Re: [RFC PATCH 14/16]: PVH xen: add xenmem_add_foreign_to_pmap()

[Mukesh Rathor <mukesh.rathor@oracle.com>]

On Thu, 14 Feb 2013 10:39:27 +0000
Tim Deegan <tim@xen.org> wrote:

> At 18:18 -0800 on 11 Feb (1360606704), Mukesh Rathor wrote:
> > > > +        if (!is_hvm_domain(fdom)) {
> > > > +            printk("mmio type for non-hvm domain. fd:%d
> > > > fgmfn:%lx gpfn:%lx\n",
> > > > +                   foreign_domid, fgmfn, gpfn);
> > > > +            return -EINVAL;
> > > > +        }
> > > > +        mfn = fgmfn;     /* map 1 to 1 */
> > > 
> > > Surely not -- you want to map the _actual_ MMIO range, right, not
> > > just whatever GFN-address the foreigh domain mapped it at?
> > 
> > Actually, fgmfn here is the machine address of the mmio page. 
> 
> I hope not!  You've just passed it to a p2m lookup a little earlier,
> so it had better be a gfn.  (And in either case it could do with a
> more explicit name: maybe fmfn if it's an mfn or fgfn if it's a gfn?

Yup, already changed the name to fgfn last week.

>It may be that the memory is genuinely not there - since qemu doesn't
>build the guest itself it doesn't necessarily know exactly where the
>builder put all the memory.

>If that's the case, then just failing the mapping is the right
>response. 

Correct, and thats what PV dom0 does. I had it blatantly wrong! Fixed. 

Thanks,
Mukesh