1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Copyright (c) 1982, 1986, 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.
|
|
|
|
*
|
1997-02-10 02:22:35 +00:00
|
|
|
* @(#)ffs_vnops.c 8.15 (Berkeley) 5/14/95
|
1998-09-07 11:50:19 +00:00
|
|
|
* $Id: ffs_vnops.c,v 1.50 1998/06/10 19:27:56 julian Exp $
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/resourcevar.h>
|
1994-10-10 01:04:55 +00:00
|
|
|
#include <sys/signalvar.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/kernel.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#include <sys/buf.h>
|
|
|
|
#include <sys/proc.h>
|
|
|
|
#include <sys/mount.h>
|
|
|
|
#include <sys/vnode.h>
|
|
|
|
|
1998-03-28 10:33:27 +00:00
|
|
|
#include <machine/limits.h>
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <vm/vm.h>
|
1995-12-07 12:48:31 +00:00
|
|
|
#include <vm/vm_prot.h>
|
1995-04-09 06:03:56 +00:00
|
|
|
#include <vm/vm_page.h>
|
|
|
|
#include <vm/vm_object.h>
|
1995-12-07 12:48:31 +00:00
|
|
|
#include <vm/vm_extern.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
#include <ufs/ufs/quota.h>
|
|
|
|
#include <ufs/ufs/inode.h>
|
1997-02-10 02:22:35 +00:00
|
|
|
#include <ufs/ufs/ufsmount.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <ufs/ufs/ufs_extern.h>
|
|
|
|
|
|
|
|
#include <ufs/ffs/fs.h>
|
|
|
|
#include <ufs/ffs/ffs_extern.h>
|
|
|
|
|
1995-12-17 21:14:36 +00:00
|
|
|
static int ffs_fsync __P((struct vop_fsync_args *));
|
|
|
|
static int ffs_getpages __P((struct vop_getpages_args *));
|
1998-02-26 06:39:59 +00:00
|
|
|
static int ffs_putpages __P((struct vop_putpages_args *));
|
1995-12-17 21:14:36 +00:00
|
|
|
static int ffs_read __P((struct vop_read_args *));
|
|
|
|
static int ffs_write __P((struct vop_write_args *));
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/* Global vfs data structures for ufs. */
|
1995-11-09 08:17:23 +00:00
|
|
|
vop_t **ffs_vnodeop_p;
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_entry_desc ffs_vnodeop_entries[] = {
|
1997-10-15 13:24:07 +00:00
|
|
|
{ &vop_default_desc, (vop_t *) ufs_vnoperate },
|
1997-10-15 09:22:02 +00:00
|
|
|
{ &vop_fsync_desc, (vop_t *) ffs_fsync },
|
1997-10-15 10:05:29 +00:00
|
|
|
{ &vop_getpages_desc, (vop_t *) ffs_getpages },
|
1998-02-26 06:39:59 +00:00
|
|
|
{ &vop_putpages_desc, (vop_t *) ffs_putpages },
|
1997-10-15 09:22:02 +00:00
|
|
|
{ &vop_read_desc, (vop_t *) ffs_read },
|
1998-03-08 09:59:44 +00:00
|
|
|
{ &vop_balloc_desc, (vop_t *) ffs_balloc },
|
1997-10-15 09:22:02 +00:00
|
|
|
{ &vop_reallocblks_desc, (vop_t *) ffs_reallocblks },
|
|
|
|
{ &vop_write_desc, (vop_t *) ffs_write },
|
1995-11-09 08:17:23 +00:00
|
|
|
{ NULL, NULL }
|
1994-05-24 10:09:53 +00:00
|
|
|
};
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_desc ffs_vnodeop_opv_desc =
|
1994-05-24 10:09:53 +00:00
|
|
|
{ &ffs_vnodeop_p, ffs_vnodeop_entries };
|
|
|
|
|
1995-11-09 08:17:23 +00:00
|
|
|
vop_t **ffs_specop_p;
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_entry_desc ffs_specop_entries[] = {
|
1997-10-15 13:24:07 +00:00
|
|
|
{ &vop_default_desc, (vop_t *) ufs_vnoperatespec },
|
1997-10-15 09:22:02 +00:00
|
|
|
{ &vop_fsync_desc, (vop_t *) ffs_fsync },
|
1995-11-09 08:17:23 +00:00
|
|
|
{ NULL, NULL }
|
1994-05-24 10:09:53 +00:00
|
|
|
};
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_desc ffs_specop_opv_desc =
|
1994-05-24 10:09:53 +00:00
|
|
|
{ &ffs_specop_p, ffs_specop_entries };
|
|
|
|
|
1995-11-09 08:17:23 +00:00
|
|
|
vop_t **ffs_fifoop_p;
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_entry_desc ffs_fifoop_entries[] = {
|
1997-10-15 13:24:07 +00:00
|
|
|
{ &vop_default_desc, (vop_t *) ufs_vnoperatefifo },
|
1997-10-15 09:22:02 +00:00
|
|
|
{ &vop_fsync_desc, (vop_t *) ffs_fsync },
|
1995-11-09 08:17:23 +00:00
|
|
|
{ NULL, NULL }
|
1994-05-24 10:09:53 +00:00
|
|
|
};
|
1995-12-17 21:14:36 +00:00
|
|
|
static struct vnodeopv_desc ffs_fifoop_opv_desc =
|
1994-05-24 10:09:53 +00:00
|
|
|
{ &ffs_fifoop_p, ffs_fifoop_entries };
|
|
|
|
|
1994-09-21 03:47:43 +00:00
|
|
|
VNODEOP_SET(ffs_vnodeop_opv_desc);
|
|
|
|
VNODEOP_SET(ffs_specop_opv_desc);
|
|
|
|
VNODEOP_SET(ffs_fifoop_opv_desc);
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <ufs/ufs/ufs_readwrite.c>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Synch an open file.
|
|
|
|
*/
|
|
|
|
/* ARGSUSED */
|
1995-12-17 21:14:36 +00:00
|
|
|
static int
|
1994-05-24 10:09:53 +00:00
|
|
|
ffs_fsync(ap)
|
|
|
|
struct vop_fsync_args /* {
|
|
|
|
struct vnode *a_vp;
|
|
|
|
struct ucred *a_cred;
|
|
|
|
int a_waitfor;
|
|
|
|
struct proc *a_p;
|
|
|
|
} */ *ap;
|
|
|
|
{
|
1998-03-08 09:59:44 +00:00
|
|
|
struct vnode *vp = ap->a_vp;
|
|
|
|
struct buf *bp;
|
1994-05-24 10:09:53 +00:00
|
|
|
struct timeval tv;
|
|
|
|
struct buf *nbp;
|
1998-03-08 09:59:44 +00:00
|
|
|
int s, error, passes, skipmeta;
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
daddr_t lbn;
|
|
|
|
|
1998-02-01 08:24:00 +00:00
|
|
|
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
if (vp->v_type == VBLK) {
|
|
|
|
lbn = INT_MAX;
|
|
|
|
} else {
|
|
|
|
struct inode *ip;
|
|
|
|
ip = VTOI(vp);
|
|
|
|
lbn = lblkno(ip->i_fs, (ip->i_size + ip->i_fs->fs_bsize - 1));
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Flush all dirty buffers associated with a vnode.
|
|
|
|
*/
|
1998-03-08 09:59:44 +00:00
|
|
|
passes = NIADDR;
|
|
|
|
skipmeta = 0;
|
|
|
|
if (ap->a_waitfor == MNT_WAIT)
|
|
|
|
skipmeta = 1;
|
1994-05-24 10:09:53 +00:00
|
|
|
loop:
|
|
|
|
s = splbio();
|
1998-03-08 09:59:44 +00:00
|
|
|
loop2:
|
1994-05-24 10:09:53 +00:00
|
|
|
for (bp = vp->v_dirtyblkhd.lh_first; bp; bp = nbp) {
|
|
|
|
nbp = bp->b_vnbufs.le_next;
|
1998-03-08 09:59:44 +00:00
|
|
|
/*
|
|
|
|
* First time through on a synchronous call,
|
|
|
|
* or if it's already scheduled, skip to the next
|
|
|
|
* buffer
|
|
|
|
*/
|
|
|
|
if ((bp->b_flags & B_BUSY) ||
|
|
|
|
((skipmeta == 1) && (bp->b_lblkno < 0)))
|
1994-05-24 10:09:53 +00:00
|
|
|
continue;
|
|
|
|
if ((bp->b_flags & B_DELWRI) == 0)
|
|
|
|
panic("ffs_fsync: not dirty");
|
1998-03-08 09:59:44 +00:00
|
|
|
/*
|
|
|
|
* If data is outstanding to another vnode, or we were
|
|
|
|
* asked to wait for everything, or it's not a file or BDEV,
|
|
|
|
* start the IO on this buffer immediatly.
|
|
|
|
*/
|
|
|
|
if (((bp->b_vp != vp) || (ap->a_waitfor == MNT_WAIT)) ||
|
|
|
|
((vp->v_type != VREG) && (vp->v_type != VBLK))) {
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
|
1995-04-09 06:03:56 +00:00
|
|
|
/*
|
1998-03-08 09:59:44 +00:00
|
|
|
* Wait for I/O associated with indirect blocks to
|
|
|
|
* complete, since there is no way to quickly wait
|
|
|
|
* for them below.
|
1995-04-09 06:03:56 +00:00
|
|
|
*/
|
1998-03-08 09:59:44 +00:00
|
|
|
if ((bp->b_vp == vp) || (ap->a_waitfor != MNT_WAIT)) {
|
1998-06-10 19:27:56 +00:00
|
|
|
if ((bp->b_flags & B_CLUSTEROK) &&
|
|
|
|
ap->a_waitfor != MNT_WAIT) {
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
(void) vfs_bio_awrite(bp);
|
1998-03-19 22:49:44 +00:00
|
|
|
splx(s);
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
} else {
|
1998-03-19 22:49:44 +00:00
|
|
|
bremfree(bp);
|
|
|
|
bp->b_flags |= B_BUSY;
|
|
|
|
splx(s);
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
(void) bawrite(bp);
|
|
|
|
}
|
|
|
|
} else {
|
1998-03-19 22:49:44 +00:00
|
|
|
bremfree(bp);
|
|
|
|
bp->b_flags |= B_BUSY;
|
|
|
|
splx(s);
|
These changes embody the support of the fully coherent merged VM buffer cache,
much higher filesystem I/O performance, and much better paging performance. It
represents the culmination of over 6 months of R&D.
The majority of the merged VM/cache work is by John Dyson.
The following highlights the most significant changes. Additionally, there are
(mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to
support the new VM/buffer scheme.
vfs_bio.c:
Significant rewrite of most of vfs_bio to support the merged VM buffer cache
scheme. The scheme is almost fully compatible with the old filesystem
interface. Significant improvement in the number of opportunities for write
clustering.
vfs_cluster.c, vfs_subr.c
Upgrade and performance enhancements in vfs layer code to support merged
VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff.
vm_object.c:
Yet more improvements in the collapse code. Elimination of some windows that
can cause list corruption.
vm_pageout.c:
Fixed it, it really works better now. Somehow in 2.0, some "enhancements"
broke the code. This code has been reworked from the ground-up.
vm_fault.c, vm_page.c, pmap.c, vm_object.c
Support for small-block filesystems with merged VM/buffer cache scheme.
pmap.c vm_map.c
Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of
kernel PTs.
vm_glue.c
Much simpler and more effective swapping code. No more gratuitous swapping.
proc.h
Fixed the problem that the p_lock flag was not being cleared on a fork.
swap_pager.c, vnode_pager.c
Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the
code doesn't need it anymore.
machdep.c
Changes to better support the parameter values for the merged VM/buffer cache
scheme.
machdep.c, kern_exec.c, vm_glue.c
Implemented a seperate submap for temporary exec string space and another one
to contain process upages. This eliminates all map fragmentation problems
that previously existed.
ffs_inode.c, ufs_inode.c, ufs_readwrite.c
Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on
busy buffers.
Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
|
|
|
(void) bwrite(bp);
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
}
|
|
|
|
} else if ((vp->v_type == VREG) && (bp->b_lblkno >= lbn)) {
|
1998-03-08 09:59:44 +00:00
|
|
|
/*
|
|
|
|
* If the buffer is for data that has been truncated
|
|
|
|
* off the file, then throw it away.
|
|
|
|
*/
|
Make our v_usecount vnode reference count work identically to the
original BSD code. The association between the vnode and the vm_object
no longer includes reference counts. The major difference is that
vm_object's are no longer freed gratuitiously from the vnode, and so
once an object is created for the vnode, it will last as long as the
vnode does.
When a vnode object reference count is incremented, then the underlying
vnode reference count is incremented also. The two "objects" are now
more intimately related, and so the interactions are now much less
complex.
When vnodes are now normally placed onto the free queue with an object still
attached. The rundown of the object happens at vnode rundown time, and
happens with exactly the same filesystem semantics of the original VFS
code. There is absolutely no need for vnode_pager_uncache and other
travesties like that anymore.
A side-effect of these changes is that SMP locking should be much simpler,
the I/O copyin/copyout optimizations work, NFS should be more ponderable,
and further work on layered filesystems should be less frustrating, because
of the totally coherent management of the vnode objects and vnodes.
Please be careful with your system while running this code, but I would
greatly appreciate feedback as soon a reasonably possible.
1998-01-06 05:26:17 +00:00
|
|
|
bremfree(bp);
|
|
|
|
bp->b_flags |= B_BUSY | B_INVAL | B_NOCACHE;
|
|
|
|
brelse(bp);
|
|
|
|
splx(s);
|
These changes embody the support of the fully coherent merged VM buffer cache,
much higher filesystem I/O performance, and much better paging performance. It
represents the culmination of over 6 months of R&D.
The majority of the merged VM/cache work is by John Dyson.
The following highlights the most significant changes. Additionally, there are
(mostly minor) changes to the various filesystem modules (nfs, msdosfs, etc) to
support the new VM/buffer scheme.
vfs_bio.c:
Significant rewrite of most of vfs_bio to support the merged VM buffer cache
scheme. The scheme is almost fully compatible with the old filesystem
interface. Significant improvement in the number of opportunities for write
clustering.
vfs_cluster.c, vfs_subr.c
Upgrade and performance enhancements in vfs layer code to support merged
VM/buffer cache. Fixup of vfs_cluster to eliminate the bogus pagemove stuff.
vm_object.c:
Yet more improvements in the collapse code. Elimination of some windows that
can cause list corruption.
vm_pageout.c:
Fixed it, it really works better now. Somehow in 2.0, some "enhancements"
broke the code. This code has been reworked from the ground-up.
vm_fault.c, vm_page.c, pmap.c, vm_object.c
Support for small-block filesystems with merged VM/buffer cache scheme.
pmap.c vm_map.c
Dynamic kernel VM size, now we dont have to pre-allocate excessive numbers of
kernel PTs.
vm_glue.c
Much simpler and more effective swapping code. No more gratuitous swapping.
proc.h
Fixed the problem that the p_lock flag was not being cleared on a fork.
swap_pager.c, vnode_pager.c
Removal of old vfs_bio cruft to support the past pseudo-coherency. Now the
code doesn't need it anymore.
machdep.c
Changes to better support the parameter values for the merged VM/buffer cache
scheme.
machdep.c, kern_exec.c, vm_glue.c
Implemented a seperate submap for temporary exec string space and another one
to contain process upages. This eliminates all map fragmentation problems
that previously existed.
ffs_inode.c, ufs_inode.c, ufs_readwrite.c
Changes for merged VM/buffer cache. Add "bypass" support for sneaking in on
busy buffers.
Submitted by: John Dyson and David Greenman
1995-01-09 16:06:02 +00:00
|
|
|
} else {
|
|
|
|
vfs_bio_awrite(bp);
|
|
|
|
splx(s);
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
goto loop;
|
|
|
|
}
|
1998-03-08 09:59:44 +00:00
|
|
|
/*
|
|
|
|
* If we were asked to do this synchronously, then go back for
|
|
|
|
* another pass, this time doing the metadata.
|
|
|
|
*/
|
|
|
|
if (skipmeta) {
|
|
|
|
skipmeta = 0;
|
|
|
|
goto loop2; /* stay within the splbio() */
|
1995-04-09 06:03:56 +00:00
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
if (ap->a_waitfor == MNT_WAIT) {
|
1998-06-10 19:27:56 +00:00
|
|
|
while (vp->v_numoutput) {
|
|
|
|
vp->v_flag |= VBWAIT;
|
|
|
|
(void) tsleep((caddr_t)&vp->v_numoutput,
|
|
|
|
PRIBIO + 4, "ffsfsn", 0);
|
|
|
|
}
|
1998-03-21 05:16:09 +00:00
|
|
|
|
1998-06-10 19:27:56 +00:00
|
|
|
/*
|
|
|
|
* Ensure that any filesystem metatdata associated
|
|
|
|
* with the vnode has been written.
|
|
|
|
*/
|
|
|
|
splx(s);
|
|
|
|
if ((error = softdep_sync_metadata(ap)) != 0)
|
|
|
|
return (error);
|
1995-04-09 06:03:56 +00:00
|
|
|
s = splbio();
|
1998-03-21 05:16:09 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
if (vp->v_dirtyblkhd.lh_first) {
|
1998-03-08 09:59:44 +00:00
|
|
|
/*
|
|
|
|
* Block devices associated with filesystems may
|
|
|
|
* have new I/O requests posted for them even if
|
|
|
|
* the vnode is locked, so no amount of trying will
|
|
|
|
* get them clean. Thus we give block devices a
|
|
|
|
* good effort, then just give up. For all other file
|
|
|
|
* types, go around and try again until it is clean.
|
|
|
|
*/
|
|
|
|
if (passes > 0) {
|
|
|
|
passes -= 1;
|
|
|
|
goto loop2;
|
|
|
|
}
|
|
|
|
#ifdef DIAGNOSTIC
|
|
|
|
if (vp->v_type != VBLK)
|
|
|
|
vprint("ffs_fsync: dirty", vp);
|
1994-05-24 10:09:53 +00:00
|
|
|
#endif
|
1998-03-08 09:59:44 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1998-06-10 19:27:56 +00:00
|
|
|
splx(s);
|
1998-03-30 09:56:58 +00:00
|
|
|
getmicrotime(&tv);
|
1998-06-10 19:27:56 +00:00
|
|
|
if ((error = UFS_UPDATE(vp, &tv, &tv, ap->a_waitfor == MNT_WAIT)) != 0)
|
1998-03-08 09:59:44 +00:00
|
|
|
return (error);
|
|
|
|
if (DOINGSOFTDEP(vp) && ap->a_waitfor == MNT_WAIT)
|
|
|
|
error = softdep_fsync(vp);
|
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|