Re: [PATCH/RFC] Shared page tables

["Ray Bryant" <raybry@mpdtxmail.amd.com>]

On Monday 23 January 2006 18:39, Andi Kleen wrote:
> On Tuesday 24 January 2006 01:16, Ray Bryant wrote:
> > On Monday 23 January 2006 17:58, Ray Bryant wrote:
> > <snip>
> >
> > > ... And what kind of alignment constraints do we end up
> > > under in order to make the sharing happen?   (My guess would be that
> > > there aren't any such constraints (well, page alignment.. :-)  if we
> > > are just sharing pte's.)
> >
> > Oh, obviously that is not right as you have to share full pte pages.   So
> > on x86_64 I'm guessing one needs 2MB alignment in order to get the
> > sharing to kick in, since a pte page maps 512 pages of 4 KB each.
>
> The new randomized mmaps will likely actively sabotate such alignment. I
> just added them for x86-64.
>
> -Andi

Hmmm, does that mean there is a fundamental conflict between the desire to 
share pte's and getting good cache coloring behavior?

Isn't it the case that if the region is large enough (say >> 2MB), that 
randomized mmaps will just cause the first partial page of pte's to not be 
shareable, and as soon as we have a full pte page mapped into the file that 
the full pte pages will be shareable, etc, until the last (partial) pte page 
is not shareable?

> -
> 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/

-- 
Ray Bryant
AMD Performance Labs                   Austin, Tx
512-602-0038 (o)                 512-507-7807 (c)

-
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/