Re: Non-linear mappings and truncate/madvise(MADV_DONTNEED)

From: Ingo Molnar
Date: Thu Apr 22 2004 - 02:57:44 EST

On Mon, 19 Apr 2004, Hugh Dickins wrote:

> rmap 6 nonlinear truncation (which never appeared on LKML, though
> sent twice) fixed most of this, and went into 2.6.6-rc1-bk4 last
> night: please check it out.
> But I just converted madvise_dontneed by rote, adding a NULL arg to
> zap_page_range, missing your point that it should respect nonlinearity.
> And I made the zap_details structure private to mm/memory.c since I
> hadn't noticed anything outside needing it: I'll fix that up later and
> post a patch.
> I'm haven't and don't intend to change the behaviour of ->populate,
> without agreement from others - Ingo? Jamie?

feel free. I've got followup work, protection bits stored in the swap pte,
thus per-page protection possible via remap_file_pages_prot(). (earlier
-mm trees had this but it clashed with objrmap which has priority.)

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at