From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Paul E. McKenney" Subject: Re: Question on clock drift with NO_HZ_FULL vs NO_HZ_IDLE Date: Sun, 23 Feb 2014 08:38:49 -0800 Message-ID: <20140223163849.GV4250@linux.vnet.ibm.com> References: Reply-To: paulmck@linux.vnet.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-rt-users@vger.kernel.org, fweisbec@gmail.com To: Joel Fernandes Return-path: Received: from e39.co.us.ibm.com ([32.97.110.160]:44044 "EHLO e39.co.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751262AbaBWQi4 (ORCPT ); Sun, 23 Feb 2014 11:38:56 -0500 Received: from /spool/local by e39.co.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Sun, 23 Feb 2014 09:38:55 -0700 Received: from b03cxnp07028.gho.boulder.ibm.com (b03cxnp07028.gho.boulder.ibm.com [9.17.130.15]) by d03dlp03.boulder.ibm.com (Postfix) with ESMTP id 2E81619D8042 for ; Sun, 23 Feb 2014 09:38:51 -0700 (MST) Received: from d03av06.boulder.ibm.com (d03av06.boulder.ibm.com [9.17.195.245]) by b03cxnp07028.gho.boulder.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id s1NGcKoO6029808 for ; Sun, 23 Feb 2014 17:38:20 +0100 Received: from d03av06.boulder.ibm.com (loopback [127.0.0.1]) by d03av06.boulder.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id s1NGgI7v012328 for ; Sun, 23 Feb 2014 09:42:19 -0700 Content-Disposition: inline In-Reply-To: Sender: linux-rt-users-owner@vger.kernel.org List-ID: On Sun, Feb 23, 2014 at 01:20:53AM -0600, Joel Fernandes wrote: > Hi, > After reading documentation, I follow that NO_HZ_FULL turns off the > scheduling-clock interrupt on all except the boot CPU, where as > NO_HZ_IDLE turns it off on all idle CPUs. > > I understand the period scheduling-clock interrupt is required to > compensate for hardware oscillator drift. I believe that's why > NO_HZ_FULL keeps it ON on atleast one CPU. That way the drift is > compensated for. > > But, with NO_HZ_IDLE, all CPUs turn off interrupt during idle. Then > how is the drift accounted for? > > CC'ing Paul as well for any guidance, thanks :) The clock code fixes things up as needed when the first CPU returns from idle. Thanx, Paul