Re: [git pull] drm fixes for 4.9-rc4

[Daniel Vetter <daniel@ffwll.ch>]

On Fri, Nov 04, 2016 at 05:58:55PM -0400, Theodore Ts'o wrote:
> On Fri, Nov 04, 2016 at 01:38:25PM -0700, Linus Torvalds wrote:
> > On Wed, Nov 2, 2016 at 5:31 PM, Dave Airlie <airlied@gmail.com> wrote:
> > >
> > > There are a set of fixes for an oops we've been seeing around MST
> > > display unplug,
> > 
> > Side note: I heard from a couple of people at the KS that said that
> > they had had problems with suspend/resume after plugging in to a 4k
> > display (but _only_ a 4k display - apparently normal FHD displays
> > didn't show this). I think at least one was USB3/Thunderbolt. Ted with
> > a Lenovo laptop (intel GPU) was one, I forget who else mentioned this.
> 
> Actually, it's after a unplugging from a Dell 30" monitor with a 3k
> display (2560 x 1920).  This is after I've carefully deactivated the
> video output to the Dell 30" monitor, unplugged the Dell 30" monitor
> (at which point the system becomes non-responsive for 2-3 seconds for
> reasons unknown), and only suspending after the system has recovered
> from the unplug.
> 
> At that point, it's a 20-30% chance that the system will never come
> back after a suspend.  So I have to make a point of saving all of my
> editor buffers, etc., since I never can know whether my laptop will
> come back.
> 
> This was happening for years and years on the T540p laptop, as well as
> my new T460 laptop.  I've complained about this in the past, and
> gotten no response, and I've just gotten used to the fact that if I'm
> transitioning from home (where I have the 30" display) to work,
> there's a good chance the resume will lock up, and I will be forced to
> push the power button for 8 seconds to forcibly power down the laptop
> to recover from the suspend.  :-(
> 
> I agree with Linus's suspicion that I probably need to bite the bullet
> and just buy a new SST monitor, and that will probably make the
> problem go away.  But if the bug can be fixed, that would be really
> great.

This pull contains a bunch of fixes for oopses on mst connector
unplugging, and iirc those started popping up with the connector
refcounting we landed in 4.8. Definitely should retest with this before
spending more time debugging.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch