Re: [RFC v1 2/4] pinctrl: add polarfire soc mssio pinctrl driver

[Linus Walleij]

On Mon, Nov 24, 2025 at 6:16 PM Conor Dooley <conor@xxxxxxxxxx> wrote:

> I was looking at the kernel part of this today, trying to figure out
> where it would make sense to actually check this, but I'm not super keen
> on what has to be done. I think doing it in parse_dt_cfg() makes the
> most sense, setting flags if the property is one we care about during
> the loop and then checking mutual exclusion at the end based on the
> flags? The gpio example you gave has it easy, since they already appear
> to have these things stored in flag properties.
> Is there somewhere else, in addition to creating the config from dt that
> this would have to be checked?


We are right now parsing with an array of
struct pinconf_generic_params:

static const struct pinconf_generic_params dt_params[] = {
    { "input-disable", PIN_CONFIG_INPUT_ENABLE, 0 },
    (...)
};

(Sorry for not using C99 .initializers on the above array)

Struct looks like so:

struct pinconf_generic_params {
        const char * const property;
        enum pin_config_param param;
        u32 default_value;
        const char * const *values;
        size_t num_values;
};

Can't we add a
    const enum pin_config_param *conflicts;
    size_t num_conflicts;

And rewrite the parsing table to be more explicit:

static const char * const input_disable_conflicts[] = {
    "input-enable",
};

static const struct pinconf_generic_params dt_params[] = {
    {
        .property = "input-disable",
        .param = PIN_CONFIG_INPUT_ENABLE,
        .default_value = 0,
        .conflicting_properties = input_disable_conflicts,
        .num_conflicting_properties = ARRAY_SIZE(input_disable_conflicts),
     },
    (...)
};

Then in the loop we can use of_property_present(np, ...) to check for
conflicting properties when we encounter something.

Yours,
Linus Walleij