Re: [Xen-devel] [PATCH v4] x86/mm: Add mem access rights to NPT

[Isaila Alexandru <aisaila@xxxxxxxxxxxxxxx>]

On Wed, 2018-07-25 at 04:37 -0600, Jan Beulich wrote:
> > > > On 25.07.18 at 11:25, <george.dunlap@xxxxxxxxxx> wrote:
> > 
> > On 07/24/2018 01:02 PM, Jan Beulich wrote:
> > > > > > On 24.07.18 at 13:26, <george.dunlap@xxxxxxxxxx> wrote:
> > > > 
> > > > On 07/24/2018 09:55 AM, Jan Beulich wrote:
> > > > > > > > On 23.07.18 at 15:48, <aisaila@xxxxxxxxxxxxxxx> wrote:
> > > > > > 
> > > > > > +        {
> > > > > > +            xfree(d->arch.monitor.msr_bitmap);
> > > > > > +            return -ENOMEM;
> > > > > > +        }
> > > > > > +        radix_tree_init(p2m->mem_access_settings);
> > > > > > +    }
> > > > > 
> > > > > What's the SVM connection here? Please don't forget that p2m-
> > > > > pt.c
> > > > > also serves the shadow case. Perhaps struct p2m_domain should
> > > > > contain a boolean indicator whether this auxiliary data
> > > > > structure is
> > > > > needed?
> > > > 
> > > > It's basically just "hap_enabled()" isn't it?
> > > 
> > > Only if we can't make it there when EPT is active.
> > 
> > It can make it here when VMX is active and shadow is enabled, but
> > it
> > shouldn't be able to get here when EPT is active.  We could add an
> > ASSERT() to that effect; it should be safe in production, as the
> > only
> > side effect should be that we do a small pointless allocation.
> 
> So I've looked a little more closely: This is being added to
> arch_monitor_init_domain(), called from vm_event_domctl(). I can't
> see why this wouldn't be reachable with EPT enabled.
> 
Hi George, 

Any input on this?

Regards, 
Alex


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel