Skip to content

Commit

Permalink
oom: use pte pages in OOM score
Browse files Browse the repository at this point in the history
PTE pages eat up memory just like anything else, but we do not account for
them in any way in the OOM scores.  They are also _guaranteed_ to get
freed up when a process is OOM killed, while RSS is not.

Reported-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: <stable@kernel.org>		[2.6.36+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
  • Loading branch information
KOSAKI Motohiro authored and Linus Torvalds committed Apr 28, 2011
1 parent 0dcecae commit f755a04
Showing 1 changed file with 6 additions and 3 deletions.
9 changes: 6 additions & 3 deletions mm/oom_kill.c
Original file line number Diff line number Diff line change
Expand Up @@ -172,10 +172,13 @@ unsigned int oom_badness(struct task_struct *p, struct mem_cgroup *mem,

/*
* The baseline for the badness score is the proportion of RAM that each
* task's rss and swap space use.
* task's rss, pagetable and swap space use.
*/
points = (get_mm_rss(p->mm) + get_mm_counter(p->mm, MM_SWAPENTS)) * 1000 /
totalpages;
points = get_mm_rss(p->mm) + p->mm->nr_ptes;
points += get_mm_counter(p->mm, MM_SWAPENTS);

points *= 1000;
points /= totalpages;
task_unlock(p);

/*
Expand Down

0 comments on commit f755a04

Please sign in to comment.