Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1946363AbWJSSuF (ORCPT ); Thu, 19 Oct 2006 14:50:05 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1946362AbWJSSuF (ORCPT ); Thu, 19 Oct 2006 14:50:05 -0400 Received: from e31.co.us.ibm.com ([32.97.110.149]:1193 "EHLO e31.co.us.ibm.com") by vger.kernel.org with ESMTP id S1946363AbWJSSuB (ORCPT ); Thu, 19 Oct 2006 14:50:01 -0400 Subject: Re: 2.6.18 and tsc clocksource on SMP From: john stultz To: Krzysztof Oledzki Cc: linux-kernel@vger.kernel.org In-Reply-To: References: Content-Type: text/plain Date: Thu, 19 Oct 2006 11:48:16 -0700 Message-Id: <1161283696.6961.137.camel@localhost> Mime-Version: 1.0 X-Mailer: Evolution 2.6.1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1605 Lines: 42 On Wed, 2006-10-18 at 21:32 +0200, Krzysztof Oledzki wrote: > Hello, > > I have just upgraded kernel (2.6.16.x->2.6.18.1) on on of my servers (Dell > PowerEdge1425SC with 2 CPUs with HT - total 4 CPUs) and noticed that > system no longer uses HPET as a clocksource. > > # cat /sys/devices/system/clocksource/clocksource0/available_clocksource > acpi_pm jiffies hpet tsc pit > > # cat /sys/devices/system/clocksource/clocksource0/current_clocksource > tsc Yep. You'll notice the hpet is still available, but by default we select the TSC (unless it is marked as unstable). > AFAIK TSC should not be used on SMP systems, shouldn't it? This was done to unify the behavior between i386 and x86_64. Previously i386 was overly cautious, selecting the TSC only if ACPI PM, or HPET were not available. However this impacted performance as the TSC is much faster. In fact, many distros reversed the order to prefer the TSC. It has been asserted that many Intel SMP systems can safely use the TSC, while AMD smp boxes for the most part cannot. On any system, if the system enters C3, or we notice cpufreq changes in the TSC, it will be disqualified and it will fall back to the hpet or the next best available clocksource. Hopefully this will always select the best clocksource. Please do let us know if you see any problems due to this change. thanks -john - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/