Re: [Qemu-ppc] PCI Express on pseries guests

[David Gibson]

On Wed, Sep 21, 2016 at 02:28:42PM +1000, Benjamin Herrenschmidt wrote:
> On Wed, 2016-09-21 at 13:35 +1000, David Gibson wrote:
> > So, the thing is that because the PCI bridge is para-virtualized on
> > the pseries platform, the distinction between PCI and PCI-E isn't
> > really meaningful.  That's why everything appears to be legacy PCI
> > within the guest.
> > 
> > BenH, can you correct me if I'm wrong here?
> 
> You are correct. On PAPR guests, the Root Complex isn't going to be
> visible, this is also true of the PowerVM hypervisor.

Note that this can give odd things like a PCI-E physical device
appearing on an apparently plain PCI bus (especially with VFIO).

It's weird, but since the hypercall interface works at a level where
the PCI and PCI-E are more or less the same, it works nonetheless.

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

signature.asc
Description: PGP signature