freebsd-dev/sys/kern/vfs_default.c

807 lines
17 KiB
C
Raw Normal View History

/*-
* Copyright (c) 1989, 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.
*
* Source: * @(#)i405_init.c 2.10 92/04/27 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.
*/
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bio.h>
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/kernel.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/mutex.h>
#include <sys/unistd.h>
#include <sys/vnode.h>
#include <sys/poll.h>
#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_page.h>
#include <vm/vm_pager.h>
#include <vm/vnode_pager.h>
2002-03-19 21:25:46 +00:00
static int vop_nolookup(struct vop_lookup_args *);
static int vop_nostrategy(struct vop_strategy_args *);
/*
* This vnode table stores what we want to do if the filesystem doesn't
* implement a particular VOP.
*
* If there is no specific entry here, we will return EOPNOTSUPP.
*
*/
struct vop_vector default_vnodeops = {
.vop_default = NULL,
.vop_bypass = VOP_EOPNOTSUPP,
.vop_advlock = VOP_EINVAL,
.vop_bmap = vop_stdbmap,
.vop_close = VOP_NULL,
.vop_createvobject = vop_stdcreatevobject,
.vop_destroyvobject = vop_stddestroyvobject,
.vop_fsync = VOP_NULL,
.vop_getpages = vop_stdgetpages,
.vop_getvobject = vop_stdgetvobject,
.vop_getwritemount = vop_stdgetwritemount,
.vop_inactive = vop_stdinactive,
.vop_ioctl = VOP_ENOTTY,
.vop_islocked = vop_stdislocked,
.vop_lease = VOP_NULL,
.vop_lock = vop_stdlock,
.vop_lookup = vop_nolookup,
.vop_open = VOP_NULL,
.vop_pathconf = VOP_EINVAL,
.vop_poll = vop_nopoll,
.vop_putpages = vop_stdputpages,
.vop_readlink = VOP_EINVAL,
.vop_revoke = VOP_PANIC,
.vop_strategy = vop_nostrategy,
.vop_unlock = vop_stdunlock,
};
/*
* Series of placeholder functions for various error returns for
* VOPs.
*/
int
vop_eopnotsupp(struct vop_generic_args *ap)
{
/*
printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name);
*/
return (EOPNOTSUPP);
}
int
vop_ebadf(struct vop_generic_args *ap)
{
return (EBADF);
}
int
vop_enotty(struct vop_generic_args *ap)
{
return (ENOTTY);
}
int
vop_einval(struct vop_generic_args *ap)
{
return (EINVAL);
}
int
vop_null(struct vop_generic_args *ap)
{
return (0);
}
/*
* Helper function to panic on some bad VOPs in some filesystems.
*/
int
vop_panic(struct vop_generic_args *ap)
{
panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name);
}
/*
* vop_std<something> and vop_no<something> are default functions for use by
* filesystems that need the "default reasonable" implementation for a
* particular operation.
*
* The documentation for the operations they implement exists (if it exists)
* in the VOP_<SOMETHING>(9) manpage (all uppercase).
*/
/*
* Default vop for filesystems that do not support name lookup
*/
static int
vop_nolookup(ap)
struct vop_lookup_args /* {
struct vnode *a_dvp;
struct vnode **a_vpp;
struct componentname *a_cnp;
} */ *ap;
{
*ap->a_vpp = NULL;
return (ENOTDIR);
}
The VFS/BIO subsystem contained a number of hacks in order to optimize piecemeal, middle-of-file writes for NFS. These hacks have caused no end of trouble, especially when combined with mmap(). I've removed them. Instead, NFS will issue a read-before-write to fully instantiate the struct buf containing the write. NFS does, however, optimize piecemeal appends to files. For most common file operations, you will not notice the difference. The sole remaining fragment in the VFS/BIO system is b_dirtyoff/end, which NFS uses to avoid cache coherency issues with read-merge-write style operations. NFS also optimizes the write-covers-entire-buffer case by avoiding the read-before-write. There is quite a bit of room for further optimization in these areas. The VM system marks pages fully-valid (AKA vm_page_t->valid = VM_PAGE_BITS_ALL) in several places, most noteably in vm_fault. This is not correct operation. The vm_pager_get_pages() code is now responsible for marking VM pages all-valid. A number of VM helper routines have been added to aid in zeroing-out the invalid portions of a VM page prior to the page being marked all-valid. This operation is necessary to properly support mmap(). The zeroing occurs most often when dealing with file-EOF situations. Several bugs have been fixed in the NFS subsystem, including bits handling file and directory EOF situations and buf->b_flags consistancy issues relating to clearing B_ERROR & B_INVAL, and handling B_DONE. getblk() and allocbuf() have been rewritten. B_CACHE operation is now formally defined in comments and more straightforward in implementation. B_CACHE for VMIO buffers is based on the validity of the backing store. B_CACHE for non-VMIO buffers is based simply on whether the buffer is B_INVAL or not (B_CACHE set if B_INVAL clear, and vise-versa). biodone() is now responsible for setting B_CACHE when a successful read completes. B_CACHE is also set when a bdwrite() is initiated and when a bwrite() is initiated. VFS VOP_BWRITE routines (there are only two - nfs_bwrite() and bwrite()) are now expected to set B_CACHE. This means that bowrite() and bawrite() also set B_CACHE indirectly. There are a number of places in the code which were previously using buf->b_bufsize (which is DEV_BSIZE aligned) when they should have been using buf->b_bcount. These have been fixed. getblk() now clears B_DONE on return because the rest of the system is so bad about dealing with B_DONE. Major fixes to NFS/TCP have been made. A server-side bug could cause requests to be lost by the server due to nfs_realign() overwriting other rpc's in the same TCP mbuf chain. The server's kernel must be recompiled to get the benefit of the fixes. Submitted by: Matthew Dillon <dillon@apollo.backplane.com>
1999-05-02 23:57:16 +00:00
/*
* vop_nostrategy:
*
* Strategy routine for VFS devices that have none.
*
* BIO_ERROR and B_INVAL must be cleared prior to calling any strategy
* routine. Typically this is done for a BIO_READ strategy call.
2003-03-10 21:55:00 +00:00
* Typically B_INVAL is assumed to already be clear prior to a write
* and should not be cleared manually unless you just made the buffer
* invalid. BIO_ERROR should be cleared either way.
The VFS/BIO subsystem contained a number of hacks in order to optimize piecemeal, middle-of-file writes for NFS. These hacks have caused no end of trouble, especially when combined with mmap(). I've removed them. Instead, NFS will issue a read-before-write to fully instantiate the struct buf containing the write. NFS does, however, optimize piecemeal appends to files. For most common file operations, you will not notice the difference. The sole remaining fragment in the VFS/BIO system is b_dirtyoff/end, which NFS uses to avoid cache coherency issues with read-merge-write style operations. NFS also optimizes the write-covers-entire-buffer case by avoiding the read-before-write. There is quite a bit of room for further optimization in these areas. The VM system marks pages fully-valid (AKA vm_page_t->valid = VM_PAGE_BITS_ALL) in several places, most noteably in vm_fault. This is not correct operation. The vm_pager_get_pages() code is now responsible for marking VM pages all-valid. A number of VM helper routines have been added to aid in zeroing-out the invalid portions of a VM page prior to the page being marked all-valid. This operation is necessary to properly support mmap(). The zeroing occurs most often when dealing with file-EOF situations. Several bugs have been fixed in the NFS subsystem, including bits handling file and directory EOF situations and buf->b_flags consistancy issues relating to clearing B_ERROR & B_INVAL, and handling B_DONE. getblk() and allocbuf() have been rewritten. B_CACHE operation is now formally defined in comments and more straightforward in implementation. B_CACHE for VMIO buffers is based on the validity of the backing store. B_CACHE for non-VMIO buffers is based simply on whether the buffer is B_INVAL or not (B_CACHE set if B_INVAL clear, and vise-versa). biodone() is now responsible for setting B_CACHE when a successful read completes. B_CACHE is also set when a bdwrite() is initiated and when a bwrite() is initiated. VFS VOP_BWRITE routines (there are only two - nfs_bwrite() and bwrite()) are now expected to set B_CACHE. This means that bowrite() and bawrite() also set B_CACHE indirectly. There are a number of places in the code which were previously using buf->b_bufsize (which is DEV_BSIZE aligned) when they should have been using buf->b_bcount. These have been fixed. getblk() now clears B_DONE on return because the rest of the system is so bad about dealing with B_DONE. Major fixes to NFS/TCP have been made. A server-side bug could cause requests to be lost by the server due to nfs_realign() overwriting other rpc's in the same TCP mbuf chain. The server's kernel must be recompiled to get the benefit of the fixes. Submitted by: Matthew Dillon <dillon@apollo.backplane.com>
1999-05-02 23:57:16 +00:00
*/
static int
vop_nostrategy (struct vop_strategy_args *ap)
{
printf("No strategy for buffer at %p\n", ap->a_bp);
vprint("vnode", ap->a_vp);
ap->a_bp->b_ioflags |= BIO_ERROR;
ap->a_bp->b_error = EOPNOTSUPP;
bufdone(ap->a_bp);
return (EOPNOTSUPP);
}
/*
* vop_stdpathconf:
2003-03-10 21:55:00 +00:00
*
* Standard implementation of POSIX pathconf, to get information about limits
* for a filesystem.
* Override per filesystem for the case where the filesystem has smaller
* limits.
*/
int
vop_stdpathconf(ap)
struct vop_pathconf_args /* {
struct vnode *a_vp;
int a_name;
int *a_retval;
} */ *ap;
{
switch (ap->a_name) {
case _PC_LINK_MAX:
*ap->a_retval = LINK_MAX;
return (0);
case _PC_MAX_CANON:
*ap->a_retval = MAX_CANON;
return (0);
case _PC_MAX_INPUT:
*ap->a_retval = MAX_INPUT;
return (0);
case _PC_PIPE_BUF:
*ap->a_retval = PIPE_BUF;
return (0);
case _PC_CHOWN_RESTRICTED:
*ap->a_retval = 1;
return (0);
case _PC_VDISABLE:
*ap->a_retval = _POSIX_VDISABLE;
return (0);
default:
return (EINVAL);
}
/* NOTREACHED */
}
/*
* Standard lock, unlock and islocked functions.
*/
int
vop_stdlock(ap)
struct vop_lock_args /* {
struct vnode *a_vp;
int a_flags;
struct thread *a_td;
} */ *ap;
2003-03-10 21:55:00 +00:00
{
struct vnode *vp = ap->a_vp;
#ifndef DEBUG_LOCKS
return (lockmgr(vp->v_vnlock, ap->a_flags, VI_MTX(vp), ap->a_td));
#else
return (debuglockmgr(vp->v_vnlock, ap->a_flags, VI_MTX(vp),
ap->a_td, "vop_stdlock", vp->filename, vp->line));
#endif
}
/* See above. */
int
vop_stdunlock(ap)
struct vop_unlock_args /* {
struct vnode *a_vp;
int a_flags;
struct thread *a_td;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
return (lockmgr(vp->v_vnlock, ap->a_flags | LK_RELEASE, VI_MTX(vp),
ap->a_td));
}
/* See above. */
int
vop_stdislocked(ap)
struct vop_islocked_args /* {
struct vnode *a_vp;
struct thread *a_td;
} */ *ap;
{
return (lockstatus(ap->a_vp->v_vnlock, ap->a_td));
}
/* Mark the vnode inactive */
int
vop_stdinactive(ap)
struct vop_inactive_args /* {
struct vnode *a_vp;
struct thread *a_td;
} */ *ap;
{
VOP_UNLOCK(ap->a_vp, 0, ap->a_td);
return (0);
}
/*
* Return true for select/poll.
*/
int
vop_nopoll(ap)
struct vop_poll_args /* {
struct vnode *a_vp;
int a_events;
struct ucred *a_cred;
struct thread *a_td;
} */ *ap;
{
/*
* Return true for read/write. If the user asked for something
* special, return POLLNVAL, so that clients have a way of
* determining reliably whether or not the extended
* functionality is present without hard-coding knowledge
* of specific filesystem implementations.
* Stay in sync with kern_conf.c::no_poll().
*/
if (ap->a_events & ~POLLSTANDARD)
return (POLLNVAL);
return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
}
/*
* Implement poll for local filesystems that support it.
*/
int
vop_stdpoll(ap)
struct vop_poll_args /* {
struct vnode *a_vp;
int a_events;
struct ucred *a_cred;
struct thread *a_td;
} */ *ap;
{
if (ap->a_events & ~POLLSTANDARD)
return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events));
return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
}
/*
* Return our mount point, as we will take charge of the writes.
*/
int
vop_stdgetwritemount(ap)
struct vop_getwritemount_args /* {
struct vnode *a_vp;
struct mount **a_mpp;
} */ *ap;
{
*(ap->a_mpp) = ap->a_vp->v_mount;
return (0);
}
/* Create the VM system backing object for this vnode */
int
vop_stdcreatevobject(ap)
struct vop_createvobject_args /* {
struct vnode *vp;
struct ucred *cred;
struct thread *td;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
struct ucred *cred = ap->a_cred;
struct thread *td = ap->a_td;
struct vattr vat;
vm_object_t object;
int error = 0;
vm_ooffset_t size;
GIANT_REQUIRED;
if (!vn_isdisk(vp, NULL) && vn_canvmio(vp) == FALSE)
return (0);
while ((object = vp->v_object) != NULL) {
VM_OBJECT_LOCK(object);
if (!(object->flags & OBJ_DEAD)) {
VM_OBJECT_UNLOCK(object);
break;
}
VOP_UNLOCK(vp, 0, td);
vm_object_set_flag(object, OBJ_DISCONNECTWNT);
msleep(object, VM_OBJECT_MTX(object), PDROP | PVM, "vodead", 0);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
}
if (object == NULL) {
if (vn_isdisk(vp, NULL)) {
/*
* This simply allocates the biggest object possible
* for a disk vnode. This should be fixed, but doesn't
* cause any problems (yet).
*/
size = IDX_TO_OFF(INT_MAX);
} else {
if ((error = VOP_GETATTR(vp, &vat, cred, td)) != 0)
return (error);
size = vat.va_size;
}
object = vnode_pager_alloc(vp, size, 0, 0);
/*
* Dereference the reference we just created. This assumes
* that the object is associated with the vp.
*/
VM_OBJECT_LOCK(object);
object->ref_count--;
VM_OBJECT_UNLOCK(object);
vrele(vp);
}
KASSERT(vp->v_object != NULL, ("vfs_object_create: NULL object"));
vp->v_vflag |= VV_OBJBUF;
return (error);
}
/* Destroy the VM system object associated with this vnode */
int
vop_stddestroyvobject(ap)
struct vop_destroyvobject_args /* {
struct vnode *vp;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
vm_object_t obj = vp->v_object;
GIANT_REQUIRED;
if (obj == NULL)
return (0);
VM_OBJECT_LOCK(obj);
if (obj->ref_count == 0) {
/*
* vclean() may be called twice. The first time
* removes the primary reference to the object,
* the second time goes one further and is a
* special-case to terminate the object.
*
* don't double-terminate the object
*/
if ((obj->flags & OBJ_DEAD) == 0)
vm_object_terminate(obj);
else
VM_OBJECT_UNLOCK(obj);
} else {
/*
* Woe to the process that tries to page now :-).
*/
vm_pager_deallocate(obj);
VM_OBJECT_UNLOCK(obj);
}
return (0);
}
/*
* Return the underlying VM object. This routine may be called with or
* without the vnode interlock held. If called without, the returned
* object is not guarenteed to be valid. The syncer typically gets the
* object without holding the interlock in order to quickly test whether
* it might be dirty before going heavy-weight. vm_object's use zalloc
* and thus stable-storage, so this is safe.
*/
int
vop_stdgetvobject(ap)
struct vop_getvobject_args /* {
struct vnode *vp;
struct vm_object **objpp;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
struct vm_object **objpp = ap->a_objpp;
if (objpp)
*objpp = vp->v_object;
return (vp->v_object ? 0 : EINVAL);
}
/* XXX Needs good comment and VOP_BMAP(9) manpage */
int
vop_stdbmap(ap)
2003-03-10 21:55:00 +00:00
struct vop_bmap_args /* {
struct vnode *a_vp;
daddr_t a_bn;
struct bufobj **a_bop;
daddr_t *a_bnp;
int *a_runp;
int *a_runb;
} */ *ap;
{
if (ap->a_bop != NULL)
*ap->a_bop = &ap->a_vp->v_bufobj;
if (ap->a_bnp != NULL)
*ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize);
if (ap->a_runp != NULL)
*ap->a_runp = 0;
if (ap->a_runb != NULL)
*ap->a_runb = 0;
return (0);
}
int
vop_stdfsync(ap)
struct vop_fsync_args /* {
struct vnode *a_vp;
struct ucred *a_cred;
int a_waitfor;
struct thread *a_td;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
struct buf *bp;
struct bufobj *bo;
struct buf *nbp;
int s, error = 0;
int maxretry = 100; /* large, arbitrarily chosen */
VI_LOCK(vp);
loop1:
/*
* MARK/SCAN initialization to avoid infinite loops.
*/
s = splbio();
TAILQ_FOREACH(bp, &vp->v_bufobj.bo_dirty.bv_hd, b_bobufs) {
bp->b_vflags &= ~BV_SCANNED;
bp->b_error = 0;
}
splx(s);
/*
* Flush all dirty buffers associated with a block device.
*/
loop2:
s = splbio();
TAILQ_FOREACH_SAFE(bp, &vp->v_bufobj.bo_dirty.bv_hd, b_bobufs, nbp) {
if ((bp->b_vflags & BV_SCANNED) != 0)
continue;
bp->b_vflags |= BV_SCANNED;
if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL))
continue;
VI_UNLOCK(vp);
if ((bp->b_flags & B_DELWRI) == 0)
panic("fsync: not dirty");
if ((vp->v_vflag & VV_OBJBUF) && (bp->b_flags & B_CLUSTEROK)) {
vfs_bio_awrite(bp);
splx(s);
} else {
bremfree(bp);
splx(s);
bawrite(bp);
}
VI_LOCK(vp);
goto loop2;
}
/*
* If synchronous the caller expects us to completely resolve all
* dirty buffers in the system. Wait for in-progress I/O to
* complete (which could include background bitmap writes), then
* retry if dirty blocks still exist.
*/
if (ap->a_waitfor == MNT_WAIT) {
bo = &vp->v_bufobj;
bufobj_wwait(bo, 0, 0);
if (bo->bo_dirty.bv_cnt > 0) {
/*
* If we are unable to write any of these buffers
* then we fail now rather than trying endlessly
* to write them out.
*/
TAILQ_FOREACH(bp, &bo->bo_dirty.bv_hd, b_bobufs)
if ((error = bp->b_error) == 0)
continue;
if (error == 0 && --maxretry >= 0) {
splx(s);
goto loop1;
}
vprint("fsync: giving up on dirty", vp);
error = EAGAIN;
}
}
VI_UNLOCK(vp);
splx(s);
2003-03-10 21:55:00 +00:00
return (error);
}
2003-03-10 21:55:00 +00:00
/* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */
int
vop_stdgetpages(ap)
struct vop_getpages_args /* {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int a_reqpage;
vm_ooffset_t a_offset;
} */ *ap;
{
return vnode_pager_generic_getpages(ap->a_vp, ap->a_m,
ap->a_count, ap->a_reqpage);
}
/* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */
int
vop_stdputpages(ap)
struct vop_putpages_args /* {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int a_sync;
int *a_rtvals;
vm_ooffset_t a_offset;
} */ *ap;
{
return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
ap->a_sync, ap->a_rtvals);
}
2003-03-10 21:55:00 +00:00
/*
* vfs default ops
* used to fill the vfs function table to get reasonable default return values.
*/
int
vfs_stdroot (mp, vpp, td)
struct mount *mp;
struct vnode **vpp;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdstatfs (mp, sbp, td)
struct mount *mp;
struct statfs *sbp;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdvptofh (vp, fhp)
struct vnode *vp;
struct fid *fhp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdstart (mp, flags, td)
struct mount *mp;
int flags;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
return (0);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdquotactl (mp, cmds, uid, arg, td)
struct mount *mp;
int cmds;
uid_t uid;
caddr_t arg;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdsync(mp, waitfor, cred, td)
struct mount *mp;
int waitfor;
struct ucred *cred;
struct thread *td;
{
struct vnode *vp, *nvp;
int error, lockreq, allerror = 0;
lockreq = LK_EXCLUSIVE | LK_INTERLOCK;
if (waitfor != MNT_WAIT)
lockreq |= LK_NOWAIT;
/*
* Force stale buffer cache information to be flushed.
*/
MNT_ILOCK(mp);
loop:
When we traverse the vnodes on a mountpoint we need to look out for our cached 'next vnode' being removed from this mountpoint. If we find that it was recycled, we restart our traversal from the start of the list. Code to do that is in all local disk filesystems (and a few other places) and looks roughly like this: MNT_ILOCK(mp); loop: for (vp = TAILQ_FIRST(&mp...); (vp = nvp) != NULL; nvp = TAILQ_NEXT(vp,...)) { if (vp->v_mount != mp) goto loop; MNT_IUNLOCK(mp); ... MNT_ILOCK(mp); } MNT_IUNLOCK(mp); The code which takes vnodes off a mountpoint looks like this: MNT_ILOCK(vp->v_mount); ... TAILQ_REMOVE(&vp->v_mount->mnt_nvnodelist, vp, v_nmntvnodes); ... MNT_IUNLOCK(vp->v_mount); ... vp->v_mount = something; (Take a moment and try to spot the locking error before you read on.) On a SMP system, one CPU could have removed nvp from our mountlist but not yet gotten to assign a new value to vp->v_mount while another CPU simultaneously get to the top of the traversal loop where it finds that (vp->v_mount != mp) is not true despite the fact that the vnode has indeed been removed from our mountpoint. Fix: Introduce the macro MNT_VNODE_FOREACH() to traverse the list of vnodes on a mountpoint while taking into account that vnodes may be removed from the list as we go. This saves approx 65 lines of duplicated code. Split the insmntque() which potentially moves a vnode from one mount point to another into delmntque() and insmntque() which does just what the names say. Fix delmntque() to set vp->v_mount to NULL while holding the mountpoint lock.
2004-07-04 08:52:35 +00:00
MNT_VNODE_FOREACH(vp, mp, nvp) {
VI_LOCK(vp);
if (vp->v_bufobj.bo_dirty.bv_cnt == 0) {
VI_UNLOCK(vp);
continue;
}
MNT_IUNLOCK(mp);
if ((error = vget(vp, lockreq, td)) != 0) {
MNT_ILOCK(mp);
if (error == ENOENT)
goto loop;
continue;
}
error = VOP_FSYNC(vp, cred, waitfor, td);
if (error)
allerror = error;
VOP_UNLOCK(vp, 0, td);
vrele(vp);
MNT_ILOCK(mp);
}
MNT_IUNLOCK(mp);
return (allerror);
}
int
vfs_stdnosync (mp, waitfor, cred, td)
struct mount *mp;
int waitfor;
2003-03-10 21:55:00 +00:00
struct ucred *cred;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
return (0);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdvget (mp, ino, flags, vpp)
struct mount *mp;
ino_t ino;
int flags;
struct vnode **vpp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdfhtovp (mp, fhp, vpp)
struct mount *mp;
struct fid *fhp;
struct vnode **vpp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
2003-03-10 21:55:00 +00:00
vfs_stdinit (vfsp)
struct vfsconf *vfsp;
{
2004-07-07 07:00:02 +00:00
return (0);
}
int
vfs_stduninit (vfsp)
struct vfsconf *vfsp;
{
2004-07-07 07:00:02 +00:00
return(0);
}
int
vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace, attrname, td)
struct mount *mp;
int cmd;
o Change the API and ABI of the Extended Attribute kernel interfaces to introduce a new argument, "namespace", rather than relying on a first- character namespace indicator. This is in line with more recent thinking on EA interfaces on various mailing lists, including the posix1e, Linux acl-devel, and trustedbsd-discuss forums. Two namespaces are defined by default, EXTATTR_NAMESPACE_SYSTEM and EXTATTR_NAMESPACE_USER, where the primary distinction lies in the access control model: user EAs are accessible based on the normal MAC and DAC file/directory protections, and system attributes are limited to kernel-originated or appropriately privileged userland requests. o These API changes occur at several levels: the namespace argument is introduced in the extattr_{get,set}_file() system call interfaces, at the vnode operation level in the vop_{get,set}extattr() interfaces, and in the UFS extended attribute implementation. Changes are also introduced in the VFS extattrctl() interface (system call, VFS, and UFS implementation), where the arguments are modified to include a namespace field, as well as modified to advoid direct access to userspace variables from below the VFS layer (in the style of recent changes to mount by adrian@FreeBSD.org). This required some cleanup and bug fixing regarding VFS locks and the VFS interface, as a vnode pointer may now be optionally submitted to the VFS_EXTATTRCTL() call. Updated documentation for the VFS interface will be committed shortly. o In the near future, the auto-starting feature will be updated to search two sub-directories to the ".attribute" directory in appropriate file systems: "user" and "system" to locate attributes intended for those namespaces, as the single filename is no longer sufficient to indicate what namespace the attribute is intended for. Until this is committed, all attributes auto-started by UFS will be placed in the EXTATTR_NAMESPACE_SYSTEM namespace. o The default POSIX.1e attribute names for ACLs and Capabilities have been updated to no longer include the '$' in their filename. As such, if you're using these features, you'll need to rename the attribute backing files to the same names without '$' symbols in front. o Note that these changes will require changes in userland, which will be committed shortly. These include modifications to the extended attribute utilities, as well as to libutil for new namespace string conversion routines. Once the matching userland changes are committed, a buildworld is recommended to update all the necessary include files and verify that the kernel and userland environments are in sync. Note: If you do not use extended attributes (most people won't), upgrading is not imperative although since the system call API has changed, the new userland extended attribute code will no longer compile with old include files. o Couple of minor cleanups while I'm there: make more code compilation conditional on FFS_EXTATTR, which should recover a bit of space on kernels running without EA's, as well as update copyright dates. Obtained from: TrustedBSD Project
2001-03-15 02:54:29 +00:00
struct vnode *filename_vp;
int attrnamespace;
const char *attrname;
struct thread *td;
{
2004-07-07 07:00:02 +00:00
if (filename_vp != NULL)
VOP_UNLOCK(filename_vp, 0, td);
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdsysctl(mp, op, req)
struct mount *mp;
fsctlop_t op;
struct sysctl_req *req;
{
return (EOPNOTSUPP);
}
/* end of vfs default ops */