Re: Note describing poor dcache utilization under high memory pressure

Daniel Phillips (phillips@bonn-fries.net)
Tue, 29 Jan 2002 09:55:02 +0100


On January 29, 2002 09:39 am, Momchil Velikov wrote:
> >>>>> "Oliver" == Oliver Xymoron <oxymoron@waste.org> writes:
> Oliver> you can't actually _share_ the page tables without marking the pages
> Oliver> themselves readonly.
>
> Of course, ptes are made COW, just like now. Which brings up the
> question how much speedup we'll gain with a code that touches every
> single pte anyway ?

It's only touching the ptes on tables that are actually used, so if a parent
with a massive amount of mapped memory forks a child that only instantiates
a small portion of it (common situation) then the saving is pretty big.

Note that I'm not counting on this to be a huge performance win, except in
the specific case that that is bothering rmap. This is already worth the
price of admission.

This code might also be helpful if we want to get into swapping page tables
at some point, and it can also be pushed in the direction of sharing page
tables for mmaps. There it could be quite helpful, for example, with glibc.
(Thanks to Suparna for that observation.)

-- 
Daniel
-
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/