Re: [Xen-devel] [PATCH v5 1/3] x86: Consolidate boolean inputs in hvm and p2m into a shared bitmap.

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 11.08.14 at 15:29, <tamas.lengyel@xxxxxxxxxxxx> wrote:
>> > --- a/xen/arch/x86/hvm/vmx/vmx.c
>> > +++ b/xen/arch/x86/hvm/vmx/vmx.c
>> > @@ -2353,6 +2353,11 @@ static void ept_handle_violation(unsigned long 
>> > qualification, paddr_t gpa)
>> >      p2m_type_t p2mt;
>> >      int ret;
>> >      struct domain *d = current->domain;
>> > +    struct mem_access_check npfec = {
>> > +        .read_access = !!(qualification & EPT_READ_VIOLATION),
>>
>> While I can see that you want this patch to not have any functional
>> change, I think correcting the mistake here (like I also did in the
>> patch that I sent you) would be rather desirable: Read-modify-write
>> instructions absolutely need to be treated as read accesses, yet
>> hardware doesn't guarantee to tell us so (they may surface as just
>> write accesses).
>>
> 
> I wasn't sure if it/s appropriate to include that fix in this patch and the
> other one you had on the AMD side for instruction/read accesses. Should I
> add another patch for that in this series or just include it in this one?

I'd certainly prefer it to be right here, with the functional change(s)
properly spelled out in the description.

Jan


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