the graphics/2_valid_opengl_renderer_AMD-GPU test failed

Bug #1940492 reported by StanleyHuang
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Checkbox Provider - Base
Expired
High
Unassigned

Bug Description

[Summary]
This test case would try to validate OpenGL renderer is AMD GPU with DRI_RPIME=1 argument.
But in desktop platform, we would connect a monitor to AMD GPU before power on the system.
At this moment the primary GPU provider is AMD GPU, that means the default renderer is AMD GPU, and the OpenGL renderer is Intel GPU with DRI_PRIME=1.

--------------[ Running job 2 / 2. Estimated time left: unknown ]---------------
------------[ Check the OpenGL renderer (AMD GPU and DRI_PRIME=1) ]-------------
ID: com.canonical.certification::graphics/2_valid_opengl_renderer_Navi_10__Radeon_Pro_W5700_
Category: com.canonical.plainbox::graphics
... 8< -------------------------------------------------------------------------
OpenGL renderer string: Mesa Intel(R) Graphics (RKL GT1)
ERROR: renderer is Intel when DRI_PRIME=1

[Steps to reproduce]
1. Connect a monitor to AMD GPU.
2. Power on system and boot into OS.
3. Install checkbox and provider
4. Run valid_opengl_renderer with below command.
Cmd> checkbox-cli run com.canonical.certification::graphics_card com.canonical.certification::graphics/2_valid_opengl_renderer_Navi_10__Radeon_Pro_W5700_

[Expected result]
The valid_opengl_renderer test passed

[Actual result]
The valid_opengl_renderer test failed

[Failure rate]
10/10

[Additional information]
CID: 202010-28295
SKU: RDT-RKL-DVT2.1-C1
Image: canonical-oem-somerville-focal-amd64-20200502-85+fossa-proposed-release+X124
bios-version: 1.2.0
GPU: 00:02.0 VGA compatible controller [0300]: Intel Corporation Device [8086:4c8a] (rev 04)
03:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Device [1002:7312]
kernel-version: 5.10.0-1043-oem

[Stage]
Issue reported and logs collected right after it happened

Revision history for this message
StanleyHuang (stanley31) wrote :

Automatically attached

Changed in plainbox-provider-checkbox:
importance: Undecided → High
Revision history for this message
StanleyHuang (stanley31) wrote :

Automatically attached

Revision history for this message
StanleyHuang (stanley31) wrote :

Automatically attached

Revision history for this message
StanleyHuang (stanley31) wrote :

Automatically attached

information type: Public → Private
Revision history for this message
StanleyHuang (stanley31) wrote :

It would be better we could check which GPU is primary.

StanleyHuang (stanley31)
information type: Private → Private Security
information type: Private Security → Public
Revision history for this message
StanleyHuang (stanley31) wrote :

I have collected the response via glxinfo and xrandr on a Desktop (has both iGPU and dGPU) in different scenario.

# Connect a monitor only to iGPU port.
1. Connect a monitor to onboard DP port.
2. Power on the system and boot into OS.
3. Open a terminal and launch following commands.
- glxinfo -B > dp_in_intel/glxinfo.log
- DRI_PRIME=1 glxinfo -B > dp_in_intel/glxinfo_with_dri_prime1.log
- xrandr --listproviders > dp_in_intel/xrandr_listproviders.log

# Connect a monitor only to AMD dGPU port.
1. Connect a monitor to AMD DP port.
2. Power on the system and boot into OS.
3. Open a terminal and launch following commands.
- glxinfo -B > dp_in_amd/glxinfo.log
- DRI_PRIME=1 glxinfo -B > dp_in_amd/glxinfo_with_dri_prime1.log
- xrandr --listproviders > dp_in_amd/xrandr_listproviders.log

# Connect two monitor to Both Intel iGPU and AMD dGPU port.
1. Connect a monitor to AMD DP port.
2. Connect a monitor to onboard DP port.
3. Power on the system and boot into OS.
4. Open a terminal and launch following commands.
- glxinfo -B > dp_in_both/glxinfo.log
- DRI_PRIME=1 glxinfo -B > dp_in_both/glxinfo_with_dri_prime1.log
- xrandr --listproviders > dp_in_both/xrandr_listproviders.log

Matias Piipari (mz2)
tags: added: cbox-13
Revision history for this message
Maksim Beliaev (beliaev-maksim) wrote :

Bug was migrated to GitHub: https://github.com/canonical/checkbox/issues/85.
Bug is no more monitored here.

Changed in plainbox-provider-checkbox:
status: New → Expired
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.