[ubuntu-x] i8xx legacy driver
Geir Ove Myhr
gomyhr at gmail.com
Wed Jul 21 20:37:16 BST 2010
On Wed, Jul 21, 2010 at 4:25 PM, Christopher James Halse Rogers
<raof at ubuntu.com> wrote:
> The wonderful world of i8xx freezes probably needs no introduction.
> What does need introduction is the xserver-xorg-video-intel driver in
> ppa:raof/aubergine¹. This construction of Chris Wilson's re-introduces
> a GEM-less UMS codepath, which should work around the GTT incoherency
> problems.
Do you have any more information about what the plan for this is? Drop
support for i8xx with the current driver and use UMS for those
chipsets? From what I gather from our bug reports, it does not really
matter if people use UMS or KMS, the problem is with UXA+GEM. The
ideal solution would probably be EXA+KMS (if that makes sense), but I
guess that means writing new code instead of recycling old one.
> This needs testing — on *all* Intel chips, not just i8xx — to both
> ensure that it doesn't break current chips and that the UMS codepath
> works successfully, and doesn't hit the same i8xx bugs that the current
> driver hits.
I suppose the libdrm in your PPA is supposed to be used with
xserver-xorg-video-intel. Do we need to use the nomodeset kernel
parameter to the kernel in order to trigger the UMS codepath?
I have tried both with and without nomodeset on my Thinkpad X61 Tablet
with 965GM chipset.
Without nomodeset it works fine in Lucid.
OpenGL-stuff don't seem to work with nomodeset on Lucid (but maybe
that is expected)
go at storhaugen:~$ glxgears
Segmentation fault
go at storhaugen:~$ glxdemo
Resize event
Resize event
Redraw event
../../intel/intel_bufmgr_gem.c:1048: Error preparing buffer map 0
(temporary VBO): Invalid argument .
Segmentation fault
go at storhaugen:~$ blender
Compiled with Python version 2.6.5.
Checking for installed Python... got it!
../../intel/intel_bufmgr_gem.c:1048: Error preparing buffer map 0
(temporary VBO): Invalid argument .
Segmentation fault
Geir Ove
More information about the Ubuntu-x
mailing list