Re: [patch] slub: default min_partial to at least highest cpus per node

From: David Rientjes
Date: Tue Apr 07 2009 - 18:04:42 EST


On Tue, 7 Apr 2009, Christoph Lameter wrote:

> > Right, but right now allocating new slabs is extremely slow compared to
> > using a partial slab.
>
> Work on the page allocator is in progress. So this is band aid until the
> work on the page allocator is complete. Maybe better focus on helping Mel?
>

The page allocator may be getting significantly faster, but that is _not_
the only reason why we use partial slabs. Higher order slabs, such as the
default for kmalloc-2048, kmalloc-4096, etc., increase the liklihood of
having cpu slabs at that order as opposed to order fallback with a
sufficiently long uptime.
--
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/