[Xorg-driver-geode] libpciaccess patch - please test!

[Néstor Amigo Cairo]

I have a sent a logfile just a few days ago. Here it is:
https://bugs.freedesktop.org/show_bug.cgi?id=14919

I think little has changed since then, but anyways, I'm going to send
some more logs when I'm able to.

Forget about BIOS, it just messed me up, that's all.

I have little knowledge about kernel internals to make the changes right
now, but I will do it when I'm able to... :-S it's not like adding some
lines to a config file, it seems quite more complicated. I suppose I
have to change the proper lines of code inside the driver to set the
correct modes, am I right?

I thank you for being so fast in replying, I see that AMD has changed to
support FLOSS, and that's a good thing, indeed.

I had a bad feeling when I couldn't get this card working; now, I'm
starting feeling better, since I understand the problem.


El mar, 11-03-2008 a las 08:22 -0600, Jordan Crouse escribió:
> On 11/03/08 12:40 +0100, Néstor Amigo Cairo wrote:
> > I have tested it further with the Modeline, Jordan. And I have applied
> > the patch to X.org AMD driver. But still does not work for me using
> > TFT panel. It complains about an incorrect horiz sync. BTW, how does
> > VESA manage so well in all situations?? Is this a stupid question for
> > some reason?? I have also posted a message to the kernel folks, of
> > course.
> > I wonder how I could know the correct Modeline (maybe in the Manual,
> > some idea??), or making X.org recognize the panel someway, or
> > something else. X.org 2D support is further more important for me than
> > Framebuffer, since I can manage this with vesa, but not X.org, because
> > I need proper 2D acceleration for some multimedia tasks that eat too
> > much processor otherwise. And TFT panel support, of course, is a must.
> > It would be fine if I could put up working X.org AMD driver with
> > TFT/panel, I have tested it with a CRT, and it worked perfectly with
> > the patch (I have not tested it before with CRT)!!
> > It wouldn't make any big difference if I had to add some Modeline or
> > similar thing, just if I was able to guess it. But I can guess the
> > patch is still Ok, even if not enough to put it working.
> > 
> > I have tested GeodeLX driver two minutes ago, and it doesn't work with
> > panel, even with vga=normal  mode... :-S And yes, you are right: it's
> > a GeodeLX, not GX, despite of the BIOS, which still thinks it's a GX.
> > I got confused by that, but I have investigated it further and BIOS is
> > mistaken, so was me.
> 
> People have explained these things to you.   Your processor is an LX,
> and your BIOS thinks its an LX, or else Linux wouldn't boot.  vga= doesn't
> work for the lxfb.  And you need to apply the timings that you were given
> to make the lxfb work correctly with a TFT panel.
> 
> As for X, we cannot speculate on what your problem without the log
> files.  Please send them.
> 
> Jordan
>