Balancing the scan rate of major caches V3

@@ -63,6 +63,9 @@ struct scan_control {
 
 	unsigned long nr_mapped;	/* From page_state */
 
+	/* How many pages shrink_cache() should reclaim */
+	int nr_to_reclaim;
+
 	/* Ask shrink_caches, or shrink_zone to scan at this priority */
 	unsigned int priority;
 

@@ -898,6 +901,7 @@ static void shrink_cache(struct zone *zo
 		if (current_is_kswapd())
 			mod_page_state(kswapd_steal, nr_freed);
 		mod_page_state_zone(zone, pgsteal, nr_freed);
+		sc->nr_to_reclaim -= nr_freed;
 
 		spin_lock_irq(&zone->lru_lock);
 		/*

@@ -1097,6 +1101,8 @@ shrink_zone(struct zone *zone, struct sc
 	else
 		nr_inactive = 0;
 
+	sc->nr_to_reclaim = sc->swap_cluster_max;
+
 	while (nr_active || nr_inactive) {
 		if (nr_active) {
 			sc->nr_to_scan = min(nr_active,

@@ -1110,6 +1116,8 @@ shrink_zone(struct zone *zone, struct sc
 					(unsigned long)sc->swap_cluster_max);
 			nr_inactive -= sc->nr_to_scan;
 			shrink_cache(zone, sc);
+			if (sc->nr_to_reclaim <= 0)
+				break;
 		}
 	}
 

@@ -1309,47 +1309,18 @@ loop_again:
 	}
 
 	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
-		int end_zone = 0;	/* Inclusive.  0 = ZONE_DMA */
 		unsigned long lru_pages = 0;
 
+		all_zones_ok = 1;
+		sc.nr_scanned = 0;
+		sc.nr_reclaimed = 0;
+		sc.priority = priority;
+		sc.swap_cluster_max = nr_pages ? nr_pages : SWAP_CLUSTER_MAX;
+
 		/* The swap token gets in the way of swapout... */
 		if (!priority)
 			disable_swap_token();
 
-		all_zones_ok = 1;
-
-		if (nr_pages == 0) {
-			/*
-			 * Scan in the highmem->dma direction for the highest
-			 * zone which needs scanning
-			 */
-			for (i = pgdat->nr_zones - 1; i >= 0; i--) {
-				struct zone *zone = pgdat->node_zones + i;
-
-				if (!populated_zone(zone))
-					continue;
-
-				if (zone->all_unreclaimable &&
-						priority != DEF_PRIORITY)
-					continue;
-
-				if (!zone_watermark_ok(zone, order,
-						zone->pages_high, 0, 0)) {
-					end_zone = i;
-					goto scan;
-				}
-			}
-			goto out;
-		} else {
-			end_zone = pgdat->nr_zones - 1;
-		}
-scan:
-		for (i = 0; i <= end_zone; i++) {
-			struct zone *zone = pgdat->node_zones + i;
-
-			lru_pages += zone->nr_active + zone->nr_inactive;
-		}
-
 		/*
 		 * Now scan the zone in the dma->highmem direction, stopping
 		 * at the last zone which needs scanning.

@@ -1359,51 +1330,49 @@ scan:
 		 * pages behind kswapd's direction of progress, which would
 		 * cause too much scanning of the lower zones.
 		 */
-		for (i = 0; i <= end_zone; i++) {
+		for (i = 0; i < pgdat->nr_zones; i++) {
 			struct zone *zone = pgdat->node_zones + i;
-			int nr_slab;
 
 			if (!populated_zone(zone))
 				continue;
 
+			if (nr_pages == 0) {	/* Not software suspend */
+				if (zone_watermark_ok(zone, order,
+					zone->pages_high, 0, 0))
+					continue;
+
+				all_zones_ok = 0;
+			}
+
 			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
 				continue;
 
-			if (nr_pages == 0) {	/* Not software suspend */
-				if (!zone_watermark_ok(zone, order,
-						zone->pages_high, end_zone, 0))
-					all_zones_ok = 0;
-			}
 			zone->temp_priority = priority;
 			if (zone->prev_priority > priority)
 				zone->prev_priority = priority;
-			sc.nr_scanned = 0;
-			sc.nr_reclaimed = 0;
-			sc.priority = priority;
-			sc.swap_cluster_max = nr_pages? nr_pages : SWAP_CLUSTER_MAX;
-			atomic_inc(&zone->reclaim_in_progress);
+			lru_pages += zone->nr_active + zone->nr_inactive;
+
 			shrink_zone(zone, &sc);
-			atomic_dec(&zone->reclaim_in_progress);
-			reclaim_state->reclaimed_slab = 0;
-			nr_slab = shrink_slab(sc.nr_scanned, GFP_KERNEL,
-						lru_pages);
-			sc.nr_reclaimed += reclaim_state->reclaimed_slab;
-			total_reclaimed += sc.nr_reclaimed;
-			total_scanned += sc.nr_scanned;
-			if (zone->all_unreclaimable)
-				continue;
-			if (nr_slab == 0 && zone->pages_scanned >=
+
+			if (zone->pages_scanned >=
 				    (zone->nr_active + zone->nr_inactive) * 4)
 				zone->all_unreclaimable = 1;
-			/*
-			 * If we've done a decent amount of scanning and
-			 * the reclaim ratio is low, start doing writepage
-			 * even in laptop mode
-			 */
-			if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
-			    total_scanned > total_reclaimed+total_reclaimed/2)
-				sc.may_writepage = 1;
 		}
+		reclaim_state->reclaimed_slab = 0;
+		shrink_slab(sc.nr_scanned, GFP_KERNEL, lru_pages);
+		sc.nr_reclaimed += reclaim_state->reclaimed_slab;
+		total_reclaimed += sc.nr_reclaimed;
+		total_scanned += sc.nr_scanned;
+
+		/*
+		 * If we've done a decent amount of scanning and
+		 * the reclaim ratio is low, start doing writepage
+		 * even in laptop mode
+		 */
+		if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
+		    total_scanned > total_reclaimed+total_reclaimed/2)
+			sc.may_writepage = 1;
+
 		if (nr_pages && to_free > total_reclaimed)
 			continue;	/* swsusp: need to do more work */
 		if (all_zones_ok)

@@ -1424,7 +1393,6 @@ scan:
 		if ((total_reclaimed >= SWAP_CLUSTER_MAX) && (!nr_pages))
 			break;
 	}
-out:
 	for (i = 0; i < pgdat->nr_zones; i++) {
 		struct zone *zone = pgdat->node_zones + i;
 

@@ -149,6 +149,20 @@ struct zone {
 	unsigned long		pages_scanned;	   /* since last reclaim */
 	int			all_unreclaimable; /* All pages pinned */
 
+	/* Fields for balanced page aging:
+	 * aging_total     - The accumulated number of activities that may
+	 *                   cause page aging, that is, make some pages closer
+	 *                   to the tail of inactive_list.
+	 * aging_milestone - A snapshot of total_scan every time a full
+	 *                   inactive_list of pages become aged.
+	 * page_age        - A normalized value showing the percent of pages
+	 *                   have been aged.  It is compared between zones to
+	 *                   balance the rate of page aging.
+	 */
+	unsigned long		aging_total;
+	unsigned long		aging_milestone;
+	unsigned long		page_age;
+
 	/*
 	 * Does the allocator try to reclaim pages from the zone as soon
 	 * as it fails a watermark_ok() in __alloc_pages?

60 lines not on this page · raw message

@@ -1522,6 +1522,8 @@ void show_free_areas(void)
 			" active:%lukB"
 			" inactive:%lukB"
 			" present:%lukB"
+			" aging:%lukB"
+			" age:%lu"
 			" pages_scanned:%lu"
 			" all_unreclaimable? %s"
 			"\n",

@@ -1533,6 +1535,8 @@ void show_free_areas(void)
 			K(zone->nr_active),
 			K(zone->nr_inactive),
 			K(zone->present_pages),
+			K(zone->aging_total),
+			zone->page_age,
 			zone->pages_scanned,
 			(zone->all_unreclaimable ? "yes" : "no")
 			);

@@ -2144,6 +2148,9 @@ static void __init free_area_init_core(s
 		zone->nr_scan_inactive = 0;
 		zone->nr_active = 0;
 		zone->nr_inactive = 0;
+		zone->aging_total = 0;
+		zone->aging_milestone = 0;
+		zone->page_age = 0;
 		atomic_set(&zone->reclaim_in_progress, 0);
 		if (!size)
 			continue;

@@ -2292,6 +2299,8 @@ static int zoneinfo_show(struct seq_file
 			   "\n        high     %lu"
 			   "\n        active   %lu"
 			   "\n        inactive %lu"
+			   "\n        aging    %lu"
+			   "\n        age      %lu"
 			   "\n        scanned  %lu (a: %lu i: %lu)"
 			   "\n        spanned  %lu"
 			   "\n        present  %lu",

@@ -2301,6 +2310,8 @@ static int zoneinfo_show(struct seq_file
 			   zone->pages_high,
 			   zone->nr_active,
 			   zone->nr_inactive,
+			   zone->aging_total,
+			   zone->page_age,
 			   zone->pages_scanned,
 			   zone->nr_scan_active, zone->nr_scan_inactive,
 			   zone->spanned_pages,

@@ -106,6 +106,12 @@ struct per_cpu_pageset {
  * ZONE_HIGHMEM	 > 896 MB	only page cache and user processes
  */
 
+struct aging {
+	unsigned long	life_span;
+	unsigned long	raw_age;
+	unsigned long	std_age;
+};
+
 struct zone {
 	/* Fields commonly accessed by the page allocator */
 	unsigned long		free_pages;

@@ -149,6 +155,8 @@ struct zone {
 	unsigned long		pages_scanned;	   /* since last reclaim */
 	int			all_unreclaimable; /* All pages pinned */
 
+	struct aging		aging;
+
 	/*
 	 * Does the allocator try to reclaim pages from the zone as soon
 	 * as it fails a watermark_ok() in __alloc_pages?

60 lines not on this page · raw message

@@ -1522,6 +1522,7 @@ void show_free_areas(void)
 			" active:%lukB"
 			" inactive:%lukB"
 			" present:%lukB"
+			" age:%lu"
 			" pages_scanned:%lu"
 			" all_unreclaimable? %s"
 			"\n",

@@ -1533,6 +1534,7 @@ void show_free_areas(void)
 			K(zone->nr_active),
 			K(zone->nr_inactive),
 			K(zone->present_pages),
+			zone->aging.std_age,
 			zone->pages_scanned,
 			(zone->all_unreclaimable ? "yes" : "no")
 			);

@@ -2144,6 +2146,7 @@ static void __init free_area_init_core(s
 		zone->nr_scan_inactive = 0;
 		zone->nr_active = 0;
 		zone->nr_inactive = 0;
+		memset(&zone->aging, 0, sizeof(struct aging));
 		atomic_set(&zone->reclaim_in_progress, 0);
 		if (!size)
 			continue;

@@ -2292,6 +2295,7 @@ static int zoneinfo_show(struct seq_file
 			   "\n        high     %lu"
 			   "\n        active   %lu"
 			   "\n        inactive %lu"
+			   "\n        age      %lu"
 			   "\n        scanned  %lu (a: %lu i: %lu)"
 			   "\n        spanned  %lu"
 			   "\n        present  %lu",

@@ -2301,6 +2305,7 @@ static int zoneinfo_show(struct seq_file
 			   zone->pages_high,
 			   zone->nr_active,
 			   zone->nr_inactive,
+			   zone->aging.std_age,
 			   zone->pages_scanned,
 			   zone->nr_scan_active, zone->nr_scan_inactive,
 			   zone->spanned_pages,

@@ -1194,6 +1194,7 @@ static void
 shrink_caches(struct zone **zones, struct scan_control *sc)
 {
 	int i;
+	struct zone *z = NULL;
 
 	for (i = 0; zones[i] != NULL; i++) {
 		struct zone *zone = zones[i];

@@ -1208,11 +1209,34 @@ shrink_caches(struct zone **zones, struc
 		if (zone->prev_priority > sc->priority)
 			zone->prev_priority = sc->priority;
 
-		if (zone->all_unreclaimable && sc->priority != DEF_PRIORITY)
+		if (zone->all_unreclaimable && sc->priority < DEF_PRIORITY)
 			continue;	/* Let kswapd poll it */
 
+		/*
+		 * Balance page aging in local zones and following headless
+		 * zones.
+		 */
+		if (sc->priority > DEF_PRIORITY) {
+			if (zone->zone_pgdat != zones[0]->zone_pgdat) {
+				cpumask_t cpu = node_to_cpumask(
+						zone->zone_pgdat->node_id);
+				if (!cpus_empty(cpu))
+					break;
+			}
+
+			if (!z)
+				z = zone;
+			else if (age_gt(z, zone))
+				z = zone;
+
+			continue;
+		}
+
 		shrink_zone(zone, sc);
 	}
+
+	if (z)
+		shrink_zone(z, sc);
 }
  
 /*

@@ -1256,7 +1280,8 @@ int try_to_free_pages(struct zone **zone
 		lru_pages += zone->nr_active + zone->nr_inactive;
 	}
 
-	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
+	/* The added 10 priorities are for scan rate balancing */
+	for (priority = DEF_PRIORITY + 10; priority >= 0; priority--) {
 		sc.nr_mapped = read_page_state(nr_mapped);
 		sc.nr_scanned = 0;
 		sc.nr_reclaimed = 0;

@@ -1290,7 +1315,7 @@ int try_to_free_pages(struct zone **zone
 		}
 
 		/* Take a nap, wait for some writeback to complete */
-		if (sc.nr_scanned && priority < DEF_PRIORITY - 2)
+		if (sc.nr_scanned && priority < DEF_PRIORITY)
 			blk_congestion_wait(WRITE, HZ/10);
 	}
 out:

@@ -1364,6 +1364,7 @@ static int balance_pgdat(pg_data_t *pgda
 	int total_scanned, total_reclaimed;
 	struct reclaim_state *reclaim_state = current->reclaim_state;
 	struct scan_control sc;
+	struct zone *prev_zone = pgdat->node_zones;
 
 loop_again:
 	total_scanned = 0;

@@ -1379,6 +1380,9 @@ loop_again:
 		struct zone *zone = pgdat->node_zones + i;
 
 		zone->temp_priority = DEF_PRIORITY;
+
+		if (populated_zone(zone))
+			prev_zone = zone;
 	}
 
 	for (priority = DEF_PRIORITY; priority >= 0; priority--) {

@@ -1409,14 +1413,34 @@ loop_again:
 			if (!populated_zone(zone))
 				continue;
 
-			if (nr_pages == 0) {	/* Not software suspend */
-				if (zone_watermark_ok(zone, order,
-					zone->pages_high, 0, 0))
-					continue;
+			if (nr_pages) 	/* software suspend */
+				goto scan_swspd;
 
-				all_zones_ok = 0;
+			if (zone_watermark_ok(zone, order,
+						zone->pages_high,
+						pgdat->nr_zones - 1, 0)) {
+				/* free pages enough, no reclaim */
+			} else if (zone < prev_zone) {
+				if (!zone_watermark_ok(zone, order,
+						zone->pages_high, 0, 0)) {
+					/* have to scan for free pages */
+					goto scan;
+				}
+				if (age_ge(prev_zone, zone)) {
+					/* catch up if falls behind */
+					goto scan;
+				}
+			} else if (!age_gt(zone, prev_zone)) {
+				/* catch up if falls behind or out of sync */
+				goto scan;
 			}
 
+			prev_zone = zone;
+			continue;
+scan:
+			prev_zone = zone;
+			all_zones_ok = 0;
+scan_swspd:
 			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
 				continue;
 

@@ -798,7 +798,9 @@ struct shrinker {
 	shrinker_t		shrinker;
 	struct list_head	list;
 	int			seeks;	/* seeks to recreate an obj */
-	long			nr;	/* objs pending delete */
+	unsigned long		aging_total;
+	unsigned long		aging_milestone;
+	unsigned long		page_age;
 	struct shrinker_stats	*s_stats;
 };
 

@@ -1080,8 +1082,7 @@ int in_gate_area_no_task(unsigned long a
 
 int drop_pagecache_sysctl_handler(struct ctl_table *, int, struct file *,
 					void __user *, size_t *, loff_t *);
-int shrink_slab(unsigned long scanned, gfp_t gfp_mask,
-			unsigned long lru_pages);
+int shrink_slab(struct zone *zone, int priority, gfp_t gfp_mask);
 
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */

198 lines not on this page · raw message

@@ -47,7 +47,7 @@ static void drop_slab(void)
 	int nr_objects;
 
 	do {
-		nr_objects = shrink_slab(1000, GFP_KERNEL, 1000);
+		nr_objects = shrink_slab(NULL, 0, GFP_KERNEL);
 	} while (nr_objects > 10);
 }
 

@@ -912,7 +912,7 @@ static void shrink_cache(struct zone *zo
 		int nr_scan;
 		int nr_freed;
 
-		nr_taken = isolate_lru_pages(sc->swap_cluster_max,
+		nr_taken = isolate_lru_pages(sc->nr_to_scan,
 					     &zone->inactive_list,
 					     &page_list, &nr_scan);
 		zone->nr_inactive -= nr_taken;

@@ -1106,56 +1106,56 @@ refill_inactive_zone(struct zone *zone, 
 
 /*
  * This is a basic per-zone page freer.  Used by both kswapd and direct reclaim.
+ * The reclaim process:
+ * a) scan always in batch of SWAP_CLUSTER_MAX pages
+ * b) scan inactive list at least one batch
+ * c) balance the scan rate of active/inactive list
+ * d) finish on either scanned or reclaimed enough pages
  */
 static void
 shrink_zone(struct zone *zone, struct scan_control *sc)
 {
+	unsigned long long next_scan_active;
 	unsigned long nr_active;
 	unsigned long nr_inactive;
 
 	atomic_inc(&zone->reclaim_in_progress);
 
+	next_scan_active = sc->nr_scanned;
+
 	/*
 	 * Add one to `nr_to_scan' just to make sure that the kernel will
 	 * slowly sift through the active list.
 	 */
-	zone->nr_scan_active += (zone->nr_active >> sc->priority) + 1;
-	nr_active = zone->nr_scan_active;
-	if (nr_active >= sc->swap_cluster_max)
-		zone->nr_scan_active = 0;
-	else
-		nr_active = 0;
-
-	zone->nr_scan_inactive += (zone->nr_inactive >> sc->priority) + 1;
-	nr_inactive = zone->nr_scan_inactive;
-	if (nr_inactive >= sc->swap_cluster_max)
-		zone->nr_scan_inactive = 0;
-	else
-		nr_inactive = 0;
+	nr_active = zone->nr_scan_active + 1;
+	nr_inactive = (zone->nr_inactive >> sc->priority) + SWAP_CLUSTER_MAX;
+	nr_inactive &= ~(SWAP_CLUSTER_MAX - 1);
 
+	sc->nr_to_scan = SWAP_CLUSTER_MAX;
 	sc->nr_to_reclaim = sc->swap_cluster_max;
 
-	while (nr_active || nr_inactive) {
-		if (nr_active) {
-			sc->nr_to_scan = min(nr_active,
-					(unsigned long)sc->swap_cluster_max);
-			nr_active -= sc->nr_to_scan;
+	while (nr_active >= SWAP_CLUSTER_MAX * 1024 || nr_inactive) {
+		if (nr_active >= SWAP_CLUSTER_MAX * 1024) {
+			nr_active -= SWAP_CLUSTER_MAX * 1024;
 			refill_inactive_zone(zone, sc);
 		}
 
 		if (nr_inactive) {
-			sc->nr_to_scan = min(nr_inactive,
-					(unsigned long)sc->swap_cluster_max);
-			nr_inactive -= sc->nr_to_scan;
+			nr_inactive -= SWAP_CLUSTER_MAX;
 			shrink_cache(zone, sc);
 			if (sc->nr_to_reclaim <= 0)
 				break;
 		}
 	}
 
-	throttle_vm_writeout();
+	next_scan_active = (sc->nr_scanned - next_scan_active) * 1024ULL *
+					(unsigned long long)zone->nr_active;
+	do_div(next_scan_active, zone->nr_inactive | 1);
+	zone->nr_scan_active = nr_active + (unsigned long)next_scan_active;
 
 	atomic_dec(&zone->reclaim_in_progress);
+
+	throttle_vm_writeout();
 }
 
 /*

@@ -1196,6 +1196,9 @@ shrink_caches(struct zone **zones, struc
 		if (zone->all_unreclaimable && sc->priority < DEF_PRIORITY)
 			continue;	/* Let kswapd poll it */
 
+		if (atomic_read(&zone->reclaim_in_progress))
+			continue;
+
 		/*
 		 * Balance page aging in local zones and following headless
 		 * zones.

@@ -1425,6 +1428,9 @@ scan_swspd:
 			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
 				continue;
 
+			if (atomic_read(&zone->reclaim_in_progress))
+				continue;
+
 			zone->temp_priority = priority;
 			if (zone->prev_priority > priority)
 				zone->prev_priority = priority;

@@ -2145,7 +2145,6 @@ static void __init free_area_init_core(s
 		INIT_LIST_HEAD(&zone->active_list);
 		INIT_LIST_HEAD(&zone->inactive_list);
 		zone->nr_scan_active = 0;
-		zone->nr_scan_inactive = 0;
 		zone->nr_active = 0;
 		zone->nr_inactive = 0;
 		zone->aging_total = 0;

@@ -2301,7 +2300,7 @@ static int zoneinfo_show(struct seq_file
 			   "\n        inactive %lu"
 			   "\n        aging    %lu"
 			   "\n        age      %lu"
-			   "\n        scanned  %lu (a: %lu i: %lu)"
+			   "\n        scanned  %lu (a: %lu)"
 			   "\n        spanned  %lu"
 			   "\n        present  %lu",
 			   zone->free_pages,

@@ -2313,7 +2312,7 @@ static int zoneinfo_show(struct seq_file
 			   zone->aging_total,
 			   zone->page_age,
 			   zone->pages_scanned,
-			   zone->nr_scan_active, zone->nr_scan_inactive,
+			   zone->nr_scan_active / 1024,
 			   zone->spanned_pages,
 			   zone->present_pages);
 		seq_printf(m,

@@ -111,7 +111,7 @@ enum {
 	SWP_SCANNING	= (1 << 8),	/* refcount in scan_swap_map */
 };
 
-#define SWAP_CLUSTER_MAX 32
+#define SWAP_CLUSTER_MAX 32		/* must be power of 2 */
 
 #define SWAP_MAP_MAX	0x7fff
 #define SWAP_MAP_BAD	0x8000

@@ -142,8 +142,7 @@ struct zone {
 	spinlock_t		lru_lock;	
 	struct list_head	active_list;
 	struct list_head	inactive_list;
-	unsigned long		nr_scan_active;
-	unsigned long		nr_scan_inactive;
+	unsigned long		nr_scan_active;	/* x1024 to be more precise */
 	unsigned long		nr_active;
 	unsigned long		nr_inactive;
 	unsigned long		pages_scanned;	   /* since last reclaim */

@@ -251,10 +251,13 @@ struct zone {
 
 /*
  * The "priority" of VM scanning is how much of the queues we will scan in one
- * go. A value of 12 for DEF_PRIORITY implies that we will scan 1/4096th of the
- * queues ("queue_length >> 12") during an aging round.
+ * go. A value of 12 for DEF_PRIORITY/PRIORITY_STEPS implies that we will
+ * scan 1/4096th of the queues ("queue_length >> 12") during an aging round.
+ * Typically we will first try to scan 1/4096, then 64 times 1/2048, then 64
+ * times 1/1024, ..., at last 64 times 1/64.
  */
-#define DEF_PRIORITY 12
+#define PRIORITY_STEPS	64
+#define DEF_PRIORITY	(12*PRIORITY_STEPS)
 
 /*
  * One allocation request operates on a zonelist. A zonelist

@@ -1006,7 +1006,7 @@ refill_inactive_zone(struct zone *zone, 
 	 * `distress' is a measure of how much trouble we're having reclaiming
 	 * pages.  0 -> no problems.  100 -> great trouble.
 	 */
-	distress = 100 >> zone->prev_priority;
+	distress = 100 >> (zone->prev_priority / PRIORITY_STEPS);
 
 	/*
 	 * The point of this algorithm is to decide when to start reclaiming

@@ -1128,7 +1128,8 @@ shrink_zone(struct zone *zone, struct sc
 	 * slowly sift through the active list.
 	 */
 	nr_active = zone->nr_scan_active + 1;
-	nr_inactive = (zone->nr_inactive >> sc->priority) + SWAP_CLUSTER_MAX;
+	nr_inactive = ((zone->nr_inactive / PRIORITY_STEPS) >>
+			(sc->priority / PRIORITY_STEPS)) + SWAP_CLUSTER_MAX;
 	nr_inactive &= ~(SWAP_CLUSTER_MAX - 1);
 
 	sc->nr_to_scan = SWAP_CLUSTER_MAX;

@@ -1265,8 +1266,13 @@ int try_to_free_pages(struct zone **zone
 		zone->temp_priority = DEF_PRIORITY;
 	}
 
-	/* The added 10 priorities are for scan rate balancing */
-	for (priority = DEF_PRIORITY + 10; priority >= 0; priority--) {
+	/*
+	 * The first PRIORITY_STEPS priorities are for scan rate balancing.
+	 * One run of shrink_zone() can create at most 1/64 imbalance, here
+	 * we first scan about 64 times 1/4096 for aging, just enough to
+	 * rebalance it, before creating new imbalance.
+	 */
+	for (priority = DEF_PRIORITY + PRIORITY_STEPS; priority >= 0; priority--) {
 		sc.nr_mapped = read_page_state(nr_mapped);
 		sc.nr_scanned = 0;
 		sc.nr_reclaimed = 0;

@@ -1301,7 +1307,7 @@ int try_to_free_pages(struct zone **zone
 		}
 
 		/* Take a nap, wait for some writeback to complete */
-		if (sc.nr_scanned && priority < DEF_PRIORITY)
+		if (sc.nr_scanned && priority < DEF_PRIORITY - PRIORITY_STEPS)
 			blk_congestion_wait(WRITE, HZ/10);
 	}
 out:

@@ -1464,7 +1470,7 @@ scan_swspd:
 		 * OK, kswapd is getting into trouble.  Take a nap, then take
 		 * another pass across the zones.
 		 */
-		if (total_scanned && priority < DEF_PRIORITY - 2)
+		if (total_scanned && priority < DEF_PRIORITY - PRIORITY_STEPS)
 			blk_congestion_wait(WRITE, HZ/10);
 
 		/*

@@ -900,63 +900,58 @@ static void shrink_cache(struct zone *zo
 {
 	LIST_HEAD(page_list);
 	struct pagevec pvec;
-	int max_scan = sc->nr_to_scan;
+	struct page *page;
+	int nr_taken;
+	int nr_scan;
+	int nr_freed;
 
 	pagevec_init(&pvec, 1);
 
 	lru_add_drain();
 	spin_lock_irq(&zone->lru_lock);
-	while (max_scan > 0) {
-		struct page *page;
-		int nr_taken;
-		int nr_scan;
-		int nr_freed;
-
-		nr_taken = isolate_lru_pages(sc->nr_to_scan,
-					     &zone->inactive_list,
-					     &page_list, &nr_scan);
-		zone->nr_inactive -= nr_taken;
-		zone->pages_scanned += nr_scan;
-		update_zone_age(zone, nr_scan);
-		spin_unlock_irq(&zone->lru_lock);
+	nr_taken = isolate_lru_pages(sc->nr_to_scan,
+				     &zone->inactive_list,
+				     &page_list, &nr_scan);
+	zone->nr_inactive -= nr_taken;
+	zone->pages_scanned += nr_scan;
+	update_zone_age(zone, nr_scan);
+	spin_unlock_irq(&zone->lru_lock);
 
-		if (nr_taken == 0)
-			goto done;
+	if (nr_taken == 0)
+		return;
 
-		max_scan -= nr_scan;
-		sc->nr_scanned += nr_scan;
-		if (current_is_kswapd())
-			mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
-		else
-			mod_page_state_zone(zone, pgscan_direct, nr_scan);
-		nr_freed = shrink_list(&page_list, sc);
-		if (current_is_kswapd())
-			mod_page_state(kswapd_steal, nr_freed);
-		mod_page_state_zone(zone, pgsteal, nr_freed);
-		sc->nr_to_reclaim -= nr_freed;
+	sc->nr_scanned += nr_scan;
+	if (current_is_kswapd())
+		mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
+	else
+		mod_page_state_zone(zone, pgscan_direct, nr_scan);
+	nr_freed = shrink_list(&page_list, sc);
+	if (current_is_kswapd())
+		mod_page_state(kswapd_steal, nr_freed);
+	mod_page_state_zone(zone, pgsteal, nr_freed);
+	sc->nr_to_reclaim -= nr_freed;
 
-		spin_lock_irq(&zone->lru_lock);
-		/*
-		 * Put back any unfreeable pages.
-		 */
-		while (!list_empty(&page_list)) {
-			page = lru_to_page(&page_list);
-			if (TestSetPageLRU(page))
-				BUG();
-			list_del(&page->lru);
-			if (PageActive(page))
-				add_page_to_active_list(zone, page);
-			else
-				add_page_to_inactive_list(zone, page);
-			if (!pagevec_add(&pvec, page)) {
-				spin_unlock_irq(&zone->lru_lock);
-				__pagevec_release(&pvec);
-				spin_lock_irq(&zone->lru_lock);
-			}
+	spin_lock_irq(&zone->lru_lock);
+	/*
+	 * Put back any unfreeable pages.
+	 */
+	while (!list_empty(&page_list)) {
+		page = lru_to_page(&page_list);
+		if (TestSetPageLRU(page))
+			BUG();
+		list_del(&page->lru);
+		if (PageActive(page))
+			add_page_to_active_list(zone, page);
+		else
+			add_page_to_inactive_list(zone, page);
+		if (!pagevec_add(&pvec, page)) {
+			spin_unlock_irq(&zone->lru_lock);
+			__pagevec_release(&pvec);
+			spin_lock_irq(&zone->lru_lock);
 		}
-  	}
+	}
 	spin_unlock_irq(&zone->lru_lock);
-done:
+
 	pagevec_release(&pvec);
 }
 

@@ -983,7 +978,6 @@ refill_inactive_zone(struct zone *zone, 
 	int pgmoved;
 	int pgdeactivate = 0;
 	int pgscanned;
-	int nr_pages = sc->nr_to_scan;
 	LIST_HEAD(l_hold);	/* The pages which were snipped off */
 	LIST_HEAD(l_inactive);	/* Pages to go onto the inactive_list */
 	LIST_HEAD(l_active);	/* Pages to go onto the active_list */

@@ -996,7 +990,7 @@ refill_inactive_zone(struct zone *zone, 
 
 	lru_add_drain();
 	spin_lock_irq(&zone->lru_lock);
-	pgmoved = isolate_lru_pages(nr_pages, &zone->active_list,
+	pgmoved = isolate_lru_pages(sc->nr_to_scan, &zone->active_list,
 				    &l_hold, &pgscanned);
 	zone->pages_scanned += pgscanned;
 	zone->nr_active -= pgmoved;

@@ -76,12 +76,6 @@ struct scan_control {
 
 	/* Can pages be swapped as part of reclaim? */
 	int may_swap;
-
-	/* This context's SWAP_CLUSTER_MAX. If freeing memory for
-	 * suspend, we effectively ignore SWAP_CLUSTER_MAX.
-	 * In this context, it doesn't matter that we scan the
-	 * whole list at once. */
-	int swap_cluster_max;
 };
 
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))

@@ -1127,7 +1121,6 @@ shrink_zone(struct zone *zone, struct sc
 	nr_inactive &= ~(SWAP_CLUSTER_MAX - 1);
 
 	sc->nr_to_scan = SWAP_CLUSTER_MAX;
-	sc->nr_to_reclaim = sc->swap_cluster_max;
 
 	while (nr_active >= SWAP_CLUSTER_MAX * 1024 || nr_inactive) {
 		if (nr_active >= SWAP_CLUSTER_MAX * 1024) {

@@ -1271,7 +1264,7 @@ int try_to_free_pages(struct zone **zone
 		sc.nr_scanned = 0;
 		sc.nr_reclaimed = 0;
 		sc.priority = priority;
-		sc.swap_cluster_max = SWAP_CLUSTER_MAX;
+		sc.nr_to_reclaim = SWAP_CLUSTER_MAX;
 		if (!priority)
 			disable_swap_token();
 		shrink_caches(zones, &sc);

@@ -1283,7 +1276,7 @@ int try_to_free_pages(struct zone **zone
 		}
 		total_scanned += sc.nr_scanned;
 		total_reclaimed += sc.nr_reclaimed;
-		if (total_reclaimed >= sc.swap_cluster_max) {
+		if (total_reclaimed >= SWAP_CLUSTER_MAX) {
 			ret = 1;
 			goto out;
 		}

@@ -1295,7 +1288,7 @@ int try_to_free_pages(struct zone **zone
 		 * that's undesirable in laptop mode, where we *want* lumpy
 		 * writeout.  So in laptop mode, write out the whole world.
 		 */
-		if (total_scanned > sc.swap_cluster_max + sc.swap_cluster_max/2) {
+		if (total_scanned > SWAP_CLUSTER_MAX * 3 / 2) {
 			wakeup_pdflush(laptop_mode ? 0 : total_scanned);
 			sc.may_writepage = 1;
 		}

@@ -1376,7 +1369,7 @@ loop_again:
 		sc.nr_scanned = 0;
 		sc.nr_reclaimed = 0;
 		sc.priority = priority;
-		sc.swap_cluster_max = nr_pages ? nr_pages : SWAP_CLUSTER_MAX;
+		sc.nr_to_reclaim = nr_pages ? nr_pages : SWAP_CLUSTER_MAX;
 
 		/* The swap token gets in the way of swapout... */
 		if (!priority)

@@ -1231,7 +1231,6 @@ int try_to_free_pages(struct zone **zone
 {
 	int priority;
 	int ret = 0;
-	int total_scanned = 0, total_reclaimed = 0;
 	struct reclaim_state *reclaim_state = current->reclaim_state;
 	struct scan_control sc;
 	int i;

@@ -1241,6 +1240,8 @@ int try_to_free_pages(struct zone **zone
 	sc.gfp_mask = gfp_mask;
 	sc.may_writepage = 0;
 	sc.may_swap = 1;
+	sc.nr_scanned = 0;
+	sc.nr_reclaimed = 0;
 
 	inc_page_state(allocstall);
 

@@ -1261,8 +1262,6 @@ int try_to_free_pages(struct zone **zone
 	 */
 	for (priority = DEF_PRIORITY + PRIORITY_STEPS; priority >= 0; priority--) {
 		sc.nr_mapped = read_page_state(nr_mapped);
-		sc.nr_scanned = 0;
-		sc.nr_reclaimed = 0;
 		sc.priority = priority;
 		sc.nr_to_reclaim = SWAP_CLUSTER_MAX;
 		if (!priority)

@@ -1274,9 +1273,7 @@ int try_to_free_pages(struct zone **zone
 			sc.nr_reclaimed += reclaim_state->reclaimed_slab;
 			reclaim_state->reclaimed_slab = 0;
 		}
-		total_scanned += sc.nr_scanned;
-		total_reclaimed += sc.nr_reclaimed;
-		if (total_reclaimed >= SWAP_CLUSTER_MAX) {
+		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX) {
 			ret = 1;
 			goto out;
 		}

@@ -1288,13 +1285,13 @@ int try_to_free_pages(struct zone **zone
 		 * that's undesirable in laptop mode, where we *want* lumpy
 		 * writeout.  So in laptop mode, write out the whole world.
 		 */
-		if (total_scanned > SWAP_CLUSTER_MAX * 3 / 2) {
-			wakeup_pdflush(laptop_mode ? 0 : total_scanned);
+		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 3 / 2) {
+			wakeup_pdflush(laptop_mode ? 0 : sc.nr_scanned);
 			sc.may_writepage = 1;
 		}
 
 		/* Take a nap, wait for some writeback to complete */
-		if (sc.nr_scanned && priority < DEF_PRIORITY - PRIORITY_STEPS)
+		if (priority < DEF_PRIORITY - PRIORITY_STEPS)
 			blk_congestion_wait(WRITE, HZ/10);
 	}
 out:

@@ -1340,18 +1337,17 @@ static int balance_pgdat(pg_data_t *pgda
 	int all_zones_ok;
 	int priority;
 	int i;
-	int total_scanned, total_reclaimed;
 	struct reclaim_state *reclaim_state = current->reclaim_state;
 	struct scan_control sc;
 	struct zone *prev_zone = pgdat->node_zones;
 
 loop_again:
-	total_scanned = 0;
-	total_reclaimed = 0;
 	sc.gfp_mask = GFP_KERNEL;
 	sc.may_writepage = 0;
 	sc.may_swap = 1;
 	sc.nr_mapped = read_page_state(nr_mapped);
+	sc.nr_scanned = 0;
+	sc.nr_reclaimed = 0;
 
 	inc_page_state(pageoutrun);
 

@@ -1366,8 +1362,6 @@ loop_again:
 
 	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
 		all_zones_ok = 1;
-		sc.nr_scanned = 0;
-		sc.nr_reclaimed = 0;
 		sc.priority = priority;
 		sc.nr_to_reclaim = nr_pages ? nr_pages : SWAP_CLUSTER_MAX;
 

@@ -1437,19 +1431,17 @@ scan_swspd:
 		reclaim_state->reclaimed_slab = 0;
 		shrink_slab(prev_zone, priority, GFP_KERNEL);
 		sc.nr_reclaimed += reclaim_state->reclaimed_slab;
-		total_reclaimed += sc.nr_reclaimed;
-		total_scanned += sc.nr_scanned;
 
 		/*
 		 * If we've done a decent amount of scanning and
 		 * the reclaim ratio is low, start doing writepage
 		 * even in laptop mode
 		 */
-		if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
-		    total_scanned > total_reclaimed+total_reclaimed/2)
+		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 2 &&
+		    sc.nr_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
 			sc.may_writepage = 1;
 
-		if (nr_pages && to_free > total_reclaimed)
+		if (nr_pages && to_free > sc.nr_reclaimed)
 			continue;	/* swsusp: need to do more work */
 		if (all_zones_ok)
 			break;		/* kswapd: all done */

@@ -1457,7 +1449,7 @@ scan_swspd:
 		 * OK, kswapd is getting into trouble.  Take a nap, then take
 		 * another pass across the zones.
 		 */
-		if (total_scanned && priority < DEF_PRIORITY - PRIORITY_STEPS)
+		if (priority < DEF_PRIORITY - PRIORITY_STEPS)
 			blk_congestion_wait(WRITE, HZ/10);
 
 		/*

@@ -1466,7 +1458,7 @@ scan_swspd:
 		 * matches the direct reclaim path behaviour in terms of impact
 		 * on zone->*_priority.
 		 */
-		if ((total_reclaimed >= SWAP_CLUSTER_MAX) && (!nr_pages))
+		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX && !nr_pages)
 			break;
 	}
 	for (i = 0; i < pgdat->nr_zones; i++) {

@@ -1479,7 +1471,7 @@ scan_swspd:
 		goto loop_again;
 	}
 
-	return total_reclaimed;
+	return sc.nr_reclaimed;
 }
 
 /*

@@ -72,12 +72,12 @@ struct scan_control {
 	/* This context's GFP mask */
 	gfp_t gfp_mask;
 
-	int may_writepage;
-
-	/* Can pages be swapped as part of reclaim? */
-	int may_swap;
+	unsigned long flags;
 };
 
+#define SC_MAY_WRITEPAGE	0x1
+#define SC_MAY_SWAP		0x2	/* Can pages be swapped as part of reclaim? */
+
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
 
 #ifdef ARCH_HAS_PREFETCH

@@ -488,7 +488,7 @@ static int shrink_list(struct list_head 
 		 * Try to allocate it some swap space here.
 		 */
 		if (PageAnon(page) && !PageSwapCache(page)) {
-			if (!sc->may_swap)
+			if (!(sc->flags & SC_MAY_SWAP))
 				goto keep_locked;
 			if (!add_to_swap(page, GFP_ATOMIC))
 				goto activate_locked;

@@ -519,7 +519,7 @@ static int shrink_list(struct list_head 
 				goto keep_locked;
 			if (!may_enter_fs)
 				goto keep_locked;
-			if (laptop_mode && !sc->may_writepage)
+			if (laptop_mode && !(sc->flags & SC_MAY_WRITEPAGE))
 				goto keep_locked;
 
 			/* Page is dirty, try to write it out here */

@@ -1238,8 +1238,7 @@ int try_to_free_pages(struct zone **zone
 	delay_prefetch();
 
 	sc.gfp_mask = gfp_mask;
-	sc.may_writepage = 0;
-	sc.may_swap = 1;
+	sc.flags = SC_MAY_SWAP;
 	sc.nr_scanned = 0;
 	sc.nr_reclaimed = 0;
 

@@ -1287,7 +1286,7 @@ int try_to_free_pages(struct zone **zone
 		 */
 		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 3 / 2) {
 			wakeup_pdflush(laptop_mode ? 0 : sc.nr_scanned);
-			sc.may_writepage = 1;
+			sc.flags |= SC_MAY_WRITEPAGE;
 		}
 
 		/* Take a nap, wait for some writeback to complete */

@@ -1343,8 +1342,7 @@ static int balance_pgdat(pg_data_t *pgda
 
 loop_again:
 	sc.gfp_mask = GFP_KERNEL;
-	sc.may_writepage = 0;
-	sc.may_swap = 1;
+	sc.flags = SC_MAY_SWAP;
 	sc.nr_mapped = read_page_state(nr_mapped);
 	sc.nr_scanned = 0;
 	sc.nr_reclaimed = 0;

@@ -1439,7 +1437,7 @@ scan_swspd:
 		 */
 		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 2 &&
 		    sc.nr_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
-			sc.may_writepage = 1;
+			sc.flags |= SC_MAY_WRITEPAGE;
 
 		if (nr_pages && to_free > sc.nr_reclaimed)
 			continue;	/* swsusp: need to do more work */

@@ -72,12 +72,12 @@ struct scan_control {
 	/* This context's GFP mask */
 	gfp_t gfp_mask;
 
-	int may_writepage;
-
-	/* Can pages be swapped as part of reclaim? */
-	int may_swap;
+	unsigned long flags;
 };
 
+#define SC_MAY_WRITEPAGE	0x1
+#define SC_MAY_SWAP		0x2	/* Can pages be swapped as part of reclaim? */
+
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
 
 #ifdef ARCH_HAS_PREFETCH

@@ -488,7 +488,7 @@ static int shrink_list(struct list_head 
 		 * Try to allocate it some swap space here.
 		 */
 		if (PageAnon(page) && !PageSwapCache(page)) {
-			if (!sc->may_swap)
+			if (!(sc->flags & SC_MAY_SWAP))
 				goto keep_locked;
 			if (!add_to_swap(page, GFP_ATOMIC))
 				goto activate_locked;

@@ -519,7 +519,7 @@ static int shrink_list(struct list_head 
 				goto keep_locked;
 			if (!may_enter_fs)
 				goto keep_locked;
-			if (laptop_mode && !sc->may_writepage)
+			if (laptop_mode && !(sc->flags & SC_MAY_WRITEPAGE))
 				goto keep_locked;
 
 			/* Page is dirty, try to write it out here */

@@ -1238,8 +1238,7 @@ int try_to_free_pages(struct zone **zone
 	delay_prefetch();
 
 	sc.gfp_mask = gfp_mask;
-	sc.may_writepage = 0;
-	sc.may_swap = 1;
+	sc.flags = SC_MAY_SWAP;
 	sc.nr_scanned = 0;
 	sc.nr_reclaimed = 0;
 

@@ -1287,7 +1286,7 @@ int try_to_free_pages(struct zone **zone
 		 */
 		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 3 / 2) {
 			wakeup_pdflush(laptop_mode ? 0 : sc.nr_scanned);
-			sc.may_writepage = 1;
+			sc.flags |= SC_MAY_WRITEPAGE;
 		}
 
 		/* Take a nap, wait for some writeback to complete */

@@ -1343,8 +1342,7 @@ static int balance_pgdat(pg_data_t *pgda
 
 loop_again:
 	sc.gfp_mask = GFP_KERNEL;
-	sc.may_writepage = 0;
-	sc.may_swap = 1;
+	sc.flags = SC_MAY_SWAP;
 	sc.nr_mapped = read_page_state(nr_mapped);
 	sc.nr_scanned = 0;
 	sc.nr_reclaimed = 0;

@@ -1439,7 +1437,7 @@ scan_swspd:
 		 */
 		if (sc.nr_scanned > SWAP_CLUSTER_MAX * 2 &&
 		    sc.nr_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
-			sc.may_writepage = 1;
+			sc.flags |= SC_MAY_WRITEPAGE;
 
 		if (nr_pages && to_free > sc.nr_reclaimed)
 			continue;	/* swsusp: need to do more work */

@@ -864,7 +864,8 @@ static int isolate_lru_pages(int nr_to_s
 	struct page *page;
 	int scan = 0;
 
-	while (scan++ < nr_to_scan && !list_empty(src)) {
+	while (scan < nr_to_scan && !list_empty(src)) {
+		scan++;
 		page = lru_to_page(src);
 		prefetchw_prev_lru_page(page, src, flags);
 

@@ -911,14 +912,15 @@ static void shrink_cache(struct zone *zo
 	update_zone_age(zone, nr_scan);
 	spin_unlock_irq(&zone->lru_lock);
 
-	if (nr_taken == 0)
-		return;
-
 	sc->nr_scanned += nr_scan;
 	if (current_is_kswapd())
 		mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
 	else
 		mod_page_state_zone(zone, pgscan_direct, nr_scan);
+
+	if (nr_taken == 0)
+		return;
+
 	nr_freed = shrink_list(&page_list, sc);
 	if (current_is_kswapd())
 		mod_page_state(kswapd_steal, nr_freed);

@@ -837,7 +837,8 @@ static int isolate_lru_pages(int nr_to_s
 	struct page *page;
 	int scan = 0;
 
-	while (scan++ < nr_to_scan && !list_empty(src)) {
+	while (scan < nr_to_scan && !list_empty(src)) {
+		scan++;
 		page = lru_to_page(src);
 		prefetchw_prev_lru_page(page, src, flags);
 

@@ -886,14 +887,15 @@ static void shrink_cache(struct zone *zo
 		zone->pages_scanned += nr_scan;
 		spin_unlock_irq(&zone->lru_lock);
 
-		if (nr_taken == 0)
-			goto done;
-
 		max_scan -= nr_scan;
 		if (current_is_kswapd())
 			mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
 		else
 			mod_page_state_zone(zone, pgscan_direct, nr_scan);
+
+		if (nr_taken == 0)
+			goto done;
+
 		nr_freed = shrink_list(&page_list, sc);
 		if (current_is_kswapd())
 			mod_page_state(kswapd_steal, nr_freed);

@@ -160,6 +160,7 @@ struct zone {
 	 */
 	unsigned long		aging_total;
 	unsigned long		aging_milestone;
+	unsigned long		aging_rounds;
 	unsigned long		page_age;
 
 	/*

@@ -1523,6 +1523,7 @@ void show_free_areas(void)
 			" inactive:%lukB"
 			" present:%lukB"
 			" aging:%lukB"
+			" aging_rounds:%lukB"
 			" age:%lu"
 			" pages_scanned:%lu"
 			" all_unreclaimable? %s"

@@ -1536,6 +1537,7 @@ void show_free_areas(void)
 			K(zone->nr_inactive),
 			K(zone->present_pages),
 			K(zone->aging_total),
+			zone->aging_rounds,
 			zone->page_age,
 			zone->pages_scanned,
 			(zone->all_unreclaimable ? "yes" : "no")

@@ -2149,6 +2151,7 @@ static void __init free_area_init_core(s
 		zone->nr_inactive = 0;
 		zone->aging_total = 0;
 		zone->aging_milestone = 0;
+		zone->aging_rounds = 0;
 		zone->page_age = 0;
 		atomic_set(&zone->reclaim_in_progress, 0);
 		if (!size)

@@ -2299,6 +2302,7 @@ static int zoneinfo_show(struct seq_file
 			   "\n        active   %lu"
 			   "\n        inactive %lu"
 			   "\n        aging    %lu"
+			   "\n        rounds   %lu"
 			   "\n        age      %lu"
 			   "\n        scanned  %lu (a: %lu)"
 			   "\n        spanned  %lu"

@@ -2310,6 +2314,7 @@ static int zoneinfo_show(struct seq_file
 			   zone->nr_active,
 			   zone->nr_inactive,
 			   zone->aging_total,
+			   zone->aging_rounds,
 			   zone->page_age,
 			   zone->pages_scanned,
 			   zone->nr_scan_active / 1024,

@@ -157,8 +157,10 @@ static inline void update_zone_age(struc
 
 	z->aging_total += nr_scan;
 
-	if (z->aging_total - z->aging_milestone > len)
+	if (z->aging_total - z->aging_milestone > len) {
 		z->aging_milestone += len;
+		z->aging_rounds++;
+	}
 
 	z->page_age = ((z->aging_total - z->aging_milestone)
 						<< PAGE_AGE_SHIFT) / len;

@@ -38,6 +38,7 @@
 #include <asm/div64.h>
 
 #include <linux/swapops.h>
+#include <linux/debugfs.h>
 
 /* possible outcome of pageout() */
 typedef enum {

@@ -78,6 +79,62 @@ struct scan_control {
 #define SC_MAY_WRITEPAGE	0x1
 #define SC_MAY_SWAP		0x2	/* Can pages be swapped as part of reclaim? */
 
+#define SC_RECLAIM_FROM_KSWAPD		0x10
+#define SC_RECLAIM_FROM_DIRECT		0x20
+#define SC_RECLAIM_FOR_WATERMARK	0x40
+#define SC_RECLAIM_FOR_AGING		0x80
+#define SC_RECLAIM_MASK			0xF0
+
+#ifdef CONFIG_DEBUG_FS
+static u32 debug_page_reclaim;
+
+static inline void debug_reclaim(struct scan_control *sc, unsigned long flags)
+{
+	sc->flags = (sc->flags & ~SC_RECLAIM_MASK) | flags;
+}
+
+static inline void debug_reclaim_report(struct scan_control *sc, struct zone *z)
+{
+	if (!debug_page_reclaim)
+		return;
+
+	printk(KERN_DEBUG "reclaim zone%d from %s for %s, "
+			"prio %d, scan-reclaimed %lu-%lu, age %lu, "
+			"active to scan %lu, "
+			"hot+cold+free pages %lu+%lu+%lu\n",
+			zone_idx(z),
+			(sc->flags & SC_RECLAIM_FROM_KSWAPD) ? "kswapd" :
+			((sc->flags & SC_RECLAIM_FROM_DIRECT) ? "direct" :
+								"early"),
+			(sc->flags & SC_RECLAIM_FOR_AGING) ?
+							"aging" : "watermark",
+			sc->priority, sc->nr_scanned, sc->nr_reclaimed,
+			z->page_age,
+			z->nr_scan_active,
+			z->nr_active, z->nr_inactive, z->free_pages);
+
+	if (atomic_read(&z->reclaim_in_progress))
+		printk(KERN_WARNING "reclaim_in_progress=%d\n",
+					atomic_read(&z->reclaim_in_progress));
+}
+
+static inline void debug_reclaim_init(void)
+{
+	debugfs_create_bool("debug_page_reclaim", 0644, NULL,
+							&debug_page_reclaim);
+}
+#else
+static inline void debug_reclaim(struct scan_control *sc, int flags)
+{
+}
+static inline void debug_reclaim_report(struct scan_control *sc, struct zone *z)
+{
+}
+static inline void debug_reclaim_init(void)
+{
+}
+#endif
+
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
 
 #ifdef ARCH_HAS_PREFETCH

@@ -1147,6 +1204,7 @@ shrink_zone(struct zone *zone, struct sc
 
 	atomic_dec(&zone->reclaim_in_progress);
 
+	debug_reclaim_report(sc, zone);
 	throttle_vm_writeout();
 }
 

@@ -1211,11 +1269,14 @@ shrink_caches(struct zone **zones, struc
 			continue;
 		}
 
+		debug_reclaim(sc, SC_RECLAIM_FROM_DIRECT);
 		shrink_zone(zone, sc);
 	}
 
-	if (z)
+	if (z) {
+		debug_reclaim(sc, SC_RECLAIM_FROM_DIRECT|SC_RECLAIM_FOR_AGING);
 		shrink_zone(z, sc);
+	}
 }
  
 /*

@@ -1397,14 +1458,22 @@ loop_again:
 				if (!zone_watermark_ok(zone, order,
 						zone->pages_high, 0, 0)) {
 					/* have to scan for free pages */
+					debug_reclaim(&sc,
+							SC_RECLAIM_FROM_KSWAPD |
+							SC_RECLAIM_FOR_WATERMARK);
 					goto scan;
 				}
 				if (age_ge(prev_zone, zone)) {
 					/* catch up if falls behind */
+					debug_reclaim(&sc,
+							SC_RECLAIM_FROM_KSWAPD |
+							SC_RECLAIM_FOR_AGING);
 					goto scan;
 				}
 			} else if (!age_gt(zone, prev_zone)) {
 				/* catch up if falls behind or out of sync */
+				debug_reclaim(&sc, SC_RECLAIM_FROM_KSWAPD |
+						   SC_RECLAIM_FOR_AGING);
 				goto scan;
 			}
 

@@ -1631,6 +1700,7 @@ static int __init kswapd_init(void)
 		= find_task_by_pid(kernel_thread(kswapd, pgdat, CLONE_KERNEL));
 	total_memory = nr_free_pagecache_pages();
 	hotcpu_notifier(cpu_callback, 0);
+	debug_reclaim_init();
 	return 0;
 }
 

@@ -1450,6 +1450,17 @@ loop_again:
 			if (nr_pages) 	/* software suspend */
 				goto scan_swspd;
 
+			if (debug_page_reclaim)
+			printk(KERN_DEBUG "zone %d%d watermark %d%d age %lu prio %d\n",
+					zone_idx(prev_zone),
+					zone_idx(zone),
+					zone_watermark_ok(zone, order,
+						zone->pages_high, 0, 0),
+					zone_watermark_ok(zone, order,
+						zone->pages_high,
+						pgdat->nr_zones - 1, 0),
+					zone->page_age, priority);
+
 			if (zone_watermark_ok(zone, order,
 						zone->pages_high,
 						pgdat->nr_zones - 1, 0)) {