Re: [Xen-devel] Live migration with MMIO pages

[Kieran Mansley <kmansley@xxxxxxxxxxxxxx>]

On Wed, 2007-10-31 at 16:44 +0000, Keir Fraser wrote:
> On 31/10/07 16:34, "Kieran Mansley" <kmansley@xxxxxxxxxxxxxx> wrote:
> 
> > Yes.  The reason it's failing is that sh_page_fault() in
> > xen/arch/x86/mm/shadow/multi.c thinks it's a bad gfn:
> > 
> >     if ( !p2m_is_valid(p2mt) || (!(p2m_is_mmio(p2mt) || mfn_valid
> > (gmfn))) )
> >     {
> >         perfc_incr(shadow_fault_bail_bad_gfn);
> >         SHADOW_PRINTK("BAD gfn=%"SH_PRI_gfn" gmfn=%"PRI_mfn"\n",
> >                       gfn_x(gfn), mfn_x(gmfn));
> >         goto not_a_shadow_fault;
> >     }
> > 
> > I think the problem is that set_mmio_p2m_entry() isn't getting called
> > when the IO mapping is established.  There are three places where
> > iomem_permit_access() is called:
> 
> No, basically that pagefault-handler check is nonsense for a PV guest. We
> don't have a p2m table in Xen for PV guests because they are not 'translated
> mode'. So there is nowhere for us to store the 'mmio' p2m type.
> 
> Perhaps Tim has a good idea what to do here. Adding a
> !shadow_mode_translate() condition to the if statement would probably work
> but I'm not sure it's the neatest answer.

Avoiding that if statement isn't sufficient to fix the problem - it
still doesn't seem to correctly re-enable writing to the page, even
though xenperf tells me that "calls to shadow fault" == "shadow_fault
fixed fault" (i.e. the code now thinks it's successfully done what it
should).

Kieran


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