From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751601Ab1GGVH1 (ORCPT ); Thu, 7 Jul 2011 17:07:27 -0400 Received: from ogre.sisk.pl ([217.79.144.158]:36656 "EHLO ogre.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750756Ab1GGVH0 (ORCPT ); Thu, 7 Jul 2011 17:07:26 -0400 From: "Rafael J. Wysocki" To: MyungJoo Ham Subject: Re: [PATCH v3 2/3] PM / DEVFREQ: add example governors Date: Thu, 7 Jul 2011 23:08:38 +0200 User-Agent: KMail/1.13.6 (Linux/3.0.0-rc5+; KDE/4.6.0; x86_64; ; ) Cc: linux-pm@lists.linux-foundation.org, linux-kernel@vger.kernel.org, Len Brown , Pavel Machek , "Greg Kroah-Hartman" , Kyungmin Park , Jiejing Zhang , Colin Cross , Nishanth Menon , Thomas Gleixner References: <1306471995-4048-1-git-send-email-myungjoo.ham@samsung.com> <201107041043.25150.rjw@sisk.pl> In-Reply-To: MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <201107072308.38622.rjw@sisk.pl> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Monday, July 04, 2011, MyungJoo Ham wrote: > On Mon, Jul 4, 2011 at 5:43 PM, Rafael J. Wysocki wrote: > > On Monday, July 04, 2011, MyungJoo Ham wrote: > >> Hello, > >> > >> 2011/7/3 Rafael J. Wysocki : > >> > Hi, > >> > > >> > On Friday, May 27, 2011, MyungJoo Ham wrote: > >> >> Three CPUFREQ-like governors are provided as examples. > >> >> > >> >> powersave: use the lowest frequency possible. The user (device) should > >> >> set the polling_ms as 0 because polling is useless for this governor. > >> >> > >> >> performance: use the highest freqeuncy possible. The user (device) > >> >> should set the polling_ms as 0 because polling is useless for this > >> >> governor. > >> >> > >> >> simple_ondemand: simplified version of CPUFREQ's ONDEMAND governor. > >> >> > >> >> When a user updates OPP entries (enable/disable/add), OPP framework > >> >> automatically notifies DEVFREQ to update operating frequency > >> >> accordingly. Thus, DEVFREQ users (device drivers) do not need to update > >> >> DEVFREQ manually with OPP entry updates or set polling_ms for powersave > >> >> , performance, or any other "static" governors. > >> >> > >> >> Signed-off-by: MyungJoo Ham > >> >> Signed-off-by: Kyungmin Park > >> [] > >> >> + > >> >> + /* Set the desired frequency based on the load */ > >> >> + a = (unsigned long long) stat.busy_time * stat.current_frequency; > >> > > >> > What's the purpose of the conversion? > >> > >> Assuming that the work speed of a device is proportional to its > >> frequency, it measures the amount of work done. > >> It's time * work/time. For example, during the last 10 second, if the > >> busy_time was 5 sec and frequency was 10MHz, > >> it's "50M", which is same as 20MHz and 2.5 sec. > > > > I understand that, but my question was why you're doing a forced conversion > > to (unsigned long long). > > Ah.. that was for the 64bit operations. > > Both busy_time and current_frequency are 32bit and current_frequency > may be a big number. > > Thus, in order to get "freq" value without losing bits (e.g., if > current_frequency = 1GHz and busy_time = 8000, we get an overflow > without 64bit operations), I've inserted 64bit operations with the > conversion. For the cosmetic reasons, it appears that "u64" looks > better though. You wouldn't need the explicit type casting if you did a = stat.busy_time; a *= stat.current_frequency; Thanks, Rafael