Bug 107514

Summary: [CI][BAT] igt@drv_selftest@live_coherency - dmesg-fail - Value[9/31] mismatch, (overwrite with gtt) wrote [gpu] fcf5ab91 read [cpu] 30a546e (inverse 30a546e), at offset cc
Product: DRI Reporter: Martin Peres <martin.peres>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED FIXED QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: normal    
Priority: medium CC: intel-gfx-bugs
Version: XOrg git   
Hardware: Other   
OS: All   
Whiteboard: ReadyForDev
i915 platform: i915 features:

Description Martin Peres 2018-08-07 12:27:25 UTC 
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_4535/fi-byt-j1900/igt@drv_selftest@live_coherency.html

[  655.777150] Value[9/31] mismatch, (overwrite with gtt) wrote [gpu] fcf5ab91 read [cpu] 30a546e (inverse 30a546e), at offset cc
[  655.777427] i915/i915_gem_coherency_live_selftests: igt_gem_coherency failed with error -22
[  655.882280] i915: probe of 0000:00:02.0 failed with error -22
Comment 1 Martin Peres 2018-08-07 12:28:06 UTC 
This may already be a duplicate of https://bugs.freedesktop.org/show_bug.cgi?id=107435.
Comment 2 Chris Wilson 2018-08-07 12:33:55 UTC 
commit f6844a85e0c96a55c61fa3e611f414999b11e4de
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Mon Jul 30 08:53:51 2018 +0100

    drm/i915/selftests: Replace opencoded clflush with drm_clflush_virt_range
    
    We occasionally see that the clflush prior to a read of GPU data is
    returning stale data, reminiscent of much earlier bugs fixed by adding a
    second clflush for serialisation. As drm_clflush_virt_range() already
    supplies the workaround, use it rather than open code the clflush
    instruction.
    
    References: 396f5d62d1a5 ("drm: Restore double clflush on the last partial cacheline")
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
    Reviewed-by: Matthew Auld <matthew.auld@intel.com>
    Link: https://patchwork.freedesktop.org/patch/msgid/20180730075351.15569-3-chris@chris-wilson.co.uk
Comment 3 James Ausmus 2018-08-21 00:23:37 UTC 
Test is now all green on fi-byt-j1900, closing
Comment 4 Martin Peres 2018-09-07 15:45:14 UTC 
(In reply to James Ausmus from comment #3)
> Test is now all green on fi-byt-j1900, closing

You are assuming that failures are consistent ;) In this particular case, the reproduction rate is lower than .5%... or it has been fixed immediately... hard to tell since Chris has been so faaaastt!

Anyway, I will trust Chris and archive the bug.
Comment 5 Chris Wilson 2018-09-07 15:57:39 UTC 
It's down to an irregular fail with gdg (107164). I feel like I've exhausted all the likely suspects... Perhaps my delay is not quite enough.

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.