Re: [Xen-devel] [PATCH 2/3] VMX: don't blindly enable descriptor table exiting control

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 13.04.17 at 17:16, <andrew.cooper3@xxxxxxxxxx> wrote:
> On 13/04/17 15:53, Jan Beulich wrote:
>> @@ -223,9 +223,15 @@ int arch_monitor_domctl_event(struct dom
>>          ad->monitor.descriptor_access_enabled = requested_status;
>>  
>>          for_each_vcpu ( d, v )
>> -            hvm_funcs.set_descriptor_access_exiting(v, requested_status);
>> +        {
>> +            ok = hvm_funcs.set_descriptor_access_exiting(v, 
>> requested_status);
>> +            if ( !ok )
>> +                break;
>> +        }
>>  
>>          domain_unpause(d);
>> +        if ( !ok )
>> +            return -EOPNOTSUPP;
> 
> While the overall purpose of the patch is clearly a good thing, this
> implementation isn't ideal.
> 
> In principle, this structure could hit a failure mid way through the
> vcpus, leaving them in an inconsistent state.
> 
> Instead, why not clobber the set_descriptor_access_exiting() function
> pointer if support isn't available, and use that (before the
> domain_pause()) to fail with EOPNOTSUPP?

Ah, yes, I'll do that. Don't know why it didn't occur to me right away
(as I didn't like this property of the code above either).

Jan


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