RE: [PATCH 3/3] pinctrl: imx: support SCMI pinctrl protocol for i.MX95

[Peng Fan]

Hi Linus,

> Subject: Re: [PATCH 3/3] pinctrl: imx: support SCMI pinctrl protocol for
> i.MX95
> 
> On Tue, May 21, 2024 at 8:17 AM Peng Fan (OSS) <peng.fan@xxxxxxxxxxx>
> wrote:
> 
> > +static int pinctrl_scmi_imx_dt_node_to_map(struct pinctrl_dev *pctldev,
> > +                                          struct device_node *np,
> > +                                          struct pinctrl_map **map,
> > +                                          unsigned int *num_maps)
> (...)
> > +static int pinctrl_scmi_imx_pinconf_set(struct pinctrl_dev *pctldev,
> > +                                       unsigned int pin,
> > +                                       unsigned long *configs,
> > +                                       unsigned int num_configs)
> 
> The code in these functions look suspiciously similar to same code in pinctrl-
> imx.c, I bet it is copy/pase/modify.

I only took the imx_pinctrl_parse_pin_mmio as example to get parse the node
and do the pinctrl_scmi_imx_dt_node_to_map here. Only the pieces:
"be32_to_cpu(*list++); "

For other parts, they are different. There is no MUX here, configs
has vendor SCMI "IMX_SCMI_PIN_X", and more.

> 
> Can you look a second time if it is possible to share code between the drivers?

I thought about this. Just trying what did for i.MX8 SCU pinctrl API by adding
IMX_USE_SCMI flag.

But because that means more if else check in pinctrl-imx.c and 
scmi requires different configs layout, which makes pinctrl-imx.c looks
messy. And scmi pinctrl requires a totally different probe function,
not imx_pinctrl_probe. So I decided to write a new driver for i.MX95.

> 
> It's not super much code, I'm mostly worried about bugs having to be fixed in
> two places.

I could switch back to my initial try to share pinctrl-imx.c, but I hope not.

> 
> What is the opinion of the othe i.MX pinctrl maintainers?

Aisheng,  Fabio, Shawn, Jacky, any comments?

Thanks,
Peng.

> 
> Yours,
> Linus Walleij