Re: [Xen-ia64-devel] VHPT implementation issues

[Arun Sharma <arun.sharma@xxxxxxxxx>]

Magenheimer, Dan (HP Labs Fort Collins) wrote:
> > We first eliminated 1 and 3 as they have scalability issues on large 
> > scale SMP systems.
>
> I forgot to ask at the time...
>
> Could you explain the scalability issues more? 
I'm assuming that you'll implement collision chains. If you don't 
implement collision chains, then it becomes hard to provide the 
architectural guarantees for forward progress.
If you do implement collision chains, then you'll have to lock them and 
this might end up with too much serialization.
Also, how are per-cpu mappings in the guest going to be handled in a 
global VHPT?
I don't have hard data comparing the two approaches - but above are some 
of the reasons we used to rule it out.
        -Arun

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel