Re: [patch 0/3] NUMA slab locking fixes

[Ravikiran G Thirumalai <kiran@scalex86.org>]

On Fri, Feb 03, 2006 at 03:06:16PM -0800, Christoph Lameter wrote:
> On Fri, 3 Feb 2006, Andrew Morton wrote:
> 
> > Could you please redo/retest these patches so that they apply on
> > 2.6.16-rc2.  Also, please arrange for any bugfixes to be staged ahead of
> > any optimisations - the optimisations we can defer until 2.6.17.
> 
> It seems that these two things are tightly connected. By changing the 
> locking he was able to address the hotplug breakage.

Yes,  they are.  The cachep->spinlock goes away at many places due to the 
colour_next patch and keeping the l3 around. (we should not have called that 
optimization I guess :)), and the irq disabling had to be moved to the 
l3 lock anyways.  Maybe I could have done away with patch 2 (merged it with 
patch3), but I thought keeping them seperate made it readable ....

Patchset against rc2 follows.

Thanks,
Kiran

PS: Many hotplug fixes are yet to be applied upstream I think.  I know
these slab patches work well with mm4 (with the x86_64 subtree and hotplug 
fixes in there) but I cannot really test cpu_up after cpu_down on rc2 
because it is broken as of now (pending merges, I guess).
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/