On Sun, Dec 05, 2021 at 01:06:07AM +0100, Linus Walleij wrote:
> On Thu, Dec 2, 2021 at 10:17 PM Andy Shevchenko
> <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote:
>
> > GPIO library does copy the of_node from the parent device of
> > the GPIO chip, there is no need to repeat this in the individual
> > drivers. Remove these assignment all at once.
> >
> > For the details one may look into the of_gpio_dev_init() implementation.
> >
> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
>
> This is definitely a patch in the right direction, as Bart says
> it can be a bit dangerous, the outliers are those drivers that
> assign the .dev to something completely different than the
> the dev where the of_node is copied from.
I carefully checked these all and this patch series is only for the cases
when I'm sure it's the same device, which is used as parent, and its of_node
supplied.
> The idea was definitely always to only assign it in the core
> *unless* there is a reason to have a completely different
> of_node for some reason.
>
> > +++ b/drivers/gpio/gpio-rda.c
> > @@ -240,8 +240,6 @@ static int rda_gpio_probe(struct platform_device *pdev)
> > rda_gpio->chip.label = dev_name(dev);
> > rda_gpio->chip.ngpio = ngpios;
> > rda_gpio->chip.base = -1;
> > - rda_gpio->chip.parent = dev;
> > - rda_gpio->chip.of_node = np;
>
> Mention in the commit message that in this driver
> you also drop the the .parent assignment because the
> core will handle it.
Okay, I will update it. Also I'll update to the last codebase (dunno if Bart
is going to pull the IB from Lee where one of the drivers is gone: da53cc634cea
("gpio: bd70528 Drop BD70528 support").
--
With Best Regards,
Andy Shevchenko
