From: Vincent Donnefort <vincent.donnefort@arm.com>
To: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>,
rjw@rjwysocki.net, vincent.guittot@linaro.org,
qperret@google.com, linux-kernel@vger.kernel.org,
ionela.voinescu@arm.com, lukasz.luba@arm.com,
dietmar.eggemann@arm.com
Subject: Re: [PATCH v2 3/3] PM / EM: Skip inefficient OPPs
Date: Tue, 1 Jun 2021 09:47:25 +0100 [thread overview]
Message-ID: <20210601084725.GA223449@e120877-lin.cambridge.arm.com> (raw)
In-Reply-To: <20210528050934.muji5bv7ed4k4t6j@vireshk-i7>
On Fri, May 28, 2021 at 10:39:34AM +0530, Viresh Kumar wrote:
> On 25-05-21, 10:48, Peter Zijlstra wrote:
> > On Fri, May 21, 2021 at 05:54:24PM +0100, Vincent Donnefort wrote:
> > > diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c
> > > index 4f09afd..5a91a2b 100644
> > > --- a/kernel/sched/cpufreq_schedutil.c
> > > +++ b/kernel/sched/cpufreq_schedutil.c
> > > @@ -10,6 +10,7 @@
> > >
> > > #include "sched.h"
> > >
> > > +#include <linux/energy_model.h>
> > > #include <linux/sched/cpufreq.h>
> > > #include <trace/events/power.h>
> > >
> > > @@ -153,6 +154,9 @@ static unsigned int get_next_freq(struct sugov_policy *sg_policy,
> > >
> > > freq = map_util_freq(util, freq, max);
> > >
> > > + /* Avoid inefficient performance states */
> > > + freq = em_pd_get_efficient_freq(em_cpu_get(policy->cpu), freq);
> > > +
> > > if (freq == sg_policy->cached_raw_freq && !sg_policy->need_freq_update)
> > > return sg_policy->next_freq;
> > >
> >
> > This seems somewhat unfortunate, it adds a loop over the OPPs only to
> > then call into cpufreq to do the exact same thing again :/
>
> And that's why I feel it needs to be done at a single place, either disable the
> OPP (which seems like a bad option based on what Lukasz and Vincent said
> earlier), or make changes in the cpufreq core itself to search for the best
> frequency (like adding another API to mark some frequencies as inefficient, and
> take that into account while selecting next freq).
>
> There is a potential of ending up selecting the wrong frequency here because
> there are too many decision making bodies here and so corner cases.
>
> --
> viresh
Hi Viresh,
Seems like no one has been really convinced about the arguments in favor of
keeping inefficiencies into EM :) Let me then give a shot with marking the OPPs
for the next version.
--
Vincent
next prev parent reply other threads:[~2021-06-01 8:47 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-21 16:54 [PATCH v2 0/3] EM / PM: Inefficient OPPs Vincent Donnefort
2021-05-21 16:54 ` [PATCH v2 1/3] PM / EM: Fix inefficient state detection Vincent Donnefort
2021-05-24 12:41 ` Lukasz Luba
2021-05-25 9:50 ` Quentin Perret
2021-05-21 16:54 ` [PATCH v2 2/3] PM / EM: Extend em_perf_domain with a flag field Vincent Donnefort
2021-05-24 12:44 ` Lukasz Luba
2021-05-25 9:54 ` Quentin Perret
2021-05-21 16:54 ` [PATCH v2 3/3] PM / EM: Skip inefficient OPPs Vincent Donnefort
2021-05-24 12:55 ` Lukasz Luba
2021-05-25 8:48 ` Peter Zijlstra
2021-05-25 9:21 ` Vincent Donnefort
2021-05-25 10:00 ` Vincent Donnefort
2021-05-28 5:09 ` Viresh Kumar
2021-06-01 8:47 ` Vincent Donnefort [this message]
2021-06-01 8:56 ` Viresh Kumar
2021-06-01 9:07 ` Quentin Perret
2021-06-01 9:13 ` Viresh Kumar
2021-05-25 9:33 ` Quentin Perret
2021-05-25 9:46 ` Vincent Donnefort
2021-05-25 11:03 ` Lukasz Luba
2021-05-25 13:06 ` Quentin Perret
2021-05-25 13:34 ` Lukasz Luba
2021-05-25 9:47 ` Vincent Donnefort
2021-05-28 5:04 ` Viresh Kumar
2021-05-28 9:00 ` Lukasz Luba
2021-05-26 3:47 ` [PATCH v2 0/3] EM / PM: Inefficient OPPs Viresh Kumar
2021-05-26 8:56 ` Lukasz Luba
2021-05-26 9:33 ` Viresh Kumar
2021-05-27 7:13 ` Lukasz Luba
2021-05-26 9:01 ` Vincent Donnefort
2021-05-26 9:38 ` Viresh Kumar
2021-05-26 9:39 ` Viresh Kumar
2021-05-26 10:24 ` Lukasz Luba
2021-05-26 10:39 ` Lukasz Luba
2021-05-26 11:50 ` Lukasz Luba
2021-05-26 13:49 ` Vincent Donnefort
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210601084725.GA223449@e120877-lin.cambridge.arm.com \
--to=vincent.donnefort@arm.com \
--cc=dietmar.eggemann@arm.com \
--cc=ionela.voinescu@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lukasz.luba@arm.com \
--cc=peterz@infradead.org \
--cc=qperret@google.com \
--cc=rjw@rjwysocki.net \
--cc=vincent.guittot@linaro.org \
--cc=viresh.kumar@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).