Re: Wrong DIF guard tag on ext2 write

[Nick Piggin <npiggin@suse.de>]

On Tue, Jun 08, 2010 at 06:47:18PM +1000, Dave Chinner wrote:
> On Tue, Jun 08, 2010 at 09:15:35AM +0200, Christof Schmitt wrote:
> > > But my original changes were along the lines of what Jan mentioned
> > > earlier (hooking into page_mkwrite and waiting for writeback.  I could
> > > have sworn that I only did it for ext[23] and that XFS waited out of the
> > > box but git proves me wrong).  Anyway, I'll try to get some benchmarking
> > > happening later this week.
> > 
> > Is there a patch with this change available somewhere? It might be
> > useful to patch a kernel with this XFS change for reliable DIF/DIX
> > testing.
> 
> Adding a wait_on_page_writeback() call to page_mkwrite() won't help
> by itself - you need to unmap the pages after transitioning them
> from dirty to writeback but before the hardware starts processing
> the IO if you want to lock out mmap writes this way....

Actually as Jan pointed out, clear_page_dirty_for_io already does
this. So it seems pretty doable. get_user_pages remains a problem
(which I believe NFS just ignores) but it could be avoided by
just disallowing it.