Re: [PATCH v3 1/7] pinctrl: Renesas RZ/A1 pin and gpio controller

[jacopo <jacopo@jmondi.org>]

Hi Linus,

On Fri, Mar 24, 2017 at 04:42:47PM +0100, Linus Walleij wrote:
> On Fri, Mar 24, 2017 at 4:22 PM, Jacopo Mondi <jacopo+renesas@jmondi.org> wrote:
> 
> I assume Geert will queue this driver even if it is outside of sh-pfc?
> 
> > Add combined gpio and pin controller driver for Renesas RZ/A1
> > r7s72100 SoC.
> >
> > Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
> > ---
> >  drivers/pinctrl/Kconfig        |  10 +
> >  drivers/pinctrl/Makefile       |   1 +
> >  drivers/pinctrl/pinctrl-rza1.c | 961 +++++++++++++++++++++++++++++++++++++++++
> 
> So this is very different from the SH-PFC family and should not
> be in drivers/pinctrl/sh-pfc?
> 

Correct. The pin controller hardware in RZ/A(1) devices configures pin
functions per-pin and not per-group as the SH/R-Car family does.

There has been an attempt to support this platforms with the existing
sh-pfc/ infrastructure in the past (https://lwn.net/Articles/573222/)
which I re-proposed for v4.9, but that driver simply does not fit
this hardware which is different from the one found in R-Car
devices.

More on the background of this series in this email thread:
https://marc.info/?l=linux-gpio&m=148536779917834&w=2

I initially created a whole new sub-directory where all the Renesas
devices with this kind of pin controller would have gone
(drivers/pinctrl/rz-pfc) but since as of now the only available
hardware of that type is RZ/A1 and some of its variants, we decided to go
with a single driver supporting that platform only.
If more will come, we'll think about supporting them through this
driver if possible, or create a dedicated directory.

Are you ok with this?

Thanks
  j

> > +config PINCTRL_RZA1
> > +       bool "Renesas RZ/A1 gpio and pinctrl driver"
> > +       depends on OF
> > +       depends on ARCH_R7S72100 || COMPILE_TEST
> > +       select GENERIC_PINCTRL_GROUPS
> > +       select GENERIC_PINMUX_FUNCTIONS
> > +       select GENERIC_PINCONF
> 
> If it is also a GPIO driver I guess it should
> select GPIOLIB as well.
> 
> This was not possible in the past, but it is possible nowadays.
> 
> > +struct gpio_chip rza1_gpiochip_template = {
> > +       .request                = rza1_gpio_request,
> > +       .free                   = rza1_gpio_free,
> > +       .get_direction          = rza1_gpio_get_direction,
> > +       .direction_input        = rza1_gpio_direction_input,
> > +       .direction_output       = rza1_gpio_direction_output,
> > +       .get                    = rza1_gpio_get,
> > +       .set                    = rza1_gpio_set,
> > +};
> 
> We now also have .set_multiple() and more interestingly
> .set_config() which can be backed by pinctrl if you want
> to e.g. support debouncing and/or open drain/open source.
> 
> Maybe this is stuff your pin controller can do, but not needed
> in the initial submission for sure.
> 
> > +static int rza1_pinmux_set(struct pinctrl_dev *pctldev, unsigned int selector,
> > +                          unsigned int group)
> 
> Please name it rza1_set_mux() to correspond with the ops field.
> 
> Yours,
> Linus Walleij