[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] PV-vNUMA issue: topology is misinterpreted by the guest

On Mon, 2015-07-27 at 11:49 +0100, Andrew Cooper wrote:
> On 27/07/15 11:41, George Dunlap wrote:

> > Can you expand a little on this?  I'm having trouble figuring out
> > exactly what user-space applications are reading and how they're using
> > it -- and, how they work currently in virtual environments, given that
> > they (typically) will be moved between physical processors even if
> > they stay on the same virtual processor.
> There are many examples of userspace application using cpuid to modify
> themselves.  Any serious application with processor optimisations will
> use the cpuid feature bits to choose the most efficient algorithm.
> hwloc is an perfect example which gathers all of the topology
> information out of cpuid to work out how to most efficiently
> pin/schedule tasks.
And all of this is broken, right now, isn't it?

Where, saying this, I'm aiming at stressing the point (as the thread is
starting to become a bit hard to follow) that we must come up with a
decent solution, working reasonably well in a bunch of (conflicting!
*sigh*) scenarios, *not* to implement something new without breaking
what we have, as what we have is broken already!

<<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)

Attachment: signature.asc
Description: This is a digitally signed message part

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.