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

/*-
 * 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>
__FBSDID("$FreeBSD: src/lib/libc/db/mpool/mpool.c,v 1.12 2004/09/10 05:41:41 kuriyama Exp $");

#include "namespace.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 "un-namespace.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;
	nr = pread(mp->fd, bp->page, mp->pagesize, off);
	if (nr != 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 (pwrite(mp->fd, bp->page, mp->pagesize, off) != 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, "%u pages in the file\n", mp->npages);
	(void)fprintf(stderr,
	    "page size %lu, cacheing %u pages of %u 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
9385eb3d A	1	/*-
9385eb3d A	2	* Copyright (c) 1990, 1993, 1994
e9ce8d39 A	3	* The Regents of the University of California. All rights reserved.
	4	*
	5	* Redistribution and use in source and binary forms, with or without
	6	* modification, are permitted provided that the following conditions
	7	* are met:
	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.
	20	*
	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
	31	* SUCH DAMAGE.
	32	*/
	33
9385eb3d A	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 */
	37	#include <sys/cdefs.h>
3d9156a7	38	__FBSDID("$FreeBSD: src/lib/libc/db/mpool/mpool.c,v 1.12 2004/09/10 05:41:41 kuriyama Exp $");
e9ce8d39	39
59e0d9fe	40	#include "namespace.h"
e9ce8d39	41	#include <sys/param.h>
9385eb3d	42	#include <sys/queue.h>
e9ce8d39 A	43	#include <sys/stat.h>
	44
	45	#include <errno.h>
	46	#include <stdio.h>
	47	#include <stdlib.h>
	48	#include <string.h>
	49	#include <unistd.h>
59e0d9fe	50	#include "un-namespace.h"
e9ce8d39 A	51
e9ce8d39 A	52	#include <db.h>
9385eb3d	53
e9ce8d39	54	#define __MPOOLINTERFACE_PRIVATE
9385eb3d	55	#include <mpool.h>
e9ce8d39	56
9385eb3d A	57	static BKT mpool_bkt(MPOOL );
	58	static BKT mpool_look(MPOOL , pgno_t);
	59	static int mpool_write(MPOOL , BKT );
e9ce8d39 A	60
e9ce8d39 A	61	/*
9385eb3d A	62	* mpool_open --
9385eb3d A	63	* Initialize a memory pool.
e9ce8d39 A	64	*/
	65	MPOOL *
	66	mpool_open(key, fd, pagesize, maxcache)
9385eb3d	67	void *key;
e9ce8d39 A	68	int fd;
	69	pgno_t pagesize, maxcache;
	70	{
	71	struct stat sb;
	72	MPOOL *mp;
	73	int entry;
	74
9385eb3d A	75	/*
	76	* Get information about the file.
	77	*
	78	* XXX
	79	* We don't currently handle pipes, although we should.
	80	*/
59e0d9fe	81	if (_fstat(fd, &sb))
e9ce8d39	82	return (NULL);
e9ce8d39 A	83	if (!S_ISREG(sb.st_mode)) {
	84	errno = ESPIPE;
	85	return (NULL);
	86	}
	87
9385eb3d A	88	/* Allocate and initialize the MPOOL cookie. */
9385eb3d A	89	if ((mp = (MPOOL *)calloc(1, sizeof(MPOOL))) == NULL)
e9ce8d39	90	return (NULL);
9385eb3d	91	TAILQ_INIT(&mp->lqh);
e9ce8d39	92	for (entry = 0; entry < HASHSIZE; ++entry)
9385eb3d	93	TAILQ_INIT(&mp->hqh[entry]);
e9ce8d39	94	mp->maxcache = maxcache;
e9ce8d39	95	mp->npages = sb.st_size / pagesize;
9385eb3d	96	mp->pagesize = pagesize;
e9ce8d39	97	mp->fd = fd;
e9ce8d39 A	98	return (mp);
	99	}
	100
	101	/*
9385eb3d A	102	* mpool_filter --
9385eb3d A	103	* Initialize input/output filters.
e9ce8d39 A	104	*/
	105	void
	106	mpool_filter(mp, pgin, pgout, pgcookie)
	107	MPOOL *mp;
9385eb3d A	108	void (pgin)(void , pgno_t, void *);
9385eb3d A	109	void (pgout)(void , pgno_t, void *);
e9ce8d39 A	110	void *pgcookie;
	111	{
	112	mp->pgin = pgin;
	113	mp->pgout = pgout;
	114	mp->pgcookie = pgcookie;
	115	}
	116
	117	/*
9385eb3d A	118	* mpool_new --
9385eb3d A	119	* Get a new page of memory.
e9ce8d39 A	120	*/
	121	void *
	122	mpool_new(mp, pgnoaddr)
	123	MPOOL *mp;
	124	pgno_t *pgnoaddr;
	125	{
9385eb3d A	126	struct _hqh *head;
9385eb3d A	127	BKT *bp;
e9ce8d39	128
9385eb3d A	129	if (mp->npages == MAX_PAGE_NUMBER) {
	130	(void)fprintf(stderr, "mpool_new: page allocation overflow.\n");
	131	abort();
	132	}
e9ce8d39 A	133	#ifdef STATISTICS
	134	++mp->pagenew;
	135	#endif
	136	/*
9385eb3d A	137	* Get a BKT from the cache. Assign a new page number, attach
	138	* it to the head of the hash chain, the tail of the lru chain,
	139	* and return.
e9ce8d39	140	*/
9385eb3d	141	if ((bp = mpool_bkt(mp)) == NULL)
e9ce8d39	142	return (NULL);
9385eb3d A	143	*pgnoaddr = bp->pgno = mp->npages++;
	144	bp->flags = MPOOL_PINNED;
	145
	146	head = &mp->hqh[HASHKEY(bp->pgno)];
	147	TAILQ_INSERT_HEAD(head, bp, hq);
	148	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	149	return (bp->page);
e9ce8d39 A	150	}
	151
	152	/*
9385eb3d A	153	* mpool_get
9385eb3d A	154	* Get a page.
e9ce8d39 A	155	*/
	156	void *
	157	mpool_get(mp, pgno, flags)
	158	MPOOL *mp;
	159	pgno_t pgno;
9385eb3d	160	u_int flags; /* XXX not used? */
e9ce8d39	161	{
9385eb3d A	162	struct _hqh *head;
9385eb3d A	163	BKT *bp;
e9ce8d39 A	164	off_t off;
	165	int nr;
	166
9385eb3d A	167	/* Check for attempt to retrieve a non-existent page. */
	168	if (pgno >= mp->npages) {
	169	errno = EINVAL;
	170	return (NULL);
	171	}
	172
e9ce8d39	173	#ifdef STATISTICS
9385eb3d	174	++mp->pageget;
e9ce8d39	175	#endif
9385eb3d A	176
	177	/* Check for a page that is cached. */
	178	if ((bp = mpool_look(mp, pgno)) != NULL) {
e9ce8d39	179	#ifdef DEBUG
9385eb3d A	180	if (bp->flags & MPOOL_PINNED) {
	181	(void)fprintf(stderr,
	182	"mpool_get: page %d already pinned\n", bp->pgno);
	183	abort();
	184	}
e9ce8d39	185	#endif
9385eb3d A	186	/*
	187	* Move the page to the head of the hash chain and the tail
	188	* of the lru chain.
	189	*/
	190	head = &mp->hqh[HASHKEY(bp->pgno)];
	191	TAILQ_REMOVE(head, bp, hq);
	192	TAILQ_INSERT_HEAD(head, bp, hq);
	193	TAILQ_REMOVE(&mp->lqh, bp, q);
	194	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	195
	196	/* Return a pinned page. */
	197	bp->flags \|= MPOOL_PINNED;
	198	return (bp->page);
e9ce8d39 A	199	}
	200
	201	/* Get a page from the cache. */
9385eb3d	202	if ((bp = mpool_bkt(mp)) == NULL)
e9ce8d39	203	return (NULL);
e9ce8d39	204
9385eb3d	205	/* Read in the contents. */
e9ce8d39 A	206	#ifdef STATISTICS
	207	++mp->pageread;
	208	#endif
e9ce8d39	209	off = mp->pagesize * pgno;
3d9156a7 A	210	nr = pread(mp->fd, bp->page, mp->pagesize, off);
3d9156a7 A	211	if (nr != mp->pagesize) {
e9ce8d39 A	212	if (nr >= 0)
	213	errno = EFTYPE;
	214	return (NULL);
	215	}
e9ce8d39	216
9385eb3d A	217	/* Set the page number, pin the page. */
	218	bp->pgno = pgno;
	219	bp->flags = MPOOL_PINNED;
	220
	221	/*
	222	* Add the page to the head of the hash chain and the tail
	223	* of the lru chain.
	224	*/
	225	head = &mp->hqh[HASHKEY(bp->pgno)];
	226	TAILQ_INSERT_HEAD(head, bp, hq);
	227	TAILQ_INSERT_TAIL(&mp->lqh, bp, q);
	228
	229	/* Run through the user's filter. */
	230	if (mp->pgin != NULL)
	231	(mp->pgin)(mp->pgcookie, bp->pgno, bp->page);
	232
	233	return (bp->page);
e9ce8d39 A	234	}
	235
	236	/*
9385eb3d A	237	* mpool_put
9385eb3d A	238	* Return a page.
e9ce8d39 A	239	*/
	240	int
	241	mpool_put(mp, page, flags)
	242	MPOOL *mp;
	243	void *page;
	244	u_int flags;
	245	{
9385eb3d	246	BKT *bp;
e9ce8d39 A	247
	248	#ifdef STATISTICS
	249	++mp->pageput;
	250	#endif
9385eb3d	251	bp = (BKT )((char )page - sizeof(BKT));
e9ce8d39	252	#ifdef DEBUG
9385eb3d A	253	if (!(bp->flags & MPOOL_PINNED)) {
	254	(void)fprintf(stderr,
	255	"mpool_put: page %d not pinned\n", bp->pgno);
	256	abort();
e9ce8d39 A	257	}
e9ce8d39 A	258	#endif
9385eb3d A	259	bp->flags &= ~MPOOL_PINNED;
9385eb3d A	260	bp->flags \|= flags & MPOOL_DIRTY;
e9ce8d39 A	261	return (RET_SUCCESS);
	262	}
	263
	264	/*
9385eb3d A	265	* mpool_close
9385eb3d A	266	* Close the buffer pool.
e9ce8d39 A	267	*/
	268	int
	269	mpool_close(mp)
	270	MPOOL *mp;
	271	{
9385eb3d	272	BKT *bp;
e9ce8d39 A	273
e9ce8d39 A	274	/* Free up any space allocated to the lru pages. */
9385eb3d A	275	while (!TAILQ_EMPTY(&mp->lqh)) {
	276	bp = TAILQ_FIRST(&mp->lqh);
	277	TAILQ_REMOVE(&mp->lqh, bp, q);
	278	free(bp);
e9ce8d39	279	}
9385eb3d A	280
9385eb3d A	281	/* Free the MPOOL cookie. */
e9ce8d39 A	282	free(mp);
	283	return (RET_SUCCESS);
	284	}
	285
	286	/*
9385eb3d A	287	* mpool_sync
9385eb3d A	288	* Sync the pool to disk.
e9ce8d39 A	289	*/
	290	int
	291	mpool_sync(mp)
	292	MPOOL *mp;
	293	{
9385eb3d	294	BKT *bp;
e9ce8d39	295
9385eb3d A	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)
e9ce8d39	300	return (RET_ERROR);
9385eb3d A	301
9385eb3d A	302	/* Sync the file descriptor. */
59e0d9fe	303	return (_fsync(mp->fd) ? RET_ERROR : RET_SUCCESS);
e9ce8d39 A	304	}
	305
	306	/*
9385eb3d A	307	* mpool_bkt
9385eb3d A	308	* Get a page from the cache (or create one).
e9ce8d39 A	309	*/
	310	static BKT *
	311	mpool_bkt(mp)
	312	MPOOL *mp;
	313	{
9385eb3d A	314	struct _hqh *head;
9385eb3d A	315	BKT *bp;
e9ce8d39	316
9385eb3d	317	/* If under the max cached, always create a new page. */
e9ce8d39 A	318	if (mp->curcache < mp->maxcache)
	319	goto new;
	320
	321	/*
9385eb3d A	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.
e9ce8d39 A	325	* The cache never shrinks.
e9ce8d39 A	326	*/
9385eb3d A	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)
e9ce8d39	332	return (NULL);
e9ce8d39 A	333	#ifdef STATISTICS
	334	++mp->pageflush;
	335	#endif
9385eb3d A	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);
e9ce8d39	340	#ifdef DEBUG
9385eb3d A	341	{ void *spage;
	342	spage = bp->page;
	343	memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
	344	bp->page = spage;
e9ce8d39 A	345	}
e9ce8d39 A	346	#endif
9385eb3d	347	return (bp);
e9ce8d39 A	348	}
e9ce8d39 A	349
9385eb3d	350	new: if ((bp = (BKT *)malloc(sizeof(BKT) + mp->pagesize)) == NULL)
e9ce8d39 A	351	return (NULL);
	352	#ifdef STATISTICS
	353	++mp->pagealloc;
	354	#endif
9385eb3d A	355	#if defined(DEBUG) \|\| defined(PURIFY)
9385eb3d A	356	memset(bp, 0xff, sizeof(BKT) + mp->pagesize);
e9ce8d39	357	#endif
9385eb3d	358	bp->page = (char *)bp + sizeof(BKT);
e9ce8d39	359	++mp->curcache;
9385eb3d	360	return (bp);
e9ce8d39 A	361	}
	362
	363	/*
9385eb3d A	364	* mpool_write
9385eb3d A	365	* Write a page to disk.
e9ce8d39 A	366	*/
e9ce8d39 A	367	static int
9385eb3d	368	mpool_write(mp, bp)
e9ce8d39	369	MPOOL *mp;
9385eb3d	370	BKT *bp;
e9ce8d39 A	371	{
	372	off_t off;
	373
e9ce8d39 A	374	#ifdef STATISTICS
	375	++mp->pagewrite;
	376	#endif
9385eb3d A	377
	378	/* Run through the user's filter. */
	379	if (mp->pgout)
	380	(mp->pgout)(mp->pgcookie, bp->pgno, bp->page);
	381
	382	off = mp->pagesize * bp->pgno;
3d9156a7	383	if (pwrite(mp->fd, bp->page, mp->pagesize, off) != mp->pagesize)
e9ce8d39	384	return (RET_ERROR);
9385eb3d A	385
9385eb3d A	386	bp->flags &= ~MPOOL_DIRTY;
e9ce8d39 A	387	return (RET_SUCCESS);
	388	}
	389
	390	/*
9385eb3d A	391	* mpool_look
9385eb3d A	392	* Lookup a page in the cache.
e9ce8d39 A	393	*/
	394	static BKT *
	395	mpool_look(mp, pgno)
	396	MPOOL *mp;
	397	pgno_t pgno;
	398	{
9385eb3d A	399	struct _hqh *head;
9385eb3d A	400	BKT *bp;
e9ce8d39	401
9385eb3d A	402	head = &mp->hqh[HASHKEY(pgno)];
	403	TAILQ_FOREACH(bp, head, hq)
	404	if (bp->pgno == pgno) {
e9ce8d39 A	405	#ifdef STATISTICS
	406	++mp->cachehit;
	407	#endif
9385eb3d	408	return (bp);
e9ce8d39 A	409	}
	410	#ifdef STATISTICS
	411	++mp->cachemiss;
	412	#endif
	413	return (NULL);
	414	}
	415
	416	#ifdef STATISTICS
	417	/*
9385eb3d A	418	* mpool_stat
9385eb3d A	419	* Print out cache statistics.
e9ce8d39 A	420	*/
	421	void
	422	mpool_stat(mp)
	423	MPOOL *mp;
	424	{
9385eb3d	425	BKT *bp;
e9ce8d39 A	426	int cnt;
	427	char *sep;
	428
3d9156a7	429	(void)fprintf(stderr, "%u pages in the file\n", mp->npages);
e9ce8d39	430	(void)fprintf(stderr,
3d9156a7	431	"page size %lu, cacheing %u pages of %u page max cache\n",
e9ce8d39 A	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);
	444
	445	sep = "";
	446	cnt = 0;
9385eb3d A	447	TAILQ_FOREACH(bp, &mp->lqh, q) {
	448	(void)fprintf(stderr, "%s%d", sep, bp->pgno);
	449	if (bp->flags & MPOOL_DIRTY)
e9ce8d39	450	(void)fprintf(stderr, "d");
9385eb3d	451	if (bp->flags & MPOOL_PINNED)
e9ce8d39 A	452	(void)fprintf(stderr, "P");
	453	if (++cnt == 10) {
	454	sep = "\n";
	455	cnt = 0;
	456	} else
	457	sep = ", ";
	458
	459	}
	460	(void)fprintf(stderr, "\n");
	461	}
	462	#endif