bufcache.c revision 1.9 1 /* $NetBSD: bufcache.c,v 1.9 2006/04/19 15:52:58 christos Exp $ */
2 /*-
3 * Copyright (c) 2003 The NetBSD Foundation, Inc.
4 * All rights reserved.
5 *
6 * This code is derived from software contributed to The NetBSD Foundation
7 * by Konrad E. Schroder <perseant (at) hhhh.org>.
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 the NetBSD
20 * Foundation, Inc. and its contributors.
21 * 4. Neither the name of The NetBSD Foundation nor the names of its
22 * contributors may be used to endorse or promote products derived
23 * from this software without specific prior written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
26 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
29 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
38 #include <sys/types.h>
39 #include <sys/param.h>
40 #include <sys/time.h>
41 #include <sys/buf.h>
42 #include <sys/queue.h>
43 #include <sys/mount.h>
44
45 #include <assert.h>
46 #include <err.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51
52 #include "bufcache.h"
53 #include "vnode.h"
54
55 /*
56 * Definitions for the buffer free lists.
57 */
58 #define BQUEUES 3 /* number of free buffer queues */
59
60 #define BQ_LOCKED 0 /* super-blocks &c */
61 #define BQ_LRU 1 /* lru, useful buffers */
62 #define BQ_AGE 2 /* rubbish */
63
64 TAILQ_HEAD(bqueues, ubuf) bufqueues[BQUEUES];
65
66 struct bufhash_struct *bufhash;
67
68 #define HASH_MAX 1024
69 int hashmax = HASH_MAX;
70 int hashmask = (HASH_MAX - 1);
71
72 int maxbufs = BUF_CACHE_SIZE;
73 int nbufs = 0;
74 int cachehits = 0;
75 int cachemisses = 0;
76 int max_depth = 0;
77 off_t locked_queue_bytes = 0;
78 int locked_queue_count = 0;
79
80 /* Simple buffer hash function */
81 static int
82 vl_hash(struct uvnode * vp, daddr_t lbn)
83 {
84 return (int)((unsigned long) vp + lbn) & hashmask;
85 }
86
87 /* Initialize buffer cache */
88 void
89 bufinit(int max)
90 {
91 int i;
92
93 if (max) {
94 for (hashmax = 1; hashmax < max; hashmax <<= 1)
95 ;
96 hashmask = hashmax - 1;
97 }
98
99 for (i = 0; i < BQUEUES; i++) {
100 TAILQ_INIT(&bufqueues[i]);
101 }
102 bufhash = (struct bufhash_struct *)malloc(hashmax * sizeof(*bufhash));
103 if (bufhash == NULL)
104 err(1, NULL);
105 for (i = 0; i < hashmax; i++)
106 LIST_INIT(&bufhash[i]);
107 }
108
109 /* Widen the hash table. */
110 void bufrehash(int max)
111 {
112 int i, newhashmax, newhashmask;
113 struct ubuf *bp, *nbp;
114 struct bufhash_struct *np;
115
116 if (max < 0 || max < hashmax)
117 return;
118
119 /* Round up to a power of two */
120 for (newhashmax = 1; newhashmax < max; newhashmax <<= 1)
121 ;
122 newhashmask = newhashmax - 1;
123
124 /* Allocate new empty hash table, if we can */
125 np = (struct bufhash_struct *)malloc(newhashmax * sizeof(*bufhash));
126 if (np == NULL)
127 return;
128 for (i = 0; i < newhashmax; i++)
129 LIST_INIT(&np[i]);
130
131 /* Now reassign all existing buffers to their new hash chains. */
132 for (i = 0; i < hashmax; i++) {
133 bp = LIST_FIRST(&bufhash[i]);
134 while(bp) {
135 nbp = LIST_NEXT(bp, b_hash);
136 LIST_REMOVE(bp, b_hash);
137 bp->b_hashval = vl_hash(bp->b_vp, bp->b_lblkno);
138 LIST_INSERT_HEAD(&np[bp->b_hashval], bp, b_hash);
139 bp = nbp;
140 }
141 }
142
143 /* Switch over and clean up */
144 free(bufhash);
145 bufhash = np;
146 hashmax = newhashmax;
147 hashmask = newhashmask;
148 }
149
150 /* Print statistics of buffer cache usage */
151 void
152 bufstats(void)
153 {
154 printf("buffer cache: %d hits %d misses (%2.2f%%); hash width %d, depth %d\n",
155 cachehits, cachemisses,
156 (cachehits * 100.0) / (cachehits + cachemisses),
157 hashmax, max_depth);
158 }
159
160 /*
161 * Remove a buffer from the cache.
162 * Caller must remove the buffer from its free list.
163 */
164 void
165 buf_destroy(struct ubuf * bp)
166 {
167 bp->b_flags |= B_NEEDCOMMIT;
168 LIST_REMOVE(bp, b_vnbufs);
169 LIST_REMOVE(bp, b_hash);
170 if (!(bp->b_flags & B_DONTFREE))
171 free(bp->b_data);
172 free(bp);
173 --nbufs;
174 }
175
176 /* Remove a buffer from its free list. */
177 void
178 bremfree(struct ubuf * bp)
179 {
180 struct bqueues *dp = NULL;
181
182 /*
183 * We only calculate the head of the freelist when removing
184 * the last element of the list as that is the only time that
185 * it is needed (e.g. to reset the tail pointer).
186 *
187 * NB: This makes an assumption about how tailq's are implemented.
188 */
189 if (bp->b_flags & B_LOCKED) {
190 locked_queue_bytes -= bp->b_bcount;
191 --locked_queue_count;
192 }
193 if (TAILQ_NEXT(bp, b_freelist) == NULL) {
194 for (dp = bufqueues; dp < &bufqueues[BQUEUES]; dp++)
195 if (dp->tqh_last == &bp->b_freelist.tqe_next)
196 break;
197 if (dp == &bufqueues[BQUEUES])
198 errx(1, "bremfree: lost tail");
199 }
200 ++bp->b_vp->v_usecount;
201 TAILQ_REMOVE(dp, bp, b_freelist);
202 }
203
204 /* Return a buffer if it is in the cache, otherwise return NULL. */
205 struct ubuf *
206 incore(struct uvnode * vp, int lbn)
207 {
208 struct ubuf *bp;
209 int hash, depth;
210
211 hash = vl_hash(vp, lbn);
212 /* XXX use a real hash instead. */
213 depth = 0;
214 LIST_FOREACH(bp, &bufhash[hash], b_hash) {
215 if (++depth > max_depth)
216 max_depth = depth;
217 assert(depth <= nbufs);
218 if (bp->b_vp == vp && bp->b_lblkno == lbn) {
219 return bp;
220 }
221 }
222 return NULL;
223 }
224
225 /*
226 * Return a buffer of the given size, lbn and uvnode.
227 * If none is in core, make a new one.
228 */
229 struct ubuf *
230 getblk(struct uvnode * vp, daddr_t lbn, int size)
231 {
232 struct ubuf *bp;
233 #ifdef DEBUG
234 static int warned;
235 #endif
236
237 /*
238 * First check the buffer cache lists.
239 * We might sometimes need to resize a buffer. If we are growing
240 * the buffer, its contents are invalid; but shrinking is okay.
241 */
242 if ((bp = incore(vp, lbn)) != NULL) {
243 assert(!(bp->b_flags & B_NEEDCOMMIT));
244 assert(!(bp->b_flags & B_BUSY));
245 bp->b_flags |= B_BUSY;
246 bremfree(bp);
247 if (bp->b_bcount == size)
248 return bp;
249 else if (bp->b_bcount > size) {
250 assert(!(bp->b_flags & B_DELWRI));
251 bp->b_bcount = size;
252 bp->b_data = realloc(bp->b_data, size);
253 if (bp->b_data == NULL)
254 err(1, NULL);
255 return bp;
256 }
257
258 buf_destroy(bp);
259 bp = NULL;
260 }
261
262 /*
263 * Not on the list.
264 * Get a new block of the appropriate size and use that.
265 * If not enough space, free blocks from the AGE and LRU lists
266 * to make room.
267 */
268 while (nbufs >= maxbufs + locked_queue_count) {
269 bp = TAILQ_FIRST(&bufqueues[BQ_AGE]);
270 if (bp)
271 TAILQ_REMOVE(&bufqueues[BQ_AGE], bp, b_freelist);
272 if (bp == NULL) {
273 bp = TAILQ_FIRST(&bufqueues[BQ_LRU]);
274 if (bp)
275 TAILQ_REMOVE(&bufqueues[BQ_LRU], bp,
276 b_freelist);
277 }
278 if (bp) {
279 if (bp->b_flags & B_DELWRI)
280 VOP_STRATEGY(bp);
281 buf_destroy(bp);
282 break;
283 }
284 #ifdef DEBUG
285 else if (!warned) {
286 warnx("allocating more than %d buffers", maxbufs);
287 ++warned;
288 }
289 #endif
290 break;
291 }
292 ++nbufs;
293 bp = (struct ubuf *) malloc(sizeof(*bp));
294 if (bp == NULL)
295 err(1, NULL);
296 memset(bp, 0, sizeof(*bp));
297 bp->b_data = malloc(size);
298 if (bp->b_data == NULL)
299 err(1, NULL);
300 memset(bp->b_data, 0, size);
301
302 bp->b_vp = vp;
303 bp->b_blkno = bp->b_lblkno = lbn;
304 bp->b_bcount = size;
305 bp->b_hashval = vl_hash(vp, lbn);
306 LIST_INSERT_HEAD(&bufhash[bp->b_hashval], bp, b_hash);
307 LIST_INSERT_HEAD(&vp->v_cleanblkhd, bp, b_vnbufs);
308 bp->b_flags = B_BUSY;
309
310 return bp;
311 }
312
313 /* Write a buffer to disk according to its strategy routine. */
314 void
315 bwrite(struct ubuf * bp)
316 {
317 bp->b_flags &= ~(B_READ | B_DONE | B_DELWRI | B_LOCKED);
318 VOP_STRATEGY(bp);
319 bp->b_flags |= B_DONE;
320 reassignbuf(bp, bp->b_vp);
321 brelse(bp);
322 }
323
324 /* Put a buffer back on its free list, clear B_BUSY. */
325 void
326 brelse(struct ubuf * bp)
327 {
328 int age;
329
330 assert(!(bp->b_flags & B_NEEDCOMMIT));
331 assert(bp->b_flags & B_BUSY);
332
333 age = bp->b_flags & B_AGE;
334 bp->b_flags &= ~(B_BUSY | B_AGE);
335 if (bp->b_flags & B_INVAL) {
336 buf_destroy(bp);
337 return;
338 }
339 if (bp->b_flags & B_LOCKED) {
340 locked_queue_bytes += bp->b_bcount;
341 ++locked_queue_count;
342 TAILQ_INSERT_TAIL(&bufqueues[BQ_LOCKED], bp, b_freelist);
343 } else if (age) {
344 TAILQ_INSERT_TAIL(&bufqueues[BQ_AGE], bp, b_freelist);
345 } else {
346 TAILQ_INSERT_TAIL(&bufqueues[BQ_LRU], bp, b_freelist);
347 }
348 --bp->b_vp->v_usecount;
349
350 /* Move to the front of the hash chain */
351 if (LIST_FIRST(&bufhash[bp->b_hashval]) != bp) {
352 LIST_REMOVE(bp, b_hash);
353 LIST_INSERT_HEAD(&bufhash[bp->b_hashval], bp, b_hash);
354 }
355 }
356
357 /* Read the given block from disk, return it B_BUSY. */
358 int
359 bread(struct uvnode * vp, daddr_t lbn, int size, void * unused,
360 struct ubuf ** bpp)
361 {
362 struct ubuf *bp;
363 daddr_t daddr;
364 int error;
365
366 bp = getblk(vp, lbn, size);
367 *bpp = bp;
368 if (bp->b_flags & (B_DELWRI | B_DONE)) {
369 ++cachehits;
370 return 0;
371 }
372 ++cachemisses;
373
374 /*
375 * Not found. Need to find that block's location on disk,
376 * and load it in.
377 */
378 daddr = -1;
379 error = VOP_BMAP(vp, lbn, &daddr);
380 bp->b_blkno = daddr;
381 if (daddr >= 0) {
382 bp->b_flags |= B_READ;
383 return VOP_STRATEGY(bp);
384 }
385 memset(bp->b_data, 0, bp->b_bcount);
386 return 0;
387 }
388
389 /* Move a buffer between dirty and clean block lists. */
390 void
391 reassignbuf(struct ubuf * bp, struct uvnode * vp)
392 {
393 LIST_REMOVE(bp, b_vnbufs);
394 if (bp->b_flags & B_DELWRI) {
395 LIST_INSERT_HEAD(&vp->v_dirtyblkhd, bp, b_vnbufs);
396 } else {
397 LIST_INSERT_HEAD(&vp->v_cleanblkhd, bp, b_vnbufs);
398 }
399 }
400
401 #ifdef DEBUG
402 void
403 dump_free_lists(void)
404 {
405 struct ubuf *bp;
406 int i;
407
408 for (i = 0; i <= BQ_LOCKED; i++) {
409 printf("==> free list %d:\n", i);
410 TAILQ_FOREACH(bp, &bufqueues[i], b_freelist) {
411 printf("vp %p lbn %" PRId64 " flags %lx\n",
412 bp->b_vp, bp->b_lblkno, bp->b_flags);
413 }
414 }
415 }
416 #endif
417