Re: [PATCH 2/4] Add replace_page(), change the mapping of pte from one page into another

[Andrea Arcangeli <aarcange@redhat.com>]

On Wed, Nov 12, 2008 at 12:17:22AM +0100, Andrea Arcangeli wrote:
> We don't have to check the page_count vs mapcount later in
> replace_page because we know if anybody started an O_DIRECT read from
> disk, it would have triggered a cow, and the pte_same check that we
> have to do for other reasons would take care of bailing out the
> replace_page.

Ah, for completeness: above I didn't mention the case of O_DIRECT
writes to disk, because we never need to care about those. They're
never a problem. If the page is replaced and the cpu writes to the
page and by doing so triggers a cow that lead to the CPU write to go
in a different page (not the one under dma) it'll be like if the write
to disk completed before the cpu overwritten the page, so result is
undefined. I don't think we've to define the case of somebody doing a
direct read from a location where there's still an o_direct write in
flight either.