[Intel-gfx] Skylake graphics regression: projector failure with 4.8-rc3

[James Bottomley]

On Mon, 2016-09-19 at 08:09 -0700, James Bottomley wrote:
> On Sun, 2016-09-18 at 13:35 +0200, Thorsten Leemhuis wrote:
> > Hi! James & Paulo: What's the current status of this?
> 
> No, the only interaction has been the suggestion below for a revert,
> which didn't fix the problem.
> 
> >  Was this issue discussed elsewhere or even fixed in between? Just 
> > asking, because this issue is on the list of regressions for 4.8.
> 
> 
> I'm just about to try out -rc7, but it's not fixed so far.

OK, with -rc7 and the i915 fixes, there seems to be a marked
improvement.  I can no longer crash the crtc by using lots of xrandr
switches, which was the principal problem.

I've so far only got one of these in the logs

[14858.635035] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR*
CPU pipe B FIFO underrun

And the only residual problem seems to be that the monitor goes blank
periodically, but this can be fixed by switching resolution a couple of
times.

I haven't seen any of the link training errors so far and I've run
through my usual battery of be nasty to the external monitor tests.

James