This commit makes my SKL-Y ultramobile SDP (UMSL7) hang hard if X is restarted: commit c9f038a1a5924352ab8e510e4a45ac57b08db391 Author: Matt Roper <matthew.d.roper@intel.com> Date: Mon Mar 9 11:06:02 2015 -0700 drm/i915: Don't assume primary & cursor are always on for wm calculation (v4)
One for Matt to look at I guess.
By "if X is restarted" do you mean the hang only happens if you run X, kill X, and then start X again? The referenced commit was intended to avoid some kernel panics (NULL deref) that we were hitting before; it sounds like I might have overlooked something in the SKL codepath since I don't think there have been any issues on non-SKL platforms. Also adding Tvrtko to Cc list in case he has any SKL-specific insight into what might be going wrong.
'systemctl restart lightdm' is how I reproduce it, I'll try other methods. And yes I noticed that the other commit mentioned in the description (59bea8822f2d2e0) was what triggered the oopses, X couldn't be killed with the earlier kernel when I was bisecting this. IIRC this only happens on the SKL-Y which has a builtin eDP panel, but not on SKL-S hooked to my DP monitor. Damien said that it's working on his system though, so perhaps I'm just running an obsolete bios version? (one that the system came with)
Do you have a serial console setup to capture some logs? I don't have a fully fledged X setup yet, but I think the same or similar hardware survives plain X restarts fine here.
I don't have a null cable needed for a serial console debugging.. yet. This could be related to / dupe of #89727 btw.
Could be a duplicate yes. You can also try netconsole.
It sounds like bug 89727 has magically resolved itself; is this one resolved now too or do we still have an issue?
oh right, I forgot this wasn't marked as duplicate. drm-intel-next-2015-04-10 was still bad, -nightly from yesterday was good. I'd still need to find what fixed it..
bisect points at 70a101f8639a, the first commit after that tag.. 4.1 needs that then?
(In reply to Timo Aaltonen from comment #9) > bisect points at 70a101f8639a, the first commit after that tag.. 4.1 needs > that then? Thanks for bisecting back to what fixed the issue. For QA, that's commit 70a101f8639a712576c3ca1d910564c7e2ec91d8 Author: Matt Roper <matthew.d.roper@intel.com> Date: Wed Apr 8 18:56:53 2015 -0700 drm/i915: Switch to full atomic helpers for plane updates/disable, take two Setting this to resolved now.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.