Created attachment 122351 [details]
Log output from /sys/class/drm/card0/error after hang

When DRI is enabled, I periodically (every 5min or so) get (like everyone elses with the same hardware it seems):

ar 15 17:14:00 xps13 kernel: [drm] stuck on render ring
Mar 15 17:14:00 xps13 kernel: [drm] GPU HANG: ecode 9:0:0x84df3ffc, in glxgears [9275], reason: Ring hung, action: reset
Mar 15 17:14:00 xps13 kernel: [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Mar 15 17:14:00 xps13 kernel: [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
Mar 15 17:14:00 xps13 kernel: [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Mar 15 17:14:00 xps13 kernel: [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
Mar 15 17:14:00 xps13 kernel: [drm] GPU crash dump saved to /sys/class/drm/card0/error
Mar 15 17:14:00 xps13 kernel: drm/i915: Resetting chip after gpu hang
Mar 15 17:14:01 xps13 kernel: [drm] RC6 on
Mar 15 17:14:08 xps13 kernel: [drm] stuck on render ring
Mar 15 17:14:08 xps13 kernel: [drm] GPU HANG: ecode 9:0:0x87f53ff9, in glxgears [9275], reason: Ring hung, action: reset
[..] and so on

Screen (eDP1/integrated laptop screen) freezes for a few seconds during the hang. Running glxgears at full speed (vsync off) helps triggering the problem more quickly though it happens either way.
Disabling rc6 on kernel command line also works around the problem.

This is on a XPS13 "2016" (ie brand new model) with a i7-6560U CPU and "Iris" graphics.
The kernel used is Linux xps13 4.5.0-1-drm-intel-nightly #1 SMP PREEMPT Tue Mar 15 13:41:07 PDT 2016 x86_64 GNU/Linux (ie built from git today) though the same problem happens on mainline 4.5.0 (also from today).

I use xorg 1.18.1-3 with the video-intel xorg driver 1:2.99.917+560+gd167280 on ArchLinux.