On Thu, 2025-09-04 at 21:57 +0200, Linus Walleij wrote:
> Hi Nuno,
>
> thanks for your patch!
>
> On Wed, Sep 3, 2025 at 12:04 PM Nuno Sá via B4 Relay
> <devnull+nuno.sa.analog.com@xxxxxxxxxx> wrote:
>
> > From: Nuno Sá <nuno.sa@xxxxxxxxxx>
> >
> > The LTC4283 device has up to 8 pins that can be configured as GPIOs.
> >
> > Note that PGIO pins are not set as GPIOs by default so if they are
> > configured to be used as GPIOs we need to make sure to initialize them
> > to a sane default. They are set as inputs by default.
> >
> > Signed-off-by: Nuno Sá <nuno.sa@xxxxxxxxxx>
>
> (...)
>
> > +config GPIO_LTC4283
> > + tristate "Analog Devices LTC4283 GPIO support"
> > + depends on SENSORS_LTC4283
>
> Could that be
> depends on REGMAP && (SENSOR_LTC4283 || COMPILE_TEST)
> ?
>
> Or does something blow up if you do that? (I guess it also needs
> AUXBUS but more on that below)
>
> should it also be
>
> default SENSOR_LTC4283
>
> Sof if that is compiled in (=y) or module (=m) then this becomes
> the same by default?
>
> > + help
> > + If you say yes here you want the GPIO function available in Analog
> > + Devices LTC4283 Negative Voltage Hot Swap Controller.
> > +
> > + This driver can also be built as a module. If so, the module will
> > + be called gpio-ltc4283.
> > +
> > config GPIO_MB86S7X
>
> This is placed among the memory-mapped drivers, but:
>
> > +#include <linux/auxiliary_bus.h>
> (...)
> > +static struct auxiliary_driver ltc4283_gpio_driver = {
> > + .probe = ltc4283_gpio_probe,
> > + .id_table = ltc4283_aux_id_table,
> > +};
> > +module_auxiliary_driver(ltc4283_gpio_driver);
>
> Create a new submenu for auxiliary bus drivers and add it
> there. We already have a submenu for MFD so why not?
>
> menu "AUXBUS GPIO expanders"
> depends on AUXILIARY_BUS
> ...
>
> Have you looked into using GPIO_REGMAP?
> I guess some specials are used here so maybe it is
> not possible.
>
Yeps I looked and I even replied to you on some other thread
https://lore.kernel.org/linux-iio/908feb42783fd182c8b0f22ae5c147de2f7a60d2.camel@xxxxxxxxx/
So for now, .init_valid_mask() is the real blocker (I think)
- Nuno Sá
> Yours,
> Linus Walleij
