On Tue, 24 Feb 2015 15:19:28 -0300
Eduardo Habkost <ehabk...@redhat.com> wrote:
> On Tue, Feb 24, 2015 at 09:01:04AM +0100, Igor Mammedov wrote:
> > On Thu, 12 Feb 2015 15:50:35 -0200
> > Eduardo Habkost <ehabk...@redhat.com> wrote:
> >
> > > Instead of silently assigning CPU to node 0 when it is omitted from the
> > > command-line, check if all CPUs up to max_cpus are present in the NUMA
> > > configuration.
> > That would also trigger warning for possible (i.e. to be hotplugged) CPUs
> > as well.
>
> And that's correct, right? As far as I can see, there's no _PXM method
> in our Processor objects, so we need to know the NUMA node for all
> possible VCPUs in the moment the SRAT is generated. Or am I missing
> something?
Yep, that's how it's now. And probably it should stay so.
Just define topology at startup and there won't be warnings.
>
>
> [...]
> > > + if (!bitmap_full(seen_cpus, max_cpus)) {
> > > + char *msg;
> > > + bitmap_complement(seen_cpus, seen_cpus, max_cpus);
> > > + msg = enumerate_cpus(seen_cpus, max_cpus);
> > > + error_report("warning: CPU(s) not present in any NUMA nodes:
> > > %s", msg);
> > Maybe add to this that all CPUs up to maxcpus must be described in numa
> > config?
>
> I will update it. Thanks!
>
