On Tue, Jun 23, 2020 at 11:40:11AM +0300, Andy Shevchenko wrote: > On Mon, Jun 22, 2020 at 07:26:51PM +0100, Mark Brown wrote: > > On Mon, Jun 22, 2020 at 01:23:04PM -0500, Pierre-Louis Bossart wrote: > > > So I explicitly added a test for all possible cases. I don't mind removing > > > the _OR_NULL if indeed it's safe, but showing this redundancy also shows an > > > intent to deal with such cases. > > Yeah, I think that's fine - it's perhaps redundant but we're not in a > > hot path and as well as the intentionality it saves the reader from > > having to know if gpiod_put() copes with NULL or not. > What the point? > We can document this instead of being a dead code, right? To a certain extent the _OR_NULL is documentation - the effect is the same with or without it. > IS_ERR() may happen there only if we assign such pointer to be error. Is it possible case here? That one is a bit more real.
Attachment:
signature.asc
Description: PGP signature
