xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

xfs: log timestamp updates

Browse Source 
Timestamps on regular files are the last metadata that XFS does not update
transactionally.  Now that we use the delaylog mode exclusively and made
the log scode scale extremly well there is no need to bypass that code for
timestamp updates.  Logging all updates allows to drop a lot of code, and
will allow for further performance improvements later on.

Note that this patch drops optimized handling of fdatasync - it will be
added back in a separate commit.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
This commit is contained in:
Christoph Hellwig
2012-02-29 09:53:52 +00:00
committed by Ben Myers
parent 281627df3e
commit 8a9c9980f2
13 changed files with 65 additions and 321 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
fs/xfs/xfs_sync.c
View File

@@ -336,32 +336,6 @@ xfs_sync_fsdata(
return error;
}
int
xfs_log_dirty_inode(
struct xfs_inode *ip,
struct xfs_perag *pag,
int flags)
{
struct xfs_mount *mp = ip->i_mount;
struct xfs_trans *tp;
int error;
if (!ip->i_update_core)
return 0;
tp = xfs_trans_alloc(mp, XFS_TRANS_FSYNC_TS);
error = xfs_trans_reserve(tp, 0, XFS_FSYNC_TS_LOG_RES(mp), 0, 0, 0);
if (error) {
xfs_trans_cancel(tp, 0);
return error;
}
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
return xfs_trans_commit(tp, 0);
}
/*
* When remounting a filesystem read-only or freezing the filesystem, we have
* two phases to execute. This first phase is syncing the data before we
@@ -385,16 +359,6 @@ xfs_quiesce_data(
{
int error, error2 = 0;
/*
* Log all pending size and timestamp updates. The vfs writeback
* code is supposed to do this, but due to its overagressive
* livelock detection it will skip inodes where appending writes
* were written out in the first non-blocking sync phase if their
* completion took long enough that it happened after taking the
* timestamp for the cut-off in the blocking phase.
*/
xfs_inode_ag_iterator(mp, xfs_log_dirty_inode, 0);
/* force out the log */
xfs_log_force(mp, XFS_LOG_SYNC);