freebsd-skq/sys/ufs/ffs/ffs_inode.c

547 lines
16 KiB
C
Raw Normal View History

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.
*
* @(#)ffs_inode.c 8.13 (Berkeley) 4/21/95
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
1996-01-05 18:31:58 +00:00
#include "opt_quota.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mount.h>
#include <sys/proc.h>
#include <sys/bio.h>
1994-05-24 10:09:53 +00:00
#include <sys/buf.h>
#include <sys/vnode.h>
#include <sys/malloc.h>
#include <sys/resourcevar.h>
Implement a low-memory deadlock solution. Removed most of the hacks that were trying to deal with low-memory situations prior to now. The new code is based on the concept that I/O must be able to function in a low memory situation. All major modules related to I/O (except networking) have been adjusted to allow allocation out of the system reserve memory pool. These modules now detect a low memory situation but rather then block they instead continue to operate, then return resources to the memory pool instead of cache them or leave them wired. Code has been added to stall in a low-memory situation prior to a vnode being locked. Thus situations where a process blocks in a low-memory condition while holding a locked vnode have been reduced to near nothing. Not only will I/O continue to operate, but many prior deadlock conditions simply no longer exist. Implement a number of VFS/BIO fixes (found by Ian): in biodone(), bogus-page replacement code, the loop was not properly incrementing loop variables prior to a continue statement. We do not believe this code can be hit anyway but we aren't taking any chances. We'll turn the whole section into a panic (as it already is in brelse()) after the release is rolled. In biodone(), the foff calculation was incorrectly clamped to the iosize, causing the wrong foff to be calculated for pages in the case of an I/O error or biodone() called without initiating I/O. The problem always caused a panic before. Now it doesn't. The problem is mainly an issue with NFS. Fixed casts for ~PAGE_MASK. This code worked properly before only because the calculations use signed arithmatic. Better to properly extend PAGE_MASK first before inverting it for the 64 bit masking op. In brelse(), the bogus_page fixup code was improperly throwing away the original contents of 'm' when it did the j-loop to fix the bogus pages. The result was that it would potentially invalidate parts of the *WRONG* page(!), leading to corruption. There may still be cases where a background bitmap write is being duplicated, causing potential corruption. We have identified a potentially serious bug related to this but the fix is still TBD. So instead this patch contains a KASSERT to detect the problem and panic the machine rather then continue to corrupt the filesystem. The problem does not occur very often.. it is very hard to reproduce, and it may or may not be the cause of the corruption people have reported. Review by: (VFS/BIO: mckusick, Ian Dowse <iedowse@maths.tcd.ie>) Testing by: (VM/Deadlock) Paul Saab <ps@yahoo-inc.com>
2000-11-18 23:06:26 +00:00
#include <sys/vmmeter.h>
#include <sys/stat.h>
1994-05-24 10:09:53 +00:00
#include <vm/vm.h>
#include <vm/vm_extern.h>
1994-05-24 10:09:53 +00:00
#include <ufs/ufs/extattr.h>
1994-05-24 10:09:53 +00:00
#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
1994-05-24 10:09:53 +00:00
#include <ufs/ufs/inode.h>
#include <ufs/ufs/ufs_extern.h>
1994-05-24 10:09:53 +00:00
#include <ufs/ffs/fs.h>
#include <ufs/ffs/ffs_extern.h>
2002-03-19 22:40:48 +00:00
static int ffs_indirtrunc(struct inode *, ufs_daddr_t, ufs_daddr_t,
ufs_daddr_t, int, long *);
1994-05-24 10:09:53 +00:00
/*
* Update the access, modified, and inode change times as specified by the
* IN_ACCESS, IN_UPDATE, and IN_CHANGE flags respectively. Write the inode
* to disk if the IN_MODIFIED flag is set (it may be set initially, or by
* the timestamp update). The IN_LAZYMOD flag is set to force a write
* later if not now. If we write now, then clear both IN_MODIFIED and
* IN_LAZYMOD to reflect the presumably successful write, and if waitfor is
* set, then wait for the write to complete.
1994-05-24 10:09:53 +00:00
*/
int
ffs_update(vp, waitfor)
struct vnode *vp;
int waitfor;
1994-05-24 10:09:53 +00:00
{
struct fs *fs;
1994-05-24 10:09:53 +00:00
struct buf *bp;
struct inode *ip;
int error;
ufs_itimes(vp);
ip = VTOI(vp);
if ((ip->i_flag & IN_MODIFIED) == 0 && waitfor == 0)
1994-05-24 10:09:53 +00:00
return (0);
ip->i_flag &= ~(IN_LAZYMOD | IN_MODIFIED);
1994-05-24 10:09:53 +00:00
fs = ip->i_fs;
if (fs->fs_ronly)
return (0);
1994-05-24 10:09:53 +00:00
/*
* Ensure that uid and gid are correct. This is a temporary
* fix until fsck has been changed to do the update.
*/
if (fs->fs_inodefmt < FS_44INODEFMT) { /* XXX */
ip->i_din.di_ouid = ip->i_uid; /* XXX */
ip->i_din.di_ogid = ip->i_gid; /* XXX */
} /* XXX */
error = bread(ip->i_devvp, fsbtodb(fs, ino_to_fsba(fs, ip->i_number)),
(int)fs->fs_bsize, NOCRED, &bp);
if (error) {
1994-05-24 10:09:53 +00:00
brelse(bp);
return (error);
}
if (DOINGSOFTDEP(vp))
softdep_update_inodeblock(ip, bp, waitfor);
else if (ip->i_effnlink != ip->i_nlink)
panic("ffs_update: bad link cnt");
1994-05-24 10:09:53 +00:00
*((struct dinode *)bp->b_data +
ino_to_fsbo(fs, ip->i_number)) = ip->i_din;
if (waitfor && !DOINGASYNC(vp)) {
1994-05-24 10:09:53 +00:00
return (bwrite(bp));
Implement a low-memory deadlock solution. Removed most of the hacks that were trying to deal with low-memory situations prior to now. The new code is based on the concept that I/O must be able to function in a low memory situation. All major modules related to I/O (except networking) have been adjusted to allow allocation out of the system reserve memory pool. These modules now detect a low memory situation but rather then block they instead continue to operate, then return resources to the memory pool instead of cache them or leave them wired. Code has been added to stall in a low-memory situation prior to a vnode being locked. Thus situations where a process blocks in a low-memory condition while holding a locked vnode have been reduced to near nothing. Not only will I/O continue to operate, but many prior deadlock conditions simply no longer exist. Implement a number of VFS/BIO fixes (found by Ian): in biodone(), bogus-page replacement code, the loop was not properly incrementing loop variables prior to a continue statement. We do not believe this code can be hit anyway but we aren't taking any chances. We'll turn the whole section into a panic (as it already is in brelse()) after the release is rolled. In biodone(), the foff calculation was incorrectly clamped to the iosize, causing the wrong foff to be calculated for pages in the case of an I/O error or biodone() called without initiating I/O. The problem always caused a panic before. Now it doesn't. The problem is mainly an issue with NFS. Fixed casts for ~PAGE_MASK. This code worked properly before only because the calculations use signed arithmatic. Better to properly extend PAGE_MASK first before inverting it for the 64 bit masking op. In brelse(), the bogus_page fixup code was improperly throwing away the original contents of 'm' when it did the j-loop to fix the bogus pages. The result was that it would potentially invalidate parts of the *WRONG* page(!), leading to corruption. There may still be cases where a background bitmap write is being duplicated, causing potential corruption. We have identified a potentially serious bug related to this but the fix is still TBD. So instead this patch contains a KASSERT to detect the problem and panic the machine rather then continue to corrupt the filesystem. The problem does not occur very often.. it is very hard to reproduce, and it may or may not be the cause of the corruption people have reported. Review by: (VFS/BIO: mckusick, Ian Dowse <iedowse@maths.tcd.ie>) Testing by: (VM/Deadlock) Paul Saab <ps@yahoo-inc.com>
2000-11-18 23:06:26 +00:00
} else if (vm_page_count_severe() || buf_dirty_count_severe()) {
return (bwrite(bp));
} else {
if (bp->b_bufsize == fs->fs_bsize)
bp->b_flags |= B_CLUSTEROK;
1994-05-24 10:09:53 +00:00
bdwrite(bp);
return (0);
}
}
#define SINGLE 0 /* index of single indirect block */
#define DOUBLE 1 /* index of double indirect block */
#define TRIPLE 2 /* index of triple indirect block */
/*
* Truncate the inode oip to at most length size, freeing the
* disk blocks.
*/
int
ffs_truncate(vp, length, flags, cred, td)
struct vnode *vp;
off_t length;
int flags;
struct ucred *cred;
struct thread *td;
1994-05-24 10:09:53 +00:00
{
struct vnode *ovp = vp;
ufs_daddr_t lastblock;
struct inode *oip;
ufs_daddr_t bn, lbn, lastiblock[NIADDR], indir_lbn[NIADDR];
ufs_daddr_t oldblks[NDADDR + NIADDR], newblks[NDADDR + NIADDR];
struct fs *fs;
1994-05-24 10:09:53 +00:00
struct buf *bp;
int offset, size, level;
long count, nblocks, blocksreleased = 0;
int i;
1994-05-24 10:09:53 +00:00
int aflags, error, allerror;
off_t osize;
oip = VTOI(ovp);
fs = oip->i_fs;
if (length < 0)
return (EINVAL);
if (length > fs->fs_maxfilesize)
return (EFBIG);
1994-05-24 10:09:53 +00:00
if (ovp->v_type == VLNK &&
(oip->i_size < ovp->v_mount->mnt_maxsymlinklen || oip->i_din.di_blocks == 0)) {
1994-05-24 10:09:53 +00:00
#ifdef DIAGNOSTIC
if (length != 0)
panic("ffs_truncate: partial truncate of symlink");
#endif
bzero((char *)&oip->i_shortlink, (u_int)oip->i_size);
oip->i_size = 0;
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (UFS_UPDATE(ovp, 1));
1994-05-24 10:09:53 +00:00
}
if (oip->i_size == length) {
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (UFS_UPDATE(ovp, 0));
1994-05-24 10:09:53 +00:00
}
if (fs->fs_ronly)
panic("ffs_truncate: read-only filesystem");
1994-05-24 10:09:53 +00:00
#ifdef QUOTA
error = getinoquota(oip);
if (error)
1994-05-24 10:09:53 +00:00
return (error);
#endif
if ((oip->i_flags & SF_SNAPSHOT) != 0)
ffs_snapremove(ovp);
ovp->v_lasta = ovp->v_clen = ovp->v_cstart = ovp->v_lastw = 0;
if (DOINGSOFTDEP(ovp)) {
if (length > 0 || softdep_slowdown(ovp)) {
/*
* If a file is only partially truncated, then
* we have to clean up the data structures
* describing the allocation past the truncation
* point. Finding and deallocating those structures
* is a lot of work. Since partial truncation occurs
* rarely, we solve the problem by syncing the file
* so that it will have no data structures left.
*/
if ((error = VOP_FSYNC(ovp, cred, MNT_WAIT,
td)) != 0)
return (error);
if (oip->i_flag & IN_SPACECOUNTED)
fs->fs_pendingblocks -= oip->i_blocks;
} else {
#ifdef QUOTA
(void) chkdq(oip, -oip->i_blocks, NOCRED, 0);
#endif
softdep_setup_freeblocks(oip, length);
vinvalbuf(ovp, 0, cred, td, 0, 0);
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (ffs_update(ovp, 0));
}
}
1994-05-24 10:09:53 +00:00
osize = oip->i_size;
/*
* Lengthen the size of the file. We must ensure that the
* last byte of the file is allocated. Since the smallest
* value of osize is 0, length will be at least 1.
1994-05-24 10:09:53 +00:00
*/
if (osize < length) {
This mega-commit is meant to fix numerous interrelated problems. There has been some bitrot and incorrect assumptions in the vfs_bio code. These problems have manifest themselves worse on NFS type filesystems, but can still affect local filesystems under certain circumstances. Most of the problems have involved mmap consistancy, and as a side-effect broke the vfs.ioopt code. This code might have been committed seperately, but almost everything is interrelated. 1) Allow (pmap_object_init_pt) prefaulting of buffer-busy pages that are fully valid. 2) Rather than deactivating erroneously read initial (header) pages in kern_exec, we now free them. 3) Fix the rundown of non-VMIO buffers that are in an inconsistent (missing vp) state. 4) Fix the disassociation of pages from buffers in brelse. The previous code had rotted and was faulty in a couple of important circumstances. 5) Remove a gratuitious buffer wakeup in vfs_vmio_release. 6) Remove a crufty and currently unused cluster mechanism for VBLK files in vfs_bio_awrite. When the code is functional, I'll add back a cleaner version. 7) The page busy count wakeups assocated with the buffer cache usage were incorrectly cleaned up in a previous commit by me. Revert to the original, correct version, but with a cleaner implementation. 8) The cluster read code now tries to keep data associated with buffers more aggressively (without breaking the heuristics) when it is presumed that the read data (buffers) will be soon needed. 9) Change to filesystem lockmgr locks so that they use LK_NOPAUSE. The delay loop waiting is not useful for filesystem locks, due to the length of the time intervals. 10) Correct and clean-up spec_getpages. 11) Implement a fully functional nfs_getpages, nfs_putpages. 12) Fix nfs_write so that modifications are coherent with the NFS data on the server disk (at least as well as NFS seems to allow.) 13) Properly support MS_INVALIDATE on NFS. 14) Properly pass down MS_INVALIDATE to lower levels of the VM code from vm_map_clean. 15) Better support the notion of pages being busy but valid, so that fewer in-transit waits occur. (use p->busy more for pageouts instead of PG_BUSY.) Since the page is fully valid, it is still usable for reads. 16) It is possible (in error) for cached pages to be busy. Make the page allocation code handle that case correctly. (It should probably be a printf or panic, but I want the system to handle coding errors robustly. I'll probably add a printf.) 17) Correct the design and usage of vm_page_sleep. It didn't handle consistancy problems very well, so make the design a little less lofty. After vm_page_sleep, if it ever blocked, it is still important to relookup the page (if the object generation count changed), and verify it's status (always.) 18) In vm_pageout.c, vm_pageout_clean had rotted, so clean that up. 19) Push the page busy for writes and VM_PROT_READ into vm_pageout_flush. 20) Fix vm_pager_put_pages and it's descendents to support an int flag instead of a boolean, so that we can pass down the invalidate bit.
1998-03-07 21:37:31 +00:00
vnode_pager_setsize(ovp, length);
1994-05-24 10:09:53 +00:00
aflags = B_CLRBUF;
if (flags & IO_SYNC)
1994-05-24 10:09:53 +00:00
aflags |= B_SYNC;
error = UFS_BALLOC(ovp, length - 1, 1,
cred, aflags, &bp);
if (error)
1994-05-24 10:09:53 +00:00
return (error);
oip->i_size = length;
if (bp->b_bufsize == fs->fs_bsize)
bp->b_flags |= B_CLUSTEROK;
if (aflags & B_SYNC)
1994-05-24 10:09:53 +00:00
bwrite(bp);
else
bawrite(bp);
oip->i_flag |= IN_CHANGE | IN_UPDATE;
return (UFS_UPDATE(ovp, 1));
1994-05-24 10:09:53 +00:00
}
/*
* Shorten the size of the file. If the file is not being
* truncated to a block boundary, the contents of the
1994-05-24 10:09:53 +00:00
* partial block following the end of the file must be
* zero'ed in case it ever becomes accessible again because
* of subsequent file growth. Directories however are not
* zero'ed as they should grow back initialized to empty.
1994-05-24 10:09:53 +00:00
*/
offset = blkoff(fs, length);
if (offset == 0) {
oip->i_size = length;
} else {
lbn = lblkno(fs, length);
aflags = B_CLRBUF;
if (flags & IO_SYNC)
1994-05-24 10:09:53 +00:00
aflags |= B_SYNC;
error = UFS_BALLOC(ovp, length - 1, 1, cred, aflags, &bp);
if (error) {
1994-05-24 10:09:53 +00:00
return (error);
}
/*
* When we are doing soft updates and the UFS_BALLOC
* above fills in a direct block hole with a full sized
* block that will be truncated down to a fragment below,
* we must flush out the block dependency with an FSYNC
* so that we do not get a soft updates inconsistency
* when we create the fragment below.
*/
if (DOINGSOFTDEP(ovp) && lbn < NDADDR &&
fragroundup(fs, blkoff(fs, length)) < fs->fs_bsize &&
(error = VOP_FSYNC(ovp, cred, MNT_WAIT, td)) != 0)
return (error);
1994-05-24 10:09:53 +00:00
oip->i_size = length;
size = blksize(fs, oip, lbn);
if (ovp->v_type != VDIR)
bzero((char *)bp->b_data + offset,
(u_int)(size - offset));
/* Kirk's code has reallocbuf(bp, size, 1) here */
allocbuf(bp, size);
if (bp->b_bufsize == fs->fs_bsize)
bp->b_flags |= B_CLUSTEROK;
if (aflags & B_SYNC)
1994-05-24 10:09:53 +00:00
bwrite(bp);
else
bawrite(bp);
}
/*
* Calculate index into inode's block list of
* last direct and indirect blocks (if any)
* which we want to keep. Lastblock is -1 when
* the file is truncated to 0.
*/
lastblock = lblkno(fs, length + fs->fs_bsize - 1) - 1;
lastiblock[SINGLE] = lastblock - NDADDR;
lastiblock[DOUBLE] = lastiblock[SINGLE] - NINDIR(fs);
lastiblock[TRIPLE] = lastiblock[DOUBLE] - NINDIR(fs) * NINDIR(fs);
nblocks = btodb(fs->fs_bsize);
/*
* Update file and block pointers on disk before we start freeing
* blocks. If we crash before free'ing blocks below, the blocks
* will be returned to the free list. lastiblock values are also
* normalized to -1 for calls to ffs_indirtrunc below.
*/
bcopy((caddr_t)&oip->i_db[0], (caddr_t)oldblks, sizeof oldblks);
for (level = TRIPLE; level >= SINGLE; level--)
if (lastiblock[level] < 0) {
oip->i_ib[level] = 0;
lastiblock[level] = -1;
}
for (i = NDADDR - 1; i > lastblock; i--)
oip->i_db[i] = 0;
oip->i_flag |= IN_CHANGE | IN_UPDATE;
allerror = UFS_UPDATE(ovp, 1);
1994-05-24 10:09:53 +00:00
/*
* Having written the new inode to disk, save its new configuration
* and put back the old block pointers long enough to process them.
* Note that we save the new block configuration so we can check it
* when we are done.
*/
bcopy((caddr_t)&oip->i_db[0], (caddr_t)newblks, sizeof newblks);
bcopy((caddr_t)oldblks, (caddr_t)&oip->i_db[0], sizeof oldblks);
oip->i_size = osize;
error = vtruncbuf(ovp, cred, td, length, fs->fs_bsize);
if (error && (allerror == 0))
allerror = error;
1994-05-24 10:09:53 +00:00
/*
* Indirect blocks first.
*/
indir_lbn[SINGLE] = -NDADDR;
indir_lbn[DOUBLE] = indir_lbn[SINGLE] - NINDIR(fs) - 1;
indir_lbn[TRIPLE] = indir_lbn[DOUBLE] - NINDIR(fs) * NINDIR(fs) - 1;
for (level = TRIPLE; level >= SINGLE; level--) {
bn = oip->i_ib[level];
if (bn != 0) {
error = ffs_indirtrunc(oip, indir_lbn[level],
fsbtodb(fs, bn), lastiblock[level], level, &count);
if (error)
allerror = error;
blocksreleased += count;
if (lastiblock[level] < 0) {
oip->i_ib[level] = 0;
ffs_blkfree(fs, oip->i_devvp, bn, fs->fs_bsize,
oip->i_number);
1994-05-24 10:09:53 +00:00
blocksreleased += nblocks;
}
}
if (lastiblock[level] >= 0)
goto done;
}
/*
* All whole direct blocks or frags.
*/
for (i = NDADDR - 1; i > lastblock; i--) {
long bsize;
1994-05-24 10:09:53 +00:00
bn = oip->i_db[i];
if (bn == 0)
continue;
oip->i_db[i] = 0;
bsize = blksize(fs, oip, i);
ffs_blkfree(fs, oip->i_devvp, bn, bsize, oip->i_number);
1994-05-24 10:09:53 +00:00
blocksreleased += btodb(bsize);
}
if (lastblock < 0)
goto done;
/*
* Finally, look for a change in size of the
* last direct block; release any frags.
*/
bn = oip->i_db[lastblock];
if (bn != 0) {
long oldspace, newspace;
/*
* Calculate amount of space we're giving
* back as old block size minus new block size.
*/
oldspace = blksize(fs, oip, lastblock);
oip->i_size = length;
newspace = blksize(fs, oip, lastblock);
if (newspace == 0)
panic("ffs_truncate: newspace");
1994-05-24 10:09:53 +00:00
if (oldspace - newspace > 0) {
/*
* Block number of space to be free'd is
* the old block # plus the number of frags
* required for the storage we're keeping.
*/
bn += numfrags(fs, newspace);
ffs_blkfree(fs, oip->i_devvp, bn, oldspace - newspace,
oip->i_number);
1994-05-24 10:09:53 +00:00
blocksreleased += btodb(oldspace - newspace);
}
}
done:
#ifdef DIAGNOSTIC
for (level = SINGLE; level <= TRIPLE; level++)
if (newblks[NDADDR + level] != oip->i_ib[level])
panic("ffs_truncate1");
1994-05-24 10:09:53 +00:00
for (i = 0; i < NDADDR; i++)
if (newblks[i] != oip->i_db[i])
panic("ffs_truncate2");
1994-05-24 10:09:53 +00:00
if (length == 0 &&
(!TAILQ_EMPTY(&ovp->v_dirtyblkhd) ||
!TAILQ_EMPTY(&ovp->v_cleanblkhd)))
panic("ffs_truncate3");
1994-05-24 10:09:53 +00:00
#endif /* DIAGNOSTIC */
/*
* Put back the real size.
*/
oip->i_size = length;
oip->i_blocks -= blocksreleased;
1994-05-24 10:09:53 +00:00
if (oip->i_blocks < 0) /* sanity */
oip->i_blocks = 0;
oip->i_flag |= IN_CHANGE;
#ifdef QUOTA
(void) chkdq(oip, -blocksreleased, NOCRED, 0);
#endif
return (allerror);
}
/*
* Release blocks associated with the inode ip and stored in the indirect
* block bn. Blocks are free'd in LIFO order up to (but not including)
* lastbn. If level is greater than SINGLE, the block is an indirect block
* and recursive calls to indirtrunc must be used to cleanse other indirect
* blocks.
*
* NB: triple indirect blocks are untested.
*/
static int
ffs_indirtrunc(ip, lbn, dbn, lastbn, level, countp)
struct inode *ip;
ufs_daddr_t lbn, lastbn;
ufs_daddr_t dbn;
1994-05-24 10:09:53 +00:00
int level;
long *countp;
{
int i;
1994-05-24 10:09:53 +00:00
struct buf *bp;
struct fs *fs = ip->i_fs;
ufs_daddr_t *bap;
1994-05-24 10:09:53 +00:00
struct vnode *vp;
ufs_daddr_t *copy = NULL, nb, nlbn, last;
1994-05-24 10:09:53 +00:00
long blkcount, factor;
int nblocks, blocksreleased = 0;
int error = 0, allerror = 0;
/*
* Calculate index in current block of last
* block to be kept. -1 indicates the entire
* block so we need not calculate the index.
*/
factor = 1;
for (i = SINGLE; i < level; i++)
factor *= NINDIR(fs);
last = lastbn;
if (lastbn > 0)
last /= factor;
nblocks = btodb(fs->fs_bsize);
/*
* Get buffer of block pointers, zero those entries corresponding
* to blocks to be free'd, and update on disk copy first. Since
* double(triple) indirect before single(double) indirect, calls
* to bmap on these blocks will fail. However, we already have
* the on disk address, so we have to set the b_blkno field
* explicitly instead of letting bread do everything for us.
*/
vp = ITOV(ip);
bp = getblk(vp, lbn, (int)fs->fs_bsize, 0, 0);
1995-03-04 03:24:45 +00:00
if ((bp->b_flags & B_CACHE) == 0) {
1994-05-24 10:09:53 +00:00
curproc->p_stats->p_ru.ru_inblock++; /* pay for read */
bp->b_iocmd = BIO_READ;
bp->b_flags &= ~B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
1994-05-24 10:09:53 +00:00
if (bp->b_bcount > bp->b_bufsize)
panic("ffs_indirtrunc: bad buffer size");
bp->b_blkno = dbn;
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
vfs_busy_pages(bp, 0);
BUF_STRATEGY(bp);
error = bufwait(bp);
1994-05-24 10:09:53 +00:00
}
if (error) {
brelse(bp);
*countp = 0;
return (error);
}
bap = (ufs_daddr_t *)bp->b_data;
if (lastbn != -1) {
MALLOC(copy, ufs_daddr_t *, fs->fs_bsize, M_TEMP, M_WAITOK);
bcopy((caddr_t)bap, (caddr_t)copy, (u_int)fs->fs_bsize);
bzero((caddr_t)&bap[last + 1],
(u_int)(NINDIR(fs) - (last + 1)) * sizeof (ufs_daddr_t));
if (DOINGASYNC(vp)) {
bawrite(bp);
} else {
error = bwrite(bp);
if (error)
allerror = error;
}
bap = copy;
}
1994-05-24 10:09:53 +00:00
/*
* Recursively free totally unused blocks.
*/
for (i = NINDIR(fs) - 1, nlbn = lbn + 1 - i * factor; i > last;
i--, nlbn += factor) {
nb = bap[i];
if (nb == 0)
continue;
if (level > SINGLE) {
if ((error = ffs_indirtrunc(ip, nlbn, fsbtodb(fs, nb),
(ufs_daddr_t)-1, level - 1, &blkcount)) != 0)
1994-05-24 10:09:53 +00:00
allerror = error;
blocksreleased += blkcount;
}
ffs_blkfree(fs, ip->i_devvp, nb, fs->fs_bsize, ip->i_number);
1994-05-24 10:09:53 +00:00
blocksreleased += nblocks;
}
/*
* Recursively free last partial block.
*/
if (level > SINGLE && lastbn >= 0) {
last = lastbn % factor;
nb = bap[i];
if (nb != 0) {
error = ffs_indirtrunc(ip, nlbn, fsbtodb(fs, nb),
last, level - 1, &blkcount);
if (error)
1994-05-24 10:09:53 +00:00
allerror = error;
blocksreleased += blkcount;
}
}
if (copy != NULL) {
FREE(copy, M_TEMP);
} else {
bp->b_flags |= B_INVAL | B_NOCACHE;
brelse(bp);
}
1994-05-24 10:09:53 +00:00
*countp = blocksreleased;
return (allerror);
}