From: Eric Dumazet Subject: Re: Regression caused by commit "netfilter: iptables: lock free counters" Date: Sun, 05 Apr 2009 12:30:17 +0200 Message-ID: <49D88839.4090902@cosmosbay.com> References: <20090329234702.4988017f@desktop.r000n.info> <8763hja8cy.fsf@newton.gmurray.org.uk> <20090405.012237.198610462.davem@davemloft.net> <49D88162.5040809@cosmosbay.com> <87ws9z8l4c.fsf@newton.gmurray.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: linux-kernel@vger.kernel.org, Linux Netdev List To: Graham Murray Return-path: Received: from gw1.cosmosbay.com ([212.99.114.194]:39006 "EHLO gw1.cosmosbay.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757681AbZDEKa3 convert rfc822-to-8bit (ORCPT ); Sun, 5 Apr 2009 06:30:29 -0400 In-Reply-To: <87ws9z8l4c.fsf@newton.gmurray.org.uk> Sender: netdev-owner@vger.kernel.org List-ID: Graham Murray a =E9crit : > Eric Dumazet writes: >=20 >>>>> 784544739a25c30637397ace5489eeb6e15d7d49 is first bad commit >>>>> commit 784544739a25c30637397ace5489eeb6e15d7d49 =20 >>>> I am seeing a different problem which also bisects to this commit.= There are >>>> no kernel messages but ip6tables fails to run. >>>> >>>> newton ~ # ip6tables -L -v >>>> FATAL: Module ip6_tables not found. >>>> ip6tables v1.4.3.1: can't initialize ip6tables table `filter': Mem= ory allocation problem >>>> Perhaps ip6tables or your kernel needs to be upgraded. >>>> >>>> I get this error no matter which ip6tables sub-command I run. Ip6t= ables >>>> is built into the kernel, not as modules. >>>> >>>> An strace shows the failure to be=20 >>>> socket(PF_INET6, SOCK_RAW, IPPROTO_RAW) =3D 3 >>>> getsockopt(3, SOL_IPV6, 0x40 /* IPV6_??? */, "filter\0\305\0w~\300= \0wb\305P\24\312\t\0009b\305\216\23\0\0\310\341/g\16"..., [84]) =3D 0 >>>> brk(0) =3D 0x8273000 >>>> brk(0x8294000) =3D 0x8294000 >> so ip6tables allocates about 128 Kbytes of ram in order to get rules= from kernel. >> >>>> getsockopt(3, SOL_IPV6, 0x41 /* IPV6_??? */, 0x8273090, 0xbfd23628= ) =3D -1 ENOMEM (Cannot allocate memory) >>>> close(3) =3D 0 >>>> >> >> This is a big problem yes, since "iptables|ip6tables" -L needs to al= locate kernel memory >> to perform the momentary swap. >> >> On x86, this is potentially a problem if vmalloc space is exhausted = or fragmented, >> (or lowmem exhausted) and/or many cpus are online/possible. >=20 > iptables gives me no problems at all, it is just ip6tables that > fails. The first indication of this is during the init scripts when > ip6tables-restore fails. I see, its a plain bug in net/ipv6/netfilter/ip6_tables.c function alloc_counters() always returns -ENOMEM Unfortunatly , its Sunday here and I have to run for lunch time with fa= mily :) If nobody beats me, I will do the fix in a couple of hours... Thank you