xref: /netbsd-src/sys/fs/msdosfs/msdosfs_fat.c (revision 5bbd2a12505d72a8177929a37b5cee489d0a1cfd)
1 /*	$NetBSD: msdosfs_fat.c,v 1.20 2012/04/09 11:10:06 hannken Exp $	*/
2 
3 /*-
4  * Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
5  * Copyright (C) 1994, 1995, 1997 TooLs GmbH.
6  * All rights reserved.
7  * Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by TooLs GmbH.
20  * 4. The name of TooLs GmbH may not be used to endorse or promote products
21  *    derived from this software without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
24  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
25  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
26  * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
27  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
28  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
29  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
30  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
31  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
32  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
33  */
34 /*
35  * Written by Paul Popelka (paulp@uts.amdahl.com)
36  *
37  * You can do anything you want with this software, just don't say you wrote
38  * it, and don't remove this notice.
39  *
40  * This software is provided "as is".
41  *
42  * The author supplies this software to be publicly redistributed on the
43  * understanding that the author is not responsible for the correct
44  * functioning of this software in any circumstances and is not liable for
45  * any damages caused by this software.
46  *
47  * October 1992
48  */
49 
50 #include <sys/cdefs.h>
51 __KERNEL_RCSID(0, "$NetBSD: msdosfs_fat.c,v 1.20 2012/04/09 11:10:06 hannken Exp $");
52 
53 /*
54  * kernel include files.
55  */
56 #include <sys/param.h>
57 #include <sys/systm.h>
58 #include <sys/buf.h>
59 #include <sys/file.h>
60 #include <sys/namei.h>
61 #include <sys/mount.h>		/* to define statvfs structure */
62 #include <sys/vnode.h>		/* to define vattr structure */
63 #include <sys/errno.h>
64 #include <sys/dirent.h>
65 #include <sys/kauth.h>
66 
67 /*
68  * msdosfs include files.
69  */
70 #include <fs/msdosfs/bpb.h>
71 #include <fs/msdosfs/msdosfsmount.h>
72 #include <fs/msdosfs/direntry.h>
73 #include <fs/msdosfs/denode.h>
74 #include <fs/msdosfs/fat.h>
75 
76 /*
77  * Fat cache stats.
78  */
79 int fc_fileextends;		/* # of file extends			 */
80 int fc_lfcempty;		/* # of time last file cluster cache entry
81 				 * was empty */
82 int fc_bmapcalls;		/* # of times pcbmap was called		 */
83 
84 #define	LMMAX	20
85 int fc_lmdistance[LMMAX];	/* counters for how far off the last
86 				 * cluster mapped entry was. */
87 int fc_largedistance;		/* off by more than LMMAX		 */
88 int fc_wherefrom, fc_whereto, fc_lastclust;
89 int pm_fatblocksize;
90 
91 #ifdef MSDOSFS_DEBUG
92 void print_fat_stats(void);
93 
94 void
95 print_fat_stats(void)
96 {
97 	int i;
98 
99 	printf("fc_fileextends=%d fc_lfcempty=%d fc_bmapcalls=%d "
100 	    "fc_largedistance=%d [%d->%d=%d] fc_lastclust=%d pm_fatblocksize=%d\n",
101 	    fc_fileextends, fc_lfcempty, fc_bmapcalls, fc_largedistance,
102 	    fc_wherefrom, fc_whereto, fc_whereto-fc_wherefrom,
103 	    fc_lastclust, pm_fatblocksize);
104 
105 	fc_fileextends = fc_lfcempty = fc_bmapcalls = 0;
106 	fc_wherefrom = fc_whereto = fc_lastclust = 0;
107 
108 	for (i = 0; i < LMMAX; i++) {
109 		printf("%d:%d ", i, fc_lmdistance[i]);
110 	fc_lmdistance[i] = 0;
111 	}
112 
113 	printf("\n");
114 }
115 #endif
116 
117 static void fatblock(struct msdosfsmount *, u_long, u_long *, u_long *,
118 			  u_long *);
119 void updatefats(struct msdosfsmount *, struct buf *, u_long);
120 static inline void usemap_free(struct msdosfsmount *, u_long);
121 static inline void usemap_alloc(struct msdosfsmount *, u_long);
122 static int fatchain(struct msdosfsmount *, u_long, u_long, u_long);
123 int chainlength(struct msdosfsmount *, u_long, u_long);
124 int chainalloc(struct msdosfsmount *, u_long, u_long, u_long, u_long *,
125 		    u_long *);
126 
127 static void
128 fatblock(struct msdosfsmount *pmp, u_long ofs, u_long *bnp, u_long *sizep, u_long *bop)
129 {
130 	u_long bn, size;
131 
132 	bn = ofs / pmp->pm_fatblocksize * pmp->pm_fatblocksec;
133 	size = min(pmp->pm_fatblocksec, pmp->pm_FATsecs - bn)
134 	    * pmp->pm_BytesPerSec;
135 	bn += pmp->pm_fatblk + pmp->pm_curfat * pmp->pm_FATsecs;
136 
137 	if (bnp)
138 		*bnp = bn;
139 	if (sizep)
140 		*sizep = size;
141 	if (bop)
142 		*bop = ofs % pmp->pm_fatblocksize;
143 
144 	pm_fatblocksize =  pmp->pm_fatblocksize;
145 }
146 
147 /*
148  * Map the logical cluster number of a file into a physical disk sector
149  * that is filesystem relative.
150  *
151  * dep	  - address of denode representing the file of interest
152  * findcn - file relative cluster whose filesystem relative cluster number
153  *	    and/or block number are/is to be found
154  * bnp	  - address of where to place the file system relative block number.
155  *	    If this pointer is null then don't return this quantity.
156  * cnp	  - address of where to place the file system relative cluster number.
157  *	    If this pointer is null then don't return this quantity.
158  *
159  * NOTE: Either bnp or cnp must be non-null.
160  * This function has one side effect.  If the requested file relative cluster
161  * is beyond the end of file, then the actual number of clusters in the file
162  * is returned in *cnp.  This is useful for determining how long a directory is.
163  *  If cnp is null, nothing is returned.
164  */
165 int
166 pcbmap(struct denode *dep, u_long findcn, daddr_t *bnp, u_long *cnp, int *sp)
167 	/* findcn:		 file relative cluster to get		 */
168 	/* bnp:		 returned filesys rel sector number	 */
169 	/* cnp:		 returned cluster number		 */
170 	/* sp:		 returned block size			 */
171 {
172 	int error;
173 	u_long i;
174 	u_long cn;
175 	u_long prevcn = 0; /* XXX: prevcn could be used unititialized */
176 	u_long byteoffset;
177 	u_long bn;
178 	u_long bo;
179 	struct buf *bp = NULL;
180 	u_long bp_bn = -1;
181 	struct msdosfsmount *pmp = dep->de_pmp;
182 	u_long bsize;
183 
184 	fc_bmapcalls++;
185 
186 	/*
187 	 * If they don't give us someplace to return a value then don't
188 	 * bother doing anything.
189 	 */
190 	if (bnp == NULL && cnp == NULL && sp == NULL)
191 		return (0);
192 
193 	cn = dep->de_StartCluster;
194 	/*
195 	 * The "file" that makes up the root directory is contiguous,
196 	 * permanently allocated, of fixed size, and is not made up of
197 	 * clusters.  If the cluster number is beyond the end of the root
198 	 * directory, then return the number of clusters in the file.
199 	 */
200 	if (cn == MSDOSFSROOT) {
201 		if (dep->de_Attributes & ATTR_DIRECTORY) {
202 			if (de_cn2off(pmp, findcn) >= dep->de_FileSize) {
203 				if (cnp)
204 					*cnp = de_bn2cn(pmp, pmp->pm_rootdirsize);
205 				return (E2BIG);
206 			}
207 			if (bnp)
208 				*bnp = pmp->pm_rootdirblk + de_cn2bn(pmp, findcn);
209 			if (cnp)
210 				*cnp = MSDOSFSROOT;
211 			if (sp)
212 				*sp = min(pmp->pm_bpcluster,
213 				    dep->de_FileSize - de_cn2off(pmp, findcn));
214 			return (0);
215 		} else {		/* just an empty file */
216 			if (cnp)
217 				*cnp = 0;
218 			return (E2BIG);
219 		}
220 	}
221 
222 	/*
223 	 * All other files do I/O in cluster sized blocks
224 	 */
225 	if (sp)
226 		*sp = pmp->pm_bpcluster;
227 
228 	/*
229 	 * Rummage around in the fat cache, maybe we can avoid tromping
230 	 * thru every fat entry for the file. And, keep track of how far
231 	 * off the cache was from where we wanted to be.
232 	 */
233 	i = 0;
234 	fc_lookup(dep, findcn, &i, &cn);
235 	if ((bn = findcn - i) >= LMMAX) {
236 		fc_largedistance++;
237 		fc_wherefrom = i;
238 		fc_whereto = findcn;
239 		fc_lastclust = dep->de_fc[FC_LASTFC].fc_frcn;
240 	} else
241 		fc_lmdistance[bn]++;
242 
243 	/*
244 	 * Handle all other files or directories the normal way.
245 	 */
246 	for (; i < findcn; i++) {
247 		/*
248 		 * Stop with all reserved clusters, not just with EOF.
249 		 */
250 		if (cn >= (CLUST_RSRVD & pmp->pm_fatmask))
251 			goto hiteof;
252 		byteoffset = FATOFS(pmp, cn);
253 		fatblock(pmp, byteoffset, &bn, &bsize, &bo);
254 		if (bn != bp_bn) {
255 			if (bp)
256 				brelse(bp, 0);
257 			bp = getblk(pmp->pm_devvp, de_bn2kb(pmp, bn), bsize,
258 			    0, 0);
259 			if (bp == NULL) {
260 				/*
261 				 * getblk() above returns NULL only iff we are
262 				 * pagedaemon.  See the implementation of getblk
263 				 * for detail.
264 				 */
265 				return ENOMEM;
266 			}
267 			if (!ISSET(bp->b_oflags, (BO_DONE | BO_DELWRI))) {
268 				SET(bp->b_flags, B_READ);
269 				BIO_SETPRIO(bp, BPRIO_TIMECRITICAL);
270 				VOP_STRATEGY(pmp->pm_devvp, bp);
271 				curlwp->l_ru.ru_inblock++;
272 				error = biowait(bp);
273 				if (error) {
274 					brelse(bp, 0);
275 					return error;
276 				}
277 			}
278 			bp_bn = bn;
279 		}
280 		prevcn = cn;
281 		if (bo >= bsize) {
282 			if (bp)
283 				brelse(bp, 0);
284 			return (EIO);
285 		}
286 		KASSERT(bp != NULL);
287 		if (FAT32(pmp))
288 			cn = getulong((char *)bp->b_data + bo);
289 		else
290 			cn = getushort((char *)bp->b_data + bo);
291 		if (FAT12(pmp) && (prevcn & 1))
292 			cn >>= 4;
293 		cn &= pmp->pm_fatmask;
294 	}
295 
296 	if (!MSDOSFSEOF(cn, pmp->pm_fatmask)) {
297 		if (bp)
298 			brelse(bp, 0);
299 		if (bnp)
300 			*bnp = cntobn(pmp, cn);
301 		if (cnp)
302 			*cnp = cn;
303 		fc_setcache(dep, FC_LASTMAP, i, cn);
304 		return (0);
305 	}
306 
307 hiteof:;
308 	if (cnp)
309 		*cnp = i;
310 	if (bp)
311 		brelse(bp, 0);
312 	/* update last file cluster entry in the fat cache */
313 	fc_setcache(dep, FC_LASTFC, i - 1, prevcn);
314 	return (E2BIG);
315 }
316 
317 /*
318  * Find the closest entry in the fat cache to the cluster we are looking
319  * for.
320  */
321 void
322 fc_lookup(struct denode *dep, u_long findcn, u_long *frcnp, u_long *fsrcnp)
323 {
324 	int i;
325 	u_long cn;
326 	struct fatcache *closest = 0;
327 
328 	for (i = 0; i < FC_SIZE; i++) {
329 		cn = dep->de_fc[i].fc_frcn;
330 		if (cn != FCE_EMPTY && cn <= findcn) {
331 			if (closest == 0 || cn > closest->fc_frcn)
332 				closest = &dep->de_fc[i];
333 		}
334 	}
335 	if (closest) {
336 		*frcnp = closest->fc_frcn;
337 		*fsrcnp = closest->fc_fsrcn;
338 	}
339 }
340 
341 /*
342  * Purge the fat cache in denode dep of all entries relating to file
343  * relative cluster frcn and beyond.
344  */
345 void
346 fc_purge(struct denode *dep, u_int frcn)
347 {
348 	int i;
349 	struct fatcache *fcp;
350 
351 	fcp = dep->de_fc;
352 	for (i = 0; i < FC_SIZE; i++, fcp++) {
353 		if (fcp->fc_frcn >= frcn)
354 			fcp->fc_frcn = FCE_EMPTY;
355 	}
356 }
357 
358 /*
359  * Update the fat.
360  * If mirroring the fat, update all copies, with the first copy as last.
361  * Else update only the current fat (ignoring the others).
362  *
363  * pmp	 - msdosfsmount structure for filesystem to update
364  * bp	 - addr of modified fat block
365  * fatbn - block number relative to begin of filesystem of the modified fat block.
366  */
367 void
368 updatefats(struct msdosfsmount *pmp, struct buf *bp, u_long fatbn)
369 {
370 	int i;
371 	struct buf *bpn;
372 
373 #ifdef MSDOSFS_DEBUG
374 	printf("updatefats(pmp %p, bp %p, fatbn %lu)\n",
375 	    pmp, bp, fatbn);
376 #endif
377 
378 	/*
379 	 * If we have an FSInfo block, update it.
380 	 */
381 	if (pmp->pm_fsinfo) {
382 		u_long cn = pmp->pm_nxtfree;
383 
384 		if (pmp->pm_freeclustercount
385 		    && (pmp->pm_inusemap[cn / N_INUSEBITS]
386 			& (1 << (cn % N_INUSEBITS)))) {
387 			/*
388 			 * The cluster indicated in FSInfo isn't free
389 			 * any longer.  Got get a new free one.
390 			 */
391 			for (cn = 0; cn < pmp->pm_maxcluster; cn++)
392 				if (pmp->pm_inusemap[cn / N_INUSEBITS] != (u_int)-1)
393 					break;
394 			pmp->pm_nxtfree = cn
395 				+ ffs(pmp->pm_inusemap[cn / N_INUSEBITS]
396 				      ^ (u_int)-1) - 1;
397 		}
398 		/*
399 		 * XXX  If the fsinfo block is stored on media with
400 		 *      2KB or larger sectors, is the fsinfo structure
401 		 *      padded at the end or in the middle?
402 		 */
403 		if (bread(pmp->pm_devvp, de_bn2kb(pmp, pmp->pm_fsinfo),
404 		    pmp->pm_BytesPerSec, NOCRED, B_MODIFY, &bpn) != 0) {
405 			/*
406 			 * Ignore the error, but turn off FSInfo update for the future.
407 			 */
408 			pmp->pm_fsinfo = 0;
409 			brelse(bpn, 0);
410 		} else {
411 			struct fsinfo *fp = (struct fsinfo *)bpn->b_data;
412 
413 			putulong(fp->fsinfree, pmp->pm_freeclustercount);
414 			putulong(fp->fsinxtfree, pmp->pm_nxtfree);
415 			if (pmp->pm_flags & MSDOSFSMNT_WAITONFAT)
416 				bwrite(bpn);
417 			else
418 				bdwrite(bpn);
419 		}
420 	}
421 
422 	if (pmp->pm_flags & MSDOSFS_FATMIRROR) {
423 		/*
424 		 * Now copy the block(s) of the modified fat to the other copies of
425 		 * the fat and write them out.  This is faster than reading in the
426 		 * other fats and then writing them back out.  This could tie up
427 		 * the fat for quite a while. Preventing others from accessing it.
428 		 * To prevent us from going after the fat quite so much we use
429 		 * delayed writes, unless they specified "synchronous" when the
430 		 * filesystem was mounted.  If synch is asked for then use
431 		 * bwrite()'s and really slow things down.
432 		 */
433 		for (i = 1; i < pmp->pm_FATs; i++) {
434 			fatbn += pmp->pm_FATsecs;
435 			/* getblk() never fails */
436 			bpn = getblk(pmp->pm_devvp, de_bn2kb(pmp, fatbn),
437 			    bp->b_bcount, 0, 0);
438 			memcpy(bpn->b_data, bp->b_data, bp->b_bcount);
439 			if (pmp->pm_flags & MSDOSFSMNT_WAITONFAT)
440 				bwrite(bpn);
441 			else
442 				bdwrite(bpn);
443 		}
444 	}
445 
446 	/*
447 	 * Write out the first (or current) fat last.
448 	 */
449 	if (pmp->pm_flags & MSDOSFSMNT_WAITONFAT)
450 		bwrite(bp);
451 	else
452 		bdwrite(bp);
453 	/*
454 	 * Maybe update fsinfo sector here?
455 	 */
456 }
457 
458 /*
459  * Updating entries in 12 bit fats is a pain in the butt.
460  *
461  * The following picture shows where nibbles go when moving from a 12 bit
462  * cluster number into the appropriate bytes in the FAT.
463  *
464  *	byte m        byte m+1      byte m+2
465  *	+----+----+   +----+----+   +----+----+
466  *	|  0    1 |   |  2    3 |   |  4    5 |   FAT bytes
467  *	+----+----+   +----+----+   +----+----+
468  *
469  *	+----+----+----+   +----+----+----+
470  *	|  3    0    1 |   |  4    5    2 |
471  *	+----+----+----+   +----+----+----+
472  *	cluster n  	   cluster n+1
473  *
474  * Where n is even. m = n + (n >> 2)
475  *
476  */
477 static inline void
478 usemap_alloc(struct msdosfsmount *pmp, u_long cn)
479 {
480 
481 	pmp->pm_inusemap[cn / N_INUSEBITS] |= 1 << (cn % N_INUSEBITS);
482 	pmp->pm_freeclustercount--;
483 }
484 
485 static inline void
486 usemap_free(struct msdosfsmount *pmp, u_long cn)
487 {
488 
489 	pmp->pm_freeclustercount++;
490 	pmp->pm_inusemap[cn / N_INUSEBITS] &= ~(1 << (cn % N_INUSEBITS));
491 }
492 
493 int
494 clusterfree(struct msdosfsmount *pmp, u_long cluster, u_long *oldcnp)
495 {
496 	int error;
497 	u_long oldcn;
498 
499 	usemap_free(pmp, cluster);
500 	error = fatentry(FAT_GET_AND_SET, pmp, cluster, &oldcn, MSDOSFSFREE);
501 	if (error) {
502 		usemap_alloc(pmp, cluster);
503 		return (error);
504 	}
505 	/*
506 	 * If the cluster was successfully marked free, then update
507 	 * the count of free clusters, and turn off the "allocated"
508 	 * bit in the "in use" cluster bit map.
509 	 */
510 	if (oldcnp)
511 		*oldcnp = oldcn;
512 	return (0);
513 }
514 
515 /*
516  * Get or Set or 'Get and Set' the cluster'th entry in the fat.
517  *
518  * function	- whether to get or set a fat entry
519  * pmp		- address of the msdosfsmount structure for the filesystem
520  *		  whose fat is to be manipulated.
521  * cn		- which cluster is of interest
522  * oldcontents	- address of a word that is to receive the contents of the
523  *		  cluster'th entry if this is a get function
524  * newcontents	- the new value to be written into the cluster'th element of
525  *		  the fat if this is a set function.
526  *
527  * This function can also be used to free a cluster by setting the fat entry
528  * for a cluster to 0.
529  *
530  * All copies of the fat are updated if this is a set function. NOTE: If
531  * fatentry() marks a cluster as free it does not update the inusemap in
532  * the msdosfsmount structure. This is left to the caller.
533  */
534 int
535 fatentry(int function, struct msdosfsmount *pmp, u_long cn, u_long *oldcontents, u_long newcontents)
536 {
537 	int error;
538 	u_long readcn;
539 	u_long bn, bo, bsize, byteoffset;
540 	struct buf *bp;
541 
542 #ifdef	MSDOSFS_DEBUG
543 	printf("fatentry(func %d, pmp %p, clust %lu, oldcon %p, newcon %lx)\n",
544 	     function, pmp, cn, oldcontents, newcontents);
545 #endif
546 
547 #ifdef DIAGNOSTIC
548 	/*
549 	 * Be sure they asked us to do something.
550 	 */
551 	if ((function & (FAT_SET | FAT_GET)) == 0) {
552 		printf("fatentry(): function code doesn't specify get or set\n");
553 		return (EINVAL);
554 	}
555 
556 	/*
557 	 * If they asked us to return a cluster number but didn't tell us
558 	 * where to put it, give them an error.
559 	 */
560 	if ((function & FAT_GET) && oldcontents == NULL) {
561 		printf("fatentry(): get function with no place to put result\n");
562 		return (EINVAL);
563 	}
564 #endif
565 
566 	/*
567 	 * Be sure the requested cluster is in the filesystem.
568 	 */
569 	if (cn < CLUST_FIRST || cn > pmp->pm_maxcluster)
570 		return (EINVAL);
571 
572 	byteoffset = FATOFS(pmp, cn);
573 	fatblock(pmp, byteoffset, &bn, &bsize, &bo);
574 	if ((error = bread(pmp->pm_devvp, de_bn2kb(pmp, bn), bsize, NOCRED,
575 	    0, &bp)) != 0) {
576 		brelse(bp, 0);
577 		return (error);
578 	}
579 
580 	if (function & FAT_GET) {
581 		if (FAT32(pmp))
582 			readcn = getulong((char *)bp->b_data + bo);
583 		else
584 			readcn = getushort((char *)bp->b_data + bo);
585 		if (FAT12(pmp) & (cn & 1))
586 			readcn >>= 4;
587 		readcn &= pmp->pm_fatmask;
588 		*oldcontents = readcn;
589 	}
590 	if (function & FAT_SET) {
591 		switch (pmp->pm_fatmask) {
592 		case FAT12_MASK:
593 			readcn = getushort((char *)bp->b_data + bo);
594 			if (cn & 1) {
595 				readcn &= 0x000f;
596 				readcn |= newcontents << 4;
597 			} else {
598 				readcn &= 0xf000;
599 				readcn |= newcontents & 0xfff;
600 			}
601 			putushort((char *)bp->b_data + bo, readcn);
602 			break;
603 		case FAT16_MASK:
604 			putushort((char *)bp->b_data + bo, newcontents);
605 			break;
606 		case FAT32_MASK:
607 			/*
608 			 * According to spec we have to retain the
609 			 * high order bits of the fat entry.
610 			 */
611 			readcn = getulong((char *)bp->b_data + bo);
612 			readcn &= ~FAT32_MASK;
613 			readcn |= newcontents & FAT32_MASK;
614 			putulong((char *)bp->b_data + bo, readcn);
615 			break;
616 		}
617 		updatefats(pmp, bp, bn);
618 		bp = NULL;
619 		pmp->pm_fmod = 1;
620 	}
621 	if (bp)
622 		brelse(bp, 0);
623 	return (0);
624 }
625 
626 /*
627  * Update a contiguous cluster chain
628  *
629  * pmp	    - mount point
630  * start    - first cluster of chain
631  * count    - number of clusters in chain
632  * fillwith - what to write into fat entry of last cluster
633  */
634 static int
635 fatchain(struct msdosfsmount *pmp, u_long start, u_long count, u_long fillwith)
636 {
637 	int error;
638 	u_long bn, bo, bsize, byteoffset, readcn, newc;
639 	struct buf *bp;
640 
641 #ifdef MSDOSFS_DEBUG
642 	printf("fatchain(pmp %p, start %lu, count %lu, fillwith %lx)\n",
643 	    pmp, start, count, fillwith);
644 #endif
645 	/*
646 	 * Be sure the clusters are in the filesystem.
647 	 */
648 	if (start < CLUST_FIRST || start + count - 1 > pmp->pm_maxcluster)
649 		return (EINVAL);
650 
651 	while (count > 0) {
652 		byteoffset = FATOFS(pmp, start);
653 		fatblock(pmp, byteoffset, &bn, &bsize, &bo);
654 		error = bread(pmp->pm_devvp, de_bn2kb(pmp, bn), bsize, NOCRED,
655 		    B_MODIFY, &bp);
656 		if (error) {
657 			brelse(bp, 0);
658 			return (error);
659 		}
660 		while (count > 0) {
661 			start++;
662 			newc = --count > 0 ? start : fillwith;
663 			switch (pmp->pm_fatmask) {
664 			case FAT12_MASK:
665 				readcn = getushort((char *)bp->b_data + bo);
666 				if (start & 1) {
667 					readcn &= 0xf000;
668 					readcn |= newc & 0xfff;
669 				} else {
670 					readcn &= 0x000f;
671 					readcn |= newc << 4;
672 				}
673 				putushort((char *)bp->b_data + bo, readcn);
674 				bo++;
675 				if (!(start & 1))
676 					bo++;
677 				break;
678 			case FAT16_MASK:
679 				putushort((char *)bp->b_data + bo, newc);
680 				bo += 2;
681 				break;
682 			case FAT32_MASK:
683 				readcn = getulong((char *)bp->b_data + bo);
684 				readcn &= ~pmp->pm_fatmask;
685 				readcn |= newc & pmp->pm_fatmask;
686 				putulong((char *)bp->b_data + bo, readcn);
687 				bo += 4;
688 				break;
689 			}
690 			if (bo >= bsize)
691 				break;
692 		}
693 		updatefats(pmp, bp, bn);
694 	}
695 	pmp->pm_fmod = 1;
696 	return (0);
697 }
698 
699 /*
700  * Check the length of a free cluster chain starting at start.
701  *
702  * pmp	 - mount point
703  * start - start of chain
704  * count - maximum interesting length
705  */
706 int
707 chainlength(struct msdosfsmount *pmp, u_long start, u_long count)
708 {
709 	u_long idx, max_idx;
710 	u_int map;
711 	u_long len;
712 
713 	max_idx = pmp->pm_maxcluster / N_INUSEBITS;
714 	idx = start / N_INUSEBITS;
715 	start %= N_INUSEBITS;
716 	map = pmp->pm_inusemap[idx];
717 	map &= ~((1 << start) - 1);
718 	if (map) {
719 		len = ffs(map) - 1 - start;
720 		return (len > count ? count : len);
721 	}
722 	len = N_INUSEBITS - start;
723 	if (len >= count)
724 		return (count);
725 	while (++idx <= max_idx) {
726 		if (len >= count)
727 			break;
728 		if ((map = pmp->pm_inusemap[idx]) != 0) {
729 			len +=  ffs(map) - 1;
730 			break;
731 		}
732 		len += N_INUSEBITS;
733 	}
734 	return (len > count ? count : len);
735 }
736 
737 /*
738  * Allocate contigous free clusters.
739  *
740  * pmp	      - mount point.
741  * start      - start of cluster chain.
742  * count      - number of clusters to allocate.
743  * fillwith   - put this value into the fat entry for the
744  *		last allocated cluster.
745  * retcluster - put the first allocated cluster's number here.
746  * got	      - how many clusters were actually allocated.
747  */
748 int
749 chainalloc(struct msdosfsmount *pmp, u_long start, u_long count, u_long fillwith, u_long *retcluster, u_long *got)
750 {
751 	int error;
752 	u_long cl, n;
753 
754 	for (cl = start, n = count; n-- > 0;)
755 		usemap_alloc(pmp, cl++);
756 	if ((error = fatchain(pmp, start, count, fillwith)) != 0)
757 		return (error);
758 #ifdef MSDOSFS_DEBUG
759 	printf("clusteralloc(): allocated cluster chain at %lu (%lu clusters)\n",
760 	    start, count);
761 #endif
762 	if (retcluster)
763 		*retcluster = start;
764 	if (got)
765 		*got = count;
766 	return (0);
767 }
768 
769 /*
770  * Allocate contiguous free clusters.
771  *
772  * pmp	      - mount point.
773  * start      - preferred start of cluster chain.
774  * count      - number of clusters requested.
775  * fillwith   - put this value into the fat entry for the
776  *		last allocated cluster.
777  * retcluster - put the first allocated cluster's number here.
778  * got	      - how many clusters were actually allocated.
779  */
780 int
781 clusteralloc(struct msdosfsmount *pmp, u_long start, u_long count, u_long *retcluster, u_long *got)
782 {
783 	u_long idx;
784 	u_long len, newst, foundl, cn, l;
785 	u_long foundcn = 0; /* XXX: foundcn could be used unititialized */
786 	u_long fillwith = CLUST_EOFE;
787 	u_int map;
788 
789 #ifdef MSDOSFS_DEBUG
790 	printf("clusteralloc(): find %lu clusters\n",count);
791 #endif
792 	if (start) {
793 		if ((len = chainlength(pmp, start, count)) >= count)
794 			return (chainalloc(pmp, start, count, fillwith, retcluster, got));
795 	} else {
796 		/*
797 		 * This is a new file, initialize start
798 		 */
799 		struct timeval tv;
800 
801 		microtime(&tv);
802 		start = (tv.tv_usec >> 10) | tv.tv_usec;
803 		len = 0;
804 	}
805 
806 	/*
807 	 * Start at a (pseudo) random place to maximize cluster runs
808 	 * under multiple writers.
809 	 */
810 	newst = (start * 1103515245 + 12345) % (pmp->pm_maxcluster + 1);
811 	foundl = 0;
812 
813 	for (cn = newst; cn <= pmp->pm_maxcluster;) {
814 		idx = cn / N_INUSEBITS;
815 		map = pmp->pm_inusemap[idx];
816 		map |= (1 << (cn % N_INUSEBITS)) - 1;
817 		if (map != (u_int)-1) {
818 			cn = idx * N_INUSEBITS + ffs(map^(u_int)-1) - 1;
819 			if ((l = chainlength(pmp, cn, count)) >= count)
820 				return (chainalloc(pmp, cn, count, fillwith, retcluster, got));
821 			if (l > foundl) {
822 				foundcn = cn;
823 				foundl = l;
824 			}
825 			cn += l + 1;
826 			continue;
827 		}
828 		cn += N_INUSEBITS - cn % N_INUSEBITS;
829 	}
830 	for (cn = 0; cn < newst;) {
831 		idx = cn / N_INUSEBITS;
832 		map = pmp->pm_inusemap[idx];
833 		map |= (1 << (cn % N_INUSEBITS)) - 1;
834 		if (map != (u_int)-1) {
835 			cn = idx * N_INUSEBITS + ffs(map^(u_int)-1) - 1;
836 			if ((l = chainlength(pmp, cn, count)) >= count)
837 				return (chainalloc(pmp, cn, count, fillwith, retcluster, got));
838 			if (l > foundl) {
839 				foundcn = cn;
840 				foundl = l;
841 			}
842 			cn += l + 1;
843 			continue;
844 		}
845 		cn += N_INUSEBITS - cn % N_INUSEBITS;
846 	}
847 
848 	if (!foundl)
849 		return (ENOSPC);
850 
851 	if (len)
852 		return (chainalloc(pmp, start, len, fillwith, retcluster, got));
853 	else
854 		return (chainalloc(pmp, foundcn, foundl, fillwith, retcluster, got));
855 }
856 
857 
858 /*
859  * Free a chain of clusters.
860  *
861  * pmp		- address of the msdosfs mount structure for the filesystem
862  *		  containing the cluster chain to be freed.
863  * startcluster - number of the 1st cluster in the chain of clusters to be
864  *		  freed.
865  */
866 int
867 freeclusterchain(struct msdosfsmount *pmp, u_long cluster)
868 {
869 	int error;
870 	struct buf *bp = NULL;
871 	u_long bn, bo, bsize, byteoffset;
872 	u_long readcn, lbn = -1;
873 
874 	while (cluster >= CLUST_FIRST && cluster <= pmp->pm_maxcluster) {
875 		byteoffset = FATOFS(pmp, cluster);
876 		fatblock(pmp, byteoffset, &bn, &bsize, &bo);
877 		if (lbn != bn) {
878 			if (bp)
879 				updatefats(pmp, bp, lbn);
880 			error = bread(pmp->pm_devvp, de_bn2kb(pmp, bn), bsize,
881 			    NOCRED, B_MODIFY, &bp);
882 			if (error) {
883 				brelse(bp, 0);
884 				return (error);
885 			}
886 			lbn = bn;
887 		}
888 		usemap_free(pmp, cluster);
889 		KASSERT(bp != NULL);
890 		switch (pmp->pm_fatmask) {
891 		case FAT12_MASK:
892 			readcn = getushort((char *)bp->b_data + bo);
893 			if (cluster & 1) {
894 				cluster = readcn >> 4;
895 				readcn &= 0x000f;
896 				readcn |= MSDOSFSFREE << 4;
897 			} else {
898 				cluster = readcn;
899 				readcn &= 0xf000;
900 				readcn |= MSDOSFSFREE & 0xfff;
901 			}
902 			putushort((char *)bp->b_data + bo, readcn);
903 			break;
904 		case FAT16_MASK:
905 			cluster = getushort((char *)bp->b_data + bo);
906 			putushort((char *)bp->b_data + bo, MSDOSFSFREE);
907 			break;
908 		case FAT32_MASK:
909 			cluster = getulong((char *)bp->b_data + bo);
910 			putulong((char *)bp->b_data + bo,
911 				 (MSDOSFSFREE & FAT32_MASK) | (cluster & ~FAT32_MASK));
912 			break;
913 		}
914 		cluster &= pmp->pm_fatmask;
915 	}
916 	if (bp)
917 		updatefats(pmp, bp, bn);
918 	return (0);
919 }
920 
921 /*
922  * Read in fat blocks looking for free clusters. For every free cluster
923  * found turn off its corresponding bit in the pm_inusemap.
924  */
925 int
926 fillinusemap(struct msdosfsmount *pmp)
927 {
928 	struct buf *bp = NULL;
929 	u_long cn, readcn;
930 	int error;
931 	u_long bn, bo, bsize, byteoffset;
932 
933 	/*
934 	 * Mark all clusters in use, we mark the free ones in the fat scan
935 	 * loop further down.
936 	 */
937 	for (cn = 0; cn < (pmp->pm_maxcluster + N_INUSEBITS) / N_INUSEBITS; cn++)
938 		pmp->pm_inusemap[cn] = (u_int)-1;
939 
940 	/*
941 	 * Figure how many free clusters are in the filesystem by ripping
942 	 * through the fat counting the number of entries whose content is
943 	 * zero.  These represent free clusters.
944 	 */
945 	pmp->pm_freeclustercount = 0;
946 	for (cn = CLUST_FIRST; cn <= pmp->pm_maxcluster; cn++) {
947 		byteoffset = FATOFS(pmp, cn);
948 		bo = byteoffset % pmp->pm_fatblocksize;
949 		if (!bo || !bp) {
950 			/* Read new FAT block */
951 			if (bp)
952 				brelse(bp, 0);
953 			fatblock(pmp, byteoffset, &bn, &bsize, NULL);
954 			error = bread(pmp->pm_devvp, de_bn2kb(pmp, bn), bsize,
955 			    NOCRED, 0, &bp);
956 			if (error) {
957 				brelse(bp, 0);
958 				return (error);
959 			}
960 		}
961 		if (FAT32(pmp))
962 			readcn = getulong((char *)bp->b_data + bo);
963 		else
964 			readcn = getushort((char *)bp->b_data + bo);
965 		if (FAT12(pmp) && (cn & 1))
966 			readcn >>= 4;
967 		readcn &= pmp->pm_fatmask;
968 
969 		if (readcn == 0)
970 			usemap_free(pmp, cn);
971 	}
972 	if (bp)
973 		brelse(bp, 0);
974 	return (0);
975 }
976 
977 /*
978  * Allocate a new cluster and chain it onto the end of the file.
979  *
980  * dep	 - the file to extend
981  * count - number of clusters to allocate
982  * bpp	 - where to return the address of the buf header for the first new
983  *	   file block
984  * ncp	 - where to put cluster number of the first newly allocated cluster
985  *	   If this pointer is 0, do not return the cluster number.
986  * flags - see fat.h
987  *
988  * NOTE: This function is not responsible for turning on the DE_UPDATE bit of
989  * the de_flag field of the denode and it does not change the de_FileSize
990  * field.  This is left for the caller to do.
991  */
992 
993 int
994 extendfile(struct denode *dep, u_long count, struct buf **bpp, u_long *ncp, int flags)
995 {
996 	int error;
997 	u_long frcn = 0, cn, got;
998 	struct msdosfsmount *pmp = dep->de_pmp;
999 	struct buf *bp;
1000 
1001 	/*
1002 	 * Don't try to extend the root directory
1003 	 */
1004 	if (dep->de_StartCluster == MSDOSFSROOT
1005 	    && (dep->de_Attributes & ATTR_DIRECTORY)) {
1006 		printf("extendfile(): attempt to extend root directory\n");
1007 		return (ENOSPC);
1008 	}
1009 
1010 	/*
1011 	 * If the "file's last cluster" cache entry is empty, and the file
1012 	 * is not empty, then fill the cache entry by calling pcbmap().
1013 	 */
1014 	fc_fileextends++;
1015 	if (dep->de_fc[FC_LASTFC].fc_frcn == FCE_EMPTY &&
1016 	    dep->de_StartCluster != 0) {
1017 		fc_lfcempty++;
1018 		error = pcbmap(dep, CLUST_END, 0, &cn, 0);
1019 		/* we expect it to return E2BIG */
1020 		if (error != E2BIG)
1021 			return (error);
1022 	}
1023 
1024 	fc_last_to_nexttolast(dep);
1025 
1026 	while (count > 0) {
1027 
1028 		/*
1029 		 * Allocate a new cluster chain and cat onto the end of the
1030 		 * file.  If the file is empty we make de_StartCluster point
1031 		 * to the new block.  Note that de_StartCluster being 0 is
1032 		 * sufficient to be sure the file is empty since we exclude
1033 		 * attempts to extend the root directory above, and the root
1034 		 * dir is the only file with a startcluster of 0 that has
1035 		 * blocks allocated (sort of).
1036 		 */
1037 
1038 		if (dep->de_StartCluster == 0)
1039 			cn = 0;
1040 		else
1041 			cn = dep->de_fc[FC_LASTFC].fc_fsrcn + 1;
1042 		error = clusteralloc(pmp, cn, count, &cn, &got);
1043 		if (error)
1044 			return (error);
1045 
1046 		count -= got;
1047 
1048 		/*
1049 		 * Give them the filesystem relative cluster number if they want
1050 		 * it.
1051 		 */
1052 		if (ncp) {
1053 			*ncp = cn;
1054 			ncp = NULL;
1055 		}
1056 
1057 		if (dep->de_StartCluster == 0) {
1058 			dep->de_StartCluster = cn;
1059 			frcn = 0;
1060 		} else {
1061 			error = fatentry(FAT_SET, pmp,
1062 					 dep->de_fc[FC_LASTFC].fc_fsrcn,
1063 					 0, cn);
1064 			if (error) {
1065 				clusterfree(pmp, cn, NULL);
1066 				return (error);
1067 			}
1068 			frcn = dep->de_fc[FC_LASTFC].fc_frcn + 1;
1069 		}
1070 
1071 		/*
1072 		 * Update the "last cluster of the file" entry in the
1073 		 * denode's fat cache.
1074 		 */
1075 
1076 		fc_setcache(dep, FC_LASTFC, frcn + got - 1, cn + got - 1);
1077 		if ((flags & DE_CLEAR) &&
1078 		    (dep->de_Attributes & ATTR_DIRECTORY)) {
1079 			while (got-- > 0) {
1080 				bp = getblk(pmp->pm_devvp,
1081 				    de_bn2kb(pmp, cntobn(pmp, cn++)),
1082 				    pmp->pm_bpcluster, 0, 0);
1083 				clrbuf(bp);
1084 				if (bpp) {
1085 					*bpp = bp;
1086 						bpp = NULL;
1087 				} else {
1088 					bdwrite(bp);
1089 				}
1090 			}
1091 		}
1092 	}
1093 
1094 	return (0);
1095 }
1096