[v1] irqchip/gic-v3: Print a warning for out-of-range interrupt numbers

[PATCH] irqchip/gic-v3: Print a warning for out-of-range interrupt numbers

Posted by Geert Uytterhoeven 1 month ago

gic_irq_domain_translate() does not check if an interrupt number lies
within the valid range of the specified interrupt type.  Add these
checks, and print a warning if the interrupt number is out of range.

This can help flagging incorrectly described Extended SPI and PPI
interrupts in DT.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
This would have prevented the issue fixed by "[PATCH] arm64: dts:
renesas: r8a78000: Fix out-of-range SPI interrupt numbers"[1].

[1] https://lore.kernel.org/1f9dd274720ea1b66617a5dd84f76c3efc829dc8.1772641415.git.geert+renesas@glider.be
---
 drivers/irqchip/irq-gic-v3.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index 20f13b686ab22faf..d75163e71bf22473 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -1603,15 +1603,27 @@ static int gic_irq_domain_translate(struct irq_domain *d,
 
 		switch (fwspec->param[0]) {
 		case 0:			/* SPI */
+			if (fwspec->param[1] > 987)
+				pr_warn_once("SPI %u out of range (use ESPI?)\n",
+					     fwspec->param[1]);
 			*hwirq = fwspec->param[1] + 32;
 			break;
 		case 1:			/* PPI */
+			if (fwspec->param[1] > 16)
+				pr_warn_once("PPI %u out of range (use EPPI?)\n",
+					     fwspec->param[1]);
 			*hwirq = fwspec->param[1] + 16;
 			break;
 		case 2:			/* ESPI */
+			if (fwspec->param[1] > 1023)
+				pr_warn_once("ESPI %u out of range\n",
+					     fwspec->param[1]);
 			*hwirq = fwspec->param[1] + ESPI_BASE_INTID;
 			break;
 		case 3:			/* EPPI */
+			if (fwspec->param[1] > 63)
+				pr_warn_once("EPPI %u out of range\n",
+					     fwspec->param[1]);
 			*hwirq = fwspec->param[1] + EPPI_BASE_INTID;
 			break;
 		case GIC_IRQ_TYPE_LPI:	/* LPI */
-- 
2.43.0

Re: [PATCH] irqchip/gic-v3: Print a warning for out-of-range interrupt numbers

Posted by Sergey Shtylyov 1 month ago

On 3/4/26 7:31 PM, Geert Uytterhoeven wrote:

> gic_irq_domain_translate() does not check if an interrupt number lies
> within the valid range of the specified interrupt type.  Add these
> checks, and print a warning if the interrupt number is out of range.
> 
> This can help flagging incorrectly described Extended SPI and PPI
> interrupts in DT.
> 
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> ---
> This would have prevented the issue fixed by "[PATCH] arm64: dts:
> renesas: r8a78000: Fix out-of-range SPI interrupt numbers"[1].
> 
> [1] https://lore.kernel.org/1f9dd274720ea1b66617a5dd84f76c3efc829dc8.1772641415.git.geert+renesas@glider.be
> ---
>  drivers/irqchip/irq-gic-v3.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
> index 20f13b686ab22faf..d75163e71bf22473 100644
> --- a/drivers/irqchip/irq-gic-v3.c
> +++ b/drivers/irqchip/irq-gic-v3.c
> @@ -1603,15 +1603,27 @@ static int gic_irq_domain_translate(struct irq_domain *d,
>  
>  		switch (fwspec->param[0]) {
>  		case 0:			/* SPI */
> +			if (fwspec->param[1] > 987)
> +				pr_warn_once("SPI %u out of range (use ESPI?)\n",
> +					     fwspec->param[1]);
>  			*hwirq = fwspec->param[1] + 32;
>  			break;
>  		case 1:			/* PPI */
> +			if (fwspec->param[1] > 16)

   Not 15? Don't PPIs use INTIDs 16-31?

> +				pr_warn_once("PPI %u out of range (use EPPI?)\n",
> +					     fwspec->param[1]);
>  			*hwirq = fwspec->param[1] + 16;
>  			break;
[...]

MBR, Sergey

Re: [PATCH] irqchip/gic-v3: Print a warning for out-of-range interrupt numbers

Posted by Geert Uytterhoeven 1 month ago

Hi Sergey,

On Wed, 4 Mar 2026 at 18:05, Sergey Shtylyov <sergei.shtylyov@gmail.com> wrote:
> On 3/4/26 7:31 PM, Geert Uytterhoeven wrote:
> > gic_irq_domain_translate() does not check if an interrupt number lies
> > within the valid range of the specified interrupt type.  Add these
> > checks, and print a warning if the interrupt number is out of range.
> >
> > This can help flagging incorrectly described Extended SPI and PPI
> > interrupts in DT.
> >
> > Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> > ---
> > This would have prevented the issue fixed by "[PATCH] arm64: dts:
> > renesas: r8a78000: Fix out-of-range SPI interrupt numbers"[1].
> >
> > [1] https://lore.kernel.org/1f9dd274720ea1b66617a5dd84f76c3efc829dc8.1772641415.git.geert+renesas@glider.be
> > ---
> >  drivers/irqchip/irq-gic-v3.c | 12 ++++++++++++
> >  1 file changed, 12 insertions(+)
> >
> > diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
> > index 20f13b686ab22faf..d75163e71bf22473 100644
> > --- a/drivers/irqchip/irq-gic-v3.c
> > +++ b/drivers/irqchip/irq-gic-v3.c
> > @@ -1603,15 +1603,27 @@ static int gic_irq_domain_translate(struct irq_domain *d,
> >
> >               switch (fwspec->param[0]) {
> >               case 0:                 /* SPI */
> > +                     if (fwspec->param[1] > 987)
> > +                             pr_warn_once("SPI %u out of range (use ESPI?)\n",
> > +                                          fwspec->param[1]);
> >                       *hwirq = fwspec->param[1] + 32;
> >                       break;
> >               case 1:                 /* PPI */
> > +                     if (fwspec->param[1] > 16)
>
>    Not 15? Don't PPIs use INTIDs 16-31?

Thank you, that is indeed an off-by-one bug.
Will fix.

>
> > +                             pr_warn_once("PPI %u out of range (use EPPI?)\n",
> > +                                          fwspec->param[1]);
> >                       *hwirq = fwspec->param[1] + 16;
> >                       break;

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds