vmscan: stop kswapd waiting on congestion when the min watermark is not being met

If reclaim fails to make sufficient progress, the priority is raised.
Once the priority is higher, kswapd starts waiting on congestion.
However, if the zone is below the min watermark then kswapd needs to
continue working without delay as there is a danger of an increased rate
of GFP_ATOMIC allocation failure.

This patch changes the conditions under which kswapd waits on congestion
by only going to sleep if the min watermarks are being met.

[mel@csn.ul.ie: add stats to track how relevant the logic is]
[mel@csn.ul.ie: make kswapd only check its own zones and rename the relevant counters]
Signed-off-by: NKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: NMel Gorman <mel@csn.ul.ie>
Reviewed-by: NRik van Riel <riel@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

include/linux/vmstat.h

@@ -40,7 +40,8 @@ enum vm_event_item { PGPGIN, PGPGOUT, PSWPIN, PSWPOUT,
 		PGSCAN_ZONE_RECLAIM_FAILED,
 #endif
 		PGINODESTEAL, SLABS_SCANNED, KSWAPD_STEAL, KSWAPD_INODESTEAL,
-		KSWAPD_PREMATURE_FAST, KSWAPD_PREMATURE_SLOW,
+		KSWAPD_LOW_WMARK_HIT_QUICKLY, KSWAPD_HIGH_WMARK_HIT_QUICKLY,
+		KSWAPD_SKIP_CONGESTION_WAIT,
 		PAGEOUTRUN, ALLOCSTALL, PGROTATED,
 #ifdef CONFIG_HUGETLB_PAGE
 		HTLB_BUDDY_PGALLOC, HTLB_BUDDY_PGALLOC_FAIL,

mm/vmscan.c

@@ -1905,19 +1905,25 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
 #endif
 
 /* is kswapd sleeping prematurely? */
-static int sleeping_prematurely(int order, long remaining)
+static int sleeping_prematurely(pg_data_t *pgdat, int order, long remaining)
 {
-	struct zone *zone;
+	int i;
 
 	/* If a direct reclaimer woke kswapd within HZ/10, it's premature */
 	if (remaining)
 		return 1;
 
 	/* If after HZ/10, a zone is below the high mark, it's premature */
-	for_each_populated_zone(zone)
+	for (i = 0; i < pgdat->nr_zones; i++) {
+		struct zone *zone = pgdat->node_zones + i;
+
+		if (!populated_zone(zone))
+			continue;
+
 		if (!zone_watermark_ok(zone, order, high_wmark_pages(zone),
 								0, 0))
 			return 1;
+	}
 
 	return 0;
 }
@@ -1979,6 +1985,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
 	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
 		int end_zone = 0;	/* Inclusive.  0 = ZONE_DMA */
 		unsigned long lru_pages = 0;
+		int has_under_min_watermark_zone = 0;
 
 		/* The swap token gets in the way of swapout... */
 		if (!priority)
@@ -2085,6 +2092,15 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
 			if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
 			    total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
 				sc.may_writepage = 1;
+
+			/*
+			 * We are still under min water mark. it mean we have
+			 * GFP_ATOMIC allocation failure risk. Hurry up!
+			 */
+			if (!zone_watermark_ok(zone, order, min_wmark_pages(zone),
+					      end_zone, 0))
+				has_under_min_watermark_zone = 1;
+
 		}
 		if (all_zones_ok)
 			break;		/* kswapd: all done */
@@ -2092,8 +2108,12 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
 		 * OK, kswapd is getting into trouble.  Take a nap, then take
 		 * another pass across the zones.
 		 */
-		if (total_scanned && priority < DEF_PRIORITY - 2)
-			congestion_wait(BLK_RW_ASYNC, HZ/10);
+		if (total_scanned && (priority < DEF_PRIORITY - 2)) {
+			if (has_under_min_watermark_zone)
+				count_vm_event(KSWAPD_SKIP_CONGESTION_WAIT);
+			else
+				congestion_wait(BLK_RW_ASYNC, HZ/10);
+		}
 
 		/*
 		 * We do this so kswapd doesn't build up large priorities for
@@ -2207,7 +2227,7 @@ static int kswapd(void *p)
 				long remaining = 0;
 
 				/* Try to sleep for a short interval */
-				if (!sleeping_prematurely(order, remaining)) {
+				if (!sleeping_prematurely(pgdat, order, remaining)) {
 					remaining = schedule_timeout(HZ/10);
 					finish_wait(&pgdat->kswapd_wait, &wait);
 					prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);
@@ -2218,13 +2238,13 @@ static int kswapd(void *p)
 				 * premature sleep. If not, then go fully
 				 * to sleep until explicitly woken up
 				 */
-				if (!sleeping_prematurely(order, remaining))
+				if (!sleeping_prematurely(pgdat, order, remaining))
 					schedule();
 				else {
 					if (remaining)
-						count_vm_event(KSWAPD_PREMATURE_FAST);
+						count_vm_event(KSWAPD_LOW_WMARK_HIT_QUICKLY);
 					else
-						count_vm_event(KSWAPD_PREMATURE_SLOW);
+						count_vm_event(KSWAPD_HIGH_WMARK_HIT_QUICKLY);
 				}
 			}
 

mm/vmstat.c

@@ -683,8 +683,9 @@ static const char * const vmstat_text[] = {
 	"slabs_scanned",
 	"kswapd_steal",
 	"kswapd_inodesteal",
-	"kswapd_slept_prematurely_fast",
-	"kswapd_slept_prematurely_slow",
+	"kswapd_low_wmark_hit_quickly",
+	"kswapd_high_wmark_hit_quickly",
+	"kswapd_skip_congestion_wait",
 	"pageoutrun",
 	"allocstall",