Re: [PATCH] Binding processes to selected CPUs

Andi Kleen (ak@muc.de)
07 Oct 1999 05:33:56 +0200


monnier+lists/linux/kernel/news/@tequila.cs.yale.edu (Stefan Monnier) writes:

> >>>>> "Harald" == Harald Koenig <koenig@tat.physik.uni-tuebingen.de> writes:
> > this array is only contigous in virtual address space.
> > but your physical pages are allocated randomly (more or less)
> > and it might even happen that all memory pages of this array
> > map to the _same_ cache lines (if your virtual array consists
> > of physical pages 128k apart or similar...).
>
> You mean that after the never-ending thread about page-coloring earlier this
> year, Linux still doesn't have any kind of page-coloring ?

Page colouring is hard to implement with the current buddy allocator, because
it would make the already bad fragmentation problems worse (and yes, it has
been tried). It probably needs a new zoned page allocator. So far nobody has
done this work.

-Andi

-- 
This is like TV. I don't like TV.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/