On Fri, Dec 09, 2022 at 11:11:29AM +0100, Jan Beulich wrote: > On 09.12.2022 10:59, Roger Pau Monné wrote: > > On Thu, Dec 08, 2022 at 12:24:54PM +0100, Jan Beulich wrote: > >> --- a/xen/arch/x86/msr.c > >> +++ b/xen/arch/x86/msr.c > >> @@ -699,12 +699,16 @@ int guest_wrmsr(struct vcpu *v, uint32_t > >> } > >> else > > > > I think you could turn this into an `else if` and check if the new > > value and the current one differ on the SSBD bit? > > I'd prefer not to: Keeping it as I have it will likely reduce code churn > if a 2nd bit wants supporting in that MSR. > > > Provided it fixes the issue: > > > > Acked-by: Roger Pau Monné <roger....@citrix.com> > > Thanks, but I'm a little puzzled by the constraint: Imo even if this > doesn't address the observed issue, it still fixes one aspect of wrong > behavior here. The sole difference then would be that the Reported-by: > would go away.
Just wanted to make sure whether there was a further issue linked with this, in a way that we might need to change the fix. Maybe do the accounting in amd_set_legacy_ssbd() and keep track of each thread status. Thanks, Roger.
