Re: [PATCH] Define wc_wmb, a write barrier for PCI write combining

[Jesse Barnes <jbarnes@virtuousgeek.org>]

On Tuesday, February 28, 2006 9:52 am, Bryan O'Sullivan wrote:
> >   (In particular I wanted this for the sysfs PCI interface.
> > Userspace apps can map PCI resources there and it would be nice if
> > they could map them with WC semantics if requested.)
>
> They already sort of can.  It just happens that most arches ignore the
> WC parameters.

Only on a per-driver basis though at this point, right?  The sysfs code 
is generic across architectures and exports PCI resources independently 
from the driver, so it needs some way of letting the user communicate 
that they want WC behavior.

> > I don't think it addresses the flushing issue you seem to be
> > concerned about though.
>
> Yes, I think I could have made my original wording a bit clearer.  I
> don't care if writes have hit the device, merely that they do so in an
> order that I control.

Ah, ok.  In that case maybe mmiowb *is* what you want.  Or at the very 
least mmiowcb :).

Jesse