>From 142b08258e4c60834602e9b0a734564208bc6397 Mon Sep 17 00:00:00 2001 From: Tetsuo Handa <[email protected]> Date: Wed, 17 Feb 2016 16:29:29 +0900 Subject: [PATCH 1/6] mm,oom: exclude TIF_MEMDIE processes from candidates.
The OOM reaper kernel thread can reclaim OOM victim's memory before the victim releases it. But it is possible that a TIF_MEMDIE thread gets stuck at down_read(&mm->mmap_sem) in exit_mm() called from do_exit() due to one of !TIF_MEMDIE threads doing a GFP_KERNEL allocation between down_write(&mm->mmap_sem) and up_write(&mm->mmap_sem) (e.g. mmap()). In that case, we need to use SysRq-f (manual invocation of the OOM killer) because down_read_trylock(&mm->mmap_sem) by the OOM reaper will not succeed. Also, there are other situations where the OOM reaper cannot reap the victim's memory (e.g. CONFIG_MMU=n, victim's memory is shared with OOM-unkillable processes) which will require manual SysRq-f for making progress. However, it is possible that the OOM killer chooses the same OOM victim forever which already has TIF_MEMDIE. This is effectively disabling SysRq-f. This patch excludes processes which has a TIF_MEMDIE thread from OOM victim candidates. Signed-off-by: Tetsuo Handa <[email protected]> --- mm/oom_kill.c | 30 +++++++++++++++++++++++++++--- 1 file changed, 27 insertions(+), 3 deletions(-) diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 871470f..27949ef 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -119,6 +119,30 @@ found: } /* + * Treat the whole process p as unkillable when one of threads has + * TIF_MEMDIE pending. Otherwise, we may end up setting TIF_MEMDIE + * on the same victim forever (e.g. making SysRq-f unusable). + */ +static struct task_struct *find_lock_non_victim_task_mm(struct task_struct *p) +{ + struct task_struct *t; + + rcu_read_lock(); + + for_each_thread(p, t) { + if (likely(!test_tsk_thread_flag(t, TIF_MEMDIE))) + continue; + t = NULL; + goto found; + } + t = find_lock_task_mm(p); + found: + rcu_read_unlock(); + + return t; +} + +/* * order == -1 means the oom kill is required by sysrq, otherwise only * for display purposes. */ @@ -165,7 +189,7 @@ unsigned long oom_badness(struct task_struct *p, struct mem_cgroup *memcg, if (oom_unkillable_task(p, memcg, nodemask)) return 0; - p = find_lock_task_mm(p); + p = find_lock_non_victim_task_mm(p); if (!p) return 0; @@ -361,7 +385,7 @@ static void dump_tasks(struct mem_cgroup *memcg, const nodemask_t *nodemask) if (oom_unkillable_task(p, memcg, nodemask)) continue; - task = find_lock_task_mm(p); + task = find_lock_non_victim_task_mm(p); if (!task) { /* * This is a kthread or all of p's threads have already @@ -562,7 +586,7 @@ void oom_kill_process(struct oom_control *oc, struct task_struct *p, } read_unlock(&tasklist_lock); - p = find_lock_task_mm(victim); + p = find_lock_non_victim_task_mm(victim); if (!p) { put_task_struct(victim); return; -- 1.8.3.1
