From mboxrd@z Thu Jan 1 00:00:00 1970 From: KOVACS Krisztian Subject: Re: list delete bug: kernel crash Date: Tue, 31 Aug 2004 12:50:37 +0200 Sender: netfilter-devel-bounces@lists.netfilter.org Message-ID: <1093949437.2435.4.camel@nienna.balabit> References: <003701c48ee9$e1ad1f40$0a00a8c0@aziz> Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <003701c48ee9$e1ad1f40$0a00a8c0@aziz> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: netfilter-devel-bounces@lists.netfilter.org Content-Type: text/plain; charset="iso-8859-1" To: security Cc: netfilter-devel , netfilter@lists.netfilter.org Hi, 2004-08-31, k keltez=E9ssel 01:34-kor security ezt =EDrta: > Aug 30 20:42:24 gateway kernel: NF_IP_ASSERT: > net/ipv4/netfilter/ip_conntrack_core.c:1115(ip_conntrack_alter_reply) > Aug 30 20:42:24 gateway kernel: NF_IP_ASSERT: > net/ipv4/netfilter/ip_conntrack_core.c:1115(ip_conntrack_alter_reply) > Aug 30 20:44:24 gateway kernel: LIST_DELETE: > net/ipv4/netfilter/ip_conntrack_core.c:300 > `&ct->tuplehash[IP_CT_DIR_REPLY]'(f3cedca4) not in > &ip_conntrack_hash[hr]. >=20 > I have search in bugzilla and google and, apparently, this bug has been= solve > since kernel 2.6.5. > But i have kernel 2.6.8.1 and i have try kernel 2.6.6 and 2.6.7: same c= rash >=20 > But it hard to "see" because i can stay 1 week without crash, or 5mn. >=20 > I have the "local NAT" activated. As always, the first thing you should try is testing the memory of your computer. The problems seems to be caused by trying to call ip_nat_setup_info() on an already confirmed connection, which is known to cause hash corruption. BTW, while running memtest, could you send us the nat table of your iptables ruleset, along with the routing setup? And also a list of loaded (iptables-related) kernel modules would be useful. --=20 Regards, Krisztian KOVACS