[Intel-gfx] GM45/G45 status

Martin Olsson mnemo at minimum.se
Fri Oct 10 23:39:43 CEST 2008


Keith Packard wrote:
> The G45 motherboard, on the other hand, is a mess
> 
>      1. Without any DRM, I'm getting GPU lockups when accessing large
>         amounts of memory. Something is funky here, and I don't
>         understand it. The error register reports a page fault, which
>         seems a bit odd.
>      2. With GEM. The driver stalls as soon as it calls the throttle
>         IOCTL, which will wait for previous commands to complete. It
>         looks like the hw status page values written by the GPU are not
>         visible to the CPU. Most concerning.
> 

Thanks for the update!

I got a Gigabyte GA-EG45M-DS2H and I'm willing to help out testing G45 bits.
I know how to compile and install xf86-video-intel and if you explain how to
build/install the GEM stuff I can test that as well. I'm on Intrepid as my base.

If there is any particular xorg.conf options you want me to test just let me
know and I will post xorg_log/stacks/dmesg back.

---

Currently, even with DRI=false and EXANoComposite I'm getting a CPU spin inside
evict_and_alloc_block() which happens while _gdm_ is starting (it's inside the
AnimCurScreenBlockHandler function). I have not tried to turn off the animated
cursor (not sure how to do it?). I'm attaching gdb+xorglog+dmesg for this config.





Martin


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xorg_2008oct10_AccelMethodEXA_DRIfalse_EXANoCompositeTrue_modeDebug_cpu_spin_before_even_getting_to_gdm_login_screen__dmesg.log
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20081010/174f09ec/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xorg_2008oct10_AccelMethodEXA_DRIfalse_EXANoCompositeTrue_modeDebug_cpu_spin_before_even_getting_to_gdm_login_screen__gdb_bt_etc.log
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20081010/174f09ec/attachment-0001.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xorg_2008oct10_AccelMethodEXA_DRIfalse_EXANoCompositeTrue_modeDebug_cpu_spin_before_even_getting_to_gdm_login_screen__xorg_log.log
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20081010/174f09ec/attachment-0002.ksh>


More information about the Intel-gfx mailing list