xfs
[Top] [All Lists]

Re: [PATCH 2/7] xfs: split inode flushing from xfs_sync_inodes_ag

To: Sujit Karataparambil <sjt.kar@xxxxxxxxx>
Subject: Re: [PATCH 2/7] xfs: split inode flushing from xfs_sync_inodes_ag
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Fri, 15 May 2009 13:22:47 -0400
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <921ca19c0905142152u2a8012a9jba6f1d5b856df7ad@xxxxxxxxxxxxxx>
References: <20090514171233.942489000@xxxxxxxxxxxxxxxxxxxxxx> <20090514171558.298098000@xxxxxxxxxxxxxxxxxxxxxx> <921ca19c0905142152u2a8012a9jba6f1d5b856df7ad@xxxxxxxxxxxxxx>
User-agent: Mutt/1.5.18 (2008-05-17)
On Fri, May 15, 2009 at 10:22:31AM +0530, Sujit Karataparambil wrote:
> here also should there be an
> xfs_sync_inode_data(ip, flags);
> xfs_iflush(ip, XFS_IFLUSH_SYNC);
> 
> in out_unlock.

No.  We only want to perform the iflush if all the preconditions are met
in the !SYNC_WAIT case.  And we certainly do not want to do a data
writeout from the metadata flush case - for one thing it does require
the iolock, but most importantly the point of this series is to separate
the two actions.

<Prev in Thread] Current Thread [Next in Thread>