On 03/10, Oleg Nesterov wrote: > > On 03/09, David Rientjes wrote: > > > > On Wed, 9 Mar 2011, Oleg Nesterov wrote: > > > > > > Using for_each_process() does not consider threads that have failed to > > > > exit after the oom killed parent and, thus, we select another innocent > > > > task to kill when we're really just waiting for those threads to exit > > > > > > How so? select_bad_process() checks TIF_MEMDIE and returns ERR_PTR() > > > if it is set. > > > > > > > TIF_MEMDIE is quite obviously a per-thread flag > Yes, and this is why I think it should be replaced. But this is not simple. I'd suggest this patch as the first step. It is not tested. I do not pretend I really understand oom-killer in all details (although oom_kill.c itself is quite trivial). The patch assumes that oom-prevent-unnecessary-oom-kills-or-kernel-panics.patch oom-skip-zombies-when-iterating-tasklist.patch are dropped. I think, this patch __might__ address the problems described in the changelog, but of course I am not sure. I am relying on your and Kosaki's review, if you disagree with this change I won't argue. But the current usage of TIF_MEMDIE can't be right. Oleg. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>