From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756383AbWK0TBE (ORCPT ); Mon, 27 Nov 2006 14:01:04 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1757833AbWK0TBD (ORCPT ); Mon, 27 Nov 2006 14:01:03 -0500 Received: from omx1-ext.sgi.com ([192.48.179.11]:36764 "EHLO omx1.sgi.com") by vger.kernel.org with ESMTP id S1756383AbWK0TBB (ORCPT ); Mon, 27 Nov 2006 14:01:01 -0500 Date: Mon, 27 Nov 2006 11:00:24 -0800 (PST) From: Christoph Lameter To: Andi Kleen cc: Ingo Molnar , linux-kernel@vger.kernel.org, Andrew Morton , Arjan van de Ven , Thomas Gleixner Subject: Re: [patch] x86: unify/rewrite SMP TSC sync code In-Reply-To: <200611241813.13205.ak@suse.de> Message-ID: References: <20061124170246.GA9956@elte.hu> <200611241813.13205.ak@suse.de> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, 24 Nov 2006, Andi Kleen wrote: > The trouble is that people are using the RDTSC anyways even if the > kernel doesn't. So some synchronization is probably a good idea. It is better to simply leave TSC alone if unsynchronized. If TSCs appear to be in sync (through some sporadic "synchronization") then people will be tempted to use RDTSC because it seems to work/ However, RDTSC will sporadically yield incoherent values (i.e. time earlier than last TSC read) if we have no full synchronization.