linux-kernel - Re: [PATCH 0/2] selftests/cgroup: better bound for cpu.max tests

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

Message-ID: <4bqk62cqsv3b4sid76zf3jwvyswdym7bl5wf7r6ouwqvmmvsfv@qztfmjdd7nvc>
Date: Wed, 2 Jul 2025 14:34:29 +0200
From: Michal Koutný <mkoutny@...e.com>
To: Shashank Balaji <shashank.mahadasyam@...y.com>
Cc: Tejun Heo <tj@...nel.org>, Johannes Weiner <hannes@...xchg.org>, 
	Shuah Khan <shuah@...nel.org>, cgroups@...r.kernel.org, linux-kselftest@...r.kernel.org, 
	linux-kernel@...r.kernel.org, Shinya Takumi <shinya.takumi@...y.com>
Subject: Re: [PATCH 0/2] selftests/cgroup: better bound for cpu.max tests

Hello Shashank.

On Tue, Jul 01, 2025 at 11:13:54PM +0900, Shashank Balaji <shashank.mahadasyam@...y.com> wrote:
> cpu.max selftests (both the normal one and the nested one) test the
> working of throttling by setting up cpu.max, running a cpu hog process
> for a specified duration, and comparing usage_usec as reported by
> cpu.stat with the duration of the cpu hog: they should be far enough.
> 
> Currently, this is done by using values_close, which has two problems:
> 
> 1. Semantic: values_close is used with an error percentage of 95%, which
>    one will not expect on seeing "values close". The intent it's
> actually going for is "values far".
> 
> 2. Accuracy: the tests can pass even if usage_usec is upto around double
>    the expected amount. That's too high of a margin for usage_usec.
> 
> Overall, this patchset improves the readability and accuracy of the
> cpu.max tests.
> 
> Signed-off-by: Shashank Balaji <shashank.mahadasyam@...y.com>

I think you're getting at an actual bug in the test definition. 

I think that the test_cpucg_max should either run hog_cpus_timed with
CPU_HOG_CLOCK_PROCESS instead of CPU_HOG_CLOCK_WALL to make sense or the
expected_usage_usec should be defined with the configured quota in mind
(i.e. 1/100).  (The latter seems to make the test more natural.)

With such defined metrics, the asserted expression could be
	values_close(usage_usec, expected_usage_usec, 10)
based on your numbers, error is around 20% so our helper's argument is
roughly half of that. (I'd be fine even with err=20 to prevent some
false positives.)

I think those changes could even be in one patch but I leave that up to
you. My comment to your 2nd patch is that I'd like to stick to relative
errors and keep positive values_close() predicate that's used in other
selftests too. (But those 95% in the current code are clumsy given two
different qualities are compared.)

Thanks,
Michal

Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)