Re: [PATCH v3 2/3] ACPI: video: Remove CONFIG_X86 handling from nvidia_wmi_ec_supported()

[Hans de Goede]

Hi,

On 16-Sep-26 09:16, Jani Nikula wrote:
> On Wed, 16 Sep 2026, Armin Wolf <W_Armin@xxxxxx> wrote:
>> Am 15.09.26 um 14:31 schrieb Jani Nikula:
>>
>>> On Tue, 15 Sep 2026, Ilpo Järvinen <ilpo.jarvinen@xxxxxxxxxxxxxxx> wrote:
>>>> GPU people heads up!!
>>>>
>>>> On Wed, 2 Sep 2026, Armin Wolf wrote:
>>>>
>>>>> Am 01.09.26 um 23:30 schrieb Deucher, Alexander:
>>>>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig
>>>>>>> b/drivers/gpu/drm/amd/amdgpu/Kconfig
>>>>>>> index 12e4a41bf1f0..50305630f533 100644
>>>>>>> --- a/drivers/gpu/drm/amd/amdgpu/Kconfig
>>>>>>> +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig
>>>>>>> @@ -30,9 +30,7 @@ config DRM_AMDGPU
>>>>>>>         # ACPI_VIDEO's dependencies must also be selected.
>>>>>>>         select INPUT if ACPI
>>>>>>>         select ACPI_VIDEO if ACPI
>>>>>>> -     # On x86 ACPI_VIDEO also needs ACPI_WMI
>>>>>>> -     select X86_PLATFORM_DEVICES if ACPI && X86
>>>>>> The change below makes sense, but this one looks like it should remain
>>>>>> unless X86_PLATFORM_DEVICES is handled some other way?
>>>>>>
>>>>>> Alex
>>>>> AFAIK the dependency on X86_PLATFORM_DEVICES was only necessary because
>>>>> ACPI-WMI depended on it in the past.
>>>>> By now ACPI-WMI neither depends on X86_PLATFORM_DEVICES or X86 anymore.
>>>> Hi all,
>>>>
>>>> Could more GPU people besides Alex take a look on this. I'd want to make
>>>> progress with this series within this cycle but would prefer to have Acks
>>>> from GPU side.
>>> I share the same concern with Alex. It seems to me we could now end up
>>> with x86 configurations without X86_PLATFORM_DEVICES, which handle
>>> e.g. backlight on a lot of platforms.
>>>
>>> I think the safer option might be to go for:
>>>
>>> 	depends on X86_PLATFORM_DEVICES || !X86
>>>
>>> in all of the cases that select/depend on X86_PLATFORM_DEVICES in drm.
>>>
>>> The alternative is to find out, and handle the fallout, if any.
>>>
>>>
>>> BR,
>>> Jani.
>>
>> Alright, should i leave the original "select X86_PLATFORM_DEVICES if ..." statements
>> as-is or should is change them to "depends on X86_PLATFORM_DEVICES || !X86" everywhere?
> 
> There are some places that depends on X86_PLATFORM_DEVICES. I think it's
> almost always a mistake to have some places select and some places
> depends on, and select should anyway be used with care. I'd go for
> depends on.

I'm a bit late to the party here, sorry.

Note that X86_PLATFORM_DEVICES is an option which enables / disables
a menu and does nothing more.

Enabling it or depending on it by itself does nothing.

The select used to be there since it must be selected to be able
to select ACPI_WMI (which is no longer the case after this series).

To depends on in XE was probably copy pasted from the existing
select-s and then changed to a depends and already was a completely
no-op since depending on ACPI_WMI was enough as that itself
dependend on ACPI_WMI.

As for X86_PLATFORM_DEVICES being needed for some backlight
interfaces, depending on it by itself does not result in those
interfaces being there. That requires enabling the actual drivers
providing those interfaces.

And I don't think that the GPU drivers should start depending
on those since that will become very unyielding.

TL;DR: I believe Armin's original patch is fine and this is
a welcome cleanup / simplification of the DRM drivers Kconfig
bits.

Regards,

Hans