Re: [PATCH v3 3/5] irqchip/mips-gic: Enable interrupt when moving affinity across clusters

From: Thomas Bogendoerfer

Date: Mon Sep 28 2026 - 04:43:05 EST


On Mon, Sep 07, 2026 at 02:46:37PM +0200, Benoît Monin wrote:
> When an interrupt's affinity is moved to a CPU in another cluster,
> gic_set_affinity() updates the routing (GIC_SH_MAP_VP) and trigger type
> in the destination cluster, but never touched the interrupt's mask state.
>
> The interrupt mask is per-cluster. After such a move the interrupt was
> left disabled in the destination cluster, so it never fires despite
> being correctly routed to its new VP.
>
> Handle the mask explicitly on a cross-cluster affinity change: in the
> old cluster, write GIC_SH_RMASK to disable the interrupt while clearing
> the route so it is no longer delivered. And in the new cluster, set the
> mask to enable the interrupt along with reconfiguring the trigger type.
>
> Fixes: 322a90638768 ("irqchip/mips-gic: Multi-cluster support")
> Signed-off-by: Benoît Monin <benoit.monin@xxxxxxxxxxx>
> ---
> drivers/irqchip/irq-mips-gic.c | 16 +++++++++++++---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c
> index f2ae60d39d66..4b76a65f12c9 100644
> --- a/drivers/irqchip/irq-mips-gic.c
> +++ b/drivers/irqchip/irq-mips-gic.c
> @@ -390,14 +390,17 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *cpumask,
>
> /*
> * If we're moving affinity between clusters, stop routing the
> - * interrupt to any VP(E) in the old cluster.
> + * interrupt to any VP(E) in the old cluster and disable
> + * the interrupt in that cluster.
> */
> if (cl != old_cl) {
> if (gic_irq_lock_cluster(d)) {
> write_gic_redir_map_vp(irq, 0);
> + write_gic_redir_rmask(irq);
> mips_cm_unlock_other();
> } else {
> write_gic_map_vp(irq, 0);
> + write_gic_rmask(irq);
> }
> }
>
> @@ -409,10 +412,17 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *cpumask,
>
> /*
> * If we're moving affinity between clusters, configure the interrupt
> - * trigger type in the new cluster.
> + * trigger type and enable the interrupt in the new cluster.
> */
> - if (cl != old_cl)
> + if (cl != old_cl) {
> gic_set_type_locked(d, irqd_get_trigger_type(d));
> + if (gic_irq_lock_cluster(d)) {
> + write_gic_redir_smask(irq);
> + mips_cm_unlock_other();
> + } else {
> + write_gic_smask(irq);
> + }
> + }

shouldn't this be done depending on the mask state in the old cluster ?

Thomas.

--
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea. [ RFC1925, 2.3 ]