From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH] net: mv643xx_eth: Add GRO support Date: Thu, 11 Apr 2013 13:31:19 -0400 (EDT) Message-ID: <20130411.133119.913809939413807690.davem@davemloft.net> References: <20130411150326.GA19978@1wt.eu> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: w@1wt.eu, andrew@lunn.ch, jason@lakedaemon.net, benh@kernel.crashing.org, linux-kernel@vger.kernel.org, florian@openwrt.org, smoch@web.de, paulus@samba.org, buytenh@wantstofly.org, dale@farnsworth.org, netdev@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, linux-arm-kernel@lists.infradead.org To: sebastian.hesselbarth@gmail.com Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org From: Sebastian Hesselbarth Date: Thu, 11 Apr 2013 17:27:03 +0200 > On Thu, Apr 11, 2013 at 5:03 PM, Willy Tarreau wrote: >> On Thu, Apr 11, 2013 at 04:47:49PM +0200, Sebastian Hesselbarth wrote: >>> I tried todays net-next on top of 3.9-rc6 without any gro patch, with >>> the initial >>> patch (Soeren) and your proposed patch (Willy). The results show that >>> both patches >>> allow a significant increase in throughput compared to >>> netif_receive_skb (!gro, !lro) >>> alone. Having gro with lro disabled gives some 2% more throughput >>> compared to lro only. >> >> Indeed this is consistent with my memories, since Eric improved the >> GRO path, it became faster than LRO on this chip. > > I don't have a strong opinion on whether Soeren's or your proposal should > be submitted. But I insist on having one of them in, as GRO significantly > improves the common use case, is enabled by default, and not as > constrained as LRO. I think, as per other drivers, LRO should be eliminated completely from all drivers, including this one, and GRO used exclusively instead.