Bug 98325

Summary:	GPU HANG: ecode 9:0:0x85dffffb
Product:	Mesa	Reporter:	Milan Bouchet-Valat <nalimilan>
Component:	Drivers/DRI/i965	Assignee:	Ian Romanick <idr>
Status:	RESOLVED WORKSFORME	QA Contact:	Intel 3D Bugs Mailing List <intel-3d-bugs>
Severity:	critical
Priority:	medium	CC:	intel-gfx-bugs
Version:	unspecified
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:
Attachments:	journalctl -b output /sys/class/drm/card0/error

Description Milan Bouchet-Valat 2016-10-19 17:43:58 UTC

I'm using Fedora 24 with kernel 4.7.5-200, GPU Intel Corporation HD Graphics 520 (rev 07). Today I've been forcefully logged out of my X session, and I've spotted this in the system logs:

oct. 19 19:34:32 mob01772 kernel: [drm] stuck on render ring
oct. 19 19:34:32 mob01772 kernel: [drm] GPU HANG: ecode 9:0:0x85dffffb, in Xorg [12939], reason: Engine(s) hung, action: reset
oct. 19 19:34:32 mob01772 kernel: [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
oct. 19 19:34:32 mob01772 kernel: [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
oct. 19 19:34:32 mob01772 kernel: [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
oct. 19 19:34:32 mob01772 kernel: [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
oct. 19 19:34:32 mob01772 kernel: [drm] GPU crash dump saved to /sys/class/drm/card0/error
oct. 19 19:34:32 mob01772 kernel: drm/i915: Resetting chip after gpu hang
[...]
oct. 19 19:34:34 mob01772 kernel: [drm] RC6 on
oct. 19 19:34:44 mob01772 kernel: [drm] stuck on render ring
oct. 19 19:34:44 mob01772 kernel: [drm] GPU HANG: ecode 9:0:0x85dffffb, in Xorg [12939], reason: Engine(s) hung, action: reset
oct. 19 19:34:44 mob01772 kernel: drm/i915: Resetting chip after gpu hang

FWIW, this happened right after starting an I/O intensive operation, which might be related to the hang.

Comment 1 Milan Bouchet-Valat 2016-10-19 17:44:31 UTC

Created attachment 127408 [details]
journalctl -b output

Comment 2 Milan Bouchet-Valat 2016-10-19 17:45:14 UTC

Created attachment 127409 [details]
/sys/class/drm/card0/error

Comment 3 Matt Turner 2016-10-25 18:55:28 UTC

We don't have enough information.

- What Mesa version are you using? (glxinfo | grep Mesa)
- What were you doing at the time of the hang? Is it reproducible?

Comment 4 Milan Bouchet-Valat 2016-10-25 20:17:01 UTC

Unfortunately, as I said, it's not reproducible. I don't really remember what I was doing, apart from the high I/O or swap pressure.

$ glxinfo | grep Mesa
client glx vendor string: Mesa Project and SGI
    Device: Mesa DRI Intel(R) HD Graphics 520 (Skylake GT2)  (0x1916)
OpenGL renderer string: Mesa DRI Intel(R) HD Graphics 520 (Skylake GT2) 
OpenGL core profile version string: 4.3 (Core Profile) Mesa 12.0.3
OpenGL version string: 3.0 Mesa 12.0.3
OpenGL ES profile version string: OpenGL ES 3.1 Mesa 12.0.3

Comment 5 Matt Turner 2016-10-25 21:00:45 UTC

Okay, thanks. There's not really anything we can do if it's not reproducible. Let us know if you find a way to trigger it again.

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.