Re: [Xen-devel] [PATCH RFC v1 45/74] x86/guest: use unpopulated memory to map the shared_info page

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Mon, Jan 08, 2018 at 04:03:50AM -0700, Jan Beulich wrote:
> >>> On 04.01.18 at 14:05, <wei.liu2@xxxxxxxxxx> wrote:
> > This prevents hardcoding a known unpopulated memory page to map
> > the shared info page. This fixes a TODO item in a previous patch.
> > 
> > Remove hypervisor_early_setup as now it is not required anymore.
> > 
> > Signed-off-by: Roger Pau Monne <roger.pau@xxxxxxxxxx>
> > Signed-off-by: Wei Liu <wei.liu2@xxxxxxxxxx>
> 
> Looks good, but one question:
> 
> > @@ -187,15 +186,12 @@ static void __init init_memmap(void)
> >      }
> >  }
> >  
> > -void __init hypervisor_early_setup(struct e820map *e820)
> > -{
> > -    map_shared_info(e820);
> > -}
> > -
> >  void __init hypervisor_setup(void)
> >  {
> >      init_memmap();
> >  
> > +    map_shared_info();
> > +
> >      init_evtchn();
> >  }
> 
> If the shared info page isn't needed as early, why was it set up
> that early originally?

Because during the setup of the shared_info the used memory address
would also be added to the e820 in order to mark it as RAM. I don't
think that was really required, since it's not needed to have a
page_info for the shared_info because it's not shared with the guest.

Roger.

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