From: Alexander Bluhm <bluhm@openbsd.org>
Subject: Re: SEV-ES guest: Only claim early GHCB pages when we actuall are SEV-ES guest
To: Mike Larkin <mlarkin@nested.page>
Cc: tech@openbsd.org
Date: Thu, 3 Jul 2025 12:56:01 +0200

On Wed, Jul 02, 2025 at 11:47:12AM -0700, Mike Larkin wrote:
> On Tue, Jul 01, 2025 at 10:37:57AM +0200, Hans-J?rg H?xer wrote:
> > Hi,
> >
> > as suggested by bluhm@, only claim GHCB pages when we are actually
> > running SEV-ES guest mode.
> >
> 
> I think this is overoptimization but ok mlarkin if this is the way
> you want to head.

When I was reading the code, I was wondering why first_avail is
incremented.  For me it is clearer if we put it within the
if(SEV_STAT_ES_ENABLED).

It is not about saving 8 KB.  For me the intension is clearer.  But
I don't insist on that if you have the opposit impression.

bluhm

> > ---
> > commit d2444d09c945baad0e5d9283a50d466d7de7b400
> > Author: Hans-Joerg Hoexer <hshoexer@genua.de>
> > Date:   Mon Jun 30 15:29:35 2025 +0200
> >
> >     SEV-ES guest: Only claim early GHCB pages when we actuall are SEV-ES guest
> >
> > diff --git a/sys/arch/amd64/amd64/machdep.c b/sys/arch/amd64/amd64/machdep.c
> > index 9b0badd610d..e0e8862d375 100644
> > --- a/sys/arch/amd64/amd64/machdep.c
> > +++ b/sys/arch/amd64/amd64/machdep.c
> > @@ -1470,9 +1470,10 @@ init_x86_64(paddr_t first_avail)
> >  	/*
> >  	 * locore0 mapped 2 pages for use as GHCB before pmap is initialized.
> >  	 */
> > -	if (ISSET(cpu_sev_guestmode, SEV_STAT_ES_ENABLED))
> > +	if (ISSET(cpu_sev_guestmode, SEV_STAT_ES_ENABLED)) {
> >  		cpu_init_early_vctrap(first_avail);
> > -	first_avail += 2 * NBPG;
> > +		first_avail += 2 * NBPG;
> > +	}
> >
> >  	/*
> >  	 * locore0 mapped 3 pages for use before the pmap is initialized
>