Re: [PATCH 5/5] KVM: arm64: vgic-v3: Indicate vgic_put_irq() may take LPI xarray lock

[Oliver Upton <oliver.upton@linux.dev>]

On Thu, Sep 04, 2025 at 11:25:08AM +0100, Ben Horgan wrote:
> Hi Oliver,
> 
> On 9/4/25 07:23, Oliver Upton wrote:
> > The release path on LPIs is quite rare, meaning it can be difficult to
> > find lock ordering bugs on the LPI xarray's spinlock. Tell lockdep that
> > vgic_put_irq() might acquire the xa_lock to make unsafe patterns more
> > obvious.
> > 
> > Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
> > ---
> >  arch/arm64/kvm/vgic/vgic.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/arch/arm64/kvm/vgic/vgic.c b/arch/arm64/kvm/vgic/vgic.c
> > index a21b482844ce..e61b95167ed1 100644
> > --- a/arch/arm64/kvm/vgic/vgic.c
> > +++ b/arch/arm64/kvm/vgic/vgic.c
> > @@ -142,6 +142,9 @@ void vgic_put_irq(struct kvm *kvm, struct vgic_irq *irq)
> >  {
> >  	struct vgic_dist *dist = &kvm->arch.vgic;
> >  
> > +	if (irq->intid > VGIC_MIN_LPI)
> > +		might_lock(&dist->lpi_xa.xa_lock);
> > +
> 
> nit: Use >= rather than > to include all LPI.

Of course, silly typo.

Thanks!
Oliver