Re: [Xen-devel] [PATCH RFC] xen: allow guest_remove_page to remove p2m_mmio_direct pages

["Jan Beulich" <jbeulich@xxxxxxxx>]

>>> Tim Deegan <tim@xxxxxxx> 06/19/14 6:08 PM >>>
>At 16:51 +0100 on 19 Jun (1403193111), Jan Beulich wrote:
>> >>> Tim Deegan <tim@xxxxxxx> 06/19/14 1:08 PM >>>
>> >At 14:36 +0100 on 18 Jun (1403098594), Jan Beulich wrote:
>> >> >>> On 18.06.14 at 13:38, <roger.pau@xxxxxxxxxx> wrote:
>> >> Finally, how is this same situation going to be handled for PVH driver
>> >> domains or disaggregated PVH control domains? The
>> >> is_hardware_domain() here seems too much of a special case too...
>> >
>> >Agreed -- that is_hardware_domain() should be removed.  And with that
>> >removed, this patch has my ack.
>> 
>> I'm inclined to nak it with that part removed: Without it, the guest can get 
>> its
>> P2M out of sync wrt qemu's view, which might end up being a security issue.
>
>The guest can already do that (and I don't see that as a secutity
>issue anyway, since we treat qemu as under guest control).  This just
>suppresses a warning from attempting to free MMIO space.

That's true - I forgot for a moment that the patch is just about suppressing a
warning message, not about altering functionality as such.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel