On 05/20, Oleg Nesterov wrote: > > On 05/19, Michal Hocko wrote: > > > > Long term I > > would like to to move this logic into the mm_struct, it would be just > > larger surgery I guess. > > Why we can't do this right now? Just another MMF_ flag set only once and > never cleared. Just in case... yes, "never cleared" is not that simple because oom_kill_process() can find a OOM_SCORE_ADJ_MIN process (or is_global_init) later. But to me this just looks like another proof that select_bad_process() must not pick such a task (mm!) as victim. Nevermind, I said many times that I do not understand OOM-killer, please ignore me. But sorry, can't resist, I do not think that "signal_struct has some holes[1] so we can stitch it there." can excuse the new member ;) Oleg. -- 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>