Bug 99696 - GPU HANG: ecode 9:0:0x87d6fffe, in X [836], reason: Hang on render ring, action: reset
Summary: GPU HANG: ecode 9:0:0x87d6fffe, in X [836], reason: Hang on render ring, acti...
Status: CLOSED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: XOrg git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-02-06 16:06 UTC by Shea Levy
Modified: 2017-07-24 23:15 UTC (History)
2 users (show)

See Also:
i915 platform: KBL
i915 features: GPU hang


Attachments
/sys/class/drm/card0/error (671.40 KB, text/plain)
2017-02-06 16:06 UTC, Shea Levy
no flags Details
dmesg output (59.27 KB, text/plain)
2017-02-06 20:58 UTC, Shea Levy
no flags Details
X.0.log (32.69 KB, text/plain)
2017-02-06 20:59 UTC, Shea Levy
no flags Details
/sys/class/drm/card0/error (28.78 KB, text/plain)
2017-02-06 22:10 UTC, Shea Levy
no flags Details
dmesg output (60.00 KB, text/plain)
2017-02-06 22:11 UTC, Shea Levy
no flags Details
X.0.log (32.48 KB, text/plain)
2017-02-06 22:13 UTC, Shea Levy
no flags Details

Description Shea Levy 2017-02-06 16:06:11 UTC
Created attachment 129366 [details]
/sys/class/drm/card0/error

System is Dell XPS 13 9360
Comment 1 Shea Levy 2017-02-06 16:07:18 UTC
Seems to happen most often when doing stuff in emacs
Comment 2 Chris Wilson 2017-02-06 17:00:23 UTC
The failure here is a little different to those I've seen so far. It would be nice to compare the error state from drm-tip (https://cgit.freedesktop.org/drm-tip) just for the same bits of extra info now captured, and could you also attach Xorg.log and dmesg from the failure (just to fill in the gaps as to what is going on).
Comment 3 Shea Levy 2017-02-06 20:58:53 UTC
Created attachment 129368 [details]
dmesg output

This is from a different boot than the first error report
Comment 4 Shea Levy 2017-02-06 20:59:27 UTC
Created attachment 129369 [details]
X.0.log

This is from a different boot than the initial error report, but the same as the dmesg output.
Comment 5 Shea Levy 2017-02-06 20:59:55 UTC
Rebuilding with drm-tip now, will report back if/when the bug gets triggered there.
Comment 6 Shea Levy 2017-02-06 21:28:57 UTC
I assume I want DRM_I915_CAPTURE_ERROR=Y. I've put DRM_I915_ALPHA_SUPPORT=N for now, let me know if that should be different.
Comment 7 Chris Wilson 2017-02-06 21:32:35 UTC
(In reply to Shea Levy from comment #6)
> I assume I want DRM_I915_CAPTURE_ERROR=Y. I've put DRM_I915_ALPHA_SUPPORT=N
> for now, let me know if that should be different.

Both are correct.
Comment 8 Shea Levy 2017-02-06 22:10:28 UTC
Created attachment 129371 [details]
/sys/class/drm/card0/error
Comment 9 Shea Levy 2017-02-06 22:11:03 UTC
Created attachment 129372 [details]
dmesg output
Comment 10 Shea Levy 2017-02-06 22:13:06 UTC
Created attachment 129373 [details]
X.0.log
Comment 11 Shea Levy 2017-02-06 22:13:31 UTC
Updated all attachments, from the same boot with drm-tip
Comment 12 Shea Levy 2017-02-06 22:19:52 UTC
Is there some workaround in the mean time? Fine with degraded graphics or whatever for now.
Comment 13 Chris Wilson 2017-02-06 22:36:32 UTC
Hmm, it does have the double 3DSTATE_VERTEX_ELEMENTS that I'm guessing is an issue from skl+, i.e. that's a patch in https://cgit.freedesktop.org/xorg/driver/xf86-video-intel/ to prevent that sequence. As you seem to be reliably hitting this, testing -intel would be invaluable feedback.

Alternatively, you can disable some acceleration using
/etc/X11/xorg.conf.d/intel.conf:
Section "Device"
  Identifier "igfx" 
  Driver "intel"
  Option "AccelMethod" "blt"
EndSection

or try modesetting, same file:
Section "Device"
  Identifier "igfx"
  Driver "modesetting"
EndSection
Comment 14 Shea Levy 2017-02-06 22:44:09 UTC
Should I test the tip of xf86-video-intel or just find that patch?
Comment 15 Shea Levy 2017-02-06 23:35:16 UTC
Currently running with xf86-video-intel at 	e4fe79cf0d9a05ee3f3a027148ef0aeb2b1b34e1. Will let you know if it has the issue or if it doesn't.
Comment 16 Shea Levy 2017-02-07 01:45:52 UTC
Normally I would have seen the issue by now, haven't so far. Still sometimes getting a brief lag, especially when switching between workspaces (in dwm)
Comment 17 Jari Tahvanainen 2017-04-11 11:19:04 UTC
Hello Shea. Based on your last comment, is the issue now fixed with the latest video-intel? If yes then please change status RESOLVED, if not then set it REOPENED.
Comment 18 Chris Wilson 2017-04-11 15:04:14 UTC
commit 4acd4a7d3d2f41227022fa7581cfb85a0b124eae
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Mon Dec 5 15:13:24 2016 +0000

    sna/gen9: Emit a dummy primitive between VertexElements
    
    References: https://bugs.freedesktop.org/show_bug.cgi?id=98999
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>


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.