Bug 98924 - [AMDGPU][SI] R9 m265x doesn't work with experimental kernel driver
Summary: [AMDGPU][SI] R9 m265x doesn't work with experimental kernel driver
Status: RESOLVED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/AMDgpu (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-12-01 10:17 UTC by Stefano Cipriani
Modified: 2017-07-17 14:51 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg (68.47 KB, text/plain)
2016-12-01 10:17 UTC, Stefano Cipriani
no flags Details
Xorg log (56.82 KB, text/plain)
2016-12-01 10:17 UTC, Stefano Cipriani
no flags Details
dmesg when glxgears runs once (82.09 KB, text/plain)
2017-02-09 10:25 UTC, Stefano Cipriani
no flags Details

Description Stefano Cipriani 2016-12-01 10:17:23 UTC
Created attachment 128297 [details]
dmesg

My laptop has an Intel CPU/iGPU + AMD dGPU
I compiled both the current master of 4.9 and drm-next-4.10-wip, I have the same problem, when I try to use the m265x with "DRI_PRIME=1 glxgears" everything freezes for about 5~10 seconds, then glxgears appears but the gears are either frozen or I just get a black window. 
Then when I try to shutdown it never completely turn off and I have to force the power off
I tried both with and without amdgpu.runpm=0 but I get the same behaviour.
Comment 1 Stefano Cipriani 2016-12-01 10:17:59 UTC
Created attachment 128298 [details]
Xorg log
Comment 2 Stefano Cipriani 2017-02-09 10:25:45 UTC
Created attachment 129431 [details]
dmesg when glxgears runs once

Small update, on drm-next-4.11-wip sometimes glxgears starts correctly the first time, but it freezes on subsequent runs. 
On dmesg I get:

asynchronous wait on fence amd_sched:gfx:19 timed out

Full dmesg attached
Comment 3 Jeroen Bollen 2017-03-28 00:07:26 UTC
I can confirm this bug also.

Hardware:
03:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Opal XT [Radeon R7 M265] (rev 81)
	Subsystem: Hewlett-Packard Company Device 814f
	Kernel driver in use: amdgpu
	Kernel modules: radeon, amdgpu

My dmesg error is mildly different though:
[  156.940353] amdgpu 0000:03:00.0: PCIE GART of 1024M enabled (table at 0x0000000000040000).
[  157.159282] [drm] ring test on 0 succeeded in 9 usecs
[  157.161319] [drm] ring test on 1 succeeded in 1 usecs
[  157.161331] [drm] ring test on 2 succeeded in 1 usecs
[  157.161346] [drm] ring test on 3 succeeded in 3 usecs
[  157.161355] [drm] ring test on 4 succeeded in 4 usecs
[  168.384040] asynchronous wait on fence amd_sched:gfx:13 timed out

Also tested with glxgears. Doing so crashes the application into a state it cannot be killed in, and often also takes down the rest of the system by freezing the entire gui.
Comment 4 Stefano Cipriani 2017-07-17 14:51:57 UTC
The GPU now works with AMDGPU driver on latest linux master. Playing some Rocket League I had a couple of crashes, but the game was just updated so I don't know if it was the kernel. Anyways, marking as fixed as it works now.


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.