[PATCH] mm: refactoring inactive_list_is_low

[Minchan Kim]

Recently, Michal Hocko added tracepoint into inactive_list_is_low
for catching why VM decided to age the active list to know
active/inacive balancing problem. With that, unfortunately, it
added "bool trace" to inactlive_list_is_low to control some place
should be prohibited tracing. It is not elegant to me so this patch
try to clean it up.

Normally, most inactive_list_is_low is used for deciding active list
demotion but one site(i.e., get_scan_count) uses for other purpose
which reclaim file LRU forcefully. Sites for deactivation calls it
with shrink_active_list. It means inactive_list_is_low could be
located in shrink_active_list.

One more thing this patch does is to remove "ratio" in the tracepoint
because we can get it by post processing in script via simple math.

Signed-off-by: Minchan Kim <minchan@xxxxxxxxxx>
---
 include/trace/events/vmscan.h |  9 +++-----
 mm/vmscan.c                   | 51 ++++++++++++++++++++++++-------------------
 2 files changed, 31 insertions(+), 29 deletions(-)

diff --git a/include/trace/events/vmscan.h b/include/trace/events/vmscan.h
index 27e8a5c..406ea95 100644
--- a/include/trace/events/vmscan.h
+++ b/include/trace/events/vmscan.h
@@ -432,9 +432,9 @@ TRACE_EVENT(mm_vmscan_inactive_list_is_low,
 	TP_PROTO(int nid, int reclaim_idx,
 		unsigned long total_inactive, unsigned long inactive,
 		unsigned long total_active, unsigned long active,
-		unsigned long ratio, int file),
+		int file),
 
-	TP_ARGS(nid, reclaim_idx, total_inactive, inactive, total_active, active, ratio, file),
+	TP_ARGS(nid, reclaim_idx, total_inactive, inactive, total_active, active, file),
 
 	TP_STRUCT__entry(
 		__field(int, nid)
@@ -443,7 +443,6 @@ TRACE_EVENT(mm_vmscan_inactive_list_is_low,
 		__field(unsigned long, inactive)
 		__field(unsigned long, total_active)
 		__field(unsigned long, active)
-		__field(unsigned long, ratio)
 		__field(int, reclaim_flags)
 	),
 
@@ -454,16 +453,14 @@ TRACE_EVENT(mm_vmscan_inactive_list_is_low,
 		__entry->inactive = inactive;
 		__entry->total_active = total_active;
 		__entry->active = active;
-		__entry->ratio = ratio;
 		__entry->reclaim_flags = trace_shrink_flags(file) & RECLAIM_WB_LRU;
 	),
 
-	TP_printk("nid=%d reclaim_idx=%d total_inactive=%ld inactive=%ld total_active=%ld active=%ld ratio=%ld flags=%s",
+	TP_printk("nid=%d reclaim_idx=%d total_inactive=%ld inactive=%ld total_active=%ld active=%ld flags=%s",
 		__entry->nid,
 		__entry->reclaim_idx,
 		__entry->total_inactive, __entry->inactive,
 		__entry->total_active, __entry->active,
-		__entry->ratio,
 		show_reclaim_flags(__entry->reclaim_flags))
 );
 #endif /* _TRACE_VMSCAN_H */
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 75cdf68..6890c21 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -150,6 +150,7 @@ unsigned long vm_total_pages;
 
 static LIST_HEAD(shrinker_list);
 static DECLARE_RWSEM(shrinker_rwsem);
+static bool inactive_list_is_low(bool file, unsigned long, unsigned long);
 
 #ifdef CONFIG_MEMCG
 static bool global_reclaim(struct scan_control *sc)
@@ -1962,6 +1963,22 @@ static void shrink_active_list(unsigned long nr_to_scan,
 	isolate_mode_t isolate_mode = 0;
 	int file = is_file_lru(lru);
 	struct pglist_data *pgdat = lruvec_pgdat(lruvec);
+	unsigned long inactive, active;
+	enum lru_list inactive_lru = file * LRU_FILE;
+	enum lru_list active_lru = file * LRU_FILE + LRU_ACTIVE;
+	bool deactivate;
+
+	inactive = lruvec_lru_size_eligibe_zones(lruvec, file * LRU_FILE,
+					sc->reclaim_idx);
+	active = lruvec_lru_size_eligibe_zones(lruvec, file * LRU_FILE +
+					LRU_ACTIVE, sc->reclaim_idx);
+	deactivate = inactive_list_is_low(file, inactive, active);
+	trace_mm_vmscan_inactive_list_is_low(pgdat->node_id,
+			sc->reclaim_idx,
+			lruvec_lru_size(lruvec, inactive_lru), inactive,
+			lruvec_lru_size(lruvec, active_lru), active, file);
+	if (!deactivate)
+		return;
 
 	lru_add_drain();
 
@@ -2073,13 +2090,10 @@ static void shrink_active_list(unsigned long nr_to_scan,
  *    1TB     101        10GB
  *   10TB     320        32GB
  */
-static bool inactive_list_is_low(struct lruvec *lruvec, bool file,
-						struct scan_control *sc, bool trace)
+static bool inactive_list_is_low(bool file,
+			unsigned long inactive, unsigned long active)
 {
 	unsigned long inactive_ratio;
-	unsigned long inactive, active;
-	enum lru_list inactive_lru = file * LRU_FILE;
-	enum lru_list active_lru = file * LRU_FILE + LRU_ACTIVE;
 	unsigned long gb;
 
 	/*
@@ -2089,22 +2103,12 @@ static bool inactive_list_is_low(struct lruvec *lruvec, bool file,
 	if (!file && !total_swap_pages)
 		return false;
 
-	inactive = lruvec_lru_size_eligibe_zones(lruvec, inactive_lru, sc->reclaim_idx);
-	active = lruvec_lru_size_eligibe_zones(lruvec, active_lru, sc->reclaim_idx);
-
 	gb = (inactive + active) >> (30 - PAGE_SHIFT);
 	if (gb)
 		inactive_ratio = int_sqrt(10 * gb);
 	else
 		inactive_ratio = 1;
 
-	if (trace)
-		trace_mm_vmscan_inactive_list_is_low(lruvec_pgdat(lruvec)->node_id,
-				sc->reclaim_idx,
-				lruvec_lru_size(lruvec, inactive_lru), inactive,
-				lruvec_lru_size(lruvec, active_lru), active,
-				inactive_ratio, file);
-
 	return inactive * inactive_ratio < active;
 }
 
@@ -2112,8 +2116,7 @@ static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
 				 struct lruvec *lruvec, struct scan_control *sc)
 {
 	if (is_active_lru(lru)) {
-		if (inactive_list_is_low(lruvec, is_file_lru(lru), sc, true))
-			shrink_active_list(nr_to_scan, lruvec, sc, lru);
+		shrink_active_list(nr_to_scan, lruvec, sc, lru);
 		return 0;
 	}
 
@@ -2153,6 +2156,7 @@ static void get_scan_count(struct lruvec *lruvec, struct mem_cgroup *memcg,
 	enum lru_list lru;
 	bool some_scanned;
 	int pass;
+	unsigned long inactive, active;
 
 	/*
 	 * If the zone or memcg is small, nr[l] can be 0.  This
@@ -2243,7 +2247,11 @@ static void get_scan_count(struct lruvec *lruvec, struct mem_cgroup *memcg,
 	 * lruvec even if it has plenty of old anonymous pages unless the
 	 * system is under heavy pressure.
 	 */
-	if (!inactive_list_is_low(lruvec, true, sc, false) &&
+	inactive = lruvec_lru_size_eligibe_zones(lruvec,
+				LRU_FILE, sc->reclaim_idx);
+	active = lruvec_lru_size_eligibe_zones(lruvec,
+				LRU_FILE + LRU_ACTIVE, sc->reclaim_idx);
+	if (!inactive_list_is_low(true, inactive, active) &&
 	    lruvec_lru_size_eligibe_zones(lruvec, LRU_INACTIVE_FILE, sc->reclaim_idx) >> sc->priority) {
 		scan_balance = SCAN_FILE;
 		goto out;
@@ -2468,9 +2476,7 @@ static void shrink_node_memcg(struct pglist_data *pgdat, struct mem_cgroup *memc
 	 * Even if we did not try to evict anon pages at all, we want to
 	 * rebalance the anon lru active/inactive ratio.
 	 */
-	if (inactive_list_is_low(lruvec, false, sc, true))
-		shrink_active_list(SWAP_CLUSTER_MAX, lruvec,
-				   sc, LRU_ACTIVE_ANON);
+	shrink_active_list(SWAP_CLUSTER_MAX, lruvec, sc, LRU_ACTIVE_ANON);
 }
 
 /* Use reclaim/compaction for costly allocs or under memory pressure */
@@ -3118,8 +3124,7 @@ static void age_active_anon(struct pglist_data *pgdat,
 	do {
 		struct lruvec *lruvec = mem_cgroup_lruvec(pgdat, memcg);
 
-		if (inactive_list_is_low(lruvec, false, sc, true))
-			shrink_active_list(SWAP_CLUSTER_MAX, lruvec,
+		shrink_active_list(SWAP_CLUSTER_MAX, lruvec,
 					   sc, LRU_ACTIVE_ANON);
 
 		memcg = mem_cgroup_iter(NULL, memcg, NULL);
-- 
2.7.4