From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: [RFC PATCH net-next-2.6] gro: drivers should feed GRO only with TCP packets Date: Wed, 1 Sep 2010 10:55:19 -0700 Message-ID: <20100901105519.603492cf@nehalam> References: <1283360686.2556.381.camel@edumazet-laptop> <20100901104235.5476a242@nehalam> <20100901.104805.35816252.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: eric.dumazet@gmail.com, netdev@vger.kernel.org To: David Miller Return-path: Received: from mail.vyatta.com ([76.74.103.46]:36323 "EHLO mail.vyatta.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755396Ab0IARzV (ORCPT ); Wed, 1 Sep 2010 13:55:21 -0400 In-Reply-To: <20100901.104805.35816252.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 01 Sep 2010 10:48:05 -0700 (PDT) David Miller wrote: > From: Stephen Hemminger > Date: Wed, 1 Sep 2010 10:42:35 -0700 > > > There was talk of doing GRO for UDP as well but it never > > got implemented. > > I suppose the most direct application would be for IP fragmentation. > > In fact I could see that working very well. > > But for non-fragmented UDP... I can't see much value to that. > > We'd have to preserve the packet boundaries, and process each chunk of > data as one packet at a time within those boundaries, to retain > datagram recvmsg() semantics. I was thinking for fragmented UDP, which still unfortunately gets used.