Windows 8 will attempt to read MSR_IA32_THERM_STATUS and panic if a
#GP fault is injected as a result:

vmx.c:3035:d8v0 RDMSR 0x0000019c unimplemented
d8v0 VIRIDIAN CRASH: 3b c0000096 fffff8061de31651 fffff4088a613720 0

So handle the MSR and return 0 instead.

Note that this is done on the generic MSR handler, and PV guest will
also get 0 back when trying to read the MSR. There doesn't seem to be
much value in handling the MSR for HVM guests only.

Fixes: 84e848fd7a1 ('x86/hvm: disallow access to unknown MSRs')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
---
 xen/arch/x86/msr.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/xen/arch/x86/msr.c b/xen/arch/x86/msr.c
index e4c4fa6127..190d6ac6c5 100644
--- a/xen/arch/x86/msr.c
+++ b/xen/arch/x86/msr.c
@@ -253,6 +253,12 @@ int guest_rdmsr(struct vcpu *v, uint32_t msr, uint64_t *val)
             break;
         goto gp_fault;
 
+    case MSR_IA32_THERM_STATUS:
+        if ( cp->x86_vendor != X86_VENDOR_INTEL )
+            goto gp_fault;
+        *val = 0;
+        break;
+
     case MSR_X2APIC_FIRST ... MSR_X2APIC_LAST:
         if ( !is_hvm_domain(d) || v != curr )
             goto gp_fault;
-- 
2.28.0

On 07.10.2020 12:20, Roger Pau Monne wrote:
> --- a/xen/arch/x86/msr.c
> +++ b/xen/arch/x86/msr.c
> @@ -253,6 +253,12 @@ int guest_rdmsr(struct vcpu *v, uint32_t msr, uint64_t *val)
>              break;
>          goto gp_fault;
>  
> +    case MSR_IA32_THERM_STATUS:
> +        if ( cp->x86_vendor != X86_VENDOR_INTEL )
> +            goto gp_fault;
> +        *val = 0;
> +        break;

I've been puzzled while applying this: The upper patch context doesn't
match what's been in master for about the last month, and hence I
wonder what version of the tree you created this patch against. In any
event please double check that I didn't screw it up.

Jan

On Wed, Oct 14, 2020 at 02:17:15PM +0200, Jan Beulich wrote:
> On 07.10.2020 12:20, Roger Pau Monne wrote:
> > --- a/xen/arch/x86/msr.c
> > +++ b/xen/arch/x86/msr.c
> > @@ -253,6 +253,12 @@ int guest_rdmsr(struct vcpu *v, uint32_t msr, uint64_t *val)
> >              break;
> >          goto gp_fault;
> >  
> > +    case MSR_IA32_THERM_STATUS:
> > +        if ( cp->x86_vendor != X86_VENDOR_INTEL )
> > +            goto gp_fault;
> > +        *val = 0;
> > +        break;
> 
> I've been puzzled while applying this: The upper patch context doesn't
> match what's been in master for about the last month, and hence I
> wonder what version of the tree you created this patch against. In any
> event please double check that I didn't screw it up.

I had this applied on top of:

https://lore.kernel.org/xen-devel/20201006162327.93055-1-roger.pau@citrix.com/

Which I will reply to now because I'm not sure how to proceed there.

Thanks for fixing the context and applied.

Roger.

On 07.10.2020 12:20, Roger Pau Monne wrote:
> Windows 8 will attempt to read MSR_IA32_THERM_STATUS and panic if a
> #GP fault is injected as a result:
> 
> vmx.c:3035:d8v0 RDMSR 0x0000019c unimplemented
> d8v0 VIRIDIAN CRASH: 3b c0000096 fffff8061de31651 fffff4088a613720 0
> 
> So handle the MSR and return 0 instead.
> 
> Note that this is done on the generic MSR handler, and PV guest will
> also get 0 back when trying to read the MSR. There doesn't seem to be
> much value in handling the MSR for HVM guests only.
> 
> Fixes: 84e848fd7a1 ('x86/hvm: disallow access to unknown MSRs')
> Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>

Acked-by: Jan Beulich <jbeulich@suse.com>