Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-11-14 16:12:02 +00:00
Code Issues Packages Projects Releases Wiki Activity
3655d54af8
linux/fs/xfs/xfs_rw.c
Christoph Hellwig 13e6d5cdde xfs: merge fsync and O_SYNC handling 
The guarantees for O_SYNC are exactly the same as the ones we need to
make for an fsync call (and given that Linux O_SYNC is O_DSYNC the
equivalent is fdadatasync, but we treat both the same in XFS), except
with a range data writeout.  Jan Kara has started unifying these two
path for filesystems using the generic helpers, and I've started to
look at XFS.

The actual transaction commited by xfs_fsync and xfs_write_sync_logforce
has a different transaction number, but actually is exactly the same.
We'll only use the fsync transaction going forward.  One major difference
is that xfs_write_sync_logforce never issues a cache flush unless we
commit a transaction causing that as a side-effect, which is an obvious
bug in the O_SYNC handling.  Second all the locking and i_update_size
vs i_update_core changes from 978b723712
never made it to xfs_write_sync_logforce, so we add them back.

To make xfs_fsync easily usable from the O_SYNC path, the filemap_fdatawait
call is moved up to xfs_file_fsync, so that we don't wait on the whole
file after we already waited for our portion in xfs_write.

We'll also use a plain call to filemap_write_and_wait_range instead
of the previous sync_page_rang which did it in two steps including
an half-hearted inode write out that doesn't help us.

Once we're done with this also remove the now useless i_update_size
tracking.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Felix Blyakher <felixb@sgi.com>
Signed-off-by: Felix Blyakher <felixb@sgi.com>
2009-09-01 12:45:57 -05:00

339 lines
8.4 KiB
C
Raw Blame History

/*
* Copyright (c) 2000-2006 Silicon Graphics, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it would be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_types.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_dir2.h"
#include "xfs_dmapi.h"
#include "xfs_mount.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_inode_item.h"
#include "xfs_itable.h"
#include "xfs_btree.h"
#include "xfs_alloc.h"
#include "xfs_ialloc.h"
#include "xfs_attr.h"
#include "xfs_bmap.h"
#include "xfs_error.h"
#include "xfs_buf_item.h"
#include "xfs_rw.h"
/*
* This is a subroutine for xfs_write() and other writers (xfs_ioctl)
* which clears the setuid and setgid bits when a file is written.
*/
int
xfs_write_clear_setuid(
xfs_inode_t *ip)
{
xfs_mount_t *mp;
xfs_trans_t *tp;
int error;
mp = ip->i_mount;
tp = xfs_trans_alloc(mp, XFS_TRANS_WRITEID);
if ((error = xfs_trans_reserve(tp, 0,
XFS_WRITEID_LOG_RES(mp),
0, 0, 0))) {
xfs_trans_cancel(tp, 0);
return error;
}
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
xfs_trans_ihold(tp, ip);
ip->i_d.di_mode &= ~S_ISUID;
/*
* Note that we don't have to worry about mandatory
* file locking being disabled here because we only
* clear the S_ISGID bit if the Group execute bit is
* on, but if it was on then mandatory locking wouldn't
* have been enabled.
*/
if (ip->i_d.di_mode & S_IXGRP) {
ip->i_d.di_mode &= ~S_ISGID;
}
xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
xfs_trans_set_sync(tp);
error = xfs_trans_commit(tp, 0);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return 0;
}
/*
* Force a shutdown of the filesystem instantly while keeping
* the filesystem consistent. We don't do an unmount here; just shutdown
* the shop, make sure that absolutely nothing persistent happens to
* this filesystem after this point.
*/
void
xfs_do_force_shutdown(
xfs_mount_t *mp,
int flags,
char *fname,
int lnnum)
{
int logerror;
logerror = flags & SHUTDOWN_LOG_IO_ERROR;
if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
cmn_err(CE_NOTE, "xfs_force_shutdown(%s,0x%x) called from "
"line %d of file %s. Return address = 0x%p",
mp->m_fsname, flags, lnnum, fname, __return_address);
}
/*
* No need to duplicate efforts.
*/
if (XFS_FORCED_SHUTDOWN(mp) && !logerror)
return;
/*
* This flags XFS_MOUNT_FS_SHUTDOWN, makes sure that we don't
* queue up anybody new on the log reservations, and wakes up
* everybody who's sleeping on log reservations to tell them
* the bad news.
*/
if (xfs_log_force_umount(mp, logerror))
return;
if (flags & SHUTDOWN_CORRUPT_INCORE) {
xfs_cmn_err(XFS_PTAG_SHUTDOWN_CORRUPT, CE_ALERT, mp,
"Corruption of in-memory data detected. Shutting down filesystem: %s",
mp->m_fsname);
if (XFS_ERRLEVEL_HIGH <= xfs_error_level) {
xfs_stack_trace();
}
} else if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
if (logerror) {
xfs_cmn_err(XFS_PTAG_SHUTDOWN_LOGERROR, CE_ALERT, mp,
"Log I/O Error Detected. Shutting down filesystem: %s",
mp->m_fsname);
} else if (flags & SHUTDOWN_DEVICE_REQ) {
xfs_cmn_err(XFS_PTAG_SHUTDOWN_IOERROR, CE_ALERT, mp,
"All device paths lost. Shutting down filesystem: %s",
mp->m_fsname);
} else if (!(flags & SHUTDOWN_REMOTE_REQ)) {
xfs_cmn_err(XFS_PTAG_SHUTDOWN_IOERROR, CE_ALERT, mp,
"I/O Error Detected. Shutting down filesystem: %s",
mp->m_fsname);
}
}
if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
cmn_err(CE_ALERT, "Please umount the filesystem, "
"and rectify the problem(s)");
}
}
/*
* Called when we want to stop a buffer from getting written or read.
* We attach the EIO error, muck with its flags, and call biodone
* so that the proper iodone callbacks get called.
*/
int
xfs_bioerror(
xfs_buf_t *bp)
{
#ifdef XFSERRORDEBUG
ASSERT(XFS_BUF_ISREAD(bp) || bp->b_iodone);
#endif
/*
* No need to wait until the buffer is unpinned.
* We aren't flushing it.
*/
xfs_buftrace("XFS IOERROR", bp);
XFS_BUF_ERROR(bp, EIO);
/*
* We're calling biodone, so delete B_DONE flag. Either way
* we have to call the iodone callback, and calling biodone
* probably is the best way since it takes care of
* GRIO as well.
*/
XFS_BUF_UNREAD(bp);
XFS_BUF_UNDELAYWRITE(bp);
XFS_BUF_UNDONE(bp);
XFS_BUF_STALE(bp);
XFS_BUF_CLR_BDSTRAT_FUNC(bp);
xfs_biodone(bp);
return (EIO);
}
/*
* Same as xfs_bioerror, except that we are releasing the buffer
* here ourselves, and avoiding the biodone call.
* This is meant for userdata errors; metadata bufs come with
* iodone functions attached, so that we can track down errors.
*/
int
xfs_bioerror_relse(
xfs_buf_t *bp)
{
int64_t fl;
ASSERT(XFS_BUF_IODONE_FUNC(bp) != xfs_buf_iodone_callbacks);
ASSERT(XFS_BUF_IODONE_FUNC(bp) != xlog_iodone);
xfs_buftrace("XFS IOERRELSE", bp);
fl = XFS_BUF_BFLAGS(bp);
/*
* No need to wait until the buffer is unpinned.
* We aren't flushing it.
*
* chunkhold expects B_DONE to be set, whether
* we actually finish the I/O or not. We don't want to
* change that interface.
*/
XFS_BUF_UNREAD(bp);
XFS_BUF_UNDELAYWRITE(bp);
XFS_BUF_DONE(bp);
XFS_BUF_STALE(bp);
XFS_BUF_CLR_IODONE_FUNC(bp);
XFS_BUF_CLR_BDSTRAT_FUNC(bp);
if (!(fl & XFS_B_ASYNC)) {
/*
* Mark b_error and B_ERROR _both_.
* Lot's of chunkcache code assumes that.
* There's no reason to mark error for
* ASYNC buffers.
*/
XFS_BUF_ERROR(bp, EIO);
XFS_BUF_FINISH_IOWAIT(bp);
} else {
xfs_buf_relse(bp);
}
return (EIO);
}
/*
* Prints out an ALERT message about I/O error.
*/
void
xfs_ioerror_alert(
char *func,
struct xfs_mount *mp,
xfs_buf_t *bp,
xfs_daddr_t blkno)
{
cmn_err(CE_ALERT,
"I/O error in filesystem (\"%s\") meta-data dev %s block 0x%llx"
" (\"%s\") error %d buf count %zd",
(!mp || !mp->m_fsname) ? "(fs name not set)" : mp->m_fsname,
XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
(__uint64_t)blkno, func,
XFS_BUF_GETERROR(bp), XFS_BUF_COUNT(bp));
}
/*
* This isn't an absolute requirement, but it is
* just a good idea to call xfs_read_buf instead of
* directly doing a read_buf call. For one, we shouldn't
* be doing this disk read if we are in SHUTDOWN state anyway,
* so this stops that from happening. Secondly, this does all
* the error checking stuff and the brelse if appropriate for
* the caller, so the code can be a little leaner.
*/
int
xfs_read_buf(
struct xfs_mount *mp,
xfs_buftarg_t *target,
xfs_daddr_t blkno,
int len,
uint flags,
xfs_buf_t **bpp)
{
xfs_buf_t *bp;
int error;
if (flags)
bp = xfs_buf_read_flags(target, blkno, len, flags);
else
bp = xfs_buf_read(target, blkno, len, flags);
if (!bp)
return XFS_ERROR(EIO);
error = XFS_BUF_GETERROR(bp);
if (bp && !error && !XFS_FORCED_SHUTDOWN(mp)) {
*bpp = bp;
} else {
*bpp = NULL;
if (error) {
xfs_ioerror_alert("xfs_read_buf", mp, bp, XFS_BUF_ADDR(bp));
} else {
error = XFS_ERROR(EIO);
}
if (bp) {
XFS_BUF_UNDONE(bp);
XFS_BUF_UNDELAYWRITE(bp);
XFS_BUF_STALE(bp);
/*
* brelse clears B_ERROR and b_error
*/
xfs_buf_relse(bp);
}
}
return (error);
}
/*
* Wrapper around bwrite() so that we can trap
* write errors, and act accordingly.
*/
int
xfs_bwrite(
struct xfs_mount *mp,
struct xfs_buf *bp)
{
int error;
/*
* XXXsup how does this work for quotas.
*/
XFS_BUF_SET_BDSTRAT_FUNC(bp, xfs_bdstrat_cb);
bp->b_mount = mp;
XFS_BUF_WRITE(bp);
if ((error = XFS_bwrite(bp))) {
ASSERT(mp);
/*
* Cannot put a buftrace here since if the buffer is not
* B_HOLD then we will brelse() the buffer before returning
* from bwrite and we could be tracing a buffer that has
* been reused.
*/
xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
}
return (error);
}
Reference in New Issue View Git Blame Copy Permalink