Another patchset from CK to lighten vm load

[kernel-bfs] / kernel-bfs-2.6.28 / debian / patches / mm-lots_watermark.diff

diff --git a/kernel-bfs-2.6.28/debian/patches/mm-lots_watermark.diff b/kernel-bfs-2.6.28/debian/patches/mm-lots_watermark.diff
new file mode 100644 (file)
index 0000000..c756bd6
--- /dev/null
+++ b/kernel-bfs-2.6.28/debian/patches/mm-lots_watermark.diff
@@ -0,0 +1,108 @@
+The vm currently performs scanning when allocating ram once the watermarks
+are below the pages_low value and tries to restore them to the pages_high
+watermark. The disadvantage of this is that we are scanning most aggresssively
+at the same time we are allocating ram regardless of the stress the vm is
+under. Add a pages_lots watermark and allow the watermark to be relaxed
+according to the stress the vm is at the time (according to the priority
+value). Thus we have more in reserve next time we are allocating ram and end
+up scanning less aggresssively. Note the actual pages_lots isn't used directly
+in this code.
+
+Signed-off-by: Con Kolivas <kernel@kolivas.org>
+
+ include/linux/mmzone.h |    2 +-
+ mm/page_alloc.c        |    3 +++
+ mm/vmscan.c            |   17 ++++++++++++++---
+ 3 files changed, 18 insertions(+), 4 deletions(-)
+
+Index: linux-2.6.22-ck1/include/linux/mmzone.h
+===================================================================
+--- linux-2.6.22-ck1.orig/include/linux/mmzone.h       2007-07-09 18:44:34.000000000 +1000
++++ linux-2.6.22-ck1/include/linux/mmzone.h    2007-07-09 18:44:39.000000000 +1000
+@@ -181,7 +181,7 @@ enum zone_type {
+ 
+ struct zone {
+       /* Fields commonly accessed by the page allocator */
+-      unsigned long           pages_min, pages_low, pages_high;
++      unsigned long           pages_min, pages_low, pages_high, pages_lots;
+       /*
+        * We don't know if the memory that we're going to allocate will be freeable
+        * or/and it will be released eventually, so to avoid totally wasting several
+Index: linux-2.6.22-ck1/mm/page_alloc.c
+===================================================================
+--- linux-2.6.22-ck1.orig/mm/page_alloc.c      2007-07-09 18:44:34.000000000 +1000
++++ linux-2.6.22-ck1/mm/page_alloc.c   2007-07-09 18:44:39.000000000 +1000
+@@ -1570,6 +1570,7 @@ void show_free_areas(void)
+                       " min:%lukB"
+                       " low:%lukB"
+                       " high:%lukB"
++                      " lots:%lukB"
+                       " active_anon:%lukB"
+                       " inactive_anon:%lukB"
+                       " active_file:%lukB"
+@@ -1581,6 +1582,7 @@ void show_free_areas(void)
+                       K(zone->pages_min),
+                       K(zone->pages_low),
+                       K(zone->pages_high),
++                      K(zone->pages_lots),
+                       K(zone_page_state(zone, NR_ACTIVE_ANON)),
+                       K(zone_page_state(zone, NR_INACTIVE_ANON)),
+                       K(zone_page_state(zone, NR_ACTIVE_FILE)),
+@@ -3142,6 +3144,7 @@ void setup_per_zone_pages_min(void)
+ 
+               zone->pages_low   = zone->pages_min + (tmp >> 2);
+               zone->pages_high  = zone->pages_min + (tmp >> 1);
++              zone->pages_lots  = zone->pages_min + tmp;
+               setup_zone_migrate_reserve(zone);
+               spin_unlock_irqrestore(&zone->lock, flags);
+       }
+Index: linux-2.6.22-ck1/mm/vmscan.c
+===================================================================
+--- linux-2.6.22-ck1.orig/mm/vmscan.c  2007-07-09 18:44:39.000000000 +1000
++++ linux-2.6.22-ck1/mm/vmscan.c       2007-07-09 18:44:39.000000000 +1000
+@@ -1171,6 +1171,7 @@ loop_again:
+                */
+               for (i = pgdat->nr_zones - 1; i >= 0; i--) {
+                       struct zone *zone = pgdat->node_zones + i;
++                      unsigned long watermark;
+ 
+                       if (!populated_zone(zone))
+                               continue;
+@@ -1178,8 +1179,14 @@ loop_again:
+                               shrink_active_list(SWAP_CLUSTER_MAX, zone,
+                                                       &sc, priority, 0);
+ 
+-                      if (!zone_watermark_ok(zone, order, zone->pages_high,
+-                                             0, 0)) {
++                      /*
++                       * The watermark is relaxed depending on the
++                       * level of "priority" till it drops to
++                       * pages_high.
++                       */
++                      watermark = zone->pages_high + (zone->pages_high *
++                                  priority / DEF_PRIORITY);
++                      if (!zone_watermark_ok(zone, order, watermark, 0, 0)) {
+                               end_zone = i;
+                               break;
+                       }
+@@ -1206,6 +1213,7 @@ loop_again:
+               for (i = 0; i <= end_zone; i++) {
+                       struct zone *zone = pgdat->node_zones + i;
+                       int nr_slab;
++                      unsigned long watermark;
+ 
+                       if (!populated_zone(zone))
+                               continue;
+@@ -1213,7 +1221,10 @@ loop_again:
+                                       priority != DEF_PRIORITY)
+                               continue;
+ 
+-                      if (!zone_watermark_ok(zone, order, zone->pages_high,
++                      watermark = zone->pages_high + (zone->pages_high *
++                                  priority / DEF_PRIORITY);
++
++                      if (!zone_watermark_ok(zone, order, watermark,
+                                              end_zone, 0))
+                               all_zones_ok = 0;
+                       temp_priority[i] = priority;
+