From mboxrd@z Thu Jan 1 00:00:00 1970 From: Edward Cree Subject: Re: [PATCH v2 net-next 4/5] fou: enable LCO in FOU and GUE Date: Thu, 7 Jan 2016 19:00:07 +0000 Message-ID: <568EB5B7.9090304@solarflare.com> References: <568E9BF9.2050001@solarflare.com> <568E9CAC.6070906@solarflare.com> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Cc: David Miller , Linux Kernel Network Developers , To: Tom Herbert Return-path: Received: from nbfkord-smmo03.seg.att.com ([209.65.160.84]:63246 "EHLO nbfkord-smmo03.seg.att.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752189AbcAGTAe (ORCPT ); Thu, 7 Jan 2016 14:00:34 -0500 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On 07/01/16 18:51, Tom Herbert wrote: > On Thu, Jan 7, 2016 at 9:13 AM, Edward Cree wrote: >> @@ -787,7 +786,7 @@ int fou_build_header(struct sk_buff *skb, struct ip_tunnel_encap *e, >> int type = csum ? SKB_GSO_UDP_TUNNEL_CSUM : SKB_GSO_UDP_TUNNEL; >> __be16 sport; >> >> - skb = iptunnel_handle_offloads(skb, csum, type); >> + skb = iptunnel_handle_offloads(skb, false, type); > It might good to have another iptunnel_handle_offloads (lifke > iptunnel_handle_offloads_nocsum) function that doesn't have csum > parameter for readability. I hope to follow up soon with a patch to use LCO for GRE, at which point I believe that all callers will be passing false anyway, so the csum_help parameter can be removed from the function entirely. -ed