Re: [Qemu-devel] Re: [PATCH 15/15] Implement the bus structure for PAPR virtual IO

[Benjamin Herrenschmidt]

On Sun, 2011-02-13 at 10:48 -0600, Anthony Liguori wrote:
> 
> We try very, very hard to make our paravirtualization look like real 
> hardware.

Sure, that makes sense when you invent new paravirt interfaces, but that
isn't the case. Note also that our current processors do not have the
ability to emulate MMIOs in all cases, ie, when doing "real" KVM in HV
mode, we cannot trap MMIO unless we redirect all page faults to the
hypervisor, which comes at a cost.

> When the paravirtualization interfaces are already defined, we have no
> choice in supporting those interfaces although sometimes we try to
> push  that to firmware (like with Xenner).  It's better from a
> security PoV.
> 
> But in this case, we have no choice but to implement the 
> paravirtualization interfaces in QEMU because of the way the hardware 
> works and because these interfaces are already well defined.

Right.

Now, in the long run, we might decide to "reflect" some of these back
into some guest co-located FW or whatever of that kind, especially if we
get to a point where linux virt-io becomes more prevalent and the sPAPR
style IOs become purely legacy backward compat stubs, but we aren't
there yet.

Cheers,
Ben.