Re: Page Colouring

["David S. Miller" <davem@jenolan.rutgers.edu>]

We could pull this off, and I've been keeping this very issue in the
back of my head as I hack on Mark Hemment's changes.

Essentially you'd have to pass into the page allocation the
destination virtual address.  Then in the coloured allocator we put a
macro which can be defined to zero on sane architectures.

        if(page_color_verify(phys_page, virt_page)) {
                free_page(phys_page);
                goto repeat;
        }

Something like this, don't worry I'm thinking about it.

Later,
David "Sparc" Miller
davem@caip.rutgers.edu