RE: [PATCH v4 3/8] irqchip/al-fic: keep the device_node instead of a cached name string

[Farber, Eliav]

On Sat, 2026-10-10 at 11:33 -0400, Radu Rendec wrote:
> It looks like this was *not* the intended behavior of of_irq_init()
> after all; it was a bug and it was fixed in commit 30724547b221
> ("of/irq: Fix remaining refcount leaks in of_irq_init()") [1].
>
> Anyway, you'll have to address this in v5 because it's clearly not
> going to work with that patch merged. I guess just take the refcount
> explicitly in al_fic_wire_init() where you store the pointer.

Agreed, thanks for catching it. 30724547b221 is already in mainline
(v7.3-rc6), so v5 is based on a tree that has it.

v5 takes the reference where the pointer is stored:

        fic->node = of_node_get(node);

and drops it on the error path:

 err_free:
        of_node_put(fic->node);
        kfree(fic);

I also reworked the commit message. It used to claim that of_irq_init()
never drops the reference on a successful init, which is no longer true.
It now says why the reference is needed and cites 30724547b221.

Thanks,
Eliav