On Wed, Nov 29, 2023 at 11:21:06AM +0300, Dan Carpenter wrote: > On Wed, Nov 29, 2023 at 01:43:35PM +0800, James Tai wrote: > > +static int realtek_intc_subset(struct device_node *node, struct realtek_intc_data *data, int index) > > +{ > > + struct realtek_intc_subset_data *subset_data = &data->subset_data[index]; > > + const struct realtek_intc_subset_cfg *cfg = &data->info->cfg[index]; > > + int irq; > > + > > + irq = irq_of_parse_and_map(node, index); > > + if (irq <= 0) > > + return irq; > > I don't think irq_of_parse_and_map() can return negatives. Only zero > on error. Returning zero on error is a historical artifact with IRQ > functions and a constant source of bugs. But here returning zero is > success. See my blog for more details: > https://staticthinking.wordpress.com/2023/08/07/writing-a-check-for-zero-irq-error-codes/ It's worse than that. The irq functions historically returned NO_IRQ on error, but that could be 0 or -1 depending on the arch. Use of_irq_get() instead. It's a bit better in that it returns an error code for most cases. But still returns 0 on mapping failure. Rob