Am Wed, 2 Jun 2021 08:59:13 +0200
schrieb Juergen Gross <jgross@suse.com>:

> > @@ -152,9 +152,8 @@ int populate_pfns(struct xc_sr_context *ctx, unsigned int count,
> >   
> >       for ( i = 0; i < count; ++i )
> >       {
> > -        if ( (!types || (types &&
> > -                         (types[i] != XEN_DOMCTL_PFINFO_XTAB &&
> > -                          types[i] != XEN_DOMCTL_PFINFO_BROKEN))) &&
> > +        if ( (!types ||
> > +              (types && page_type_has_stream_data(types[i]) == true)) &&  
> 
> What about XEN_DOMCTL_PFINFO_XALLOC? Is this case impossible here, or
> are you changing behavior?

I guess this needs to be handled somehow, a large enough HVM domU will have XEN_DOMCTL_PFINFO_XALLOC in the stream. Not sure why this was thrown away with the v2 format.

Olaf