Re: [Xen-devel] Re: changeset 22526:7a5ee3800417

[George Dunlap <George.Dunlap@xxxxxxxxxxxxx>]

On Wed, Mar 9, 2011 at 8:44 AM, Tim Deegan <Tim.Deegan@xxxxxxxxxx> wrote:
> At 16:57 +0000 on 07 Mar (1299517021), George Dunlap wrote:
>> > Better to use old_entry.mfn, in the spirit of the original cset
>> > ("access-once semantics")?
>>
>> I started to do that, but the one below didn't have an old_entry
>> already.
>>
>> > In fact, I suspect that to be safe, you need
>> > to do an atomic RMW instead of just an atomic set, and then decide
>> > whether the VT-d tables will need to be synced.
>>
>> Are we not holding the p2m lock when writing entries?
>
> Good point. :)  I would prefer to use old_entry in both places anyway,
> just for consistency with the general approach of reading once.  It
> won't be any slower.

Sure, why not. :-)

>
> Is this patch intended for 4.1.0?

It looks like by accident, the bug will only cause a performance
degradation -- it will unnecessarily flush the vtd table even if the
mfn for a page isn't changing.  Presumably this will mainly have an
impact on guests which are migrating with PCI pass-through -- is that
even possible?

 -George

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel