[RFC PATCH 0/9] Bus layer cleanups

[Matt Turner]

On Tue, Sep 21, 2010 at 7:28 PM, Adam Jackson <ajax at redhat.com> wrote:
> This attempts to tackle Tiago's (justified) hatred for the IOADDRESS type
> in a slightly more elegant way.  We move more of the VGA hacks to vgahw,
> port it and int10 over to pciaccess' legacy I/O API, and then clean up
> some of the remaining detritus.
>
> THIS IS AN ABI BREAK.  Like, a pretty massive one.  Please do not pull
> this yet.  I'm mostly looking for r-b's here that the strategy looks
> sane.  The driver work needs to land before this lands, and it's pretty
> trivial, but I don't want to bother unless this is actually going to
> merge.
>
> Once (if) this lands, the only remaining direct usage of {in,out}[bwl]
> in xserver will be xf86SlowBcopy (jokes).  This also makes it plausible
> to convert the rest of the drivers to the pciaccess API, which then
> means we can drop roughly 137% of the content of compiler.h.

\o/

I'm not completely familiar with all the code, so take it with a grain of salt,
Reviewed-by: Matt Turner <mattst88 at gmail.com>

I tried to kill xf86SlowBcopy one time and it got reverted.
(aa07957373fd7cbe67458a001e4afd6a7f1ea37f). xf86-video-vesa uses
xf86SlowBcopy, so we can't get rid of it? This doesn't seem right.

Please kill compiler.h.

Matt