linux-kernel - RE: [PATCH] cpuidle: Add 'high' and 'low' idle state metrics

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <006901d48cef$5bb88a00$13299e00$@net>
Date:   Wed, 5 Dec 2018 15:07:54 -0800
From:   "Doug Smythies" <dsmythies@...us.net>
To:     "'Rafael J. Wysocki'" <rjw@...ysocki.net>
Cc:     "'LKML'" <linux-kernel@...r.kernel.org>,
        "'Linux Documentation'" <linux-doc@...r.kernel.org>,
        "'Peter Zijlstra'" <peterz@...radead.org>,
        "'Daniel Lezcano'" <daniel.lezcano@...aro.org>,
        "'Giovanni Gherdovich'" <ggherdovich@...e.cz>,
        "'Linux PM'" <linux-pm@...r.kernel.org>,
        "Doug Smythies" <dsmythies@...us.net>
Subject: RE: [PATCH] cpuidle: Add 'high' and 'low' idle state metrics

Hi Rafael,

On 2018.12.03 04:32 Rafael J. Wysocki wrote:

> Add two new metrics for CPU idle states, "high" and "low", to count
> the number of times the given state had been asked for (or entered
> from the kernel's perspective), but the observed idle duration turned
> out to be too high or too low for it (respectively).

I wonder about the "high" "low" terminology here.

> These mertics help to estimat the quality of the CPU idle governor
> in use.

Yes, very useful. Thanks.

Here the terms are mixed with "deep" and "shallow"

> +	unsigned long long	high; /* Number of times it's been too deep */
> +	unsigned long long	low;  /* Number of times it's been too shallow */

> +``high``
> +	Total number of times this idle state had been asked for, but the
> +	observed idle duration was too short to match its target residency.
> +

O.K. To avoid ambiguity, how about naming them "too_short" and "too_long"?

> +``low``
> +	Total number of times this idle state had been asked for, but a deeper
> +	idle state would have been a better match for the observed idle duration.

Even though I read the patch, by the time I actually looked at the numbers I had
forgotten the meaning. I had look at idle state 0 and 4 (the deepest for my computer)
to figure it out:

doug@s15:~/c$ cat /sys/devices/system/cpu/cpu0/cpuidle/state0/low
259871
doug@s15:~/c$ cat /sys/devices/system/cpu/cpu0/cpuidle/state0/high
0
doug@s15:~/c$ cat /sys/devices/system/cpu/cpu0/cpuidle/state4/low
0
doug@s15:~/c$ cat /sys/devices/system/cpu/cpu0/cpuidle/state4/high
5584

Because state 0 can not be too short and state 4 can not be too long.

... Doug