Re: [PATCH v3] mm: make expand_downwards symmetrical toexpand_upwards

[Christoph Lameter]

On Wed, 20 Apr 2011, Matthew Wilcox wrote:

> On Wed, Apr 20, 2011 at 10:34:23AM +0300, Pekka Enberg wrote:
> > That part makes me think the best option is to make parisc do
> > CONFIG_NUMA as well regardless of the historical intent was.
>
> But it's not just parisc.  It's six other architectures as well, some
> of which aren't even SMP.  Does !SMP && NUMA make any kind of sense?

Of course not.

> I think really, this is just a giant horrible misunderstanding on the part
> of the MM people.  There's no reason why an ARM chip with 16MB of memory
> at 0 and 16MB of memory at 1GB should be saddled with all the NUMA gunk.

DISCONTIG has fallen out of favor in the last years. SPARSEMEM has largely
replaced it. ARM uses that and does not suffer from these issue.

No one considered the issues of having a !NUMA configuration with
nodes (which DISCONTIG seems to create) when developing core code in the
last years. The implicit assumption has always been that page_to_nid(x)
etc is always zero on a !NUMA configuration.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/