freebsd-nq/sys/kern/vfs_cache.c

/*
 * Copyright (c) 1989, 1993
 *	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.
 *
 *	@(#)vfs_cache.c	8.3 (Berkeley) 8/22/94
 * $Id$
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/time.h>
#include <sys/mount.h>
#include <sys/vnode.h>
#include <sys/namei.h>
#include <sys/errno.h>
#include <sys/malloc.h>

/*
 * Name caching works as follows:
 *
 * Names found by directory scans are retained in a cache
 * for future reference.  It is managed LRU, so frequently
 * used names will hang around.  Cache is indexed by hash value
 * obtained from (vp, name) where vp refers to the directory
 * containing name.
 *
 * For simplicity (and economy of storage), names longer than
 * a maximum length of NCHNAMLEN are not cached; they occur
 * infrequently in any case, and are almost never of interest.
 *
 * Upon reaching the last segment of a path, if the reference
 * is for DELETE, or NOCACHE is set (rewrite), and the
 * name is located in the cache, it will be dropped.
 */

/*
 * Structures associated with name cacheing.
 */
LIST_HEAD(nchashhead, namecache) *nchashtbl;	/* Hash Table */
u_long	nchash;				/* size of hash table - 1 */
long	numcache;			/* number of cache entries allocated */
TAILQ_HEAD(, namecache) nclruhead;		/* LRU chain */
struct	nchstats nchstats;		/* cache effectiveness statistics */

int doingcache = 1;			/* 1 => enable the cache */

/*
 * Look for a the name in the cache. We don't do this
 * if the segment name is long, simply so the cache can avoid
 * holding long names (which would either waste space, or
 * add greatly to the complexity).
 *
 * Lookup is called with ni_dvp pointing to the directory to search,
 * ni_ptr pointing to the name of the entry being sought, ni_namelen
 * tells the length of the name, and ni_hash contains a hash of
 * the name. If the lookup succeeds, the vnode is returned in ni_vp
 * and a status of -1 is returned. If the lookup determines that
 * the name does not exist (negative cacheing), a status of ENOENT
 * is returned. If the lookup fails, a status of zero is returned.
 */
int
cache_lookup(dvp, vpp, cnp)
	struct vnode *dvp;
	struct vnode **vpp;
	struct componentname *cnp;
{
	register struct namecache *ncp,*nnp;
	register struct nchashhead *ncpp;

	if (!doingcache) {
		cnp->cn_flags &= ~MAKEENTRY;
		return (0);
	}
	if (cnp->cn_namelen > NCHNAMLEN) {
		nchstats.ncs_long++;
		cnp->cn_flags &= ~MAKEENTRY;
		return (0);
	}
	ncpp = &nchashtbl[cnp->cn_hash & nchash];
	for (ncp = ncpp->lh_first; ncp != 0; ncp = nnp) {
		if (ncp->nc_dvp == dvp &&
		    ncp->nc_dvpid == dvp->v_id &&
		    ncp->nc_nlen == cnp->cn_namelen &&
		    !bcmp(ncp->nc_name, cnp->cn_nameptr, (u_int)ncp->nc_nlen))
			break;
		nnp = ncp->nc_hash.le_next;
#ifdef not_yet_phk
		/*
		 * if not already last and one of the vp's are invalid,
		 * move to head of LRU
		 */
		if ((ncp->nc_lru.tqe_next != 0) &&
		   ( (ncp->nc_dvpid != ncp->nc_dvp->v_id) ||
			(ncp->nc_vp && (ncp->nc_vpid != ncp->nc_vp->v_id)))) {
			LIST_REMOVE(ncp, nc_hash);
			ncp->nc_hash.le_prev = 0;
			TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
			TAILQ_INSERT_HEAD(&nclruhead, ncp, nc_lru);
		}
#endif /* not_yet_phk */
	}
	if (ncp == 0) {
		nchstats.ncs_miss++;
		return (0);
	}
	if ((cnp->cn_flags & MAKEENTRY) == 0) {
		nchstats.ncs_badhits++;
	} else if (ncp->nc_vp == NULL) {
		if (cnp->cn_nameiop != CREATE) {
			nchstats.ncs_neghits++;
			/*
			 * Move this slot to end of LRU chain,
			 * if not already there.
			 */
			if (ncp->nc_lru.tqe_next != 0) {
				TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
				TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
			}
			return (ENOENT);
		}
	} else if (ncp->nc_vpid != ncp->nc_vp->v_id) {
		nchstats.ncs_falsehits++;
	} else {
		nchstats.ncs_goodhits++;
		/*
		 * move this slot to end of LRU chain, if not already there
		 */
		if (ncp->nc_lru.tqe_next != 0) {
			TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
			TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
		}
#ifdef not_yet_phk
		/* Touch the parent vnode */
		vtouch(ncp->nc_dvp);
#endif /* not_yet_phk */
		*vpp = ncp->nc_vp;
		return (-1);
	}

	/*
	 * Last component and we are renaming or deleting,
	 * the cache entry is invalid, or otherwise don't
	 * want cache entry to exist.
	 */
	TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
	LIST_REMOVE(ncp, nc_hash);
	ncp->nc_hash.le_prev = 0;
	TAILQ_INSERT_HEAD(&nclruhead, ncp, nc_lru);
	return (0);
}

/*
 * Add an entry to the cache
 */
void
cache_enter(dvp, vp, cnp)
	struct vnode *dvp;
	struct vnode *vp;
	struct componentname *cnp;
{
	register struct namecache *ncp;
	register struct nchashhead *ncpp;

#ifdef DIAGNOSTIC
	if (cnp->cn_namelen > NCHNAMLEN)
		panic("cache_enter: name too long");
#endif
	if (!doingcache)
		return;
	/*
	 * Free the cache slot at head of lru chain.
	 */
	if (numcache < desiredvnodes) {
		ncp = (struct namecache *)
			malloc((u_long)sizeof *ncp, M_CACHE, M_WAITOK);
		bzero((char *)ncp, sizeof *ncp);
		numcache++;
	} else if (ncp = nclruhead.tqh_first) {
		TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
		if (ncp->nc_hash.le_prev != 0) {
			LIST_REMOVE(ncp, nc_hash);
			ncp->nc_hash.le_prev = 0;
		}
	} else
		return;
	/* grab the vnode we just found */
	ncp->nc_vp = vp;
	if (vp)
		ncp->nc_vpid = vp->v_id;
	else
		ncp->nc_vpid = 0;
	/* fill in cache info */
	ncp->nc_dvp = dvp;
	ncp->nc_dvpid = dvp->v_id;
	ncp->nc_nlen = cnp->cn_namelen;
	bcopy(cnp->cn_nameptr, ncp->nc_name, (unsigned)ncp->nc_nlen);
	TAILQ_INSERT_TAIL(&nclruhead, ncp, nc_lru);
	ncpp = &nchashtbl[cnp->cn_hash & nchash];
	LIST_INSERT_HEAD(ncpp, ncp, nc_hash);
}

/*
 * Name cache initialization, from vfs_init() when we are booting
 */
void
nchinit()
{

	TAILQ_INIT(&nclruhead);
	nchashtbl = hashinit(desiredvnodes, M_CACHE, &nchash);
}

/*
 * Cache flush, a particular vnode; called when a vnode is renamed to
 * hide entries that would now be invalid
 */
void
cache_purge(vp)
	struct vnode *vp;
{
	struct namecache *ncp;
	struct nchashhead *ncpp;

	vp->v_id = ++nextvnodeid;
	if (nextvnodeid != 0)
		return;
	for (ncpp = &nchashtbl[nchash]; ncpp >= nchashtbl; ncpp--) {
		for (ncp = ncpp->lh_first; ncp != 0; ncp = ncp->nc_hash.le_next) {
			ncp->nc_vpid = 0;
			ncp->nc_dvpid = 0;
		}
	}
	vp->v_id = ++nextvnodeid;
}

/*
 * Cache flush, a whole filesystem; called when filesys is umounted to
 * remove entries that would now be invalid
 *
 * The line "nxtcp = nchhead" near the end is to avoid potential problems
 * if the cache lru chain is modified while we are dumping the
 * inode.  This makes the algorithm O(n^2), but do you think I care?
 */
void
cache_purgevfs(mp)
	struct mount *mp;
{
	register struct namecache *ncp, *nxtcp;

	for (ncp = nclruhead.tqh_first; ncp != 0; ncp = nxtcp) {
		if (ncp->nc_dvp == NULL || ncp->nc_dvp->v_mount != mp) {
			nxtcp = ncp->nc_lru.tqe_next;
			continue;
		}
		/* free the resources we had */
		ncp->nc_vp = NULL;
		ncp->nc_dvp = NULL;
		TAILQ_REMOVE(&nclruhead, ncp, nc_lru);
		if (ncp->nc_hash.le_prev != 0) {
			LIST_REMOVE(ncp, nc_hash);
			ncp->nc_hash.le_prev = 0;
		}
		/* cause rescan of list, it may have altered */
		nxtcp = nclruhead.tqh_first;
		TAILQ_INSERT_HEAD(&nclruhead, ncp, nc_lru);
	}
}