2 * Copyright (c) 1990, 1993, 1994
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
34 #if defined(LIBC_SCCS) && !defined(lint)
35 static char sccsid
[] = "@(#)mpool.c 8.5 (Berkeley) 7/26/94";
36 #endif /* LIBC_SCCS and not lint */
38 #include <sys/param.h>
39 #include <sys/queue.h>
50 #define __MPOOLINTERFACE_PRIVATE
53 static BKT
*mpool_bkt
__P((MPOOL
*));
54 static BKT
*mpool_look
__P((MPOOL
*, pgno_t
));
55 static int mpool_write
__P((MPOOL
*, BKT
*));
59 * Initialize a memory pool.
62 mpool_open(key
, fd
, pagesize
, maxcache
)
65 pgno_t pagesize
, maxcache
;
72 * Get information about the file.
75 * We don't currently handle pipes, although we should.
79 if (!S_ISREG(sb
.st_mode
)) {
84 /* Allocate and initialize the MPOOL cookie. */
85 if ((mp
= (MPOOL
*)calloc(1, sizeof(MPOOL
))) == NULL
)
87 CIRCLEQ_INIT(&mp
->lqh
);
88 for (entry
= 0; entry
< HASHSIZE
; ++entry
)
89 CIRCLEQ_INIT(&mp
->hqh
[entry
]);
90 mp
->maxcache
= maxcache
;
91 mp
->npages
= sb
.st_size
/ pagesize
;
92 mp
->pagesize
= pagesize
;
99 * Initialize input/output filters.
102 mpool_filter(mp
, pgin
, pgout
, pgcookie
)
104 void (*pgin
) __P((void *, pgno_t
, void *));
105 void (*pgout
) __P((void *, pgno_t
, void *));
110 mp
->pgcookie
= pgcookie
;
115 * Get a new page of memory.
118 mpool_new(mp
, pgnoaddr
)
125 if (mp
->npages
== MAX_PAGE_NUMBER
) {
126 (void)fprintf(stderr
, "mpool_new: page allocation overflow.\n");
133 * Get a BKT from the cache. Assign a new page number, attach
134 * it to the head of the hash chain, the tail of the lru chain,
137 if ((bp
= mpool_bkt(mp
)) == NULL
)
139 *pgnoaddr
= bp
->pgno
= mp
->npages
++;
140 bp
->flags
= MPOOL_PINNED
;
142 head
= &mp
->hqh
[HASHKEY(bp
->pgno
)];
143 CIRCLEQ_INSERT_HEAD(head
, bp
, hq
);
144 CIRCLEQ_INSERT_TAIL(&mp
->lqh
, bp
, q
);
153 mpool_get(mp
, pgno
, flags
)
156 u_int flags
; /* XXX not used? */
163 /* Check for attempt to retrieve a non-existent page. */
164 if (pgno
>= mp
->npages
) {
173 /* Check for a page that is cached. */
174 if ((bp
= mpool_look(mp
, pgno
)) != NULL
) {
176 if (bp
->flags
& MPOOL_PINNED
) {
177 (void)fprintf(stderr
,
178 "mpool_get: page %d already pinned\n", bp
->pgno
);
183 * Move the page to the head of the hash chain and the tail
186 head
= &mp
->hqh
[HASHKEY(bp
->pgno
)];
187 CIRCLEQ_REMOVE(head
, bp
, hq
);
188 CIRCLEQ_INSERT_HEAD(head
, bp
, hq
);
189 CIRCLEQ_REMOVE(&mp
->lqh
, bp
, q
);
190 CIRCLEQ_INSERT_TAIL(&mp
->lqh
, bp
, q
);
192 /* Return a pinned page. */
193 bp
->flags
|= MPOOL_PINNED
;
197 /* Get a page from the cache. */
198 if ((bp
= mpool_bkt(mp
)) == NULL
)
201 /* Read in the contents. */
205 off
= mp
->pagesize
* pgno
;
206 if (lseek(mp
->fd
, off
, SEEK_SET
) != off
)
208 if ((nr
= read(mp
->fd
, bp
->page
, mp
->pagesize
)) != mp
->pagesize
) {
214 /* Set the page number, pin the page. */
216 bp
->flags
= MPOOL_PINNED
;
219 * Add the page to the head of the hash chain and the tail
222 head
= &mp
->hqh
[HASHKEY(bp
->pgno
)];
223 CIRCLEQ_INSERT_HEAD(head
, bp
, hq
);
224 CIRCLEQ_INSERT_TAIL(&mp
->lqh
, bp
, q
);
226 /* Run through the user's filter. */
227 if (mp
->pgin
!= NULL
)
228 (mp
->pgin
)(mp
->pgcookie
, bp
->pgno
, bp
->page
);
238 mpool_put(mp
, page
, flags
)
248 bp
= (BKT
*)((char *)page
- sizeof(BKT
));
250 if (!(bp
->flags
& MPOOL_PINNED
)) {
251 (void)fprintf(stderr
,
252 "mpool_put: page %d not pinned\n", bp
->pgno
);
256 bp
->flags
&= ~MPOOL_PINNED
;
257 bp
->flags
|= flags
& MPOOL_DIRTY
;
258 return (RET_SUCCESS
);
263 * Close the buffer pool.
271 /* Free up any space allocated to the lru pages. */
272 while ((bp
= mp
->lqh
.cqh_first
) != (void *)&mp
->lqh
) {
273 CIRCLEQ_REMOVE(&mp
->lqh
, mp
->lqh
.cqh_first
, q
);
277 /* Free the MPOOL cookie. */
279 return (RET_SUCCESS
);
284 * Sync the pool to disk.
292 /* Walk the lru chain, flushing any dirty pages to disk. */
293 for (bp
= mp
->lqh
.cqh_first
;
294 bp
!= (void *)&mp
->lqh
; bp
= bp
->q
.cqe_next
)
295 if (bp
->flags
& MPOOL_DIRTY
&&
296 mpool_write(mp
, bp
) == RET_ERROR
)
299 /* Sync the file descriptor. */
300 return (fsync(mp
->fd
) ? RET_ERROR
: RET_SUCCESS
);
305 * Get a page from the cache (or create one).
314 /* If under the max cached, always create a new page. */
315 if (mp
->curcache
< mp
->maxcache
)
319 * If the cache is max'd out, walk the lru list for a buffer we
320 * can flush. If we find one, write it (if necessary) and take it
321 * off any lists. If we don't find anything we grow the cache anyway.
322 * The cache never shrinks.
324 for (bp
= mp
->lqh
.cqh_first
;
325 bp
!= (void *)&mp
->lqh
; bp
= bp
->q
.cqe_next
)
326 if (!(bp
->flags
& MPOOL_PINNED
)) {
327 /* Flush if dirty. */
328 if (bp
->flags
& MPOOL_DIRTY
&&
329 mpool_write(mp
, bp
) == RET_ERROR
)
334 /* Remove from the hash and lru queues. */
335 head
= &mp
->hqh
[HASHKEY(bp
->pgno
)];
336 CIRCLEQ_REMOVE(head
, bp
, hq
);
337 CIRCLEQ_REMOVE(&mp
->lqh
, bp
, q
);
341 memset(bp
, 0xff, sizeof(BKT
) + mp
->pagesize
);
348 new: if ((bp
= (BKT
*)malloc(sizeof(BKT
) + mp
->pagesize
)) == NULL
)
353 #if defined(DEBUG) || defined(PURIFY)
354 memset(bp
, 0xff, sizeof(BKT
) + mp
->pagesize
);
356 bp
->page
= (char *)bp
+ sizeof(BKT
);
363 * Write a page to disk.
376 /* Run through the user's filter. */
378 (mp
->pgout
)(mp
->pgcookie
, bp
->pgno
, bp
->page
);
380 off
= mp
->pagesize
* bp
->pgno
;
381 if (lseek(mp
->fd
, off
, SEEK_SET
) != off
)
383 if (write(mp
->fd
, bp
->page
, mp
->pagesize
) != mp
->pagesize
)
386 bp
->flags
&= ~MPOOL_DIRTY
;
387 return (RET_SUCCESS
);
392 * Lookup a page in the cache.
402 head
= &mp
->hqh
[HASHKEY(pgno
)];
403 for (bp
= head
->cqh_first
; bp
!= (void *)head
; bp
= bp
->hq
.cqe_next
)
404 if (bp
->pgno
== pgno
) {
419 * Print out cache statistics.
429 (void)fprintf(stderr
, "%lu pages in the file\n", mp
->npages
);
430 (void)fprintf(stderr
,
431 "page size %lu, cacheing %lu pages of %lu page max cache\n",
432 mp
->pagesize
, mp
->curcache
, mp
->maxcache
);
433 (void)fprintf(stderr
, "%lu page puts, %lu page gets, %lu page new\n",
434 mp
->pageput
, mp
->pageget
, mp
->pagenew
);
435 (void)fprintf(stderr
, "%lu page allocs, %lu page flushes\n",
436 mp
->pagealloc
, mp
->pageflush
);
437 if (mp
->cachehit
+ mp
->cachemiss
)
438 (void)fprintf(stderr
,
439 "%.0f%% cache hit rate (%lu hits, %lu misses)\n",
440 ((double)mp
->cachehit
/ (mp
->cachehit
+ mp
->cachemiss
))
441 * 100, mp
->cachehit
, mp
->cachemiss
);
442 (void)fprintf(stderr
, "%lu page reads, %lu page writes\n",
443 mp
->pageread
, mp
->pagewrite
);
447 for (bp
= mp
->lqh
.cqh_first
;
448 bp
!= (void *)&mp
->lqh
; bp
= bp
->q
.cqe_next
) {
449 (void)fprintf(stderr
, "%s%d", sep
, bp
->pgno
);
450 if (bp
->flags
& MPOOL_DIRTY
)
451 (void)fprintf(stderr
, "d");
452 if (bp
->flags
& MPOOL_PINNED
)
453 (void)fprintf(stderr
, "P");
461 (void)fprintf(stderr
, "\n");