[patch 0/3] mm: vmscan: followup fixes to cleanups in -mm

[patch 0/3] mm: vmscan: followup fixes to cleanups in -mm

[Johannes Weiner]

Hi Andrew,

here is a follow-up to feedback on patches you already have in -mm.
This series is not linear: the first two patches are fixlets according
to their name, the third one could be placed after "mm: vmscan: move
swappiness out of scan_control".

Thanks!

[patch 0/3] mm: vmscan: followup fixes to cleanups in -mm

[Johannes Weiner]

Hi Andrew,

here is a follow-up to feedback on patches you already have in -mm.
This series is not linear: the first two patches are fixlets according
to their name, the third one could be placed after "mm: vmscan: move
swappiness out of scan_control".

Thanks!

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Johannes Weiner]

As per Mel, replace out label with breaks from the loop.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 35747a75bf08..6f43df4a5253 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2496,10 +2496,10 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 
 		total_scanned += sc->nr_scanned;
 		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
-			goto out;
+			break;
 
 		if (sc->compaction_ready)
-			goto out;
+			break;
 
 		/*
 		 * If we're getting trouble reclaiming, start doing
@@ -2523,7 +2523,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		}
 	} while (--sc->priority >= 0);
 
-out:
 	delayacct_freepages_end();
 
 	if (sc->nr_reclaimed)
-- 
2.0.0

[patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Johannes Weiner]

As per Mel, replace out label with breaks from the loop.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 35747a75bf08..6f43df4a5253 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2496,10 +2496,10 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 
 		total_scanned += sc->nr_scanned;
 		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
-			goto out;
+			break;
 
 		if (sc->compaction_ready)
-			goto out;
+			break;
 
 		/*
 		 * If we're getting trouble reclaiming, start doing
@@ -2523,7 +2523,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		}
 	} while (--sc->priority >= 0);
 
-out:
 	delayacct_freepages_end();
 
 	if (sc->nr_reclaimed)
-- 
2.0.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Johannes Weiner]

As per Mel, use bool for reclaimability throughout and simplify the
reclaimability tracking in shrink_zones().

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 6dac1310e5e4..74a9e0ae09b0 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2244,10 +2244,10 @@ static inline bool should_continue_reclaim(struct zone *zone,
 	}
 }
 
-static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
+static bool shrink_zone(struct zone *zone, struct scan_control *sc)
 {
 	unsigned long nr_reclaimed, nr_scanned;
-	unsigned long zone_reclaimed = 0;
+	bool reclaimable = false;
 
 	do {
 		struct mem_cgroup *root = sc->target_mem_cgroup;
@@ -2291,12 +2291,13 @@ static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
 			   sc->nr_scanned - nr_scanned,
 			   sc->nr_reclaimed - nr_reclaimed);
 
-		zone_reclaimed += sc->nr_reclaimed - nr_reclaimed;
+		if (sc->nr_reclaimed - nr_reclaimed)
+			reclaimable = true;
 
 	} while (should_continue_reclaim(zone, sc->nr_reclaimed - nr_reclaimed,
 					 sc->nr_scanned - nr_scanned, sc));
 
-	return zone_reclaimed;
+	return reclaimable;
 }
 
 /* Returns true if compaction should go ahead for a high-order request */
@@ -2346,7 +2347,7 @@ static inline bool compaction_ready(struct zone *zone, int order)
  * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
  *
- * Returns whether the zones overall are reclaimable or not.
+ * Returns true if a zone was reclaimable.
  */
 static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 {
@@ -2361,7 +2362,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 		.gfp_mask = sc->gfp_mask,
 	};
 	enum zone_type requested_highidx = gfp_zone(sc->gfp_mask);
-	bool all_unreclaimable = true;
+	bool reclaimable = false;
 
 	/*
 	 * If the number of buffer_heads in the machine exceeds the maximum
@@ -2376,8 +2377,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 
 	for_each_zone_zonelist_nodemask(zone, z, zonelist,
 					gfp_zone(sc->gfp_mask), sc->nodemask) {
-		unsigned long zone_reclaimed = 0;
-
 		if (!populated_zone(zone))
 			continue;
 		/*
@@ -2424,15 +2423,17 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 						&nr_soft_scanned);
 			sc->nr_reclaimed += nr_soft_reclaimed;
 			sc->nr_scanned += nr_soft_scanned;
-			zone_reclaimed += nr_soft_reclaimed;
+			if (nr_soft_reclaimed)
+				reclaimable = true;
 			/* need some check for avoid more shrink_zone() */
 		}
 
-		zone_reclaimed += shrink_zone(zone, sc);
+		if (shrink_zone(zone, sc))
+			reclaimable = true;
 
-		if (zone_reclaimed ||
-		    (global_reclaim(sc) && zone_reclaimable(zone)))
-			all_unreclaimable = false;
+		if (global_reclaim(sc) &&
+		    !reclaimable && zone_reclaimable(zone))
+			reclaimable = true;
 	}
 
 	/*
@@ -2455,7 +2456,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 	 */
 	sc->gfp_mask = orig_mask;
 
-	return !all_unreclaimable;
+	return reclaimable;
 }
 
 /*
-- 
2.0.0

[patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Johannes Weiner]

As per Mel, use bool for reclaimability throughout and simplify the
reclaimability tracking in shrink_zones().

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 6dac1310e5e4..74a9e0ae09b0 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2244,10 +2244,10 @@ static inline bool should_continue_reclaim(struct zone *zone,
 	}
 }
 
-static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
+static bool shrink_zone(struct zone *zone, struct scan_control *sc)
 {
 	unsigned long nr_reclaimed, nr_scanned;
-	unsigned long zone_reclaimed = 0;
+	bool reclaimable = false;
 
 	do {
 		struct mem_cgroup *root = sc->target_mem_cgroup;
@@ -2291,12 +2291,13 @@ static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
 			   sc->nr_scanned - nr_scanned,
 			   sc->nr_reclaimed - nr_reclaimed);
 
-		zone_reclaimed += sc->nr_reclaimed - nr_reclaimed;
+		if (sc->nr_reclaimed - nr_reclaimed)
+			reclaimable = true;
 
 	} while (should_continue_reclaim(zone, sc->nr_reclaimed - nr_reclaimed,
 					 sc->nr_scanned - nr_scanned, sc));
 
-	return zone_reclaimed;
+	return reclaimable;
 }
 
 /* Returns true if compaction should go ahead for a high-order request */
@@ -2346,7 +2347,7 @@ static inline bool compaction_ready(struct zone *zone, int order)
  * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
  *
- * Returns whether the zones overall are reclaimable or not.
+ * Returns true if a zone was reclaimable.
  */
 static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 {
@@ -2361,7 +2362,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 		.gfp_mask = sc->gfp_mask,
 	};
 	enum zone_type requested_highidx = gfp_zone(sc->gfp_mask);
-	bool all_unreclaimable = true;
+	bool reclaimable = false;
 
 	/*
 	 * If the number of buffer_heads in the machine exceeds the maximum
@@ -2376,8 +2377,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 
 	for_each_zone_zonelist_nodemask(zone, z, zonelist,
 					gfp_zone(sc->gfp_mask), sc->nodemask) {
-		unsigned long zone_reclaimed = 0;
-
 		if (!populated_zone(zone))
 			continue;
 		/*
@@ -2424,15 +2423,17 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 						&nr_soft_scanned);
 			sc->nr_reclaimed += nr_soft_reclaimed;
 			sc->nr_scanned += nr_soft_scanned;
-			zone_reclaimed += nr_soft_reclaimed;
+			if (nr_soft_reclaimed)
+				reclaimable = true;
 			/* need some check for avoid more shrink_zone() */
 		}
 
-		zone_reclaimed += shrink_zone(zone, sc);
+		if (shrink_zone(zone, sc))
+			reclaimable = true;
 
-		if (zone_reclaimed ||
-		    (global_reclaim(sc) && zone_reclaimable(zone)))
-			all_unreclaimable = false;
+		if (global_reclaim(sc) &&
+		    !reclaimable && zone_reclaimable(zone))
+			reclaimable = true;
 	}
 
 	/*
@@ -2455,7 +2456,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 	 */
 	sc->gfp_mask = orig_mask;
 
-	return !all_unreclaimable;
+	return reclaimable;
 }
 
 /*
-- 
2.0.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[patch 3/3] mm: vmscan: clean up struct scan_control

[Johannes Weiner]

Reorder the members by input and output, then turn the individual
integers for may_writepage, may_unmap, may_swap, compaction_ready,
hibernation_mode into flags that fit into a single integer.

Stack delta: +72/-296 -224                   old     new   delta
kswapd                                       104     176     +72
try_to_free_pages                             80      56     -24
try_to_free_mem_cgroup_pages                  80      56     -24
shrink_all_memory                             88      64     -24
reclaim_clean_pages_from_list                168     144     -24
mem_cgroup_shrink_node_zone                  104      80     -24
__zone_reclaim                               176     152     -24
balance_pgdat                                152       -    -152

   text    data     bss     dec     hex filename
  38151    5641      16   43808    ab20 mm/vmscan.o.old
  38047    5641      16   43704    aab8 mm/vmscan.o

Suggested-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
 1 file changed, 78 insertions(+), 80 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index c28b8981e56a..73d8e69ff3eb 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -58,36 +58,28 @@
 #define CREATE_TRACE_POINTS
 #include <trace/events/vmscan.h>
 
-struct scan_control {
-	/* Incremented by the number of inactive pages that were scanned */
-	unsigned long nr_scanned;
-
-	/* Number of pages freed so far during a call to shrink_zones() */
-	unsigned long nr_reclaimed;
-
-	/* One of the zones is ready for compaction */
-	int compaction_ready;
+/* Scan control flags */
+#define MAY_WRITEPAGE		0x1
+#define MAY_UNMAP		0x2
+#define MAY_SWAP		0x4
+#define MAY_SKIP_CONGESTION	0x8
+#define COMPACTION_READY	0x10
 
+struct scan_control {
 	/* How many pages shrink_list() should reclaim */
 	unsigned long nr_to_reclaim;
 
-	unsigned long hibernation_mode;
-
 	/* This context's GFP mask */
 	gfp_t gfp_mask;
 
-	int may_writepage;
-
-	/* Can mapped pages be reclaimed? */
-	int may_unmap;
-
-	/* Can pages be swapped as part of reclaim? */
-	int may_swap;
-
+	/* Allocation order */
 	int order;
 
-	/* Scan (total_size >> priority) pages at once */
-	int priority;
+	/*
+	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
+	 * are scanned.
+	 */
+	nodemask_t	*nodemask;
 
 	/*
 	 * The memory cgroup that hit its limit and as a result is the
@@ -95,11 +87,17 @@ struct scan_control {
 	 */
 	struct mem_cgroup *target_mem_cgroup;
 
-	/*
-	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
-	 * are scanned.
-	 */
-	nodemask_t	*nodemask;
+	/* Scan (total_size >> priority) pages at once */
+	int priority;
+
+	/* Scan control flags; see above */
+	unsigned int flags;
+
+	/* Incremented by the number of inactive pages that were scanned */
+	unsigned long nr_scanned;
+
+	/* Number of pages freed so far during a call to shrink_zones() */
+	unsigned long nr_reclaimed;
 };
 
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
@@ -840,7 +838,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		if (unlikely(!page_evictable(page)))
 			goto cull_mlocked;
 
-		if (!sc->may_unmap && page_mapped(page))
+		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))
 			goto keep_locked;
 
 		/* Double the slab pressure for mapped and swapcache pages */
@@ -1014,7 +1012,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 				goto keep_locked;
 			if (!may_enter_fs)
 				goto keep_locked;
-			if (!sc->may_writepage)
+			if (!(sc->flags & MAY_WRITEPAGE))
 				goto keep_locked;
 
 			/* Page is dirty, try to write it out here */
@@ -1146,7 +1144,7 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
 	struct scan_control sc = {
 		.gfp_mask = GFP_KERNEL,
 		.priority = DEF_PRIORITY,
-		.may_unmap = 1,
+		.flags = MAY_UNMAP,
 	};
 	unsigned long ret, dummy1, dummy2, dummy3, dummy4, dummy5;
 	struct page *page, *next;
@@ -1489,9 +1487,9 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
 
 	lru_add_drain();
 
-	if (!sc->may_unmap)
+	if (!(sc->flags & MAY_UNMAP))
 		isolate_mode |= ISOLATE_UNMAPPED;
-	if (!sc->may_writepage)
+	if (!(sc->flags & MAY_WRITEPAGE))
 		isolate_mode |= ISOLATE_CLEAN;
 
 	spin_lock_irq(&zone->lru_lock);
@@ -1593,7 +1591,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
 	 * is congested. Allow kswapd to continue until it starts encountering
 	 * unqueued dirty pages or cycling through the LRU too quickly.
 	 */
-	if (!sc->hibernation_mode && !current_is_kswapd() &&
+	if (!(sc->flags & MAY_SKIP_CONGESTION) && !current_is_kswapd() &&
 	    current_may_throttle())
 		wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10);
 
@@ -1683,9 +1681,9 @@ static void shrink_active_list(unsigned long nr_to_scan,
 
 	lru_add_drain();
 
-	if (!sc->may_unmap)
+	if (!(sc->flags & MAY_UNMAP))
 		isolate_mode |= ISOLATE_UNMAPPED;
-	if (!sc->may_writepage)
+	if (!(sc->flags & MAY_WRITEPAGE))
 		isolate_mode |= ISOLATE_CLEAN;
 
 	spin_lock_irq(&zone->lru_lock);
@@ -1897,7 +1895,7 @@ static void get_scan_count(struct lruvec *lruvec, int swappiness,
 		force_scan = true;
 
 	/* If we have no swap space, do not bother scanning anon pages. */
-	if (!sc->may_swap || (get_nr_swap_pages() <= 0)) {
+	if (!(sc->flags & MAY_SWAP) || (get_nr_swap_pages() <= 0)) {
 		scan_balance = SCAN_FILE;
 		goto out;
 	}
@@ -2406,7 +2404,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 			    sc->order > PAGE_ALLOC_COSTLY_ORDER &&
 			    zonelist_zone_idx(z) <= requested_highidx &&
 			    compaction_ready(zone, sc->order)) {
-				sc->compaction_ready = true;
+				sc->flags |= COMPACTION_READY;
 				continue;
 			}
 
@@ -2496,7 +2494,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
 			break;
 
-		if (sc->compaction_ready)
+		if (sc->flags & COMPACTION_READY)
 			break;
 
 		/*
@@ -2504,7 +2502,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		 * writepage even in laptop mode.
 		 */
 		if (sc->priority < DEF_PRIORITY - 2)
-			sc->may_writepage = 1;
+			sc->flags |= MAY_WRITEPAGE;
 
 		/*
 		 * Try to write back as many pages as we just scanned.  This
@@ -2517,7 +2515,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		if (total_scanned > writeback_threshold) {
 			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned,
 						WB_REASON_TRY_TO_FREE_PAGES);
-			sc->may_writepage = 1;
+			sc->flags |= MAY_WRITEPAGE;
 		}
 	} while (--sc->priority >= 0);
 
@@ -2527,7 +2525,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		return sc->nr_reclaimed;
 
 	/* Aborted reclaim to try compaction? don't OOM, then */
-	if (sc->compaction_ready)
+	if (sc->flags & COMPACTION_READY)
 		return 1;
 
 	/* Any of the zones still reclaimable?  Don't OOM. */
@@ -2668,17 +2666,17 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 {
 	unsigned long nr_reclaimed;
 	struct scan_control sc = {
-		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
-		.may_writepage = !laptop_mode,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.may_unmap = 1,
-		.may_swap = 1,
+		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
 		.order = order,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = NULL,
 		.nodemask = nodemask,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP | MAY_SWAP,
 	};
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+
 	/*
 	 * Do not enter reclaim if fatal signal was delivered while throttled.
 	 * 1 is returned so that the page allocator does not OOM kill at this
@@ -2688,7 +2686,7 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 		return 1;
 
 	trace_mm_vmscan_direct_reclaim_begin(order,
-				sc.may_writepage,
+				sc.flags & MAY_WRITEPAGE,
 				gfp_mask);
 
 	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
@@ -2706,23 +2704,22 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
 						unsigned long *nr_scanned)
 {
 	struct scan_control sc = {
-		.nr_scanned = 0,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.may_writepage = !laptop_mode,
-		.may_unmap = 1,
-		.may_swap = !noswap,
-		.order = 0,
-		.priority = 0,
+		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
+		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
 		.target_mem_cgroup = memcg,
+		.flags = MAY_UNMAP,
 	};
 	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
 	int swappiness = mem_cgroup_swappiness(memcg);
 
-	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
-			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+	if (!noswap)
+		sc.flags |= MAY_SWAP;
 
 	trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order,
-						      sc.may_writepage,
+						      sc.flags & MAY_WRITEPAGE,
 						      sc.gfp_mask);
 
 	/*
@@ -2748,18 +2745,19 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 	unsigned long nr_reclaimed;
 	int nid;
 	struct scan_control sc = {
-		.may_writepage = !laptop_mode,
-		.may_unmap = 1,
-		.may_swap = !noswap,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.order = 0,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = memcg,
-		.nodemask = NULL, /* we don't care the placement */
 		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
-				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
+		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
+		.target_mem_cgroup = memcg,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP,
 	};
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+	if (!noswap)
+		sc.flags |= MAY_SWAP;
+
 	/*
 	 * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
 	 * take care of from where we get pages. So the node where we start the
@@ -2770,7 +2768,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 	zonelist = NODE_DATA(nid)->node_zonelists;
 
 	trace_mm_vmscan_memcg_reclaim_begin(0,
-					    sc.may_writepage,
+					    sc.flags & MAY_WRITEPAGE,
 					    sc.gfp_mask);
 
 	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
@@ -3015,15 +3013,15 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
 	unsigned long nr_soft_scanned;
 	struct scan_control sc = {
 		.gfp_mask = GFP_KERNEL,
-		.priority = DEF_PRIORITY,
-		.may_unmap = 1,
-		.may_swap = 1,
-		.may_writepage = !laptop_mode,
 		.order = order,
-		.target_mem_cgroup = NULL,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP | MAY_SWAP,
 	};
 	count_vm_event(PAGEOUTRUN);
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+
 	do {
 		unsigned long lru_pages = 0;
 		unsigned long nr_attempted = 0;
@@ -3104,7 +3102,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
 		 * even in laptop mode.
 		 */
 		if (sc.priority < DEF_PRIORITY - 2)
-			sc.may_writepage = 1;
+			sc.flags |= MAY_WRITEPAGE;
 
 		/*
 		 * Now scan the zone in the dma->highmem direction, stopping
@@ -3401,14 +3399,11 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
 {
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
-		.gfp_mask = GFP_HIGHUSER_MOVABLE,
-		.may_swap = 1,
-		.may_unmap = 1,
-		.may_writepage = 1,
 		.nr_to_reclaim = nr_to_reclaim,
-		.hibernation_mode = 1,
-		.order = 0,
+		.gfp_mask = GFP_HIGHUSER_MOVABLE,
 		.priority = DEF_PRIORITY,
+		.flags = MAY_WRITEPAGE | MAY_UNMAP | MAY_SWAP |
+		         MAY_SKIP_CONGESTION,
 	};
 	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
 	struct task_struct *p = current;
@@ -3588,19 +3583,22 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 	struct task_struct *p = current;
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
-		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
-		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
-		.may_swap = 1,
 		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
 		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
 		.order = order,
 		.priority = ZONE_RECLAIM_PRIORITY,
+		.flags = MAY_SWAP,
 	};
 	struct shrink_control shrink = {
 		.gfp_mask = sc.gfp_mask,
 	};
 	unsigned long nr_slab_pages0, nr_slab_pages1;
 
+	if (zone_reclaim_mode & RECLAIM_WRITE)
+		sc.flags |= MAY_WRITEPAGE;
+	if (zone_reclaim_mode & RECLAIM_SWAP)
+		sc.flags |= MAY_UNMAP;
+
 	cond_resched();
 	/*
 	 * We need to be able to allocate from the reserves for RECLAIM_SWAP
-- 
2.0.0

[patch 3/3] mm: vmscan: clean up struct scan_control

[Johannes Weiner]

Reorder the members by input and output, then turn the individual
integers for may_writepage, may_unmap, may_swap, compaction_ready,
hibernation_mode into flags that fit into a single integer.

Stack delta: +72/-296 -224                   old     new   delta
kswapd                                       104     176     +72
try_to_free_pages                             80      56     -24
try_to_free_mem_cgroup_pages                  80      56     -24
shrink_all_memory                             88      64     -24
reclaim_clean_pages_from_list                168     144     -24
mem_cgroup_shrink_node_zone                  104      80     -24
__zone_reclaim                               176     152     -24
balance_pgdat                                152       -    -152

   text    data     bss     dec     hex filename
  38151    5641      16   43808    ab20 mm/vmscan.o.old
  38047    5641      16   43704    aab8 mm/vmscan.o

Suggested-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
 1 file changed, 78 insertions(+), 80 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index c28b8981e56a..73d8e69ff3eb 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -58,36 +58,28 @@
 #define CREATE_TRACE_POINTS
 #include <trace/events/vmscan.h>
 
-struct scan_control {
-	/* Incremented by the number of inactive pages that were scanned */
-	unsigned long nr_scanned;
-
-	/* Number of pages freed so far during a call to shrink_zones() */
-	unsigned long nr_reclaimed;
-
-	/* One of the zones is ready for compaction */
-	int compaction_ready;
+/* Scan control flags */
+#define MAY_WRITEPAGE		0x1
+#define MAY_UNMAP		0x2
+#define MAY_SWAP		0x4
+#define MAY_SKIP_CONGESTION	0x8
+#define COMPACTION_READY	0x10
 
+struct scan_control {
 	/* How many pages shrink_list() should reclaim */
 	unsigned long nr_to_reclaim;
 
-	unsigned long hibernation_mode;
-
 	/* This context's GFP mask */
 	gfp_t gfp_mask;
 
-	int may_writepage;
-
-	/* Can mapped pages be reclaimed? */
-	int may_unmap;
-
-	/* Can pages be swapped as part of reclaim? */
-	int may_swap;
-
+	/* Allocation order */
 	int order;
 
-	/* Scan (total_size >> priority) pages at once */
-	int priority;
+	/*
+	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
+	 * are scanned.
+	 */
+	nodemask_t	*nodemask;
 
 	/*
 	 * The memory cgroup that hit its limit and as a result is the
@@ -95,11 +87,17 @@ struct scan_control {
 	 */
 	struct mem_cgroup *target_mem_cgroup;
 
-	/*
-	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
-	 * are scanned.
-	 */
-	nodemask_t	*nodemask;
+	/* Scan (total_size >> priority) pages at once */
+	int priority;
+
+	/* Scan control flags; see above */
+	unsigned int flags;
+
+	/* Incremented by the number of inactive pages that were scanned */
+	unsigned long nr_scanned;
+
+	/* Number of pages freed so far during a call to shrink_zones() */
+	unsigned long nr_reclaimed;
 };
 
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
@@ -840,7 +838,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 		if (unlikely(!page_evictable(page)))
 			goto cull_mlocked;
 
-		if (!sc->may_unmap && page_mapped(page))
+		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))
 			goto keep_locked;
 
 		/* Double the slab pressure for mapped and swapcache pages */
@@ -1014,7 +1012,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
 				goto keep_locked;
 			if (!may_enter_fs)
 				goto keep_locked;
-			if (!sc->may_writepage)
+			if (!(sc->flags & MAY_WRITEPAGE))
 				goto keep_locked;
 
 			/* Page is dirty, try to write it out here */
@@ -1146,7 +1144,7 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
 	struct scan_control sc = {
 		.gfp_mask = GFP_KERNEL,
 		.priority = DEF_PRIORITY,
-		.may_unmap = 1,
+		.flags = MAY_UNMAP,
 	};
 	unsigned long ret, dummy1, dummy2, dummy3, dummy4, dummy5;
 	struct page *page, *next;
@@ -1489,9 +1487,9 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
 
 	lru_add_drain();
 
-	if (!sc->may_unmap)
+	if (!(sc->flags & MAY_UNMAP))
 		isolate_mode |= ISOLATE_UNMAPPED;
-	if (!sc->may_writepage)
+	if (!(sc->flags & MAY_WRITEPAGE))
 		isolate_mode |= ISOLATE_CLEAN;
 
 	spin_lock_irq(&zone->lru_lock);
@@ -1593,7 +1591,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
 	 * is congested. Allow kswapd to continue until it starts encountering
 	 * unqueued dirty pages or cycling through the LRU too quickly.
 	 */
-	if (!sc->hibernation_mode && !current_is_kswapd() &&
+	if (!(sc->flags & MAY_SKIP_CONGESTION) && !current_is_kswapd() &&
 	    current_may_throttle())
 		wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10);
 
@@ -1683,9 +1681,9 @@ static void shrink_active_list(unsigned long nr_to_scan,
 
 	lru_add_drain();
 
-	if (!sc->may_unmap)
+	if (!(sc->flags & MAY_UNMAP))
 		isolate_mode |= ISOLATE_UNMAPPED;
-	if (!sc->may_writepage)
+	if (!(sc->flags & MAY_WRITEPAGE))
 		isolate_mode |= ISOLATE_CLEAN;
 
 	spin_lock_irq(&zone->lru_lock);
@@ -1897,7 +1895,7 @@ static void get_scan_count(struct lruvec *lruvec, int swappiness,
 		force_scan = true;
 
 	/* If we have no swap space, do not bother scanning anon pages. */
-	if (!sc->may_swap || (get_nr_swap_pages() <= 0)) {
+	if (!(sc->flags & MAY_SWAP) || (get_nr_swap_pages() <= 0)) {
 		scan_balance = SCAN_FILE;
 		goto out;
 	}
@@ -2406,7 +2404,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 			    sc->order > PAGE_ALLOC_COSTLY_ORDER &&
 			    zonelist_zone_idx(z) <= requested_highidx &&
 			    compaction_ready(zone, sc->order)) {
-				sc->compaction_ready = true;
+				sc->flags |= COMPACTION_READY;
 				continue;
 			}
 
@@ -2496,7 +2494,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
 			break;
 
-		if (sc->compaction_ready)
+		if (sc->flags & COMPACTION_READY)
 			break;
 
 		/*
@@ -2504,7 +2502,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		 * writepage even in laptop mode.
 		 */
 		if (sc->priority < DEF_PRIORITY - 2)
-			sc->may_writepage = 1;
+			sc->flags |= MAY_WRITEPAGE;
 
 		/*
 		 * Try to write back as many pages as we just scanned.  This
@@ -2517,7 +2515,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		if (total_scanned > writeback_threshold) {
 			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned,
 						WB_REASON_TRY_TO_FREE_PAGES);
-			sc->may_writepage = 1;
+			sc->flags |= MAY_WRITEPAGE;
 		}
 	} while (--sc->priority >= 0);
 
@@ -2527,7 +2525,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		return sc->nr_reclaimed;
 
 	/* Aborted reclaim to try compaction? don't OOM, then */
-	if (sc->compaction_ready)
+	if (sc->flags & COMPACTION_READY)
 		return 1;
 
 	/* Any of the zones still reclaimable?  Don't OOM. */
@@ -2668,17 +2666,17 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 {
 	unsigned long nr_reclaimed;
 	struct scan_control sc = {
-		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
-		.may_writepage = !laptop_mode,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.may_unmap = 1,
-		.may_swap = 1,
+		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
 		.order = order,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = NULL,
 		.nodemask = nodemask,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP | MAY_SWAP,
 	};
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+
 	/*
 	 * Do not enter reclaim if fatal signal was delivered while throttled.
 	 * 1 is returned so that the page allocator does not OOM kill at this
@@ -2688,7 +2686,7 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 		return 1;
 
 	trace_mm_vmscan_direct_reclaim_begin(order,
-				sc.may_writepage,
+				sc.flags & MAY_WRITEPAGE,
 				gfp_mask);
 
 	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
@@ -2706,23 +2704,22 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
 						unsigned long *nr_scanned)
 {
 	struct scan_control sc = {
-		.nr_scanned = 0,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.may_writepage = !laptop_mode,
-		.may_unmap = 1,
-		.may_swap = !noswap,
-		.order = 0,
-		.priority = 0,
+		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
+		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
 		.target_mem_cgroup = memcg,
+		.flags = MAY_UNMAP,
 	};
 	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
 	int swappiness = mem_cgroup_swappiness(memcg);
 
-	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
-			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+	if (!noswap)
+		sc.flags |= MAY_SWAP;
 
 	trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order,
-						      sc.may_writepage,
+						      sc.flags & MAY_WRITEPAGE,
 						      sc.gfp_mask);
 
 	/*
@@ -2748,18 +2745,19 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 	unsigned long nr_reclaimed;
 	int nid;
 	struct scan_control sc = {
-		.may_writepage = !laptop_mode,
-		.may_unmap = 1,
-		.may_swap = !noswap,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.order = 0,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = memcg,
-		.nodemask = NULL, /* we don't care the placement */
 		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
-				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
+		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
+		.target_mem_cgroup = memcg,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP,
 	};
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+	if (!noswap)
+		sc.flags |= MAY_SWAP;
+
 	/*
 	 * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
 	 * take care of from where we get pages. So the node where we start the
@@ -2770,7 +2768,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 	zonelist = NODE_DATA(nid)->node_zonelists;
 
 	trace_mm_vmscan_memcg_reclaim_begin(0,
-					    sc.may_writepage,
+					    sc.flags & MAY_WRITEPAGE,
 					    sc.gfp_mask);
 
 	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
@@ -3015,15 +3013,15 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
 	unsigned long nr_soft_scanned;
 	struct scan_control sc = {
 		.gfp_mask = GFP_KERNEL,
-		.priority = DEF_PRIORITY,
-		.may_unmap = 1,
-		.may_swap = 1,
-		.may_writepage = !laptop_mode,
 		.order = order,
-		.target_mem_cgroup = NULL,
+		.priority = DEF_PRIORITY,
+		.flags = MAY_UNMAP | MAY_SWAP,
 	};
 	count_vm_event(PAGEOUTRUN);
 
+	if (!laptop_mode)
+		sc.flags |= MAY_WRITEPAGE;
+
 	do {
 		unsigned long lru_pages = 0;
 		unsigned long nr_attempted = 0;
@@ -3104,7 +3102,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
 		 * even in laptop mode.
 		 */
 		if (sc.priority < DEF_PRIORITY - 2)
-			sc.may_writepage = 1;
+			sc.flags |= MAY_WRITEPAGE;
 
 		/*
 		 * Now scan the zone in the dma->highmem direction, stopping
@@ -3401,14 +3399,11 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
 {
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
-		.gfp_mask = GFP_HIGHUSER_MOVABLE,
-		.may_swap = 1,
-		.may_unmap = 1,
-		.may_writepage = 1,
 		.nr_to_reclaim = nr_to_reclaim,
-		.hibernation_mode = 1,
-		.order = 0,
+		.gfp_mask = GFP_HIGHUSER_MOVABLE,
 		.priority = DEF_PRIORITY,
+		.flags = MAY_WRITEPAGE | MAY_UNMAP | MAY_SWAP |
+		         MAY_SKIP_CONGESTION,
 	};
 	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
 	struct task_struct *p = current;
@@ -3588,19 +3583,22 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 	struct task_struct *p = current;
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
-		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
-		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
-		.may_swap = 1,
 		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
 		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
 		.order = order,
 		.priority = ZONE_RECLAIM_PRIORITY,
+		.flags = MAY_SWAP,
 	};
 	struct shrink_control shrink = {
 		.gfp_mask = sc.gfp_mask,
 	};
 	unsigned long nr_slab_pages0, nr_slab_pages1;
 
+	if (zone_reclaim_mode & RECLAIM_WRITE)
+		sc.flags |= MAY_WRITEPAGE;
+	if (zone_reclaim_mode & RECLAIM_SWAP)
+		sc.flags |= MAY_UNMAP;
+
 	cond_resched();
 	/*
 	 * We need to be able to allocate from the reserves for RECLAIM_SWAP
-- 
2.0.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Rik van Riel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 07/14/2014 09:20 AM, Johannes Weiner wrote:
> As per Mel, replace out label with breaks from the loop.
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Rik van Riel <riel@redhat.com>

- -- 
All rights reversed
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTw+SUAAoJEM553pKExN6Dj4wH/imfBq+85Kpjrw4NQltD+uUt
0pAzt/SX9IfiUcowi/1i1jWUKhMAfrY4SCG14g3ErKnIprMT8oa9ujRGCpnnZud2
eqLDFIHM8BlLNfIOV6a96+i1JpFDLbNL8WBjlew6X7ZDZamUG6j+0XxBOtwVemn6
Yj+cubH6mgPtovGRHdEDnyb4JOw5eue4/vIpumdTak3mnKghRpAxdN5tq7h13e1a
w/tweAWFspYHBkUj6FjeGiXrttNF7ToOy0cJeypZJZJfZFRwHBYStTe81iLa7jld
eNmfr48eLjnHrvZAN+lFm/DPuqU4ISuoCnL9N67OCudLrj8YbBzw8tOYNvff7fw=
=bbE9
-----END PGP SIGNATURE-----

Re: [patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Rik van Riel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 07/14/2014 09:20 AM, Johannes Weiner wrote:
> As per Mel, replace out label with breaks from the loop.
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Rik van Riel <riel@redhat.com>

- -- 
All rights reversed
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTw+SUAAoJEM553pKExN6Dj4wH/imfBq+85Kpjrw4NQltD+uUt
0pAzt/SX9IfiUcowi/1i1jWUKhMAfrY4SCG14g3ErKnIprMT8oa9ujRGCpnnZud2
eqLDFIHM8BlLNfIOV6a96+i1JpFDLbNL8WBjlew6X7ZDZamUG6j+0XxBOtwVemn6
Yj+cubH6mgPtovGRHdEDnyb4JOw5eue4/vIpumdTak3mnKghRpAxdN5tq7h13e1a
w/tweAWFspYHBkUj6FjeGiXrttNF7ToOy0cJeypZJZJfZFRwHBYStTe81iLa7jld
eNmfr48eLjnHrvZAN+lFm/DPuqU4ISuoCnL9N67OCudLrj8YbBzw8tOYNvff7fw=
=bbE9
-----END PGP SIGNATURE-----

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Rik van Riel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 07/14/2014 09:20 AM, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify
> the reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Rik van Riel <riel@redhat.com>


- -- 
All rights reversed
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTw+S+AAoJEM553pKExN6DtvYH/RXaEx/lWC9UR5eRQ8nQy2L4
V87wVoWPXauuIeJrGurTV28cvqUW/JXNAmnONuGdRI//9jE2vMZmVi2X5V4CnGv2
zpMsM2MhIn7tzBKW7AlBLHBC9nUEIHpo+OA3IvCvQsgG5qWNkdWOTUv1xkiqTuuQ
Nu7pxiNH360Dp+g2VCuFU2+nrjcKKSolsMBqEvGGP+Dh3/G5EQpQ/lQJ0/a/4q1y
/ew0HCYRfH2/kCMKxixTtUXR7QcMw4L4AkD0fHoisRoyuVAws4QwYA4zB1FKpxgr
XIsVPLftQc8f+++Djone+RvPaPmUOvXrf8UdbxTG5wrfBJ9aPB5AuecWqOOO22g=
=3M3e
-----END PGP SIGNATURE-----

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Rik van Riel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 07/14/2014 09:20 AM, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify
> the reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Rik van Riel <riel@redhat.com>


- -- 
All rights reversed
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTw+S+AAoJEM553pKExN6DtvYH/RXaEx/lWC9UR5eRQ8nQy2L4
V87wVoWPXauuIeJrGurTV28cvqUW/JXNAmnONuGdRI//9jE2vMZmVi2X5V4CnGv2
zpMsM2MhIn7tzBKW7AlBLHBC9nUEIHpo+OA3IvCvQsgG5qWNkdWOTUv1xkiqTuuQ
Nu7pxiNH360Dp+g2VCuFU2+nrjcKKSolsMBqEvGGP+Dh3/G5EQpQ/lQJ0/a/4q1y
/ew0HCYRfH2/kCMKxixTtUXR7QcMw4L4AkD0fHoisRoyuVAws4QwYA4zB1FKpxgr
XIsVPLftQc8f+++Djone+RvPaPmUOvXrf8UdbxTG5wrfBJ9aPB5AuecWqOOO22g=
=3M3e
-----END PGP SIGNATURE-----

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Hugh Dickins]

On Mon, 14 Jul 2014, Johannes Weiner wrote:

> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into flags that fit into a single integer.
> 
> Stack delta: +72/-296 -224                   old     new   delta
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
>    text    data     bss     dec     hex filename
>   38151    5641      16   43808    ab20 mm/vmscan.o.old
>   38047    5641      16   43704    aab8 mm/vmscan.o
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
>  1 file changed, 78 insertions(+), 80 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c28b8981e56a..73d8e69ff3eb 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -58,36 +58,28 @@
>  #define CREATE_TRACE_POINTS
>  #include <trace/events/vmscan.h>
>  
> -struct scan_control {
> -	/* Incremented by the number of inactive pages that were scanned */
> -	unsigned long nr_scanned;
> -
> -	/* Number of pages freed so far during a call to shrink_zones() */
> -	unsigned long nr_reclaimed;
> -
> -	/* One of the zones is ready for compaction */
> -	int compaction_ready;
> +/* Scan control flags */
> +#define MAY_WRITEPAGE		0x1
> +#define MAY_UNMAP		0x2
> +#define MAY_SWAP		0x4
> +#define MAY_SKIP_CONGESTION	0x8
> +#define COMPACTION_READY	0x10
>  
> +struct scan_control {
>  	/* How many pages shrink_list() should reclaim */
>  	unsigned long nr_to_reclaim;
>  
> -	unsigned long hibernation_mode;
> -
>  	/* This context's GFP mask */
>  	gfp_t gfp_mask;
>  
> -	int may_writepage;
> -
> -	/* Can mapped pages be reclaimed? */
> -	int may_unmap;
> -
> -	/* Can pages be swapped as part of reclaim? */
> -	int may_swap;
> -
> +	/* Allocation order */
>  	int order;
>  
> -	/* Scan (total_size >> priority) pages at once */
> -	int priority;
> +	/*
> +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> +	 * are scanned.
> +	 */
> +	nodemask_t	*nodemask;
>  
>  	/*
>  	 * The memory cgroup that hit its limit and as a result is the
> @@ -95,11 +87,17 @@ struct scan_control {
>  	 */
>  	struct mem_cgroup *target_mem_cgroup;
>  
> -	/*
> -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> -	 * are scanned.
> -	 */
> -	nodemask_t	*nodemask;
> +	/* Scan (total_size >> priority) pages at once */
> +	int priority;
> +
> +	/* Scan control flags; see above */
> +	unsigned int flags;

This seems to result in a fair amount of unnecessary churn:
why not just put may_writepage etc into an unsigned int bitfield,
then you get the saving without changing all the rest of the code.

Hugh

> +
> +	/* Incremented by the number of inactive pages that were scanned */
> +	unsigned long nr_scanned;
> +
> +	/* Number of pages freed so far during a call to shrink_zones() */
> +	unsigned long nr_reclaimed;
>  };
>  
>  #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
> @@ -840,7 +838,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  		if (unlikely(!page_evictable(page)))
>  			goto cull_mlocked;
>  
> -		if (!sc->may_unmap && page_mapped(page))
> +		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))
>  			goto keep_locked;
>  
>  		/* Double the slab pressure for mapped and swapcache pages */
> @@ -1014,7 +1012,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  				goto keep_locked;
>  			if (!may_enter_fs)
>  				goto keep_locked;
> -			if (!sc->may_writepage)
> +			if (!(sc->flags & MAY_WRITEPAGE))
>  				goto keep_locked;
>  
>  			/* Page is dirty, try to write it out here */
> @@ -1146,7 +1144,7 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
>  		.priority = DEF_PRIORITY,
> -		.may_unmap = 1,
> +		.flags = MAY_UNMAP,
>  	};
>  	unsigned long ret, dummy1, dummy2, dummy3, dummy4, dummy5;
>  	struct page *page, *next;
> @@ -1489,9 +1487,9 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
>  
>  	lru_add_drain();
>  
> -	if (!sc->may_unmap)
> +	if (!(sc->flags & MAY_UNMAP))
>  		isolate_mode |= ISOLATE_UNMAPPED;
> -	if (!sc->may_writepage)
> +	if (!(sc->flags & MAY_WRITEPAGE))
>  		isolate_mode |= ISOLATE_CLEAN;
>  
>  	spin_lock_irq(&zone->lru_lock);
> @@ -1593,7 +1591,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
>  	 * is congested. Allow kswapd to continue until it starts encountering
>  	 * unqueued dirty pages or cycling through the LRU too quickly.
>  	 */
> -	if (!sc->hibernation_mode && !current_is_kswapd() &&
> +	if (!(sc->flags & MAY_SKIP_CONGESTION) && !current_is_kswapd() &&
>  	    current_may_throttle())
>  		wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10);
>  
> @@ -1683,9 +1681,9 @@ static void shrink_active_list(unsigned long nr_to_scan,
>  
>  	lru_add_drain();
>  
> -	if (!sc->may_unmap)
> +	if (!(sc->flags & MAY_UNMAP))
>  		isolate_mode |= ISOLATE_UNMAPPED;
> -	if (!sc->may_writepage)
> +	if (!(sc->flags & MAY_WRITEPAGE))
>  		isolate_mode |= ISOLATE_CLEAN;
>  
>  	spin_lock_irq(&zone->lru_lock);
> @@ -1897,7 +1895,7 @@ static void get_scan_count(struct lruvec *lruvec, int swappiness,
>  		force_scan = true;
>  
>  	/* If we have no swap space, do not bother scanning anon pages. */
> -	if (!sc->may_swap || (get_nr_swap_pages() <= 0)) {
> +	if (!(sc->flags & MAY_SWAP) || (get_nr_swap_pages() <= 0)) {
>  		scan_balance = SCAN_FILE;
>  		goto out;
>  	}
> @@ -2406,7 +2404,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  			    sc->order > PAGE_ALLOC_COSTLY_ORDER &&
>  			    zonelist_zone_idx(z) <= requested_highidx &&
>  			    compaction_ready(zone, sc->order)) {
> -				sc->compaction_ready = true;
> +				sc->flags |= COMPACTION_READY;
>  				continue;
>  			}
>  
> @@ -2496,7 +2494,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
>  			break;
>  
> -		if (sc->compaction_ready)
> +		if (sc->flags & COMPACTION_READY)
>  			break;
>  
>  		/*
> @@ -2504,7 +2502,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		 * writepage even in laptop mode.
>  		 */
>  		if (sc->priority < DEF_PRIORITY - 2)
> -			sc->may_writepage = 1;
> +			sc->flags |= MAY_WRITEPAGE;
>  
>  		/*
>  		 * Try to write back as many pages as we just scanned.  This
> @@ -2517,7 +2515,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		if (total_scanned > writeback_threshold) {
>  			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned,
>  						WB_REASON_TRY_TO_FREE_PAGES);
> -			sc->may_writepage = 1;
> +			sc->flags |= MAY_WRITEPAGE;
>  		}
>  	} while (--sc->priority >= 0);
>  
> @@ -2527,7 +2525,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		return sc->nr_reclaimed;
>  
>  	/* Aborted reclaim to try compaction? don't OOM, then */
> -	if (sc->compaction_ready)
> +	if (sc->flags & COMPACTION_READY)
>  		return 1;
>  
>  	/* Any of the zones still reclaimable?  Don't OOM. */
> @@ -2668,17 +2666,17 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  {
>  	unsigned long nr_reclaimed;
>  	struct scan_control sc = {
> -		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
> -		.may_writepage = !laptop_mode,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.may_unmap = 1,
> -		.may_swap = 1,
> +		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = NULL,
>  		.nodemask = nodemask,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP | MAY_SWAP,
>  	};
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +
>  	/*
>  	 * Do not enter reclaim if fatal signal was delivered while throttled.
>  	 * 1 is returned so that the page allocator does not OOM kill at this
> @@ -2688,7 +2686,7 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  		return 1;
>  
>  	trace_mm_vmscan_direct_reclaim_begin(order,
> -				sc.may_writepage,
> +				sc.flags & MAY_WRITEPAGE,
>  				gfp_mask);
>  
>  	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
> @@ -2706,23 +2704,22 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
>  						unsigned long *nr_scanned)
>  {
>  	struct scan_control sc = {
> -		.nr_scanned = 0,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
> -		.order = 0,
> -		.priority = 0,
> +		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> +		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
>  		.target_mem_cgroup = memcg,
> +		.flags = MAY_UNMAP,
>  	};
>  	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
>  	int swappiness = mem_cgroup_swappiness(memcg);
>  
> -	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> -			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (!noswap)
> +		sc.flags |= MAY_SWAP;
>  
>  	trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order,
> -						      sc.may_writepage,
> +						      sc.flags & MAY_WRITEPAGE,
>  						      sc.gfp_mask);
>  
>  	/*
> @@ -2748,18 +2745,19 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	unsigned long nr_reclaimed;
>  	int nid;
>  	struct scan_control sc = {
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = memcg,
> -		.nodemask = NULL, /* we don't care the placement */
>  		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> -				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		.target_mem_cgroup = memcg,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP,
>  	};
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (!noswap)
> +		sc.flags |= MAY_SWAP;
> +
>  	/*
>  	 * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
>  	 * take care of from where we get pages. So the node where we start the
> @@ -2770,7 +2768,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	zonelist = NODE_DATA(nid)->node_zonelists;
>  
>  	trace_mm_vmscan_memcg_reclaim_begin(0,
> -					    sc.may_writepage,
> +					    sc.flags & MAY_WRITEPAGE,
>  					    sc.gfp_mask);
>  
>  	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
> @@ -3015,15 +3013,15 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  	unsigned long nr_soft_scanned;
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
> -		.priority = DEF_PRIORITY,
> -		.may_unmap = 1,
> -		.may_swap = 1,
> -		.may_writepage = !laptop_mode,
>  		.order = order,
> -		.target_mem_cgroup = NULL,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP | MAY_SWAP,
>  	};
>  	count_vm_event(PAGEOUTRUN);
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +
>  	do {
>  		unsigned long lru_pages = 0;
>  		unsigned long nr_attempted = 0;
> @@ -3104,7 +3102,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  		 * even in laptop mode.
>  		 */
>  		if (sc.priority < DEF_PRIORITY - 2)
> -			sc.may_writepage = 1;
> +			sc.flags |= MAY_WRITEPAGE;
>  
>  		/*
>  		 * Now scan the zone in the dma->highmem direction, stopping
> @@ -3401,14 +3399,11 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
>  {
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.gfp_mask = GFP_HIGHUSER_MOVABLE,
> -		.may_swap = 1,
> -		.may_unmap = 1,
> -		.may_writepage = 1,
>  		.nr_to_reclaim = nr_to_reclaim,
> -		.hibernation_mode = 1,
> -		.order = 0,
> +		.gfp_mask = GFP_HIGHUSER_MOVABLE,
>  		.priority = DEF_PRIORITY,
> +		.flags = MAY_WRITEPAGE | MAY_UNMAP | MAY_SWAP |
> +		         MAY_SKIP_CONGESTION,
>  	};
>  	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
>  	struct task_struct *p = current;
> @@ -3588,19 +3583,22 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
>  	struct task_struct *p = current;
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> -		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> -		.may_swap = 1,
>  		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
>  		.priority = ZONE_RECLAIM_PRIORITY,
> +		.flags = MAY_SWAP,
>  	};
>  	struct shrink_control shrink = {
>  		.gfp_mask = sc.gfp_mask,
>  	};
>  	unsigned long nr_slab_pages0, nr_slab_pages1;
>  
> +	if (zone_reclaim_mode & RECLAIM_WRITE)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (zone_reclaim_mode & RECLAIM_SWAP)
> +		sc.flags |= MAY_UNMAP;
> +
>  	cond_resched();
>  	/*
>  	 * We need to be able to allocate from the reserves for RECLAIM_SWAP
> -- 
> 2.0.0

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Hugh Dickins]

On Mon, 14 Jul 2014, Johannes Weiner wrote:

> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into flags that fit into a single integer.
> 
> Stack delta: +72/-296 -224                   old     new   delta
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
>    text    data     bss     dec     hex filename
>   38151    5641      16   43808    ab20 mm/vmscan.o.old
>   38047    5641      16   43704    aab8 mm/vmscan.o
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
>  1 file changed, 78 insertions(+), 80 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c28b8981e56a..73d8e69ff3eb 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -58,36 +58,28 @@
>  #define CREATE_TRACE_POINTS
>  #include <trace/events/vmscan.h>
>  
> -struct scan_control {
> -	/* Incremented by the number of inactive pages that were scanned */
> -	unsigned long nr_scanned;
> -
> -	/* Number of pages freed so far during a call to shrink_zones() */
> -	unsigned long nr_reclaimed;
> -
> -	/* One of the zones is ready for compaction */
> -	int compaction_ready;
> +/* Scan control flags */
> +#define MAY_WRITEPAGE		0x1
> +#define MAY_UNMAP		0x2
> +#define MAY_SWAP		0x4
> +#define MAY_SKIP_CONGESTION	0x8
> +#define COMPACTION_READY	0x10
>  
> +struct scan_control {
>  	/* How many pages shrink_list() should reclaim */
>  	unsigned long nr_to_reclaim;
>  
> -	unsigned long hibernation_mode;
> -
>  	/* This context's GFP mask */
>  	gfp_t gfp_mask;
>  
> -	int may_writepage;
> -
> -	/* Can mapped pages be reclaimed? */
> -	int may_unmap;
> -
> -	/* Can pages be swapped as part of reclaim? */
> -	int may_swap;
> -
> +	/* Allocation order */
>  	int order;
>  
> -	/* Scan (total_size >> priority) pages at once */
> -	int priority;
> +	/*
> +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> +	 * are scanned.
> +	 */
> +	nodemask_t	*nodemask;
>  
>  	/*
>  	 * The memory cgroup that hit its limit and as a result is the
> @@ -95,11 +87,17 @@ struct scan_control {
>  	 */
>  	struct mem_cgroup *target_mem_cgroup;
>  
> -	/*
> -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> -	 * are scanned.
> -	 */
> -	nodemask_t	*nodemask;
> +	/* Scan (total_size >> priority) pages at once */
> +	int priority;
> +
> +	/* Scan control flags; see above */
> +	unsigned int flags;

This seems to result in a fair amount of unnecessary churn:
why not just put may_writepage etc into an unsigned int bitfield,
then you get the saving without changing all the rest of the code.

Hugh

> +
> +	/* Incremented by the number of inactive pages that were scanned */
> +	unsigned long nr_scanned;
> +
> +	/* Number of pages freed so far during a call to shrink_zones() */
> +	unsigned long nr_reclaimed;
>  };
>  
>  #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
> @@ -840,7 +838,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  		if (unlikely(!page_evictable(page)))
>  			goto cull_mlocked;
>  
> -		if (!sc->may_unmap && page_mapped(page))
> +		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))
>  			goto keep_locked;
>  
>  		/* Double the slab pressure for mapped and swapcache pages */
> @@ -1014,7 +1012,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
>  				goto keep_locked;
>  			if (!may_enter_fs)
>  				goto keep_locked;
> -			if (!sc->may_writepage)
> +			if (!(sc->flags & MAY_WRITEPAGE))
>  				goto keep_locked;
>  
>  			/* Page is dirty, try to write it out here */
> @@ -1146,7 +1144,7 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
>  		.priority = DEF_PRIORITY,
> -		.may_unmap = 1,
> +		.flags = MAY_UNMAP,
>  	};
>  	unsigned long ret, dummy1, dummy2, dummy3, dummy4, dummy5;
>  	struct page *page, *next;
> @@ -1489,9 +1487,9 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
>  
>  	lru_add_drain();
>  
> -	if (!sc->may_unmap)
> +	if (!(sc->flags & MAY_UNMAP))
>  		isolate_mode |= ISOLATE_UNMAPPED;
> -	if (!sc->may_writepage)
> +	if (!(sc->flags & MAY_WRITEPAGE))
>  		isolate_mode |= ISOLATE_CLEAN;
>  
>  	spin_lock_irq(&zone->lru_lock);
> @@ -1593,7 +1591,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
>  	 * is congested. Allow kswapd to continue until it starts encountering
>  	 * unqueued dirty pages or cycling through the LRU too quickly.
>  	 */
> -	if (!sc->hibernation_mode && !current_is_kswapd() &&
> +	if (!(sc->flags & MAY_SKIP_CONGESTION) && !current_is_kswapd() &&
>  	    current_may_throttle())
>  		wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10);
>  
> @@ -1683,9 +1681,9 @@ static void shrink_active_list(unsigned long nr_to_scan,
>  
>  	lru_add_drain();
>  
> -	if (!sc->may_unmap)
> +	if (!(sc->flags & MAY_UNMAP))
>  		isolate_mode |= ISOLATE_UNMAPPED;
> -	if (!sc->may_writepage)
> +	if (!(sc->flags & MAY_WRITEPAGE))
>  		isolate_mode |= ISOLATE_CLEAN;
>  
>  	spin_lock_irq(&zone->lru_lock);
> @@ -1897,7 +1895,7 @@ static void get_scan_count(struct lruvec *lruvec, int swappiness,
>  		force_scan = true;
>  
>  	/* If we have no swap space, do not bother scanning anon pages. */
> -	if (!sc->may_swap || (get_nr_swap_pages() <= 0)) {
> +	if (!(sc->flags & MAY_SWAP) || (get_nr_swap_pages() <= 0)) {
>  		scan_balance = SCAN_FILE;
>  		goto out;
>  	}
> @@ -2406,7 +2404,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  			    sc->order > PAGE_ALLOC_COSTLY_ORDER &&
>  			    zonelist_zone_idx(z) <= requested_highidx &&
>  			    compaction_ready(zone, sc->order)) {
> -				sc->compaction_ready = true;
> +				sc->flags |= COMPACTION_READY;
>  				continue;
>  			}
>  
> @@ -2496,7 +2494,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
>  			break;
>  
> -		if (sc->compaction_ready)
> +		if (sc->flags & COMPACTION_READY)
>  			break;
>  
>  		/*
> @@ -2504,7 +2502,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		 * writepage even in laptop mode.
>  		 */
>  		if (sc->priority < DEF_PRIORITY - 2)
> -			sc->may_writepage = 1;
> +			sc->flags |= MAY_WRITEPAGE;
>  
>  		/*
>  		 * Try to write back as many pages as we just scanned.  This
> @@ -2517,7 +2515,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		if (total_scanned > writeback_threshold) {
>  			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned,
>  						WB_REASON_TRY_TO_FREE_PAGES);
> -			sc->may_writepage = 1;
> +			sc->flags |= MAY_WRITEPAGE;
>  		}
>  	} while (--sc->priority >= 0);
>  
> @@ -2527,7 +2525,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>  		return sc->nr_reclaimed;
>  
>  	/* Aborted reclaim to try compaction? don't OOM, then */
> -	if (sc->compaction_ready)
> +	if (sc->flags & COMPACTION_READY)
>  		return 1;
>  
>  	/* Any of the zones still reclaimable?  Don't OOM. */
> @@ -2668,17 +2666,17 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  {
>  	unsigned long nr_reclaimed;
>  	struct scan_control sc = {
> -		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
> -		.may_writepage = !laptop_mode,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.may_unmap = 1,
> -		.may_swap = 1,
> +		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = NULL,
>  		.nodemask = nodemask,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP | MAY_SWAP,
>  	};
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +
>  	/*
>  	 * Do not enter reclaim if fatal signal was delivered while throttled.
>  	 * 1 is returned so that the page allocator does not OOM kill at this
> @@ -2688,7 +2686,7 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  		return 1;
>  
>  	trace_mm_vmscan_direct_reclaim_begin(order,
> -				sc.may_writepage,
> +				sc.flags & MAY_WRITEPAGE,
>  				gfp_mask);
>  
>  	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
> @@ -2706,23 +2704,22 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
>  						unsigned long *nr_scanned)
>  {
>  	struct scan_control sc = {
> -		.nr_scanned = 0,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
> -		.order = 0,
> -		.priority = 0,
> +		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> +		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
>  		.target_mem_cgroup = memcg,
> +		.flags = MAY_UNMAP,
>  	};
>  	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
>  	int swappiness = mem_cgroup_swappiness(memcg);
>  
> -	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> -			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (!noswap)
> +		sc.flags |= MAY_SWAP;
>  
>  	trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order,
> -						      sc.may_writepage,
> +						      sc.flags & MAY_WRITEPAGE,
>  						      sc.gfp_mask);
>  
>  	/*
> @@ -2748,18 +2745,19 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	unsigned long nr_reclaimed;
>  	int nid;
>  	struct scan_control sc = {
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = memcg,
> -		.nodemask = NULL, /* we don't care the placement */
>  		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
> -				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		            (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		.target_mem_cgroup = memcg,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP,
>  	};
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (!noswap)
> +		sc.flags |= MAY_SWAP;
> +
>  	/*
>  	 * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
>  	 * take care of from where we get pages. So the node where we start the
> @@ -2770,7 +2768,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	zonelist = NODE_DATA(nid)->node_zonelists;
>  
>  	trace_mm_vmscan_memcg_reclaim_begin(0,
> -					    sc.may_writepage,
> +					    sc.flags & MAY_WRITEPAGE,
>  					    sc.gfp_mask);
>  
>  	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
> @@ -3015,15 +3013,15 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  	unsigned long nr_soft_scanned;
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
> -		.priority = DEF_PRIORITY,
> -		.may_unmap = 1,
> -		.may_swap = 1,
> -		.may_writepage = !laptop_mode,
>  		.order = order,
> -		.target_mem_cgroup = NULL,
> +		.priority = DEF_PRIORITY,
> +		.flags = MAY_UNMAP | MAY_SWAP,
>  	};
>  	count_vm_event(PAGEOUTRUN);
>  
> +	if (!laptop_mode)
> +		sc.flags |= MAY_WRITEPAGE;
> +
>  	do {
>  		unsigned long lru_pages = 0;
>  		unsigned long nr_attempted = 0;
> @@ -3104,7 +3102,7 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  		 * even in laptop mode.
>  		 */
>  		if (sc.priority < DEF_PRIORITY - 2)
> -			sc.may_writepage = 1;
> +			sc.flags |= MAY_WRITEPAGE;
>  
>  		/*
>  		 * Now scan the zone in the dma->highmem direction, stopping
> @@ -3401,14 +3399,11 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
>  {
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.gfp_mask = GFP_HIGHUSER_MOVABLE,
> -		.may_swap = 1,
> -		.may_unmap = 1,
> -		.may_writepage = 1,
>  		.nr_to_reclaim = nr_to_reclaim,
> -		.hibernation_mode = 1,
> -		.order = 0,
> +		.gfp_mask = GFP_HIGHUSER_MOVABLE,
>  		.priority = DEF_PRIORITY,
> +		.flags = MAY_WRITEPAGE | MAY_UNMAP | MAY_SWAP |
> +		         MAY_SKIP_CONGESTION,
>  	};
>  	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
>  	struct task_struct *p = current;
> @@ -3588,19 +3583,22 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
>  	struct task_struct *p = current;
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> -		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> -		.may_swap = 1,
>  		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
>  		.priority = ZONE_RECLAIM_PRIORITY,
> +		.flags = MAY_SWAP,
>  	};
>  	struct shrink_control shrink = {
>  		.gfp_mask = sc.gfp_mask,
>  	};
>  	unsigned long nr_slab_pages0, nr_slab_pages1;
>  
> +	if (zone_reclaim_mode & RECLAIM_WRITE)
> +		sc.flags |= MAY_WRITEPAGE;
> +	if (zone_reclaim_mode & RECLAIM_SWAP)
> +		sc.flags |= MAY_UNMAP;
> +
>  	cond_resched();
>  	/*
>  	 * We need to be able to allocate from the reserves for RECLAIM_SWAP
> -- 
> 2.0.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Andrew Morton]

On Mon, 14 Jul 2014 09:20:49 -0400 Johannes Weiner <hannes@cmpxchg.org> wrote:

> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into flags that fit into a single integer.

bitfields would be a pretty good fit here.  I usually don't like them
because of locking concerns with the RMWs, but scan_control is never
accessed from another thread.

> -		if (!sc->may_unmap && page_mapped(page))
> +		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))

Then edits such as this are unneeded.

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Andrew Morton]

On Mon, 14 Jul 2014 09:20:49 -0400 Johannes Weiner <hannes@cmpxchg.org> wrote:

> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into flags that fit into a single integer.

bitfields would be a pretty good fit here.  I usually don't like them
because of locking concerns with the RMWs, but scan_control is never
accessed from another thread.

> -		if (!sc->may_unmap && page_mapped(page))
> +		if (!(sc->flags & MAY_UNMAP) && page_mapped(page))

Then edits such as this are unneeded.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Michal Hocko]

On Mon 14-07-14 09:20:48, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify the
> reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Looks good to me and it fits better with my low/min limit patches which I
hopefully post soon.

> ---
>  mm/vmscan.c | 29 +++++++++++++++--------------
>  1 file changed, 15 insertions(+), 14 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 6dac1310e5e4..74a9e0ae09b0 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -2244,10 +2244,10 @@ static inline bool should_continue_reclaim(struct zone *zone,
>  	}
>  }
>  
> -static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
> +static bool shrink_zone(struct zone *zone, struct scan_control *sc)
>  {
>  	unsigned long nr_reclaimed, nr_scanned;
> -	unsigned long zone_reclaimed = 0;
> +	bool reclaimable = false;
>  
>  	do {
>  		struct mem_cgroup *root = sc->target_mem_cgroup;
> @@ -2291,12 +2291,13 @@ static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
>  			   sc->nr_scanned - nr_scanned,
>  			   sc->nr_reclaimed - nr_reclaimed);
>  
> -		zone_reclaimed += sc->nr_reclaimed - nr_reclaimed;
> +		if (sc->nr_reclaimed - nr_reclaimed)
> +			reclaimable = true;
>  
>  	} while (should_continue_reclaim(zone, sc->nr_reclaimed - nr_reclaimed,
>  					 sc->nr_scanned - nr_scanned, sc));
>  
> -	return zone_reclaimed;
> +	return reclaimable;
>  }
>  
>  /* Returns true if compaction should go ahead for a high-order request */
> @@ -2346,7 +2347,7 @@ static inline bool compaction_ready(struct zone *zone, int order)
>   * If a zone is deemed to be full of pinned pages then just give it a light
>   * scan then give up on it.
>   *
> - * Returns whether the zones overall are reclaimable or not.
> + * Returns true if a zone was reclaimable.
>   */
>  static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  {
> @@ -2361,7 +2362,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  		.gfp_mask = sc->gfp_mask,
>  	};
>  	enum zone_type requested_highidx = gfp_zone(sc->gfp_mask);
> -	bool all_unreclaimable = true;
> +	bool reclaimable = false;
>  
>  	/*
>  	 * If the number of buffer_heads in the machine exceeds the maximum
> @@ -2376,8 +2377,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  
>  	for_each_zone_zonelist_nodemask(zone, z, zonelist,
>  					gfp_zone(sc->gfp_mask), sc->nodemask) {
> -		unsigned long zone_reclaimed = 0;
> -
>  		if (!populated_zone(zone))
>  			continue;
>  		/*
> @@ -2424,15 +2423,17 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  						&nr_soft_scanned);
>  			sc->nr_reclaimed += nr_soft_reclaimed;
>  			sc->nr_scanned += nr_soft_scanned;
> -			zone_reclaimed += nr_soft_reclaimed;
> +			if (nr_soft_reclaimed)
> +				reclaimable = true;
>  			/* need some check for avoid more shrink_zone() */
>  		}
>  
> -		zone_reclaimed += shrink_zone(zone, sc);
> +		if (shrink_zone(zone, sc))
> +			reclaimable = true;
>  
> -		if (zone_reclaimed ||
> -		    (global_reclaim(sc) && zone_reclaimable(zone)))
> -			all_unreclaimable = false;
> +		if (global_reclaim(sc) &&
> +		    !reclaimable && zone_reclaimable(zone))
> +			reclaimable = true;
>  	}
>  
>  	/*
> @@ -2455,7 +2456,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  	 */
>  	sc->gfp_mask = orig_mask;
>  
> -	return !all_unreclaimable;
> +	return reclaimable;
>  }
>  
>  /*
> -- 
> 2.0.0
> 

-- 
Michal Hocko
SUSE Labs

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Michal Hocko]

On Mon 14-07-14 09:20:48, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify the
> reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Looks good to me and it fits better with my low/min limit patches which I
hopefully post soon.

> ---
>  mm/vmscan.c | 29 +++++++++++++++--------------
>  1 file changed, 15 insertions(+), 14 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 6dac1310e5e4..74a9e0ae09b0 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -2244,10 +2244,10 @@ static inline bool should_continue_reclaim(struct zone *zone,
>  	}
>  }
>  
> -static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
> +static bool shrink_zone(struct zone *zone, struct scan_control *sc)
>  {
>  	unsigned long nr_reclaimed, nr_scanned;
> -	unsigned long zone_reclaimed = 0;
> +	bool reclaimable = false;
>  
>  	do {
>  		struct mem_cgroup *root = sc->target_mem_cgroup;
> @@ -2291,12 +2291,13 @@ static unsigned long shrink_zone(struct zone *zone, struct scan_control *sc)
>  			   sc->nr_scanned - nr_scanned,
>  			   sc->nr_reclaimed - nr_reclaimed);
>  
> -		zone_reclaimed += sc->nr_reclaimed - nr_reclaimed;
> +		if (sc->nr_reclaimed - nr_reclaimed)
> +			reclaimable = true;
>  
>  	} while (should_continue_reclaim(zone, sc->nr_reclaimed - nr_reclaimed,
>  					 sc->nr_scanned - nr_scanned, sc));
>  
> -	return zone_reclaimed;
> +	return reclaimable;
>  }
>  
>  /* Returns true if compaction should go ahead for a high-order request */
> @@ -2346,7 +2347,7 @@ static inline bool compaction_ready(struct zone *zone, int order)
>   * If a zone is deemed to be full of pinned pages then just give it a light
>   * scan then give up on it.
>   *
> - * Returns whether the zones overall are reclaimable or not.
> + * Returns true if a zone was reclaimable.
>   */
>  static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  {
> @@ -2361,7 +2362,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  		.gfp_mask = sc->gfp_mask,
>  	};
>  	enum zone_type requested_highidx = gfp_zone(sc->gfp_mask);
> -	bool all_unreclaimable = true;
> +	bool reclaimable = false;
>  
>  	/*
>  	 * If the number of buffer_heads in the machine exceeds the maximum
> @@ -2376,8 +2377,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  
>  	for_each_zone_zonelist_nodemask(zone, z, zonelist,
>  					gfp_zone(sc->gfp_mask), sc->nodemask) {
> -		unsigned long zone_reclaimed = 0;
> -
>  		if (!populated_zone(zone))
>  			continue;
>  		/*
> @@ -2424,15 +2423,17 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  						&nr_soft_scanned);
>  			sc->nr_reclaimed += nr_soft_reclaimed;
>  			sc->nr_scanned += nr_soft_scanned;
> -			zone_reclaimed += nr_soft_reclaimed;
> +			if (nr_soft_reclaimed)
> +				reclaimable = true;
>  			/* need some check for avoid more shrink_zone() */
>  		}
>  
> -		zone_reclaimed += shrink_zone(zone, sc);
> +		if (shrink_zone(zone, sc))
> +			reclaimable = true;
>  
> -		if (zone_reclaimed ||
> -		    (global_reclaim(sc) && zone_reclaimable(zone)))
> -			all_unreclaimable = false;
> +		if (global_reclaim(sc) &&
> +		    !reclaimable && zone_reclaimable(zone))
> +			reclaimable = true;
>  	}
>  
>  	/*
> @@ -2455,7 +2456,7 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
>  	 */
>  	sc->gfp_mask = orig_mask;
>  
> -	return !all_unreclaimable;
> +	return reclaimable;
>  }
>  
>  /*
> -- 
> 2.0.0
> 

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Johannes Weiner]

On Mon, Jul 14, 2014 at 12:46:21PM -0700, Hugh Dickins wrote:
> On Mon, 14 Jul 2014, Johannes Weiner wrote:
> 
> > Reorder the members by input and output, then turn the individual
> > integers for may_writepage, may_unmap, may_swap, compaction_ready,
> > hibernation_mode into flags that fit into a single integer.
> > 
> > Stack delta: +72/-296 -224                   old     new   delta
> > kswapd                                       104     176     +72
> > try_to_free_pages                             80      56     -24
> > try_to_free_mem_cgroup_pages                  80      56     -24
> > shrink_all_memory                             88      64     -24
> > reclaim_clean_pages_from_list                168     144     -24
> > mem_cgroup_shrink_node_zone                  104      80     -24
> > __zone_reclaim                               176     152     -24
> > balance_pgdat                                152       -    -152
> > 
> >    text    data     bss     dec     hex filename
> >   38151    5641      16   43808    ab20 mm/vmscan.o.old
> >   38047    5641      16   43704    aab8 mm/vmscan.o
> > 
> > Suggested-by: Mel Gorman <mgorman@suse.de>
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> > ---
> >  mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
> >  1 file changed, 78 insertions(+), 80 deletions(-)
> > 
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index c28b8981e56a..73d8e69ff3eb 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -58,36 +58,28 @@
> >  #define CREATE_TRACE_POINTS
> >  #include <trace/events/vmscan.h>
> >  
> > -struct scan_control {
> > -	/* Incremented by the number of inactive pages that were scanned */
> > -	unsigned long nr_scanned;
> > -
> > -	/* Number of pages freed so far during a call to shrink_zones() */
> > -	unsigned long nr_reclaimed;
> > -
> > -	/* One of the zones is ready for compaction */
> > -	int compaction_ready;
> > +/* Scan control flags */
> > +#define MAY_WRITEPAGE		0x1
> > +#define MAY_UNMAP		0x2
> > +#define MAY_SWAP		0x4
> > +#define MAY_SKIP_CONGESTION	0x8
> > +#define COMPACTION_READY	0x10
> >  
> > +struct scan_control {
> >  	/* How many pages shrink_list() should reclaim */
> >  	unsigned long nr_to_reclaim;
> >  
> > -	unsigned long hibernation_mode;
> > -
> >  	/* This context's GFP mask */
> >  	gfp_t gfp_mask;
> >  
> > -	int may_writepage;
> > -
> > -	/* Can mapped pages be reclaimed? */
> > -	int may_unmap;
> > -
> > -	/* Can pages be swapped as part of reclaim? */
> > -	int may_swap;
> > -
> > +	/* Allocation order */
> >  	int order;
> >  
> > -	/* Scan (total_size >> priority) pages at once */
> > -	int priority;
> > +	/*
> > +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> > +	 * are scanned.
> > +	 */
> > +	nodemask_t	*nodemask;
> >  
> >  	/*
> >  	 * The memory cgroup that hit its limit and as a result is the
> > @@ -95,11 +87,17 @@ struct scan_control {
> >  	 */
> >  	struct mem_cgroup *target_mem_cgroup;
> >  
> > -	/*
> > -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> > -	 * are scanned.
> > -	 */
> > -	nodemask_t	*nodemask;
> > +	/* Scan (total_size >> priority) pages at once */
> > +	int priority;
> > +
> > +	/* Scan control flags; see above */
> > +	unsigned int flags;
> 
> This seems to result in a fair amount of unnecessary churn:
> why not just put may_writepage etc into an unsigned int bitfield,
> then you get the saving without changing all the rest of the code.

Good point, I didn't even think of that.  Thanks!

Andrew, could you please replace this patch with the following?

---
>From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
From: Johannes Weiner <hannes@cmpxchg.org>
Date: Mon, 14 Jul 2014 08:51:54 -0400
Subject: [patch] mm: vmscan: clean up struct scan_control

Reorder the members by input and output, then turn the individual
integers for may_writepage, may_unmap, may_swap, compaction_ready,
hibernation_mode into bit fields to save stack space:

+72/-296 -224
kswapd                                       104     176     +72
try_to_free_pages                             80      56     -24
try_to_free_mem_cgroup_pages                  80      56     -24
shrink_all_memory                             88      64     -24
reclaim_clean_pages_from_list                168     144     -24
mem_cgroup_shrink_node_zone                  104      80     -24
__zone_reclaim                               176     152     -24
balance_pgdat                                152       -    -152

Suggested-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmscan.c | 99 ++++++++++++++++++++++++++++---------------------------------
 1 file changed, 46 insertions(+), 53 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index c28b8981e56a..81dd858b9d17 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -59,35 +59,20 @@
 #include <trace/events/vmscan.h>
 
 struct scan_control {
-	/* Incremented by the number of inactive pages that were scanned */
-	unsigned long nr_scanned;
-
-	/* Number of pages freed so far during a call to shrink_zones() */
-	unsigned long nr_reclaimed;
-
-	/* One of the zones is ready for compaction */
-	int compaction_ready;
-
 	/* How many pages shrink_list() should reclaim */
 	unsigned long nr_to_reclaim;
 
-	unsigned long hibernation_mode;
-
 	/* This context's GFP mask */
 	gfp_t gfp_mask;
 
-	int may_writepage;
-
-	/* Can mapped pages be reclaimed? */
-	int may_unmap;
-
-	/* Can pages be swapped as part of reclaim? */
-	int may_swap;
-
+	/* Allocation order */
 	int order;
 
-	/* Scan (total_size >> priority) pages at once */
-	int priority;
+	/*
+	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
+	 * are scanned.
+	 */
+	nodemask_t	*nodemask;
 
 	/*
 	 * The memory cgroup that hit its limit and as a result is the
@@ -95,11 +80,27 @@ struct scan_control {
 	 */
 	struct mem_cgroup *target_mem_cgroup;
 
-	/*
-	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
-	 * are scanned.
-	 */
-	nodemask_t	*nodemask;
+	/* Scan (total_size >> priority) pages at once */
+	int priority;
+
+	unsigned int may_writepage:1;
+
+	/* Can mapped pages be reclaimed? */
+	unsigned int may_unmap:1;
+
+	/* Can pages be swapped as part of reclaim? */
+	unsigned int may_swap:1;
+
+	unsigned int hibernation_mode:1;
+
+	/* One of the zones is ready for compaction */
+	unsigned int compaction_ready:1;
+
+	/* Incremented by the number of inactive pages that were scanned */
+	unsigned long nr_scanned;
+
+	/* Number of pages freed so far during a call to shrink_zones() */
+	unsigned long nr_reclaimed;
 };
 
 #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
@@ -2668,15 +2669,14 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 {
 	unsigned long nr_reclaimed;
 	struct scan_control sc = {
+		.nr_to_reclaim = SWAP_CLUSTER_MAX,
 		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
+		.order = order,
+		.nodemask = nodemask,
+		.priority = DEF_PRIORITY,
 		.may_writepage = !laptop_mode,
-		.nr_to_reclaim = SWAP_CLUSTER_MAX,
 		.may_unmap = 1,
 		.may_swap = 1,
-		.order = order,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = NULL,
-		.nodemask = nodemask,
 	};
 
 	/*
@@ -2706,14 +2706,11 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
 						unsigned long *nr_scanned)
 {
 	struct scan_control sc = {
-		.nr_scanned = 0,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
+		.target_mem_cgroup = memcg,
 		.may_writepage = !laptop_mode,
 		.may_unmap = 1,
 		.may_swap = !noswap,
-		.order = 0,
-		.priority = 0,
-		.target_mem_cgroup = memcg,
 	};
 	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
 	int swappiness = mem_cgroup_swappiness(memcg);
@@ -2748,16 +2745,14 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 	unsigned long nr_reclaimed;
 	int nid;
 	struct scan_control sc = {
-		.may_writepage = !laptop_mode,
-		.may_unmap = 1,
-		.may_swap = !noswap,
 		.nr_to_reclaim = SWAP_CLUSTER_MAX,
-		.order = 0,
-		.priority = DEF_PRIORITY,
-		.target_mem_cgroup = memcg,
-		.nodemask = NULL, /* we don't care the placement */
 		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
 				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
+		.target_mem_cgroup = memcg,
+		.priority = DEF_PRIORITY,
+		.may_writepage = !laptop_mode,
+		.may_unmap = 1,
+		.may_swap = !noswap,
 	};
 
 	/*
@@ -3015,12 +3010,11 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
 	unsigned long nr_soft_scanned;
 	struct scan_control sc = {
 		.gfp_mask = GFP_KERNEL,
+		.order = order,
 		.priority = DEF_PRIORITY,
+		.may_writepage = !laptop_mode,
 		.may_unmap = 1,
 		.may_swap = 1,
-		.may_writepage = !laptop_mode,
-		.order = order,
-		.target_mem_cgroup = NULL,
 	};
 	count_vm_event(PAGEOUTRUN);
 
@@ -3401,14 +3395,13 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
 {
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
+		.nr_to_reclaim = nr_to_reclaim,
 		.gfp_mask = GFP_HIGHUSER_MOVABLE,
-		.may_swap = 1,
-		.may_unmap = 1,
+		.priority = DEF_PRIORITY,
 		.may_writepage = 1,
-		.nr_to_reclaim = nr_to_reclaim,
+		.may_unmap = 1,
+		.may_swap = 1,
 		.hibernation_mode = 1,
-		.order = 0,
-		.priority = DEF_PRIORITY,
 	};
 	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
 	struct task_struct *p = current;
@@ -3588,13 +3581,13 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 	struct task_struct *p = current;
 	struct reclaim_state reclaim_state;
 	struct scan_control sc = {
-		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
-		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
-		.may_swap = 1,
 		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
 		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
 		.order = order,
 		.priority = ZONE_RECLAIM_PRIORITY,
+		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
+		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
+		.may_swap = 1,
 	};
 	struct shrink_control shrink = {
 		.gfp_mask = sc.gfp_mask,
-- 
2.0.0

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Johannes Weiner]

On Mon, Jul 14, 2014 at 12:46:21PM -0700, Hugh Dickins wrote:
> On Mon, 14 Jul 2014, Johannes Weiner wrote:
> 
> > Reorder the members by input and output, then turn the individual
> > integers for may_writepage, may_unmap, may_swap, compaction_ready,
> > hibernation_mode into flags that fit into a single integer.
> > 
> > Stack delta: +72/-296 -224                   old     new   delta
> > kswapd                                       104     176     +72
> > try_to_free_pages                             80      56     -24
> > try_to_free_mem_cgroup_pages                  80      56     -24
> > shrink_all_memory                             88      64     -24
> > reclaim_clean_pages_from_list                168     144     -24
> > mem_cgroup_shrink_node_zone                  104      80     -24
> > __zone_reclaim                               176     152     -24
> > balance_pgdat                                152       -    -152
> > 
> >    text    data     bss     dec     hex filename
> >   38151    5641      16   43808    ab20 mm/vmscan.o.old
> >   38047    5641      16   43704    aab8 mm/vmscan.o
> > 
> > Suggested-by: Mel Gorman <mgorman@suse.de>
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> > ---
> >  mm/vmscan.c | 158 ++++++++++++++++++++++++++++++------------------------------
> >  1 file changed, 78 insertions(+), 80 deletions(-)
> > 
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index c28b8981e56a..73d8e69ff3eb 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -58,36 +58,28 @@
> >  #define CREATE_TRACE_POINTS
> >  #include <trace/events/vmscan.h>
> >  
> > -struct scan_control {
> > -	/* Incremented by the number of inactive pages that were scanned */
> > -	unsigned long nr_scanned;
> > -
> > -	/* Number of pages freed so far during a call to shrink_zones() */
> > -	unsigned long nr_reclaimed;
> > -
> > -	/* One of the zones is ready for compaction */
> > -	int compaction_ready;
> > +/* Scan control flags */
> > +#define MAY_WRITEPAGE		0x1
> > +#define MAY_UNMAP		0x2
> > +#define MAY_SWAP		0x4
> > +#define MAY_SKIP_CONGESTION	0x8
> > +#define COMPACTION_READY	0x10
> >  
> > +struct scan_control {
> >  	/* How many pages shrink_list() should reclaim */
> >  	unsigned long nr_to_reclaim;
> >  
> > -	unsigned long hibernation_mode;
> > -
> >  	/* This context's GFP mask */
> >  	gfp_t gfp_mask;
> >  
> > -	int may_writepage;
> > -
> > -	/* Can mapped pages be reclaimed? */
> > -	int may_unmap;
> > -
> > -	/* Can pages be swapped as part of reclaim? */
> > -	int may_swap;
> > -
> > +	/* Allocation order */
> >  	int order;
> >  
> > -	/* Scan (total_size >> priority) pages at once */
> > -	int priority;
> > +	/*
> > +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> > +	 * are scanned.
> > +	 */
> > +	nodemask_t	*nodemask;
> >  
> >  	/*
> >  	 * The memory cgroup that hit its limit and as a result is the
> > @@ -95,11 +87,17 @@ struct scan_control {
> >  	 */
> >  	struct mem_cgroup *target_mem_cgroup;
> >  
> > -	/*
> > -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> > -	 * are scanned.
> > -	 */
> > -	nodemask_t	*nodemask;
> > +	/* Scan (total_size >> priority) pages at once */
> > +	int priority;
> > +
> > +	/* Scan control flags; see above */
> > +	unsigned int flags;
> 
> This seems to result in a fair amount of unnecessary churn:
> why not just put may_writepage etc into an unsigned int bitfield,
> then you get the saving without changing all the rest of the code.

Good point, I didn't even think of that.  Thanks!

Andrew, could you please replace this patch with the following?

---

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Michal Hocko]

On Thu 17-07-14 09:26:04, Johannes Weiner wrote:
> From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Mon, 14 Jul 2014 08:51:54 -0400
> Subject: [patch] mm: vmscan: clean up struct scan_control
> 
> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into bit fields to save stack space:
> 
> +72/-296 -224
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Looks nice to me.
Acked-by: Michal Hocko <mhocko@suse.cz>

> ---
>  mm/vmscan.c | 99 ++++++++++++++++++++++++++++---------------------------------
>  1 file changed, 46 insertions(+), 53 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c28b8981e56a..81dd858b9d17 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -59,35 +59,20 @@
>  #include <trace/events/vmscan.h>
>  
>  struct scan_control {
> -	/* Incremented by the number of inactive pages that were scanned */
> -	unsigned long nr_scanned;
> -
> -	/* Number of pages freed so far during a call to shrink_zones() */
> -	unsigned long nr_reclaimed;
> -
> -	/* One of the zones is ready for compaction */
> -	int compaction_ready;
> -
>  	/* How many pages shrink_list() should reclaim */
>  	unsigned long nr_to_reclaim;
>  
> -	unsigned long hibernation_mode;
> -
>  	/* This context's GFP mask */
>  	gfp_t gfp_mask;
>  
> -	int may_writepage;
> -
> -	/* Can mapped pages be reclaimed? */
> -	int may_unmap;
> -
> -	/* Can pages be swapped as part of reclaim? */
> -	int may_swap;
> -
> +	/* Allocation order */
>  	int order;
>  
> -	/* Scan (total_size >> priority) pages at once */
> -	int priority;
> +	/*
> +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> +	 * are scanned.
> +	 */
> +	nodemask_t	*nodemask;
>  
>  	/*
>  	 * The memory cgroup that hit its limit and as a result is the
> @@ -95,11 +80,27 @@ struct scan_control {
>  	 */
>  	struct mem_cgroup *target_mem_cgroup;
>  
> -	/*
> -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> -	 * are scanned.
> -	 */
> -	nodemask_t	*nodemask;
> +	/* Scan (total_size >> priority) pages at once */
> +	int priority;
> +
> +	unsigned int may_writepage:1;
> +
> +	/* Can mapped pages be reclaimed? */
> +	unsigned int may_unmap:1;
> +
> +	/* Can pages be swapped as part of reclaim? */
> +	unsigned int may_swap:1;
> +
> +	unsigned int hibernation_mode:1;
> +
> +	/* One of the zones is ready for compaction */
> +	unsigned int compaction_ready:1;
> +
> +	/* Incremented by the number of inactive pages that were scanned */
> +	unsigned long nr_scanned;
> +
> +	/* Number of pages freed so far during a call to shrink_zones() */
> +	unsigned long nr_reclaimed;
>  };
>  
>  #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
> @@ -2668,15 +2669,14 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  {
>  	unsigned long nr_reclaimed;
>  	struct scan_control sc = {
> +		.nr_to_reclaim = SWAP_CLUSTER_MAX,
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
> +		.order = order,
> +		.nodemask = nodemask,
> +		.priority = DEF_PRIORITY,
>  		.may_writepage = !laptop_mode,
> -		.nr_to_reclaim = SWAP_CLUSTER_MAX,
>  		.may_unmap = 1,
>  		.may_swap = 1,
> -		.order = order,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = NULL,
> -		.nodemask = nodemask,
>  	};
>  
>  	/*
> @@ -2706,14 +2706,11 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
>  						unsigned long *nr_scanned)
>  {
>  	struct scan_control sc = {
> -		.nr_scanned = 0,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> +		.target_mem_cgroup = memcg,
>  		.may_writepage = !laptop_mode,
>  		.may_unmap = 1,
>  		.may_swap = !noswap,
> -		.order = 0,
> -		.priority = 0,
> -		.target_mem_cgroup = memcg,
>  	};
>  	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
>  	int swappiness = mem_cgroup_swappiness(memcg);
> @@ -2748,16 +2745,14 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	unsigned long nr_reclaimed;
>  	int nid;
>  	struct scan_control sc = {
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = memcg,
> -		.nodemask = NULL, /* we don't care the placement */
>  		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
>  				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		.target_mem_cgroup = memcg,
> +		.priority = DEF_PRIORITY,
> +		.may_writepage = !laptop_mode,
> +		.may_unmap = 1,
> +		.may_swap = !noswap,
>  	};
>  
>  	/*
> @@ -3015,12 +3010,11 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  	unsigned long nr_soft_scanned;
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
> +		.order = order,
>  		.priority = DEF_PRIORITY,
> +		.may_writepage = !laptop_mode,
>  		.may_unmap = 1,
>  		.may_swap = 1,
> -		.may_writepage = !laptop_mode,
> -		.order = order,
> -		.target_mem_cgroup = NULL,
>  	};
>  	count_vm_event(PAGEOUTRUN);
>  
> @@ -3401,14 +3395,13 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
>  {
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> +		.nr_to_reclaim = nr_to_reclaim,
>  		.gfp_mask = GFP_HIGHUSER_MOVABLE,
> -		.may_swap = 1,
> -		.may_unmap = 1,
> +		.priority = DEF_PRIORITY,
>  		.may_writepage = 1,
> -		.nr_to_reclaim = nr_to_reclaim,
> +		.may_unmap = 1,
> +		.may_swap = 1,
>  		.hibernation_mode = 1,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
>  	};
>  	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
>  	struct task_struct *p = current;
> @@ -3588,13 +3581,13 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
>  	struct task_struct *p = current;
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> -		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> -		.may_swap = 1,
>  		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
>  		.priority = ZONE_RECLAIM_PRIORITY,
> +		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> +		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> +		.may_swap = 1,
>  	};
>  	struct shrink_control shrink = {
>  		.gfp_mask = sc.gfp_mask,
> -- 
> 2.0.0
> 

-- 
Michal Hocko
SUSE Labs

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Michal Hocko]

On Thu 17-07-14 09:26:04, Johannes Weiner wrote:
> From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Mon, 14 Jul 2014 08:51:54 -0400
> Subject: [patch] mm: vmscan: clean up struct scan_control
> 
> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into bit fields to save stack space:
> 
> +72/-296 -224
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Looks nice to me.
Acked-by: Michal Hocko <mhocko@suse.cz>

> ---
>  mm/vmscan.c | 99 ++++++++++++++++++++++++++++---------------------------------
>  1 file changed, 46 insertions(+), 53 deletions(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c28b8981e56a..81dd858b9d17 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -59,35 +59,20 @@
>  #include <trace/events/vmscan.h>
>  
>  struct scan_control {
> -	/* Incremented by the number of inactive pages that were scanned */
> -	unsigned long nr_scanned;
> -
> -	/* Number of pages freed so far during a call to shrink_zones() */
> -	unsigned long nr_reclaimed;
> -
> -	/* One of the zones is ready for compaction */
> -	int compaction_ready;
> -
>  	/* How many pages shrink_list() should reclaim */
>  	unsigned long nr_to_reclaim;
>  
> -	unsigned long hibernation_mode;
> -
>  	/* This context's GFP mask */
>  	gfp_t gfp_mask;
>  
> -	int may_writepage;
> -
> -	/* Can mapped pages be reclaimed? */
> -	int may_unmap;
> -
> -	/* Can pages be swapped as part of reclaim? */
> -	int may_swap;
> -
> +	/* Allocation order */
>  	int order;
>  
> -	/* Scan (total_size >> priority) pages at once */
> -	int priority;
> +	/*
> +	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> +	 * are scanned.
> +	 */
> +	nodemask_t	*nodemask;
>  
>  	/*
>  	 * The memory cgroup that hit its limit and as a result is the
> @@ -95,11 +80,27 @@ struct scan_control {
>  	 */
>  	struct mem_cgroup *target_mem_cgroup;
>  
> -	/*
> -	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
> -	 * are scanned.
> -	 */
> -	nodemask_t	*nodemask;
> +	/* Scan (total_size >> priority) pages at once */
> +	int priority;
> +
> +	unsigned int may_writepage:1;
> +
> +	/* Can mapped pages be reclaimed? */
> +	unsigned int may_unmap:1;
> +
> +	/* Can pages be swapped as part of reclaim? */
> +	unsigned int may_swap:1;
> +
> +	unsigned int hibernation_mode:1;
> +
> +	/* One of the zones is ready for compaction */
> +	unsigned int compaction_ready:1;
> +
> +	/* Incremented by the number of inactive pages that were scanned */
> +	unsigned long nr_scanned;
> +
> +	/* Number of pages freed so far during a call to shrink_zones() */
> +	unsigned long nr_reclaimed;
>  };
>  
>  #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
> @@ -2668,15 +2669,14 @@ unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>  {
>  	unsigned long nr_reclaimed;
>  	struct scan_control sc = {
> +		.nr_to_reclaim = SWAP_CLUSTER_MAX,
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
> +		.order = order,
> +		.nodemask = nodemask,
> +		.priority = DEF_PRIORITY,
>  		.may_writepage = !laptop_mode,
> -		.nr_to_reclaim = SWAP_CLUSTER_MAX,
>  		.may_unmap = 1,
>  		.may_swap = 1,
> -		.order = order,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = NULL,
> -		.nodemask = nodemask,
>  	};
>  
>  	/*
> @@ -2706,14 +2706,11 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg,
>  						unsigned long *nr_scanned)
>  {
>  	struct scan_control sc = {
> -		.nr_scanned = 0,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> +		.target_mem_cgroup = memcg,
>  		.may_writepage = !laptop_mode,
>  		.may_unmap = 1,
>  		.may_swap = !noswap,
> -		.order = 0,
> -		.priority = 0,
> -		.target_mem_cgroup = memcg,
>  	};
>  	struct lruvec *lruvec = mem_cgroup_zone_lruvec(zone, memcg);
>  	int swappiness = mem_cgroup_swappiness(memcg);
> @@ -2748,16 +2745,14 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  	unsigned long nr_reclaimed;
>  	int nid;
>  	struct scan_control sc = {
> -		.may_writepage = !laptop_mode,
> -		.may_unmap = 1,
> -		.may_swap = !noswap,
>  		.nr_to_reclaim = SWAP_CLUSTER_MAX,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
> -		.target_mem_cgroup = memcg,
> -		.nodemask = NULL, /* we don't care the placement */
>  		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
>  				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
> +		.target_mem_cgroup = memcg,
> +		.priority = DEF_PRIORITY,
> +		.may_writepage = !laptop_mode,
> +		.may_unmap = 1,
> +		.may_swap = !noswap,
>  	};
>  
>  	/*
> @@ -3015,12 +3010,11 @@ static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
>  	unsigned long nr_soft_scanned;
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
> +		.order = order,
>  		.priority = DEF_PRIORITY,
> +		.may_writepage = !laptop_mode,
>  		.may_unmap = 1,
>  		.may_swap = 1,
> -		.may_writepage = !laptop_mode,
> -		.order = order,
> -		.target_mem_cgroup = NULL,
>  	};
>  	count_vm_event(PAGEOUTRUN);
>  
> @@ -3401,14 +3395,13 @@ unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
>  {
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> +		.nr_to_reclaim = nr_to_reclaim,
>  		.gfp_mask = GFP_HIGHUSER_MOVABLE,
> -		.may_swap = 1,
> -		.may_unmap = 1,
> +		.priority = DEF_PRIORITY,
>  		.may_writepage = 1,
> -		.nr_to_reclaim = nr_to_reclaim,
> +		.may_unmap = 1,
> +		.may_swap = 1,
>  		.hibernation_mode = 1,
> -		.order = 0,
> -		.priority = DEF_PRIORITY,
>  	};
>  	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
>  	struct task_struct *p = current;
> @@ -3588,13 +3581,13 @@ static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
>  	struct task_struct *p = current;
>  	struct reclaim_state reclaim_state;
>  	struct scan_control sc = {
> -		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> -		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> -		.may_swap = 1,
>  		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
>  		.gfp_mask = (gfp_mask = memalloc_noio_flags(gfp_mask)),
>  		.order = order,
>  		.priority = ZONE_RECLAIM_PRIORITY,
> +		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
> +		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> +		.may_swap = 1,
>  	};
>  	struct shrink_control shrink = {
>  		.gfp_mask = sc.gfp_mask,
> -- 
> 2.0.0
> 

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Hugh Dickins]

On Thu, 17 Jul 2014, Michal Hocko wrote:
> On Thu 17-07-14 09:26:04, Johannes Weiner wrote:
> > From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> > From: Johannes Weiner <hannes@cmpxchg.org>
> > Date: Mon, 14 Jul 2014 08:51:54 -0400
> > Subject: [patch] mm: vmscan: clean up struct scan_control
> > 
> > Reorder the members by input and output, then turn the individual
> > integers for may_writepage, may_unmap, may_swap, compaction_ready,
> > hibernation_mode into bit fields to save stack space:
> > 
> > +72/-296 -224
> > kswapd                                       104     176     +72
> > try_to_free_pages                             80      56     -24
> > try_to_free_mem_cgroup_pages                  80      56     -24
> > shrink_all_memory                             88      64     -24
> > reclaim_clean_pages_from_list                168     144     -24
> > mem_cgroup_shrink_node_zone                  104      80     -24
> > __zone_reclaim                               176     152     -24
> > balance_pgdat                                152       -    -152
> > 
> > Suggested-by: Mel Gorman <mgorman@suse.de>
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> 
> Looks nice to me.
> Acked-by: Michal Hocko <mhocko@suse.cz>

Yes, looks nice to me too; and I agree that it was worthwhile to make
those initialization orders consistent, and drop the 0 initializations.

Acked-by: Hugh Dickins <hughd@google.com>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Hugh Dickins]

On Thu, 17 Jul 2014, Michal Hocko wrote:
> On Thu 17-07-14 09:26:04, Johannes Weiner wrote:
> > From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> > From: Johannes Weiner <hannes@cmpxchg.org>
> > Date: Mon, 14 Jul 2014 08:51:54 -0400
> > Subject: [patch] mm: vmscan: clean up struct scan_control
> > 
> > Reorder the members by input and output, then turn the individual
> > integers for may_writepage, may_unmap, may_swap, compaction_ready,
> > hibernation_mode into bit fields to save stack space:
> > 
> > +72/-296 -224
> > kswapd                                       104     176     +72
> > try_to_free_pages                             80      56     -24
> > try_to_free_mem_cgroup_pages                  80      56     -24
> > shrink_all_memory                             88      64     -24
> > reclaim_clean_pages_from_list                168     144     -24
> > mem_cgroup_shrink_node_zone                  104      80     -24
> > __zone_reclaim                               176     152     -24
> > balance_pgdat                                152       -    -152
> > 
> > Suggested-by: Mel Gorman <mgorman@suse.de>
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> 
> Looks nice to me.
> Acked-by: Michal Hocko <mhocko@suse.cz>

Yes, looks nice to me too; and I agree that it was worthwhile to make
those initialization orders consistent, and drop the 0 initializations.

Acked-by: Hugh Dickins <hughd@google.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Mel Gorman]

On Mon, Jul 14, 2014 at 09:20:47AM -0400, Johannes Weiner wrote:
> As per Mel, replace out label with breaks from the loop.
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

Re: [patch 1/3] mm: vmscan: rework compaction-ready signaling in direct reclaim fix

[Mel Gorman]

On Mon, Jul 14, 2014 at 09:20:47AM -0400, Johannes Weiner wrote:
> As per Mel, replace out label with breaks from the loop.
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Mel Gorman]

On Mon, Jul 14, 2014 at 09:20:48AM -0400, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify the
> reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

Re: [patch 2/3] mm: vmscan: remove all_unreclaimable() fix

[Mel Gorman]

On Mon, Jul 14, 2014 at 09:20:48AM -0400, Johannes Weiner wrote:
> As per Mel, use bool for reclaimability throughout and simplify the
> reclaimability tracking in shrink_zones().
> 
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Mel Gorman]

On Thu, Jul 17, 2014 at 09:26:04AM -0400, Johannes Weiner wrote:
> <SNIP>
> 
> Andrew, could you please replace this patch with the following?
> 
> ---
> From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Mon, 14 Jul 2014 08:51:54 -0400
> Subject: [patch] mm: vmscan: clean up struct scan_control
> 
> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into bit fields to save stack space:
> 
> +72/-296 -224
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

Re: [patch 3/3] mm: vmscan: clean up struct scan_control

[Mel Gorman]

On Thu, Jul 17, 2014 at 09:26:04AM -0400, Johannes Weiner wrote:
> <SNIP>
> 
> Andrew, could you please replace this patch with the following?
> 
> ---
> From bbe8c1645c77297a96ecd5d64d659ddcd6984d03 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Mon, 14 Jul 2014 08:51:54 -0400
> Subject: [patch] mm: vmscan: clean up struct scan_control
> 
> Reorder the members by input and output, then turn the individual
> integers for may_writepage, may_unmap, may_swap, compaction_ready,
> hibernation_mode into bit fields to save stack space:
> 
> +72/-296 -224
> kswapd                                       104     176     +72
> try_to_free_pages                             80      56     -24
> try_to_free_mem_cgroup_pages                  80      56     -24
> shrink_all_memory                             88      64     -24
> reclaim_clean_pages_from_list                168     144     -24
> mem_cgroup_shrink_node_zone                  104      80     -24
> __zone_reclaim                               176     152     -24
> balance_pgdat                                152       -    -152
> 
> Suggested-by: Mel Gorman <mgorman@suse.de>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>