Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
68aacb8f70
NetBSD/sys/ufs/ffs/ffs_vnops.c
chs 64c6d1d2dc a whole bunch of changes to improve performance and robustness under load: 
- remove special treatment of pager_map mappings in pmaps.  this is
   required now, since I've removed the globals that expose the address range.
   pager_map now uses pmap_kenter_pa() instead of pmap_enter(), so there's
   no longer any need to special-case it.
 - eliminate struct uvm_vnode by moving its fields into struct vnode.
 - rewrite the pageout path.  the pager is now responsible for handling the
   high-level requests instead of only getting control after a bunch of work
   has already been done on its behalf.  this will allow us to UBCify LFS,
   which needs tighter control over its pages than other filesystems do.
   writing a page to disk no longer requires making it read-only, which
   allows us to write wired pages without causing all kinds of havoc.
 - use a new PG_PAGEOUT flag to indicate that a page should be freed
   on behalf of the pagedaemon when it's unlocked.  this flag is very similar
   to PG_RELEASED, but unlike PG_RELEASED, PG_PAGEOUT can be cleared if the
   pageout fails due to eg. an indirect-block buffer being locked.
   this allows us to remove the "version" field from struct vm_page,
   and together with shrinking "loan_count" from 32 bits to 16,
   struct vm_page is now 4 bytes smaller.
 - no longer use PG_RELEASED for swap-backed pages.  if the page is busy
   because it's being paged out, we can't release the swap slot to be
   reallocated until that write is complete, but unlike with vnodes we
   don't keep a count of in-progress writes so there's no good way to
   know when the write is done.  instead, when we need to free a busy
   swap-backed page, just sleep until we can get it busy ourselves.
 - implement a fast-path for extending writes which allows us to avoid
   zeroing new pages.  this substantially reduces cpu usage.
 - encapsulate the data used by the genfs code in a struct genfs_node,
   which must be the first element of the filesystem-specific vnode data
   for filesystems which use genfs_{get,put}pages().
 - eliminate many of the UVM pagerops, since they aren't needed anymore
   now that the pager "put" operation is a higher-level operation.
 - enhance the genfs code to allow NFS to use the genfs_{get,put}pages
   instead of a modified copy.
 - clean up struct vnode by removing all the fields that used to be used by
   the vfs_cluster.c code (which we don't use anymore with UBC).
 - remove kmem_object and mb_object since they were useless.
   instead of allocating pages to these objects, we now just allocate
   pages with no object.  such pages are mapped in the kernel until they
   are freed, so we can use the mapping to find the page to free it.
   this allows us to remove splvm() protection in several places.

The sum of all these changes improves write throughput on my
decstation 5000/200 to within 1% of the rate of NetBSD 1.5
and reduces the elapsed time for "make release" of a NetBSD 1.5
source tree on my 128MB pc to 10% less than a 1.5 kernel took.
2001-09-15 20:36:31 +00:00

515 lines
16 KiB
C
Raw Blame History

/* $NetBSD: ffs_vnops.c,v 1.39 2001/09/15 20:36:43 chs Exp $ */
/*
* 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.
*
* @(#)ffs_vnops.c 8.15 (Berkeley) 5/14/95
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/resourcevar.h>
#include <sys/kernel.h>
#include <sys/file.h>
#include <sys/stat.h>
#include <sys/buf.h>
#include <sys/proc.h>
#include <sys/mount.h>
#include <sys/vnode.h>
#include <sys/pool.h>
#include <sys/signalvar.h>
#include <miscfs/fifofs/fifo.h>
#include <miscfs/genfs/genfs.h>
#include <miscfs/specfs/specdev.h>
#include <ufs/ufs/quota.h>
#include <ufs/ufs/inode.h>
#include <ufs/ufs/dir.h>
#include <ufs/ufs/ufs_extern.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ffs/fs.h>
#include <ufs/ffs/ffs_extern.h>
static int ffs_full_fsync __P((void *));
/* Global vfs data structures for ufs. */
int (**ffs_vnodeop_p) __P((void *));
const struct vnodeopv_entry_desc ffs_vnodeop_entries[] = {
{ &vop_default_desc, vn_default_error },
{ &vop_lookup_desc, ufs_lookup }, /* lookup */
{ &vop_create_desc, ufs_create }, /* create */
{ &vop_whiteout_desc, ufs_whiteout }, /* whiteout */
{ &vop_mknod_desc, ufs_mknod }, /* mknod */
{ &vop_open_desc, ufs_open }, /* open */
{ &vop_close_desc, ufs_close }, /* close */
{ &vop_access_desc, ufs_access }, /* access */
{ &vop_getattr_desc, ufs_getattr }, /* getattr */
{ &vop_setattr_desc, ufs_setattr }, /* setattr */
{ &vop_read_desc, ffs_read }, /* read */
{ &vop_write_desc, ffs_write }, /* write */
{ &vop_lease_desc, ufs_lease_check }, /* lease */
{ &vop_ioctl_desc, ufs_ioctl }, /* ioctl */
{ &vop_fcntl_desc, ufs_fcntl }, /* fcntl */
{ &vop_poll_desc, ufs_poll }, /* poll */
{ &vop_revoke_desc, ufs_revoke }, /* revoke */
{ &vop_mmap_desc, ufs_mmap }, /* mmap */
{ &vop_fsync_desc, ffs_fsync }, /* fsync */
{ &vop_seek_desc, ufs_seek }, /* seek */
{ &vop_remove_desc, ufs_remove }, /* remove */
{ &vop_link_desc, ufs_link }, /* link */
{ &vop_rename_desc, ufs_rename }, /* rename */
{ &vop_mkdir_desc, ufs_mkdir }, /* mkdir */
{ &vop_rmdir_desc, ufs_rmdir }, /* rmdir */
{ &vop_symlink_desc, ufs_symlink }, /* symlink */
{ &vop_readdir_desc, ufs_readdir }, /* readdir */
{ &vop_readlink_desc, ufs_readlink }, /* readlink */
{ &vop_abortop_desc, ufs_abortop }, /* abortop */
{ &vop_inactive_desc, ufs_inactive }, /* inactive */
{ &vop_reclaim_desc, ffs_reclaim }, /* reclaim */
{ &vop_lock_desc, ufs_lock }, /* lock */
{ &vop_unlock_desc, ufs_unlock }, /* unlock */
{ &vop_bmap_desc, ufs_bmap }, /* bmap */
{ &vop_strategy_desc, ufs_strategy }, /* strategy */
{ &vop_print_desc, ufs_print }, /* print */
{ &vop_islocked_desc, ufs_islocked }, /* islocked */
{ &vop_pathconf_desc, ufs_pathconf }, /* pathconf */
{ &vop_advlock_desc, ufs_advlock }, /* advlock */
{ &vop_blkatoff_desc, ffs_blkatoff }, /* blkatoff */
{ &vop_valloc_desc, ffs_valloc }, /* valloc */
{ &vop_balloc_desc, ffs_balloc }, /* balloc */
{ &vop_reallocblks_desc, ffs_reallocblks }, /* reallocblks */
{ &vop_vfree_desc, ffs_vfree }, /* vfree */
{ &vop_truncate_desc, ffs_truncate }, /* truncate */
{ &vop_update_desc, ffs_update }, /* update */
{ &vop_bwrite_desc, vn_bwrite }, /* bwrite */
{ &vop_getpages_desc, ffs_getpages }, /* getpages */
{ &vop_putpages_desc, genfs_putpages }, /* putpages */
{ NULL, NULL }
};
const struct vnodeopv_desc ffs_vnodeop_opv_desc =
{ &ffs_vnodeop_p, ffs_vnodeop_entries };
int (**ffs_specop_p) __P((void *));
const struct vnodeopv_entry_desc ffs_specop_entries[] = {
{ &vop_default_desc, vn_default_error },
{ &vop_lookup_desc, spec_lookup }, /* lookup */
{ &vop_create_desc, spec_create }, /* create */
{ &vop_mknod_desc, spec_mknod }, /* mknod */
{ &vop_open_desc, spec_open }, /* open */
{ &vop_close_desc, ufsspec_close }, /* close */
{ &vop_access_desc, ufs_access }, /* access */
{ &vop_getattr_desc, ufs_getattr }, /* getattr */
{ &vop_setattr_desc, ufs_setattr }, /* setattr */
{ &vop_read_desc, ufsspec_read }, /* read */
{ &vop_write_desc, ufsspec_write }, /* write */
{ &vop_lease_desc, spec_lease_check }, /* lease */
{ &vop_ioctl_desc, spec_ioctl }, /* ioctl */
{ &vop_fcntl_desc, ufs_fcntl }, /* fcntl */
{ &vop_poll_desc, spec_poll }, /* poll */
{ &vop_revoke_desc, spec_revoke }, /* revoke */
{ &vop_mmap_desc, spec_mmap }, /* mmap */
{ &vop_fsync_desc, ffs_fsync }, /* fsync */
{ &vop_seek_desc, spec_seek }, /* seek */
{ &vop_remove_desc, spec_remove }, /* remove */
{ &vop_link_desc, spec_link }, /* link */
{ &vop_rename_desc, spec_rename }, /* rename */
{ &vop_mkdir_desc, spec_mkdir }, /* mkdir */
{ &vop_rmdir_desc, spec_rmdir }, /* rmdir */
{ &vop_symlink_desc, spec_symlink }, /* symlink */
{ &vop_readdir_desc, spec_readdir }, /* readdir */
{ &vop_readlink_desc, spec_readlink }, /* readlink */
{ &vop_abortop_desc, spec_abortop }, /* abortop */
{ &vop_inactive_desc, ufs_inactive }, /* inactive */
{ &vop_reclaim_desc, ffs_reclaim }, /* reclaim */
{ &vop_lock_desc, ufs_lock }, /* lock */
{ &vop_unlock_desc, ufs_unlock }, /* unlock */
{ &vop_bmap_desc, spec_bmap }, /* bmap */
{ &vop_strategy_desc, spec_strategy }, /* strategy */
{ &vop_print_desc, ufs_print }, /* print */
{ &vop_islocked_desc, ufs_islocked }, /* islocked */
{ &vop_pathconf_desc, spec_pathconf }, /* pathconf */
{ &vop_advlock_desc, spec_advlock }, /* advlock */
{ &vop_blkatoff_desc, spec_blkatoff }, /* blkatoff */
{ &vop_valloc_desc, spec_valloc }, /* valloc */
{ &vop_reallocblks_desc, spec_reallocblks }, /* reallocblks */
{ &vop_vfree_desc, ffs_vfree }, /* vfree */
{ &vop_truncate_desc, spec_truncate }, /* truncate */
{ &vop_update_desc, ffs_update }, /* update */
{ &vop_bwrite_desc, vn_bwrite }, /* bwrite */
{ &vop_getpages_desc, spec_getpages }, /* getpages */
{ &vop_putpages_desc, spec_putpages }, /* putpages */
{ NULL, NULL }
};
const struct vnodeopv_desc ffs_specop_opv_desc =
{ &ffs_specop_p, ffs_specop_entries };
int (**ffs_fifoop_p) __P((void *));
const struct vnodeopv_entry_desc ffs_fifoop_entries[] = {
{ &vop_default_desc, vn_default_error },
{ &vop_lookup_desc, fifo_lookup }, /* lookup */
{ &vop_create_desc, fifo_create }, /* create */
{ &vop_mknod_desc, fifo_mknod }, /* mknod */
{ &vop_open_desc, fifo_open }, /* open */
{ &vop_close_desc, ufsfifo_close }, /* close */
{ &vop_access_desc, ufs_access }, /* access */
{ &vop_getattr_desc, ufs_getattr }, /* getattr */
{ &vop_setattr_desc, ufs_setattr }, /* setattr */
{ &vop_read_desc, ufsfifo_read }, /* read */
{ &vop_write_desc, ufsfifo_write }, /* write */
{ &vop_lease_desc, fifo_lease_check }, /* lease */
{ &vop_ioctl_desc, fifo_ioctl }, /* ioctl */
{ &vop_fcntl_desc, ufs_fcntl }, /* fcntl */
{ &vop_poll_desc, fifo_poll }, /* poll */
{ &vop_revoke_desc, fifo_revoke }, /* revoke */
{ &vop_mmap_desc, fifo_mmap }, /* mmap */
{ &vop_fsync_desc, ffs_fsync }, /* fsync */
{ &vop_seek_desc, fifo_seek }, /* seek */
{ &vop_remove_desc, fifo_remove }, /* remove */
{ &vop_link_desc, fifo_link }, /* link */
{ &vop_rename_desc, fifo_rename }, /* rename */
{ &vop_mkdir_desc, fifo_mkdir }, /* mkdir */
{ &vop_rmdir_desc, fifo_rmdir }, /* rmdir */
{ &vop_symlink_desc, fifo_symlink }, /* symlink */
{ &vop_readdir_desc, fifo_readdir }, /* readdir */
{ &vop_readlink_desc, fifo_readlink }, /* readlink */
{ &vop_abortop_desc, fifo_abortop }, /* abortop */
{ &vop_inactive_desc, ufs_inactive }, /* inactive */
{ &vop_reclaim_desc, ffs_reclaim }, /* reclaim */
{ &vop_lock_desc, ufs_lock }, /* lock */
{ &vop_unlock_desc, ufs_unlock }, /* unlock */
{ &vop_bmap_desc, fifo_bmap }, /* bmap */
{ &vop_strategy_desc, fifo_strategy }, /* strategy */
{ &vop_print_desc, ufs_print }, /* print */
{ &vop_islocked_desc, ufs_islocked }, /* islocked */
{ &vop_pathconf_desc, fifo_pathconf }, /* pathconf */
{ &vop_advlock_desc, fifo_advlock }, /* advlock */
{ &vop_blkatoff_desc, fifo_blkatoff }, /* blkatoff */
{ &vop_valloc_desc, fifo_valloc }, /* valloc */
{ &vop_reallocblks_desc, fifo_reallocblks }, /* reallocblks */
{ &vop_vfree_desc, ffs_vfree }, /* vfree */
{ &vop_truncate_desc, fifo_truncate }, /* truncate */
{ &vop_update_desc, ffs_update }, /* update */
{ &vop_bwrite_desc, vn_bwrite }, /* bwrite */
{ NULL, NULL }
};
const struct vnodeopv_desc ffs_fifoop_opv_desc =
{ &ffs_fifoop_p, ffs_fifoop_entries };
int doclusterread = 1;
int doclusterwrite = 1;
#include <ufs/ufs/ufs_readwrite.c>
int
ffs_fsync(v)
void *v;
{
struct vop_fsync_args /* {
struct vnode *a_vp;
struct ucred *a_cred;
int a_flags;
off_t offlo;
off_t offhi;
struct proc *a_p;
} */ *ap = v;
struct buf *bp;
int s, num, error, i;
struct indir ia[NIADDR + 1];
int bsize;
daddr_t blk_low, blk_high;
struct vnode *vp;
/*
* XXX no easy way to sync a range in a file with softdep.
*/
if ((ap->a_offlo == 0 && ap->a_offhi == 0) || DOINGSOFTDEP(ap->a_vp))
return ffs_full_fsync(v);
vp = ap->a_vp;
bsize = ap->a_vp->v_mount->mnt_stat.f_iosize;
blk_low = ap->a_offlo / bsize;
blk_high = ap->a_offhi / bsize;
if (ap->a_offhi % bsize != 0)
blk_high++;
s = splbio();
/*
* First, flush all pages in range.
*/
simple_lock(&vp->v_uobj.vmobjlock);
error = (vp->v_uobj.pgops->pgo_put)(&vp->v_uobj,
ap->a_offlo, ap->a_offhi, PGO_CLEANIT|PGO_SYNCIO);
if (error) {
return error;
}
/*
* Then, flush indirect blocks.
*/
if (!(ap->a_flags & FSYNC_DATAONLY) && blk_high >= NDADDR) {
error = ufs_getlbns(vp, blk_high, ia, &num);
if (error)
return error;
for (i = 0; i < num; i++) {
bp = incore(vp, ia[i].in_lbn);
if (bp != NULL && !(bp->b_flags & B_BUSY) &&
(bp->b_flags & B_DELWRI)) {
bp->b_flags |= B_BUSY | B_VFLUSH;
splx(s);
bawrite(bp);
s = splbio();
}
}
}
if (ap->a_flags & FSYNC_WAIT) {
while (vp->v_numoutput > 0) {
vp->v_flag |= VBWAIT;
tsleep(&vp->v_numoutput, PRIBIO + 1, "fsync_range", 0);
}
}
splx(s);
return (VOP_UPDATE(vp, NULL, NULL,
(ap->a_flags & FSYNC_WAIT) ? UPDATE_WAIT : 0));
}
/*
* Synch an open file.
*/
/* ARGSUSED */
static int
ffs_full_fsync(v)
void *v;
{
struct vop_fsync_args /* {
struct vnode *a_vp;
struct ucred *a_cred;
int a_flags;
off_t offlo;
off_t offhi;
struct proc *a_p;
} */ *ap = v;
struct vnode *vp = ap->a_vp;
struct buf *bp, *nbp;
int s, error, passes, skipmeta;
struct uvm_object *uobj;
if (vp->v_type == VBLK &&
vp->v_specmountpoint != NULL &&
(vp->v_specmountpoint->mnt_flag & MNT_SOFTDEP))
softdep_fsync_mountdev(vp);
/*
* Flush all dirty data associated with a vnode.
*/
if (vp->v_type == VREG) {
uobj = &vp->v_uobj;
simple_lock(&uobj->vmobjlock);
error = (uobj->pgops->pgo_put)(uobj, 0, 0,
PGO_ALLPAGES|PGO_CLEANIT|
((ap->a_flags & FSYNC_WAIT) ? PGO_SYNCIO : 0));
if (error) {
return error;
}
}
passes = NIADDR + 1;
skipmeta = 0;
if (ap->a_flags & (FSYNC_DATAONLY|FSYNC_WAIT))
skipmeta = 1;
s = splbio();
loop:
LIST_FOREACH(bp, &vp->v_dirtyblkhd, b_vnbufs)
bp->b_flags &= ~B_SCANNED;
for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) {
nbp = LIST_NEXT(bp, b_vnbufs);
if (bp->b_flags & (B_BUSY | B_SCANNED))
continue;
if ((bp->b_flags & B_DELWRI) == 0)
panic("ffs_fsync: not dirty");
if (skipmeta && bp->b_lblkno < 0)
continue;
bp->b_flags |= B_BUSY | B_VFLUSH | B_SCANNED;
splx(s);
/*
* On our final pass through, do all I/O synchronously
* so that we can find out if our flush is failing
* because of write errors.
*/
if (passes > 0 || !(ap->a_flags & FSYNC_WAIT))
(void) bawrite(bp);
else if ((error = bwrite(bp)) != 0)
return (error);
s = splbio();
/*
* Since we may have slept during the I/O, we need
* to start from a known point.
*/
nbp = LIST_FIRST(&vp->v_dirtyblkhd);
}
if (skipmeta && !(ap->a_flags & FSYNC_DATAONLY)) {
skipmeta = 0;
goto loop;
}
if (ap->a_flags & FSYNC_WAIT) {
while (vp->v_numoutput) {
vp->v_flag |= VBWAIT;
(void) tsleep(&vp->v_numoutput, PRIBIO + 1,
"ffsfsync", 0);
}
splx(s);
if (ap->a_flags & FSYNC_DATAONLY)
return (0);
/*
* Ensure that any filesystem metadata associated
* with the vnode has been written.
*/
if ((error = softdep_sync_metadata(ap)) != 0)
return (error);
s = splbio();
if (!LIST_EMPTY(&vp->v_dirtyblkhd)) {
/*
* 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--;
goto loop;
}
#ifdef DIAGNOSTIC
if (vp->v_type != VBLK)
vprint("ffs_fsync: dirty", vp);
#endif
}
}
splx(s);
return (VOP_UPDATE(vp, NULL, NULL,
(ap->a_flags & FSYNC_WAIT) ? UPDATE_WAIT : 0));
}
/*
* Reclaim an inode so that it can be used for other purposes.
*/
int
ffs_reclaim(v)
void *v;
{
struct vop_reclaim_args /* {
struct vnode *a_vp;
struct proc *a_p;
} */ *ap = v;
struct vnode *vp = ap->a_vp;
int error;
if ((error = ufs_reclaim(vp, ap->a_p)) != 0)
return (error);
/*
* XXX MFS ends up here, too, to free an inode. Should we create
* XXX a separate pool for MFS inodes?
*/
pool_put(&ffs_inode_pool, vp->v_data);
vp->v_data = NULL;
return (0);
}
int
ffs_getpages(void *v)
{
struct vop_getpages_args /* {
struct vnode *a_vp;
voff_t a_offset;
struct vm_page **a_m;
int *a_count;
int a_centeridx;
vm_prot_t a_access_type;
int a_advice;
int a_flags;
} */ *ap = v;
struct vnode *vp = ap->a_vp;
struct inode *ip = VTOI(vp);
struct fs *fs = ip->i_fs;
/*
* don't allow a softdep write to create pages for only part of a block.
* the dependency tracking requires that all pages be in memory for
* a block involved in a dependency.
*/
if (ap->a_flags & PGO_OVERWRITE &&
(blkoff(fs, ap->a_offset) != 0 ||
blkoff(fs, *ap->a_count << PAGE_SHIFT) != 0) &&
DOINGSOFTDEP(ap->a_vp)) {
if ((ap->a_flags & PGO_LOCKED) == 0) {
simple_unlock(&vp->v_uobj.vmobjlock);
}
return EINVAL;
}
return genfs_getpages(v);
}
/*
* Return the last logical file offset that should be written for this file
* if we're doing a write that ends at "size".
*/
void
ffs_gop_size(struct vnode *vp, off_t size, off_t *eobp)
{
struct inode *ip = VTOI(vp);
struct fs *fs = ip->i_fs;
ufs_lbn_t olbn, nlbn;
olbn = lblkno(fs, ip->i_ffs_size);
nlbn = lblkno(fs, size);
if (nlbn < NDADDR && olbn <= nlbn) {
*eobp = fragroundup(fs, size);
} else {
*eobp = blkroundup(fs, size);
}
}
Reference in New Issue View Git Blame Copy Permalink