From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eduardo Valentin Subject: Re: [PATCH 0/2] thermal, OPP: move the CPU power estimation to the OPP library Date: Wed, 10 Jan 2018 11:34:33 -0800 Message-ID: <20180110193431.GE3837@localhost.localdomain> References: <20180109110252.13557-1-quentin.perret@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mail-pg0-f65.google.com ([74.125.83.65]:46164 "EHLO mail-pg0-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752790AbeAJTe2 (ORCPT ); Wed, 10 Jan 2018 14:34:28 -0500 Received: by mail-pg0-f65.google.com with SMTP id r2so122703pgq.13 for ; Wed, 10 Jan 2018 11:34:28 -0800 (PST) Content-Disposition: inline In-Reply-To: <20180109110252.13557-1-quentin.perret@arm.com> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Quentin Perret Cc: linux-pm@vger.kernel.org, rjw@rjwysocki.net, vireshk@kernel.org, nm@ti.com, sboyd@codeaurora.org, sudeep.holla@arm.com, amit.kachhap@gmail.com, javi.merino@kernel.org, rui.zhang@intel.com, matthias.bgg@gmail.com, dietmar.eggemann@arm.com, morten.rasmussen@arm.com, patrick.bellasi@arm.com, ionela.voinescu@arm.com On Tue, Jan 09, 2018 at 11:02:50AM +0000, Quentin Perret wrote: > Currently, IPA estimates the power dissipated by a CPU at each available OPP > using its capacitance (the dynamic-power-coefficient DT binding). This series > relocates this feature into the OPP library as a preparation for future > changes. More specifically: > > 1. The current DT-based approach for power estimation will need deep > changes to support SCMI-provided power values. While the thermal > subsystem is not necessarily the best place to hide multiple power > estimation methods, the OPP library appears to be a good candidate to > implement the required platform abstraction. > 2. The energy models of CPUs will be needed by other clients in the future > (such as the task scheduler or CPUFreq governors for example) in order > to make energy-aware decisions. The relocation to the OPP library will > enable code re-use and all clients will benefit form the platform > abstraction mentioned previously. To be quite frank, I am happy to see this leaving thermal subsystem. However, a few concerns with the patch set as it is. First, I am not convinced PM OPP is the right place to put this, nor I see a good explanation put in the patch set why it must be part of PM OPP. Second, looks like we are following ARM "good" practice of fixing problems of the future. I would only really sign off for this series when we see real "other future users", otherwise we end up with the infamous static power scenario in 2-3 years down the row. If we currently do not have users of this IN MAINLINE KERNEL, then the series is not for upstream. > > Quentin Perret (2): > PM / OPP: introduce an OPP power estimation helper > thermal: cpu_cooling: use power models from the OPP library > > drivers/cpufreq/arm_big_little.c | 2 ++ > drivers/cpufreq/cpufreq-dt.c | 2 ++ > drivers/cpufreq/mediatek-cpufreq.c | 2 ++ > drivers/opp/core.c | 40 +++++++++++++++++++++++++ > drivers/opp/of.c | 61 ++++++++++++++++++++++++++++++++++++++ > drivers/opp/opp.h | 4 +++ > drivers/thermal/cpu_cooling.c | 33 ++++++--------------- > include/linux/pm_opp.h | 20 +++++++++++++ > 8 files changed, 140 insertions(+), 24 deletions(-) > > -- > 2.15.1 >