On Thu, Nov 10, 2022 at 03:48:02PM +0200, Andy Shevchenko wrote:
> On Wed, Nov 09, 2022 at 11:08:07AM -0800, Dmitry Torokhov wrote:
> > On Wed, Nov 09, 2022 at 01:20:46PM +0200, Andy Shevchenko wrote:
> > > On Tue, Nov 08, 2022 at 04:26:51PM -0800, Dmitry Torokhov wrote:
>
> ...
>
> > > > + pr_debug("%s: parsed '%s' property of node '%pfwP[%d]' - status (%d)\n",
> > > > + __func__, propname, fwnode, idx, PTR_ERR_OR_ZERO(desc));
> > >
> > > %pe ?
> >
> > "/* %pe with a non-ERR_PTR gets treated as plain %p */".
> >
> > I do not think users are interested in the address on success.
>
> Hmm... Perhaps we can teach %pe to behave differently with, e.g. %pe0,
> modification.
Yes, and maybe we could even have %e for normal errors ;)
> But this is another story. So, let's go with your variant.
>
> ...
>
> > > > +#include <dt-bindings/gpio/gpio.h>
> > >
> > > Not sure why we have this here.
> >
> > For convenience - so that users have access to GPIO_ACTIVE_HIGH/
> > GPIO_ACTIVE_LOW and other flags.
>
> Okay, would we make this as a guarantee then?
>
> In such case a comment before this inclusion should be added to explain why
> we do that without any actual user to be present in the header file.
Just to close the loop - I added a comment reflecting this in v3.
Thanks.
--
Dmitry
