freebsd-skq/sys/fs/nullfs/null_subr.c

322 lines
8.2 KiB
C
Raw Normal View History

/*-
1994-05-24 10:09:53 +00:00
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software donated to Berkeley by
* Jan-Simon Pendry.
*
* 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.
* 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.
*
* @(#)null_subr.c 8.7 (Berkeley) 5/14/95
1994-05-24 10:09:53 +00:00
*
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mutex.h>
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/proc.h>
#include <sys/vnode.h>
#include <fs/nullfs/null.h>
1994-05-24 10:09:53 +00:00
/*
* Null layer cache:
* Each cache entry holds a reference to the lower vnode
* along with a pointer to the alias vnode. When an
* entry is added the lower vnode is VREF'd. When the
* alias is removed the lower vnode is vrele'd.
*/
#define NULL_NHASH(vp) (&null_node_hashtbl[vfs_hash_index(vp) & null_hash_mask])
static LIST_HEAD(null_node_hashhead, null_node) *null_node_hashtbl;
static struct mtx null_hashmtx;
static u_long null_hash_mask;
static MALLOC_DEFINE(M_NULLFSHASH, "nullfs_hash", "NULLFS hash table");
MALLOC_DEFINE(M_NULLFSNODE, "nullfs_node", "NULLFS vnode private part");
1994-05-24 10:09:53 +00:00
static struct vnode * null_hashins(struct mount *, struct null_node *);
1994-05-24 10:09:53 +00:00
/*
* Initialise cache headers
*/
int
nullfs_init(vfsp)
struct vfsconf *vfsp;
1994-05-24 10:09:53 +00:00
{
null_node_hashtbl = hashinit(desiredvnodes, M_NULLFSHASH,
&null_hash_mask);
mtx_init(&null_hashmtx, "nullhs", NULL, MTX_DEF);
return (0);
}
int
nullfs_uninit(vfsp)
struct vfsconf *vfsp;
{
mtx_destroy(&null_hashmtx);
hashdestroy(null_node_hashtbl, M_NULLFSHASH, null_hash_mask);
return (0);
1994-05-24 10:09:53 +00:00
}
/*
* Return a VREF'ed alias for lower vnode if already exists, else 0.
* Lower vnode should be locked on entry and will be left locked on exit.
1994-05-24 10:09:53 +00:00
*/
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
struct vnode *
null_hashget(mp, lowervp)
struct mount *mp;
1994-05-24 10:09:53 +00:00
struct vnode *lowervp;
{
struct null_node_hashhead *hd;
1994-05-24 10:09:53 +00:00
struct null_node *a;
struct vnode *vp;
ASSERT_VOP_LOCKED(lowervp, "null_hashget");
1994-05-24 10:09:53 +00:00
/*
* Find hash base, and then search the (two-way) linked
* list looking for a null_node structure which is referencing
* the lower vnode. If found, the increment the null_node
* reference count (but NOT the lower vnode's VREF counter).
*/
hd = NULL_NHASH(lowervp);
mtx_lock(&null_hashmtx);
LIST_FOREACH(a, hd, null_hash) {
if (a->null_lowervp == lowervp && NULLTOV(a)->v_mount == mp) {
/*
* Since we have the lower node locked the nullfs
* node can not be in the process of recycling. If
* it had been recycled before we grabed the lower
* lock it would not have been found on the hash.
*/
vp = NULLTOV(a);
vref(vp);
mtx_unlock(&null_hashmtx);
1994-05-24 10:09:53 +00:00
return (vp);
}
}
mtx_unlock(&null_hashmtx);
return (NULLVP);
1994-05-24 10:09:53 +00:00
}
/*
* Act like null_hashget, but add passed null_node to hash if no existing
* node found.
*/
static struct vnode *
null_hashins(mp, xp)
struct mount *mp;
struct null_node *xp;
{
struct null_node_hashhead *hd;
struct null_node *oxp;
struct vnode *ovp;
hd = NULL_NHASH(xp->null_lowervp);
mtx_lock(&null_hashmtx);
LIST_FOREACH(oxp, hd, null_hash) {
if (oxp->null_lowervp == xp->null_lowervp &&
NULLTOV(oxp)->v_mount == mp) {
/*
* See null_hashget for a description of this
* operation.
*/
ovp = NULLTOV(oxp);
vref(ovp);
mtx_unlock(&null_hashmtx);
return (ovp);
}
}
LIST_INSERT_HEAD(hd, xp, null_hash);
mtx_unlock(&null_hashmtx);
return (NULLVP);
}
1994-05-24 10:09:53 +00:00
static void
null_destroy_proto(struct vnode *vp, void *xp)
{
lockmgr(&vp->v_lock, LK_EXCLUSIVE, NULL);
VI_LOCK(vp);
vp->v_data = NULL;
vp->v_vnlock = &vp->v_lock;
vp->v_op = &dead_vnodeops;
VI_UNLOCK(vp);
vgone(vp);
vput(vp);
free(xp, M_NULLFSNODE);
}
static void
null_insmntque_dtr(struct vnode *vp, void *xp)
{
vput(((struct null_node *)xp)->null_lowervp);
null_destroy_proto(vp, xp);
}
1994-05-24 10:09:53 +00:00
/*
* Make a new or get existing nullfs node.
* Vp is the alias vnode, lowervp is the lower vnode.
*
* The lowervp assumed to be locked and having "spare" reference. This routine
* vrele lowervp if nullfs node was taken from hash. Otherwise it "transfers"
* the caller's "spare" reference to created nullfs vnode.
1994-05-24 10:09:53 +00:00
*/
int
null_nodeget(mp, lowervp, vpp)
1994-05-24 10:09:53 +00:00
struct mount *mp;
struct vnode *lowervp;
struct vnode **vpp;
{
struct null_node *xp;
struct vnode *vp;
1994-05-24 10:09:53 +00:00
int error;
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
ASSERT_VOP_LOCKED(lowervp, "lowervp");
KASSERT(lowervp->v_usecount >= 1, ("Unreferenced vnode %p", lowervp));
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
/* Lookup the hash firstly. */
*vpp = null_hashget(mp, lowervp);
if (*vpp != NULL) {
vrele(lowervp);
return (0);
}
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
/*
* The insmntque1() call below requires the exclusive lock on
* the nullfs vnode. Upgrade the lock now if hash failed to
* provide ready to use vnode.
*/
if (VOP_ISLOCKED(lowervp) != LK_EXCLUSIVE) {
KASSERT((MOUNTTONULLMOUNT(mp)->nullm_flags & NULLM_CACHE) != 0,
("lowervp %p is not excl locked and cache is disabled",
lowervp));
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
vn_lock(lowervp, LK_UPGRADE | LK_RETRY);
if ((lowervp->v_iflag & VI_DOOMED) != 0) {
vput(lowervp);
return (ENOENT);
}
}
/*
* We do not serialize vnode creation, instead we will check for
* duplicates later, when adding new vnode to hash.
* Note that duplicate can only appear in hash if the lowervp is
* locked LK_SHARED.
*/
Allow shared lookups for nullfs mounts, if lower filesystem supports it. There are two problems which shall be addressed for shared lookups use to have measurable effect on nullfs scalability: 1. When vfs_lookup() calls VOP_LOOKUP() for nullfs, which passes lookup operation to lower fs, resulting vnode is often only shared-locked. Then null_nodeget() cannot instantiate covering vnode for lower vnode, since insmntque1() and null_hashins() require exclusive lock on the lower. Change the assert that lower vnode is exclusively locked to only require any lock. If null hash failed to find pre-existing nullfs vnode for lower vnode and the vnode is shared-locked, the lower vnode lock is upgraded. 2. Nullfs reclaims its vnodes on deactivation. This is due to nullfs inability to detect reclamation of the lower vnode. Reclamation of a nullfs vnode at deactivation time prevents a reference to the lower vnode to become stale. Change nullfs VOP_INACTIVE to not reclaim the vnode, instead use the VFS_RECLAIM_LOWERVP to get notification and reclaim upper vnode together with the reclamation of the lower vnode. Note that nullfs reclamation procedure calls vput() on the lowervp vnode, temporary unlocking the vnode being reclaimed. This seems to be fine for MPSAFE filesystems, but not-MPSAFE code often put partially initialized vnode on some globally visible list, and later can decide that half-constructed vnode is not needed. If nullfs mount is created above such filesystem, then other threads might catch such not properly initialized vnode. Instead of trying to overcome this case, e.g. by recursing the lower vnode lock in null_reclaim_lowervp(), I decided to rely on nearby removal of the support for non-MPSAFE filesystems. In collaboration with: pho MFC after: 3 weeks
2012-09-09 19:20:23 +00:00
xp = malloc(sizeof(struct null_node), M_NULLFSNODE, M_WAITOK);
error = getnewvnode("null", mp, &null_vnodeops, &vp);
if (error) {
vput(lowervp);
free(xp, M_NULLFSNODE);
1994-05-24 10:09:53 +00:00
return (error);
}
1994-05-24 10:09:53 +00:00
xp->null_vnode = vp;
xp->null_lowervp = lowervp;
vp->v_type = lowervp->v_type;
vp->v_data = xp;
vp->v_vnlock = lowervp->v_vnlock;
error = insmntque1(vp, mp, null_insmntque_dtr, xp);
if (error != 0)
return (error);
/*
* Atomically insert our new node into the hash or vget existing
* if someone else has beaten us to it.
*/
*vpp = null_hashins(mp, xp);
if (*vpp != NULL) {
vrele(lowervp);
null_destroy_proto(vp, xp);
return (0);
1994-05-24 10:09:53 +00:00
}
*vpp = vp;
1994-05-24 10:09:53 +00:00
return (0);
}
1997-08-27 08:44:43 +00:00
/*
* Remove node from hash.
*/
void
null_hashrem(xp)
struct null_node *xp;
{
mtx_lock(&null_hashmtx);
LIST_REMOVE(xp, null_hash);
mtx_unlock(&null_hashmtx);
}
#ifdef DIAGNOSTIC
1994-05-24 10:09:53 +00:00
struct vnode *
null_checkvp(vp, fil, lno)
struct vnode *vp;
char *fil;
int lno;
{
struct null_node *a = VTONULL(vp);
1994-05-24 10:09:53 +00:00
#ifdef notyet
/*
* Can't do this check because vop_reclaim runs
* with a funny vop vector.
*/
if (vp->v_op != null_vnodeop_p) {
printf ("null_checkvp: on non-null-node\n");
panic("null_checkvp");
}
1994-05-24 10:09:53 +00:00
#endif
1997-04-17 11:17:30 +00:00
if (a->null_lowervp == NULLVP) {
1994-05-24 10:09:53 +00:00
/* Should never happen */
panic("null_checkvp %p", vp);
1994-05-24 10:09:53 +00:00
}
VI_LOCK_FLAGS(a->null_lowervp, MTX_DUPOK);
if (a->null_lowervp->v_usecount < 1)
panic ("null with unref'ed lowervp, vp %p lvp %p",
vp, a->null_lowervp);
VI_UNLOCK(a->null_lowervp);
1994-05-24 10:09:53 +00:00
#ifdef notyet
printf("null %x/%d -> %x/%d [%s, %d]\n",
NULLTOV(a), vrefcnt(NULLTOV(a)),
a->null_lowervp, vrefcnt(a->null_lowervp),
1994-05-24 10:09:53 +00:00
fil, lno);
#endif
return (a->null_lowervp);
1994-05-24 10:09:53 +00:00
}
#endif