Re: [Xen-devel] [PATCH 01/10] VMX: Enable EPT A/D bit support

[Kai Huang <kaih.linux@xxxxxxxxx>]

On Fri, Apr 10, 2015 at 5:51 PM, Tim Deegan <tim@xxxxxxx> wrote:
> At 17:26 +0800 on 10 Apr (1428686789), Kai Huang wrote:
>>
>>
>> On 04/10/2015 04:54 PM, Tim Deegan wrote:
>> > At 14:40 +0800 on 10 Apr (1428676858), Kai Huang wrote:
>> >>> In fact, thinking about that, we should probably be setting _all_ the
>> >>> A bits to 1, even in leaf entries, to avoid the overhead of the MMU
>> >>> having to write them later.
>> >> I suppose you mean A bits of leaf entries of new shattered 4K pages? Yes
>> >> we can, but looks it's more reasonable to inherit according to original
>> >> A bit of super page.
>> > Yes, inheriting is best, but we should have already set the A bit of the
>> > superpage to 1 as well, for the same reasons.  IOW we should set every A
>> > bit in the whole system to 1 all the time.
>> >
>> > BTW, this is what we do with A/D bits in normal pagetables in Xen too
>> > -- the default is to set them all the time, so the MMU doesn't have to
>> > fix them up later.
>> Sure then we can set all A bit to 1, which can be done by setting A bit
>> properly in ept_p2m_type_to_flags I think.
>>
>> What does normal pagetables mean here? I don't think you mean shadow
>> page table, as shadow page table needs to sync A/D bits with guest
>> table, right?
>
> I meant xen's own pagetables, that it maps its own text and data with.
> But IIRC we also set the Accessed bit in all shadow pagetable entries
> -- fixing up the A/D bits in the guest PTEs happens in the pagefault
> handler.
Interesting. Thanks for pointing out. I should read that code when I
get spare time :)

Thanks,
-Kai
>
> Tim.
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel



-- 
Thanks,
-Kai

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