On Wed 09-09-15 14:55:59, David Rientjes wrote: > On Sun, 6 Sep 2015, Tetsuo Handa wrote: > > > It was observed that setting TIF_MEMDIE before sending SIGKILL at > > oom_kill_process() allows memory reserves to be depleted by allocations > > which are not needed for terminating the OOM victim. > > > > I don't understand what you are trying to fix. Sending a SIGKILL first > does not guarantee that it is handling that signal before accessing memory > reserves. Yes it doesn't guarantee that but it kicks the task from userspace so that it cannot deplete the memory reserves from the _userspace_ under its controll. It still can consume some amount of memory from the kernel but that shouldn't be under control of the task. -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>