From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754750Ab3AXLL4 (ORCPT ); Thu, 24 Jan 2013 06:11:56 -0500 Received: from cantor2.suse.de ([195.135.220.15]:35384 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754412Ab3AXLK7 (ORCPT ); Thu, 24 Jan 2013 06:10:59 -0500 Date: Thu, 24 Jan 2013 12:13:40 +0100 From: Borislav Petkov To: Michal Marek Cc: Alec Bickerton , linux-kbuild@vger.kernel.org, x86@kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH] Add Intel Ivy bridge CPU flags Message-ID: <20130124111340.GB13471@pd.tnic> Mail-Followup-To: Borislav Petkov , Michal Marek , Alec Bickerton , linux-kbuild@vger.kernel.org, x86@kernel.org, linux-kernel@vger.kernel.org References: <51002135.2060302@theangrymob.co.uk> <20130124105343.GA30701@sepie.suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20130124105343.GA30701@sepie.suse.cz> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Jan 24, 2013 at 11:53:43AM +0100, Michal Marek wrote: > On Wed, Jan 23, 2013 at 06:43:17PM +0100, Alec Bickerton wrote: > > Hi, > > > > I'd like to submit the attached patch to add Ivy Bridge processor support to the > > processor type menu in menuconfig. Setting this configures the appropriate > > CFLAGS (core-avx-i) for ivy bridge. > > > > As this is my first kbuild patch, could somebody review it and let me know what > > I've done wrong. ;-) > > You sent it to linux-kbuild instead of x86@kernel.org. The kbuild > maintainer is not really knowledgeable about gcc optimisation options ;). > Nevertheless, some random comments are below. Note that I am not > saying if the config option is useful or not. Normally, such optimizations have a little effect on the kernel because it doesn't normally use fpu stuff and the rest of the instructions we need, we add ourselves (see arch/x86/include/asm/archrandom.h for RDRAND, for example). So, first we'd need persuasive benchmark numbers which show that a kernel built with core-avx-i is faster at some representative workload than mtune=generic one... Thanks. -- Regards/Gruss, Boris. Sent from a fat crate under my desk. Formatting is fine. --