From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chanwoo Choi Subject: Re: [RESEND][PATCH] cpufreq: stats: Add 'load_table' sysfs file to show accumulated data of CPU Date: Fri, 14 Jun 2013 11:11:39 +0900 Message-ID: <51BA7BDB.9070801@samsung.com> References: <1370419882-16831-1-git-send-email-cw00.choi@samsung.com> <11735265.3zXb8Z81XT@vostro.rjw.lan> <2801243.ZfLEluCnpU@vostro.rjw.lan> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: In-reply-to: <2801243.ZfLEluCnpU@vostro.rjw.lan> Sender: linux-kernel-owner@vger.kernel.org To: "Rafael J. Wysocki" Cc: Viresh Kumar , linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org, cpufreq@vger.kernel.org, kyungmin.park@samsung.com, myungjoo.ham@samsung.com List-Id: linux-pm@vger.kernel.org On 06/12/2013 08:05 PM, Rafael J. Wysocki wrote: > On Wednesday, June 12, 2013 09:32:16 AM Viresh Kumar wrote: >> On 12 June 2013 03:44, Rafael J. Wysocki wrote: >>> On Wednesday, June 05, 2013 05:11:22 PM Chanwoo Choi wrote: >>>> This patch add new sysfs file to show previous accumulated data of CPU load >>>> as following path. This sysfs file is used to judge the correct system state >>>> or determine suitable system resource on user-space. >>>> - /sys/devices/system/cpu/cpu0/cpufreq/stats/load_table >>>> >>>> This sysfs file include following data: >>>> - Measurement point of time >>>> - CPU frequency >>>> - Per-CPU load >>>> >>>> Signed-off-by: Chanwoo Choi >>>> Signed-off-by: Myungjoo Ham >>>> Signed-off-by: Kyungmin Park >>> >>> Well, first of all, there is the "one value per file" rule for sysfs attributes >>> which is evidently violated by this code. >> >> Even this was enclosed in CONFIG_CPU_FREQ_STAT_DETAILS, >> so even sysfs isn't that bad as we already had something similar here. > > Yes, we did, and yes, it was a mistake. It should have been in debugfs from > the very beginning. > >>> Second, this looks like a feature needed to handle one particular platform, so >>> why do you want to add it to the cpufreq core? >> >> I really felt this would be useful to others. They can track the load on >> all cores for some time and that will really be useful. People can >> understand their loads and system more easily with this patch in. > > If it were in debugfs, I'd have no objections. > OK, I will reimplement load_table node by using debugfs instead of sysfs. Thanks. Best Regards, Chanwoo Choi