Re: Found the commit that causes the OOMs

[Johannes Weiner]

On Sat, Jun 27, 2009 at 10:50:25PM +0900, Minchan Kim wrote:
> Hi, Hannes.
> 
> On Sat, Jun 27, 2009 at 9:54 PM, Johannes Weiner<hannes@xxxxxxxxxxx> wrote:
> > On Sat, Jun 27, 2009 at 08:12:49AM +0100, David Howells wrote:
> >>
> >> I've managed to bisect things to find the commit that causes the OOMs. ÂIt's:
> >>
> >> Â Â Â commit 69c854817566db82c362797b4a6521d0b00fe1d8
> >> Â Â Â Author: MinChan Kim <minchan.kim@xxxxxxxxx>
> >> Â Â Â Date: Â Tue Jun 16 15:32:44 2009 -0700
> >>
> >> Â Â Â Â Â vmscan: prevent shrinking of active anon lru list in case of no swap space V3
> >>
> >> Â Â Â Â Â shrink_zone() can deactivate active anon pages even if we don't have a
> >> Â Â Â Â Â swap device. ÂMany embedded products don't have a swap device. ÂSo the
> >> Â Â Â Â Â deactivation of anon pages is unnecessary.
> >>
> >> Â Â Â Â Â This patch prevents unnecessary deactivation of anon lru pages. ÂBut, it
> >> Â Â Â Â Â don't prevent aging of anon pages to swap out.
> >>
> >> Â Â Â Â Â Signed-off-by: Minchan Kim <minchan.kim@xxxxxxxxx>
> >> Â Â Â Â Â Acked-by: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx>
> >> Â Â Â Â Â Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
> >> Â Â Â Â Â Acked-by: Rik van Riel <riel@xxxxxxxxxx>
> >> Â Â Â Â Â Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> >> Â Â Â Â Â Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> >>
> >> This exhibits the problem. ÂThe previous commit:
> >>
> >> Â Â Â commit 35282a2de4e5e4e173ab61aa9d7015886021a821
> >> Â Â Â Author: Brice Goglin <Brice.Goglin@xxxxxxxxxxxx>
> >> Â Â Â Date: Â Tue Jun 16 15:32:43 2009 -0700
> >>
> >> Â Â Â Â Â migration: only migrate_prep() once per move_pages()
> >>
> >> survives 16 iterations of the LTP syscall testsuite without exhibiting the
> >> problem.
> >
> > Here is the patch in question:
> >
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index 7592d8e..879d034 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -1570,7 +1570,7 @@ static void shrink_zone(int priority, struct zone *zone,
> > Â Â Â Â * Even if we did not try to evict anon pages at all, we want to
> > Â Â Â Â * rebalance the anon lru active/inactive ratio.
> > Â Â Â Â */
> > - Â Â Â if (inactive_anon_is_low(zone, sc))
> > + Â Â Â if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0)
> > Â Â Â Â Â Â Â Âshrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);
> >
> > Â Â Â Âthrottle_vm_writeout(sc->gfp_mask);
> >
> > When this was discussed, I think we missed that nr_swap_pages can
> > actually get zero on swap systems as well and this should have been
> > total_swap_pages - otherwise we also stop balancing the two anon lists
> > when swap is _full_ which was not the intention of this change at all.
> 
> At that time we considered it so that we didn't prevent anon list
> aging for background reclaim.
> Do you think it is not enough ?

With a heavy multiprocess anon load, direct reclaimers will likely
reuse the reclaimed pages for anon mappings, so you have a handful of
processes shuffling pages on the active list and only one thread that
tries to balance.  I can imagine that it can not keep up for long.
--
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/