Re: [Qemu-devel] [PATCH] i386: Add support to get/set/migrate MSR (33H)

[Daniel P . Berrangé]

On Wed, Jul 04, 2018 at 04:43:27PM -0300, Eduardo Habkost wrote:
> On Wed, Jul 04, 2018 at 09:21:06PM +0800, Jingqi Liu wrote:
> > The MSR (33H) controls support for #AC exception
> > for split locked accesses. When bit 29 of the MSR (33H)
> > is set, the processor causes an #AC exception to
> > be issued instead of suppressing LOCK on bus
> > (during split lock access).
> > 
> > Signed-off-by: Jingqi Liu <address@hidden>
> [...]
> > diff --git a/target/i386/machine.c b/target/i386/machine.c
> > index 4d98d36..c82dc0d 100644
> > --- a/target/i386/machine.c
> > +++ b/target/i386/machine.c
> > @@ -935,6 +935,25 @@ static const VMStateDescription vmstate_msr_virt_ssbd 
> > = {
> >      }
> >  };
> >  
> > +static bool split_lock_ctrl_needed(void *opaque)
> > +{
> > +    X86CPU *cpu = opaque;
> > +    CPUX86State *env = &cpu->env;
> > +
> > +    return env->split_lock_ctrl != 0;
> > +}
> 
> Based on the Linux patch at [1], guests may try to detect the
> feature by writing to the MSR unconditionally.
> 
> If this happens, KVM needs to provide a mechanism to
> enable/disable the MSR emulation.  Otherwise users will end up
> with VMs that can't be migrated to older hosts even if they are
> using older machine-types.

Is there really no CPUID flag that can be used to detect the feature ?
Unconditionally probing for existance of arbitrary MSRs seems to be
just re-inventing CPUID feature detection, but worse because as you
say we need to now invent a way to control existance of individual
MSRs too :-(

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|