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 $
34 * $DragonFly: src/lib/libc/db/mpool/mpool.c,v 1.4 2005/01/31 22:29:11 dillon Exp $
36 * @(#)mpool.c 8.5 (Berkeley) 7/26/94
39 #include "namespace.h"
40 #include <sys/param.h>
41 #include <sys/queue.h>
49 #include "un-namespace.h"
53 #define __MPOOLINTERFACE_PRIVATE
56 static BKT *mpool_bkt (MPOOL *);
57 static BKT *mpool_look (MPOOL *, pgno_t);
58 static int mpool_write (MPOOL *, BKT *);
62 * Initialize a memory pool.
65 mpool_open(key, fd, pagesize, maxcache)
68 pgno_t pagesize, maxcache;
75 * Get information about the file.
78 * We don't currently handle pipes, although we should.
82 if (!S_ISREG(sb.st_mode)) {
87 /* Allocate and initialize the MPOOL cookie. */
88 if ((mp = (MPOOL *)calloc(1, sizeof(MPOOL))) == NULL)
91 for (entry = 0; entry < HASHSIZE; ++entry)
92 TAILQ_INIT(&mp->hqh[entry]);
93 mp->maxcache = maxcache;
94 mp->npages = sb.st_size / pagesize;
95 mp->pagesize = pagesize;
102 * Initialize input/output filters.
105 mpool_filter(mp, pgin, pgout, pgcookie)
107 void (*pgin) (void *, pgno_t, void *);
108 void (*pgout) (void *, pgno_t, void *);
113 mp->pgcookie = pgcookie;
118 * Get a new page of memory.
121 mpool_new(mp, pgnoaddr)
128 if (mp->npages == MAX_PAGE_NUMBER) {
129 (void)fprintf(stderr, "mpool_new: page allocation overflow.\n");
136 * Get a BKT from the cache. Assign a new page number, attach
137 * it to the head of the hash chain, the tail of the lru chain,
140 if ((bp = mpool_bkt(mp)) == NULL)
142 *pgnoaddr = bp->pgno = mp->npages++;
143 bp->flags = MPOOL_PINNED;
145 head = &mp->hqh[HASHKEY(bp->pgno)];
146 TAILQ_INSERT_HEAD(head, bp, hq);
147 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
156 mpool_get(mp, pgno, flags)
159 u_int flags; /* XXX not used? */
166 /* Check for attempt to retrieve a non-existent page. */
167 if (pgno >= mp->npages) {
176 /* Check for a page that is cached. */
177 if ((bp = mpool_look(mp, pgno)) != NULL) {
179 if (bp->flags & MPOOL_PINNED) {
180 (void)fprintf(stderr,
181 "mpool_get: page %d already pinned\n", bp->pgno);
186 * Move the page to the head of the hash chain and the tail
189 head = &mp->hqh[HASHKEY(bp->pgno)];
190 TAILQ_REMOVE(head, bp, hq);
191 TAILQ_INSERT_HEAD(head, bp, hq);
192 TAILQ_REMOVE(&mp->lqh, bp, q);
193 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
195 /* Return a pinned page. */
196 bp->flags |= MPOOL_PINNED;
200 /* Get a page from the cache. */
201 if ((bp = mpool_bkt(mp)) == NULL)
204 /* Read in the contents. */
208 off = mp->pagesize * pgno;
209 if (lseek(mp->fd, off, SEEK_SET) != off)
211 if ((nr = _read(mp->fd, bp->page, mp->pagesize)) != mp->pagesize) {
217 /* Set the page number, pin the page. */
219 bp->flags = MPOOL_PINNED;
222 * Add the page to the head of the hash chain and the tail
225 head = &mp->hqh[HASHKEY(bp->pgno)];
226 TAILQ_INSERT_HEAD(head, bp, hq);
227 TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
229 /* Run through the user's filter. */
230 if (mp->pgin != NULL)
231 (mp->pgin)(mp->pgcookie, bp->pgno, bp->page);
241 mpool_put(mp, page, flags)
251 bp = (BKT *)((char *)page - sizeof(BKT));
253 if (!(bp->flags & MPOOL_PINNED)) {
254 (void)fprintf(stderr,
255 "mpool_put: page %d not pinned\n", bp->pgno);
259 bp->flags &= ~MPOOL_PINNED;
260 bp->flags |= flags & MPOOL_DIRTY;
261 return (RET_SUCCESS);
266 * Close the buffer pool.
274 /* Free up any space allocated to the lru pages. */
275 while (!TAILQ_EMPTY(&mp->lqh)) {
276 bp = TAILQ_FIRST(&mp->lqh);
277 TAILQ_REMOVE(&mp->lqh, bp, q);
281 /* Free the MPOOL cookie. */
283 return (RET_SUCCESS);
288 * Sync the pool to disk.
296 /* Walk the lru chain, flushing any dirty pages to disk. */
297 TAILQ_FOREACH(bp, &mp->lqh, q)
298 if (bp->flags & MPOOL_DIRTY &&
299 mpool_write(mp, bp) == RET_ERROR)
302 /* Sync the file descriptor. */
303 return (_fsync(mp->fd) ? RET_ERROR : RET_SUCCESS);
308 * Get a page from the cache (or create one).
317 /* If under the max cached, always create a new page. */
318 if (mp->curcache < mp->maxcache)
322 * If the cache is max'd out, walk the lru list for a buffer we
323 * can flush. If we find one, write it (if necessary) and take it
324 * off any lists. If we don't find anything we grow the cache anyway.
325 * The cache never shrinks.
327 TAILQ_FOREACH(bp, &mp->lqh, q)
328 if (!(bp->flags & MPOOL_PINNED)) {
329 /* Flush if dirty. */
330 if (bp->flags & MPOOL_DIRTY &&
331 mpool_write(mp, bp) == RET_ERROR)
336 /* Remove from the hash and lru queues. */
337 head = &mp->hqh[HASHKEY(bp->pgno)];
338 TAILQ_REMOVE(head, bp, hq);
339 TAILQ_REMOVE(&mp->lqh, bp, q);
343 memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
350 new: if ((bp = (BKT *)malloc(sizeof(BKT) + mp->pagesize)) == NULL)
355 #if defined(DEBUG) || defined(PURIFY)
356 memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
358 bp->page = (char *)bp + sizeof(BKT);
365 * Write a page to disk.
378 /* Run through the user's filter. */
380 (mp->pgout)(mp->pgcookie, bp->pgno, bp->page);
382 off = mp->pagesize * bp->pgno;
383 if (lseek(mp->fd, off, SEEK_SET) != off)
385 if (_write(mp->fd, bp->page, mp->pagesize) != mp->pagesize)
388 bp->flags &= ~MPOOL_DIRTY;
389 return (RET_SUCCESS);
394 * Lookup a page in the cache.
404 head = &mp->hqh[HASHKEY(pgno)];
405 TAILQ_FOREACH(bp, head, hq)
406 if (bp->pgno == pgno) {
421 * Print out cache statistics.
431 (void)fprintf(stderr, "%lu pages in the file\n", mp->npages);
432 (void)fprintf(stderr,
433 "page size %lu, cacheing %lu pages of %lu page max cache\n",
434 mp->pagesize, mp->curcache, mp->maxcache);
435 (void)fprintf(stderr, "%lu page puts, %lu page gets, %lu page new\n",
436 mp->pageput, mp->pageget, mp->pagenew);
437 (void)fprintf(stderr, "%lu page allocs, %lu page flushes\n",
438 mp->pagealloc, mp->pageflush);
439 if (mp->cachehit + mp->cachemiss)
440 (void)fprintf(stderr,
441 "%.0f%% cache hit rate (%lu hits, %lu misses)\n",
442 ((double)mp->cachehit / (mp->cachehit + mp->cachemiss))
443 * 100, mp->cachehit, mp->cachemiss);
444 (void)fprintf(stderr, "%lu page reads, %lu page writes\n",
445 mp->pageread, mp->pagewrite);
449 TAILQ_FOREACH(bp, &mp->lqh, q) {
450 (void)fprintf(stderr, "%s%d", sep, bp->pgno);
451 if (bp->flags & MPOOL_DIRTY)
452 (void)fprintf(stderr, "d");
453 if (bp->flags & MPOOL_PINNED)
454 (void)fprintf(stderr, "P");
462 (void)fprintf(stderr, "\n");