Xgl server

[Benjamin Herrenschmidt]

On Fri, 2004-11-05 at 14:27 -0500, Jon Smirl wrote:
> On Fri, 05 Nov 2004 01:03:05 +0100, David Reveman <c99drn at cs.umu.se> wrote:
> > I've been doing some work to get an X server running on top of
> > OpenGL/glitz and I've got something that works pretty well now. I stuck
> > the code into the "xserver" tree and './configure --enable-xglserver'
> 
> Have you tried it on a fast ATI/NVidia driver, how's performance?
> Did you use pbuffers?
> What is an example of something that can't be accelerated?
> 
> I'm still working on mesa-solo. There is still a lot of work to do
> there. Main problem is that mesa-solo relies on fbdev for mode suppot.
> fbdev only supports one head so I'm in the process of adding second
> head support to fbdev. I also need to merge fbdev and drm to make them
> stop fighting over the hardware.  DRM is in pretty good shape now,
> most of the work needed is in fbdev. All hardware drivers in mesa are
> missing pbuffer support. If any one wants to help with this there is
> plenty of work to do.
> 
> I would really like to see an XGL running on mesa-solo. It is Linux's
> answer to Longhorn.

Where are you of your work on dual head fbdev ? I'm working on radeonfb
a bit lately, mostly power management stuff, but I want to add support
for mirroring using the second CRTC and may dual head. The problem with
the later is the fbdev API itself isn't very good at it. Userland need
to know what head is "linked" to what head, among others, plus the need
to dynamically add/remove the second head when user enables mirroring
for example.

(I also found some TV output code for mobility chips in open-beos that
will eventually add to radeonfb as well).

Ben.