[PATCH 1/1] Uncomment PCI access functions inside server

[Jesse Barnes]

On Sun, 03 May 2009 21:48:19 -0300
Tiago Vignatti <vignatti at freedesktop.org> wrote:

> Hi,
> 
> Jesse Barnes escreveu:
> > On Tue, 28 Apr 2009 22:33:08 -0300
> > Tiago Vignatti <vignatti at freedesktop.org> wrote:
> > 
> >> This reintroduces secondary cards initialization.
> >>
> >> We're currently living in a completely mess regarding the resources
> >> access control. Some accesses are done by libpciaccess, other by
> >> Xorg and other simply don't exist yet (e.g. arbitration of legacy
> >> VGA). We can discuss about a beautiful redesign later. In this
> >> moment I just want to put back the multiple card, multiple head
> >> functionality. The first step is to reintroduce secondary cards
> >> initialization. I'll keep the work.
> >>
> >> PS.: AFAIK the currently implementation of int10's x86emu backend
> >> is not working to POST secondary cards (BUG). I POSTed using vm86
> >> instead.
> >>
> >> Signed-off-by: Tiago Vignatti <vignatti at freedesktop.org>
> >> ---
> >>   hw/xfree86/common/xf86pciBus.c |   33
> >> --------------------------------- 1 files changed, 0 insertions(+),
> > 
> > Can we create a new define for these and put this code under that?
> 
> Are you saying to put these functions inside the pci library? I'm not 
> sure if I got you corretly. Can you be more specific, Jesse?

That would be one option (probably better).  I was actually thinking of
making all the #if 0s you removed into #if XF86_RAC or something; we
could put all the resource management stuff that shouldn't be necessary
with libpciaccess & vga arbitration under that.


-- 
Jesse Barnes, Intel Open Source Technology Center