Bug 97248

Summary: [regression] [amdgpu] New regression in 4.8-rc1
Product: DRI Reporter: Mike Lothian <mike>
Component: DRM/AMDgpuAssignee: Default DRI bug account <dri-devel>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: medium CC: mike
Version: DRI git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
dmesg
none
acpi switcherooo
none
acpi off
none
amdgpu switcheroo none

Description Mike Lothian 2016-08-08 19:17:30 UTC
Created attachment 125605 [details]
dmesg

I'm seeing a new regression since 4.8-rc1 on Linus's tree, drm-next, agd5f's 4.9-next-wip since 4.8-rc1 was merged. I'm about to start bisecting

[   12.644973] amdgpu 0000:01:00.0: Refused to change power state, currently in D3
[   12.720886] amdgpu 0000:01:00.0: Refused to change power state, currently in D3
[   12.733891] amdgpu 0000:01:00.0: Refused to change power state, currently in D3
[   12.958093] amdgpu 0000:01:00.0: Wait for MC idle timedout !
[   13.071062] amdgpu 0000:01:00.0: Wait for MC idle timedout !
[   13.081607] [ powerplay ] Invalid VramInfo table.Failed to initialize MC reg table!
[   13.770954] [drm:gfx_v8_0_ring_test_ring] *ERROR* amdgpu: ring 0 test failed (scratch(0xC040)=0xFFFFFFFF)
[   13.772803] [drm:amdgpu_resume] *ERROR* resume of IP block <gfx_v8_0> failed -22
[   13.774713] [drm:amdgpu_resume_kms] *ERROR* amdgpu_resume failed (-22).
[   13.776643] amdgpu 0000:01:00.0: couldn't schedule ib
[   13.778602] [drm:amdgpu_job_run] *ERROR* Error scheduling IBs (-22)
[   13.805924] amdgpu 0000:01:00.0: couldn't schedule ib
[   13.805928] [drm:amdgpu_job_run] *ERROR* Error scheduling IBs (-22)
[   18.934066] [drm:amdgpu_suspend] *ERROR* suspend of IP block <vce_v3_0> failed -110
[   18.943133] [drm:amdgpu_suspend] *ERROR* set_clockgating_state(ungate) of IP block <uvd_v5_0> failed -16

These are probably the most relevant messages
Comment 1 Alex Deucher 2016-08-08 19:24:20 UTC
does reverting c63695cc5e5f685e924e25a8f9555f6e846f1fc6 help?

drm/amdgpu: work around lack of upstream ACPI support for D3cold

Until Dave's patch to support the new hybrid gfx ACPI method goes
upstream, we can fallback to the old ATPX method which seems to
still work.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

The d3cold patches went upstream in 4.8, so we should drop the workarounds.
Comment 2 Mike Lothian 2016-08-08 19:29:50 UTC
I'll revert that on the 4.9-next-wip branch and see if it makes a difference

I'm bisecting on linus's tree at the moment to see where the regression crept in

I've also been runnig with these local patches to use Dave's D3cold work (I reverted those before reporting) I'll attach them to the bug though
Comment 3 Mike Lothian 2016-08-08 19:30:16 UTC
Created attachment 125606 [details] [review]
acpi switcherooo
Comment 4 Mike Lothian 2016-08-08 19:30:32 UTC
Created attachment 125607 [details] [review]
acpi off
Comment 5 Mike Lothian 2016-08-08 19:31:01 UTC
Created attachment 125608 [details] [review]
amdgpu switcheroo
Comment 6 Mike Lothian 2016-08-08 19:33:44 UTC
Yes reverting that patch seems to work
Comment 7 Alex Deucher 2016-08-08 19:35:11 UTC
(In reply to Mike Lothian from comment #2)
> I'll revert that on the 4.9-next-wip branch and see if it makes a difference
> 
> I'm bisecting on linus's tree at the moment to see where the regression
> crept in

It's probably due to the d3cold support that when upstream in 4.8.

> 
> I've also been runnig with these local patches to use Dave's D3cold work (I
> reverted those before reporting) I'll attach them to the bug though

You shouldn't need those (or the workaround from my tree referenced in comment 1) now that the d3cold support has gone upstream via the pci tree.

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.