From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nicolas Dichtel Subject: [v3.18] Fixes for ip_vti and ip6_vti Date: Tue, 25 Aug 2015 15:21:19 +0200 Message-ID: <55DC6BCF.1040206@6wind.com> Reply-To: nicolas.dichtel@6wind.com Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Cc: David Miller , Alexander Duyck , Steffen Klassert To: stable , netdev , Sasha Levin Return-path: Received: from mail-wi0-f178.google.com ([209.85.212.178]:36789 "EHLO mail-wi0-f178.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755549AbbHYNVX (ORCPT ); Tue, 25 Aug 2015 09:21:23 -0400 Received: by wicja10 with SMTP id ja10so15039957wic.1 for ; Tue, 25 Aug 2015 06:21:21 -0700 (PDT) Sender: netdev-owner@vger.kernel.org List-ID: Hi, since commit ca7c7b9059e3 ("skbuff: Do not scrub skb mark within the same name space", backport of upstream commit 213dd74aee76), the following three upstream commits need also to be backported in the 3.18 branch: - cd5279c194f8 ("ip_vti/ip6_vti: Do not touch skb->mark on xmit") - 049f8e2e28d9 ("xfrm: Override skb->mark with tunnel->parm.i_key in xfrm_input") - d55c670cbc54 ("ip_vti/ip6_vti: Preserve skb->mark after rcv_cb call") When the packet is handled by the ip_vti interface, it is temporarily marked with the ip_vti interface key. Once the process is completed, the packet is received via the ip_vti interface and its original mark should be restored (typically to 0). Before the backported patch, ip_vti counted on skb_scrub_packet to reset the mark, but skb_scrub_packet nomore does. Since the packet is still marked, it will match the outbound SP and so be encrypted again. Regards, Nicolas