Re: [PATCH v2 09/10] x86/SVM: Hook up miscellaneous AVIC functions

[Suravee Suthikulpanit <Suravee.Suthikulpanit@amd.com>]

On 01/10/2017 04:00 PM, Jan Beulich wrote:
>>>> On 10.01.17 at 09:35, <Suravee.Suthikulpanit@amd.com> wrote:
>> On 01/05/2017 11:05 PM, Jan Beulich wrote:
>>>>>> On 31.12.16 at 06:46, <suravee.suthikulpanit@amd.com> wrote:
>>>> --- a/xen/arch/x86/hvm/svm/svm.c
>>>> +++ b/xen/arch/x86/hvm/svm/svm.c
>>>> @@ -1438,6 +1438,11 @@ static int svm_cpu_up(void)
>>>>      return 0;
>>>>  }
>>>>
>>>> +static inline int svm_avic_enabled(void)
>>>
>>> bool?
>>
>> Actually, I declared this as int because the
>> hvm_function_table.virtual_intr_delivery_enabled() is returning int.
>
> Oh, that's used as a hook function. That's pretty un-obvious for a
> function declared inline. Of course in that case you need to match
> the hook function type, even if that ought to have return type bool.
> Even farther - I question the need for a function here in the first
> place, as both VMX and now SVM AVIC return a static value. This
> could therefore be a bool flag alongside the various others we
> already have near the beginning of the structure, if you're up to
> such a change. If you'd rather stick with what's there now, we can
> always put together a cleanup patch later on.
>
> Jan
>

Sure, I can incorporate the changes to make this a bool flag in this patch.

S

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