xf86-video-intel 2.2.1 release testing

[Jin, Gordon]

Ben E. Hard wrote on Wednesday, January 23, 2008 12:49 AM:
> Am Dienstag, 22. Januar 2008 schrieb Michel Dänzer:
>> On Tue, 2008-01-22 at 14:51 +0100, Ben E. Hard wrote:
>>> Am Dienstag, 22. Januar 2008 schrieb Jin, Gordon:
>>>> Jesse is planning to release 2.2.1 soon in this week. So it will be
>>>> good if you could test xf86-video-intel-2.2-branch at this point
>>>> and report bugs asap. You could use the latest released
>>>> xserver(1.4), Mesa(7.0.2 or 7.0.3-rc1), DRM(in the recent kernel).
>>> 
>>> I tried to compile xf86-video-intel from git on debian lenny and
>>> updated xserver-xorg[-input, dev] first from unstable. I compiled
>>> the drm, libdrm, intel module and mesa as descriped on
>>> http://intellinuxgraphics.org/install.html.
>>> 
>>> Everthing compiled and I got it working, but bug #13722 still
>>> exists and, secondly, rendering is not working anymore. I found
>>> some probably related message in Xorg.0.log: (EE) AIGLX error:
>>> dlsym for __driCreateNewScreen_20050727 failed
>>> (/usr/lib/dri/i915_dri.so: undefined symbol:
>>> __driCreateNewScreen_20050727) (EE) AIGLX: reverting to software
>>> rendering. 
>> 
>> mesa Git requires xserver Git for AIGLX to work.
>> 
>> You could use mesa_7_0_branch as suggested above or
>> leave /usr/lib/dri/i915_dri.so unchanged from your distro and use the
>> mesa Git build from another location on the client side though.
> 
> Thanks, that helped. I installed xserver-xorg and dependend packages
> from unstable and xf86-video-intel from git. Now i get
> ~$ glxinfo | grep renderMesa DRI Intel(R)
> Failed to initialize TTM buffer manager.  Falling back to classic.
> direct rendering: Yes
> OpenGL renderer string: Mesa DRI Intel(R) 830M 20061102
> 
> don't know about the TTM stuff, but at least rendering is ok.

Thanks for testing. But I was not suggesting to try mesa/drm master tip for this release testing as they are unstable at this point.
The TTM failure seems as http://bugs.freedesktop.org/show_bug.cgi?id=13922, probably a cross-component compat issue.

> And xrandr still thinks, the LVDS is activated, while i810switch
> tells me, that it is off (which is correct).

Is there a bug tracker for this?

-Gordon