Re: DRI not working with libgl1-mesa-dri (r300)
On Thu, 2006-01-12 at 14:50 +0100, Ruben wrote:
> At Thu, 12 Jan 2006 12:15:34 +0100, Michel Dänzer wrote:
> >
> > Maybe, but the X server creates the device nodes on the fly as necessary
> > anyway.
>
> Not in my case it seems. I tried creating the device node manually
> (mknod /dev/dri/card0 c 226 0) before starting X, but then X removes
> it.
The X server creates and removes the nodes as they are (not) needed. If
the missing node was the problem, the error would be 'No such file or
directory'.
> > > drmOpenDevice: node name is /dev/dri/card0
> > > drmOpenDevice: open result is -1, (No such device)
> >
> > 'No such device' usually occurs when the kernel module isn't loaded.
>
> That was what I thought at first, but I did have DRM+Radeon enabled in
> the kernel. I even changed my kernel configuration to have those as
> modules, but that doesn't change a thing (and it shouldn't).
You're right that it shouldn't in theory, but in reality the modular
case is much better tested.
> X does load the radeon module though, but still no entry in /dev/dri.
What about /proc/dri/?
> ruben@guava ~ $ dmesg | grep drm
> [drm] Initialized drm 1.0.0 20040925
There should be some output from the radeon module here. Please post the
kernel output that results from
modprobe -r radeon; modprobe radeon
(or the full kernel output, if in doubt)
> ruben@guava ~ $ gunzip -c /proc/config.gz | grep DRM
> CONFIG_DRM=m
> # CONFIG_DRM_TDFX is not set
> # CONFIG_DRM_R128 is not set
> CONFIG_DRM_RADEON=m
This is fine.
Have you verified that the 'device scape' patch you're using doesn't
cause the problem? I must confess I haven't got around to running 2.6.15
yet, so there's the possibility of radeon DRM breakage there, although
I'd expect more noise about that on this list. :)
--
Earthling Michel Dänzer | Debian (powerpc), X and DRI developer
Libre software enthusiast | http://svcs.affero.net/rm.php?r=daenzer
Reply to: