Re: [RFC] Badness in __mutex_unlock_slowpath with XFS stress tests

[Suparna Bhattacharya <suparna@in.ibm.com>]

On Fri, Mar 10, 2006 at 03:49:25PM +0000, Christoph Hellwig wrote:
> On Fri, Mar 10, 2006 at 11:50:20AM +1100, Nathan Scott wrote:
> > Something like this (works OK for me)...
> 
> Yeah, that should work for now.  But long-term we really need to redo
> direct I/O locking to have a common scheme for all filesystems.  I've heard
> birds whistling RH patches yet another scheme into RHEL4 for GFS an it's
> definitly already far too complex now.

Yup, getting rid of the need for all these confusing locking
modes was one of the objectives in mind for DIO simplification.
(http://www.kernel.org/pub/linux/kernel/people/suparna/DIO-simplify.txt)
Once we have an efficient range locking or similar mechanism in place
(Chris Mason is working on a patch), then it should be possible to push
out all of the i_mutex locking to higher level routines, outside of
direct-io.c.

Longer term, it would be nice to be able to rethink and further simplify
the whole _nolock equiv versions for VFS write methods. Especially the
percolation down to sync_page_range_nolock, etc.

Regards
Suparna

> 
> --
> To unsubscribe, send a message with 'unsubscribe linux-aio' in
> the body to majordomo@kvack.org.  For more info on Linux AIO,
> see: http://www.kvack.org/aio/
> Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>

-- 
Suparna Bhattacharya (suparna@in.ibm.com)
Linux Technology Center
IBM Software Lab, India