memcg: fix calculation of active_ratio
authorKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Thu, 8 Jan 2009 02:08:25 +0000 (18:08 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 8 Jan 2009 16:31:09 +0000 (08:31 -0800)
Currently, inactive_ratio of memcg is calculated at setting limit.
because page_alloc.c does so and current implementation is straightforward
porting.

However, memcg introduced hierarchy feature recently.  In hierarchy
restriction, memory limit is not only decided memory.limit_in_bytes of
current cgroup, but also parent limit and sibling memory usage.

Then, The optimal inactive_ratio is changed frequently.  So, everytime
calculation is better.

Tested-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/memcontrol.h
mm/memcontrol.c
mm/vmscan.c

index 8752052da8dfad8308f376114319f32c06d2553d..056cf82c0e863a2bfa92b5af0ec96580c0d91a2e 100644 (file)
@@ -97,8 +97,7 @@ extern void mem_cgroup_note_reclaim_priority(struct mem_cgroup *mem,
                                                        int priority);
 extern void mem_cgroup_record_reclaim_priority(struct mem_cgroup *mem,
                                                        int priority);
-int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg,
-                                   struct zone *zone);
+int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg);
 unsigned long mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg,
                                       struct zone *zone,
                                       enum lru_list lru);
@@ -252,7 +251,7 @@ static inline bool mem_cgroup_oom_called(struct task_struct *task)
 }
 
 static inline int
-mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg, struct zone *zone)
+mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg)
 {
        return 1;
 }
index ab2ecbb95b8daf23168aaa79261885d8a1972ac1..c7d78eca8363bfcb6b4c1ff8b674af5516f583d5 100644 (file)
@@ -166,9 +166,6 @@ struct mem_cgroup {
 
        unsigned int    swappiness;
 
-
-       unsigned int inactive_ratio;
-
        /*
         * statistics. This must be placed at the end of memcg.
         */
@@ -432,15 +429,43 @@ void mem_cgroup_record_reclaim_priority(struct mem_cgroup *mem, int priority)
        spin_unlock(&mem->reclaim_param_lock);
 }
 
-int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg, struct zone *zone)
+static int calc_inactive_ratio(struct mem_cgroup *memcg, unsigned long *present_pages)
 {
        unsigned long active;
        unsigned long inactive;
+       unsigned long gb;
+       unsigned long inactive_ratio;
 
        inactive = mem_cgroup_get_all_zonestat(memcg, LRU_INACTIVE_ANON);
        active = mem_cgroup_get_all_zonestat(memcg, LRU_ACTIVE_ANON);
 
-       if (inactive * memcg->inactive_ratio < active)
+       gb = (inactive + active) >> (30 - PAGE_SHIFT);
+       if (gb)
+               inactive_ratio = int_sqrt(10 * gb);
+       else
+               inactive_ratio = 1;
+
+       if (present_pages) {
+               present_pages[0] = inactive;
+               present_pages[1] = active;
+       }
+
+       return inactive_ratio;
+}
+
+int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg)
+{
+       unsigned long active;
+       unsigned long inactive;
+       unsigned long present_pages[2];
+       unsigned long inactive_ratio;
+
+       inactive_ratio = calc_inactive_ratio(memcg, present_pages);
+
+       inactive = present_pages[0];
+       active = present_pages[1];
+
+       if (inactive * inactive_ratio < active)
                return 1;
 
        return 0;
@@ -1432,29 +1457,6 @@ int mem_cgroup_shrink_usage(struct mm_struct *mm, gfp_t gfp_mask)
        return 0;
 }
 
-/*
- * The inactive anon list should be small enough that the VM never has to
- * do too much work, but large enough that each inactive page has a chance
- * to be referenced again before it is swapped out.
- *
- * this calculation is straightforward porting from
- * page_alloc.c::setup_per_zone_inactive_ratio().
- * it describe more detail.
- */
-static void mem_cgroup_set_inactive_ratio(struct mem_cgroup *memcg)
-{
-       unsigned int gb, ratio;
-
-       gb = res_counter_read_u64(&memcg->res, RES_LIMIT) >> 30;
-       if (gb)
-               ratio = int_sqrt(10 * gb);
-       else
-               ratio = 1;
-
-       memcg->inactive_ratio = ratio;
-
-}
-
 static DEFINE_MUTEX(set_limit_mutex);
 
 static int mem_cgroup_resize_limit(struct mem_cgroup *memcg,
@@ -1496,9 +1498,6 @@ static int mem_cgroup_resize_limit(struct mem_cgroup *memcg,
                if (!progress)                  retry_count--;
        }
 
-       if (!ret)
-               mem_cgroup_set_inactive_ratio(memcg);
-
        return ret;
 }
 
@@ -1858,7 +1857,7 @@ static int mem_control_stat_show(struct cgroup *cont, struct cftype *cft,
        }
 
 #ifdef CONFIG_DEBUG_VM
-       cb->fill(cb, "inactive_ratio", mem_cont->inactive_ratio);
+       cb->fill(cb, "inactive_ratio", calc_inactive_ratio(mem_cont, NULL));
 
        {
                int nid, zid;
@@ -2150,7 +2149,6 @@ mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
                res_counter_init(&mem->res, NULL);
                res_counter_init(&mem->memsw, NULL);
        }
-       mem_cgroup_set_inactive_ratio(mem);
        mem->last_scanned_child = NULL;
        spin_lock_init(&mem->reclaim_param_lock);
 
index ece2f405187f6703871bfe5ebf185971d810f7d7..9a27c44aa327f8f46ddbde07776df8644db82de5 100644 (file)
@@ -1340,7 +1340,7 @@ static int inactive_anon_is_low(struct zone *zone, struct scan_control *sc)
        if (scanning_global_lru(sc))
                low = inactive_anon_is_low_global(zone);
        else
-               low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup, zone);
+               low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
        return low;
 }