[PATCH 3/4] irqchip/gic: Use GIC_* DT binding definitions

Marc Zyngier maz at kernel.org
Thu Mar 5 02:38:53 PST 2026


On Thu, 05 Mar 2026 10:24:23 +0000,
Geert Uytterhoeven <geert at linux-m68k.org> wrote:
> 
> Hi Marc,
> 
> On Thu, 5 Mar 2026 at 11:13, Marc Zyngier <maz at kernel.org> wrote:
> > On Wed, 04 Mar 2026 17:21:58 +0000,
> > Geert Uytterhoeven <geert+renesas at glider.be> wrote:
> > > Replace magic numbers by symbolic DT binding definitions.  This improves
> > > readability, and makes it easier to find where the various GIC
> > > interrupts types are handled.
> > >
> > > Signed-off-by: Geert Uytterhoeven <geert+renesas at glider.be>
> 
> > > --- a/drivers/irqchip/irq-gic-v3.c
> > > +++ b/drivers/irqchip/irq-gic-v3.c
> > > @@ -35,6 +35,8 @@
> > >  #include <asm/smp_plat.h>
> > >  #include <asm/virt.h>
> > >
> > > +#include <dt-bindings/interrupt-controller/arm-gic.h>
> > > +
> > >  #include "irq-gic-common.h"
> > >
> > >  static u8 dist_prio_irq __ro_after_init = GICV3_PRIO_IRQ;
> > > @@ -1602,25 +1604,25 @@ static int gic_irq_domain_translate(struct irq_domain *d,
> > >                       return -EINVAL;
> > >
> > >               switch (fwspec->param[0]) {
> > > -             case 0:                 /* SPI */
> > > +             case GIC_SPI:
> >
> > I'd rather not do that. I use *numeric* values on purpose, because
> > that's what the DT *binding* describes, and I have no control over
> > what lives in that include file (it gets changed without me being even
> > Cc'd).
> >
> > So I want to stick to the binding, and not to the interpretation of
> > it. If you want symbolic values to be used, describe them in the
> > binding, have a tool to generate the values from the binding, and use
> > that everywhere.
> 
> This sounds more like a philosophical debate, so I'd like to defer
> to the DT maintainers...

That's not philosophical.

That's a pragmatic approach to having a common source of information,
and a unique reference. Carrying an extra copy that can be
independently changed is a source of errors, which I've been trying to
reduce in other parts of the kernel (system register description and
encoding, for example).

> 
> About you not being notified: that can be fixed easily ;-)
> 
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -2110,6 +2110,7 @@ F:        arch/arm64/include/asm/arch_gicv3.h
>  F:     drivers/irqchip/irq-gic*.[ch]
>  F:     include/linux/irqchip/arm-gic*.h
>  F:     include/linux/irqchip/arm-vgic-info.h
> +F:     include/dt-bindings/interrupt-controller/arm-gic.h

I'm actively trying to *remove* myself from the kernel, not to grab
more stuff.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.



More information about the Linux-mediatek mailing list