2005-02-10 02:02:10 +03:00
|
|
|
/* $NetBSD: ext2fs_balloc.c,v 1.23 2005/02/09 23:02:10 ws Exp $ */
|
1997-06-11 13:33:37 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 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.
|
2003-08-07 20:26:28 +04:00
|
|
|
* 3. 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_balloc.c 8.4 (Berkeley) 9/23/93
|
|
|
|
* Modified for ext2fs by Manuel Bouyer.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Copyright (c) 1997 Manuel Bouyer.
|
|
|
|
*
|
|
|
|
* 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.
|
1997-06-11 13:33:37 +04:00
|
|
|
* 3. All advertising materials mentioning features or use of this software
|
|
|
|
* must display the following acknowledgement:
|
2003-10-05 21:48:49 +04:00
|
|
|
* This product includes software developed by Manuel Bouyer.
|
|
|
|
* 4. The name of the author may not be used to endorse or promote products
|
|
|
|
* derived from this software without specific prior written permission.
|
1997-06-11 13:33:37 +04:00
|
|
|
*
|
2004-03-22 22:23:08 +03:00
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
|
1997-06-11 13:33:37 +04:00
|
|
|
*
|
|
|
|
* @(#)ffs_balloc.c 8.4 (Berkeley) 9/23/93
|
|
|
|
* Modified for ext2fs by Manuel Bouyer.
|
|
|
|
*/
|
|
|
|
|
2001-11-08 05:39:06 +03:00
|
|
|
#include <sys/cdefs.h>
|
2005-02-10 02:02:10 +03:00
|
|
|
__KERNEL_RCSID(0, "$NetBSD: ext2fs_balloc.c,v 1.23 2005/02/09 23:02:10 ws Exp $");
|
2001-11-08 05:39:06 +03:00
|
|
|
|
2001-05-30 15:57:16 +04:00
|
|
|
#if defined(_KERNEL_OPT)
|
2000-11-27 11:39:39 +03:00
|
|
|
#include "opt_uvmhist.h"
|
|
|
|
#endif
|
|
|
|
|
1997-06-11 13:33:37 +04:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/buf.h>
|
|
|
|
#include <sys/proc.h>
|
|
|
|
#include <sys/file.h>
|
|
|
|
#include <sys/vnode.h>
|
2000-11-27 11:39:39 +03:00
|
|
|
#include <sys/mount.h>
|
|
|
|
|
|
|
|
#include <uvm/uvm.h>
|
1997-06-11 13:33:37 +04:00
|
|
|
|
|
|
|
#include <ufs/ufs/inode.h>
|
|
|
|
#include <ufs/ufs/ufs_extern.h>
|
|
|
|
|
|
|
|
#include <ufs/ext2fs/ext2fs.h>
|
|
|
|
#include <ufs/ext2fs/ext2fs_extern.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Balloc defines the structure of file system storage
|
|
|
|
* by allocating the physical blocks on a device given
|
|
|
|
* the inode and the logical block number in a file.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
ext2fs_balloc(ip, bn, size, cred, bpp, flags)
|
2000-03-30 16:41:09 +04:00
|
|
|
struct inode *ip;
|
2003-01-25 00:55:02 +03:00
|
|
|
daddr_t bn;
|
1997-06-11 13:33:37 +04:00
|
|
|
int size;
|
|
|
|
struct ucred *cred;
|
|
|
|
struct buf **bpp;
|
|
|
|
int flags;
|
|
|
|
{
|
2000-03-30 16:41:09 +04:00
|
|
|
struct m_ext2fs *fs;
|
2003-01-25 00:55:02 +03:00
|
|
|
daddr_t nb;
|
1997-06-11 13:33:37 +04:00
|
|
|
struct buf *bp, *nbp;
|
|
|
|
struct vnode *vp = ITOV(ip);
|
|
|
|
struct indir indirs[NIADDR + 2];
|
2003-01-25 00:55:02 +03:00
|
|
|
daddr_t newb, lbn, pref;
|
|
|
|
int32_t *bap; /* XXX ondisk32 */
|
1997-06-11 13:33:37 +04:00
|
|
|
int num, i, error;
|
1998-03-01 05:20:01 +03:00
|
|
|
u_int deallocated;
|
2003-01-25 00:55:02 +03:00
|
|
|
daddr_t *blkp, *allocblk, allociblk[NIADDR + 1];
|
|
|
|
int32_t *allocib; /* XXX ondisk32 */
|
2000-05-28 12:44:32 +04:00
|
|
|
int unwindidx = -1;
|
2000-11-27 11:39:39 +03:00
|
|
|
UVMHIST_FUNC("ext2fs_balloc"); UVMHIST_CALLED(ubchist);
|
1997-06-11 13:33:37 +04:00
|
|
|
|
2000-11-27 11:39:39 +03:00
|
|
|
UVMHIST_LOG(ubchist, "bn 0x%x", bn,0,0,0);
|
|
|
|
|
|
|
|
if (bpp != NULL) {
|
|
|
|
*bpp = NULL;
|
|
|
|
}
|
1997-06-11 13:33:37 +04:00
|
|
|
if (bn < 0)
|
|
|
|
return (EFBIG);
|
|
|
|
fs = ip->i_e2fs;
|
|
|
|
lbn = bn;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The first NDADDR blocks are direct blocks
|
|
|
|
*/
|
|
|
|
if (bn < NDADDR) {
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
1997-10-09 19:42:19 +04:00
|
|
|
nb = fs2h32(ip->i_e2fs_blocks[bn]);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (nb != 0) {
|
2000-11-27 11:39:39 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* the block is already allocated, just read it.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (bpp != NULL) {
|
|
|
|
error = bread(vp, bn, fs->e2fs_bsize, NOCRED,
|
|
|
|
&bp);
|
|
|
|
if (error) {
|
|
|
|
brelse(bp);
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
*bpp = bp;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
|
|
|
return (0);
|
2000-11-27 11:39:39 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* allocate a new direct block.
|
|
|
|
*/
|
|
|
|
|
|
|
|
error = ext2fs_alloc(ip, bn,
|
|
|
|
ext2fs_blkpref(ip, bn, bn, &ip->i_e2fs_blocks[0]),
|
|
|
|
cred, &newb);
|
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
ip->i_e2fs_last_lblk = lbn;
|
|
|
|
ip->i_e2fs_last_blk = newb;
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
|
|
|
ip->i_e2fs_blocks[bn] = h2fs32((int32_t)newb);
|
2000-11-27 11:39:39 +03:00
|
|
|
ip->i_flag |= IN_CHANGE | IN_UPDATE;
|
|
|
|
if (bpp != NULL) {
|
1997-06-11 13:33:37 +04:00
|
|
|
bp = getblk(vp, bn, fs->e2fs_bsize, 0, 0);
|
|
|
|
bp->b_blkno = fsbtodb(fs, newb);
|
|
|
|
if (flags & B_CLRBUF)
|
|
|
|
clrbuf(bp);
|
2000-11-27 11:39:39 +03:00
|
|
|
*bpp = bp;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Determine the number of levels of indirection.
|
|
|
|
*/
|
|
|
|
pref = 0;
|
|
|
|
if ((error = ufs_getlbns(vp, bn, indirs, &num)) != 0)
|
|
|
|
return(error);
|
|
|
|
#ifdef DIAGNOSTIC
|
|
|
|
if (num < 1)
|
|
|
|
panic ("ext2fs_balloc: ufs_getlbns returned indirect block\n");
|
|
|
|
#endif
|
|
|
|
/*
|
|
|
|
* Fetch the first indirect block allocating if necessary.
|
|
|
|
*/
|
|
|
|
--num;
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
1997-10-09 19:42:19 +04:00
|
|
|
nb = fs2h32(ip->i_e2fs_blocks[NDADDR + indirs[0].in_off]);
|
1998-03-01 05:20:01 +03:00
|
|
|
allocib = NULL;
|
|
|
|
allocblk = allociblk;
|
1997-06-11 13:33:37 +04:00
|
|
|
if (nb == 0) {
|
2003-01-25 00:55:02 +03:00
|
|
|
pref = ext2fs_blkpref(ip, lbn, 0, (int32_t *)0);
|
2000-05-28 12:44:32 +04:00
|
|
|
error = ext2fs_alloc(ip, lbn, pref, cred, &newb);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
nb = newb;
|
1998-03-01 05:20:01 +03:00
|
|
|
*allocblk++ = nb;
|
1997-06-11 13:33:37 +04:00
|
|
|
ip->i_e2fs_last_blk = newb;
|
|
|
|
bp = getblk(vp, indirs[1].in_lbn, fs->e2fs_bsize, 0, 0);
|
|
|
|
bp->b_blkno = fsbtodb(fs, newb);
|
|
|
|
clrbuf(bp);
|
|
|
|
/*
|
|
|
|
* Write synchronously so that indirect blocks
|
|
|
|
* never point at garbage.
|
|
|
|
*/
|
1998-03-01 05:20:01 +03:00
|
|
|
if ((error = bwrite(bp)) != 0)
|
|
|
|
goto fail;
|
2000-05-28 12:44:32 +04:00
|
|
|
unwindidx = 0;
|
1998-03-01 05:20:01 +03:00
|
|
|
allocib = &ip->i_e2fs_blocks[NDADDR + indirs[0].in_off];
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
|
|
|
*allocib = h2fs32((int32_t)newb);
|
1997-06-11 13:33:37 +04:00
|
|
|
ip->i_flag |= IN_CHANGE | IN_UPDATE;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Fetch through the indirect blocks, allocating as necessary.
|
|
|
|
*/
|
|
|
|
for (i = 1;;) {
|
|
|
|
error = bread(vp,
|
2000-05-28 12:44:32 +04:00
|
|
|
indirs[i].in_lbn, (int)fs->e2fs_bsize, NOCRED, &bp);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (error) {
|
|
|
|
brelse(bp);
|
1998-03-01 05:20:01 +03:00
|
|
|
goto fail;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
2003-01-25 00:55:02 +03:00
|
|
|
bap = (int32_t *)bp->b_data; /* XXX ondisk32 */
|
1997-10-09 19:42:19 +04:00
|
|
|
nb = fs2h32(bap[indirs[i].in_off]);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (i == num)
|
|
|
|
break;
|
2000-05-28 12:44:32 +04:00
|
|
|
i++;
|
1997-06-11 13:33:37 +04:00
|
|
|
if (nb != 0) {
|
|
|
|
brelse(bp);
|
|
|
|
continue;
|
|
|
|
}
|
2003-01-25 00:55:02 +03:00
|
|
|
pref = ext2fs_blkpref(ip, lbn, 0, (int32_t *)0);
|
2000-05-28 12:44:32 +04:00
|
|
|
error = ext2fs_alloc(ip, lbn, pref, cred, &newb);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (error) {
|
|
|
|
brelse(bp);
|
1998-03-01 05:20:01 +03:00
|
|
|
goto fail;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
|
|
|
nb = newb;
|
1998-03-01 05:20:01 +03:00
|
|
|
*allocblk++ = nb;
|
1997-06-11 13:33:37 +04:00
|
|
|
ip->i_e2fs_last_blk = newb;
|
|
|
|
nbp = getblk(vp, indirs[i].in_lbn, fs->e2fs_bsize, 0, 0);
|
|
|
|
nbp->b_blkno = fsbtodb(fs, nb);
|
|
|
|
clrbuf(nbp);
|
|
|
|
/*
|
|
|
|
* Write synchronously so that indirect blocks
|
|
|
|
* never point at garbage.
|
|
|
|
*/
|
|
|
|
if ((error = bwrite(nbp)) != 0) {
|
|
|
|
brelse(bp);
|
1998-03-01 05:20:01 +03:00
|
|
|
goto fail;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
2000-05-28 12:44:32 +04:00
|
|
|
if (unwindidx < 0)
|
|
|
|
unwindidx = i - 1;
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
|
|
|
bap[indirs[i - 1].in_off] = h2fs32((int32_t)nb);
|
1997-06-11 13:33:37 +04:00
|
|
|
/*
|
|
|
|
* If required, write synchronously, otherwise use
|
|
|
|
* delayed write.
|
|
|
|
*/
|
|
|
|
if (flags & B_SYNC) {
|
|
|
|
bwrite(bp);
|
|
|
|
} else {
|
|
|
|
bdwrite(bp);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Get the data block, allocating if necessary.
|
|
|
|
*/
|
|
|
|
if (nb == 0) {
|
2000-05-28 12:44:32 +04:00
|
|
|
pref = ext2fs_blkpref(ip, lbn, indirs[num].in_off, &bap[0]);
|
|
|
|
error = ext2fs_alloc(ip, lbn, pref, cred, &newb);
|
1997-06-11 13:33:37 +04:00
|
|
|
if (error) {
|
|
|
|
brelse(bp);
|
1998-03-01 05:20:01 +03:00
|
|
|
goto fail;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
|
|
|
nb = newb;
|
1998-03-01 05:20:01 +03:00
|
|
|
*allocblk++ = nb;
|
1997-06-11 13:33:37 +04:00
|
|
|
ip->i_e2fs_last_lblk = lbn;
|
|
|
|
ip->i_e2fs_last_blk = newb;
|
2003-01-25 00:55:02 +03:00
|
|
|
/* XXX ondisk32 */
|
|
|
|
bap[indirs[num].in_off] = h2fs32((int32_t)nb);
|
1997-06-11 13:33:37 +04:00
|
|
|
/*
|
|
|
|
* If required, write synchronously, otherwise use
|
|
|
|
* delayed write.
|
|
|
|
*/
|
|
|
|
if (flags & B_SYNC) {
|
|
|
|
bwrite(bp);
|
|
|
|
} else {
|
|
|
|
bdwrite(bp);
|
|
|
|
}
|
2000-11-27 11:39:39 +03:00
|
|
|
if (bpp != NULL) {
|
|
|
|
nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0);
|
|
|
|
nbp->b_blkno = fsbtodb(fs, nb);
|
|
|
|
if (flags & B_CLRBUF)
|
|
|
|
clrbuf(nbp);
|
|
|
|
*bpp = nbp;
|
|
|
|
}
|
1997-06-11 13:33:37 +04:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
brelse(bp);
|
2000-11-27 11:39:39 +03:00
|
|
|
if (bpp != NULL) {
|
|
|
|
if (flags & B_CLRBUF) {
|
|
|
|
error = bread(vp, lbn, (int)fs->e2fs_bsize, NOCRED,
|
|
|
|
&nbp);
|
|
|
|
if (error) {
|
|
|
|
brelse(nbp);
|
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
nbp = getblk(vp, lbn, fs->e2fs_bsize, 0, 0);
|
|
|
|
nbp->b_blkno = fsbtodb(fs, nb);
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
2000-11-27 11:39:39 +03:00
|
|
|
*bpp = nbp;
|
1997-06-11 13:33:37 +04:00
|
|
|
}
|
|
|
|
return (0);
|
1998-03-01 05:20:01 +03:00
|
|
|
fail:
|
|
|
|
/*
|
|
|
|
* If we have failed part way through block allocation, we
|
|
|
|
* have to deallocate any indirect blocks that we have allocated.
|
|
|
|
*/
|
|
|
|
for (deallocated = 0, blkp = allociblk; blkp < allocblk; blkp++) {
|
|
|
|
ext2fs_blkfree(ip, *blkp);
|
|
|
|
deallocated += fs->e2fs_bsize;
|
|
|
|
}
|
2000-05-28 12:44:32 +04:00
|
|
|
if (unwindidx >= 0) {
|
|
|
|
if (unwindidx == 0) {
|
|
|
|
*allocib = 0;
|
|
|
|
} else {
|
|
|
|
int r;
|
|
|
|
|
|
|
|
r = bread(vp, indirs[unwindidx].in_lbn,
|
|
|
|
(int)fs->e2fs_bsize, NOCRED, &bp);
|
|
|
|
if (r) {
|
|
|
|
panic("Could not unwind indirect block, error %d", r);
|
|
|
|
brelse(bp);
|
|
|
|
} else {
|
2003-01-25 00:55:02 +03:00
|
|
|
bap = (int32_t *)bp->b_data; /* XXX ondisk32 */
|
2000-05-28 12:44:32 +04:00
|
|
|
bap[indirs[unwindidx].in_off] = 0;
|
|
|
|
if (flags & B_SYNC)
|
|
|
|
bwrite(bp);
|
|
|
|
else
|
|
|
|
bdwrite(bp);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
for (i = unwindidx + 1; i <= num; i++) {
|
|
|
|
bp = getblk(vp, indirs[i].in_lbn, (int)fs->e2fs_bsize,
|
|
|
|
0, 0);
|
|
|
|
bp->b_flags |= B_INVAL;
|
|
|
|
brelse(bp);
|
|
|
|
}
|
|
|
|
}
|
1998-03-01 05:20:01 +03:00
|
|
|
if (deallocated) {
|
|
|
|
ip->i_e2fs_nblock -= btodb(deallocated);
|
|
|
|
ip->i_e2fs_flags |= IN_CHANGE | IN_UPDATE;
|
|
|
|
}
|
2000-11-27 11:39:39 +03:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
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.
2001-09-16 00:36:31 +04:00
|
|
|
ext2fs_gop_alloc(struct vnode *vp, off_t off, off_t len, int flags,
|
|
|
|
struct ucred *cred)
|
2000-11-27 11:39:39 +03:00
|
|
|
{
|
|
|
|
struct inode *ip = VTOI(vp);
|
|
|
|
struct m_ext2fs *fs = ip->i_e2fs;
|
|
|
|
int error, delta, bshift, bsize;
|
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.
2001-09-16 00:36:31 +04:00
|
|
|
UVMHIST_FUNC("ext2fs_gop_alloc"); UVMHIST_CALLED(ubchist);
|
2000-11-27 11:39:39 +03:00
|
|
|
|
|
|
|
bshift = fs->e2fs_bshift;
|
|
|
|
bsize = 1 << bshift;
|
|
|
|
|
|
|
|
delta = off & (bsize - 1);
|
|
|
|
off -= delta;
|
|
|
|
len += delta;
|
|
|
|
|
|
|
|
while (len > 0) {
|
|
|
|
bsize = min(bsize, len);
|
|
|
|
UVMHIST_LOG(ubchist, "off 0x%x len 0x%x bsize 0x%x",
|
|
|
|
off, len, bsize, 0);
|
|
|
|
|
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.
2001-09-16 00:36:31 +04:00
|
|
|
error = ext2fs_balloc(ip, lblkno(fs, off), bsize, cred,
|
|
|
|
NULL, flags);
|
2000-11-27 11:39:39 +03:00
|
|
|
if (error) {
|
|
|
|
UVMHIST_LOG(ubchist, "error %d", error, 0,0,0);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* increase file size now, VOP_BALLOC() requires that
|
|
|
|
* EOF be up-to-date before each call.
|
|
|
|
*/
|
|
|
|
|
2005-02-10 02:02:10 +03:00
|
|
|
if (ext2fs_size(ip) < off + bsize) {
|
|
|
|
UVMHIST_LOG(ubchist, "old 0x%lx%8lx new 0x%lx%8lx",
|
|
|
|
/* Note that arguments are always cast to u_long. */
|
|
|
|
ext2fs_size(ip) >> 32,
|
|
|
|
ext2fs_size(ip) & 0xffffffff,
|
|
|
|
(off + bsize) >> 32,
|
|
|
|
(off + bsize) & 0xffffffff);
|
|
|
|
error = ext2fs_setsize(ip, off + bsize);
|
|
|
|
if (error) {
|
|
|
|
UVMHIST_LOG(ubcist, "error %d", error, 0,0,0);
|
|
|
|
return error;
|
|
|
|
}
|
2000-11-27 11:39:39 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
off += bsize;
|
|
|
|
len -= bsize;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|