for (i = frags; i < fs->fs_frag; i++)
937 			setbit(cg_blksfree(cgp), bpref + i);
938 		i = fs->fs_frag - frags;
939 		cgp->cg_cs.cs_nffree += i;
940 		fs->fs_cstotal.cs_nffree += i;
941 		fs->fs_cs(fs, cg).cs_nffree += i;
942 		fs->fs_fmod = 1;
943 		cgp->cg_frsum[i]++;
944 		bdwrite(bp);
945 		return (bno);
946 	}
947 	bno = ffs_mapsearch(fs, cgp, bpref, allocsiz);
948 	if (bno < 0) {
949 		brelse(bp);
950 		return (0);
951 	}
952 	for (i = 0; i < frags; i++)
953 		clrbit(cg_blksfree(cgp), bno + i);
954 	cgp->cg_cs.cs_nffree -= frags;
955 	fs->fs_cstotal.cs_nffree -= frags;
956 	fs->fs_cs(fs, cg).cs_nffree -= frags;
957 	fs->fs_fmod = 1;
958 	cgp->cg_frsum[allocsiz]--;
959 	if (frags != allocsiz)
960 		cgp->cg_frsum[allocsiz - frags]++;
961 	blkno = cg * fs->fs_fpg + bno;
962 	if (DOINGSOFTDEP(ITOV(ip)))
963 		softdep_setup_blkmapdep(bp, fs, blkno);
964 	bdwrite(bp);
965 	return ((u_long)blkno);
966 }
967 
968 /*
969  * Allocate a block in a cylinder group.
970  *
971  * This algorithm implements the following policy:
972  *   1) allocate the requested block.
973  *   2) allocate a rotationally optimal block in the same cylinder.
974  *   3) allocate the next available block on the block rotor for the
975  *      specified cylinder group.
976  * Note that this routine only allocates fs_bsize blocks; these
977  * blocks may be fragmented by the routine that allocates them.
978  */
979 static ufs_daddr_t
980 ffs_alloccgblk(ip, bp, bpref)
981 	struct inode *ip;
982 	struct buf *bp;
983 	ufs_daddr_t bpref;
984 {
985 	struct fs *fs;
986 	struct cg *cgp;
987 	ufs_daddr_t bno, blkno;
988 	int cylno, pos, delta;
989 	short *cylbp;
990 	register int i;
991 
992 	fs = ip->i_fs;
993 	cgp = (struct cg *)bp->b_data;
994 	if (bpref == 0 || dtog(fs, bpref) != cgp->cg_cgx) {
995 		bpref = cgp->cg_rotor;
996 		goto norot;
997 	}
998 	bpref = blknum(fs, bpref);
999 	bpref = dtogd(fs, bpref);
1000 	/*
1001 	 * if the requested block is available, use it
1002 	 */
1003 	if (ffs_isblock(fs, cg_blksfree(cgp), fragstoblks(fs, bpref))) {
1004 		bno = bpref;
1005 		goto gotit;
1006 	}
1007 	if (fs->fs_nrpos <= 1 || fs->fs_cpc == 0) {
1008 		/*
1009 		 * Block layout information is not available.
1010 		 * Leaving bpref unchanged means we take the
1011 		 * next available free block following the one
1012 		 * we just allocated. Hopefully this will at
1013 		 * least hit a track cache on drives of unknown
1014 		 * geometry (e.g. SCSI).
1015 		 */
1016 		goto norot;
1017 	}
1018 	/*
1019 	 * check for a block available on the same cylinder
1020 	 */
1021 	cylno = cbtocylno(fs, bpref);
1022 	if (cg_blktot(cgp)[cylno] == 0)
1023 		goto norot;
1024 	/*
1025 	 * check the summary information to see if a block is
1026 	 * available in the requested cylinder starting at the
1027 	 * requested rotational position and proceeding around.
1028 	 */
1029 	cylbp = cg_blks(fs, cgp, cylno);
1030 	pos = cbtorpos(fs, bpref);
1031 	for (i = pos; i < fs->fs_nrpos; i++)
1032 		if (cylbp[i] > 0)
1033 			break;
1034 	if (i == fs->fs_nrpos)
1035 		for (i = 0; i < pos; i++)
1036 			if (cylbp[i] > 0)
1037 				break;
1038 	if (cylbp[i] > 0) {
1039 		/*
1040 		 * found a rotational position, now find the actual
1041 		 * block. A panic if none is actually there.
1042 		 */
1043 		pos = cylno % fs->fs_cpc;
1044 		bno = (cylno - pos) * fs->fs_spc / NSPB(fs);
1045 		if (fs_postbl(fs, pos)[i] == -1) {
1046 			printf("pos = %d, i = %d, fs = %s\n",
1047 			    pos, i, fs->fs_fsmnt);
1048 			panic("ffs_alloccgblk: cyl groups corrupted");
1049 		}
1050 		for (i = fs_postbl(fs, pos)[i];; ) {
1051 			if (ffs_isblock(fs, cg_blksfree(cgp), bno + i)) {
1052 				bno = blkstofrags(fs, (bno + i));
1053 				goto gotit;
1054 			}
1055 			delta = fs_rotbl(fs)[i];
1056 			if (delta <= 0 ||
1057 			    delta + i > fragstoblks(fs, fs->fs_fpg))
1058 				break;
1059 			i += delta;
1060 		}
1061 		printf("pos = %d, i = %d, fs = %s\n", pos, i, fs->fs_fsmnt);
1062 		panic("ffs_alloccgblk: can't find blk in cyl");
1063 	}
1064 norot:
1065 	/*
1066 	 * no blocks in the requested cylinder, so take next
1067 	 * available one in this cylinder group.
1068 	 */
1069 	bno = ffs_mapsearch(fs, cgp, bpref, (int)fs->fs_frag);
1070 	if (bno < 0)
1071 		return (0);
1072 	cgp->cg_rotor = bno;
1073 gotit:
1074 	blkno = fragstoblks(fs, bno);
1075 	ffs_clrblock(fs, cg_blksfree(cgp), (long)blkno);
1076 	ffs_clusteracct(fs, cgp, blkno, -1);
1077 	cgp->cg_cs.cs_nbfree--;
1078 	fs->fs_cstotal.cs_nbfree--;
1079 	fs->fs_cs(fs, cgp->cg_cgx).cs_nbfree--;
1080 	cylno = cbtocylno(fs, bno);
1081 	cg_blks(fs, cgp, cylno)[cbtorpos(fs, bno)]--;
1082 	cg_blktot(cgp)[cylno]--;
1083 	fs->fs_fmod = 1;
1084 	blkno = cgp->cg_cgx * fs->fs_fpg + bno;
1085 	if (DOINGSOFTDEP(ITOV(ip)))
1086 		softdep_setup_blkmapdep(bp, fs, blkno);
1087 	return (blkno);
1088 }
1089 
1090 #ifdef notyet
1091 /*
1092  * Determine whether a cluster can be allocated.
1093  *
1094  * We do not currently check for optimal rotational layout if there
1095  * are multiple choices in the same cylinder group. Instead we just
1096  * take the first one that we find following bpref.
1097  */
1098 static ufs_daddr_t
1099 ffs_clusteralloc(ip, cg, bpref, len)
1100 	struct inode *ip;
1101 	int cg;
1102 	ufs_daddr_t bpref;
1103 	int len;
1104 {
1105 	register struct fs *fs;
1106 	register struct cg *cgp;
1107 	struct buf *bp;
1108 	int i, got, run, bno, bit, map;
1109 	u_char *mapp;
1110 	int32_t *lp;
1111 
1112 	fs = ip->i_fs;
1113 	if (fs->fs_maxcluster[cg] < len)
1114 		return (NULL);
1115 	if (bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize,
1116 	    NOCRED, &bp))
1117 		goto fail;
1118 	cgp = (struct cg *)bp->b_data;
1119 	if (!cg_chkmagic(cgp))
1120 		goto fail;
1121 	/*
1122 	 * Check to see if a cluster of the needed size (or bigger) is
1123 	 * available in this cylinder group.
1124 	 */
1125 	lp = &cg_clustersum(cgp)[len];
1126 	for (i = len; i <= fs->fs_contigsumsize; i++)
1127 		if (*lp++ > 0)
1128 			break;
1129 	if (i > fs->fs_contigsumsize) {
1130 		/*
1131 		 * This is the first time looking for a cluster in this
1132 		 * cylinder group. Update the cluster summary information
1133 		 * to reflect the true maximum sized cluster so that
1134 		 * future cluster allocation requests can avoid reading
1135 		 * the cylinder group map only to find no clusters.
1136 		 */
1137 		lp = &cg_clustersum(cgp)[len - 1];
1138 		for (i = len - 1; i > 0; i--)
1139 			if (*lp-- > 0)
1140 				break;
1141 		fs->fs_maxcluster[cg] = i;
1142 		goto fail;
1143 	}
1144 	/*
1145 	 * Search the cluster map to find a big enough cluster.
1146 	 * We take the first one that we find, even if it is larger
1147 	 * than we need as we prefer to get one close to the previous
1148 	 * block allocation. We do not search before the current
1149 	 * preference point as we do not want to allocate a block
1150 	 * that is allocated before the previous one (as we will
1151 	 * then have to wait for another pass of the elevator
1152 	 * algorithm before it will be read). We prefer to fail and
1153 	 * be recalled to try an allocation in the next cylinder group.
1154 	 */
1155 	if (dtog(fs, bpref) != cg)
1156 		bpref = 0;
1157 	else
1158 		bpref = fragstoblks(fs, dtogd(fs, blknum(fs, bpref)));
1159 	mapp = &cg_clustersfree(cgp)[bpref / NBBY];
1160 	map = *mapp++;
1161 	bit = 1 << (bpref % NBBY);
1162 	for (run = 0, got = bpref; got < cgp->cg_nclusterblks; got++) {
1163 		if ((map & bit) == 0) {
1164 			run = 0;
1165 		} else {
1166 			run++;
1167 			if (run == len)
1168 				break;
1169 		}
1170 		if ((got & (NBBY - 1)) != (NBBY - 1)) {
1171 			bit <<= 1;
1172 		} else {
1173 			map = *mapp++;
1174 			bit = 1;
1175 		}
1176 	}
1177 	if (got >= cgp->cg_nclusterblks)
1178 		goto fail;
1179 	/*
1180 	 * Allocate the cluster that we have found.
1181 	 */
1182 	for (i = 1; i <= len; i++)
1183 		if (!ffs_isblock(fs, cg_blksfree(cgp), got - run + i))
1184 			panic("ffs_clusteralloc: map mismatch");
1185 	bno = cg * fs->fs_fpg + blkstofrags(fs, got - run + 1);
1186 	if (dtog(fs, bno) != cg)
1187 		panic("ffs_clusteralloc: allocated out of group");
1188 	len = blkstofrags(fs, len);
1189 	for (i = 0; i < len; i += fs->fs_frag)
1190 		if ((got = ffs_alloccgblk(ip, bp, bno + i)) != bno + i)
1191 			panic("ffs_clusteralloc: lost block");
1192 	bdwrite(bp);
1193 	return (bno);
1194 
1195 fail:
1196 	brelse(bp);
1197 	return (0);
1198 }
1199 #endif
1200 
1201 /*
1202  * Determine whether an inode can be allocated.
1203  *
1204  * Check to see if an inode is available, and if it is,
1205  * allocate it using the following policy:
1206  *   1) allocate the requested inode.
1207  *   2) allocate the next available inode after the requested
1208  *      inode in the specified cylinder group.
1209  */
1210 static ino_t
1211 ffs_nodealloccg(ip, cg, ipref, mode)
1212 	struct inode *ip;
1213 	int cg;
1214 	ufs_daddr_t ipref;
1215 	int mode;
1216 {
1217 	register struct fs *fs;
1218 	register struct cg *cgp;
1219 	struct buf *bp;
1220 	int error, start, len, loc, map, i;
1221 
1222 	fs = ip->i_fs;
1223 	if (fs->fs_cs(fs, cg).cs_nifree == 0)
1224 		return (0);
1225 	error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1226 		(int)fs->fs_cgsize, NOCRED, &bp);
1227 	if (error) {
1228 		brelse(bp);
1229 		return (0);
1230 	}
1231 	cgp = (struct cg *)bp->b_data;
1232 	if (!cg_chkmagic(cgp) || cgp->cg_cs.cs_nifree == 0) {
1233 		brelse(bp);
1234 		return (0);
1235 	}
1236 	cgp->cg_time = time_second;
1237 	if (ipref) {
1238 		ipref %= fs->fs_ipg;
1239 		if (isclr(cg_inosused(cgp), ipref))
1240 			goto gotit;
1241 	}
1242 	start = cgp->cg_irotor / NBBY;
1243 	len = howmany(fs->fs_ipg - cgp->cg_irotor, NBBY);
1244 	loc = skpc(0xff, len, &cg_inosused(cgp)[start]);
1245 	if (loc == 0) {
1246 		len = start + 1;
1247 		start = 0;
1248 		loc = skpc(0xff, len, &cg_inosused(cgp)[0]);
1249 		if (loc == 0) {
1250 			printf("cg = %d, irotor = %ld, fs = %s\n",
1251 			    cg, (long)cgp->cg_irotor, fs->fs_fsmnt);
1252 			panic("ffs_nodealloccg: map corrupted");
1253 			/* NOTREACHED */
1254 		}
1255 	}
1256 	i = start + len - loc;
1257 	map = cg_inosused(cgp)[i];
1258 	ipref = i * NBBY;
1259 	for (i = 1; i < (1 << NBBY); i <<= 1, ipref++) {
1260 		if ((map & i) == 0) {
1261 			cgp->cg_irotor = ipref;
1262 			goto gotit;
1263 		}
1264 	}
1265 	printf("fs = %s\n", fs->fs_fsmnt);
1266 	panic("ffs_nodealloccg: block not in map");
1267 	/* NOTREACHED */
1268 gotit:
1269 	if (DOINGSOFTDEP(ITOV(ip)))
1270 		softdep_setup_inomapdep(bp, ip, cg * fs->fs_ipg + ipref);
1271 	setbit(cg_inosused(cgp), ipref);
1272 	cgp->cg_cs.cs_nifree--;
1273 	fs->fs_cstotal.cs_nifree--;
1274 	fs->fs_cs(fs, cg).cs_nifree--;
1275 	fs->fs_fmod = 1;
1276 	if ((mode & IFMT) == IFDIR) {
1277 		cgp->cg_cs.cs_ndir++;
1278 		fs->fs_cstotal.cs_ndir++;
1279 		fs->fs_cs(fs, cg).cs_ndir++;
1280 	}
1281 	bdwrite(bp);
1282 	return (cg * fs->fs_ipg + ipref);
1283 }
1284 
1285 /*
1286  * Free a block or fragment.
1287  *
1288  * The specified block or fragment is placed back in the
1289  * free map. If a fragment is deallocated, a possible
1290  * block reassembly is checked.
1291  */
1292 void
1293 ffs_blkfree(ip, bno, size)
1294 	register struct inode *ip;
1295 	ufs_daddr_t bno;
1296 	long size;
1297 {
1298 	register struct fs *fs;
1299 	register struct cg *cgp;
1300 	struct buf *bp;
1301 	ufs_daddr_t blkno;
1302 	int i, error, cg, blk, frags, bbase;
1303 
1304 	fs = ip->i_fs;
1305 	VOP_FREEBLKS(ip->i_devvp, fsbtodb(fs, bno), size);
1306 	if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0 ||
1307 	    fragnum(fs, bno) + numfrags(fs, size) > fs->fs_frag) {
1308 		printf("dev=0x%lx, bno = %d, bsize = %d, size = %ld, fs = %s\n",
1309 		    (u_long)ip->i_dev, bno, fs->fs_bsize, size, fs->fs_fsmnt);
1310 		panic("ffs_blkfree: bad size");
1311 	}
1312 	cg = dtog(fs, bno);
1313 	if ((u_int)bno >= fs->fs_size) {
1314 		printf("bad block %ld, ino %lu\n",
1315 		    (long)bno, (u_long)ip->i_number);
1316 		ffs_fserr(fs, ip->i_uid, "bad block");
1317 		return;
1318 	}
1319 	error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1320 		(int)fs->fs_cgsize, NOCRED, &bp);
1321 	if (error) {
1322 		brelse(bp);
1323 		return;
1324 	}
1325 	cgp = (struct cg *)bp->b_data;
1326 	if (!cg_chkmagic(cgp)) {
1327 		brelse(bp);
1328 		return;
1329 	}
1330 	cgp->cg_time = time_second;
1331 	bno = dtogd(fs, bno);
1332 	if (size == fs->fs_bsize) {
1333 		blkno = fragstoblks(fs, bno);
1334 		if (!ffs_isfreeblock(fs, cg_blksfree(cgp), blkno)) {
1335 			printf("dev = 0x%lx, block = %ld, fs = %s\n",
1336 			    (u_long)ip->i_dev, (long)bno, fs->fs_fsmnt);
1337 			panic("ffs_blkfree: freeing free block");
1338 		}
1339 		ffs_setblock(fs, cg_blksfree(cgp), blkno);
1340 		ffs_clusteracct(fs, cgp, blkno, 1);
1341 		cgp->cg_cs.cs_nbfree++;
1342 		fs->fs_cstotal.cs_nbfree++;
1343 		fs->fs_cs(fs, cg).cs_nbfree++;
1344 		i = cbtocylno(fs, bno);
1345 		cg_blks(fs, cgp, i)[cbtorpos(fs, bno)]++;
1346 		cg_blktot(cgp)[i]++;
1347 	} else {
1348 		bbase = bno - fragnum(fs, bno);
1349 		/*
1350 		 * decrement the counts associated with the old frags
1351 		 */
1352 		blk = blkmap(fs, cg_blksfree(cgp), bbase);
1353 		ffs_fragacct(fs, blk, cgp->cg_frsum, -1);
1354 		/*
1355 		 * deallocate the fragment
1356 		 */
1357 		frags = numfrags(fs, size);
1358 		for (i = 0; i < frags; i++) {
1359 			if (isset(cg_blksfree(cgp), bno + i)) {
1360 				printf("dev = 0x%lx, block = %ld, fs = %s\n",
1361 				    (u_long)ip->i_dev, (long)(bno + i),
1362 				    fs->fs_fsmnt);
1363 				panic("ffs_blkfree: freeing free frag");
1364 			}
1365 			setbit(cg_blksfree(cgp), bno + i);
1366 		}
1367 		cgp->cg_cs.cs_nffree += i;
1368 		fs->fs_cstotal.cs_nffree += i;
1369 		fs->fs_cs(fs, cg).cs_nffree += i;
1370 		/*
1371 		 * add back in counts associated with the new frags
1372 		 */
1373 		blk = blkmap(fs, cg_blksfree(cgp), bbase);
1374 		ffs_fragacct(fs, blk, cgp->cg_frsum, 1);
1375 		/*
1376 		 * if a complete block has been reassembled, account for it
1377 		 */
1378 		blkno = fragstoblks(fs, bbase);
1379 		if (ffs_isblock(fs, cg_blksfree(cgp), blkno)) {
1380 			cgp->cg_cs.cs_nffree -= fs->fs_frag;
1381 			fs->fs_cstotal.cs_nffree -= fs->fs_frag;
1382 			fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag;
1383 			ffs_clusteracct(fs, cgp, blkno, 1);
1384 			cgp->cg_cs.cs_nbfree++;
1385 			fs->fs_cstotal.cs_nbfree++;
1386 			fs->fs_cs(fs, cg).cs_nbfree++;
1387 			i = cbtocylno(fs, bbase);
1388 			cg_blks(fs, cgp, i)[cbtorpos(fs, bbase)]++;
1389 			cg_blktot(cgp)[i]++;
1390 		}
1391 	}
1392 	fs->fs_fmod = 1;
1393 	bdwrite(bp);
1394 }
1395 
1396 #ifdef DIAGNOSTIC
1397 /*
1398  * Verify allocation of a block or fragment. Returns true if block or
1399  * fragment is allocated, false if it is free.
1400  */
1401 static int
1402 ffs_checkblk(ip, bno, size)
1403 	struct inode *ip;
1404 	ufs_daddr_t bno;
1405 	long size;
1406 {
1407 	struct fs *fs;
1408 	struct cg *cgp;
1409 	struct buf *bp;
1410 	int i, error, frags, free;
1411 
1412 	fs = ip->i_fs;
1413 	if ((u_int)size > fs->fs_bsize || fragoff(fs, size) != 0) {
1414 		printf("bsize = %ld, size = %ld, fs = %s\n",
1415 		    (long)fs->fs_bsize, size, fs->fs_fsmnt);
1416 		panic("ffs_checkblk: bad size");
1417 	}
1418 	if ((u_int)bno >= fs->fs_size)
1419 		panic("ffs_checkblk: bad block %d", bno);
1420 	error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, dtog(fs, bno))),
1421 		(int)fs->fs_cgsize, NOCRED, &bp);
1422 	if (error)
1423 		panic("ffs_checkblk: cg bread failed");
1424 	cgp = (struct cg *)bp->b_data;
1425 	if (!cg_chkmagic(cgp))
1426 		panic("ffs_checkblk: cg magic mismatch");
1427 	bno = dtogd(fs, bno);
1428 	if (size == fs->fs_bsize) {
1429 		free = ffs_isblock(fs, cg_blksfree(cgp), fragstoblks(fs, bno));
1430 	} else {
1431 		frags = numfrags(fs, size);
1432 		for (free = 0, i = 0; i < frags; i++)
1433 			if (isset(cg_blksfree(cgp), bno + i))
1434 				free++;
1435 		if (free != 0 && free != frags)
1436 			panic("ffs_checkblk: partially free fragment");
1437 	}
1438 	brelse(bp);
1439 	return (!free);
1440 }
1441 #endif /* DIAGNOSTIC */
1442 
1443 /*
1444  * Free an inode.
1445  */
1446 int
1447 ffs_vfree( pvp, ino, mode)
1448 	struct vnode *pvp;
1449 	ino_t ino;
1450 	int mode;
1451 {
1452 	if (DOINGSOFTDEP(pvp)) {
1453 		softdep_freefile(pvp, ino, mode);
1454 		return (0);
1455 	}
1456 	return (ffs_freefile(pvp, ino, mode));
1457 }
1458 
1459 /*
1460  * Do the actual free operation.
1461  * The specified inode is placed back in the free map.
1462  */
1463  int
1464  ffs_freefile( pvp, ino, mode)
1465 	struct vnode *pvp;
1466 	ino_t ino;
1467 	int mode;
1468 {
1469 	register struct fs *fs;
1470 	register struct cg *cgp;
1471 	register struct inode *pip;
1472 	struct buf *bp;
1473 	int error, cg;
1474 
1475 	pip = VTOI(pvp);
1476 	fs = pip->i_fs;
1477 	if ((u_int)ino >= fs->fs_ipg * fs->fs_ncg)
1478 		panic("ffs_vfree: range: dev = 0x%x, ino = %d, fs = %s",
1479 		    pip->i_dev, ino, fs->fs_fsmnt);
1480 	cg = ino_to_cg(fs, ino);
1481 	error = bread(pip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
1482 		(int)fs->fs_cgsize, NOCRED, &bp);
1483 	if (error) {
1484 		brelse(bp);
1485 		return (error);
1486 	}
1487 	cgp = (struct cg *)bp->b_data;
1488 	if (!cg_chkmagic(cgp)) {
1489 		brelse(bp);
1490 		return (0);
1491 	}
1492 	cgp->cg_time = time_second;
1493 	ino %= fs->fs_ipg;
1494 	if (isclr(cg_inosused(cgp), ino)) {
1495 		printf("dev = 0x%lx, ino = %lu, fs = %s\n",
1496 		    (u_long)pip->i_dev, (u_long)ino, fs->fs_fsmnt);
1497 		if (fs->fs_ronly == 0)
1498 			panic("ffs_vfree: freeing free inode");
1499 	}
1500 	clrbit(cg_inosused(cgp), ino);
1501 	if (ino < cgp->cg_irotor)
1502 		cgp->cg_irotor = ino;
1503 	cgp->cg_cs.cs_nifree++;
1504 	fs->fs_cstotal.cs_nifree++;
1505 	fs->fs_cs(fs, cg).cs_nifree++;
1506 	if ((mode & IFMT) == IFDIR) {
1507 		cgp->cg_cs.cs_ndir--;
1508 		fs->fs_cstotal.cs_ndir--;
1509 		fs->fs_cs(fs, cg).cs_ndir--;
1510 	}
1511 	fs->fs_fmod = 1;
1512 	bdwrite(bp);
1513 	return (0);
1514 }
1515 
1516 /*
1517  * Find a block of the specified size in the specified cylinder group.
1518  *
1519  * It is a panic if a request is made to find a block if none are
1520  * available.
1521  */
1522 static ufs_daddr_t
1523 ffs_mapsearch(fs, cgp, bpref, allocsiz)
1524 	register struct fs *fs;
1525 	register struct cg *cgp;
1526 	ufs_daddr_t bpref;
1527 	int allocsiz;
1528 {
1529 	ufs_daddr_t bno;
1530 	int start, len, loc, i;
1531 	int blk, field, subfield, pos;
1532 
1533 	/*
1534 	 * find the fragment by searching through the free block
1535 	 * map for an appropriate bit pattern
1536 	 */
1537 	if (bpref)
1538 		start = dtogd(fs, bpref) / NBBY;
1539 	else
1540 		start = cgp->cg_frotor / NBBY;
1541 	len = howmany(fs->fs_fpg, NBBY) - start;
1542 	loc = scanc((u_int)len, (u_char *)&cg_blksfree(cgp)[start],
1543 		(u_char *)fragtbl[fs->fs_frag],
1544 		(u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
1545 	if (loc == 0) {
1546 		len = start + 1;
1547 		start = 0;
1548 		loc = scanc((u_int)len, (u_char *)&cg_blksfree(cgp)[0],
1549 			(u_char *)fragtbl[fs->fs_frag],
1550 			(u_char)(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
1551 		if (loc == 0) {
1552 			printf("start = %d, len = %d, fs = %s\n",
1553 			    start, len, fs->fs_fsmnt);
1554 			panic("ffs_alloccg: map corrupted");
1555 			/* NOTREACHED */
1556 		}
1557 	}
1558 	bno = (start + len - loc) * NBBY;
1559 	cgp->cg_frotor = bno;
1560 	/*
1561 	 * found the byte in the map
1562 	 * sift through the bits to find the selected frag
1563 	 */
1564 	for (i = bno + NBBY; bno < i; bno += fs->fs_frag) {
1565 		blk = blkmap(fs, cg_blksfree(cgp), bno);
1566 		blk <<= 1;
1567 		field = around[allocsiz];
1568 		subfield = inside[allocsiz];
1569 		for (pos = 0; pos <= fs->fs_frag - allocsiz; pos++) {
1570 			if ((blk & field) == subfield)
1571 				return (bno + pos);
1572 			field <<= 1;
1573 			subfield <<= 1;
1574 		}
1575 	}
1576 	printf("bno = %lu, fs = %s\n", (u_long)bno, fs->fs_fsmnt);
1577 	panic("ffs_alloccg: block not in map");
1578 	return (-1);
1579 }
1580 
1581 /*
1582  * Update the cluster map because of an allocation or free.
1583  *
1584  * Cnt == 1 means free; cnt == -1 means allocating.
1585  */
1586 static void
1587 ffs_clusteracct(fs, cgp, blkno, cnt)
1588 	struct fs *fs;
1589 	struct cg *cgp;
1590 	ufs_daddr_t blkno;
1591 	int cnt;
1592 {
1593 	int32_t *sump;
1594 	int32_t *lp;
1595 	u_char *freemapp, *mapp;
1596 	int i, start, end, forw, back, map, bit;
1597 
1598 	if (fs->fs_contigsumsize <= 0)
1599 		return;
1600 	freemapp = cg_clustersfree(cgp);
1601 	sump = cg_clustersum(cgp);
1602 	/*
1603 	 * Allocate or clear the actual block.
1604 	 */
1605 	if (cnt > 0)
1606 		setbit(freemapp, blkno);
1607 	else
1608 		clrbit(freemapp, blkno);
1609 	/*
1610 	 * Find the size of the cluster going forward.
1611 	 */
1612 	start = blkno + 1;
1613 	end = start + fs->fs_contigsumsize;
1614 	if (end >= cgp->cg_nclusterblks)
1615 		end = cgp->cg_nclusterblks;
1616 	mapp = &freemapp[start / NBBY];
1617 	map = *mapp++;
1618 	bit = 1 << (start % NBBY);
1619 	for (i = start; i < end; i++) {
1620 		if ((map & bit) == 0)
1621 			break;
1622 		if ((i & (NBBY - 1)) != (NBBY - 1)) {
1623 			bit <<= 1;
1624 		} else {
1625 			map = *mapp++;
1626 			bit = 1;
1627 		}
1628 	}
1629 	forw = i - start;
1630 	/*
1631 	 * Find the size of the cluster going backward.
1632 	 */
1633 	start = blkno - 1;
1634 	end = start - fs->fs_contigsumsize;
1635 	if (end < 0)
1636 		end = -1;
1637 	mapp = &freemapp[start / NBBY];
1638 	map = *mapp--;
1639 	bit = 1 << (start % NBBY);
1640 	for (i = start; i > end; i--) {
1641 		if ((map & bit) == 0)
1642 			break;
1643 		if ((i & (NBBY - 1)) != 0) {
1644 			bit >>= 1;
1645 		} else {
1646 			map = *mapp--;
1647 			bit = 1 << (NBBY - 1);
1648 		}
1649 	}
1650 	back = start - i;
1651 	/*
1652 	 * Account for old cluster and the possibly new forward and
1653 	 * back clusters.
1654 	 */
1655 	i = back + forw + 1;
1656 	if (i > fs->fs_contigsumsize)
1657 		i = fs->fs_contigsumsize;
1658 	sump[i] += cnt;
1659 	if (back > 0)
1660 		sump[back] -= cnt;
1661 	if (forw > 0)
1662 		sump[forw] -= cnt;
1663 	/*
1664 	 * Update cluster summary information.
1665 	 */
1666 	lp = &sump[fs->fs_contigsumsize];
1667 	for (i = fs->fs_contigsumsize; i > 0; i--)
1668 		if (*lp-- > 0)
1669 			break;
1670 	fs->fs_maxcluster[cgp->cg_cgx] = i;
1671 }
1672 
1673 /*
1674  * Fserr prints the name of a file system with an error diagnostic.
1675  *
1676  * The form of the error message is:
1677  *	fs: error message
1678  */
1679 static void
1680 ffs_fserr(fs, uid, cp)
1681 	struct fs *fs;
1682 	u_int uid;
1683 	char *cp;
1684 {
1685 	struct proc *p = curproc;	/* XXX */
1686 
1687 	log(LOG_ERR, "pid %d (%s), uid %d on %s: %s\n", p ? p->p_pid : -1,
1688 			p ? p->p_comm : "-", uid, fs->fs_fsmnt, cp);
1689 }
1690