On Fri, Feb 2, 2024, at 11:07, Geert Uytterhoeven wrote:
> On Fri, Feb 2, 2024 at 10:51 AM Marek Szyprowski <m.szyprowski@xxxxxxxxxxx> wrote:
>> core, because the DRM core is set to be compiled-in in this defconfig.
>> This leaves all DRM display panels without integrated backlight control,
>> even if the needed modules have been properly loaded and probed.
>
> Hmm, that's bad.
>
> Is there any way to fix this in DRM?
> A quick grep shows that DRM is using the full monty of
> IS_{BUILTIN,ENABLED,MODULE,REACHABLE}(CONFIG_BACKLIGHT_CLASS_DEVICE).
> Probably not all of them are in perfect sync?
The IS_REACHABLE() ones are almost certainly bugs, as are the
'select BACKLIGHT_CLASS_DEVICE' ones we have in drivers/gpu.
> Several DRM drivers do select BACKLIGHT_CLASS_DEVICE, but if that
> does not work in the modular case, it should be fixed.
The select should do the right thing in principle, but mixing
it with depends is what causes circular dependencies. Unfortunately
trying to fix it likely also causes those, but I think it's worth
revisiting.
It should be possible to change it like this:
- change all DRM drivers that require the class to 'depends on
BACKLIGHT_CLASS_DEVICE'
- change all those drivers that can optionally use it to
'depends on BACKLIGHT_CLASS_DEVICE || !BACKLIGHT_CLASS_DEVICE'
to avoid the dependency on a loadable module
- Make BACKLIGHT_CLASS_DEVICE itself default to 'DRM' in order
to avoid regressions in defconfig files but still make it
possible to turn it off.
>> Fix this by selecting BACKLIGHT_CLASS_DEVICE to be compiled-in in
>> multi_v7_defconfig.
>>
>> Signed-off-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
>
> Sounds like a good interim solution.
>
> Acked-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
Thanks, I've applied it to the soc/defconfig branch now.
Arnd
