Re: [PATCH 00/11] pinctrl: scmi: fixes and GPIO function support

From: Linus Walleij

Date: Fri Nov 13 2026 - 11:37:35 EST


On Fri, Oct 30, 2026 at 4:58 PM Clément Le Goffic
<clegoffic@xxxxxxxxxxxx> wrote:

> Five fixes: the drive strength went out in milliamperes, the schmitt
> trigger enable was mapped to the input buffer, the settings Linux
> carries without argument went out as disabled, a setting the platform
> reports off read as set, and the protocol layer read past a
> PINCTRL_LIST_ASSOCIATIONS reply larger than its receive buffer.
>
> Then GPIO function support: report the pin-only and GPIO attributes of
> a function, give pin-only functions no groups, mark the GPIO functions,
> and select the GPIO function of a pin on a GPIO request.
>
> Tested on R-Car X5H with a v6.18 backport that needed adapting, all but
> the schmitt mapping, which the platform does not implement. An LLM was
> used to review the patches and to set up the tests.
>
> ---
> Clément Le Goffic (11):
> pinctrl: scmi: send the drive strength in microamperes
> pinctrl: scmi: map the schmitt trigger enable to its SCMI type
> pinctrl: scmi: send the settings without argument as enabled
(...)
> pinctrl: scmi: give no groups to pin-only functions

As noted:

These four patches applied as fixes and I will also get them to Torvalds
for v7.4 ASAP.

I think we may have one or two more SCMI fixes that need to go into
v7.4 but it's good to start with these.

Yours,
Linus Walleij