Bug 78000 - [pnv] Invalid overlay PTE
Summary: [pnv] Invalid overlay PTE
Status: CLOSED WONTFIX
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: low normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-04-27 13:48 UTC by Ingo Saitz
Modified: 2017-07-24 22:54 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:

Attachments
GPU crash dump saved to /sys/class/drm/card0/error (675.15 KB, text/plain)
2014-04-27 13:48 UTC, Ingo Saitz 		no flags Details
Xorg.0.log (30.47 KB, text/plain)
2014-04-27 13:49 UTC, Ingo Saitz 		no flags Details
lspci output (23.42 KB, text/plain)
2014-04-27 13:50 UTC, Ingo Saitz 		no flags Details
dmesg output from boot (33.85 KB, text/plain)
2014-04-27 13:51 UTC, Ingo Saitz 		no flags Details
xorg.conf configuration file (251 bytes, text/plain)
2014-04-27 13:52 UTC, Ingo Saitz 		no flags Details
View All

Description Ingo Saitz 2014-04-27 13:48:51 UTC 
Created attachment 98072 [details]
GPU crash dump saved to /sys/class/drm/card0/error

[drm] GPU crash dump saved to /sys/class/drm/card0/error
[drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
i915: render error detected, EIR: 0x00000010
i915: page table error
i915:   PGTBL_ER: 0x00010000
[drm:i915_report_and_clear_eir] *ERROR* EIR stuck: 0x00000010, masking
i915: render error detected, EIR: 0x00000010
i915: page table error
i915:   PGTBL_ER: 0x00010000

After this bug GLX applications (e.g. glxgears) threw core dumps on startup, after displaying an empty black window.

After applying the fence counting patch (attachment 75820 [details] [review] from https://bugs.freedesktop.org/show_bug.cgi?id=59771 ), dri applications continue to work, and the Xorg server process outputs a fence error almost on startup ("Fixing up fence counts; was -1, expected 0" - the patch only displays this error once).

I don't know how to trigger the error, other than using 3d-accelerated programs for some time.

Package versions (current debian unstable):
mesa 10.1.0-5
libdrm 2.4.52-1 + fence counting patch applied
xorg-xserver 2:1.15.1-1

linux-kernel 3.14.1 (own config)
CONFIG_DRM_I915=m
CONFIG_DRM_I915_KMS=y
CONFIG_DRM_I915_FBDEV=y
# CONFIG_DRM_I915_PRELIMINARY_HW_SUPPORT is not set
# CONFIG_DRM_I915_UMS is not set
Comment 1 Ingo Saitz 2014-04-27 13:49:56 UTC 
Created attachment 98073 [details]
Xorg.0.log
Comment 2 Ingo Saitz 2014-04-27 13:50:30 UTC 
Created attachment 98074 [details]
lspci output
Comment 3 Ingo Saitz 2014-04-27 13:51:48 UTC 
Created attachment 98075 [details]
dmesg output from boot
Comment 4 Ingo Saitz 2014-04-27 13:52:13 UTC 
Created attachment 98076 [details]
xorg.conf configuration file
Comment 5 Chris Wilson 2014-04-27 15:35:03 UTC 
The overlay itself is not even active, nor changing. Presumably something in the update of neighbouring PTE upset the hw.
Comment 6 Rodrigo Vivi 2014-10-07 20:49:39 UTC 
Could you please test with latest drm-intel-nightly to see if the issue is still present and paste new logs?
Comment 7 Rodrigo Vivi 2015-01-14 23:56:20 UTC 
timeout. Feel free to reopen if you are still able to reproduce on recent kernels updating new logs.

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.