Re: [PATCH v6 07/13] gpio: mxc: convert pad wakeup compatible checks to hwdata flags

[Linus Walleij]

On Wed, Oct 14, 2026 at 10:30 AM Peng Fan <peng.fan@xxxxxxxxxxx> wrote:
> On Wed, Oct 14, 2026 at 01:10:48AM +0200, Linus Walleij wrote:

> >Hi Peng,
> >
> >On Sun, Oct 11, 2026 at 1:51 PM Peng Fan (OSS) <peng.fan@xxxxxxxxxxx> wrote:
> >
> >>  static bool mxc_gpio_generic_config(struct mxc_gpio_port *port,
> >>                 unsigned int offset, unsigned long conf)
> >>  {
> >> -       struct device_node *np = port->dev->of_node;
> >> +       int ret;
> >> +
> >> +       if (!mxc_gpio_has_pad_wakeup(port))
> >> +               return false;
> >
> >Why is the entire call to mxc_gpio_generic_config()
> >bailing out here if the port does not have pad wakeup?
> >
> >> -       if (of_device_is_compatible(np, "fsl,imx8dxl-gpio") ||
> >> -           of_device_is_compatible(np, "fsl,imx8qxp-gpio") ||
> >> -           of_device_is_compatible(np, "fsl,imx8qm-gpio"))
> >> -               return (gpiochip_generic_config(&port->gen_gc.gc,
> >> -                                               offset, conf) == 0);
> >> +       ret = gpiochip_generic_config(&port->gen_gc.gc, offset, conf);
> >> +       if (ret)
> >> +               return false;
> >
> >Then no generic config can happen.
> >
> >This looks weird, isn't it just some of the configs that
> >don't work?
>
> The early return is correct and matches the original behavior - only
> i.MX8 SoCs with SCU have pad wakeup hardware.
>
> gpiochip_generic_config() here calls into the pinctrl backend's
> pin_config_set(), which on i.MX8 programs the SCU firmware with the
> PAD wakeup source type.  Non-i.MX8 SoCs don't have an SCU pinctrl backend.
>
> The function name is misleading - its only caller is mxc_gpio_set_pad_wakeup().

This is just a big hack and I want it removed.

Implicit in this reasoning is: the only config we support on
the pinctrl back-end is wakeups so let's gate the entire function
call if we don't support wakeup.

This is a *generic* back-end like the function says:
gpiochip_generic_config() so the GPIO driver can't make
assumptions about what the pinctrl backend can and
cannot do. In a new SoC using the same driver the
pinctrl back-end could be something different.

So this is breaking modularization and encapsulation.

It is better to let the pin controller return -ENOTSUPP
or whatever is appropriate instead of gating the
call like this,

Yours,
Linus Walleij