Re: [PATCH v3 1/3] leds: ktd2692: move ExpressWire code to library

[Daniel Thompson <daniel.thompson@linaro.org>]

On Mon, Jan 22, 2024 at 05:24:51PM +0100, Duje Mihanović wrote:
> On Monday, January 22, 2024 11:19:26 AM CET Daniel Thompson wrote:
> > On Sat, Jan 20, 2024 at 10:26:43PM +0100, Duje Mihanović wrote:
> > > diff --git a/drivers/leds/Kconfig b/drivers/leds/Kconfig
> > > index 6292fddcc55c..d29b6823e7d1 100644
> > > --- a/drivers/leds/Kconfig
> > > +++ b/drivers/leds/Kconfig
> > > @@ -181,6 +181,9 @@ config LEDS_EL15203000
> > >
> > >  	  To compile this driver as a module, choose M here: the module
> > >  	  will be called leds-el15203000.
> > >
> > > +config LEDS_EXPRESSWIRE
> > > +	bool
> > > +
> >
> > Shouldn't there be a "select GPIOLIB" here? It seems odd to make the
> > clients responsible for the dependencies.
> >
> > BTW there seems to be very little consistency across the kernel between
> > "depends on GPIOLIB" and "select GPIOLIB".. but select is marginally
> > more popular (283 vs. 219 in the kernel I checked).
>
> I believe a "select" would be more appropriate here unless these backlights
> should be hidden if GPIOLIB is disabled. The catch with "select" is that there
> seems to be no way to throw in the "|| COMPILE_TEST" other GPIO-based
> backlights have and I'm not sure what to do about that.

I think the "|| COMPILE_TEST" might just be a copy 'n paste'ism (in fact I
may even have been guilty off propagating it in reviews when checking
for inconsistencies).

AFAICT nothing will inhibit setting GPIOLIB so allyes- and allmodconfig
builds will always end up with GPIOLIB enabled. If we are happy to
select it then I think that is enough!


Daniel.