[Intel-gfx] screen freeze with 5.2-rc6 Dell XPS-13 skylake i915

[James Bottomley]

On Wed, 2019-07-10 at 18:45 +0200, Paul Bolle wrote:
> James Bottomley schreef op wo 10-07-2019 om 09:32 [-0700]:
> > You seem to be getting it to happen much more often than I can.
> > Last
> > night, on the below pull request it took me a good hour to see the
> > freeze.
> 
> Yes. Sometimes within a minute of resuming. Typing stuff into
> evolution seems to help with triggering this. It's all a bit
> mysterious, but this message alone frooze my laptop a few times.
> Seriously!
> 
> > Sure, my current testing indicates it's somewhere inside this pull
> > request:
> > 
> > Merge: 89c3b37af87e eb85d03e01c3
> > Author: Linus Torvalds <torvalds at linux-foundation.org>
> > Date:   Wed May 8 21:35:19 2019 -0700
> > 
> >     Merge tag 'drm-next-2019-05-09' of
> > git://anongit.freedesktop.org/drm/drm
> >     
> >     Pull drm updates from Dave Airlie:
> 
> Lazy question: how does one determine the first and last commit
> inside a merge request? Can I simply do
>     good   a2d635decbfa9c1e4ae15cb05b68b2559f7f827c^
>     bad    a2d635decbfa9c1e4ae15cb05b68b2559f7f827c
> 
> for git bisect?

I think so.  I actually did

	bad	a2d635decbfa9c1e4ae15cb05b68b2559f7f827c
	good	89c3b37af87ec183b666d83428cb28cc421671a6

But I think ^ steps down in the main branch.  Note, I've only done a
couple of hours run on what I think is the good commit ... and actually
I'd already marked a2d635decbfa9c1e4ae15cb05b68b2559f7f827c as good
until the screen froze just after I'd built the new bisect kernel, so
there's still some chance 89c3b37af87ec183b666d83428cb28cc421671a6 is
bad.

> > So I was about to test out the i915 changes in that but since my
> > laptop is what I use for daily work, it's a bit hard (can't freeze
> > up on video calls for instance).
> 
> I usually use one of the ThinkPads from my embarrassing pile of
> outdated hardware to do nasty bisects, but I'm not about to loose any
> income if my much appreciated XPS 13 is out of order for a while.

I can get back to it this afternoon, when I'm done with the meeting
requirements and doing other dev stuff.

James