From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nishanth Menon Subject: Re: [PATCH] cpufreq: stats: Do not populate stats when policy->cur has no exact match Date: Mon, 18 Nov 2013 09:34:11 -0600 Message-ID: <528A3373.7040205@ti.com> References: <1384561243-30391-1-git-send-email-nm@ti.com> <1423628.uM8DtnYzDe@vostro.rjw.lan> <5287010F.9000508@linaro.org> <528A2D59.3080001@ti.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-pm-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Viresh Kumar Cc: "Rafael J. Wysocki" , "cpufreq@vger.kernel.org" , "linux-pm@vger.kernel.org" , Linux Kernel Mailing List , Carlos Hernandez , Tobias Diedrich , Konrad Rzeszutek Wilk , Dave Jones On 11/18/2013 09:22 AM, Viresh Kumar wrote: > On 18 November 2013 20:38, Nishanth Menon wrote: >> On 11/15/2013 11:22 PM, viresh kumar wrote: > >>> I have a untested patch for this. If this looks okay, Nishant can you please try >>> below patch? With some fixups from your side :) >> >> http://pastebin.mozilla.org/3628975 >> >> I agree that it does show something > > Were you required to update it or fix it? Or you used it as is? as is - though manually applied ;) > >> but would we not rather prefer to >> stick with the entries available in freq_table than have to deal with >> invalid frequencies that may be provided by the driver? for example: >> how do we in stat know that there will only be one invalid frequency >> request? > > We don't have to. Actually there is nothing like an invalid freq, as system > is able to run with it. Its only if kernel is allowed to switch to that > frequency or not. depends on how we look at it.. if we consider freq_table as a list of valid frequencies, anything that is not in there is an invalid entry. why should stat support it when freq_table is enforced by cpufreq layer? > > This information will be useful for platform developers to know that kernel > was running on some out of freq-table freq for some time, what that > frequency was their job to find out and kernel doesn't need to print all > out of range frequencies. Keeping a single entry for that is more than > sufficient.. yeah, I guessed that.. no strong feelings other wise.. > > Though I personally didn't like: 4294967295 to be printed there.. Will > see if I can improve that, but its obviously looks better than failing > in cpufreq-stats.. > >> there are some patch wrapping that thunderbird tends to do - I prefer >> mutt that way, when I need to send inline patches :). > > There are steps in Documentation/email-clients.txt to get that fixed.. I > have used thunderbird for several years and never found anything > wrong with sending inline patches.. :) -> https://patchwork.kernel.org/patch/3192211/ shows how confused my git am feels ;) -- Regards, Nishanth Menon