https://bugs.freedesktop.org/show_bug.cgi?id=92993

Bug ID: 92993 Summary: Powerplay: Thermal controller fails to start on R9 Nano (Fiji) Product: DRI Version: DRI git Hardware: Other OS: All Status: NEW Severity: normal Priority: medium Component: DRM/AMDgpu Assignee: dri-devel@lists.freedesktop.org Reporter: master.homer@gmail.com

Just got a Sapphire R9 Nano and noticed some problems while running with the latest powerplay kernel from agd5f (commit 256ca780500bc445752763f32b246dc9ee396b62). I don't know how well implemented this is just yet, but it seems like the thermal controller fails to start on my card due to hwmgr->start_thermal_controller->function_list being NULL: [ 2.057417] powerplay: phm_start_thermal_controller [ 2.057419] [ powerplay ] phm_dispatch_table Invalid Parameter! ffff880425cd2000 ffff880425cd2230 (null)

The first address refers to hwmgr, the second to rt_table and the last to rt_table->function_list. This, in turn, causes the hwmon interface to not work, and sensors gives me a 0.0 temperature. I'm unsure if this causes the fan to not speed up, but I haven't noticed any sound difference when running valley for example.

I've noticed another (possibly unrelated) thing too, checking amdgpu_pm_info I can see that the mclk never changes, just the sclk. I did notice that it fails to find the virtual voltage 65281 in the vdd list, but of course I'm not sure this is related: [ 2.006036] powerplay: us_vdd 900 vs virtual_voltage_id 65281 [ 2.006037] powerplay: us_vdd 65282 vs virtual_voltage_id 65281 [ 2.006038] powerplay: us_vdd 65283 vs virtual_voltage_id 65281 [ 2.006038] powerplay: us_vdd 65284 vs virtual_voltage_id 65281 [ 2.006039] powerplay: us_vdd 65285 vs virtual_voltage_id 65281 [ 2.006040] powerplay: us_vdd 65286 vs virtual_voltage_id 65281 [ 2.006040] powerplay: us_vdd 65287 vs virtual_voltage_id 65281 [ 2.006041] powerplay: us_vdd 65288 vs virtual_voltage_id 65281 [ 2.006042] Can't find requested voltage id in vdd_dep_on_sclk table!