From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751095AbWG0IUU (ORCPT ); Thu, 27 Jul 2006 04:20:20 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750967AbWG0IUT (ORCPT ); Thu, 27 Jul 2006 04:20:19 -0400 Received: from dsl027-180-168.sfo1.dsl.speakeasy.net ([216.27.180.168]:32415 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S1750762AbWG0IUS (ORCPT ); Thu, 27 Jul 2006 04:20:18 -0400 Date: Thu, 27 Jul 2006 01:20:37 -0700 (PDT) Message-Id: <20060727.012037.78156999.davem@davemloft.net> To: axboe@suse.de Cc: johnpol@2ka.mipt.ru, drepper@redhat.com, linux-kernel@vger.kernel.org, netdev@vger.kernel.org Subject: Re: async network I/O, event channels, etc From: David Miller In-Reply-To: <20060727081114.GH5282@suse.de> References: <20060727.010255.87351515.davem@davemloft.net> <20060727080901.GG5282@suse.de> <20060727081114.GH5282@suse.de> X-Mailer: Mew version 4.2 on Emacs 21.4 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org From: Jens Axboe Date: Thu, 27 Jul 2006 10:11:15 +0200 > Ownership transition from user -> kernel that is, what I'm trying to say > that returning ownership to the user again is the tricky part. Yes, it is important that for TCP, for example, we don't give the user the event until the data is acknowledged and the skb's referencing that data are fully freed. This is further complicated by the fact that packetization boundaries are going to be different from AIO buffer boundaries. I think this is what you are alluding to.