Bug 97123 - [AMDGPU][CIK] Suspend to disk deadlocks system on resume always (Linus master + drm-next-4.9-wip)
Summary: [AMDGPU][CIK] Suspend to disk deadlocks system on resume always (Linus master...
Status: RESOLVED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/AMDgpu (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-07-29 03:20 UTC by Shawn Starr
Modified: 2016-10-31 05:51 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
corruption seen during freeze() in hibernation (6.74 MB, image/jpeg)
2016-08-24 04:24 UTC, Shawn Starr
no flags Details

Description Shawn Starr 2016-07-29 03:20:47 UTC
I've been testing AMDGPU with CIK (Bonaire), in drm-next-4.8-wip / drm-next-4.8, hibernation sometime worked sometime not.

With drm-next-4.9-wip (with Linus master merged), hibernation works, resume fails after image is loaded, system is deadlocked.

Is there any debugging I can do testing path?
Comment 1 Shawn Starr 2016-08-24 04:24:12 UTC
Created attachment 125989 [details]
corruption seen during freeze() in hibernation
Comment 2 Shawn Starr 2016-08-24 04:25:52 UTC
With latest patches: 
[PATCH 3/3] drm/amdgpu: reset the asic when hibernating v2

I am not noticing deadlock on resume, still continue to test.

When hibernating the first time, no corruption occurs on freeze state of display, subsequent hibernations show same corruption pattern on freeze.

When resuming no such corruption shows once the GPU finishes init and we resume successfully.
Comment 3 Shawn Starr 2016-08-31 03:54:24 UTC
[PATCH 2/3] drm/amd/amdgpu: sdma resume fail during S4 on CI 
[PATCH 3/3] drm/amd/amdgpu: compute ring test fail during S4 on CI

With patch #2 specifically not upstream yet, the corruption with the image attached is now gone on subsequent freeze()/thaw() cycles.
Comment 4 Shawn Starr 2016-09-04 06:07:02 UTC
Still had a lockup during a freeze() --> Reboot cycle

Will try patch:
[PATCH v2] drm/amd/amdgpu: S4 issue for amdgpu
Comment 5 Shawn Starr 2016-09-09 18:44:24 UTC
[PATCH v2] drm/amd/amdgpu: S4 issue for amdgpu - Works

Still have corruption on freeze() but its non-fatal for thaw() on resume from reboot hibernation.

Will Continue to test additional patches, no lockups so far.
Comment 6 Shawn Starr 2016-10-31 05:51:40 UTC
The original issue is fixed, but a new regression so new bug being filed.


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.