Re: [PATCH 2/11] Sched clock paravirt op

[Zachary Amsden <zach@vmware.com>]

Jeremy Fitzhardinge wrote:
> Zachary Amsden wrote:
>   
>> Jeremy Fitzhardinge wrote:
>>     
>>> Zachary Amsden wrote:
>>>  
>>>       
>>>> Scheduled (or available) time and real time are good notions.  Stolen
>>>> time is debatable.  But TSC is basically just always wrong.  That's
>>>> why I don't want to overload the rdtsc operation.     
>>>>         
>>> Well, in the Xen case it is actually guaranteed to be correct and useful
>>> as real time, but that's definitely not something we can expect in
>>> general.  But you're talking specifically about schedulable vcpu time
>>> here, right?
>>>   
>>>       
>> Not schedulable time, scheduled time (schedulable - scheduled) = stolen
>>     
>
> I meant "schedulable" (perhaps "usable" would be better) from the
> guest's perspective: total amount of real cpu time each vcpu gets.  ie:
> (real - schedulable) = stolen.  So I think we're talking about the same
> thing.
>   

Yes, I think so.  The point though, is the for Xen, TSC is real time.  
So it won't do for the scheduler, which must be schedulable time.

Zach