[Xen-devel] Re: VT-d scalability issue

["Zhai, Edwin" <edwin.zhai@xxxxxxxxx>]

On Tue, Sep 09, 2008 at 11:22:15AM +0100, Keir Fraser wrote:
> 
> 
> 
> On 9/9/08 10:04, "Zhai, Edwin" <edwin.zhai@xxxxxxxxx> wrote:
> 
> > following code in vmx_do_resume try to invalidate orig processor's cache 
> > when
> > 14 migration if this domain has pass-through device and no support for 
> > wbinvd
> > vmexit.          
> > 16 if ( has_arch_pdevs(v->domain) && !cpu_has_wbinvd_exiting )
> > {
> >     int cpu = v->arch.hvm_vmx.active_cpu;
> >     if ( cpu != -1 )
> >         on_selected_cpus(cpumask_of_cpu(cpu), wbinvd_ipi, NULL, 1,
> > 
> > }
> > 
> > So we want to pin vcpu to free processor for domains with pass-through 
> > device
> > in
> > creation process, just like what we did for NUMA system.
> 
> (a) pinning support already exists.Maybe list this as 'best practice' but I
> don't see any need for xend changes, for example.

So end user need explicitly call "xm vcpu-pin" for VTd domain. But where to put 
this 'best practice'?

> (b) presumably your upcoming (and existing current?) processors support
> wbinvd exiting anyway?

Yes, but exist system has such problem.

> 
>  -- Keir
> 
> 

-- 
best rgds,
edwin

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