[patch] mm: memcg: shorten preempt-disabled section around eventchecks

[Johannes Weiner]

-rt ran into a problem with the soft limit spinlock inside the
non-preemptible section, because that is sleeping inside an atomic
context.  But I think it makes sense for vanilla, too, to keep the
non-preemptible section as short as possible.  Also, -3 lines.

Yong, Luis, could you add your Tested-bys?

---
Only the ratelimit checks themselves have to run with preemption
disabled, the resulting actions - checking for usage thresholds,
updating the soft limit tree - can and should run with preemption
enabled.

Signed-off-by: Johannes Weiner <jweiner@xxxxxxxxxx>
Reported-by: Yong Zhang <yong.zhang0@xxxxxxxxx>
Reported-by: Luis Henriques <henrix@xxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
---
 mm/memcontrol.c |   73 ++++++++++++++++++++++++++----------------------------
 1 files changed, 35 insertions(+), 38 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 6aff93c..8e62d3e 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -683,37 +683,32 @@ static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg,
 	return total;
 }
 
-static bool __memcg_event_check(struct mem_cgroup *memcg, int target)
+static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
+				       enum mem_cgroup_events_target target)
 {
 	unsigned long val, next;
 
 	val = __this_cpu_read(memcg->stat->events[MEM_CGROUP_EVENTS_COUNT]);
 	next = __this_cpu_read(memcg->stat->targets[target]);
 	/* from time_after() in jiffies.h */
-	return ((long)next - (long)val < 0);
-}
-
-static void __mem_cgroup_target_update(struct mem_cgroup *memcg, int target)
-{
-	unsigned long val, next;
-
-	val = __this_cpu_read(memcg->stat->events[MEM_CGROUP_EVENTS_COUNT]);
-
-	switch (target) {
-	case MEM_CGROUP_TARGET_THRESH:
-		next = val + THRESHOLDS_EVENTS_TARGET;
-		break;
-	case MEM_CGROUP_TARGET_SOFTLIMIT:
-		next = val + SOFTLIMIT_EVENTS_TARGET;
-		break;
-	case MEM_CGROUP_TARGET_NUMAINFO:
-		next = val + NUMAINFO_EVENTS_TARGET;
-		break;
-	default:
-		return;
+	if ((long)next - (long)val < 0) {
+		switch (target) {
+		case MEM_CGROUP_TARGET_THRESH:
+			next = val + THRESHOLDS_EVENTS_TARGET;
+			break;
+		case MEM_CGROUP_TARGET_SOFTLIMIT:
+			next = val + SOFTLIMIT_EVENTS_TARGET;
+			break;
+		case MEM_CGROUP_TARGET_NUMAINFO:
+			next = val + NUMAINFO_EVENTS_TARGET;
+			break;
+		default:
+			break;
+		}
+		__this_cpu_write(memcg->stat->targets[target], next);
+		return true;
 	}
-
-	__this_cpu_write(memcg->stat->targets[target], next);
+	return false;
 }
 
 /*
@@ -724,25 +719,27 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
 {
 	preempt_disable();
 	/* threshold event is triggered in finer grain than soft limit */
-	if (unlikely(__memcg_event_check(memcg, MEM_CGROUP_TARGET_THRESH))) {
+	if (unlikely(mem_cgroup_event_ratelimit(memcg,
+						MEM_CGROUP_TARGET_THRESH))) {
+		bool do_softlimit, do_numainfo;
+
+		do_softlimit = mem_cgroup_event_ratelimit(memcg,
+						MEM_CGROUP_TARGET_SOFTLIMIT);
+#if MAX_NUMNODES > 1
+		do_numainfo = mem_cgroup_event_ratelimit(memcg,
+						MEM_CGROUP_TARGET_NUMAINFO);
+#endif
+		preempt_enable();
+
 		mem_cgroup_threshold(memcg);
-		__mem_cgroup_target_update(memcg, MEM_CGROUP_TARGET_THRESH);
-		if (unlikely(__memcg_event_check(memcg,
-			     MEM_CGROUP_TARGET_SOFTLIMIT))) {
+		if (unlikely(do_softlimit))
 			mem_cgroup_update_tree(memcg, page);
-			__mem_cgroup_target_update(memcg,
-						   MEM_CGROUP_TARGET_SOFTLIMIT);
-		}
 #if MAX_NUMNODES > 1
-		if (unlikely(__memcg_event_check(memcg,
-			MEM_CGROUP_TARGET_NUMAINFO))) {
+		if (unlikely(do_numainfo))
 			atomic_inc(&memcg->numainfo_events);
-			__mem_cgroup_target_update(memcg,
-				MEM_CGROUP_TARGET_NUMAINFO);
-		}
 #endif
-	}
-	preempt_enable();
+	} else
+		preempt_enable();
 }
 
 static struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
--
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/