From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from ucspost.ucs.co.za (IDENT:root@mailgw1.ucs.co.za [196.23.43.254]) by ucs.co.za (8.11.2/8.11.2) with ESMTP id f6SCMgQ18195 for ; Sat, 28 Jul 2001 14:22:42 +0200 Received: from jhb.ucs.co.za (IDENT:root@jhb.ucs.co.za [172.31.1.73]) by ucspost.ucs.co.za (8.9.3/8.7.3) with ESMTP id OAA21992 for ; Sat, 28 Jul 2001 14:22:42 +0200 Received: from bds.ucs.co.za (bds@bds.ucs.co.za [172.31.1.36]) by jhb.ucs.co.za (8.9.3/8.8.7) with ESMTP id OAA23543 for ; Sat, 28 Jul 2001 14:25:11 +0200 Subject: oom-kill looping From: Berend De Schouwer Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: 28 Jul 2001 14:22:41 +0200 Message-Id: <996322962.1350.43.camel@bds.ucs.co.za> Mime-Version: 1.0 Sender: owner-linux-mm@kvack.org Return-Path: To: linux-mm List-ID: Hi, I've got a machine that has been crashing more-or-less weekly due to running out of memory. I've noticed an oddity in oom-kill. It loops 'Out of Memory: Killed process 12345 (abc).' The same PID is listed continuously. This is with kernel 2.4.x. With the 2.2 series it just hung. Is it a good idea to add checks oom_kill.c in badness() like: if is_not_runnable(p) return 0; if has_pending_term_or_kill(p) return 0; and if so, would (p->state != 0) and signal_pending(p) be the way to do it? How would you check for just the SIGKILL or SIGTERM signal? It takes about a week to trigger this OOM, so trials are slow. -- Berend De Schouwer -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/