On Sat, May 1, 2021 at 7:13 AM Linus Walleij [off-list ref] wrote:

On Mon, Apr 26, 2021 at 11:54 AM Thomas Bogendoerfer
[off-list ref] wrote:

quoted

Add YAML devicetree binding for IDT 79RC32434 GPIO controller

Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
---
Changes in v4:
 - renamed to idt,32434-gpio this time for real

Overall looks good to me.

quoted

+required:

(...)

quoted

+  - ngpios

Is there a *technical* reason why this is required?

Can't the driver just default to 32 gpios when not specified?

quoted

+  - interrupt-controller
+  - "#interrupt-cells"
+  - interrupts

Why can't interrupt support be made optional?

It is fine if the driver errors out if not provided, but
for the bindings this feels optional.

Or does the thing break unless you handle the IRQs?

If the hardware has interrupts, then we should describe that. It's the
OS driver that may or may not support interrupts.

Rob