On Thu, Feb 20, 2025 at 05:51:42PM +0200, Raag Jadav wrote: > On Thu, Feb 20, 2025 at 05:40:24PM +0200, Andy Shevchenko wrote: > > On Thu, Feb 20, 2025 at 05:30:07PM +0200, Raag Jadav wrote: > > > On Thu, Feb 20, 2025 at 03:44:59PM +0200, Andy Shevchenko wrote: ... > > > > +/* Same as devm_remove_action(), but doesn't WARN() if action wasn't added before */ > > > > +static inline > > > > +void devm_remove_action_optional(struct device *dev, void (*action)(void *), void *data) > > > > +{ > > > > + int ret; > > > > + > > > > + ret = devm_remove_action_nowarn(dev, action, data); > > > > + if (ret == -ENOENT) > > > > + return; > > > > + > > > > + WARN_ON(ret); > > > > +} > > > > > > Trying to wrap my head around this one, can't the user simply do > > > > > > if (devm_is_action_added()) > > > devm_remove_action/_nowarn(); > > > > Hmm... Actually it sounds like a good point. I will check > > (and I like the idea of dropping this patch). > > And perhaps > > s/devm_is_action_added/devm_action_is_added > > But whichever you think _is best_ ;) I thought about that and that's why I would like to stick to the my variant. Thanks for the review! -- With Best Regards, Andy Shevchenko