Re: [PATCH 2/2] mm/migration: Make rmap_walk_anon() andtry_to_unmap_anon() more scalable

[Mel Gorman]

On Tue, Dec 04, 2012 at 06:37:41AM -0800, Michel Lespinasse wrote:
> On Mon, Dec 3, 2012 at 6:17 AM, Mel Gorman <mgorman@xxxxxxx> wrote:
> > On Sat, Dec 01, 2012 at 09:15:38PM +0100, Ingo Molnar wrote:
> >> @@ -732,7 +732,7 @@ static int page_referenced_anon(struct p
> >>       struct anon_vma_chain *avc;
> >>       int referenced = 0;
> >>
> >> -     anon_vma = page_lock_anon_vma(page);
> >> +     anon_vma = page_lock_anon_vma_read(page);
> >>       if (!anon_vma)
> >>               return referenced;
> >
> > This is a slightly trickier one as this path is called from reclaim. It does
> > open the possibility that reclaim can stall something like a parallel fork
> > or anything that requires the anon_vma rwsem for a period of time. I very
> > severely doubt it'll really be a problem but keep an eye out for bug reports
> > related to delayed mmap/fork/anything_needing_write_lock during page reclaim.
> 
> I don't see why this would be a problem - rwsem does implement
> reader/writer fairness, so having some sites do a read lock instead of
> a write lock shouldn't cause the write lock sites to starve. Is this
> what you were worried about ?
> 

Yes. I did not expect they would be starved forever, just delayed longer
than they might have been before. I would be very surprised if there is
anything other than a synthetic case that will really care but I've been
"very surprised" before :)

-- 
Mel Gorman
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/