Re: [PATCH 34/36] treewide: rename pinctrl_gpio_direction_input_new()

From: Bartosz Golaszewski
Date: Wed Oct 04 2023 - 10:46:30 EST


On Wed, Oct 4, 2023 at 2:53 PM Andy Shevchenko
<andy.shevchenko@xxxxxxxxx> wrote:
>
> On Wed, Oct 04, 2023 at 03:49:52PM +0300, Andy Shevchenko wrote:
> > On Tue, Oct 03, 2023 at 09:08:41PM +0200, Bartosz Golaszewski wrote:
> > > On Tue, Oct 3, 2023 at 5:16 PM Andy Shevchenko
> > > <andy.shevchenko@xxxxxxxxx> wrote:
> > > > On Tue, Oct 3, 2023 at 5:51 PM Bartosz Golaszewski <brgl@xxxxxxxx> wrote:
> > > > >
> > > > > From: Bartosz Golaszewski <bartosz.golaszewski@xxxxxxxxxx>
> > > > >
> > > > > Now that pinctrl_gpio_direction_input() is no longer used, let's drop the
> > > > > '_new' suffix from its improved variant.
> > > >
> > > > This and other "treewide" patches in the series are redundant. Just
> > > > name the functions better to begin with.
> > >
> > > I don't want to rename these functions. They have perfectly fine
> > > names. I want to change their signatures and the renaming part is
> > > there only to make the reviewing easier.
> >
> > So, then you can make them macros and based on the type of the first argument
> > use either old or new API, which you then make public. The all noise with
> > "_new" is not needed, really.
>
> To elaborate:
>
> First patch splits existing functions to the macros that use _Generic(), and
> old ones become renamed only on the implementation side (to _old or whatever).
> Then you add a new ones and apply them to these macros as a second (default?)
> choice. Then you convert users one-by-one or whatever preference is, and at
> the end you kill macros with old functions and rename "_new" again in a single
> place (pinctrl core).
>

At which point do we go from "unsigned gpio" to "struct gpio_chip *gc,
unsigned int offset" as arguments here?

Bart