From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751139AbVHYPQw (ORCPT ); Thu, 25 Aug 2005 11:16:52 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751163AbVHYPQw (ORCPT ); Thu, 25 Aug 2005 11:16:52 -0400 Received: from atlrel7.hp.com ([156.153.255.213]:2446 "EHLO atlrel7.hp.com") by vger.kernel.org with ESMTP id S1751139AbVHYPQt (ORCPT ); Thu, 25 Aug 2005 11:16:49 -0400 Subject: HPET drift question From: Alex Williamson To: venkatesh.pallipadi@intel.com Cc: robert.picco@hp.com, linux-kernel@vger.kernel.org Content-Type: text/plain Organization: LOSL Date: Thu, 25 Aug 2005 09:17:08 -0600 Message-Id: <1124983028.5331.15.camel@tdi> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi Venki, I'm confused by the calculation of the drift value in the hpet driver. The specs defines the recommended minimum hardware implementation is a frequency drift of 0.05% or 500ppm. However, the drift passed in when registering with the time interpolator is: ti->drift = ti->frequency * HPET_DRIFT / 1000000; Isn't that absolute number of ticks per second drift? The time interpolator defines the drift in parts per million. Shouldn't this simply be: ti->drift = HPET_DRIFT; The current code seems to greatly penalize any hpet timer with greater than a 1MHz frequency. Thanks, Alex -- Alex Williamson HP Linux & Open Source Lab