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
33 * $FreeBSD: src/lib/libc/db/mpool/mpool.c,v 1.5.2.1 2001/03/05 23:05:01 obrien Exp $
36 #if defined(LIBC_SCCS) && !defined(lint)
37 static char sccsid[] = "@(#)mpool.c 8.5 (Berkeley) 7/26/94";
38 #endif /* LIBC_SCCS and not lint */
40 #include <sys/param.h>
41 #include <sys/queue.h>
52 #define __MPOOLINTERFACE_PRIVATE
55 static BKT *mpool_bkt __P((MPOOL *));
56 static BKT *mpool_look __P((MPOOL *, pgno_t));
57 static int mpool_write __P((MPOOL *, BKT *));
61 * Initialize a memory pool.
64 mpool_open(key, fd, pagesize, maxcache)
67 pgno_t pagesize, maxcache;
74 * Get information about the file.
77 * We don't currently handle pipes, although we should.
81 if (!S_ISREG(sb.st_mode)) {
86 /* Allocate and initialize the MPOOL cookie. */
87 if ((mp = (MPOOL *)calloc(1, sizeof(MPOOL))) == NULL)
90 for (entry = 0; entry < HASHSIZE; ++entry)
91 TAILQ_INIT(&mp->hqh[entry]);
92 mp->maxcache = maxcache;
93 mp->npages = sb.st_size / pagesize;
94 mp->pagesize = pagesize;
101 * Initialize input/output filters.
104 mpool_filter(mp, pgin, pgout, pgcookie)
106 void (*pgin) __P((void *, pgno_t, void *));
107 void (*pgout) __P((void *, pgno_t, void *));
112 mp->pgcookie = pgcookie;
117 * Get a new page of memory.
120 mpool_new(mp, pgnoaddr)
127 if (mp->npages == MAX_PAGE_NUMBER) {
128 (void)fprintf(stderr, "mpool_new: page allocation overflow.\n");
135 * Get a BKT from the cache. Assign a new page number, attach
136 * it to the head of the hash chain, the tail of the lru chain,
139 if ((bp = mpool_bkt(mp)) == NULL)
141 *pgnoaddr = bp->pgno = mp->npages++;
142 bp->flags = MPOOL_PINNED;
144 head = &mp->hqh[HASHKEY(bp->pgno)];
145 TAILQ_INSERT_HEAD(head, bp, hq);
146 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
155 mpool_get(mp, pgno, flags)
158 u_int flags; /* XXX not used? */
165 /* Check for attempt to retrieve a non-existent page. */
166 if (pgno >= mp->npages) {
175 /* Check for a page that is cached. */
176 if ((bp = mpool_look(mp, pgno)) != NULL) {
178 if (bp->flags & MPOOL_PINNED) {
179 (void)fprintf(stderr,
180 "mpool_get: page %d already pinned\n", bp->pgno);
185 * Move the page to the head of the hash chain and the tail
188 head = &mp->hqh[HASHKEY(bp->pgno)];
189 TAILQ_REMOVE(head, bp, hq);
190 TAILQ_INSERT_HEAD(head, bp, hq);
191 TAILQ_REMOVE(&mp->lqh, bp, q);
192 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
194 /* Return a pinned page. */
195 bp->flags |= MPOOL_PINNED;
199 /* Get a page from the cache. */
200 if ((bp = mpool_bkt(mp)) == NULL)
203 /* Read in the contents. */
207 off = mp->pagesize * pgno;
208 if (lseek(mp->fd, off, SEEK_SET) != off)
210 if ((nr = _read(mp->fd, bp->page, mp->pagesize)) != mp->pagesize) {
216 /* Set the page number, pin the page. */
218 bp->flags = MPOOL_PINNED;
221 * Add the page to the head of the hash chain and the tail
224 head = &mp->hqh[HASHKEY(bp->pgno)];
225 TAILQ_INSERT_HEAD(head, bp, hq);
226 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
228 /* Run through the user's filter. */
229 if (mp->pgin != NULL)
230 (mp->pgin)(mp->pgcookie, bp->pgno, bp->page);
240 mpool_put(mp, page, flags)
250 bp = (BKT *)((char *)page - sizeof(BKT));
252 if (!(bp->flags & MPOOL_PINNED)) {
253 (void)fprintf(stderr,
254 "mpool_put: page %d not pinned\n", bp->pgno);
258 bp->flags &= ~MPOOL_PINNED;
259 bp->flags |= flags & MPOOL_DIRTY;
260 return (RET_SUCCESS);
265 * Close the buffer pool.
273 /* Free up any space allocated to the lru pages. */
274 while (!TAILQ_EMPTY(&mp->lqh)) {
275 bp = TAILQ_FIRST(&mp->lqh);
276 TAILQ_REMOVE(&mp->lqh, bp, q);
280 /* Free the MPOOL cookie. */
282 return (RET_SUCCESS);
287 * Sync the pool to disk.
295 /* Walk the lru chain, flushing any dirty pages to disk. */
296 TAILQ_FOREACH(bp, &mp->lqh, q)
297 if (bp->flags & MPOOL_DIRTY &&
298 mpool_write(mp, bp) == RET_ERROR)
301 /* Sync the file descriptor. */
302 return (_fsync(mp->fd) ? RET_ERROR : RET_SUCCESS);
307 * Get a page from the cache (or create one).
316 /* If under the max cached, always create a new page. */
317 if (mp->curcache < mp->maxcache)
321 * If the cache is max'd out, walk the lru list for a buffer we
322 * can flush. If we find one, write it (if necessary) and take it
323 * off any lists. If we don't find anything we grow the cache anyway.
324 * The cache never shrinks.
326 TAILQ_FOREACH(bp, &mp->lqh, q)
327 if (!(bp->flags & MPOOL_PINNED)) {
328 /* Flush if dirty. */
329 if (bp->flags & MPOOL_DIRTY &&
330 mpool_write(mp, bp) == RET_ERROR)
335 /* Remove from the hash and lru queues. */
336 head = &mp->hqh[HASHKEY(bp->pgno)];
337 TAILQ_REMOVE(head, bp, hq);
338 TAILQ_REMOVE(&mp->lqh, bp, q);
342 memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
349 new: if ((bp = (BKT *)malloc(sizeof(BKT) + mp->pagesize)) == NULL)
354 #if defined(DEBUG) || defined(PURIFY)
355 memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
357 bp->page = (char *)bp + sizeof(BKT);
364 * Write a page to disk.
377 /* Run through the user's filter. */
379 (mp->pgout)(mp->pgcookie, bp->pgno, bp->page);
381 off = mp->pagesize * bp->pgno;
382 if (lseek(mp->fd, off, SEEK_SET) != off)
384 if (_write(mp->fd, bp->page, mp->pagesize) != mp->pagesize)
387 bp->flags &= ~MPOOL_DIRTY;
388 return (RET_SUCCESS);
393 * Lookup a page in the cache.
403 head = &mp->hqh[HASHKEY(pgno)];
404 TAILQ_FOREACH(bp, head, hq)
405 if (bp->pgno == pgno) {
420 * Print out cache statistics.
430 (void)fprintf(stderr, "%lu pages in the file\n", mp->npages);
431 (void)fprintf(stderr,
432 "page size %lu, cacheing %lu pages of %lu page max cache\n",
433 mp->pagesize, mp->curcache, mp->maxcache);
434 (void)fprintf(stderr, "%lu page puts, %lu page gets, %lu page new\n",
435 mp->pageput, mp->pageget, mp->pagenew);
436 (void)fprintf(stderr, "%lu page allocs, %lu page flushes\n",
437 mp->pagealloc, mp->pageflush);
438 if (mp->cachehit + mp->cachemiss)
439 (void)fprintf(stderr,
440 "%.0f%% cache hit rate (%lu hits, %lu misses)\n",
441 ((double)mp->cachehit / (mp->cachehit + mp->cachemiss))
442 * 100, mp->cachehit, mp->cachemiss);
443 (void)fprintf(stderr, "%lu page reads, %lu page writes\n",
444 mp->pageread, mp->pagewrite);
448 TAILQ_FOREACH(bp, &mp->lqh, q) {
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");