Ocfs2 devel - Feb 2016 - [PATCH 2/3] xfs: don't use ioends for direct write completions

On Wed, Feb 03, 2016 at 07:40:15PM +0100, Christoph Hellwig
wrote:
> We only need to communicate two bits of information to the direct I/O
> completion handler:
> 
>  (1) do we need to convert any unwritten extents in the range
>  (2) do we need to check if we need to update the inode size based
>      on the range passed to the completion handler
> 
> We can use the private data passed to the get_block handler and the
> completion handler as a simple bitmask to communicate this information
> instead of the current complicated infrastructure reusing the ioends
> from the buffer I/O path, and thus avoiding a memory allocation and
> a context switch for any non-trivial direct write.  As a nice side
> effect we also decouple the direct I/O path implementation from that
> of the buffered I/O path.
> 
> Signed-off-by: Christoph Hellwig <hch at lst.de>
> Reviewed-by: Brian Foster <bfoster at redhat.com>
This change is now dependent on the preceeding direct IO API
changes. Do I a) take the DIO API change through the XFS tree, or
b) use the older version of the patch that didn't have this
dependency and let somebody else deal with the API change and merge
issues?

I'm happy to take the DIO API change through the XFS tree, if that's
the fastest/easiest way to get the necessary DIO subsystem fixes
into the mainline tree for XFS. As such, the for-next tree that I'm
building right now will include the DIO API change patch....

-Dave.

-- 
Dave Chinner
david at fromorbit.com

On Mon, Feb 08, 2016 at 12:00:26PM +1100, Dave Chinner
wrote:
> On Wed, Feb 03, 2016 at 07:40:15PM +0100, Christoph Hellwig wrote:
> > We only need to communicate two bits of information to the direct I/O
> > completion handler:
> > 
> >  (1) do we need to convert any unwritten extents in the range
> >  (2) do we need to check if we need to update the inode size based
> >      on the range passed to the completion handler
> > 
> > We can use the private data passed to the get_block handler and the
> > completion handler as a simple bitmask to communicate this information
> > instead of the current complicated infrastructure reusing the ioends
> > from the buffer I/O path, and thus avoiding a memory allocation and
> > a context switch for any non-trivial direct write.  As a nice side
> > effect we also decouple the direct I/O path implementation from that
> > of the buffered I/O path.
> > 
> > Signed-off-by: Christoph Hellwig <hch at lst.de>
> > Reviewed-by: Brian Foster <bfoster at redhat.com>
> 
> This change is now dependent on the preceeding direct IO API
> changes. Do I a) take the DIO API change through the XFS tree, or
> b) use the older version of the patch that didn't have this
> dependency and let somebody else deal with the API change and merge
> issues?
> 
> I'm happy to take the DIO API change through the XFS tree, if
that's
> the fastest/easiest way to get the necessary DIO subsystem fixes
> into the mainline tree for XFS. As such, the for-next tree that I'm
> building right now will include the DIO API change patch....
Right now this series is in a stable branch in the XFS tree:

git://git.kernel.org/pub/scm/linux/kernel/git/dgc/linux-xfs.git xfs-dio-fix-4.6

If you want to push it through some other tree, please let me know
when/where it is committed so I can rebuild the XFS for-next branch
appropriately from a stable commit/branch...

Cheers,

Dave.
-- 
Dave Chinner
david at fromorbit.com