Skip to contentSkip to conversation

memcg reclaim tracepoint v2

PATCH In review KOSAKI Motohiro 3 participants 11 messages Jul 26, 2010

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

cover memcg reclaim tracepoint v2

Recently, Mel Gorman added some vmscan tracepoint. but they can't trace memcg. So, This patch series does.

Changelog since v1
o drop all memcg bugfix and cleanups

diffstat
=========================
KOSAKI Motohiro (4):
vmscan: convert direct reclaim tracepoint to DEFINE_TRACE memcg, vmscan: add memcg reclaim tracepoint
vmscan: convert mm_vmscan_lru_isolate to DEFINE_EVENT
memcg: add mm_vmscan_memcg_isolate tracepoint

 include/trace/events/vmscan.h |   79 +++++++++++++++++++++++++++++++++++++++--
 mm/memcontrol.c               |    6 +++
 mm/vmscan.c                   |   20 ++++++++++-
 3 files changed, 101 insertions(+), 4 deletions(-)

Sameple output is here.
=========================

              dd-1851  [001]   158.837763: mm_vmscan_memcg_reclaim_begin: order=0 may_writepage=1 gfp_flags=GFP_HIGHUSER_MOVABLE
              dd-1851  [001]   158.837783: mm_vmscan_memcg_isolate: isolate_mode=0 order=0 nr_requested=32 nr_scanned=32 nr_taken=32 contig_taken=0 contig_dirty=0 contig_failed=0
              dd-1851  [001]   158.837860: mm_vmscan_memcg_reclaim_end: nr_reclaimed=32
  (...)
              dd-1970  [000]   266.608235: mm_vmscan_wakeup_kswapd: nid=0 zid=1 order=0
              dd-1970  [000]   266.608239: mm_vmscan_wakeup_kswapd: nid=1 zid=1 order=0
              dd-1970  [000]   266.608248: mm_vmscan_wakeup_kswapd: nid=2 zid=1 order=0
         kswapd1-348   [001]   266.608254: mm_vmscan_kswapd_wake: nid=1 order=0
              dd-1970  [000]   266.608254: mm_vmscan_wakeup_kswapd: nid=3 zid=1 order=0
         kswapd3-350   [000]   266.608266: mm_vmscan_kswapd_wake: nid=3 order=0
  (...)
         kswapd0-347   [001]   267.328891: mm_vmscan_memcg_softlimit_reclaim_begin: order=0 may_writepage=1 gfp_flags=GFP_HIGHUSER_MOVABLE
         kswapd0-347   [001]   267.328897: mm_vmscan_memcg_isolate: isolate_mode=0 order=0 nr_requested=32 nr_scanned=32 nr_taken=32 contig_taken=0 contig_dirty=0 contig_failed=0
         kswapd0-347   [001]   267.328915: mm_vmscan_memcg_isolate: isolate_mode=0 order=0 nr_requested=32 nr_scanned=32 nr_taken=32 contig_taken=0 contig_dirty=0 contig_failed=0
         kswapd0-347   [001]   267.328989: mm_vmscan_memcg_softlimit_reclaim_end: nr_reclaimed=32
         kswapd0-347   [001]   267.329019: mm_vmscan_lru_isolate: isolate_mode=1 order=0 nr_requested=32 nr_scanned=32 nr_taken=32 contig_taken=0 contig_dirty=0 contig_failed=0
         kswapd0-347   [001]   267.330562: mm_vmscan_lru_isolate: isolate_mode=1 order=0 nr_requested=32 nr_scanned=32 nr_taken=32 contig_taken=0 contig_dirty=0 contig_failed=0
  (...)
         kswapd2-349   [001]   267.407081: mm_vmscan_kswapd_sleep: nid=2
         kswapd3-350   [001]   267.408077: mm_vmscan_kswapd_sleep: nid=3
         kswapd1-348   [000]   267.427858: mm_vmscan_kswapd_sleep: nid=1
         kswapd0-347   [001]   267.430064: mm_vmscan_kswapd_sleep: nid=0
#1 lore raw

PATCH 1/4 vmscan: convert direct reclaim tracepoint to DEFINE_TRACE

Mel Gorman recently added some vmscan tracepoints. Unfortunately they are covered only global reclaim. But we want to trace memcg reclaim too.

Thus, this patch convert them to DEFINE_TRACE macro. it help to reuse tracepoint definition for other similar usage (i.e. memcg). This patch have no functionally change.

---
 include/trace/events/vmscan.h |   19 +++++++++++++++++--
 1 files changed, 17 insertions(+), 2 deletions(-)
include/trace/events/vmscan.h +17 -2
index f2da66a..a601763 100644
@@ -68,7 +68,7 @@ TRACE_EVENT(mm_vmscan_wakeup_kswapd,
 		__entry->order)
 );
 
-TRACE_EVENT(mm_vmscan_direct_reclaim_begin,
+DECLARE_EVENT_CLASS(mm_vmscan_direct_reclaim_begin_template,
 
 	TP_PROTO(int order, int may_writepage, gfp_t gfp_flags),
 
@@ -92,7 +92,15 @@ TRACE_EVENT(mm_vmscan_direct_reclaim_begin,
 		show_gfp_flags(__entry->gfp_flags))
 );
 
-TRACE_EVENT(mm_vmscan_direct_reclaim_end,
+DEFINE_EVENT(mm_vmscan_direct_reclaim_begin_template, mm_vmscan_direct_reclaim_begin,
+
+	TP_PROTO(int order, int may_writepage, gfp_t gfp_flags),
+
+	TP_ARGS(order, may_writepage, gfp_flags)
+);
+
+
+DECLARE_EVENT_CLASS(mm_vmscan_direct_reclaim_end_template,
 
 	TP_PROTO(unsigned long nr_reclaimed),
 
@@ -109,6 +117,13 @@ TRACE_EVENT(mm_vmscan_direct_reclaim_end,
 	TP_printk("nr_reclaimed=%lu", __entry->nr_reclaimed)
 );
 
+DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_direct_reclaim_end,
+
+	TP_PROTO(unsigned long nr_reclaimed),
+
+	TP_ARGS(nr_reclaimed)
+);
+
 TRACE_EVENT(mm_vmscan_lru_isolate,
 
 	TP_PROTO(int order,
Signature · 9 lines
1.6.5.2



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 1/4
#2 lore raw

Re: [PATCH 1/4] vmscan: convert direct reclaim tracepoint to DEFINE_TRACE

Quoted 10 lines · KOSAKI Motohiro

Mel Gorman recently added some vmscan tracepoints. Unfortunately they are covered only global reclaim. But we want to trace memcg reclaim too.

Thus, this patch convert them to DEFINE_TRACE macro. it help to reuse tracepoint definition for other similar usage (i.e. memcg). This patch have no functionally change.

Balbir

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

#3 lore raw

PATCH 2/4 memcg, vmscan: add memcg reclaim tracepoint

In reply to KOSAKI Motohiro

Memcg also need to trace reclaim progress as direct reclaim. This patch add it.

---
 include/trace/events/vmscan.h |   28 ++++++++++++++++++++++++++++
 mm/vmscan.c                   |   20 +++++++++++++++++++-
 2 files changed, 47 insertions(+), 1 deletions(-)
include/trace/events/vmscan.h +28 -0
index a601763..c35e905 100644
@@ -99,6 +99,19 @@ DEFINE_EVENT(mm_vmscan_direct_reclaim_begin_template, mm_vmscan_direct_reclaim_b
 	TP_ARGS(order, may_writepage, gfp_flags)
 );
 
+DEFINE_EVENT(mm_vmscan_direct_reclaim_begin_template, mm_vmscan_memcg_reclaim_begin,
+
+	TP_PROTO(int order, int may_writepage, gfp_t gfp_flags),
+
+	TP_ARGS(order, may_writepage, gfp_flags)
+);
+
+DEFINE_EVENT(mm_vmscan_direct_reclaim_begin_template, mm_vmscan_memcg_softlimit_reclaim_begin,
+
+	TP_PROTO(int order, int may_writepage, gfp_t gfp_flags),
+
+	TP_ARGS(order, may_writepage, gfp_flags)
+);
 
 DECLARE_EVENT_CLASS(mm_vmscan_direct_reclaim_end_template,
 
@@ -124,6 +137,21 @@ DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_direct_reclaim_end
 	TP_ARGS(nr_reclaimed)
 );
 
+DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_memcg_reclaim_end,
+
+	TP_PROTO(unsigned long nr_reclaimed),
+
+	TP_ARGS(nr_reclaimed)
+);
+
+DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_memcg_softlimit_reclaim_end,
+
+	TP_PROTO(unsigned long nr_reclaimed),
+
+	TP_ARGS(nr_reclaimed)
+);
+
+
 TRACE_EVENT(mm_vmscan_lru_isolate,
 
 	TP_PROTO(int order,
mm/vmscan.c +19 -1
index 97170eb..c691967 100644
@@ -1949,6 +1949,11 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
 	sc.nodemask = &nm;
 	sc.nr_reclaimed = 0;
 	sc.nr_scanned = 0;
+
+	trace_mm_vmscan_memcg_softlimit_reclaim_begin(0,
+						      sc.may_writepage,
+						      sc.gfp_mask);
+
 	/*
 	 * NOTE: Although we can get the priority field, using it
 	 * here is not a good idea, since it limits the pages we can scan.
@@ -1957,6 +1962,9 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
 	 * the priority and make it zero.
 	 */
 	shrink_zone(0, zone, &sc);
+
+	trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed);
+
 	return sc.nr_reclaimed;
 }
 
@@ -1966,6 +1974,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
 					   unsigned int swappiness)
 {
 	struct zonelist *zonelist;
+	unsigned long nr_reclaimed;
 	struct scan_control sc = {
 		.may_writepage = !laptop_mode,
 		.may_unmap = 1,
@@ -1980,7 +1989,16 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
 	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
 			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
 	zonelist = NODE_DATA(numa_node_id())->node_zonelists;
-	return do_try_to_free_pages(zonelist, &sc);
+
+	trace_mm_vmscan_memcg_reclaim_begin(0,
+					    sc.may_writepage,
+					    sc.gfp_mask);
+
+	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
+
+	trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed);
+
+	return nr_reclaimed;
 }
 #endif
 
Signature · 9 lines
1.6.5.2



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#4 lore raw

PATCH 3/4 vmscan: convert mm_vmscan_lru_isolate to DEFINE_EVENT

In reply to KOSAKI Motohiro

Mel Gorman recently added some vmscan tracepoints. Unfortunately they are covered only global reclaim. But we want to trace memcg reclaim too.

Thus, this patch convert them to DEFINE_TRACE macro. it help to reuse tracepoint definition for other similar usage (i.e. memcg). This patch have no functionally change.

---
 include/trace/events/vmscan.h |   17 ++++++++++++++++-
 1 files changed, 16 insertions(+), 1 deletions(-)
include/trace/events/vmscan.h +16 -1
index c35e905..b97a3db 100644
@@ -152,7 +152,7 @@ DEFINE_EVENT(mm_vmscan_direct_reclaim_end_template, mm_vmscan_memcg_softlimit_re
 );
 
 
-TRACE_EVENT(mm_vmscan_lru_isolate,
+DECLARE_EVENT_CLASS(mm_vmscan_lru_isolate_template,
 
 	TP_PROTO(int order,
 		unsigned long nr_requested,
@@ -198,6 +198,21 @@ TRACE_EVENT(mm_vmscan_lru_isolate,
 		__entry->nr_lumpy_failed)
 );
 
+DEFINE_EVENT(mm_vmscan_lru_isolate_template, mm_vmscan_lru_isolate,
+
+	TP_PROTO(int order,
+		unsigned long nr_requested,
+		unsigned long nr_scanned,
+		unsigned long nr_taken,
+		unsigned long nr_lumpy_taken,
+		unsigned long nr_lumpy_dirty,
+		unsigned long nr_lumpy_failed,
+		int isolate_mode),
+
+	TP_ARGS(order, nr_requested, nr_scanned, nr_taken, nr_lumpy_taken, nr_lumpy_dirty, nr_lumpy_failed, isolate_mode)
+
+);
+
 TRACE_EVENT(mm_vmscan_writepage,
 
 	TP_PROTO(struct page *page,
Signature · 9 lines
1.6.5.2



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 3/4
#5 lore raw

Re: [PATCH 3/4] vmscan: convert mm_vmscan_lru_isolate to DEFINE_EVENT

Quoted 10 lines · KOSAKI Motohiro

Mel Gorman recently added some vmscan tracepoints. Unfortunately they are covered only global reclaim. But we want to trace memcg reclaim too.

Thus, this patch convert them to DEFINE_TRACE macro. it help to reuse tracepoint definition for other similar usage (i.e. memcg). This patch have no functionally change.

Balbir

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

Acked-by on v1 3/4
#6 lore raw

Re: [PATCH 3/4] vmscan: convert mm_vmscan_lru_isolate to DEFINE_EVENT

In reply to KOSAKI Motohiro
Quoted 10 lines · KOSAKI Motohiro

Mel Gorman recently added some vmscan tracepoints. Unfortunately they are covered only global reclaim. But we want to trace memcg reclaim too.

Thus, this patch convert them to DEFINE_TRACE macro. it help to reuse tracepoint definition for other similar usage (i.e. memcg). This patch have no functionally change.

Signature · 3 lines
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
#7 lore raw

PATCH 4/4 memcg: add mm_vmscan_memcg_isolate tracepoint

In reply to KOSAKI Motohiro

Memcg also need to trace page isolation information as global reclaim. This patch does it.

---
 include/trace/events/vmscan.h |   15 +++++++++++++++
 mm/memcontrol.c               |    6 ++++++
 2 files changed, 21 insertions(+), 0 deletions(-)
include/trace/events/vmscan.h +15 -0
index b97a3db..776f92b 100644
@@ -213,6 +213,21 @@ DEFINE_EVENT(mm_vmscan_lru_isolate_template, mm_vmscan_lru_isolate,
 
 );
 
+DEFINE_EVENT(mm_vmscan_lru_isolate_template, mm_vmscan_memcg_isolate,
+
+	TP_PROTO(int order,
+		unsigned long nr_requested,
+		unsigned long nr_scanned,
+		unsigned long nr_taken,
+		unsigned long nr_lumpy_taken,
+		unsigned long nr_lumpy_dirty,
+		unsigned long nr_lumpy_failed,
+		int isolate_mode),
+
+	TP_ARGS(order, nr_requested, nr_scanned, nr_taken, nr_lumpy_taken, nr_lumpy_dirty, nr_lumpy_failed, isolate_mode)
+
+);
+
 TRACE_EVENT(mm_vmscan_writepage,
 
 	TP_PROTO(struct page *page,
mm/memcontrol.c +6 -0
index 2b648ce..2600776 100644
@@ -52,6 +52,8 @@
 
 #include <asm/uaccess.h>
 
+#include <trace/events/vmscan.h>
+
 struct cgroup_subsys mem_cgroup_subsys __read_mostly;
 #define MEM_CGROUP_RECLAIM_RETRIES	5
 struct mem_cgroup *root_mem_cgroup __read_mostly;
@@ -1011,6 +1013,10 @@ unsigned long mem_cgroup_isolate_pages(unsigned long nr_to_scan,
 	}
 
 	*scanned = scan;
+
+	trace_mm_vmscan_memcg_isolate(0, nr_to_scan, scan, nr_taken,
+				      0, 0, 0, mode);
+
 	return nr_taken;
 }
 
Signature · 9 lines
1.6.5.2



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 4/4
#9 lore raw

Re: [PATCH 4/4] memcg: add mm_vmscan_memcg_isolate tracepoint

In reply to KOSAKI Motohiro
Quoted 5 lines · KOSAKI Motohiro

Memcg also need to trace page isolation information as global reclaim. This patch does it.

Signature · 3 lines
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab