Re: [PATCH v1] gpiolib: Don't WARN on gpiod_put() for optional GPIO

[Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>]

On Tue, Mar 26, 2019 at 06:48:25PM +0100, Bartosz Golaszewski wrote:
> wt., 26 mar 2019 o 16:21 Andy Shevchenko
> <andriy.shevchenko@xxxxxxxxxxxxxxx> napisał(a):
> >
> > In case of debug and optional GPIO requested, the gpiod_put() is not aware of
> > and will WARN, which is not the case.
> >
> > Make gpiod_put() NULL-aware to keep silent for optional GPIOs.
> >
> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> > ---
> >  drivers/gpio/gpiolib.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > index 144af0733581..36445e24ee89 100644
> > --- a/drivers/gpio/gpiolib.c
> > +++ b/drivers/gpio/gpiolib.c
> > @@ -4616,7 +4616,8 @@ EXPORT_SYMBOL_GPL(gpiod_get_array_optional);
> >   */
> >  void gpiod_put(struct gpio_desc *desc)
> >  {
> > -       gpiod_free(desc);
> > +       if (desc)
> > +               gpiod_free(desc);
> >  }
> >  EXPORT_SYMBOL_GPL(gpiod_put);
> >
> > --
> > 2.20.1
> >
> 
> In that case I don't think we need to keep checking desc for NULL in
> gpiod_free().

I didn't remove the check there because it's called more than just from
gpiod_put(). I didn't check all code paths myself. So, if it seems good for you
to remove the check I can do it in next version.

-- 
With Best Regards,
Andy Shevchenko