linux-kernel - Re: [PATCH] selftests: timers: improve adjtick output readability

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: <877bzoihsb.ffs@tglx>
Date: Thu, 31 Jul 2025 15:58:12 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Vishal Parmar <vishistriker@...il.com>, John Stultz <jstultz@...gle.com>
Cc: shuah@...nel.org, anna-maria@...utronix.de, frederic@...nel.org,
 sboyd@...nel.org, linux-kernel@...r.kernel.org,
 linux-kselftest@...r.kernel.org
Subject: Re: [PATCH] selftests: timers: improve adjtick output readability

Vishal!

On Wed, Jul 30 2025 at 23:35, Vishal Parmar wrote:

Please do not top-post and trim your replies.

> The intent behind this change is to make output useful as is.
> for example, to provide a performance report in case of regression.

The point John was making:

>> So it might be worth looking into getting the output to be happy with
>> TAP while you're tweaking things here.

The kernel selftests are converting over to standardized TAP output
format, which is intended to aid automated testing.

So if we change the outpot format of this test, then we switch it over to
TAP format and do not invent yet another randomized output scheme.

> CSV format is also a good alternative if the maintainer prefers that.

The most important information is whether the test succeeded or not and
CSV format is not helping either to conform with the test output
standards.

For the success case, the actual numbers are uninteresting. In the
failure case it's sufficient to emit:

        ksft_test_result_fail("Req: NNNN, Exp: $MMMM, Res: $LLLL\n", ...);

In case of regressions (fail), a report providing this output is good
enough for the relevant maintainer/developer to start investigating.

No?

Thanks,

        tglx