From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: [PATCH] netfilter: nf_nat: fix crash in nf_nat_csum Date: Wed, 08 Jun 2011 11:36:18 +0200 Message-ID: <4DEF4292.2070904@trash.net> References: <4DEE22A4.8060202@trash.net> <4DEEADCB.8000803@trash.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: netfilter-devel@vger.kernel.org To: Julian Anastasov Return-path: Received: from stinky.trash.net ([213.144.137.162]:41273 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751289Ab1FHJg2 (ORCPT ); Wed, 8 Jun 2011 05:36:28 -0400 In-Reply-To: <4DEEADCB.8000803@trash.net> Sender: netfilter-devel-owner@vger.kernel.org List-ID: On 08.06.2011 01:01, Patrick McHardy wrote: > On 07.06.2011 22:02, Julian Anastasov wrote: >> >> Hello, >> >> On Tue, 7 Jun 2011, Patrick McHardy wrote: >> >>> On 29.05.2011 22:42, Julian Anastasov wrote: >>>> >>>> Fix crash in nf_nat_csum when mangling packets >>>> in OUTPUT hook where skb->dev is not defined, it is set >>>> later before POSTROUTING. Problem happens for CHECKSUM_NONE. >>>> We can check device from rt but using CHECKSUM_PARTIAL >>>> should be safe (skb_checksum_help). >>> >>> We don't call helpers in LOCAL_OUT, how can this happen? >> >> I think, this happened with IPVS only in OUTPUT hook: >> tcp_snat_handler -> ip_vs_ftp_out -> nf_nat_mangle_tcp_packet -> >> nf_nat_csum. This is again PASV FTP via loopback. > > I see, thanks. I'll add a comment and will apply this tommorrow. I must have overlooked the message, just noticed that this patch is already applied,