RE: [RFC v3 1/2] pinctrl: core: Add pinctrl_mux_gpio_request_enable

[Fabrizio Castro <fabrizio.castro@xxxxxxxxxxxxxx>]

Hello Geert,

Thank you for your feedback!

> From: linux-renesas-soc-owner@xxxxxxxxxxxxxxx <linux-renesas-soc-owner@xxxxxxxxxxxxxxx> On Behalf Of Geert Uytterhoeven
> Sent: 01 March 2019 13:28
> Subject: Re: [RFC v3 1/2] pinctrl: core: Add pinctrl_mux_gpio_request_enable
> 
> Hi Fabrizio,
> 
> On Thu, Dec 6, 2018 at 10:47 AM Fabrizio Castro
> <fabrizio.castro@xxxxxxxxxxxxxx> wrote:
> > > From: Linus Walleij <linus.walleij@xxxxxxxxxx>
> > > Sent: 05 December 2018 21:46
> > > Subject: Re: [RFC v3 1/2] pinctrl: core: Add pinctrl_mux_gpio_request_enable
> > >
> > > On Tue, Nov 20, 2018 at 4:19 PM Fabrizio Castro
> > > <fabrizio.castro@xxxxxxxxxxxxxx> wrote:
> > >
> > > > Sometimes there is the need to change the muxing of a pin to make it
> > > > a GPIO without going through gpiolib.
> > > > This patch adds pinctrl_mux_gpio_request_enable to deal with this new
> > > > use case from code that has nothing to do with pinctrl.
> > >
> > > It has a lot to do with pinctrl I think, so I get confused by this
> > > commit message.
> >
> > I can improve that
> >
> > >
> > > >  extern int pinctrl_gpio_request(unsigned gpio);
> > > > +extern int pinctrl_mux_gpio_request_enable(unsigned gpio);
> > >
> > > What's wrong with just using the existing call
> > > pinctrl_gpio_request() right above your new one?
> > >
> > > It's not like we're reference counting or something, it's just
> > > a callback. Sprinkle some comments to show what's going
> > > on.
> >
> > I tried that, and it was working for me, then something changed lately
> > in gpiolib that broke that solution, and Geert picked it up on his end.
> > Please see this:
> > https://patchwork.kernel.org/patch/10671325/
> >
> > This patch was made to overcome the problems of the previous patch.
> >
> > >
> > > If you for some reason need a new call for this specific
> > > use case, it needs to be named after the use case,
> > > like pinctrl_gpio_request_for_irq()
> > > so it is obvious what the function is doing.
> >
> > I can do that, but I would like to hear from Geert first, no point in going
> > around in circle if this solution is not acceptable to him.
> >
> > Geert, what do you think?
> 
> /me gives v3 a try on Koelsch, Salvator-XS, and Ebisu-4D:
> 
>     BUG: sleeping function called from invalid context
> 
> for mmc, adv7511, gpio-keys, and Ethernet PHY.

Doh! It was running smoothly on my iwg23s when I tested it, I am going to
have another look at this at a later stage.

Thank you for looking into this!

Cheers,
Fab

> 
> Gr{oetje,eeting}s,
> 
>                         Geert
> 
> 
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
> 
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                 -- Linus Torvalds