2005-01-06 18:10:42 +00:00
|
|
|
/*-
|
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 contributed to Berkeley by
|
|
|
|
* John Heidemann of the UCLA Ficus project.
|
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*
|
1997-02-10 02:22:35 +00:00
|
|
|
* @(#)null_vnops.c 8.6 (Berkeley) 5/27/95
|
|
|
|
*
|
|
|
|
* Ancestors:
|
|
|
|
* @(#)lofs_vnops.c 1.2 (Berkeley) 6/18/92
|
|
|
|
* ...and...
|
|
|
|
* @(#)null_vnodeops.c 1.20 92/07/07 UCLA Ficus project
|
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
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Null Layer
|
|
|
|
*
|
2001-05-24 13:17:47 +00:00
|
|
|
* (See mount_nullfs(8) for more information.)
|
1994-05-24 10:09:53 +00:00
|
|
|
*
|
2002-05-16 21:28:32 +00:00
|
|
|
* The null layer duplicates a portion of the filesystem
|
1994-05-24 10:09:53 +00:00
|
|
|
* name space under a new name. In this respect, it is
|
2002-05-16 21:28:32 +00:00
|
|
|
* similar to the loopback filesystem. It differs from
|
1994-05-24 10:09:53 +00:00
|
|
|
* the loopback fs in two respects: it is implemented using
|
1998-04-17 22:37:19 +00:00
|
|
|
* a stackable layers techniques, and its "null-node"s stack above
|
1994-05-24 10:09:53 +00:00
|
|
|
* all lower-layer vnodes, not just over directory vnodes.
|
|
|
|
*
|
|
|
|
* The null layer has two purposes. First, it serves as a demonstration
|
|
|
|
* of layering by proving a layer which does nothing. (It actually
|
2002-05-16 21:28:32 +00:00
|
|
|
* does everything the loopback filesystem does, which is slightly
|
1994-05-24 10:09:53 +00:00
|
|
|
* more than nothing.) Second, the null layer can serve as a prototype
|
|
|
|
* layer. Since it provides all necessary layer framework,
|
2002-05-16 21:28:32 +00:00
|
|
|
* new filesystem layers can be created very easily be starting
|
1994-05-24 10:09:53 +00:00
|
|
|
* with a null layer.
|
|
|
|
*
|
|
|
|
* The remainder of this man page examines the null layer as a basis
|
|
|
|
* for constructing new layers.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* INSTANTIATING NEW NULL LAYERS
|
|
|
|
*
|
2001-05-24 13:17:47 +00:00
|
|
|
* New null layers are created with mount_nullfs(8).
|
|
|
|
* Mount_nullfs(8) takes two arguments, the pathname
|
1994-05-24 10:09:53 +00:00
|
|
|
* of the lower vfs (target-pn) and the pathname where the null
|
|
|
|
* layer will appear in the namespace (alias-pn). After
|
|
|
|
* the null layer is put into place, the contents
|
|
|
|
* of target-pn subtree will be aliased under alias-pn.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* OPERATION OF A NULL LAYER
|
|
|
|
*
|
2002-05-16 21:28:32 +00:00
|
|
|
* The null layer is the minimum filesystem layer,
|
1994-05-24 10:09:53 +00:00
|
|
|
* simply bypassing all possible operations to the lower layer
|
|
|
|
* for processing there. The majority of its activity centers
|
1997-06-26 16:13:56 +00:00
|
|
|
* on the bypass routine, through which nearly all vnode operations
|
1994-05-24 10:09:53 +00:00
|
|
|
* pass.
|
|
|
|
*
|
|
|
|
* The bypass routine accepts arbitrary vnode operations for
|
|
|
|
* handling by the lower layer. It begins by examing vnode
|
|
|
|
* operation arguments and replacing any null-nodes by their
|
|
|
|
* lower-layer equivlants. It then invokes the operation
|
|
|
|
* on the lower layer. Finally, it replaces the null-nodes
|
|
|
|
* in the arguments and, if a vnode is return by the operation,
|
|
|
|
* stacks a null-node on top of the returned vnode.
|
|
|
|
*
|
1997-02-10 02:22:35 +00:00
|
|
|
* Although bypass handles most operations, vop_getattr, vop_lock,
|
|
|
|
* vop_unlock, vop_inactive, vop_reclaim, and vop_print are not
|
|
|
|
* bypassed. Vop_getattr must change the fsid being returned.
|
|
|
|
* Vop_lock and vop_unlock must handle any locking for the
|
|
|
|
* current vnode as well as pass the lock request down.
|
1994-05-24 10:09:53 +00:00
|
|
|
* Vop_inactive and vop_reclaim are not bypassed so that
|
1997-02-10 02:22:35 +00:00
|
|
|
* they can handle freeing null-layer specific data. Vop_print
|
|
|
|
* is not bypassed to avoid excessive debugging information.
|
|
|
|
* Also, certain vnode operations change the locking state within
|
|
|
|
* the operation (create, mknod, remove, link, rename, mkdir, rmdir,
|
|
|
|
* and symlink). Ideally these operations should not change the
|
|
|
|
* lock state, but should be changed to let the caller of the
|
|
|
|
* function unlock them. Otherwise all intermediate vnode layers
|
|
|
|
* (such as union, umapfs, etc) must catch these functions to do
|
|
|
|
* the necessary locking at their layer.
|
1994-05-24 10:09:53 +00:00
|
|
|
*
|
|
|
|
*
|
|
|
|
* INSTANTIATING VNODE STACKS
|
|
|
|
*
|
|
|
|
* Mounting associates the null layer with a lower layer,
|
|
|
|
* effect stacking two VFSes. Vnode stacks are instead
|
|
|
|
* created on demand as files are accessed.
|
|
|
|
*
|
|
|
|
* The initial mount creates a single vnode stack for the
|
|
|
|
* root of the new null layer. All other vnode stacks
|
|
|
|
* are created as a result of vnode operations on
|
|
|
|
* this or other null vnode stacks.
|
|
|
|
*
|
|
|
|
* New vnode stacks come into existance as a result of
|
1995-05-30 08:16:23 +00:00
|
|
|
* an operation which returns a vnode.
|
1994-05-24 10:09:53 +00:00
|
|
|
* The bypass routine stacks a null-node above the new
|
|
|
|
* vnode before returning it to the caller.
|
|
|
|
*
|
|
|
|
* For example, imagine mounting a null layer with
|
2001-05-24 13:17:47 +00:00
|
|
|
* "mount_nullfs /usr/include /dev/layer/null".
|
1994-05-24 10:09:53 +00:00
|
|
|
* Changing directory to /dev/layer/null will assign
|
|
|
|
* the root null-node (which was created when the null layer was mounted).
|
|
|
|
* Now consider opening "sys". A vop_lookup would be
|
|
|
|
* done on the root null-node. This operation would bypass through
|
1995-05-30 08:16:23 +00:00
|
|
|
* to the lower layer which would return a vnode representing
|
1994-05-24 10:09:53 +00:00
|
|
|
* the UFS "sys". Null_bypass then builds a null-node
|
|
|
|
* aliasing the UFS "sys" and returns this to the caller.
|
|
|
|
* Later operations on the null-node "sys" will repeat this
|
|
|
|
* process when constructing other vnode stacks.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* CREATING OTHER FILE SYSTEM LAYERS
|
|
|
|
*
|
2002-05-16 21:28:32 +00:00
|
|
|
* One of the easiest ways to construct new filesystem layers is to make
|
1994-05-24 10:09:53 +00:00
|
|
|
* a copy of the null layer, rename all files and variables, and
|
|
|
|
* then begin modifing the copy. Sed can be used to easily rename
|
|
|
|
* all variables.
|
|
|
|
*
|
1995-05-30 08:16:23 +00:00
|
|
|
* The umap layer is an example of a layer descended from the
|
1994-05-24 10:09:53 +00:00
|
|
|
* null layer.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* INVOKING OPERATIONS ON LOWER LAYERS
|
|
|
|
*
|
1995-05-30 08:16:23 +00:00
|
|
|
* There are two techniques to invoke operations on a lower layer
|
1994-05-24 10:09:53 +00:00
|
|
|
* when the operation cannot be completely bypassed. Each method
|
|
|
|
* is appropriate in different situations. In both cases,
|
|
|
|
* it is the responsibility of the aliasing layer to make
|
|
|
|
* the operation arguments "correct" for the lower layer
|
2002-12-30 21:18:15 +00:00
|
|
|
* by mapping a vnode arguments to the lower layer.
|
1994-05-24 10:09:53 +00:00
|
|
|
*
|
|
|
|
* The first approach is to call the aliasing layer's bypass routine.
|
|
|
|
* This method is most suitable when you wish to invoke the operation
|
1997-06-26 17:12:59 +00:00
|
|
|
* currently being handled on the lower layer. It has the advantage
|
1994-05-24 10:09:53 +00:00
|
|
|
* that the bypass routine already must do argument mapping.
|
|
|
|
* An example of this is null_getattrs in the null layer.
|
|
|
|
*
|
1997-06-26 17:12:59 +00:00
|
|
|
* A second approach is to directly invoke vnode operations on
|
1994-05-24 10:09:53 +00:00
|
|
|
* the lower layer with the VOP_OPERATIONNAME interface.
|
|
|
|
* The advantage of this method is that it is easy to invoke
|
|
|
|
* arbitrary operations on the lower layer. The disadvantage
|
1997-06-26 17:12:59 +00:00
|
|
|
* is that vnode arguments must be manualy mapped.
|
1994-05-24 10:09:53 +00:00
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
2000-09-25 15:38:32 +00:00
|
|
|
#include <sys/conf.h>
|
2001-05-01 08:13:21 +00:00
|
|
|
#include <sys/kernel.h>
|
|
|
|
#include <sys/lock.h>
|
|
|
|
#include <sys/malloc.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/mount.h>
|
2001-05-01 08:13:21 +00:00
|
|
|
#include <sys/mutex.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/namei.h>
|
2001-05-01 08:13:21 +00:00
|
|
|
#include <sys/sysctl.h>
|
|
|
|
#include <sys/vnode.h>
|
|
|
|
|
2001-05-23 09:42:29 +00:00
|
|
|
#include <fs/nullfs/null.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
#include <vm/vm.h>
|
|
|
|
#include <vm/vm_extern.h>
|
|
|
|
#include <vm/vm_object.h>
|
|
|
|
#include <vm/vnode_pager.h>
|
|
|
|
|
1995-12-11 09:24:58 +00:00
|
|
|
static int null_bug_bypass = 0; /* for debugging: enables bypass printf'ing */
|
|
|
|
SYSCTL_INT(_debug, OID_AUTO, nullfs_bug_bypass, CTLFLAG_RW,
|
|
|
|
&null_bug_bypass, 0, "");
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This is the 10-Apr-92 bypass routine.
|
|
|
|
* This version has been optimized for speed, throwing away some
|
|
|
|
* safety checks. It should still always work, but it's not as
|
|
|
|
* robust to programmer errors.
|
|
|
|
*
|
|
|
|
* In general, we map all vnodes going down and unmap them on the way back.
|
|
|
|
* As an exception to this, vnodes can be marked "unmapped" by setting
|
|
|
|
* the Nth bit in operation's vdesc_flags.
|
|
|
|
*
|
|
|
|
* Also, some BSD vnode operations have the side effect of vrele'ing
|
|
|
|
* their arguments. With stacking, the reference counts are held
|
|
|
|
* by the upper node, not the lower one, so we must handle these
|
|
|
|
* side-effects here. This is not of concern in Sun-derived systems
|
|
|
|
* since there are no such side-effects.
|
|
|
|
*
|
|
|
|
* This makes the following assumptions:
|
|
|
|
* - only one returned vpp
|
|
|
|
* - no INOUT vpp's (Sun's vop_open has one of these)
|
|
|
|
* - the vnode operation vector of the first vnode should be used
|
|
|
|
* to determine what implementation of the op should be invoked
|
|
|
|
* - all mapped vnodes are of our vnode-type (NEEDSWORK:
|
|
|
|
* problems on rmdir'ing mount points and renaming?)
|
1995-05-30 08:16:23 +00:00
|
|
|
*/
|
1997-02-10 02:22:35 +00:00
|
|
|
int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_bypass(struct vop_generic_args *ap)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2005-01-24 12:37:51 +00:00
|
|
|
struct vnode **this_vp_p;
|
1994-05-24 10:09:53 +00:00
|
|
|
int error;
|
|
|
|
struct vnode *old_vps[VDESC_MAX_VPS];
|
|
|
|
struct vnode **vps_p[VDESC_MAX_VPS];
|
|
|
|
struct vnode ***vppp;
|
|
|
|
struct vnodeop_desc *descp = ap->a_desc;
|
|
|
|
int reles, i;
|
|
|
|
|
|
|
|
if (null_bug_bypass)
|
|
|
|
printf ("null_bypass: %s\n", descp->vdesc_name);
|
|
|
|
|
1999-08-30 07:08:04 +00:00
|
|
|
#ifdef DIAGNOSTIC
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* We require at least one vp.
|
|
|
|
*/
|
|
|
|
if (descp->vdesc_vp_offsets == NULL ||
|
|
|
|
descp->vdesc_vp_offsets[0] == VDESC_NO_OFFSET)
|
1999-08-30 07:08:04 +00:00
|
|
|
panic ("null_bypass: no vp's in map");
|
1994-05-24 10:09:53 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Map the vnodes going in.
|
|
|
|
* Later, we'll invoke the operation based on
|
|
|
|
* the first mapped vnode's operation vector.
|
|
|
|
*/
|
|
|
|
reles = descp->vdesc_flags;
|
|
|
|
for (i = 0; i < VDESC_MAX_VPS; reles >>= 1, i++) {
|
|
|
|
if (descp->vdesc_vp_offsets[i] == VDESC_NO_OFFSET)
|
|
|
|
break; /* bail out at end of list */
|
1995-05-30 08:16:23 +00:00
|
|
|
vps_p[i] = this_vp_p =
|
1994-05-24 10:09:53 +00:00
|
|
|
VOPARG_OFFSETTO(struct vnode**,descp->vdesc_vp_offsets[i],ap);
|
|
|
|
/*
|
|
|
|
* We're not guaranteed that any but the first vnode
|
|
|
|
* are of our type. Check for and don't map any
|
|
|
|
* that aren't. (We must always map first vp or vclean fails.)
|
|
|
|
*/
|
1997-04-17 11:17:30 +00:00
|
|
|
if (i && (*this_vp_p == NULLVP ||
|
2004-12-01 23:16:38 +00:00
|
|
|
(*this_vp_p)->v_op != &null_vnodeops)) {
|
1997-04-17 11:17:30 +00:00
|
|
|
old_vps[i] = NULLVP;
|
1994-05-24 10:09:53 +00:00
|
|
|
} else {
|
|
|
|
old_vps[i] = *this_vp_p;
|
|
|
|
*(vps_p[i]) = NULLVPTOLOWERVP(*this_vp_p);
|
|
|
|
/*
|
|
|
|
* XXX - Several operations have the side effect
|
|
|
|
* of vrele'ing their vp's. We must account for
|
|
|
|
* that. (This should go away in the future.)
|
|
|
|
*/
|
2000-09-25 15:38:32 +00:00
|
|
|
if (reles & VDESC_VP0_WILLRELE)
|
1994-05-24 10:09:53 +00:00
|
|
|
VREF(*this_vp_p);
|
|
|
|
}
|
1995-05-30 08:16:23 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Call the operation on the lower layer
|
|
|
|
* with the modified argument structure.
|
|
|
|
*/
|
2000-09-25 15:38:32 +00:00
|
|
|
if (vps_p[0] && *vps_p[0])
|
2005-01-13 07:53:01 +00:00
|
|
|
error = VCALL(ap);
|
2000-09-25 15:38:32 +00:00
|
|
|
else {
|
|
|
|
printf("null_bypass: no map for %s\n", descp->vdesc_name);
|
|
|
|
error = EINVAL;
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Maintain the illusion of call-by-value
|
|
|
|
* by restoring vnodes in the argument structure
|
|
|
|
* to their original value.
|
|
|
|
*/
|
|
|
|
reles = descp->vdesc_flags;
|
|
|
|
for (i = 0; i < VDESC_MAX_VPS; reles >>= 1, i++) {
|
|
|
|
if (descp->vdesc_vp_offsets[i] == VDESC_NO_OFFSET)
|
|
|
|
break; /* bail out at end of list */
|
|
|
|
if (old_vps[i]) {
|
|
|
|
*(vps_p[i]) = old_vps[i];
|
2000-09-25 15:38:32 +00:00
|
|
|
#if 0
|
|
|
|
if (reles & VDESC_VP0_WILLUNLOCK)
|
2008-01-13 14:44:15 +00:00
|
|
|
VOP_UNLOCK(*(vps_p[i]), 0);
|
2000-09-25 15:38:32 +00:00
|
|
|
#endif
|
|
|
|
if (reles & VDESC_VP0_WILLRELE)
|
1994-05-24 10:09:53 +00:00
|
|
|
vrele(*(vps_p[i]));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Map the possible out-going vpp
|
|
|
|
* (Assumes that the lower layer always returns
|
|
|
|
* a VREF'ed vpp unless it gets an error.)
|
|
|
|
*/
|
|
|
|
if (descp->vdesc_vpp_offset != VDESC_NO_OFFSET &&
|
|
|
|
!(descp->vdesc_flags & VDESC_NOMAP_VPP) &&
|
|
|
|
!error) {
|
|
|
|
/*
|
|
|
|
* XXX - even though some ops have vpp returned vp's,
|
|
|
|
* several ops actually vrele this before returning.
|
|
|
|
* We must avoid these ops.
|
|
|
|
* (This should go away when these ops are regularized.)
|
|
|
|
*/
|
|
|
|
if (descp->vdesc_flags & VDESC_VPP_WILLRELE)
|
|
|
|
goto out;
|
|
|
|
vppp = VOPARG_OFFSETTO(struct vnode***,
|
|
|
|
descp->vdesc_vpp_offset,ap);
|
1997-09-18 18:33:23 +00:00
|
|
|
if (*vppp)
|
2002-06-13 21:49:09 +00:00
|
|
|
error = null_nodeget(old_vps[0]->v_mount, **vppp, *vppp);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
1997-02-10 02:22:35 +00:00
|
|
|
/*
|
|
|
|
* We have to carry on the locking protocol on the null layer vnodes
|
|
|
|
* as we progress through the tree. We also have to enforce read-only
|
|
|
|
* if this layer is mounted read-only.
|
|
|
|
*/
|
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_lookup(struct vop_lookup_args *ap)
|
1997-02-10 02:22:35 +00:00
|
|
|
{
|
|
|
|
struct componentname *cnp = ap->a_cnp;
|
2000-09-25 15:38:32 +00:00
|
|
|
struct vnode *dvp = ap->a_dvp;
|
1997-02-10 02:22:35 +00:00
|
|
|
int flags = cnp->cn_flags;
|
2000-09-25 15:38:32 +00:00
|
|
|
struct vnode *vp, *ldvp, *lvp;
|
1997-02-10 02:22:35 +00:00
|
|
|
int error;
|
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
if ((flags & ISLASTCN) && (dvp->v_mount->mnt_flag & MNT_RDONLY) &&
|
1997-02-10 02:22:35 +00:00
|
|
|
(cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME))
|
|
|
|
return (EROFS);
|
2000-09-25 15:38:32 +00:00
|
|
|
/*
|
|
|
|
* Although it is possible to call null_bypass(), we'll do
|
|
|
|
* a direct call to reduce overhead
|
|
|
|
*/
|
|
|
|
ldvp = NULLVPTOLOWERVP(dvp);
|
|
|
|
vp = lvp = NULL;
|
|
|
|
error = VOP_LOOKUP(ldvp, &lvp, cnp);
|
1997-02-10 02:22:35 +00:00
|
|
|
if (error == EJUSTRETURN && (flags & ISLASTCN) &&
|
2000-09-25 15:38:32 +00:00
|
|
|
(dvp->v_mount->mnt_flag & MNT_RDONLY) &&
|
1997-02-10 02:22:35 +00:00
|
|
|
(cnp->cn_nameiop == CREATE || cnp->cn_nameiop == RENAME))
|
|
|
|
error = EROFS;
|
2000-09-25 15:38:32 +00:00
|
|
|
|
|
|
|
if ((error == 0 || error == EJUSTRETURN) && lvp != NULL) {
|
|
|
|
if (ldvp == lvp) {
|
|
|
|
*ap->a_vpp = dvp;
|
|
|
|
VREF(dvp);
|
|
|
|
vrele(lvp);
|
|
|
|
} else {
|
2002-06-13 21:49:09 +00:00
|
|
|
error = null_nodeget(dvp->v_mount, lvp, &vp);
|
|
|
|
if (error) {
|
|
|
|
/* XXX Cleanup needed... */
|
|
|
|
panic("null_nodeget failed");
|
|
|
|
}
|
|
|
|
*ap->a_vpp = vp;
|
2000-09-25 15:38:32 +00:00
|
|
|
}
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
2005-01-24 22:56:24 +00:00
|
|
|
static int
|
|
|
|
null_open(struct vop_open_args *ap)
|
|
|
|
{
|
|
|
|
int retval;
|
|
|
|
struct vnode *vp, *ldvp;
|
|
|
|
|
|
|
|
vp = ap->a_vp;
|
|
|
|
ldvp = NULLVPTOLOWERVP(vp);
|
|
|
|
retval = null_bypass(&ap->a_gen);
|
|
|
|
if (retval == 0)
|
|
|
|
vp->v_object = ldvp->v_object;
|
|
|
|
return (retval);
|
|
|
|
}
|
|
|
|
|
1997-02-10 02:22:35 +00:00
|
|
|
/*
|
|
|
|
* Setattr call. Disallow write attempts if the layer is mounted read-only.
|
|
|
|
*/
|
2002-10-16 08:00:32 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_setattr(struct vop_setattr_args *ap)
|
1997-02-10 02:22:35 +00:00
|
|
|
{
|
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
struct vattr *vap = ap->a_vap;
|
|
|
|
|
|
|
|
if ((vap->va_flags != VNOVAL || vap->va_uid != (uid_t)VNOVAL ||
|
1997-02-12 14:55:01 +00:00
|
|
|
vap->va_gid != (gid_t)VNOVAL || vap->va_atime.tv_sec != VNOVAL ||
|
|
|
|
vap->va_mtime.tv_sec != VNOVAL || vap->va_mode != (mode_t)VNOVAL) &&
|
1997-02-10 02:22:35 +00:00
|
|
|
(vp->v_mount->mnt_flag & MNT_RDONLY))
|
|
|
|
return (EROFS);
|
|
|
|
if (vap->va_size != VNOVAL) {
|
|
|
|
switch (vp->v_type) {
|
|
|
|
case VDIR:
|
|
|
|
return (EISDIR);
|
|
|
|
case VCHR:
|
|
|
|
case VBLK:
|
|
|
|
case VSOCK:
|
|
|
|
case VFIFO:
|
1998-06-10 06:34:57 +00:00
|
|
|
if (vap->va_flags != VNOVAL)
|
|
|
|
return (EOPNOTSUPP);
|
1997-02-10 02:22:35 +00:00
|
|
|
return (0);
|
|
|
|
case VREG:
|
|
|
|
case VLNK:
|
|
|
|
default:
|
|
|
|
/*
|
|
|
|
* Disallow write attempts if the filesystem is
|
|
|
|
* mounted read-only.
|
|
|
|
*/
|
|
|
|
if (vp->v_mount->mnt_flag & MNT_RDONLY)
|
|
|
|
return (EROFS);
|
|
|
|
}
|
|
|
|
}
|
2000-09-25 15:38:32 +00:00
|
|
|
|
1997-02-12 18:06:08 +00:00
|
|
|
return (null_bypass((struct vop_generic_args *)ap));
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We handle getattr only to change the fsid.
|
|
|
|
*/
|
1995-12-11 09:24:58 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_getattr(struct vop_getattr_args *ap)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
|
|
|
int error;
|
1997-02-10 02:22:35 +00:00
|
|
|
|
1999-01-28 00:57:57 +00:00
|
|
|
if ((error = null_bypass((struct vop_generic_args *)ap)) != 0)
|
1994-05-24 10:09:53 +00:00
|
|
|
return (error);
|
2000-09-05 09:02:07 +00:00
|
|
|
|
|
|
|
ap->a_vap->va_fsid = ap->a_vp->v_mount->mnt_stat.f_fsid.val[0];
|
1994-05-24 10:09:53 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
/*
|
|
|
|
* Handle to disallow write access if mounted read-only.
|
|
|
|
*/
|
1995-12-11 09:24:58 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_access(struct vop_access_args *ap)
|
1997-02-10 02:22:35 +00:00
|
|
|
{
|
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
mode_t mode = ap->a_mode;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Disallow write attempts on read-only layers;
|
|
|
|
* unless the file is a socket, fifo, or a block or
|
2002-05-16 21:28:32 +00:00
|
|
|
* character device resident on the filesystem.
|
1997-02-10 02:22:35 +00:00
|
|
|
*/
|
|
|
|
if (mode & VWRITE) {
|
|
|
|
switch (vp->v_type) {
|
|
|
|
case VDIR:
|
|
|
|
case VLNK:
|
|
|
|
case VREG:
|
|
|
|
if (vp->v_mount->mnt_flag & MNT_RDONLY)
|
|
|
|
return (EROFS);
|
|
|
|
break;
|
1999-01-27 22:42:27 +00:00
|
|
|
default:
|
|
|
|
break;
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
|
|
|
}
|
1997-02-12 18:06:08 +00:00
|
|
|
return (null_bypass((struct vop_generic_args *)ap));
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
|
|
|
|
2000-09-05 09:02:07 +00:00
|
|
|
/*
|
|
|
|
* We handle this to eliminate null FS to lower FS
|
|
|
|
* file moving. Don't know why we don't allow this,
|
|
|
|
* possibly we should.
|
|
|
|
*/
|
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_rename(struct vop_rename_args *ap)
|
2000-09-05 09:02:07 +00:00
|
|
|
{
|
|
|
|
struct vnode *tdvp = ap->a_tdvp;
|
|
|
|
struct vnode *fvp = ap->a_fvp;
|
|
|
|
struct vnode *fdvp = ap->a_fdvp;
|
|
|
|
struct vnode *tvp = ap->a_tvp;
|
|
|
|
|
|
|
|
/* Check for cross-device rename. */
|
|
|
|
if ((fvp->v_mount != tdvp->v_mount) ||
|
|
|
|
(tvp && (fvp->v_mount != tvp->v_mount))) {
|
|
|
|
if (tdvp == tvp)
|
|
|
|
vrele(tdvp);
|
|
|
|
else
|
|
|
|
vput(tdvp);
|
|
|
|
if (tvp)
|
|
|
|
vput(tvp);
|
|
|
|
vrele(fdvp);
|
|
|
|
vrele(fvp);
|
|
|
|
return (EXDEV);
|
|
|
|
}
|
|
|
|
|
|
|
|
return (null_bypass((struct vop_generic_args *)ap));
|
|
|
|
}
|
|
|
|
|
1997-02-10 02:22:35 +00:00
|
|
|
/*
|
|
|
|
* We need to process our own vnode lock and then clear the
|
|
|
|
* interlock flag as it applies only to our vnode, not the
|
|
|
|
* vnodes below us on the stack.
|
|
|
|
*/
|
1997-02-12 14:55:01 +00:00
|
|
|
static int
|
2007-05-18 13:02:13 +00:00
|
|
|
null_lock(struct vop_lock1_args *ap)
|
1997-02-10 02:22:35 +00:00
|
|
|
{
|
2000-09-25 15:38:32 +00:00
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
int flags = ap->a_flags;
|
2005-03-15 13:49:33 +00:00
|
|
|
struct null_node *nn;
|
2000-09-25 15:38:32 +00:00
|
|
|
struct vnode *lvp;
|
|
|
|
int error;
|
1997-02-10 02:22:35 +00:00
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
|
2005-03-13 12:18:25 +00:00
|
|
|
if ((flags & LK_INTERLOCK) == 0) {
|
|
|
|
VI_LOCK(vp);
|
2005-03-15 13:49:33 +00:00
|
|
|
ap->a_flags = flags |= LK_INTERLOCK;
|
2005-03-13 12:18:25 +00:00
|
|
|
}
|
2005-03-15 13:49:33 +00:00
|
|
|
nn = VTONULL(vp);
|
|
|
|
/*
|
|
|
|
* If we're still active we must ask the lower layer to
|
|
|
|
* lock as ffs has special lock considerations in it's
|
|
|
|
* vop lock.
|
|
|
|
*/
|
|
|
|
if (nn != NULL && (lvp = NULLVPTOLOWERVP(vp)) != NULL) {
|
2005-04-22 22:42:44 +00:00
|
|
|
VI_LOCK_FLAGS(lvp, MTX_DUPOK);
|
2003-06-17 08:52:45 +00:00
|
|
|
VI_UNLOCK(vp);
|
2000-09-25 15:38:32 +00:00
|
|
|
/*
|
2005-03-15 13:49:33 +00:00
|
|
|
* We have to hold the vnode here to solve a potential
|
|
|
|
* reclaim race. If we're forcibly vgone'd while we
|
|
|
|
* still have refs, a thread could be sleeping inside
|
|
|
|
* the lowervp's vop_lock routine. When we vgone we will
|
|
|
|
* drop our last ref to the lowervp, which would allow it
|
|
|
|
* to be reclaimed. The lowervp could then be recycled,
|
|
|
|
* in which case it is not legal to be sleeping in it's VOP.
|
|
|
|
* We prevent it from being recycled by holding the vnode
|
|
|
|
* here.
|
2000-09-25 15:38:32 +00:00
|
|
|
*/
|
2005-03-15 13:49:33 +00:00
|
|
|
vholdl(lvp);
|
2008-01-13 14:44:15 +00:00
|
|
|
error = VOP_LOCK(lvp, flags);
|
2005-09-15 19:21:26 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We might have slept to get the lock and someone might have
|
|
|
|
* clean our vnode already, switching vnode lock from one in
|
|
|
|
* lowervp to v_lock in our own vnode structure. Handle this
|
|
|
|
* case by reacquiring correct lock in requested mode.
|
|
|
|
*/
|
|
|
|
if (VTONULL(vp) == NULL && error == 0) {
|
|
|
|
ap->a_flags &= ~(LK_TYPE_MASK | LK_INTERLOCK);
|
|
|
|
switch (flags & LK_TYPE_MASK) {
|
|
|
|
case LK_SHARED:
|
|
|
|
ap->a_flags |= LK_SHARED;
|
|
|
|
break;
|
|
|
|
case LK_UPGRADE:
|
|
|
|
case LK_EXCLUSIVE:
|
|
|
|
ap->a_flags |= LK_EXCLUSIVE;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
panic("Unsupported lock request %d\n",
|
|
|
|
ap->a_flags);
|
|
|
|
}
|
2008-01-13 14:44:15 +00:00
|
|
|
VOP_UNLOCK(lvp, 0);
|
2005-09-15 19:21:26 +00:00
|
|
|
error = vop_stdlock(ap);
|
|
|
|
}
|
2005-03-15 13:49:33 +00:00
|
|
|
vdrop(lvp);
|
|
|
|
} else
|
|
|
|
error = vop_stdlock(ap);
|
|
|
|
|
|
|
|
return (error);
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to process our own vnode unlock and then clear the
|
|
|
|
* interlock flag as it applies only to our vnode, not the
|
|
|
|
* vnodes below us on the stack.
|
|
|
|
*/
|
1997-02-12 14:55:01 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_unlock(struct vop_unlock_args *ap)
|
1997-02-10 02:22:35 +00:00
|
|
|
{
|
2000-09-25 15:38:32 +00:00
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
int flags = ap->a_flags;
|
2007-10-14 13:57:11 +00:00
|
|
|
int mtxlkflag = 0;
|
2005-03-15 13:49:33 +00:00
|
|
|
struct null_node *nn;
|
2000-10-03 04:25:53 +00:00
|
|
|
struct vnode *lvp;
|
2005-03-15 13:49:33 +00:00
|
|
|
int error;
|
2000-09-25 15:38:32 +00:00
|
|
|
|
2007-10-14 13:57:11 +00:00
|
|
|
if ((flags & LK_INTERLOCK) != 0)
|
|
|
|
mtxlkflag = 1;
|
|
|
|
else if (mtx_owned(VI_MTX(vp)) == 0) {
|
|
|
|
VI_LOCK(vp);
|
|
|
|
mtxlkflag = 2;
|
2000-09-25 15:38:32 +00:00
|
|
|
}
|
2005-03-15 13:49:33 +00:00
|
|
|
nn = VTONULL(vp);
|
2007-10-14 13:57:11 +00:00
|
|
|
if (nn != NULL && (lvp = NULLVPTOLOWERVP(vp)) != NULL) {
|
|
|
|
VI_LOCK_FLAGS(lvp, MTX_DUPOK);
|
|
|
|
flags |= LK_INTERLOCK;
|
|
|
|
vholdl(lvp);
|
|
|
|
VI_UNLOCK(vp);
|
2008-01-13 14:44:15 +00:00
|
|
|
error = VOP_UNLOCK(lvp, flags);
|
2007-10-14 13:57:11 +00:00
|
|
|
vdrop(lvp);
|
|
|
|
if (mtxlkflag == 0)
|
|
|
|
VI_LOCK(vp);
|
|
|
|
} else {
|
|
|
|
if (mtxlkflag == 2)
|
|
|
|
VI_UNLOCK(vp);
|
2005-03-15 13:49:33 +00:00
|
|
|
error = vop_stdunlock(ap);
|
2007-10-14 13:57:11 +00:00
|
|
|
}
|
2005-03-15 13:49:33 +00:00
|
|
|
|
|
|
|
return (error);
|
2000-09-25 15:38:32 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_islocked(struct vop_islocked_args *ap)
|
2000-09-25 15:38:32 +00:00
|
|
|
{
|
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
|
2008-02-25 18:45:57 +00:00
|
|
|
return (lockstatus(vp->v_vnlock));
|
1997-02-10 02:22:35 +00:00
|
|
|
}
|
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
/*
|
|
|
|
* There is no way to tell that someone issued remove/rmdir operation
|
2008-09-11 20:15:34 +00:00
|
|
|
* on the underlying filesystem. For now we just have to release lowervp
|
2000-09-25 15:38:32 +00:00
|
|
|
* as soon as possible.
|
2002-06-13 21:49:09 +00:00
|
|
|
*
|
|
|
|
* Note, we can't release any resources nor remove vnode from hash before
|
|
|
|
* appropriate VXLOCK stuff is is done because other process can find this
|
|
|
|
* vnode in hash during inactivation and may be sitting in vget() and waiting
|
|
|
|
* for null_inactive to unlock vnode. Thus we will do all those in VOP_RECLAIM.
|
2000-09-25 15:38:32 +00:00
|
|
|
*/
|
1997-02-12 14:55:01 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_inactive(struct vop_inactive_args *ap)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
1997-10-21 21:01:34 +00:00
|
|
|
struct vnode *vp = ap->a_vp;
|
2002-06-13 17:30:40 +00:00
|
|
|
struct thread *td = ap->a_td;
|
|
|
|
|
2005-02-07 09:23:34 +00:00
|
|
|
vp->v_object = NULL;
|
2000-09-25 15:38:32 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2002-03-18 05:39:04 +00:00
|
|
|
* If this is the last reference, then free up the vnode
|
|
|
|
* so as not to tie up the lower vnodes.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2005-01-28 13:08:21 +00:00
|
|
|
vrecycle(vp, td);
|
2002-06-13 17:30:40 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2000-09-25 15:38:32 +00:00
|
|
|
/*
|
2002-06-13 21:49:09 +00:00
|
|
|
* Now, the VXLOCK is in force and we're free to destroy the null vnode.
|
2000-09-25 15:38:32 +00:00
|
|
|
*/
|
1995-12-11 09:24:58 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_reclaim(struct vop_reclaim_args *ap)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
|
|
|
struct vnode *vp = ap->a_vp;
|
2002-03-18 05:39:04 +00:00
|
|
|
struct null_node *xp = VTONULL(vp);
|
|
|
|
struct vnode *lowervp = xp->null_lowervp;
|
2005-03-15 11:28:45 +00:00
|
|
|
struct lock *vnlock;
|
2002-03-18 05:39:04 +00:00
|
|
|
|
2006-02-22 06:17:31 +00:00
|
|
|
if (lowervp)
|
|
|
|
null_hashrem(xp);
|
2005-03-17 12:00:05 +00:00
|
|
|
/*
|
|
|
|
* Use the interlock to protect the clearing of v_data to
|
|
|
|
* prevent faults in null_lock().
|
|
|
|
*/
|
|
|
|
VI_LOCK(vp);
|
|
|
|
vp->v_data = NULL;
|
2006-02-22 06:17:31 +00:00
|
|
|
vp->v_object = NULL;
|
2005-09-15 19:21:26 +00:00
|
|
|
vnlock = vp->v_vnlock;
|
|
|
|
vp->v_vnlock = &vp->v_lock;
|
2005-09-02 15:49:55 +00:00
|
|
|
if (lowervp) {
|
2008-01-24 12:34:30 +00:00
|
|
|
lockmgr(vp->v_vnlock, LK_EXCLUSIVE | LK_INTERLOCK, VI_MTX(vp));
|
2005-09-02 15:49:55 +00:00
|
|
|
vput(lowervp);
|
|
|
|
} else
|
2008-09-11 20:15:34 +00:00
|
|
|
panic("null_reclaim: reclaiming a node with no lowervp");
|
2008-10-23 15:53:51 +00:00
|
|
|
free(xp, M_NULLFSNODE);
|
2000-09-25 15:38:32 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
1995-12-11 09:24:58 +00:00
|
|
|
static int
|
2005-01-24 11:49:41 +00:00
|
|
|
null_print(struct vop_print_args *ap)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2005-01-24 12:37:51 +00:00
|
|
|
struct vnode *vp = ap->a_vp;
|
2006-02-22 06:17:31 +00:00
|
|
|
|
2003-03-03 19:15:40 +00:00
|
|
|
printf("\tvp=%p, lowervp=%p\n", vp, NULLVPTOLOWERVP(vp));
|
1994-05-24 10:09:53 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2006-03-12 04:58:18 +00:00
|
|
|
/* ARGSUSED */
|
|
|
|
static int
|
|
|
|
null_getwritemount(struct vop_getwritemount_args *ap)
|
|
|
|
{
|
|
|
|
struct null_node *xp;
|
|
|
|
struct vnode *lowervp;
|
|
|
|
struct vnode *vp;
|
|
|
|
|
|
|
|
vp = ap->a_vp;
|
|
|
|
VI_LOCK(vp);
|
|
|
|
xp = VTONULL(vp);
|
|
|
|
if (xp && (lowervp = xp->null_lowervp)) {
|
|
|
|
VI_LOCK_FLAGS(lowervp, MTX_DUPOK);
|
|
|
|
VI_UNLOCK(vp);
|
|
|
|
vholdl(lowervp);
|
|
|
|
VI_UNLOCK(lowervp);
|
|
|
|
VOP_GETWRITEMOUNT(lowervp, ap->a_mpp);
|
|
|
|
vdrop(lowervp);
|
|
|
|
} else {
|
|
|
|
VI_UNLOCK(vp);
|
|
|
|
*(ap->a_mpp) = NULL;
|
|
|
|
}
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2007-02-15 22:08:35 +00:00
|
|
|
static int
|
|
|
|
null_vptofh(struct vop_vptofh_args *ap)
|
|
|
|
{
|
|
|
|
struct vnode *lvp;
|
|
|
|
|
|
|
|
lvp = NULLVPTOLOWERVP(ap->a_vp);
|
|
|
|
return VOP_VPTOFH(lvp, ap->a_fhp);
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Global vfs data structures
|
|
|
|
*/
|
2004-12-01 23:16:38 +00:00
|
|
|
struct vop_vector null_vnodeops = {
|
|
|
|
.vop_bypass = null_bypass,
|
|
|
|
.vop_access = null_access,
|
|
|
|
.vop_bmap = VOP_EOPNOTSUPP,
|
|
|
|
.vop_getattr = null_getattr,
|
2006-03-12 04:58:18 +00:00
|
|
|
.vop_getwritemount = null_getwritemount,
|
2004-12-01 23:16:38 +00:00
|
|
|
.vop_inactive = null_inactive,
|
|
|
|
.vop_islocked = null_islocked,
|
2007-05-18 13:02:13 +00:00
|
|
|
.vop_lock1 = null_lock,
|
2004-12-01 23:16:38 +00:00
|
|
|
.vop_lookup = null_lookup,
|
2005-01-24 22:56:24 +00:00
|
|
|
.vop_open = null_open,
|
2004-12-01 23:16:38 +00:00
|
|
|
.vop_print = null_print,
|
|
|
|
.vop_reclaim = null_reclaim,
|
|
|
|
.vop_rename = null_rename,
|
|
|
|
.vop_setattr = null_setattr,
|
|
|
|
.vop_strategy = VOP_EOPNOTSUPP,
|
|
|
|
.vop_unlock = null_unlock,
|
2007-02-15 22:08:35 +00:00
|
|
|
.vop_vptofh = null_vptofh,
|
1994-05-24 10:09:53 +00:00
|
|
|
};
|