On Fri, Dec 14, 2012 at 4:07 AM, Michal Hocko [off-list ref] wrote:

On Thu 13-12-12 17:14:13, Ying Han wrote:
[...]

quoted

I haven't tried this patch set yet. Before I am doing that, I am
curious whether changing the target reclaim to be consistent with
global reclaim something worthy to consider based my last reply:

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 53dcde9..3f158c5 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c

@@ -1911,20 +1911,6 @@ static void shrink_zone(struct zone *zone,

struct scan_control *sc)

                shrink_lruvec(lruvec, sc);

-               /*
-                * Limit reclaim has historically picked one memcg and
-                * scanned it with decreasing priority levels until
-                * nr_to_reclaim had been reclaimed.  This priority
-                * cycle is thus over after a single memcg.
-                *
-                * Direct reclaim and kswapd, on the other hand, have
-                * to scan all memory cgroups to fulfill the overall
-                * scan target for the zone.
-                */
-               if (!global_reclaim(sc)) {
-                       mem_cgroup_iter_break(root, memcg);
-                       break;
-               }
                memcg = mem_cgroup_iter(root, memcg, &reclaim);

This wouldn't work because you would over-reclaim proportionally to the
number of groups in the hierarchy.

Don't get it and especially of why it is different from global
reclaim? I view the global reclaim should be viewed as target reclaim,
just a matter of the root cgroup is under memory pressure.

Anyway, don't want to distract you from working on the next post. So
feel free to not follow up on this.

--Ying

quoted

        } while (memcg);
 }

--Ying

--
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>