[PATCH v2] irqchip/gic-v5: Install root IRQ handler last in gicv5_init_common()

[PATCH v2] irqchip/gic-v5: Install root IRQ handler last in gicv5_init_common()

[Jiangshan Yi]

set_handle_irq() cannot be undone: once a handler is installed, any
further call returns -EBUSY, so the set_handle_irq(NULL) in the error
path has never worked.  Drop it, and install the root handler only
after gicv5_irs_enable() has succeeded.

set_handle_irq() only fails if another root handler is already
installed, which cannot happen on a GICv5 system.  Should it ever
fail, the system is unusable: panic instead of unwinding.

Link: https://lore.kernel.org/r/87fqzrmywu.wl-maz@kernel.org
Signed-off-by: Jiangshan Yi <yijiangshan@kylinos.cn>
---
 drivers/irqchip/irq-gic-v5.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/drivers/irqchip/irq-gic-v5.c b/drivers/irqchip/irq-gic-v5.c
index ac2d423b1723..c365f8150300 100644
--- a/drivers/irqchip/irq-gic-v5.c
+++ b/drivers/irqchip/irq-gic-v5.c
@@ -1166,21 +1166,18 @@ static int __init gicv5_init_common(struct fwnode_handle *parent_domain)
 	if (ret)
 		goto out_int;
 
-	ret = set_handle_irq(gicv5_handle_irq);
+	ret = gicv5_irs_enable();
 	if (ret)
 		goto out_int;
 
-	ret = gicv5_irs_enable();
-	if (ret)
-		goto out_handle;
+	if (set_handle_irq(gicv5_handle_irq))
+		panic("GICv5: unable to install root IRQ handler\n");
 
 	gicv5_smp_init();
 
 	gicv5_irs_its_probe();
 	return 0;
 
-out_handle:
-	set_handle_irq(NULL);
 out_int:
 	gicv5_cpu_disable_interrupts();
 	gicv5_free_domains();
-- 
2.25.1

Re: [PATCH v2] irqchip/gic-v5: Install root IRQ handler last in gicv5_init_common()

[Lorenzo Pieralisi]

On Thu, Sep 03, 2026 at 06:40:14PM +0800, Jiangshan Yi wrote:
> set_handle_irq() cannot be undone: once a handler is installed, any
> further call returns -EBUSY, so the set_handle_irq(NULL) in the error
> path has never worked.  Drop it, and install the root handler only
> after gicv5_irs_enable() has succeeded.

Thank you.

Yes, that's what I should have done instead of trying to undo a fatal error.

> set_handle_irq() only fails if another root handler is already
> installed, which cannot happen on a GICv5 system.  Should it ever

Nit: define "cannot happen". It should not happen.

> fail, the system is unusable: panic instead of unwinding.

Yes that's what needs to be done instead of trying to fix things up that are
clearly fatal.

Apologies for this churn.

> Link: https://lore.kernel.org/r/87fqzrmywu.wl-maz@kernel.org
> Signed-off-by: Jiangshan Yi <yijiangshan@kylinos.cn>
> ---
>  drivers/irqchip/irq-gic-v5.c | 9 +++------
>  1 file changed, 3 insertions(+), 6 deletions(-)

Reviewed-by: Lorenzo Pieralisi <lpieralisi@kernel.org>

> diff --git a/drivers/irqchip/irq-gic-v5.c b/drivers/irqchip/irq-gic-v5.c
> index ac2d423b1723..c365f8150300 100644
> --- a/drivers/irqchip/irq-gic-v5.c
> +++ b/drivers/irqchip/irq-gic-v5.c
> @@ -1166,21 +1166,18 @@ static int __init gicv5_init_common(struct fwnode_handle *parent_domain)
>  	if (ret)
>  		goto out_int;
>  
> -	ret = set_handle_irq(gicv5_handle_irq);
> +	ret = gicv5_irs_enable();
>  	if (ret)
>  		goto out_int;
>  
> -	ret = gicv5_irs_enable();
> -	if (ret)
> -		goto out_handle;
> +	if (set_handle_irq(gicv5_handle_irq))
> +		panic("GICv5: unable to install root IRQ handler\n");
>  
>  	gicv5_smp_init();
>  
>  	gicv5_irs_its_probe();
>  	return 0;
>  
> -out_handle:
> -	set_handle_irq(NULL);
>  out_int:
>  	gicv5_cpu_disable_interrupts();
>  	gicv5_free_domains();
> -- 
> 2.25.1
>