Re: [Qemu-devel] [Xen-devel] [PATCH][XSA-126] xen: limit guest control of PCI command register

[Michael S. Tsirkin]

On Wed, Jun 10, 2015 at 08:08:37AM +0100, Jan Beulich wrote:
> >>> On 08.06.15 at 11:30, <address@hidden> wrote:
> > What happens if you disable SERR# in the command register
> > of 83:00.1?
> 
> We've just been told that with SERR not enabled in any of the
> sibling endpoints the NMI still occurs. Not really surprising with
> us now assuming that it's the root port that generates the SERR
> in response to the UR coming back from an endpoint. But otoh
> in conflict with what we see in the ITP log (where SERR clearly
> is enabled on the endpoints, and the information we got says
> that it _is_ disabled, not that they had to do anything to disable
> it).
> 
> > 2.  Has a driver initialized this endpoint? What if you don't
> >     load a driver before sending the transaction resulting in the UR?
> 
> They now tried at least without loading a driver in Dom0, which
> didn't make a difference. Did you mean to also not load any driver
> in the guest? Otoh I can't really see what difference this makes,
> as the cleanup after the guest inside the hypervisor doesn't
> really depend much on whether it actively used any of the MSI-X
> entries.
> 
> Jan

I don't really know. The idea would be that device
is not designed for memory to be disabled when it's
active, and starts behaving in broken ways.

-- 
MST