Re: [Xen-devel] Proxying Hyper-V hypercalls from L2 to L0

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Tue, Dec 31, 2019 at 04:09:09PM +0000, Wei Liu wrote:
> On Mon, 30 Dec 2019 at 16:24, Wei Liu <wl@xxxxxxx> wrote:
> >
> > Hi all
> >
> > As much as I try to avoid writing code to proxy Hyper-V hypercalls, it
> > seems unavoidable for PV guests, because Hyper-V requires hypercalls
> > to be issued with CPL=0.
> >
> > This means for PV Dom0 I will need to add code in Xen to support
> > Hyper-V's ABIs, along with appropriate validations.
> >
> > How much do you care about running a PV Dom0 in this Xen on Hyper-V
> > setup? I personally would certainly go full on PVH if possible. :-)
> >
> 
> I tested PVH Dom0 in Xen on Hyper-V setup just now. Unfortunately
> Hyper-V doesn't support vIOMMU yet. Dom0 didn't get any interrupt in
> that setup.

Without DMA remapping I/O devices won't be able to get the real
addresses, and thus won't work correctly or at all. The only option in
such scenario lacking an IOMMU would be to build dom0 with an identity
memory map, but I would avoid going that route, as I think it's
going to cause more pain that just using a PV dom0.

Roger.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel