RE: Unblock xserver-xorg-video-geode
> -----Original Message-----
> From: Julien Cristau [mailto:jcristau@debian.org]
> Sent: 2010?9?3? 16:46
> To: Huang, FrankR
> Cc: q-funk@iki.fi; Otavio Salvador; debian-release; Cui, Hunk
> Subject: Re: Unblock xserver-xorg-video-geode
>
> On Fri, Sep 3, 2010 at 09:33:19 +0800, Huang, FrankR wrote:
>
> > Otavio,
> >
> > I saw this thread from top to the end.
> > Is it because the patch
> > http://cgit.freedesktop.org/xorg/driver/xf86-video-
> geode/commit/?id=e9447f5335681a78cf87ebf8c9659a6fecfc9746
> > block the geode driver to be accepted?
> > If that is so, can we agree with the suggestion from Julien?
> > if ((pGeode->Output & (OUTPUT_PANEL | OUTPUT_DCON)) ==
> > (OUTPUT_PANEL | OUTPUT_DCON))
> >
> Well, or you could make it if (pGeode->Output == (OUTPUT_PANEL |
> OUTPUT_DCON)). Or alternatively, explain why the existing code is
> correct.
Otavio can explain more on that I think. He is more familiar with this
part than me.
>
> Another question is whether your EXA composite paths are reliable
enough
> now, or whether we'd better disable them for release.
I know you are not sure if the all rendering commits are reasonable with
current driver. Nearly all the rendering bugs are fixed by me in the
last three monthes with the help of others. And if you have joined the
xorg-driver-geode mail list, you can see this process I added these
patches. I have to admit that these files have not been reviewed by
enough users to prove that is right(It related to HW part of geode and
rendering knowledge, so not quite too people have time to do that). But
you should know after Jordan left in 2008, this part code is not
maintained for quite a long time and is broken severely. Users complain
much on that.
I refer to rendercheck in freedesktop as my debug tool and use ATI
driver as a reference too. At the same time, I ask lot of help from
Janothon on xorg-devel mail list for the last three monthes. It is not
easy to fix the rendering bug. By the way, some users on IRC and OLPC
users have applied the rendering patches and it looks fine now. How
could I prove more? Is it enough?
By the way, maybe you are not familiar with AMD strategy. We will
support geode for several years later. So please take it easy, this part
work is going on. I am now in charge of this part work and definitely we
are lack of experience. But we will try our best with the help of the
community to make our driver workable and better.
Thanks,
Frank
>
> Cheers,
> Julien
Reply to: