Re: [Xen-devel] [PATCH 3/4] x86/vvmx: correctly emulate VMWRITE

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 02/02/17 12:52, Jan Beulich wrote:
>>>> On 31.01.17 at 12:20, <sergey.dyasli@xxxxxxxxxx> wrote:
>> --- a/xen/arch/x86/hvm/vmx/vvmx.c
>> +++ b/xen/arch/x86/hvm/vmx/vvmx.c
>> @@ -264,7 +264,7 @@ u64 get_vvmcs_real(const struct vcpu *v, u32 encoding)
>>      return virtual_vmcs_vmread(v, encoding);
>>  }
>>  
>> -void set_vvmcs_virtual(void *vvmcs, u32 vmcs_encoding, u64 val)
>> +unsigned long set_vvmcs_virtual(void *vvmcs, u32 vmcs_encoding, u64 val)
>>  {
>>      union vmcs_encoding enc;
>>      u64 *content = (u64 *) vvmcs;
>> @@ -298,11 +298,13 @@ void set_vvmcs_virtual(void *vvmcs, u32 vmcs_encoding, 
>> u64 val)
>>      }
>>  
>>      content[offset] = res;
>> +
>> +    return 0;
>>  }
> Is it reasonable for this to never fail?

No.  Fields and values need to be audited against the features offered
to the guest in the VT-x MSRs.

However, it is not worth trying to fix that now before the MSR levelling
work is started, at which point we will have a cpuid_policy-like object
to query.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel