Re: [PATCH/RFC] Add node states sysfs class attributeS - V5

[mel@skynet.ie (Mel Gorman)]

On (15/09/07 00:00), Paul Mundt didst pronounce:
> On Fri, Sep 14, 2007 at 03:43:00PM +0100, Mel Gorman wrote:
> > On (14/09/07 03:50), Andrew Morton didst pronounce:
> > > So how do we get it tested with CONFIG_HIGHMEM=y?  Needs an i386
> > > numa machine, yes?  Perhaps Andy or Martin can remember to do this
> > > sometime, but they'll need a test plan ;)
> > 
> > As an aside, 32 Bit NUMA usually means we turn the NUMAQ into a whipping boy
> > and give the problem lip service. However, I'd be interested in hearing if
> > superh has dependencies on 32 bit NUMA working properly, including HIGHMEM
> > issues.
> > 
> > I've cc'd Paul Mundt. Paul, does superh use 32 bit NUMA? Is it used with
> > with HIGHMEM?
> > 
> We do use 32-bit NUMA, yes. Not with highmem at the moment, though.
> highmem support is something that will be coming soon, so the 32-bit NUMA
> + highmem assertion will be true on SH in the not-so-distant future.
> 
> This is something that quite a few CPUs and boards are depending on, and
> these are all using static sparsemem exclusively at present. It's also
> something that's tested with current git on a close to daily basis,
> albeit with page migration generally disabled, due to the small size of
> the non-system memory nodes.
> 

Ok, thanks Paul. It confirms that 32 bit NUMA is not something that will
disappear when the NUMAQs finally kick the bucket.

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>