Re: [PATCH RFC 1/7] regulator: devres: add APIs for reference supplies

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Thu, Mar 28, 2024 at 06:18:32PM +0000, Mark Brown wrote:
> On Thu, Mar 28, 2024 at 11:03:23AM -0700, Dmitry Torokhov wrote:
> 
> > So because we decided that we could not have devm_regulator_enable()
> > because of (IMO) contrived example of someone totally mixing up the devm
> > and non-devm APIs we now have to make more and more devm- variants
> > simply because we do not have access to the regulator structure with
> > devm_regulator_get_enable() and so all normal APIs are not available.
> 
> I don't follow what you're saying here?  What normal APIs are not
> available?  AFAICT this has nothing to do with a devm enable, it's a
> combined operation which reports the voltage for the regulator if one is
> available which would still be being added even if it used a devm
> enable.

You can not do devm_regulator_get_enable() and then call
regulator_get_voltage(), you need a new combined API.

> 
> > This is quite bad honestly. Mark, could we please reverse this
> > shortsighted decision and have normal devm_regulator_enable() operating
> > on a regulator?
> 
> Nothing has changed here.

Yes, unfortunately. We could have:

	reg = devm_regulator_get(...);
	...
	err = devm_regulator_enable(dev, reg);
	...
	err = regulator_get_voltage(reg);

and not multiply APIs, but we do not have devm_regulator_enable().

Thanks.

-- 
Dmitry