Re: [Xen-devel] [PATCH] x86/HVM: don't leak PFEC_implict to guests

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 06/04/17 15:59, Boris Ostrovsky wrote:
>> --- a/xen/arch/x86/hvm/svm/svm.c
>> +++ b/xen/arch/x86/hvm/svm/svm.c
>> @@ -269,13 +269,23 @@ static int svm_vmcb_restore(struct vcpu
>>      struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb;
>>      struct p2m_domain *p2m = p2m_get_hostp2m(v->domain);
>>  
>> -    if ( c->pending_valid &&
>> -         ((c->pending_type == 1) || (c->pending_type > 6) ||
>> -          (c->pending_reserved != 0)) )
>> +    if ( c->pending_valid )
>>      {
>> -        gdprintk(XENLOG_ERR, "Invalid pending event %#"PRIx32".\n",
>> -                 c->pending_event);
>> -        return -EINVAL;
>> +       if ( (c->pending_type == 1) || (c->pending_type > 6) ||
> Shouldn't this be >=5? The only valid types are 0, 2, 3 and 4.

That is a good point.  SVM has fewer valid types than VT-x.  I guess
this was a copy&paste mistake in the past.

~Andrew

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