Re: [patch 15/21] multiple pte pointers per pte_chain

Jamie Lokier (lk@tantalophile.demon.co.uk)
Sun, 11 Aug 2002 15:51:22 +0100


Andrew Morton wrote:
> Pages which are mapped by only a single process continue to not have a
> pte_chain. The pointer in struct page points directly at the mapping
> pte (a "PageDirect" pte pointer). Once the page is shared a pte_chain
> is allocated and both the new and old pte pointers are moved into it.

May I suggest that the final pte in the list of ptes for a page is
_always_ pointed to directly?

In other words, a pte_chain looks like this:

struct page -> pte_chain -> pte_chain -> pte_chain -> pte
| | |
v v v
pte pte pte

pte_chain vs. pte would be distinguished by the least significant bit of
the pointer, or something similar.

This adds a special case in the list walker -- on the other hand, it
also removes two special cases ("PageDirect" is no longer required, and
there is no 0 to indicate end-of-list). But the best part is: it saves
more memory, has no cache line cost, and reduces the amount of work
needed to share a page :-)

-- Jamie

-
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/