linux-kernel - Re: [PATCH 2/3] bcache: Convert to lib/time

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: <20220831173639.sb6bhg5xvn4gbkiu@moria.home.lan>
Date:   Wed, 31 Aug 2022 13:36:39 -0400
From:   Kent Overstreet <kent.overstreet@...ux.dev>
To:     Coly Li <colyli@...e.de>
Cc:     linux-kernel@...r.kernel.org, linux-bcache@...r.kernel.org
Subject: Re: [PATCH 2/3] bcache: Convert to lib/time_stats

On Thu, Sep 01, 2022 at 01:02:37AM +0800, Coly Li wrote:
> Can I understand that rate = 1/frequency ?  Then frequency 4 is around to rate 0.

Correct, rate = 1/frequency.

> > The quantiles are for the duration, they give you an idea of the statistical
> > distribution, see https://en.wikipedia.org/wiki/Quantile
> 
> I wanted to ask how to read the quantiles line. Does it mean that 1 ns is
> equally divided by 15 segments, and the counter values are for the divided
> 1/15 ns segments?

It is the 15 boundaries between 16 equal segments, yes.

> Quantiles are much humane, standard deviation is scared… Just FYI LOL.

That's valuable feedback then, maybe we'll keep both. The more expensive part of
time stats is just ingesting data points, but we've got a percpu buffer frontend
for that - just doing a bit more moth doesn't cost much.