Experimental options

[Vladimir Dergachev]

On Sat, 11 Dec 2004, Michel [ISO-8859-1] Dänzer wrote:

> On Sat, 2004-12-11 at 11:19 -0500, Vladimir Dergachev wrote:
>>>> The reason this is useful is that such trick greatly accelerates 3d on
>>>                                                                   ^ 2? :)
>>>> large resolution screens as 2d driver can also use DRM driver and this
>>>> works perfectly well. This also facilitates work on 3d driver.
>>>>
>>>>      The 2d speedup is quite considerable, however one would expect all GL
>>>> apps to break because of this, as even software rendering won't work.
>>>
>>> libGL should fall back to GLX when the *_dri.so isn't around.  Is it
>>> not?  Even so, it would be nice to make some sort of change to make
>>
>> I don't think it does. At the very least this was broken for me - or maybe
>> I screwed up something else.
>
> It should definitely fall back to indirect rendering, just like when
> there's a problem loading any of the existing 3D drivers.

Well, I have just committed the code, protected by X_R300_DRM option - if 
not enabled (which is default) the DRM is not used in any way.

I double-checked (removing r300_dri.so from /usr/X11/lib/modules/dri) and
I still get glxgears lockup. Also, glxinfo returns info as for direct 
rendering driver - as, I suppose, it should.

I looked around in radeon_dri.c and could not find any place to gracefully 
disable DRI without disabling its use by Xserver.

Perhaps a better (and long-term) solution is to an early version of 
Nicolai Haehnle's driver that does not crash - it effectively implements
all-fallback DRI driver.

Suggestions (and patches:) ) are very welcome !

                         best

                            Vladimir Dergachev


>
>
> -- 
> Earthling Michel Dänzer      |     Debian (powerpc), X and DRI developer
> Libre software enthusiast    |   http://svcs.affero.net/rm.php?r=daenzer
>