Messages in this thread | | | Date | Fri, 5 Dec 2008 22:27:10 +0900 (JST) | Subject | Re: [RFC][PATCH -mmotm 3/4] memcg: avoid dead lock caused by racebetween oom and cpuset_attach | From | "KAMEZAWA Hiroyuki" <> |
| |
Daisuke Nishimura said: > 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). > good catch.
> If those two paths race, dead lock can happen. > > This patch avoid this dead lock by: > - remove cgroup_lock() from mem_cgroup_out_of_memory(). agree to this.
> - define new mutex (memcg_tasklist) and serialize mem_cgroup_move_task() > (->attach handler of memory cgroup) and mem_cgroup_out_of_memory. > Hmm...seems temporal fix (and adding new global lock...) But ok, we need fix. revist this later.
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu,com>
| |