From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH net-next-2.6] net: pskb_expand_head() optimization Date: Sun, 12 Sep 2010 13:24:45 -0700 (PDT) Message-ID: <20100912.132445.112598025.davem@davemloft.net> References: <20100912104534.GA2056@del.dom.local> <20100912.085833.226777368.davem@davemloft.net> <87zkvmvir9.fsf@benpfaff.org> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: jarkao2@gmail.com, eric.dumazet@gmail.com, netdev@vger.kernel.org To: blp@cs.stanford.edu Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:42707 "EHLO sunset.davemloft.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752802Ab0ILUY2 (ORCPT ); Sun, 12 Sep 2010 16:24:28 -0400 In-Reply-To: <87zkvmvir9.fsf@benpfaff.org> Sender: netdev-owner@vger.kernel.org List-ID: From: Ben Pfaff Date: Sun, 12 Sep 2010 12:55:54 -0700 > David Miller writes: > >>> 3) since skb_clone has some cost too, why e.g. saving only the pointer >>> to the tail of the list in skb_shared_info isn't enough? >> >> Then we won't get the rest of the advantages of using list_head such >> as prefetching during traversals, automatic debugging facilities, et al. > > Did you see the recent patch from Andi Kleen where he proposes > removing this prefetching in most situations because the costs > outweigh the benefits on most modern architectures? > http://permalink.gmane.org/gmane.linux.kernel/1033281 > > I'm not saying that list_head doesn't have other advantages > though. Yes I saw it and I somewhat disagree with him, but don't care enough to argue with him about it. There are much more important things to apply my mind and time to at the moment :-)