Re: [PATCH kvmtool v4 3/7] arm64: nested: Add support for setting maintenance IRQ

[Marc Zyngier <maz@kernel.org>]

On Fri, 16 Jan 2026 18:10:11 +0000,
Sascha Bischoff <Sascha.Bischoff@arm.com> wrote:
> 
> On Wed, 2025-09-24 at 14:45 +0100, Andre Przywara wrote:
> > Uses the new VGIC KVM device attribute to set the maintenance IRQ.
> > This is fixed to use PPI 9, as a platform decision made by kvmtool,
> > matching the SBSA recommendation.
> > Use the opportunity to pass the kvm pointer to
> > gic__generate_fdt_nodes(),
> > as this simplifies the call and allows us access to the nested_virt
> > on
> > the way.
> > 
> > Signed-off-by: Andre Przywara <andre.przywara@arm.com>
> > ---
> >  arm64/arm-cpu.c         |  2 +-
> >  arm64/gic.c             | 26 ++++++++++++++++++++++++--
> >  arm64/include/kvm/gic.h |  2 +-
> >  3 files changed, 26 insertions(+), 4 deletions(-)
> > 
> > diff --git a/arm64/arm-cpu.c b/arm64/arm-cpu.c
> > index 69bb2cb2c..0843ac051 100644
> > --- a/arm64/arm-cpu.c
> > +++ b/arm64/arm-cpu.c
> > @@ -14,7 +14,7 @@ static void generate_fdt_nodes(void *fdt, struct
> > kvm *kvm)
> >  {
> >  	int timer_interrupts[4] = {13, 14, 11, 10};
> >  
> > -	gic__generate_fdt_nodes(fdt, kvm->cfg.arch.irqchip);
> > +	gic__generate_fdt_nodes(fdt, kvm);
> >  	timer__generate_fdt_nodes(fdt, kvm, timer_interrupts);
> >  	pmu__generate_fdt_nodes(fdt, kvm);
> >  }
> > diff --git a/arm64/gic.c b/arm64/gic.c
> > index b0d3a1abb..e35986c06 100644
> > --- a/arm64/gic.c
> > +++ b/arm64/gic.c
> > @@ -11,6 +11,8 @@
> >  
> >  #define IRQCHIP_GIC 0
> >  
> > +#define GIC_MAINT_IRQ	9
> > +
> >  static int gic_fd = -1;
> >  static u64 gic_redists_base;
> >  static u64 gic_redists_size;
> > @@ -302,10 +304,15 @@ static int gic__init_gic(struct kvm *kvm)
> >  
> >  	int lines = irq__get_nr_allocated_lines();
> >  	u32 nr_irqs = ALIGN(lines, 32) + GIC_SPI_IRQ_BASE;
> > +	u32 maint_irq = GIC_PPI_IRQ_BASE + GIC_MAINT_IRQ;
> >  	struct kvm_device_attr nr_irqs_attr = {
> >  		.group	= KVM_DEV_ARM_VGIC_GRP_NR_IRQS,
> >  		.addr	= (u64)(unsigned long)&nr_irqs,
> >  	};
> > +	struct kvm_device_attr maint_irq_attr = {
> > +		.group	= KVM_DEV_ARM_VGIC_GRP_MAINT_IRQ,
> > +		.addr	= (u64)(unsigned long)&maint_irq,
> > +	};
> >  	struct kvm_device_attr vgic_init_attr = {
> >  		.group	= KVM_DEV_ARM_VGIC_GRP_CTRL,
> >  		.attr	= KVM_DEV_ARM_VGIC_CTRL_INIT,
> > @@ -325,6 +332,13 @@ static int gic__init_gic(struct kvm *kvm)
> >  			return ret;
> >  	}
> >  
> > +	if (kvm->cfg.arch.nested_virt &&
> > +	    !ioctl(gic_fd, KVM_HAS_DEVICE_ATTR, &maint_irq_attr)) {
> > +		ret = ioctl(gic_fd, KVM_SET_DEVICE_ATTR,
> > &maint_irq_attr);
> > +		if (ret)
> > +			return ret;
> > +	}
> 
> With GICv3 are things not a little broken if we're trying to do nested
> but don't have the ability to set the maint IRQ? It feels to me as if
> an error should be returned if the attr doesn't exist.
> 
> Also, the way that the FDT is generated means that we'd still generate
> the property for the maint IRQ, even if we can't set it here.

Yup, this definitely need fixing. Thanks for catching this.

	M.

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