From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [patch] Performance enhancement patches for SB1250 MAC Date: Mon, 20 Nov 2006 15:58:25 -0800 (PST) Message-ID: <20061120.155825.59657887.davem@davemloft.net> References: <20061120153548.71f90b7a@freekitty> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: steve.yang@windriver.com, netdev@vger.kernel.org Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:7099 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S966869AbWKTX6V (ORCPT ); Mon, 20 Nov 2006 18:58:21 -0500 To: shemminger@osdl.org In-Reply-To: <20061120153548.71f90b7a@freekitty> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org From: Stephen Hemminger Date: Mon, 20 Nov 2006 15:35:48 -0800 > I am not the final decider on this. Ask Jeff/David? Maybe there is some > more general work that can be done to speed up skb allocation. We used to have a skb_head per-cpu cache in skbuff.c but we killed it because SLAB was nearly as good and using generic code always trumps localized hacks like this. I suggest we don't regress in this area.