Michal Hocko wrote: > > I assert that we should fix af5679fbc669f31f. > > If you can come up with reasonable patch which doesn't complicate the > code and it is a clear win for both this particular workload as well as > others then why not. Why can't we do "at least MMF_OOM_SKIP should be set under the lock to prevent from races" ? diff --git a/mm/mmap.c b/mm/mmap.c index 5f2b2b1..e096bb8 100644 --- a/mm/mmap.c +++ b/mm/mmap.c @@ -3065,7 +3065,9 @@ void exit_mmap(struct mm_struct *mm) */ (void)__oom_reap_task_mm(mm); + mutex_lock(&oom_lock); set_bit(MMF_OOM_SKIP, &mm->flags); + mutex_unlock(&oom_lock); down_write(&mm->mmap_sem); up_write(&mm->mmap_sem); } diff --git a/mm/oom_kill.c b/mm/oom_kill.c index f10aa53..b2a94c1 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -606,7 +606,9 @@ static void oom_reap_task(struct task_struct *tsk) * Hide this mm from OOM killer because it has been either reaped or * somebody can't call up_write(mmap_sem). */ + mutex_lock(&oom_lock); set_bit(MMF_OOM_SKIP, &mm->flags); + mutex_unlock(&oom_lock); /* Drop a reference taken by wake_oom_reaper */ put_task_struct(tsk);
