Re: [RESEND PATCH v3 4/4] leds: Add a multicolor LED driver to group monochromatic LEDs

[Andy Shevchenko]

On Fri, Oct 7, 2022 at 3:03 PM Jean-Jacques Hiblot
<jjhiblot@xxxxxxxxxxxxxxx> wrote:
> On 07/10/2022 10:53, Andy Shevchenko wrote:
> > On Fri, Oct 7, 2022 at 9:34 AM Jean-Jacques Hiblot
> > <jjhiblot@xxxxxxxxxxxxxxx> wrote:
> >> On 17/09/2022 10:37, Andy Shevchenko wrote:
> >>> On Sat, Sep 17, 2022 at 11:14 AM Jean-Jacques Hiblot
> >>> <jjhiblot@xxxxxxxxxxxxxxx> wrote:

...

> >>>> +               led_cdev = devm_of_led_get(dev, count);
> >>> Why _of_ variant? Please, make this OF independent since it's
> >>> pretending to cover not only OF-based systems.
> >> This is not OF independent. It could be, but that will wait until
> >> someone needs it. I don't know much about ACPI and have no hardware to
> >> test it on.
> >>
> >> I'll add the missing  dependency on OF in the Kconfig.
> > No, please consider getting rid of OF-centric API usage.
>
> The trouble is that the OF-agnostic API for leds doesn't exist yet and I
> don't really want to add it without any way to test it.

Yeah, that might be a problem due to unestablished descriptions
outside DT. Anyway, it seems harmless to call that function when there
is no OF dependency. In such cases it will fail with a deferred probe.

-- 
With Best Regards,
Andy Shevchenko