On Wed, Jun 12, 2013 at 10:56:46AM -0600, Stephen Warren wrote: > On 06/12/2013 10:46 AM, Mark Brown wrote:
> > Unfortunately gpio_is_valid() is unhelpful for platform data since > > often zero is a valid GPIO but it's also the default "do nothing" > > platform data. It's therefore better to either include a check for > > non-zero as well or have code that takes a zero in the platform > > data and sets it to a negative value instead. > I think people filling in platform data should simply be required to > put a valid/correct value in that field. There aren't any users of > this, so it's not like adding a new field where > backwards-compatibility might be a concern (and even then, updating > all users of this platform data type wouldn't be hard), and it should > be obvious if you get it wrong. It's not idiomatic to do it this way, it'll catch people out - sadly the effects are often non-obvious, depending on what actually happens with the GPIO. Were it not for renumbering pain it'd probably be most sensible to just make 0 never a valid GPIO :/
signature.asc
Description: Digital signature
_______________________________________________ devicetree-discuss mailing list devicetree-discuss@lists.ozlabs.org https://lists.ozlabs.org/listinfo/devicetree-discuss