I've had this across several applications and several kernel/mesa/llvm versions and I'm not sure I get everything here right, but here it goes. Most easily I can reproduce the GPU hang on SOMA when starting a new game - it will hang while the intro video plays or shortly after, in the game - other and very demanding games work without hangs. It would hang like this: [Di Nov 29 19:46:21 2016] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx timeout, last signaled seq=3374996, last emitted seq=3374998 [Di Nov 29 19:46:21 2016] [drm] IP block:5 is hang! Just now is the first time it actually recovered, usually I need to hard reset the PC. Now this GPU is factory overclocked a little bit: /sys/class/drm/card0/device/pp_dpm_sclk 0: 300Mhz 1: 608Mhz 2: 910Mhz 3: 1077Mhz 4: 1145Mhz 5: 1191Mhz 6: 1236Mhz 7: 1288Mhz * I am reasonably confident that when I set a power level with this command: echo manual > /sys/class/drm/card0/device/power_dpm_force_performance_level echo 7 > /sys/class/drm/card0/device/pp_dpm_sclk the GPU will not hang while the power level is fixed. Therefore I think that the GPU hang is related to switching between power levels. It's possible that this problem is specific to this factory overclocked model.
As far as I can tell after a couple of days this is "fixed" by RMA'ing the GPU and getting a new one (same model). The entire issue was probably caused by bad hardware.
*** Bug 98162 has been marked as a duplicate of this bug. ***
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.