[Bug 108854] [polaris11] - Failed GPU reset after hang

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=108854

            Bug ID: 108854
           Summary: [polaris11] - Failed GPU reset after hang
           Product: DRI
           Version: DRI git
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: normal
          Priority: medium
         Component: DRM/AMDgpu
          Assignee: dri-devel at lists.freedesktop.org
          Reporter: tseewald at gmail.com

Created attachment 142604
  --> https://bugs.freedesktop.org/attachment.cgi?id=142604&action=edit
dmesg showing the hang and failed gpu reset

Problem:

While running RuneLite [1] with GPU acceleration enabled, the system hangs
after  several minutes of seemingly normal operation. Once the GPU hangs, it
attempts to reset itself but fails with the following message:

[drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* Failed to initialize parser -125!

This hang causes the system to lock up and ssh is the only access possible.
There is no graphical corruption, the displays are simply frozen.

System Information:

GPU: POLARIS11 - RX 560 4GB (1002:67ff)
Mesa: 18.0.5
X11: 1.19.6
Firmware files should be the latest as I've pulled them from adg5f's repo [2].

Kernel parameters: "quiet splash scsi_mod.use_blk_mq=1 apparmor=2
security=apparmor amdgpu.gpu_recovery=1 spectre_v2=off"

I have reproduced this issue on:

4.20-rc3
amd-staging-drm-next (as of commit 1179994039abc10aab0d2f0ecfc4c65dfbd77438)

[1] https://github.com/runelite/runelite
[2] https://people.freedesktop.org/~agd5f/radeon_ucode/

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20181124/52510bfc/attachment.html>