From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Thu, 2 Aug 2001 18:01:51 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 2 Aug 2001 18:01:41 -0400 Received: from front1.mail.megapathdsl.net ([66.80.60.31]:56335 "EHLO front1.mail.megapathdsl.net") by vger.kernel.org with ESMTP id convert rfc822-to-8bit; Thu, 2 Aug 2001 18:01:35 -0400 Subject: Re: Ongoing 2.4 VM suckage From: Miles Lane To: "Jeffrey W. Baker" Cc: Jakob =?ISO-8859-1?Q?=D8stergaard?= , linux-kernel@vger.kernel.org In-Reply-To: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Mailer: Evolution/0.12.99 (Preview Release) Date: 02 Aug 2001 14:56:29 -0700 Message-Id: <996789389.12934.25.camel@stomata.megapathdsl.net> Mime-Version: 1.0 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On 02 Aug 2001 14:52:11 -0700, Jeffrey W. Baker wrote: > On Thu, 2 Aug 2001, Jakob Østergaard wrote: > > > You fill up mem and you fill up swap, and you complain the box is > > acting funny ?? > > The kernel should save whatever memory it needs to do its work. It isn't > my problem, from userland, to worry that I take the last page in the > machine. If the kernel needs pages to operate efficiently, it had better > reserve them and not just hand them out until it locks up. > > > This is a clear case of "Doctor it hurts when I ..." - Don't do it ! > > > > I'm interested in hearing how you would accomplish graceful > > performance degradation in a situation where you have used up any > > possible resource on the machine. Transparent process back-tracking ? > > What ? > > Gosh, here's an idea: if there is no memory left and someone malloc()s > some more, have malloc() fail? Kill the process that required the memory? > I can't believe the attitude I am hearing. Userland processes should be > able to go around doing whaever the fuck they want and the box should stay > alive. Currently, if a userland process runs amok, the kernel goes into > self-fucking mode for the rest of the week. Hmm. What about the OOM process killer? Shouldn't that kick in? Miles