Re: [Xen-devel] [PATCH RFC v2 5/7] libxl/vNUMA: VM config parsing functions

[Elena Ufimtseva <ufimtseva@xxxxxxxxx>]

On Thu, Oct 10, 2013 at 4:38 AM, Dario Faggioli
<dario.faggioli@xxxxxxxxxx> wrote:
> On mer, 2013-10-09 at 14:45 -0400, Elena Ufimtseva wrote:
>> The other simplified way is to specify only two values. One for the
>> distance between same node and another between not the same nodes.
>> like this:
>>
>> vdistance = [10, 20]
>>
> Mmm... I like it...
>
>> and this will expand to
>>
>> 10 20 20 20
>> 20 10 20 20
>> 20 20 10 20
>> 20 20 20 10
>>
>> for 4 nodes.
>>
> Nice.
>
>> There is a loss in flexibility of course and other parsing can remain
>> the same or as George proposed, using lists.
>>
> Well, can't we support both? I mean, even if you configure 2 nodes you'd
> need 4 values, right? So, there are no cases where you only specify 2
> values.
>
> Well, let's make it like this: if you provide 2 values, it acts as you
> say above; otherwise you have to specify all of them. If you should have
> specified 4 (or 16) values, and you specify, say, 3 (or, say, 10) all
> the unspecified ones will have the same default value (e.g., 10). How do
> you like this?

Yes, two have both ways is better.

I would even add the third way - if the number of values is power of
number of nodes,
take them as it is and expand by rows (as it is right now).

so we have distances:

1) [10, 20] - only two values => same node 10, othes - 20;

2) did you mean, that if vnodes = 4, vdistance = [10, 10, 10]; the
rest of it should be 10?
Then it means all distances are the same?

3) if vdistance number of elements = vnodes * vnodes, take as it is.

Elena







>
> Dario
>
> --
> <<This happens because I choose it to happen!>> (Raistlin Majere)
> -----------------------------------------------------------------
> Dario Faggioli, Ph.D, http://about.me/dario.faggioli
> Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)
>



-- 
Elena

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