Re: [Xen-devel] [PATCH] x86/mm: relax the check in get_pg_owner

[Wei Liu <wei.liu2@xxxxxxxxxx>]

On Fri, Mar 24, 2017 at 05:17:51AM -0600, Jan Beulich wrote:
> >>> On 24.03.17 at 12:08, <wei.liu2@xxxxxxxxxx> wrote:
> > On Fri, Mar 24, 2017 at 02:12:47AM -0600, Jan Beulich wrote:
> >> >>> On 23.03.17 at 19:08, <wei.liu2@xxxxxxxxxx> wrote:
> >> > PVH guest is actually an translated guest. It should be able to
> >> > manipulate page table for other domains when acting as Dom0.
> >> 
> >> The same was true for PVHv1, so I'm afraid there's a little more to
> >> this.
> >> 
> >> > --- a/xen/arch/x86/mm.c
> >> > +++ b/xen/arch/x86/mm.c
> >> > @@ -3041,12 +3041,6 @@ static struct domain *get_pg_owner(domid_t domid)
> >> >          goto out;
> >> >      }
> >> >  
> >> > -    if ( unlikely(paging_mode_translate(curr)) )
> >> > -    {
> >> > -        MEM_LOG("Cannot mix foreign mappings with translated domains");
> >> > -        goto out;
> >> > -    }
> >> 
> >> Prior to Roger's recent removal of PVHv1 code this was
> >> 
> >>     if ( !is_pvh_domain(curr) && unlikely(paging_mode_translate(curr)) )
> >> 
> >> Instead of removing the left side, I think it should have been
> >> converted to !is_hvm_domain() (or is_pv_domain()).
> >> 
> > 
> > DYM:
> > 
> >     if ( !is_hvm_domain(curr) && unlikely(paging_mode_translate(curr)) )
> > 
> > ?
> 
> Yes.
> 
> > Here is something I'm not entirely sure of: is autotranslated PV guest
> > still a thing? I thought Andrew removed it already?
> 
> From the tools, iirc, but not from the hypervisor, at least not
> entirely (as you see from the change you want to make).
> 

OK, I will update the patch per your suggestion.

> Jan
> 

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