Re: [RFC PATCH 2/5] mfd: add 88pm88x driver

[Lee Jones <lee@kernel.org>]

Linus,

On Fri, 02 Feb 2024, Karel Balej wrote:

> Lee Jones, 2024-02-02T12:45:50+00:00:
> > On Thu, 01 Feb 2024, Karel Balej wrote:
> >
> > > Lee Jones, 2024-01-31T11:03:11+00:00:
> > > > On Sun, 28 Jan 2024, Karel Balej wrote:
> > > > > > > +	/* GPIO1: DVC, GPIO0: input */
> > > > > > > +	REG_SEQ0(PM88X_REG_GPIO_CTRL1, 0x40),

Do we have a precedent for drivers setting up their own pins like this?

> > > > > > Shouldn't you set these up using Pintrl?
> > > > > 
> > > > > You mean to add a new MFD cell for the pins and write the respective
> > > > > driver? The downstream implementation has no such thing so I'm not sure
> > > > > if I would be able to do that from scratch.
> > > >
> > > > This is not a Pinctrl driver.
> > > >
> > > > Isn't there a generic API you can use?
> > > 
> > > I'm sorry, I don't think I understand what you mean.
> >
> > Perhaps I misunderstand the code.  It looks like this regmap patch hack
> > is configuring pins and a bunch of other things.  Would that be a
> > correct assessment?
> 
> Yes, that sounds correct.
> 
> > If so, where do we draw the line here?  Do we accept a 1000 line driver
> > which configures a large SoC with a bunch of bespoke register writes?
> 
> I understand, I just don't know what you mean by "a generic API". I'm
> also not clear on whether what you have in mind is simply adding a
> dedicated driver for the pins as a new subdevice of this MFD.

-- 
Lee Jones [李琼斯]