Re: [Xen-devel] [RFC PATCH v1 10/10] xen/arm: GICv3 device tree parsing

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Sat, 2014-03-22 at 16:00 +0530, Vijay Kilari wrote:
> Hi,
> 
>  Julien,
> 
> On Thu, Mar 20, 2014 at 9:38 PM, Julien Grall <julien.grall@xxxxxxxxxx> wrote:
> > Hello Vijay,
> >
> > Thanks for the patch. I think the name of this patch is wrong. You are
> > not parsing the device tree but creating the node for DOM0.
> >
> OK
> 
> > On 03/19/2014 02:17 PM, vijay.kilari@xxxxxxxxx wrote:
> >> From: Vijaya Kumar K <Vijaya.Kumar@xxxxxxxxxxxxxxxxxx>
> >>
> >> GICv3 supports system register access to GIC cpu interface.
> >> So no need to read device tree for cpu interface.
> >> GICv3 adds Re-distributor region and redistributor stride
> >> which are parsed.
> >
> > If I understand correctly, DOM0 kernel must support GICv3, right? What
> > about booting DOM0 with only GICv2 support, e.g. because the kernel is
> > only using less than 8 CPUs?
> >
> Yes, correct.
> If DOM0 supports only V2, then Xen should load vgic v2 driver?
> if so, how to detect it?

There are four cases I suppose:

Two obvious ones:
HW: GICv2 + Xen: GICv2 + Dom0: GICv2
HW: GICv3 + Xen: GICv3 + Dom0: GICv3

Then:
HW: GICv3 + Xen: GICv2 (h/w compat mode) + Dom0: GICv2

Which I think we effectively aren't implementing, since we will always
use the GICv3 driver on GICv3 hardware. Or maybe this is just a case of
the first one above when the DT describes a GICv2 even though the h/w is
actually v3. I think we can ignore this case really.

Which leaves:

HW: GICv3 + Xen: GICv3 + Dom0: GICv2

Which would require extra support in Xen I suppose, to provide the v2
emulation in terms of v3?

There is also the question of what guests use, which I suppose it mostly
the same as dom0, but without the expectation that the guest kernel has
to boot natively on a specific platform.

I'm not sure what to suggest here, but I think for dom0 we can rely on
"dom0 kernel should be usable as a native kernel on this platform" and
we can punt for now on what to do about guests.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel