Re: [PATCH RFC RFT] drm/msm: adreno: attach the GMU device to a driver

[Neil Armstrong <neil.armstrong@linaro.org>]

On 10/24/25 10:55, Konrad Dybcio wrote:
> On 10/23/25 10:27 AM, Neil Armstrong wrote:
>> On 10/22/25 19:09, Konrad Dybcio wrote:
>>> On 10/22/25 2:44 PM, Neil Armstrong wrote:
>>>> Due to the sync_state is enabled by default in pmdomain & CCF since v6.17,
>>>> the GCC and GPUCC sync_state would stay pending, leaving the resources in
>>>> full performance:
>>>> gcc-x1e80100 100000.clock-controller: sync_state() pending due to 3d6a000.gmu
>>>> gpucc-x1e80100 3d90000.clock-controller: sync_state() pending due to 3d6a000.gmu
>>>
>>> Does this *actually* cause any harm, by the way?
>>
>> ?
>>
>>>
>>> For example on x1e, GMU refers to 2 GPU_CC GDSCs, GPU_CC refers
>>> to a pair of GCC clocks and GCC refers to VDD_CX
>>>
>>> and I see these prints, yet:
>>>
>>> /sys/kernel/debug/pm_genpd/gpu_cx_gdsc/current_state:off-0
>>> /sys/kernel/debug/pm_genpd/gpu_gx_gdsc/current_state:off-0
>>>
>>> /sys/kernel/debug/pm_genpd/cx/current_state:on
>>> /sys/kernel/debug/pm_genpd/cx/perf_state:256 # because of USB3 votes
>>>
>>> I'm not super sure where that sync_state comes from either (maybe
>>> dev_set_drv_sync_state in pmdomain/core?)
>>
>> The way we handle the GMU so far is wrong, it abuses the driver model.
>>
>> And this is a symptom, whatever the impact it has, it needs to be fixed
>> in a proper way.
>>
>> The sync_state is retained because the gmu device is never probed but
>> has some clocks and power domains attached to it, which is clearly wrong.
> 
> Yes I agree, however I'm only debating the commit message claims of
> 'leaving the resources in full performance', which doesn't seem to be
> true

OK so the wording may be a little bit extreme, perhaps something like:
the GCC and GPUCC sync_state would stay pending, leaving the unused
power domains enabled for the lifetime of the system.

Neil

> 
> Konrad