[apple/libc.git] / db / mpool / mpool.c

/*
 * Copyright (c) 2003 Apple Computer, Inc. All rights reserved.
 *
 * @APPLE_LICENSE_HEADER_START@
 * 
 * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
 * 
 * This file contains Original Code and/or Modifications of Original Code
 * as defined in and that are subject to the Apple Public Source License
 * Version 2.0 (the 'License'). You may not use this file except in
 * compliance with the License. Please obtain a copy of the License at
 * http://www.opensource.apple.com/apsl/ and read it before using this
 * file.
 * 
 * The Original Code and all software distributed under the License are
 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
 * Please see the License for the specific language governing rights and
 * limitations under the License.
 * 
 * @APPLE_LICENSE_HEADER_END@
 */
/*-
 * Copyright (c) 1990, 1993, 1994
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#if defined(LIBC_SCCS) && !defined(lint)
static char sccsid[] = "@(#)mpool.c	8.5 (Berkeley) 7/26/94";
#endif /* LIBC_SCCS and not lint */
#include <sys/cdefs.h>

#include <sys/param.h>
#include <sys/queue.h>
#include <sys/stat.h>

#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <db.h>

#define	__MPOOLINTERFACE_PRIVATE
#include <mpool.h>

static BKT *mpool_bkt(MPOOL *);
static BKT *mpool_look(MPOOL *, pgno_t);
static int  mpool_write(MPOOL *, BKT *);

/*
 * mpool_open --
 *	Initialize a memory pool.
 */
MPOOL *
mpool_open(key, fd, pagesize, maxcache)
	void *key;
	int fd;
	pgno_t pagesize, maxcache;
{
	struct stat sb;
	MPOOL *mp;
	int entry;

	/*
	 * Get information about the file.
	 *
	 * XXX
	 * We don't currently handle pipes, although we should.
	 */
	if (fstat(fd, &sb))
		return (NULL);
	if (!S_ISREG(sb.st_mode)) {
		errno = ESPIPE;
		return (NULL);
	}

	/* Allocate and initialize the MPOOL cookie. */
	if ((mp = (MPOOL *)calloc(1, sizeof(MPOOL))) == NULL)
		return (NULL);
	TAILQ_INIT(&mp->lqh);
	for (entry = 0; entry < HASHSIZE; ++entry)
		TAILQ_INIT(&mp->hqh[entry]);
	mp->maxcache = maxcache;
	mp->npages = sb.st_size / pagesize;
	mp->pagesize = pagesize;
	mp->fd = fd;
	return (mp);
}

/*
 * mpool_filter --
 *	Initialize input/output filters.
 */
void
mpool_filter(mp, pgin, pgout, pgcookie)
	MPOOL *mp;
	void (*pgin)(void *, pgno_t, void *);
	void (*pgout)(void *, pgno_t, void *);
	void *pgcookie;
{
	mp->pgin = pgin;
	mp->pgout = pgout;
	mp->pgcookie = pgcookie;
}
	
/*
 * mpool_new --
 *	Get a new page of memory.
 */
void *
mpool_new(mp, pgnoaddr)
	MPOOL *mp;
	pgno_t *pgnoaddr;
{
	struct _hqh *head;
	BKT *bp;

	if (mp->npages == MAX_PAGE_NUMBER) {
		(void)fprintf(stderr, "mpool_new: page allocation overflow.\n");
		abort();
	}
#ifdef STATISTICS
	++mp->pagenew;
#endif
	/*
	 * Get a BKT from the cache.  Assign a new page number, attach
	 * it to the head of the hash chain, the tail of the lru chain,
	 * and return.
	 */
	if ((bp = mpool_bkt(mp)) == NULL)
		return (NULL);
	*pgnoaddr = bp->pgno = mp->npages++;
	bp->flags = MPOOL_PINNED;

	head = &mp->hqh[HASHKEY(bp->pgno)];
	TAILQ_INSERT_HEAD(head, bp, hq);
	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	return (bp->page);
}

/*
 * mpool_get
 *	Get a page.
 */
void *
mpool_get(mp, pgno, flags)
	MPOOL *mp;
	pgno_t pgno;
	u_int flags;				/* XXX not used? */
{
	struct _hqh *head;
	BKT *bp;
	off_t off;
	int nr;

	/* Check for attempt to retrieve a non-existent page. */
	if (pgno >= mp->npages) {
		errno = EINVAL;
		return (NULL);
	}

#ifdef STATISTICS
	++mp->pageget;
#endif

	/* Check for a page that is cached. */
	if ((bp = mpool_look(mp, pgno)) != NULL) {
#ifdef DEBUG
		if (bp->flags & MPOOL_PINNED) {
			(void)fprintf(stderr,
			    "mpool_get: page %d already pinned\n", bp->pgno);
			abort();
		}
#endif
		/*
		 * Move the page to the head of the hash chain and the tail
		 * of the lru chain.
		 */
		head = &mp->hqh[HASHKEY(bp->pgno)];
		TAILQ_REMOVE(head, bp, hq);
		TAILQ_INSERT_HEAD(head, bp, hq);
		TAILQ_REMOVE(&mp->lqh, bp, q);
		TAILQ_INSERT_TAIL(&mp->lqh, bp, q);

		/* Return a pinned page. */
		bp->flags |= MPOOL_PINNED;
		return (bp->page);
	}

	/* Get a page from the cache. */
	if ((bp = mpool_bkt(mp)) == NULL)
		return (NULL);

	/* Read in the contents. */
#ifdef STATISTICS
	++mp->pageread;
#endif
	off = mp->pagesize * pgno;
	if (lseek(mp->fd, off, SEEK_SET) != off)
		return (NULL);
	if ((nr = read(mp->fd, bp->page, mp->pagesize)) != mp->pagesize) {
		if (nr >= 0)
			errno = EFTYPE;
		return (NULL);
	}

	/* Set the page number, pin the page. */
	bp->pgno = pgno;
	bp->flags = MPOOL_PINNED;

	/*
	 * Add the page to the head of the hash chain and the tail
	 * of the lru chain.
	 */
	head = &mp->hqh[HASHKEY(bp->pgno)];
	TAILQ_INSERT_HEAD(head, bp, hq);
	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);

	/* Run through the user's filter. */
	if (mp->pgin != NULL)
		(mp->pgin)(mp->pgcookie, bp->pgno, bp->page);

	return (bp->page);
}

/*
 * mpool_put
 *	Return a page.
 */
int
mpool_put(mp, page, flags)
	MPOOL *mp;
	void *page;
	u_int flags;
{
	BKT *bp;

#ifdef STATISTICS
	++mp->pageput;
#endif
	bp = (BKT *)((char *)page - sizeof(BKT));
#ifdef DEBUG
	if (!(bp->flags & MPOOL_PINNED)) {
		(void)fprintf(stderr,
		    "mpool_put: page %d not pinned\n", bp->pgno);
		abort();
	}
#endif
	bp->flags &= ~MPOOL_PINNED;
	bp->flags |= flags & MPOOL_DIRTY;
	return (RET_SUCCESS);
}

/*
 * mpool_close
 *	Close the buffer pool.
 */
int
mpool_close(mp)
	MPOOL *mp;
{
	BKT *bp;

	/* Free up any space allocated to the lru pages. */
	while (!TAILQ_EMPTY(&mp->lqh)) {
		bp = TAILQ_FIRST(&mp->lqh);
		TAILQ_REMOVE(&mp->lqh, bp, q);
		free(bp);
	}

	/* Free the MPOOL cookie. */
	free(mp);
	return (RET_SUCCESS);
}

/*
 * mpool_sync
 *	Sync the pool to disk.
 */
int
mpool_sync(mp)
	MPOOL *mp;
{
	BKT *bp;

	/* Walk the lru chain, flushing any dirty pages to disk. */
	TAILQ_FOREACH(bp, &mp->lqh, q)
		if (bp->flags & MPOOL_DIRTY &&
		    mpool_write(mp, bp) == RET_ERROR)
			return (RET_ERROR);

	/* Sync the file descriptor. */
	return (fsync(mp->fd) ? RET_ERROR : RET_SUCCESS);
}

/*
 * mpool_bkt
 *	Get a page from the cache (or create one).
 */
static BKT *
mpool_bkt(mp)
	MPOOL *mp;
{
	struct _hqh *head;
	BKT *bp;

	/* If under the max cached, always create a new page. */
	if (mp->curcache < mp->maxcache)
		goto new;

	/*
	 * If the cache is max'd out, walk the lru list for a buffer we
	 * can flush.  If we find one, write it (if necessary) and take it
	 * off any lists.  If we don't find anything we grow the cache anyway.
	 * The cache never shrinks.
	 */
	TAILQ_FOREACH(bp, &mp->lqh, q)
		if (!(bp->flags & MPOOL_PINNED)) {
			/* Flush if dirty. */
			if (bp->flags & MPOOL_DIRTY &&
			    mpool_write(mp, bp) == RET_ERROR)
				return (NULL);
#ifdef STATISTICS
			++mp->pageflush;
#endif
			/* Remove from the hash and lru queues. */
			head = &mp->hqh[HASHKEY(bp->pgno)];
			TAILQ_REMOVE(head, bp, hq);
			TAILQ_REMOVE(&mp->lqh, bp, q);
#ifdef DEBUG
			{ void *spage;
				spage = bp->page;
				memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
				bp->page = spage;
			}
#endif
			return (bp);
		}

new:	if ((bp = (BKT *)malloc(sizeof(BKT) + mp->pagesize)) == NULL)
		return (NULL);
#ifdef STATISTICS
	++mp->pagealloc;
#endif
#if defined(DEBUG) || defined(PURIFY)
	memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
#endif
	bp->page = (char *)bp + sizeof(BKT);
	++mp->curcache;
	return (bp);
}

/*
 * mpool_write
 *	Write a page to disk.
 */
static int
mpool_write(mp, bp)
	MPOOL *mp;
	BKT *bp;
{
	off_t off;

#ifdef STATISTICS
	++mp->pagewrite;
#endif

	/* Run through the user's filter. */
	if (mp->pgout)
		(mp->pgout)(mp->pgcookie, bp->pgno, bp->page);

	off = mp->pagesize * bp->pgno;
	if (lseek(mp->fd, off, SEEK_SET) != off)
		return (RET_ERROR);
	if (write(mp->fd, bp->page, mp->pagesize) != mp->pagesize)
		return (RET_ERROR);

	bp->flags &= ~MPOOL_DIRTY;
	return (RET_SUCCESS);
}

/*
 * mpool_look
 *	Lookup a page in the cache.
 */
static BKT *
mpool_look(mp, pgno)
	MPOOL *mp;
	pgno_t pgno;
{
	struct _hqh *head;
	BKT *bp;

	head = &mp->hqh[HASHKEY(pgno)];
	TAILQ_FOREACH(bp, head, hq)
		if (bp->pgno == pgno) {
#ifdef STATISTICS
			++mp->cachehit;
#endif
			return (bp);
		}
#ifdef STATISTICS
	++mp->cachemiss;
#endif
	return (NULL);
}

#ifdef STATISTICS
/*
 * mpool_stat
 *	Print out cache statistics.
 */
void
mpool_stat(mp)
	MPOOL *mp;
{
	BKT *bp;
	int cnt;
	char *sep;

	(void)fprintf(stderr, "%lu pages in the file\n", mp->npages);
	(void)fprintf(stderr,
	    "page size %lu, cacheing %lu pages of %lu page max cache\n",
	    mp->pagesize, mp->curcache, mp->maxcache);
	(void)fprintf(stderr, "%lu page puts, %lu page gets, %lu page new\n",
	    mp->pageput, mp->pageget, mp->pagenew);
	(void)fprintf(stderr, "%lu page allocs, %lu page flushes\n",
	    mp->pagealloc, mp->pageflush);
	if (mp->cachehit + mp->cachemiss)
		(void)fprintf(stderr,
		    "%.0f%% cache hit rate (%lu hits, %lu misses)\n", 
		    ((double)mp->cachehit / (mp->cachehit + mp->cachemiss))
		    * 100, mp->cachehit, mp->cachemiss);
	(void)fprintf(stderr, "%lu page reads, %lu page writes\n",
	    mp->pageread, mp->pagewrite);

	sep = "";
	cnt = 0;
	TAILQ_FOREACH(bp, &mp->lqh, q) {
		(void)fprintf(stderr, "%s%d", sep, bp->pgno);
		if (bp->flags & MPOOL_DIRTY)
			(void)fprintf(stderr, "d");
		if (bp->flags & MPOOL_PINNED)
			(void)fprintf(stderr, "P");
		if (++cnt == 10) {
			sep = "\n";
			cnt = 0;
		} else
			sep = ", ";
			
	}
	(void)fprintf(stderr, "\n");
}
#endif
Commit	Line	Data
e9ce8d39	1	/*
9385eb3d	2	* Copyright (c) 2003 Apple Computer, Inc. All rights reserved.
e9ce8d39 A	3	*
	4	* @APPLE_LICENSE_HEADER_START@
	5	*
734aad71	6	* Copyright (c) 1999-2003 Apple Computer, Inc. All Rights Reserved.
e9ce8d39	7	*
734aad71 A	8	* This file contains Original Code and/or Modifications of Original Code
	9	* as defined in and that are subject to the Apple Public Source License
	10	* Version 2.0 (the 'License'). You may not use this file except in
	11	* compliance with the License. Please obtain a copy of the License at
	12	* http://www.opensource.apple.com/apsl/ and read it before using this
	13	* file.
	14	*
	15	* The Original Code and all software distributed under the License are
	16	* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
e9ce8d39 A	17	* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
e9ce8d39 A	18	* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
734aad71 A	19	* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
	20	* Please see the License for the specific language governing rights and
	21	* limitations under the License.
e9ce8d39 A	22	*
	23	* @APPLE_LICENSE_HEADER_END@
	24	*/
9385eb3d A	25	/*-
9385eb3d A	26	* Copyright (c) 1990, 1993, 1994
e9ce8d39 A	27	* The Regents of the University of California. All rights reserved.
	28	*
	29	* Redistribution and use in source and binary forms, with or without
	30	* modification, are permitted provided that the following conditions
	31	* are met:
	32	* 1. Redistributions of source code must retain the above copyright
	33	* notice, this list of conditions and the following disclaimer.
	34	* 2. Redistributions in binary form must reproduce the above copyright
	35	* notice, this list of conditions and the following disclaimer in the
	36	* documentation and/or other materials provided with the distribution.
	37	* 3. All advertising materials mentioning features or use of this software
	38	* must display the following acknowledgement:
	39	* This product includes software developed by the University of
	40	* California, Berkeley and its contributors.
	41	* 4. Neither the name of the University nor the names of its contributors
	42	* may be used to endorse or promote products derived from this software
	43	* without specific prior written permission.
	44	*
	45	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
	46	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	47	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	48	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
	49	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	50	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	51	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	52	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	53	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	54	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	55	* SUCH DAMAGE.
	56	*/
	57
9385eb3d A	58	#if defined(LIBC_SCCS) && !defined(lint)
	59	static char sccsid[] = "@(#)mpool.c 8.5 (Berkeley) 7/26/94";
	60	#endif /* LIBC_SCCS and not lint */
	61	#include <sys/cdefs.h>
e9ce8d39 A	62
e9ce8d39 A	63	#include <sys/param.h>
9385eb3d	64	#include <sys/queue.h>
e9ce8d39 A	65	#include <sys/stat.h>
	66
	67	#include <errno.h>
	68	#include <stdio.h>
	69	#include <stdlib.h>
	70	#include <string.h>
	71	#include <unistd.h>
	72
	73	#include <db.h>
9385eb3d	74
e9ce8d39	75	#define __MPOOLINTERFACE_PRIVATE
9385eb3d	76	#include <mpool.h>
e9ce8d39	77
9385eb3d A	78	static BKT mpool_bkt(MPOOL );
	79	static BKT mpool_look(MPOOL , pgno_t);
	80	static int mpool_write(MPOOL , BKT );
e9ce8d39 A	81
e9ce8d39 A	82	/*
9385eb3d A	83	* mpool_open --
9385eb3d A	84	* Initialize a memory pool.
e9ce8d39 A	85	*/
	86	MPOOL *
	87	mpool_open(key, fd, pagesize, maxcache)
9385eb3d	88	void *key;
e9ce8d39 A	89	int fd;
	90	pgno_t pagesize, maxcache;
	91	{
	92	struct stat sb;
	93	MPOOL *mp;
	94	int entry;
	95
9385eb3d A	96	/*
	97	* Get information about the file.
	98	*
	99	* XXX
	100	* We don't currently handle pipes, although we should.
	101	*/
e9ce8d39 A	102	if (fstat(fd, &sb))
e9ce8d39 A	103	return (NULL);
e9ce8d39 A	104	if (!S_ISREG(sb.st_mode)) {
	105	errno = ESPIPE;
	106	return (NULL);
	107	}
	108
9385eb3d A	109	/* Allocate and initialize the MPOOL cookie. */
9385eb3d A	110	if ((mp = (MPOOL *)calloc(1, sizeof(MPOOL))) == NULL)
e9ce8d39	111	return (NULL);
9385eb3d	112	TAILQ_INIT(&mp->lqh);
e9ce8d39	113	for (entry = 0; entry < HASHSIZE; ++entry)
9385eb3d	114	TAILQ_INIT(&mp->hqh[entry]);
e9ce8d39	115	mp->maxcache = maxcache;
e9ce8d39	116	mp->npages = sb.st_size / pagesize;
9385eb3d	117	mp->pagesize = pagesize;
e9ce8d39	118	mp->fd = fd;
e9ce8d39 A	119	return (mp);
	120	}
	121
	122	/*
9385eb3d A	123	* mpool_filter --
9385eb3d A	124	* Initialize input/output filters.
e9ce8d39 A	125	*/
	126	void
	127	mpool_filter(mp, pgin, pgout, pgcookie)
	128	MPOOL *mp;
9385eb3d A	129	void (pgin)(void , pgno_t, void *);
9385eb3d A	130	void (pgout)(void , pgno_t, void *);
e9ce8d39 A	131	void *pgcookie;
	132	{
	133	mp->pgin = pgin;
	134	mp->pgout = pgout;
	135	mp->pgcookie = pgcookie;
	136	}
	137
	138	/*
9385eb3d A	139	* mpool_new --
9385eb3d A	140	* Get a new page of memory.
e9ce8d39 A	141	*/
	142	void *
	143	mpool_new(mp, pgnoaddr)
	144	MPOOL *mp;
	145	pgno_t *pgnoaddr;
	146	{
9385eb3d A	147	struct _hqh *head;
9385eb3d A	148	BKT *bp;
e9ce8d39	149
9385eb3d A	150	if (mp->npages == MAX_PAGE_NUMBER) {
	151	(void)fprintf(stderr, "mpool_new: page allocation overflow.\n");
	152	abort();
	153	}
e9ce8d39 A	154	#ifdef STATISTICS
	155	++mp->pagenew;
	156	#endif
	157	/*
9385eb3d A	158	* Get a BKT from the cache. Assign a new page number, attach
	159	* it to the head of the hash chain, the tail of the lru chain,
	160	* and return.
e9ce8d39	161	*/
9385eb3d	162	if ((bp = mpool_bkt(mp)) == NULL)
e9ce8d39	163	return (NULL);
9385eb3d A	164	*pgnoaddr = bp->pgno = mp->npages++;
	165	bp->flags = MPOOL_PINNED;
	166
	167	head = &mp->hqh[HASHKEY(bp->pgno)];
	168	TAILQ_INSERT_HEAD(head, bp, hq);
	169	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	170	return (bp->page);
e9ce8d39 A	171	}
	172
	173	/*
9385eb3d A	174	* mpool_get
9385eb3d A	175	* Get a page.
e9ce8d39 A	176	*/
	177	void *
	178	mpool_get(mp, pgno, flags)
	179	MPOOL *mp;
	180	pgno_t pgno;
9385eb3d	181	u_int flags; /* XXX not used? */
e9ce8d39	182	{
9385eb3d A	183	struct _hqh *head;
9385eb3d A	184	BKT *bp;
e9ce8d39 A	185	off_t off;
	186	int nr;
	187
9385eb3d A	188	/* Check for attempt to retrieve a non-existent page. */
	189	if (pgno >= mp->npages) {
	190	errno = EINVAL;
	191	return (NULL);
	192	}
	193
e9ce8d39	194	#ifdef STATISTICS
9385eb3d	195	++mp->pageget;
e9ce8d39	196	#endif
9385eb3d A	197
	198	/* Check for a page that is cached. */
	199	if ((bp = mpool_look(mp, pgno)) != NULL) {
e9ce8d39	200	#ifdef DEBUG
9385eb3d A	201	if (bp->flags & MPOOL_PINNED) {
	202	(void)fprintf(stderr,
	203	"mpool_get: page %d already pinned\n", bp->pgno);
	204	abort();
	205	}
e9ce8d39	206	#endif
9385eb3d A	207	/*
	208	* Move the page to the head of the hash chain and the tail
	209	* of the lru chain.
	210	*/
	211	head = &mp->hqh[HASHKEY(bp->pgno)];
	212	TAILQ_REMOVE(head, bp, hq);
	213	TAILQ_INSERT_HEAD(head, bp, hq);
	214	TAILQ_REMOVE(&mp->lqh, bp, q);
	215	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	216
	217	/* Return a pinned page. */
	218	bp->flags \|= MPOOL_PINNED;
	219	return (bp->page);
e9ce8d39 A	220	}
	221
	222	/* Get a page from the cache. */
9385eb3d	223	if ((bp = mpool_bkt(mp)) == NULL)
e9ce8d39	224	return (NULL);
e9ce8d39	225
9385eb3d	226	/* Read in the contents. */
e9ce8d39 A	227	#ifdef STATISTICS
	228	++mp->pageread;
	229	#endif
e9ce8d39 A	230	off = mp->pagesize * pgno;
	231	if (lseek(mp->fd, off, SEEK_SET) != off)
	232	return (NULL);
9385eb3d	233	if ((nr = read(mp->fd, bp->page, mp->pagesize)) != mp->pagesize) {
e9ce8d39 A	234	if (nr >= 0)
	235	errno = EFTYPE;
	236	return (NULL);
	237	}
e9ce8d39	238
9385eb3d A	239	/* Set the page number, pin the page. */
	240	bp->pgno = pgno;
	241	bp->flags = MPOOL_PINNED;
	242
	243	/*
	244	* Add the page to the head of the hash chain and the tail
	245	* of the lru chain.
	246	*/
	247	head = &mp->hqh[HASHKEY(bp->pgno)];
	248	TAILQ_INSERT_HEAD(head, bp, hq);
	249	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	250
	251	/* Run through the user's filter. */
	252	if (mp->pgin != NULL)
	253	(mp->pgin)(mp->pgcookie, bp->pgno, bp->page);
	254
	255	return (bp->page);
e9ce8d39 A	256	}
	257
	258	/*
9385eb3d A	259	* mpool_put
9385eb3d A	260	* Return a page.
e9ce8d39 A	261	*/
	262	int
	263	mpool_put(mp, page, flags)
	264	MPOOL *mp;
	265	void *page;
	266	u_int flags;
	267	{
9385eb3d	268	BKT *bp;
e9ce8d39 A	269
	270	#ifdef STATISTICS
	271	++mp->pageput;
	272	#endif
9385eb3d	273	bp = (BKT )((char )page - sizeof(BKT));
e9ce8d39	274	#ifdef DEBUG
9385eb3d A	275	if (!(bp->flags & MPOOL_PINNED)) {
	276	(void)fprintf(stderr,
	277	"mpool_put: page %d not pinned\n", bp->pgno);
	278	abort();
e9ce8d39 A	279	}
e9ce8d39 A	280	#endif
9385eb3d A	281	bp->flags &= ~MPOOL_PINNED;
9385eb3d A	282	bp->flags \|= flags & MPOOL_DIRTY;
e9ce8d39 A	283	return (RET_SUCCESS);
	284	}
	285
	286	/*
9385eb3d A	287	* mpool_close
9385eb3d A	288	* Close the buffer pool.
e9ce8d39 A	289	*/
	290	int
	291	mpool_close(mp)
	292	MPOOL *mp;
	293	{
9385eb3d	294	BKT *bp;
e9ce8d39 A	295
e9ce8d39 A	296	/* Free up any space allocated to the lru pages. */
9385eb3d A	297	while (!TAILQ_EMPTY(&mp->lqh)) {
	298	bp = TAILQ_FIRST(&mp->lqh);
	299	TAILQ_REMOVE(&mp->lqh, bp, q);
	300	free(bp);
e9ce8d39	301	}
9385eb3d A	302
9385eb3d A	303	/* Free the MPOOL cookie. */
e9ce8d39 A	304	free(mp);
	305	return (RET_SUCCESS);
	306	}
	307
	308	/*
9385eb3d A	309	* mpool_sync
9385eb3d A	310	* Sync the pool to disk.
e9ce8d39 A	311	*/
	312	int
	313	mpool_sync(mp)
	314	MPOOL *mp;
	315	{
9385eb3d	316	BKT *bp;
e9ce8d39	317
9385eb3d A	318	/* Walk the lru chain, flushing any dirty pages to disk. */
	319	TAILQ_FOREACH(bp, &mp->lqh, q)
	320	if (bp->flags & MPOOL_DIRTY &&
	321	mpool_write(mp, bp) == RET_ERROR)
e9ce8d39	322	return (RET_ERROR);
9385eb3d A	323
9385eb3d A	324	/* Sync the file descriptor. */
e9ce8d39 A	325	return (fsync(mp->fd) ? RET_ERROR : RET_SUCCESS);
	326	}
	327
	328	/*
9385eb3d A	329	* mpool_bkt
9385eb3d A	330	* Get a page from the cache (or create one).
e9ce8d39 A	331	*/
	332	static BKT *
	333	mpool_bkt(mp)
	334	MPOOL *mp;
	335	{
9385eb3d A	336	struct _hqh *head;
9385eb3d A	337	BKT *bp;
e9ce8d39	338
9385eb3d	339	/* If under the max cached, always create a new page. */
e9ce8d39 A	340	if (mp->curcache < mp->maxcache)
	341	goto new;
	342
	343	/*
9385eb3d A	344	* If the cache is max'd out, walk the lru list for a buffer we
	345	* can flush. If we find one, write it (if necessary) and take it
	346	* off any lists. If we don't find anything we grow the cache anyway.
e9ce8d39 A	347	* The cache never shrinks.
e9ce8d39 A	348	*/
9385eb3d A	349	TAILQ_FOREACH(bp, &mp->lqh, q)
	350	if (!(bp->flags & MPOOL_PINNED)) {
	351	/* Flush if dirty. */
	352	if (bp->flags & MPOOL_DIRTY &&
	353	mpool_write(mp, bp) == RET_ERROR)
e9ce8d39	354	return (NULL);
e9ce8d39 A	355	#ifdef STATISTICS
	356	++mp->pageflush;
	357	#endif
9385eb3d A	358	/* Remove from the hash and lru queues. */
	359	head = &mp->hqh[HASHKEY(bp->pgno)];
	360	TAILQ_REMOVE(head, bp, hq);
	361	TAILQ_REMOVE(&mp->lqh, bp, q);
e9ce8d39	362	#ifdef DEBUG
9385eb3d A	363	{ void *spage;
	364	spage = bp->page;
	365	memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
	366	bp->page = spage;
e9ce8d39 A	367	}
e9ce8d39 A	368	#endif
9385eb3d	369	return (bp);
e9ce8d39 A	370	}
e9ce8d39 A	371
9385eb3d	372	new: if ((bp = (BKT *)malloc(sizeof(BKT) + mp->pagesize)) == NULL)
e9ce8d39 A	373	return (NULL);
	374	#ifdef STATISTICS
	375	++mp->pagealloc;
	376	#endif
9385eb3d A	377	#if defined(DEBUG) \|\| defined(PURIFY)
9385eb3d A	378	memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
e9ce8d39	379	#endif
9385eb3d	380	bp->page = (char *)bp + sizeof(BKT);
e9ce8d39	381	++mp->curcache;
9385eb3d	382	return (bp);
e9ce8d39 A	383	}
	384
	385	/*
9385eb3d A	386	* mpool_write
9385eb3d A	387	* Write a page to disk.
e9ce8d39 A	388	*/
e9ce8d39 A	389	static int
9385eb3d	390	mpool_write(mp, bp)
e9ce8d39	391	MPOOL *mp;
9385eb3d	392	BKT *bp;
e9ce8d39 A	393	{
	394	off_t off;
	395
e9ce8d39 A	396	#ifdef STATISTICS
	397	++mp->pagewrite;
	398	#endif
9385eb3d A	399
	400	/* Run through the user's filter. */
	401	if (mp->pgout)
	402	(mp->pgout)(mp->pgcookie, bp->pgno, bp->page);
	403
	404	off = mp->pagesize * bp->pgno;
e9ce8d39 A	405	if (lseek(mp->fd, off, SEEK_SET) != off)
e9ce8d39 A	406	return (RET_ERROR);
9385eb3d	407	if (write(mp->fd, bp->page, mp->pagesize) != mp->pagesize)
e9ce8d39	408	return (RET_ERROR);
9385eb3d A	409
9385eb3d A	410	bp->flags &= ~MPOOL_DIRTY;
e9ce8d39 A	411	return (RET_SUCCESS);
	412	}
	413
	414	/*
9385eb3d A	415	* mpool_look
9385eb3d A	416	* Lookup a page in the cache.
e9ce8d39 A	417	*/
	418	static BKT *
	419	mpool_look(mp, pgno)
	420	MPOOL *mp;
	421	pgno_t pgno;
	422	{
9385eb3d A	423	struct _hqh *head;
9385eb3d A	424	BKT *bp;
e9ce8d39	425
9385eb3d A	426	head = &mp->hqh[HASHKEY(pgno)];
	427	TAILQ_FOREACH(bp, head, hq)
	428	if (bp->pgno == pgno) {
e9ce8d39 A	429	#ifdef STATISTICS
	430	++mp->cachehit;
	431	#endif
9385eb3d	432	return (bp);
e9ce8d39 A	433	}
	434	#ifdef STATISTICS
	435	++mp->cachemiss;
	436	#endif
	437	return (NULL);
	438	}
	439
	440	#ifdef STATISTICS
	441	/*
9385eb3d A	442	* mpool_stat
9385eb3d A	443	* Print out cache statistics.
e9ce8d39 A	444	*/
	445	void
	446	mpool_stat(mp)
	447	MPOOL *mp;
	448	{
9385eb3d	449	BKT *bp;
e9ce8d39 A	450	int cnt;
	451	char *sep;
	452
	453	(void)fprintf(stderr, "%lu pages in the file\n", mp->npages);
	454	(void)fprintf(stderr,
	455	"page size %lu, cacheing %lu pages of %lu page max cache\n",
	456	mp->pagesize, mp->curcache, mp->maxcache);
	457	(void)fprintf(stderr, "%lu page puts, %lu page gets, %lu page new\n",
	458	mp->pageput, mp->pageget, mp->pagenew);
	459	(void)fprintf(stderr, "%lu page allocs, %lu page flushes\n",
	460	mp->pagealloc, mp->pageflush);
	461	if (mp->cachehit + mp->cachemiss)
	462	(void)fprintf(stderr,
	463	"%.0f%% cache hit rate (%lu hits, %lu misses)\n",
	464	((double)mp->cachehit / (mp->cachehit + mp->cachemiss))
	465	* 100, mp->cachehit, mp->cachemiss);
	466	(void)fprintf(stderr, "%lu page reads, %lu page writes\n",
	467	mp->pageread, mp->pagewrite);
	468
	469	sep = "";
	470	cnt = 0;
9385eb3d A	471	TAILQ_FOREACH(bp, &mp->lqh, q) {
	472	(void)fprintf(stderr, "%s%d", sep, bp->pgno);
	473	if (bp->flags & MPOOL_DIRTY)
e9ce8d39	474	(void)fprintf(stderr, "d");
9385eb3d	475	if (bp->flags & MPOOL_PINNED)
e9ce8d39 A	476	(void)fprintf(stderr, "P");
	477	if (++cnt == 10) {
	478	sep = "\n";
	479	cnt = 0;
	480	} else
	481	sep = ", ";
	482
	483	}
	484	(void)fprintf(stderr, "\n");
	485	}
	486	#endif