On Mon, Dec 20, 2021 at 8:08 PM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:
> On Mon, Dec 20, 2021 at 3:19 PM Daniel Lezcano
> <daniel.lezcano@xxxxxxxxxx> wrote:
> > On 20/12/2021 14:48, Geert Uytterhoeven wrote:
> > > On Mon, Dec 20, 2021 at 1:29 PM Daniel Lezcano
> > > <daniel.lezcano@xxxxxxxxxx> wrote:
> > >> On 18/12/2021 15:41, Lad Prabhakar wrote:
> > >>> + irq = platform_get_irq_optional(pdev, i);
> > >>> + if (irq <= 0 && irq != -ENXIO) {
> > >>> + ret = irq ? irq : -ENXIO;
> > >>> + goto error_unregister;
> > >>> + }
> > >>> + if (irq == -ENXIO)
> > >>> continue;
> > >>
> > >> Why not invert the conditions?
> > >>
> > >> if (irq == -ENXIO)
> > >> continue;
> > >
> > > And this can be break.
> > >
> > >>
> > >> if (irq <= 0) {
> > >> ret = irq ? irq : -ENXIO;
> > >
> > > irq == 0 cannot happen.
Even if it's so, it adds a burden on my shoulders in the future.
> > >> goto out_unregister;
> > >> }
> I think so, as I don't see your point, neither ;-)
>
> I meant (a) there is no need to continue the loop when there are no
> more interrupts present, and (b) irq == 0 cannot happen, so the cod
> can be simplified to:
>
> if (irq == -ENXIO)
> break;
This should be a better check to include 0 as no IRQ case. It will
allow the platform_get_irq_optional() API to be aligned with other
_optional() APIs.
> if (irq < 0) {
> ret = irq;
> goto out_unregister;
> }
--
With Best Regards,
Andy Shevchenko
