Re: [RFC PATCH-for-9.1? 1/2] target/arm: Move qmp_query_gic_capabilities() to hw/intc/

[Peter Maydell <peter.maydell@linaro.org>]

On Thu, 8 Aug 2024 at 12:54, Daniel P. Berrangé <berrange@redhat.com> wrote:
>
> On Thu, Aug 08, 2024 at 12:32:35PM +0100, Peter Maydell wrote:
> > On Thu, 8 Aug 2024 at 12:23, Daniel P. Berrangé <berrange@redhat.com> wrote:
> > > On Thu, Aug 08, 2024 at 11:15:17AM +0100, Peter Maydell wrote:
> > > > It's not possible to use KVM with that machine type, so the
> > > > question is a bit moot. (This also indicates that the
> > > > interface is not very helpful -- it purports to tell the
> > > > management layer whether it can use an accelerated in-kernel
> > > > GIC, but because it doesn't specifiy the board type there's
> > > > no way to provide an accurate answer. It would be useful
> > > > to know exactly what libvirt/etc actually use this for...)
> > >
> > > Libvirt uses this exclusively with the arm 'virt' machine type.
> > >
> > > If the user didn't express any GIC preference, then if KVM is in use,
> > > we'll pick the highest GIC version QEMU reports as supported.
> >
> > You can get that without querying QEMU by asking for 'gic-version=max'
> > if you like.
>
> This isn't in the VM startup path. It is when we expand the user
> provided XML config into an ABI stable XML config by filling in
> the blanks left by the user. So we need to actually query the
> values available.
>
> > > If TCG
> > > is in use we'll always pick v2, even if QEMU reports v3 is emulatable
> > > due to the v3 impl lacking MSI controller which we need for PCI-e
> >
> > Our emulated GICv3 supports the ITS which has MSI support, so I'm not
> > sure what forcing GICv2 is getting you here. Looking at the linked
> > RHEL bugzilla bug, I suspect this is an out-of-date policy from before
> > we added the ITS emulation in 2021 (it's present by default
> > in virt-2.8 and later machine types). So that is something that
> > libvirt should update I think.
>
> It looks like it is virt-6.2 or later, and libvirt can probe it
> by looking for existence of the arm-gicv3-its QOM type IIUC.

Yes. (I was confused by the 'no_its' setting in virt-2.8, but
that is for "is there a KVM ITS", and no_tcg_its is separate
and as you say added with 6.2.)

-- PMM