On Mon, Aug 26, 2024 at 11:01:50AM +0100, Jonathan Cameron wrote: > On Sat, 24 Aug 2024 14:02:22 +0200 > Vasileios Amoiridis <vassilisamir@xxxxxxxxx> wrote: > > On Fri, Aug 23, 2024 at 11:06:28PM +0300, Andy Shevchenko wrote: > > > On Fri, Aug 23, 2024 at 08:17:13PM +0200, Vasileios Amoiridis wrote: ... > > > > + fwnode = dev_fwnode(data->dev); > > > > + if (!fwnode) > > > > + return -ENODEV; > > > > > > Why do you need this? The below will fail anyway. > > > > Because If I don't make this check then fwnode might be garbage and I will > > pass garbage to the fwnode_irq_get() function. Or do I miss something? > It checks for NULL which is all it can actually be and returns a suitable > error code if it is. Actually not. It may be NULL, error pointer, or valid. So, for a bare minimum this check is not full (and again, fwnode APIs should validate fwnode before accessing them where it makes sense; if fwnode_irq_get() does not do that or misses the case(s), it has to be improved). -- With Best Regards, Andy Shevchenko
