在 2026-01-22星期四的 08:31 +0100，Thomas Zimmermann写道：
> Hi
>
> Am 22.01.26 um 08:06 schrieb Icenowy Zheng:
> > 在 2026-01-21星期三的 13:56 +0100，Thomas Zimmermann写道：
> >
> > =============== 8< =================
> > > > +
> > > > + if (irqs)
> > > > + pr_warn("Unknown Verisilicon DC interrupt 0x%x
> > > > fired!\n", irqs);
> > > I'd avoid pr_warn() here. This can fill your logs quickly.
> > Then is pr_warn_once() okay?
>
> Yeah, warn_once should be fine.
>
> >
> > > > +
> > > > + return IRQ_HANDLED;
> > > > +}
> > =============== 8< =================
>
> Maybe consider returning IRQ_NONE if the (irq) tests succeeds.
Well, the IRQs are read from the device's register, and the read
operation automatically ACKs the IRQ, so it sounds more appropriate to
always return IRQ_HANDLED.
(Well maybe this function should be made return void, and returning
IRQ_HANDLED should be done in the caller of this function,
vs_dc_irq_handler() )
Thanks,
Icenowy
