On 2018/08/06 23:58, Michal Hocko wrote: >> Since I can't find mm related changes between next-20180803 (syzbot can >> reproduce) and >> next-20180806 (syzbot has not reproduced), I can't guess what makes this >> problem go away. > > Hmm, but original report was against 4.18.0-rc6-next-20180725+ kernel. > And that one had the old group oom code. /me confused. >
Yes. But I confirmed that syzbot can reproduce this problem with next-20180803 which already dropped the old group oom code. Therefore, I think that syzbot is hitting a problem other than the old group oom code.