Re: [PATCH v4 6/6] Documentation: KVM: arm64: Describe VGICv3 registers writable pre-init

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

On Mon, Jul 14, 2025 at 05:06:10PM +0200, Eric Auger wrote:
> Hi,
> 
> On 7/9/25 11:14 PM, Oliver Upton wrote:
> > KVM allows userspace to control GICD_IIDR.Revision and
> > GICD_TYPER2.nASSGIcap prior to initialization for the sake of
> > provisioning the guest-visible feature set. Document the userspace
> > expectations surrounding accesses to these registers.
> > 
> > Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
> > ---
> >  Documentation/virt/kvm/devices/arm-vgic-v3.rst | 17 +++++++++++++++++
> >  1 file changed, 17 insertions(+)
> > 
> > diff --git a/Documentation/virt/kvm/devices/arm-vgic-v3.rst b/Documentation/virt/kvm/devices/arm-vgic-v3.rst
> > index e860498b1e35..c7a1cd22d814 100644
> > --- a/Documentation/virt/kvm/devices/arm-vgic-v3.rst
> > +++ b/Documentation/virt/kvm/devices/arm-vgic-v3.rst
> > @@ -78,6 +78,8 @@ Groups:
> >      -ENXIO   The group or attribute is unknown/unsupported for this device
> >               or hardware support is missing.
> >      -EFAULT  Invalid user pointer for attr->addr.
> > +    -EBUSY   Attempt to write a register that is read-only after
> > +             initialization
> >      =======  =============================================================
> >  
> >  
> > @@ -120,6 +122,15 @@ Groups:
> >      Note that distributor fields are not banked, but return the same value
> >      regardless of the mpidr used to access the register.
> >  
> > +    Userspace is allowed to write the following register fields prior to
> > +    initialization of the VGIC:
> > +
> > +      =====================
> > +      GICD_IIDR.Revision
> > +      GICD_TYPER2.nASSGIcap
> > +      =====================
> > +
> > +
> >      GICD_IIDR.Revision is updated when the KVM implementation is changed in a
> >      way directly observable by the guest or userspace.  Userspace should read
> >      GICD_IIDR from KVM and write back the read value to confirm its expected
> > @@ -128,6 +139,12 @@ Groups:
> >      behavior.
> >  
> >  
> > +    GICD_TYPER2.nASSGIcap allows userspace to control the support of SGIs
> > +    without an active state. At VGIC creation the field resets to the
> In [PATCH v4 4/6] KVM: arm64: vgic-v3: Allow userspace to write
> GICD_TYPER2.nASSGIcap commit message it was said:
> 
> "For convenience, bundle support for vLPIs and vSGIs behind this
> feature,allowing userspace to control vPE allocation for VMs in
> environments that may be constrained on vPE IDs."
> 
> which, I understand goes beyond the simple support of vSGIs. Maybe worth
> a clarification.

I would like to omit that language from the documentation. Giving
userspace the impression of controlling the host allocation is a bit odd
and nothing stops us from implementing nASSGIcap for software-injected
SGIs.

vPE allocation should be a non-issue for correctly provisioned
implementations.

Thanks,
Oliver