[libvirt] Problem of host CPU topology parsing

[Jiri Denemark]

On Fri, May 11, 2012 at 10:47:06 +0200, Michal Privoznik wrote:
> On 11.05.2012 10:40, Osier Yang wrote:
> >     /* nodeinfo->sockets is supposed to be a number of sockets per NUMA
> > node,
> >      * however if NUMA nodes are not composed of whole sockets, we just lie
> >      * about the number of NUMA nodes and force apps to check
> > capabilities XML
> >      * for the actual NUMA topology.
> >      */
> >     if (nodeinfo->sockets % nodeinfo->nodes == 0)
> >         nodeinfo->sockets /= nodeinfo->nodes;
> >     else
> >         nodeinfo->nodes = 1;
> > 
> > Jirka said this was for a fix, but I don't quite understand it,
> > what does the "nodeinfo.nodes" mean actually? Shouldn't it
> > be 8 (for the 48 CPUs machine) instead? But then we will be
> > wrong again with using VIR_NODEINFO_MAXCPUS.
> 
> Why do you think it will be wrong? My understanding is that
> VIR_NODEINFO_MAXCPUS just tell the max number of possible cpus not the
> actual. So if it's over 48 we are safe.

Not really, the macro should count exactly the number of CPUs available to
host, otherwise lots of other issues (incl. backward compatibility) appear. It
is just a badly named macro that should never exist but we can't do anything
with it since it is our public API.

> Btw: the code above seems like a hack to me.

Yes, it is a hack but it's unfortunately required because we can't change the
macro.

Anyway, I agree with Daniel that the bug most likely lies somewhere in the
code that populates nodeinfo structure.

Jirka