From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754391Ab2LDQEA (ORCPT ); Tue, 4 Dec 2012 11:04:00 -0500 Received: from mga14.intel.com ([143.182.124.37]:3924 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752272Ab2LDQD7 (ORCPT ); Tue, 4 Dec 2012 11:03:59 -0500 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.84,215,1355126400"; d="scan'208";a="176029760" Date: Tue, 4 Dec 2012 16:06:18 +0000 From: Alan Cox To: "Winkler, Tomas" Cc: "gregkh@linuxfoundation.org" , Alan Cox , "arnd@arndb.de" , "linux-kernel@vger.kernel.org" , Stanislaw Gruszka Subject: Re: [char-misc-next 07/15] mei: use wrietl/readl instead of io wrappers Message-ID: <20121204160618.585ac53d@bob.linux.org.uk> In-Reply-To: <5B8DA87D05A7694D9FA63FD143655C1B01E9DF12@HASMSX106.ger.corp.intel.com> References: <1354629884-3202-1-git-send-email-tomas.winkler@intel.com> <1354629884-3202-7-git-send-email-tomas.winkler@intel.com> <20121204143151.1cce530a@pyramind.ukuu.org.uk> <5B8DA87D05A7694D9FA63FD143655C1B01E9DED6@HASMSX106.ger.corp.intel.com> <20121204153716.322b2cc7@bob.linux.org.uk> <5B8DA87D05A7694D9FA63FD143655C1B01E9DF12@HASMSX106.ger.corp.intel.com> Organization: Intel X-Mailer: Claws Mail 3.8.1 (GTK+ 2.24.13; x86_64-redhat-linux-gnu) Organisation: Intel Corporation UK Ltd, registered no. 1134945 (England), Registered office Pipers Way, Swindon, SN3 1RJ Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 4 Dec 2012 16:02:17 +0000 "Winkler, Tomas" wrote: > > > > As to overhead. I'd love to see a measurement that can detect > > > > the difference. Do you have a benchmark that shows it ? > > > > > > Don't have any numbers but it seems reasonable to me. > > > > Processors are pretty good at prediction these days. I've not seen > > any evidence to support the assumption. > > I don't have strong opinion about it, yet what would be benefit of > keeping the io abstraction layer, In case we are sure it is not used > with io ports? Less changes to existing working code being the obvious one. Alan