NetBSD/sys/ufs/lfs/lfs_balloc.c

431 lines
13 KiB
C
Raw Normal View History

/* $NetBSD: lfs_balloc.c,v 1.35 2003/01/24 21:55:26 fvdl Exp $ */
/*-
Various bug-fixes to LFS, to wit: Kernel: * Add runtime quantity lfs_ravail, the number of disk-blocks reserved for writing. Writes to the filesystem first reserve a maximum amount of blocks before their write is allowed to proceed; after the blocks are allocated the reserved total is reduced by a corresponding amount. If the lfs_reserve function cannot immediately reserve the requested number of blocks, the inode is unlocked, and the thread sleeps until the cleaner has made enough space available for the blocks to be reserved. In this way large files can be written to the filesystem (or, smaller files can be written to a nearly-full but thoroughly clean filesystem) and the cleaner can still function properly. * Remove explicit switching on dlfs_minfreeseg from the kernel code; it is now merely a fs-creation parameter used to compute dlfs_avail and dlfs_bfree (and used by fsck_lfs(8) to check their accuracy). Its former role is better assumed by a properly computed dlfs_avail. * Bounds-check inode numbers submitted through lfs_bmapv and lfs_markv. This prevents a panic, but, if the cleaner is feeding the filesystem the wrong data, you are still in a world of hurt. * Cleanup: remove explicit references of DEV_BSIZE in favor of btodb()/dbtob(). lfs_cleanerd: * Make -n mean "send N segments' blocks through a single call to lfs_markv". Previously it had meant "clean N segments though N calls to lfs_markv, before looking again to see if more need to be cleaned". The new behavior gives better packing of direct data on disk with as little metadata as possible, largely alleviating the problem that the cleaner can consume more disk through inefficient use of metadata than it frees by moving dirty data away from clean "holes" to produce entirely clean segments. * Make -b mean "read as many segments as necessary to write N segments of dirty data back to disk", rather than its former meaning of "read as many segments as necessary to free N segments worth of space". The new meaning, combined with the new -n behavior described above, further aids in cleaning storage efficiency as entire segments can be written at once, using as few blocks as possible for segment summaries and inode blocks. * Make the cleaner take note of segments which could not be cleaned due to error, and not attempt to clean them until they are entirely free of dirty blocks. This prevents the case in which a cleanerd running with -n 1 and without -b (formerly the default) would spin trying repeatedly to clean a corrupt segment, while the remaining space filled and deadlocked the filesystem. * Update the lfs_cleanerd manual page to describe all the options, including the changes mentioned here (in particular, the -b and -n flags were previously undocumented). fsck_lfs: * Check, and optionally fix, lfs_avail (to an exact figure) and lfs_bfree (within a margin of error) in pass 5. newfs_lfs: * Reduce the default dlfs_minfreeseg to 1/20 of the total segments. * Add a warning if the sgs disklabel field is 16 (the default for FFS' cpg, but not usually desirable for LFS' sgs: 5--8 is a better range). * Change the calculation of lfs_avail and lfs_bfree, corresponding to the kernel changes mentioned above. mount_lfs: * Add -N and -b options to pass corresponding -n and -b options to lfs_cleanerd. * Default to calling lfs_cleanerd with "-b -n 4". [All of these changes were largely tested in the 1.5 branch, with the idea that they (along with previous un-pulled-up work) could be applied to the branch while it was still in ALPHA2; however my test system has experienced corruption on another filesystem (/dev/console has gone missing :^), and, while I believe this unrelated to the LFS changes, I cannot with good conscience request that the changes be pulled up.]
2000-09-09 08:49:54 +04:00
* Copyright (c) 1999, 2000 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Konrad E. Schroder <perseant@hhhh.org>.
*
* 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 NetBSD
* Foundation, Inc. and its contributors.
* 4. Neither the name of The NetBSD Foundation 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
*/
/*
* Copyright (c) 1989, 1991, 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.
*
1998-03-01 05:20:01 +03:00
* @(#)lfs_balloc.c 8.4 (Berkeley) 5/8/95
*/
1998-06-08 08:27:50 +04:00
2001-11-08 05:39:06 +03:00
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: lfs_balloc.c,v 1.35 2003/01/24 21:55:26 fvdl Exp $");
2001-11-08 05:39:06 +03:00
2001-05-30 15:57:16 +04:00
#if defined(_KERNEL_OPT)
1998-06-08 08:27:50 +04:00
#include "opt_quota.h"
#endif
1998-06-08 08:27:50 +04:00
#include <sys/param.h>
1996-02-10 01:28:45 +03:00
#include <sys/systm.h>
#include <sys/buf.h>
#include <sys/proc.h>
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/resourcevar.h>
#include <sys/trace.h>
#include <miscfs/specfs/specdev.h>
#include <ufs/ufs/quota.h>
#include <ufs/ufs/inode.h>
#include <ufs/ufs/ufsmount.h>
1996-02-10 01:28:45 +03:00
#include <ufs/ufs/ufs_extern.h>
#include <ufs/lfs/lfs.h>
#include <ufs/lfs/lfs_extern.h>
int lfs_fragextend(struct vnode *, int, int, daddr_t, struct buf **, struct ucred *);
1998-03-01 05:20:01 +03:00
/*
* Allocate a block, and to inode and filesystem block accounting for it
* and for any indirect blocks the may need to be created in order for
* this block to be created.
*
* Blocks which have never been accounted for (i.e., which "do not exist")
* have disk address 0, which is translated by ufs_bmap to the special value
* UNASSIGNED == -1, as in the historical UFS.
*
* Blocks which have been accounted for but which have not yet been written
* to disk are given the new special disk address UNWRITTEN == -2, so that
* they can be differentiated from completely new blocks.
*/
Various bug-fixes to LFS, to wit: Kernel: * Add runtime quantity lfs_ravail, the number of disk-blocks reserved for writing. Writes to the filesystem first reserve a maximum amount of blocks before their write is allowed to proceed; after the blocks are allocated the reserved total is reduced by a corresponding amount. If the lfs_reserve function cannot immediately reserve the requested number of blocks, the inode is unlocked, and the thread sleeps until the cleaner has made enough space available for the blocks to be reserved. In this way large files can be written to the filesystem (or, smaller files can be written to a nearly-full but thoroughly clean filesystem) and the cleaner can still function properly. * Remove explicit switching on dlfs_minfreeseg from the kernel code; it is now merely a fs-creation parameter used to compute dlfs_avail and dlfs_bfree (and used by fsck_lfs(8) to check their accuracy). Its former role is better assumed by a properly computed dlfs_avail. * Bounds-check inode numbers submitted through lfs_bmapv and lfs_markv. This prevents a panic, but, if the cleaner is feeding the filesystem the wrong data, you are still in a world of hurt. * Cleanup: remove explicit references of DEV_BSIZE in favor of btodb()/dbtob(). lfs_cleanerd: * Make -n mean "send N segments' blocks through a single call to lfs_markv". Previously it had meant "clean N segments though N calls to lfs_markv, before looking again to see if more need to be cleaned". The new behavior gives better packing of direct data on disk with as little metadata as possible, largely alleviating the problem that the cleaner can consume more disk through inefficient use of metadata than it frees by moving dirty data away from clean "holes" to produce entirely clean segments. * Make -b mean "read as many segments as necessary to write N segments of dirty data back to disk", rather than its former meaning of "read as many segments as necessary to free N segments worth of space". The new meaning, combined with the new -n behavior described above, further aids in cleaning storage efficiency as entire segments can be written at once, using as few blocks as possible for segment summaries and inode blocks. * Make the cleaner take note of segments which could not be cleaned due to error, and not attempt to clean them until they are entirely free of dirty blocks. This prevents the case in which a cleanerd running with -n 1 and without -b (formerly the default) would spin trying repeatedly to clean a corrupt segment, while the remaining space filled and deadlocked the filesystem. * Update the lfs_cleanerd manual page to describe all the options, including the changes mentioned here (in particular, the -b and -n flags were previously undocumented). fsck_lfs: * Check, and optionally fix, lfs_avail (to an exact figure) and lfs_bfree (within a margin of error) in pass 5. newfs_lfs: * Reduce the default dlfs_minfreeseg to 1/20 of the total segments. * Add a warning if the sgs disklabel field is 16 (the default for FFS' cpg, but not usually desirable for LFS' sgs: 5--8 is a better range). * Change the calculation of lfs_avail and lfs_bfree, corresponding to the kernel changes mentioned above. mount_lfs: * Add -N and -b options to pass corresponding -n and -b options to lfs_cleanerd. * Default to calling lfs_cleanerd with "-b -n 4". [All of these changes were largely tested in the 1.5 branch, with the idea that they (along with previous un-pulled-up work) could be applied to the branch while it was still in ALPHA2; however my test system has experienced corruption on another filesystem (/dev/console has gone missing :^), and, while I believe this unrelated to the LFS changes, I cannot with good conscience request that the changes be pulled up.]
2000-09-09 08:49:54 +04:00
/* VOP_BWRITE NIADDR+2 times */
int
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
lfs_balloc(void *v)
{
struct vop_balloc_args /* {
struct vnode *a_vp;
off_t a_startoffset;
int a_size;
struct ucred *a_cred;
int a_flags;
struct buf *a_bpp;
} */ *ap = v;
struct vnode *vp;
1998-03-01 05:20:01 +03:00
int offset;
u_long iosize;
daddr_t daddr, idaddr;
struct buf *ibp, *bp;
struct inode *ip;
struct lfs *fs;
struct indir indirs[NIADDR+2], *idp;
daddr_t lbn, lastblock;
int bb, bcount;
int error, frags, i, nsize, osize, num;
vp = ap->a_vp;
ip = VTOI(vp);
fs = ip->i_lfs;
offset = blkoff(fs, ap->a_startoffset);
iosize = ap->a_size;
lbn = lblkno(fs, ap->a_startoffset);
(void)lfs_check(vp, lbn, 0);
/*
* Three cases: it's a block beyond the end of file, it's a block in
* the file that may or may not have been assigned a disk address or
* we're writing an entire block.
*
* Note, if the daddr is UNWRITTEN, the block already exists in
* the cache (it was read or written earlier). If so, make sure
* we don't count it as a new block or zero out its contents. If
* it did not, make sure we allocate any necessary indirect
* blocks.
*
1998-03-01 05:20:01 +03:00
* If we are writing a block beyond the end of the file, we need to
* check if the old last block was a fragment. If it was, we need
1998-03-01 05:20:01 +03:00
* to rewrite it.
*/
*ap->a_bpp = NULL;
1998-03-01 05:20:01 +03:00
/* Check for block beyond end of file and fragment extension needed. */
lastblock = lblkno(fs, ip->i_ffs_size);
if (lastblock < NDADDR && lastblock < lbn) {
osize = blksize(fs, ip, lastblock);
if (osize < fs->lfs_bsize && osize > 0) {
if ((error = lfs_fragextend(vp, osize, fs->lfs_bsize,
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
lastblock, &bp,
ap->a_cred)))
return (error);
1998-03-01 05:20:01 +03:00
ip->i_ffs_size = (lastblock + 1) * fs->lfs_bsize;
uvm_vnp_setsize(vp, ip->i_ffs_size);
1998-03-01 05:20:01 +03:00
ip->i_flag |= IN_CHANGE | IN_UPDATE;
Various bug-fixes to LFS, to wit: Kernel: * Add runtime quantity lfs_ravail, the number of disk-blocks reserved for writing. Writes to the filesystem first reserve a maximum amount of blocks before their write is allowed to proceed; after the blocks are allocated the reserved total is reduced by a corresponding amount. If the lfs_reserve function cannot immediately reserve the requested number of blocks, the inode is unlocked, and the thread sleeps until the cleaner has made enough space available for the blocks to be reserved. In this way large files can be written to the filesystem (or, smaller files can be written to a nearly-full but thoroughly clean filesystem) and the cleaner can still function properly. * Remove explicit switching on dlfs_minfreeseg from the kernel code; it is now merely a fs-creation parameter used to compute dlfs_avail and dlfs_bfree (and used by fsck_lfs(8) to check their accuracy). Its former role is better assumed by a properly computed dlfs_avail. * Bounds-check inode numbers submitted through lfs_bmapv and lfs_markv. This prevents a panic, but, if the cleaner is feeding the filesystem the wrong data, you are still in a world of hurt. * Cleanup: remove explicit references of DEV_BSIZE in favor of btodb()/dbtob(). lfs_cleanerd: * Make -n mean "send N segments' blocks through a single call to lfs_markv". Previously it had meant "clean N segments though N calls to lfs_markv, before looking again to see if more need to be cleaned". The new behavior gives better packing of direct data on disk with as little metadata as possible, largely alleviating the problem that the cleaner can consume more disk through inefficient use of metadata than it frees by moving dirty data away from clean "holes" to produce entirely clean segments. * Make -b mean "read as many segments as necessary to write N segments of dirty data back to disk", rather than its former meaning of "read as many segments as necessary to free N segments worth of space". The new meaning, combined with the new -n behavior described above, further aids in cleaning storage efficiency as entire segments can be written at once, using as few blocks as possible for segment summaries and inode blocks. * Make the cleaner take note of segments which could not be cleaned due to error, and not attempt to clean them until they are entirely free of dirty blocks. This prevents the case in which a cleanerd running with -n 1 and without -b (formerly the default) would spin trying repeatedly to clean a corrupt segment, while the remaining space filled and deadlocked the filesystem. * Update the lfs_cleanerd manual page to describe all the options, including the changes mentioned here (in particular, the -b and -n flags were previously undocumented). fsck_lfs: * Check, and optionally fix, lfs_avail (to an exact figure) and lfs_bfree (within a margin of error) in pass 5. newfs_lfs: * Reduce the default dlfs_minfreeseg to 1/20 of the total segments. * Add a warning if the sgs disklabel field is 16 (the default for FFS' cpg, but not usually desirable for LFS' sgs: 5--8 is a better range). * Change the calculation of lfs_avail and lfs_bfree, corresponding to the kernel changes mentioned above. mount_lfs: * Add -N and -b options to pass corresponding -n and -b options to lfs_cleanerd. * Default to calling lfs_cleanerd with "-b -n 4". [All of these changes were largely tested in the 1.5 branch, with the idea that they (along with previous un-pulled-up work) could be applied to the branch while it was still in ALPHA2; however my test system has experienced corruption on another filesystem (/dev/console has gone missing :^), and, while I believe this unrelated to the LFS changes, I cannot with good conscience request that the changes be pulled up.]
2000-09-09 08:49:54 +04:00
(void) VOP_BWRITE(bp);
1998-03-01 05:20:01 +03:00
}
}
/*
* If the block we are writing is a direct block, it's the last
* block in the file, and offset + iosize is less than a full
* block, we can write one or more fragments. There are two cases:
* the block is brand new and we should allocate it the correct
* size or it already exists and contains some fragments and
* may need to extend it.
*/
if (lbn < NDADDR && lblkno(fs, ip->i_ffs_size) <= lbn) {
osize = blksize(fs, ip, lbn);
1998-03-01 05:20:01 +03:00
nsize = fragroundup(fs, offset + iosize);
if (lblktosize(fs, lbn) >= ip->i_ffs_size) {
1998-03-01 05:20:01 +03:00
/* Brand new block or fragment */
frags = numfrags(fs, nsize);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
bb = fragstofsb(fs, frags);
*ap->a_bpp = bp = getblk(vp, lbn, nsize, 0, 0);
if (ap->a_flags & B_CLRBUF)
clrbuf(bp);
ip->i_lfs_effnblks += bb;
ip->i_lfs->lfs_bfree -= bb;
ip->i_ffs_db[lbn] = bp->b_blkno = UNWRITTEN;
} else {
if (nsize <= osize) {
/* No need to extend */
if ((error = bread(vp, lbn, osize, NOCRED, &bp)))
return error;
} else {
/* Extend existing block */
if ((error =
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
lfs_fragextend(vp, osize, nsize, lbn, &bp,
ap->a_cred)))
return error;
}
*ap->a_bpp = bp;
1998-03-01 05:20:01 +03:00
}
return 0;
}
error = ufs_bmaparray(vp, lbn, &daddr, &indirs[0], &num, NULL );
if (error)
return (error);
/*
* Do byte accounting all at once, so we can gracefully fail *before*
* we start assigning blocks.
*/
bb = VFSTOUFS(vp->v_mount)->um_seqinc;
bcount = 0;
if (daddr == UNASSIGNED) {
bcount = bb;
}
for (i = 1; i < num; ++i) {
if (!indirs[i].in_exists) {
bcount += bb;
}
}
if (ISSPACE(fs, bcount, ap->a_cred)) {
ip->i_lfs->lfs_bfree -= bcount;
ip->i_lfs_effnblks += bcount;
1998-03-01 05:20:01 +03:00
} else {
return ENOSPC;
}
if (daddr == UNASSIGNED) {
if (num > 0 && ip->i_ffs_ib[indirs[0].in_off] == 0) {
ip->i_ffs_ib[indirs[0].in_off] = UNWRITTEN;
}
1998-03-01 05:20:01 +03:00
/*
* Create new indirect blocks if necessary
1998-03-01 05:20:01 +03:00
*/
if (num > 1)
idaddr = ip->i_ffs_ib[indirs[0].in_off];
for (i = 1; i < num; ++i) {
ibp = getblk(vp, indirs[i].in_lbn, fs->lfs_bsize, 0,0);
if (!indirs[i].in_exists) {
clrbuf(ibp);
ibp->b_blkno = UNWRITTEN;
} else if (!(ibp->b_flags & (B_DELWRI | B_DONE))) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
ibp->b_blkno = fsbtodb(fs, idaddr);
ibp->b_flags |= B_READ;
VOP_STRATEGY(ibp);
biowait(ibp);
}
/*
* This block exists, but the next one may not.
* If that is the case mark it UNWRITTEN to keep
* the accounting straight.
*/
/* XXX ondisk32 */
if (((int32_t *)ibp->b_data)[indirs[i].in_off] == 0)
((int32_t *)ibp->b_data)[indirs[i].in_off] =
UNWRITTEN;
/* XXX ondisk32 */
idaddr = ((int32_t *)ibp->b_data)[indirs[i].in_off];
if ((error = VOP_BWRITE(ibp))) {
return error;
}
}
}
/*
* Get the existing block from the cache.
*/
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
frags = fsbtofrags(fs, bb);
*ap->a_bpp = bp = getblk(vp, lbn, blksize(fs, ip, lbn), 0, 0);
1998-03-01 05:20:01 +03:00
/*
* The block we are writing may be a brand new block
* in which case we need to do accounting.
*
* We can tell a truly new block because ufs_bmaparray will say
* it is UNASSIGNED. Once we allocate it we will assign it the
* disk address UNWRITTEN.
1998-03-01 05:20:01 +03:00
*/
if (daddr == UNASSIGNED) {
if (ap->a_flags & B_CLRBUF)
clrbuf(bp);
/* Note the new address */
bp->b_blkno = UNWRITTEN;
switch (num) {
case 0:
ip->i_ffs_db[lbn] = UNWRITTEN;
break;
case 1:
ip->i_ffs_ib[indirs[0].in_off] = UNWRITTEN;
break;
default:
idp = &indirs[num - 1];
if (bread(vp, idp->in_lbn, fs->lfs_bsize, NOCRED,
&ibp))
panic("lfs_balloc: bread bno %lld",
(long long)idp->in_lbn);
/* XXX ondisk32 */
((int32_t *)ibp->b_data)[idp->in_off] = UNWRITTEN;
VOP_BWRITE(ibp);
}
} else if (!(bp->b_flags & (B_DONE|B_DELWRI))) {
/*
* Not a brand new block, also not in the cache;
* read it in from disk.
*/
if (iosize == fs->lfs_bsize)
1998-03-01 05:20:01 +03:00
/* Optimization: I/O is unnecessary. */
bp->b_blkno = daddr;
else {
1998-03-01 05:20:01 +03:00
/*
* We need to read the block to preserve the
* existing bytes.
*/
bp->b_blkno = daddr;
bp->b_flags |= B_READ;
VOP_STRATEGY(bp);
return (biowait(bp));
}
}
1998-03-01 05:20:01 +03:00
return (0);
}
Various bug-fixes to LFS, to wit: Kernel: * Add runtime quantity lfs_ravail, the number of disk-blocks reserved for writing. Writes to the filesystem first reserve a maximum amount of blocks before their write is allowed to proceed; after the blocks are allocated the reserved total is reduced by a corresponding amount. If the lfs_reserve function cannot immediately reserve the requested number of blocks, the inode is unlocked, and the thread sleeps until the cleaner has made enough space available for the blocks to be reserved. In this way large files can be written to the filesystem (or, smaller files can be written to a nearly-full but thoroughly clean filesystem) and the cleaner can still function properly. * Remove explicit switching on dlfs_minfreeseg from the kernel code; it is now merely a fs-creation parameter used to compute dlfs_avail and dlfs_bfree (and used by fsck_lfs(8) to check their accuracy). Its former role is better assumed by a properly computed dlfs_avail. * Bounds-check inode numbers submitted through lfs_bmapv and lfs_markv. This prevents a panic, but, if the cleaner is feeding the filesystem the wrong data, you are still in a world of hurt. * Cleanup: remove explicit references of DEV_BSIZE in favor of btodb()/dbtob(). lfs_cleanerd: * Make -n mean "send N segments' blocks through a single call to lfs_markv". Previously it had meant "clean N segments though N calls to lfs_markv, before looking again to see if more need to be cleaned". The new behavior gives better packing of direct data on disk with as little metadata as possible, largely alleviating the problem that the cleaner can consume more disk through inefficient use of metadata than it frees by moving dirty data away from clean "holes" to produce entirely clean segments. * Make -b mean "read as many segments as necessary to write N segments of dirty data back to disk", rather than its former meaning of "read as many segments as necessary to free N segments worth of space". The new meaning, combined with the new -n behavior described above, further aids in cleaning storage efficiency as entire segments can be written at once, using as few blocks as possible for segment summaries and inode blocks. * Make the cleaner take note of segments which could not be cleaned due to error, and not attempt to clean them until they are entirely free of dirty blocks. This prevents the case in which a cleanerd running with -n 1 and without -b (formerly the default) would spin trying repeatedly to clean a corrupt segment, while the remaining space filled and deadlocked the filesystem. * Update the lfs_cleanerd manual page to describe all the options, including the changes mentioned here (in particular, the -b and -n flags were previously undocumented). fsck_lfs: * Check, and optionally fix, lfs_avail (to an exact figure) and lfs_bfree (within a margin of error) in pass 5. newfs_lfs: * Reduce the default dlfs_minfreeseg to 1/20 of the total segments. * Add a warning if the sgs disklabel field is 16 (the default for FFS' cpg, but not usually desirable for LFS' sgs: 5--8 is a better range). * Change the calculation of lfs_avail and lfs_bfree, corresponding to the kernel changes mentioned above. mount_lfs: * Add -N and -b options to pass corresponding -n and -b options to lfs_cleanerd. * Default to calling lfs_cleanerd with "-b -n 4". [All of these changes were largely tested in the 1.5 branch, with the idea that they (along with previous un-pulled-up work) could be applied to the branch while it was still in ALPHA2; however my test system has experienced corruption on another filesystem (/dev/console has gone missing :^), and, while I believe this unrelated to the LFS changes, I cannot with good conscience request that the changes be pulled up.]
2000-09-09 08:49:54 +04:00
/* VOP_BWRITE 1 time */
1998-03-01 05:20:01 +03:00
int
lfs_fragextend(struct vnode *vp, int osize, int nsize, daddr_t lbn, struct buf **bpp, struct ucred *cred)
1998-03-01 05:20:01 +03:00
{
struct inode *ip;
struct lfs *fs;
long bb;
int error;
extern long locked_queue_bytes;
size_t obufsize;
1998-03-01 05:20:01 +03:00
ip = VTOI(vp);
fs = ip->i_lfs;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
bb = (long)fragstofsb(fs, numfrags(fs, nsize - osize));
error = 0;
/*
* Get the seglock so we don't enlarge blocks or change the segment
* accounting information while a segment is being written.
*/
top:
#ifdef LFS_MALLOC_SEGLOCK
lfs_seglock(fs, SEGM_PROT);
#else
lockmgr(&fs->lfs_fraglock, LK_SHARED, 0);
#endif
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (!ISSPACE(fs, bb, cred)) {
error = ENOSPC;
goto out;
1998-03-01 05:20:01 +03:00
}
if ((error = bread(vp, lbn, osize, NOCRED, bpp))) {
brelse(*bpp);
goto out;
1998-03-01 05:20:01 +03:00
}
#ifdef QUOTA
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if ((error = chkdq(ip, bb, cred, 0))) {
brelse(*bpp);
goto out;
}
#endif
/*
* Adjust accounting for lfs_avail. If there's not enough room,
* we will have to wait for the cleaner, which we can't do while
* holding a block busy or while holding the seglock. In that case,
* release both and start over after waiting.
*/
if ((*bpp)->b_flags & B_DELWRI) {
if (!lfs_fits(fs, bb)) {
brelse(*bpp);
#ifdef QUOTA
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
chkdq(ip, -bb, cred, 0);
#endif
#ifdef LFS_FRAGSIZE_SEGLOCK
lfs_segunlock(fs);
#else
lockmgr(&fs->lfs_fraglock, LK_RELEASE, 0);
#endif
lfs_availwait(fs, bb);
goto top;
}
fs->lfs_avail -= bb;
}
fs->lfs_bfree -= bb;
ip->i_lfs_effnblks += bb;
1998-03-01 05:20:01 +03:00
ip->i_flag |= IN_CHANGE | IN_UPDATE;
LFS_DEBUG_COUNTLOCKED("frag1");
obufsize = (*bpp)->b_bufsize;
1998-03-01 05:20:01 +03:00
allocbuf(*bpp, nsize);
/* Adjust locked-list accounting */
if (((*bpp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED)
locked_queue_bytes += (*bpp)->b_bufsize - obufsize;
LFS_DEBUG_COUNTLOCKED("frag2");
1998-03-01 05:20:01 +03:00
bzero((char *)((*bpp)->b_data) + osize, (u_int)(nsize - osize));
out:
#ifdef LFS_FRAGSIZE_SEGLOCK
lfs_segunlock(fs);
#else
lockmgr(&fs->lfs_fraglock, LK_RELEASE, 0);
#endif
return (error);
}