Re: ovmf / PCI passthrough impaired due to very limiting PCI64 aperture

[Dr. David Alan Gilbert]

* Gerd Hoffmann (kraxel@redhat.com) wrote:
>   Hi,
> 
> > > If we can somehow make a *trustable* physbits value available to the
> > > guest, then yes, we can go that route.  But the guest physbits we have
> > > today unfortunately don't cut it.
> > 
> > In downstream RH qemu, we run with host-physbits as default; so it's 
> > reasonably
> > trustworthy;
> 
> Can the guest figure somehow whenever it is trustworthy or not?

At any one point in time there may be things that it can try and see how
the CPU responds but I'm not 100% sure.
I know there are some bodges in to make some MSR values 1 padded by the
right amount when crossing sizes that generally work.
(were those PAM registers or something - vague memories of an old
bug)

> > of course that doesn't help you across a migration between
> > hosts with different sizes (e.g. an E5 Xeon to an E3).
> 
> Making physbits configurable for migration compatibility is fine if qemu
> outlaws the problematic guest physbits > host physbits case and throws
> an error in that case.

I'm not sure that guest < host is entirely safe either though; although
it seems to work.

Dave

> take care,
>   Gerd
> 
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK