Commit 7f4d454dee2e0bdd21bafd413d1c53e443a26540

Authored by Daisuke Nishimura
Committed by Linus Torvalds
1 parent a5e924f5f8

memcg: avoid deadlock caused by race between oom and cpuset_attach

mpol_rebind_mm(), which can be called from cpuset_attach(), does
down_write(mm->mmap_sem).  This means down_write(mm->mmap_sem) can be
called under cgroup_mutex.

OTOH, page fault path does down_read(mm->mmap_sem) and calls
mem_cgroup_try_charge_xxx(), which may eventually calls
mem_cgroup_out_of_memory().  And mem_cgroup_out_of_memory() calls
cgroup_lock().  This means cgroup_lock() can be called under
down_read(mm->mmap_sem).

If those two paths race, deadlock can happen.

This patch avoid this deadlock by:
  - remove cgroup_lock() from mem_cgroup_out_of_memory().
  - define new mutex (memcg_tasklist) and serialize mem_cgroup_move_task()
    (->attach handler of memory cgroup) and mem_cgroup_out_of_memory.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: 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>

Showing 2 changed files with 5 additions and 2 deletions Side-by-side Diff

... ... @@ -51,6 +51,7 @@
51 51 #define do_swap_account (0)
52 52 #endif
53 53  
  54 +static DEFINE_MUTEX(memcg_tasklist); /* can be hold under cgroup_mutex */
54 55  
55 56 /*
56 57 * Statistics for memory cgroup.
57 58  
... ... @@ -827,7 +828,9 @@
827 828  
828 829 if (!nr_retries--) {
829 830 if (oom) {
  831 + mutex_lock(&memcg_tasklist);
830 832 mem_cgroup_out_of_memory(mem_over_limit, gfp_mask);
  833 + mutex_unlock(&memcg_tasklist);
831 834 mem_over_limit->last_oom_jiffies = jiffies;
832 835 }
833 836 goto nomem;
834 837  
... ... @@ -2211,10 +2214,12 @@
2211 2214 struct cgroup *old_cont,
2212 2215 struct task_struct *p)
2213 2216 {
  2217 + mutex_lock(&memcg_tasklist);
2214 2218 /*
2215 2219 * FIXME: It's better to move charges of this process from old
2216 2220 * memcg to new memcg. But it's just on TODO-List now.
2217 2221 */
  2222 + mutex_unlock(&memcg_tasklist);
2218 2223 }
2219 2224  
2220 2225 struct cgroup_subsys mem_cgroup_subsys = {
... ... @@ -429,7 +429,6 @@
429 429 unsigned long points = 0;
430 430 struct task_struct *p;
431 431  
432   - cgroup_lock();
433 432 read_lock(&tasklist_lock);
434 433 retry:
435 434 p = select_bad_process(&points, mem);
... ... @@ -444,7 +443,6 @@
444 443 goto retry;
445 444 out:
446 445 read_unlock(&tasklist_lock);
447   - cgroup_unlock();
448 446 }
449 447 #endif
450 448