Re: [PATCH v2 1/3] serial: mctrl_gpio: Avoid probe failures in case of missing gpiolib

[Uwe Kleine-König]

Hello,

On Fri, Aug 02, 2019 at 07:56:54AM +0000, Schrempf Frieder wrote:
> On 01.08.19 22:33, Uwe Kleine-König wrote:
> > On Thu, Aug 01, 2019 at 06:45:21PM +0000, Schrempf Frieder wrote:
> >> diff --git a/drivers/tty/serial/serial_mctrl_gpio.c b/drivers/tty/serial/serial_mctrl_gpio.c
> >> index 2b400189be91..54c43e02e375 100644
> >> --- a/drivers/tty/serial/serial_mctrl_gpio.c
> >> +++ b/drivers/tty/serial/serial_mctrl_gpio.c
> >> @@ -61,6 +61,9 @@ EXPORT_SYMBOL_GPL(mctrl_gpio_set);
> >>   struct gpio_desc *mctrl_gpio_to_gpiod(struct mctrl_gpios *gpios,
> >>   				      enum mctrl_gpio_idx gidx)
> >>   {
> >> +	if (gpios == NULL)
> >> +		return NULL;
> >> +
> > 
> > I wonder why you need this. If GPIOLIB is off this code isn't active and
> > with GPIOLIB calling mctrl_gpio_to_gpiod with a gpios == NULL is a bug
> > that IMHO should not be silently ignored.
> > 
> > Am I missing something (again)?
> 
> No, you're right. My thoughts were, that if the mctrl_gpio functions are 
> allowed to be passed a NULL pointer in general, they all should have a 
> NULL check, even if in the current context (GPIOLIB disabled) this code 
> is not active. Apparently there are other cases when a NULL pointer is 
> passed, see [1]. So you can't really consider gpios == NULL to be a bug 
> as this seems to be allowed in general.

OK, then this is another separate commit, right? Preferably with a
comment pointing to drivers that use mctrl_gpio before being
initialized.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |