From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: TCP transmit performance regression Date: Thu, 05 Jul 2012 16:28:49 +0200 Message-ID: <1341498529.2583.4155.camel@edumazet-glaptop> References: <1341474192.2583.3299.camel@edumazet-glaptop> <1341477192.2583.3415.camel@edumazet-glaptop> <1341481760.2583.3579.camel@edumazet-glaptop> <1341484919.2583.3688.camel@edumazet-glaptop> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: Network Development , David Miller To: Ming Lei Return-path: Received: from mail-ey0-f174.google.com ([209.85.215.174]:34251 "EHLO mail-ey0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751191Ab2GEO2x (ORCPT ); Thu, 5 Jul 2012 10:28:53 -0400 Received: by eaak11 with SMTP id k11so3269049eaa.19 for ; Thu, 05 Jul 2012 07:28:52 -0700 (PDT) In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 2012-07-05 at 22:01 +0800, Ming Lei wrote: > Basically, copy path will be bypassed since hard_header_len > has included the 'overhead' already. Unfortunately this is not done correctly. needed_headroom should be set to SMSC95XX_TX_OVERHEAD_CSUM