nishimura at mxp
Nov 24, 2009, 9:32 PM
Post #1 of 3
(71 views)
Permalink
|
|
[BUGFIX][PATCH v2 -stable] memcg: avoid oom-killing innocent task in case of use_hierarchy
|
|
> Hmm. Maybe not-expected behavior...could you add comment ?
>
How about this ?
> Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu [at] jp>
> (*) I'm sorry I can't work enough in these days.
>
BTW, this patch conflict with oom-dump-stack-and-vm-state-when-oom-killer-panics.patch
in current mmotm(that's why I post mmotm version separately), so this bug will not be fixed
till 2.6.33 in linus-tree.
So I think this patch should go in 2.6.32.y too.
===
From: Daisuke Nishimura <nishimura [at] mxp>
task_in_mem_cgroup(), which is called by select_bad_process() to check whether
a task can be a candidate for being oom-killed from memcg's limit, checks
"curr->use_hierarchy"("curr" is the mem_cgroup the task belongs to).
But this check return true(it's false positive) when:
<some path>/00 use_hierarchy == 0 <- hitting limit
<some path>/00/aa use_hierarchy == 1 <- "curr"
This leads to killing an innocent task in 00/aa. This patch is a fix for this
bug. And this patch also fixes the arg for mem_cgroup_print_oom_info(). We
should print information of mem_cgroup which the task being killed, not current,
belongs to.
Signed-off-by: Daisuke Nishimura <nishimura [at] mxp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu [at] jp>
Reviewed-by: Balbir Singh <balbir [at] linux>
---
mm/memcontrol.c | 8 +++++++-
mm/oom_kill.c | 2 +-
2 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index fd4529d..566925e 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -496,7 +496,13 @@ int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem)
task_unlock(task);
if (!curr)
return 0;
- if (curr->use_hierarchy)
+ /*
+ * We should check use_hierarchy of "mem" not "curr". Because checking
+ * use_hierarchy of "curr" here make this function true if hierarchy is
+ * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
+ * hierarchy(even if use_hierarchy is disabled in "mem").
+ */
+ if (mem->use_hierarchy)
ret = css_is_ancestor(&curr->css, &mem->css);
else
ret = (curr == mem);
diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index a7b2460..ed452e9 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -400,7 +400,7 @@ static int oom_kill_process(struct task_struct *p, gfp_t gfp_mask, int order,
cpuset_print_task_mems_allowed(current);
task_unlock(current);
dump_stack();
- mem_cgroup_print_oom_info(mem, current);
+ mem_cgroup_print_oom_info(mem, p);
show_mem();
if (sysctl_oom_dump_tasks)
dump_tasks(mem);
--
1.5.6.1
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo [at] vger
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
|
