On Thu, Apr 26, 2012 at 02:40:48PM -0700, David Rientjes wrote: > On Thu, 26 Apr 2012, Dave Jones wrote: > > > On a test machine that was running my system call fuzzer, I just saw > > the oom killer take out everything but the process that was doing all > > the memory exhausting. > > > > Would it be possible to try the below patch? It should kill the thread > using the most memory (which happens to only be a couple more megabytes on > your system), but it might just delay the inevitable since the system is > still in a pretty bad state. > > KOSAKI-san suggested doing this before and I think it's the best direction > to go in anyway. Sure, I'll give it a shot when I reboot. However, see my follow-up message. I think there are two bugs here. 1) The over-aggressive oom-killer, and 2) ksmd going mental. /sys/kernel/mm/ksm/full_scans is increasing constantly full_scans: 146370 pages_shared: 1 pages_sharing: 4 pages_to_scan: 1250 pages_unshared: 867 pages_volatile: 1 run: 1 sleep_millisecs: 20 everything in /sys/kernel/mm/hugepages/hugepages-2048kB, is 0. /sys/kernel/mm/transparent_hugepage/khugepaged: alloc_sleep_millisecs 60000 defrag 1 full_scans 15 max_ptes_none 511 pages_collapsed 6 pages_to_scan 4096 scan_sleep_millisecs 10000 Dave -- 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/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>