[Intel-gfx] [Bug 51061] GPU Hang using VAAPI

[Daniel Vetter]

On Wed, Jun 20, 2012 at 09:21:07PM +0200, Angela wrote:
> > > Updated xorg-edgers, updated kernel (>3.5-rc3), merged
> > > drm-intel-next-queued. "Fixed" the two conflicts, are they OK ?
> > >
> > > No console, X starts in Vesa Mode.
> > >
> > > Added the new kernel dmesg, my "fixed" conflict's and a good kernel
> > dmesg.
> > >
> > > What's next.
> > 
> > Looks like the kernel couldn't communicate with the chip at all, all reads
> just
> > fail. I've pushed out a for-angela branch to my kernel playground at
> > 
> > http://cgit.freedesktop.org/~danvet/drm
> > 
> > Which contains recent drm-intel-next-queued with rc3 merged. That one
> > boots at least correctly here. Although your merge resolution looks
> correct,
> > so I have no idea what exactly went wrong.
> > -Daniel
> > 
> 
> Used head kernel and merged for-angela. Console and X are fine now, for drm
> dmesg see below.
> 
> > contains recent drm-intel-next-queued with rc3 merged
> I could also have taken for-angela alone, does it matter ?
> 
> Still have the VAAPI problem and still cannot grab error_state
> 
> [   416.677] (WW) intel(0): flip queue failed: Input/output error
> 
> cat: /sys/kernel/debug/dri/0/i915_error_state: Cannot allocate memory
> 
> With ppa kernel I didn't had a console and the system hung (could not ssh),
> will you add the "couldn't communicate with the chip at all" patch to
> drm-intel-next-queued, so I could check the error_state once again with the
> generated ppa 3.4 kernel. Will we be more lucky ? When will it be ready to
> try ?

It should all work, but something is seriously broken for you. Could you
bisect where this started to blow up for you maybe?
-Daniel
-- 
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365 57 48