Re: [Xen-devel] [PATCH] VMX: XSA-60 workaround

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 20.08.13 at 16:27, Matt Wilson <msw@xxxxxxxxxx> wrote:
> On Tue, Aug 20, 2013 at 08:22:41AM +0100, Jan Beulich wrote:
>> >>> On 19.08.13 at 20:27, Matt Wilson <msw@xxxxxxxxxx> wrote:
>> > On Tue, Aug 13, 2013 at 05:36:17PM +0100, Jan Beulich wrote:
>> >> Considering that there's still no real progress towards a resolution
>> >> for XSA-60, I'd like to propose turning off the probelamtic code by
>> >> default, allowing it to be turned back on via command line option.
>> > 
>> > Apologies for a late reply, I've been on holiday for the past week.
>> > 
>> > I think it'd be really handy to make this a per-domain configuration,
>> > perhaps with a system-wide default set by boot command line.
>> 
>> Remember this is a temporary workaround, not a solution. I'm
>> not going to waste time trying to make the logic more sophisticated,
>> but I wouldn't mind if someone else took the patch further.
> 
> I'd like to help with a solution. What's the current thinking on the
> best way to handle the problem?

All that we seem to agree about with the Intel engineers is that the
current model of going through the entire EPT tables is wrong.
There hasn't been a good proposal for a solution yet; the one
route I proposed was confirmed to be disallowed in hyper-
threaded environments. In fact, it's not even clear whether the
problematic code wasn't in fact added "just in case", without
actually knowing of an environment where it would be needed
(GFX passthrough is the most likely candidate of a user of all this).

Jan


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