Bug 106828

Summary: [bsw] GPU hang on first user batch
Product: DRI Reporter: Torsten Fichtner <torsten>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED WORKSFORME QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: critical    
Priority: high CC: intel-gfx-bugs
Version: XOrg git   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard: Triaged
i915 platform: BSW/CHT i915 features: GPU hang
Attachments:
Description Flags
/sys/class/drm/card0/error none

Description Torsten Fichtner 2018-06-05 19:56:17 UTC
Created attachment 140035 [details]
/sys/class/drm/card0/error

GPU crash dump from /sys/class/drm/card0/error as requested

This happens on Debian Stretch with Linux Kernel 4.16 and libdrm from Backports Repository.
Comment 1 Chris Wilson 2018-06-05 20:05:37 UTC
Hmm, very suspicious that it's the first batch and claims an invalid PTE. Do you have a few recent kernels you can test?
Comment 2 Torsten Fichtner 2018-06-05 20:30:51 UTC
the stable debian kernel 4.9 works without any problems
Comment 3 Chris Wilson 2018-06-05 20:39:48 UTC
Could you please test the in-between kernel packages to give an approximate date for when this failed?
Comment 4 Chris Wilson 2018-06-05 20:41:07 UTC
Or you can skip to the end and do a bisection on the kernel git repository. :)
Comment 5 Francesco Balestrieri 2018-06-06 04:10:34 UTC
Does it happen at every boot?
Comment 6 Torsten Fichtner 2018-06-06 04:22:10 UTC
Yes it happens at every reboot.
Comment 7 Francesco Balestrieri 2018-06-06 04:39:21 UTC
Could you test this with the latest drm-tip and also attach dmesg with debug logs enabled?
Comment 8 Jani Saarinen 2018-06-25 10:08:45 UTC
Reported, have you tested using https://cgit.freedesktop.org/drm-tip and send dmesg with drm.debug=0x1e log_buf_len=4M?
Comment 9 Jani Saarinen 2018-06-26 06:26:11 UTC
(In reply to Jani Saarinen from comment #8)
> Reported, have you tested using https://cgit.freedesktop.org/drm-tip and
> send dmesg with drm.debug=0x1e log_buf_len=4M?

Meant to say reporter, have you tested using https://cgit.freedesktop.org/drm-tip and send dmesg with drm.debug=0x1e log_buf_len=4M?
Comment 10 Jani Saarinen 2018-08-13 09:43:41 UTC
No feedback, closing. Please re-open is still the issue and visible on latest
https://cgit.freedesktop.org/drm-tip and send dmesg with drm.debug=0x1e log_buf_len=4M?

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.