Dave Hansen wrote: [Mon Mar 07 2005, 12:54:05PM EST]
> On Sun, 2005-03-06 at 10:57 -0500, Bob Picco wrote:
> > +void __init *alloc_remap(int nid, unsigned long size)
> > +{
> > + return vmem_map;
> > +}
> > +
>
> Oooh. That's nice. What else was required to get the NUMA configs to
> work?
>
> -- Dave
>
>
Nothing was required for DISCONTIGMEM NUMA. The SPARSEMEM NUMA issue we have
already addressed in other emails regarding conditional approach or weak
symbol. One caveat, I really only booted on a single node. I could boot
on simulated NUMA two node should you like. The simulated NUMA test has been
done before.
bob
