The patch titled
Subject: mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix
has been removed from the -mm tree. Its filename was
mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix.patch
This patch was dropped because it was folded into mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2.patch
------------------------------------------------------
From: Andrey Ryabinin <aryabinin@xxxxxxxxxxxxx>
Subject: mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix
On 04/06/2018 05:13 AM, Shakeel Butt wrote:
> Question: Should this 'flags' be per-node? Is it ok for a congested
> memcg to call wait_iff_congested for all nodes?
Indeed, congestion state should be per-node. If memcg on node A is
congested, there is no point is stalling memcg reclaim from node B.
Make congestion state per-cgroup-per-node and record it in
'struct mem_cgroup_per_node'.
Link: http://lkml.kernel.org/r/20180406135215.10057-1-aryabinin@xxxxxxxxxxxxx
Signed-off-by: Andrey Ryabinin <aryabinin@xxxxxxxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxxxxx>
Cc: Shakeel Butt <shakeelb@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
diff -puN include/linux/memcontrol.h~mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix include/linux/memcontrol.h
--- a/include/linux/memcontrol.h~mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix
+++ a/include/linux/memcontrol.h
@@ -120,6 +120,9 @@ struct mem_cgroup_per_node {
unsigned long usage_in_excess;/* Set to the value by which */
/* the soft limit is exceeded*/
bool on_tree;
+ bool congested; /* memcg has many dirty pages */
+ /* backed by a congested BDI */
+
struct mem_cgroup *memcg; /* Back pointer, we cannot */
/* use container_of */
};
@@ -189,8 +192,6 @@ struct mem_cgroup {
/* vmpressure notifications */
struct vmpressure vmpressure;
- unsigned long flags;
-
/*
* Should the accounting and control be hierarchical, per subtree?
*/
diff -puN mm/vmscan.c~mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix mm/vmscan.c
--- a/mm/vmscan.c~mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2-fix
+++ a/mm/vmscan.c
@@ -201,16 +201,27 @@ static bool sane_reclaim(struct scan_con
return false;
}
-static void set_memcg_bit(enum pgdat_flags flag,
- struct mem_cgroup *memcg)
+static void set_memcg_congestion(pg_data_t *pgdat,
+ struct mem_cgroup *memcg,
+ bool congested)
{
- set_bit(flag, &memcg->flags);
+ struct mem_cgroup_per_node *mz;
+
+ if (!memcg)
+ return;
+
+ mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+ WRITE_ONCE(mz->congested, congested);
}
-static int test_memcg_bit(enum pgdat_flags flag,
+static bool memcg_congested(pg_data_t *pgdat,
struct mem_cgroup *memcg)
{
- return test_bit(flag, &memcg->flags);
+ struct mem_cgroup_per_node *mz;
+
+ mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+ return READ_ONCE(mz->congested);
+
}
#else
static bool global_reclaim(struct scan_control *sc)
@@ -223,15 +234,16 @@ static bool sane_reclaim(struct scan_con
return true;
}
-static inline void set_memcg_bit(enum pgdat_flags flag,
- struct mem_cgroup *memcg)
+static inline void set_memcg_congestion(struct pglist_data *pgdat,
+ struct mem_cgroup *memcg, bool congested)
{
}
-static inline int test_memcg_bit(enum pgdat_flags flag,
- struct mem_cgroup *memcg)
+static inline bool memcg_congested(struct pglist_data *pgdat,
+ struct mem_cgroup *memcg)
{
- return 0;
+ return false;
+
}
#endif
@@ -2500,7 +2512,7 @@ static inline bool should_continue_recla
static bool pgdat_memcg_congested(pg_data_t *pgdat, struct mem_cgroup *memcg)
{
return test_bit(PGDAT_CONGESTED, &pgdat->flags) ||
- (memcg && test_memcg_bit(PGDAT_CONGESTED, memcg));
+ (memcg && memcg_congested(pgdat, memcg));
}
static bool shrink_node(pg_data_t *pgdat, struct scan_control *sc)
@@ -2635,7 +2647,7 @@ static bool shrink_node(pg_data_t *pgdat
*/
if (!global_reclaim(sc) && sane_reclaim(sc) &&
sc->nr.dirty && sc->nr.dirty == sc->nr.congested)
- set_memcg_bit(PGDAT_CONGESTED, root);
+ set_memcg_congestion(pgdat, root, true);
/*
* Stall direct reclaim for IO completions if underlying BDIs
@@ -2862,6 +2874,7 @@ retry:
continue;
last_pgdat = zone->zone_pgdat;
snapshot_refaults(sc->target_mem_cgroup, zone->zone_pgdat);
+ set_memcg_congestion(last_pgdat, sc->target_mem_cgroup, false);
}
delayacct_freepages_end();
@@ -3085,7 +3098,6 @@ unsigned long mem_cgroup_shrink_node(str
* the priority and make it zero.
*/
shrink_node_memcg(pgdat, memcg, &sc, &lru_pages);
- clear_bit(PGDAT_CONGESTED, &memcg->flags);
trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed);
@@ -3131,7 +3143,6 @@ unsigned long try_to_free_mem_cgroup_pag
noreclaim_flag = memalloc_noreclaim_save();
nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
memalloc_noreclaim_restore(noreclaim_flag);
- clear_bit(PGDAT_CONGESTED, &memcg->flags);
trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed);
_
Patches currently in -mm which might be from aryabinin@xxxxxxxxxxxxx are
mm-vmscan-update-stale-comments.patch
mm-vmscan-remove-redundant-current_may_throttle-check.patch
mm-vmscan-dont-change-pgdat-state-on-base-of-a-single-lru-list-state-v2.patch
mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v2.patch
mm-vmscan-dont-mess-with-pgdat-flags-in-memcg-reclaim-v3.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
