[RFC PATCH 2/2] ARM: SMP support for mach-virt

[Will Deacon]

On Tue, Dec 11, 2012 at 05:14:11PM +0000, Stefano Stabellini wrote:
> On Tue, 11 Dec 2012, Will Deacon wrote:
> > > > > On the Xen side there are not going to be any problems related to SMP
> > > > > and holding pens, because secondary CPUs only come into existence after
> > > > > the first CPU calls the HYPERVISOR_vcpu_op hypercall (still missing on
> > > > > ARM).
> > > > 
> > > > Ok. Would you be willing/able to wrap this in a PSCI interface when you do
> > > > add SMP support for ARM?
> > > 
> > > I grepped for PSCI in the Linux tree but I couldn't find anything.
> > > Also I googled for ARM PSCI and nothing of value came up. Can I find a
> > > doc that explains what PSCI is anywhere?
> > 
> > There is an initial version of the document here (requires registration):
> > 
> >   https://silver.arm.com/download/download.tm?pv=1303201
> > 
> > Note that there are some updates due to clarify things a bit in areas such
> > as function ID numbers and the PCS. Marc and I also plan to hack something up
> > in kvm/mach-virt for booting secondary CPUs.
> 
> Ideally the functionalities would be discoverable via DT, so that PSCI,
> PSCI-HVC and a paravirtualized Xen interface based on hypercalls and
> xenbus could all coexist without issues, reusing the same internal Linux
> APIs.

Agreed, I'll propose a binding once we have something off the ground.

Will