From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shilpasri G Bhat Subject: Re: [PATCH v7 6/6] cpufreq: powernv: Add sysfs attributes to show throttle stats Date: Thu, 28 Jan 2016 15:06:54 +0530 Message-ID: <56A9E136.8090900@linux.vnet.ibm.com> References: <1453965941-7363-1-git-send-email-shilpa.bhat@linux.vnet.ibm.com> <1453965941-7363-7-git-send-email-shilpa.bhat@linux.vnet.ibm.com> <20160128084041.GK3935@vireshk> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Return-path: Received: from e23smtp01.au.ibm.com ([202.81.31.143]:41382 "EHLO e23smtp01.au.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754690AbcA1JiD (ORCPT ); Thu, 28 Jan 2016 04:38:03 -0500 Received: from localhost by e23smtp01.au.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 28 Jan 2016 19:38:00 +1000 In-Reply-To: <20160128084041.GK3935@vireshk> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Viresh Kumar Cc: linuxppc-dev@ozlabs.org, linux-kernel@vger.kernel.org, rjw@rjwysocki.net, linux-pm@vger.kernel.org, pc@us.ibm.com, anton@samba.org, ego@linux.vnet.ibm.com, shreyas@linux.vnet.ibm.com, bsingharora@gmail.com, mpe@ellerman.id.au, linux-api@vger.kernel.org Hi Viresh, On 01/28/2016 02:10 PM, Viresh Kumar wrote: > On 28-01-16, 12:55, Shilpasri G Bhat wrote: >> diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu >> index b683e8e..dea4620 100644 >> --- a/Documentation/ABI/testing/sysfs-devices-system-cpu >> +++ b/Documentation/ABI/testing/sysfs-devices-system-cpu >> @@ -271,3 +271,48 @@ Description: Parameters for the CPU cache attributes >> - WriteBack: data is written only to the cache line and >> the modified cache line is written to main >> memory only when it is replaced >> + >> +What: /sys/devices/system/cpu/cpufreq/chip*/throttle_stats > > What about the chip directory ? Shouldn't that be documented? And > shouldn't that mention that this is just for powerpc ? > > And before that, I don't think that you are doing this properly. I am > sorry that I never came to a point where I could review it, and you > continued with it, version after version. > > But, I really have strong objections to the way this is done. And you > are making things more complex then they are. > > So, these stats are per-policy, right ? First of all sorry about the version log. No these stats are not per-policy. They are per-chip. The throttle event is common for all cores in the chip. > > Then why aren't they added on the policy->kobj instead, just like > cpufreq-stats? And maybe inside cpufreq-stats folder only? > > That will solve many complexities you have in place here and will look > sane as well. > > Right now, you have stats as two places, cpu/cpufreq/chip/ and > cpu/cpuX/cpufreq/stats/, which doesn't look wise and adds to > confusion. > > What do you say? > Yes agree that it will be much cleaner with policy->kobj. But using policy->kobj will result in multiple copies of the throttle-chip stats exported for each policy in the chip. And moving it to cpu/cpuX/cpufreq/stats/ will add a dependency on CONFIG_CPU_FREQ_STAT We want throttle attributes to be either in cpu/cpufreq or cpu/cpuX/cpufreq. If multiple copies is not an issue, then I will move it to cpu/cpuX/cpufreq. Thanks and Regards, Shilpa