freebsd-dev/sys/kern/vfs_hash.c
Konstantin Belousov f19063ab02 vfs_hash_rehash(): require the vnode to be exclusively locked
Rehash updates v_hash.  Also, rehash moves the vnode to different hash
bucket, which should be noticed in vfs_hash_get() after sleeping for
the vnode lock.

Reviewed by:	mckusick, rmacklem
Tested by:	pho
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
Differential revision:	https://reviews.freebsd.org/D31464
2021-08-27 18:39:45 +03:00

244 lines
6.2 KiB
C

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2005 Poul-Henning Kamp
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/rwlock.h>
#include <sys/vnode.h>
static MALLOC_DEFINE(M_VFS_HASH, "vfs_hash", "VFS hash table");
static LIST_HEAD(vfs_hash_head, vnode) *vfs_hash_tbl;
static LIST_HEAD(,vnode) vfs_hash_side;
static u_long vfs_hash_mask;
static struct rwlock __exclusive_cache_line vfs_hash_lock;
static void
vfs_hashinit(void *dummy __unused)
{
vfs_hash_tbl = hashinit(desiredvnodes, M_VFS_HASH, &vfs_hash_mask);
rw_init(&vfs_hash_lock, "vfs hash");
LIST_INIT(&vfs_hash_side);
}
/* Must be SI_ORDER_SECOND so desiredvnodes is available */
SYSINIT(vfs_hash, SI_SUB_VFS, SI_ORDER_SECOND, vfs_hashinit, NULL);
u_int
vfs_hash_index(struct vnode *vp)
{
return (vp->v_hash + vp->v_mount->mnt_hashseed);
}
static struct vfs_hash_head *
vfs_hash_bucket(const struct mount *mp, u_int hash)
{
return (&vfs_hash_tbl[(hash + mp->mnt_hashseed) & vfs_hash_mask]);
}
int
vfs_hash_get(const struct mount *mp, u_int hash, int flags, struct thread *td,
struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
{
struct vnode *vp;
enum vgetstate vs;
int error;
while (1) {
rw_rlock(&vfs_hash_lock);
LIST_FOREACH(vp, vfs_hash_bucket(mp, hash), v_hashlist) {
if (vp->v_hash != hash)
continue;
if (vp->v_mount != mp)
continue;
if (fn != NULL && fn(vp, arg))
continue;
vs = vget_prep(vp);
rw_runlock(&vfs_hash_lock);
error = vget_finish(vp, flags, vs);
if (error == ENOENT && (flags & LK_NOWAIT) == 0)
break;
if (error != 0)
return (error);
if (vp->v_hash != hash ||
(fn != NULL && fn(vp, arg))) {
vput(vp);
/* Restart the bucket walk. */
break;
}
*vpp = vp;
return (0);
}
if (vp == NULL) {
rw_runlock(&vfs_hash_lock);
*vpp = NULL;
return (0);
}
}
}
void
vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
{
struct vnode *vp;
while (1) {
rw_rlock(&vfs_hash_lock);
LIST_FOREACH(vp, vfs_hash_bucket(mp, hash), v_hashlist) {
if (vp->v_hash != hash)
continue;
if (vp->v_mount != mp)
continue;
if (fn != NULL && fn(vp, arg))
continue;
vhold(vp);
rw_runlock(&vfs_hash_lock);
vref(vp);
vdrop(vp);
*vpp = vp;
return;
}
if (vp == NULL) {
rw_runlock(&vfs_hash_lock);
*vpp = NULL;
return;
}
}
}
void
vfs_hash_remove(struct vnode *vp)
{
rw_wlock(&vfs_hash_lock);
LIST_REMOVE(vp, v_hashlist);
rw_wunlock(&vfs_hash_lock);
}
int
vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
{
struct vnode *vp2;
enum vgetstate vs;
int error;
*vpp = NULL;
while (1) {
rw_wlock(&vfs_hash_lock);
LIST_FOREACH(vp2,
vfs_hash_bucket(vp->v_mount, hash), v_hashlist) {
if (vp2->v_hash != hash)
continue;
if (vp2->v_mount != vp->v_mount)
continue;
if (fn != NULL && fn(vp2, arg))
continue;
vs = vget_prep(vp2);
rw_wunlock(&vfs_hash_lock);
error = vget_finish(vp2, flags, vs);
if (error == ENOENT && (flags & LK_NOWAIT) == 0)
break;
rw_wlock(&vfs_hash_lock);
LIST_INSERT_HEAD(&vfs_hash_side, vp, v_hashlist);
rw_wunlock(&vfs_hash_lock);
vgone(vp);
vput(vp);
if (!error)
*vpp = vp2;
return (error);
}
if (vp2 == NULL)
break;
}
vp->v_hash = hash;
LIST_INSERT_HEAD(vfs_hash_bucket(vp->v_mount, hash), vp, v_hashlist);
rw_wunlock(&vfs_hash_lock);
return (0);
}
void
vfs_hash_rehash(struct vnode *vp, u_int hash)
{
ASSERT_VOP_ELOCKED(vp, "rehash requires excl lock");
rw_wlock(&vfs_hash_lock);
LIST_REMOVE(vp, v_hashlist);
LIST_INSERT_HEAD(vfs_hash_bucket(vp->v_mount, hash), vp, v_hashlist);
vp->v_hash = hash;
rw_wunlock(&vfs_hash_lock);
}
void
vfs_hash_changesize(u_long newmaxvnodes)
{
struct vfs_hash_head *vfs_hash_newtbl, *vfs_hash_oldtbl;
u_long vfs_hash_newmask, vfs_hash_oldmask;
struct vnode *vp;
int i;
vfs_hash_newtbl = hashinit(newmaxvnodes, M_VFS_HASH,
&vfs_hash_newmask);
/* If same hash table size, nothing to do */
if (vfs_hash_mask == vfs_hash_newmask) {
free(vfs_hash_newtbl, M_VFS_HASH);
return;
}
/*
* Move everything from the old hash table to the new table.
* None of the vnodes in the table can be recycled because to
* do so, they have to be removed from the hash table.
*/
rw_wlock(&vfs_hash_lock);
vfs_hash_oldtbl = vfs_hash_tbl;
vfs_hash_oldmask = vfs_hash_mask;
vfs_hash_tbl = vfs_hash_newtbl;
vfs_hash_mask = vfs_hash_newmask;
for (i = 0; i <= vfs_hash_oldmask; i++) {
while ((vp = LIST_FIRST(&vfs_hash_oldtbl[i])) != NULL) {
LIST_REMOVE(vp, v_hashlist);
LIST_INSERT_HEAD(
vfs_hash_bucket(vp->v_mount, vp->v_hash),
vp, v_hashlist);
}
}
rw_wunlock(&vfs_hash_lock);
free(vfs_hash_oldtbl, M_VFS_HASH);
}