From: Andreas Herrmann <aherrmann@suse.com>
To: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: linux-pm@vger.kernel.org,
"Rafael J. Wysocki" <rafael.j.wysocki@intel.com>,
Peter Zijlstra <peterz@infradead.org>,
Frederic Weisbecker <frederic@kernel.org>,
Viresh Kumar <viresh.kumar@linaro.org>,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] cpufreq: pcc-cpufreq: Disable dynamic scaling on many-CPU systems
Date: Tue, 17 Jul 2018 22:13:23 +0200 [thread overview]
Message-ID: <20180717201323.6rwb4yklq46cmtom@suselix> (raw)
In-Reply-To: <5423012.ZZnfdYddaT@aspire.rjw.lan>
On Tue, Jul 17, 2018 at 06:14:58PM +0200, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
>
> The firmware interface used by the pcc-cpufreq driver is
> fundamentally not scalable and using it for dynamic CPU performance
> scaling on systems with many CPUs leads to degraded performance.
>
> For this reason, disable dynamic CPU performance scaling on systems
> with pcc-cpufreq where the number of CPUs present at the driver init
> time is greater than 4. Also make the driver print corresponding
> complaints to the kernel log.
>
> Reported-by: Andreas Herrmann <aherrmann@suse.com>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> ---
>
> -> v2: Rework the messages printed in the problematic case.
I've tested this patch. Effect is as expected: driver loads but use of
ondemand governor is not allowed. Sample output:
[ 40.757519] pcc-cpufreq: (v1.10.00) driver loaded with frequency limits: 1200 MHz, 2800 MHz
[ 40.831705] pcc_cpufreq_init: Too many CPUs, dynamic performance scaling disabled
[ 40.898353] pcc_cpufreq_init: Try to enable a different scaling driver through BIOS settings
[ 40.972327] pcc_cpufreq_init: and complain to the system vendor
[ 41.025620] cpufreq: Can't use ondemand governor as dynamic switching is disallowed. Fallback to performance governor
...
[ 41.187928] cpufreq: Can't use ondemand governor as dynamic switching is disallowed. Fallback to performance governor
Last message is shown for each online CPU in the system (ie. 120x).
Looks good to me.
Andreas
> ---
> drivers/cpufreq/pcc-cpufreq.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> Index: linux-pm/drivers/cpufreq/pcc-cpufreq.c
> ===================================================================
> --- linux-pm.orig/drivers/cpufreq/pcc-cpufreq.c
> +++ linux-pm/drivers/cpufreq/pcc-cpufreq.c
> @@ -589,6 +589,15 @@ static int __init pcc_cpufreq_init(void)
> return ret;
> }
>
> + if (num_present_cpus() > 4) {
> + pcc_cpufreq_driver.flags |= CPUFREQ_NO_AUTO_DYNAMIC_SWITCHING;
> + pr_err("%s: Too many CPUs, dynamic performance scaling disabled\n",
> + __func__);
> + pr_err("%s: Try to enable a different scaling driver through BIOS settings\n",
> + __func__);
> + pr_err("%s: and complain to the system vendor\n", __func__);
> + }
> +
> ret = cpufreq_register_driver(&pcc_cpufreq_driver);
>
> return ret;
>
>
next prev parent reply other threads:[~2018-07-18 7:37 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-17 6:50 Commit 554c8aa8ecad causing severe performance degression with pcc-cpufreq Andreas Herrmann
2018-07-17 7:33 ` Rafael J. Wysocki
2018-07-17 8:03 ` Rafael J. Wysocki
2018-07-17 8:50 ` Andreas Herrmann
2018-07-17 8:58 ` Rafael J. Wysocki
2018-07-17 9:06 ` Rafael J. Wysocki
2018-07-17 9:11 ` Andreas Herrmann
2018-07-17 9:23 ` Rafael J. Wysocki
2018-07-17 9:27 ` Andreas Herrmann
2018-07-17 9:36 ` Andreas Herrmann
2018-07-17 10:09 ` Rafael J. Wysocki
2018-07-17 10:21 ` Andreas Herrmann
2018-07-17 10:23 ` Rafael J. Wysocki
2018-07-17 14:03 ` Andreas Herrmann
2018-07-17 15:29 ` Rafael J. Wysocki
2018-07-17 16:13 ` [PATCH] cpufreq: intel_pstate: Load when ACPI PCCH is present Rafael J. Wysocki
2018-07-17 17:23 ` Srinivas Pandruvada
2018-07-17 17:28 ` Rafael J. Wysocki
2018-07-17 18:06 ` [PATCH] cpufreq: intel_pstate: Register " Rafael J. Wysocki
2018-07-18 10:43 ` Andreas Herrmann
2018-07-18 10:51 ` Rafael J. Wysocki
2018-07-17 10:18 ` Commit 554c8aa8ecad causing severe performance degression with pcc-cpufreq Andreas Herrmann
2018-07-17 8:08 ` Daniel Lezcano
2018-07-17 8:36 ` Andreas Herrmann
2018-07-17 8:52 ` Rafael J. Wysocki
2018-07-17 8:15 ` Peter Zijlstra
2018-07-17 9:05 ` Andreas Herrmann
2018-07-17 12:02 ` [PATCH] cpufreq: pcc-cpufreq: Disable dynamic scaling on many-CPU systems Rafael J. Wysocki
2018-07-17 16:14 ` [PATCH v2] " Rafael J. Wysocki
2018-07-17 20:13 ` Andreas Herrmann [this message]
2018-07-18 7:44 ` Rafael J. Wysocki
2018-07-18 8:23 ` Peter Zijlstra
2018-07-18 9:34 ` Andreas Herrmann
2018-07-18 15:25 ` Commit 554c8aa8ecad causing severe performance degression with pcc-cpufreq Andreas Herrmann
2018-07-18 15:31 ` Andreas Herrmann
2018-07-19 11:04 ` Andreas Herrmann
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=20180717201323.6rwb4yklq46cmtom@suselix \
--to=aherrmann@suse.com \
--cc=frederic@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=rafael.j.wysocki@intel.com \
--cc=rjw@rjwysocki.net \
--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).