On Fri, Sep 04, 2026 at 12:15:26PM +0200, Philippe Mathieu-Daudé wrote:
> On 4/9/26 09:57, Bin Meng wrote:
> > Hi Philippe,
> > 
> > On Fri, Sep 4, 2026 at 12:56 AM Philippe Mathieu-Daudé
> > <[email protected]> wrote:
> > > 
> > > On 3/9/26 13:24, Bin Meng wrote:
> > > > Add the base topology for the Phytium Pi board built around the
> > > > E2000Q SoC, with two FTC310 and two FTC664 CPUs, RAM, GICv3/ITS,
> > > > and PL011 UARTs.
> > > > 
> > > > Describe the CPUs as three non-uniform clusters matching the vendor
> > > > Linux device tree. Preserve the board CPU slot order so firmware
> > > > MPIDR 0x200 maps to QEMU CPU index 2.
> > > > 
> > > > Signed-off-by: Bin Meng <[email protected]>
> > > > ---
> > > > 
> > > >    hw/arm/Kconfig         |   8 +
> > > >    hw/arm/meson.build     |   1 +
> > > >    hw/arm/phytium_e2000.c | 406 
> > > > +++++++++++++++++++++++++++++++++++++++++
> > > >    3 files changed, 415 insertions(+)
> > > >    create mode 100644 hw/arm/phytium_e2000.c
> 
> 
> > > > +static void phytium_pi_init(MachineState *ms)
> > > > +{
> > > > +    PhytiumE2000State *s = PHYTIUM_PI(ms);
> > > > +    int i;
> > > > +
> > > > +    if (kvm_enabled()) {
> > > > +        error_report("phytium-pi: KVM is not supported");
> > > 
> > > Why is it relevant?
> > 
> > This is because the machine relies on heterogeneous named
> > FTC310/FTC664 TCG CPU models and Arm KVM generally requires the host
> > CPU model. KVM is rejected explicitly instead of failing later during
> > vCPU initialization.

We have several other hardware accelerators for other platforms.
It is right to be checking KVM specifically, or should we have
general check that only passes for TCG/QTest ?

> Hmm OK. Cc'ing Pierrick to add to our list of things to think about for
> heterogenous emulation. I'd rather have this handled in a generic way,
> not per-machine. Hopefully TypeInfo::is_available will be sufficient
> to not expose the machine if KVM is not available. No need to think
> about that now.

We need to use QAPI MachineInfo / MachineClass to report to mgmt apps
whether a machine is capable of using hardware acceleration or not in
response to "query-machines".

Currently if a target supports HW accel, then apps assume that all
machines in that target can use acceleration. This was a convenient
short cut assumption, but this new machine suggests we can no make
do with that assumption, and need to explicitly report it per-machine.

With regards,
Daniel
-- 
|: https://berrange.com       ~~        https://hachyderm.io/@berrange :|
|: https://libvirt.org          ~~          https://entangle-photo.org :|
|: https://pixelfed.art/berrange   ~~    https://fstop138.berrange.com :|


Reply via email to