[Savage40] New binary snapshots

[Carl S. Isgro]

Felix,

Ok I commented out the video ram line, turned on the shadow status and
now DRI comes up and everything seems very quick.

The autoprobe of memory picked the correct size. The setting was a
remnant from XFree86 config. This machine was upgrade many times.

So it looks like the original problem I reported came from the video ram
size line in the xorg.conf file.

Thanks for all the help and I will report any other issues going
forward.


On Thu, 2005-02-03 at 17:37 +0100, Felix Kühling wrote:
> Am Donnerstag, den 03.02.2005, 11:16 -0500 schrieb Carl S. Isgro:
> > Felix,
> > 
> > Attached is the kernel log and a Xorg log with the debug=1 option set on
> > module load.
> > 
> > Hope it helps.
> 
> I found your problem. I already stumbled over your odd video RAM size
> before (set to 32767 in your xorg.conf), but didn't think it was a
> problem. Your video RAM size is not a multiple of 4KB (the page size),
> therefore the DRM refuses to create a map of that size. You should
> either set it to 32768 or leave it to the auto-detection. You'll also
> notice that all the MTRR splitting messages will disappear from your
> Xorg.0.log, because 32MB can be covered by a single MTRR.
> 
> I'll add something to the 2D driver that catches non-page-aligned video
> RAM sizes.
> 
> > 
> > My boot up text console is not a framebuffer console. Does that matter?
> > If so I can recompile kernel to use framebuffer console. I originally
> > disabled it because I thought it may be interfering.
> 
> That's ok. Vesa framebuffer should work too, but the new SavageFB is
> problematic at the moment.
> 
> > 
> > Another piece of info that may help is the binary snapshot from
> > September had everything working except occasional artifacts would
> > appear on the screen.
> > 
> > Also, with the current one and the settings in the present state I have
> > had no lockups.
> 
> Without shadow status you may get lockups with 3D applications once DRI
> is enabled. Could you try shadow status again with DRI enabled? My guess
> is that it will work correctly then.
> 
> > 
> > 
> > On Thu, 2005-02-03 at 01:40 +0100, Felix Kühling wrote:
> [snip]
> 
-- 
Carl S. Isgro
carl at csisgro.com