Re: [RFC PATCH 0/3] pinctrl: can_sleep and pinctrl_gpio_direction

[Linus Walleij <linus.walleij@xxxxxxxxxx>]

On Tue, Dec 7, 2021 at 11:14 PM Andy Shevchenko
<andy.shevchenko@xxxxxxxxx> wrote:
> On Monday, December 6, 2021, Hans Verkuil <hverkuil-cisco@xxxxxxxxx> wrote:
>>
>> Hi all,
>>
>> Based on this discussion:
>>
>> https://lore.kernel.org/linux-gpio/CACRpkdb3q4-9O3dHS6QDWnZZ5JJjXWXS9KPvwXVaowLMRhcejA@xxxxxxxxxxxxxx/T/#t
>>
>> I propose this RFC series.
>
>
> When I first saw your report I was thinking about actually adding a new callback ->set_direction_atomic()
> and then make pinctrl use it, otherwise like you do, I.e. issue a warning when it’s called in atomic context

The problem is inside of pinctrl core, not in any driver. It takes a
mutex when going over
the GPIO ranges.

I suggested maybe just replacing these mutexes with spinlocks, or RCU.

Yours,
Linus Walleij