Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265276AbUAESQ0 (ORCPT ); Mon, 5 Jan 2004 13:16:26 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265285AbUAESQ0 (ORCPT ); Mon, 5 Jan 2004 13:16:26 -0500 Received: from scrat.hensema.net ([62.212.82.150]:53148 "EHLO scrat.hensema.net") by vger.kernel.org with ESMTP id S265276AbUAESQY (ORCPT ); Mon, 5 Jan 2004 13:16:24 -0500 Date: Mon, 5 Jan 2004 19:16:10 +0100 From: Erik Hensema To: Ricky Beam Cc: Roland Dreier , linux-kernel@vger.kernel.org Subject: Re: 2.6.0: something is leaking memory Message-ID: <20040105181610.GB13674@bender.home.hensema.net> Reply-To: erik@hensema.net References: <52wu87wfzp.fsf@topspin.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1107 Lines: 24 On Mon, Jan 05, 2004 at 11:18:52AM -0500, Ricky Beam wrote: > On 4 Jan 2004, Roland Dreier wrote: > >Yup, looks like IPv6 is leaking memory (since your netstat shows > >nowhere near 19K sockets): > > > > > tcp6_sock 19729 19732 1024 4 1 : tunables 54 27 0 : slabdata 4933 4933 0 > > > >Now to figure out why... > > Of course, there are a lot of sockets in CLOSE_WAIT. That's where I'd > start looking... if I actually cared about IPv6 :-) There used to be the > same sort of problem for IPv4 (sockets would never go away), but no one > ever fixed it -- a rewrite of the network stack made it go away. Those sockets in CLOSE_WAIT state are due to a filedescriptor leak in nscd (or nss_ldap), which is a userspace problem. It does however make the leak in kernelspace more visible it seems. -- Erik Hensema (erik@hensema.net) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/