Re: [PATCH] ide: add dcache flushing after PIO

[Russell King <rmk@arm.linux.org.uk>]

On Wed, Dec 21, 2005 at 05:54:16PM +0000, Russell King wrote:
> On Wed, Dec 21, 2005 at 04:57:05PM +0100, Bartlomiej Zolnierkiewicz wrote:
> > On 12/21/05, Tejun Heo <htejun@gmail.com> wrote:
> > > clears all cpu caches before IO and, as DMA IO doesn't touch any cpu
> > > caches, it doesn't do anything after IO.  The previous patch adds a
> > > flush_dcache_page after IO which makes sure that the kernel cache line
> > > is gone, but it doesn't do anything to make sure that there's no dirty
> > > user-mapped cachelines hanging around before IO.
> > >
> > > I couldn't find an exemplary driver doing this kind of things with page
> > > caches.  Most other flush_dcache_page usages I looked at didn't deal
> > > with user mapped page caches.
> > 
> > After reading excellent explanation by rmk I think that it should be
> > fixed at VM layer (filemap_nopage() perhaps).
> 
> DaveM vetoed that with good reason - the VM layer doesn't know whether
> the driver is doing PIO or not.  If it is doing PIO, it needs the
> cache flush.  If it's doing DMA, the cache flush is entirely a
> performance bottleneck.

Bart,

Did you miss my message, or have you decided to pay no further
attention to this issue?

If the latter, I will mark IDE with a dependency of BROKEN || !ARM
until it's fixed.  It's not fair on folk to lead them to think that
PIO-based IDE might work for them when the subsystem is technically
broken on ARM architectures.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core