netdev - Re: [net-next 03/15] net/mlx5: Add adjphase function to support hardware-only offset control

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: <Y83tvS/b0NeSShpv@hoboy.vegasvil.org>
Date:   Sun, 22 Jan 2023 18:15:25 -0800
From:   Richard Cochran <richardcochran@...il.com>
To:     Jacob Keller <jacob.e.keller@...el.com>
Cc:     Rahul Rameshbabu <rrameshbabu@...dia.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Saeed Mahameed <saeed@...nel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Paolo Abeni <pabeni@...hat.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Saeed Mahameed <saeedm@...dia.com>, netdev@...r.kernel.org,
        Tariq Toukan <tariqt@...dia.com>,
        Gal Pressman <gal@...dia.com>,
        Vincent Cheng <vincent.cheng.xh@...esas.com>
Subject: Re: [net-next 03/15] net/mlx5: Add adjphase function to support
 hardware-only offset control

Wow, lots of confusion in this thread!

On Fri, Jan 20, 2023 at 03:58:25PM -0800, Jacob Keller wrote:

> Sure. I guess what I don't understand is why or when one would want to
> use adjphase instead of just performing frequency adjustment via the
> .adjfine operation...

The difference is in where the frequency adjust is calculated.  There
are two possibilities:

1. It may be done in user space.
   This is NTP timex's ADJ_FREQUENCY.
   PHC implements .adjfine

2. It may be left to kernel space.
   This is NTP timex's ADJ_OFFSET.
   PHC implements .adjphase

In case of #2, the hardware implements some kind of clock servo.  The
inputs to the servo are the reported phase offsets.  The output of the
servo is a frequency adjustment.

> Especially since "gradual adjustment over time" means it will still be
> slow to converge (just like adjusting frequency is today).

It depends on the servo parameters.

> We should definitely improve the doc to explain the diff between them
> and make sure that its more clear to driver implementations.

Sure.

> It also makes it harder to justify mapping small .adjtime to .adjphase,

adjtime is totally different.  Don't mix those two up:

- adjtime  sets the time (phase), plain and simple.

- adjphase feeds the phase offsets into a servo algorithm in hardware.

FWIW, the PHC subsystem handles the semantics of timex ADJ_FREQUENCY
and ADJ_OFFSET in exactly the same way as the NTP subsystem.

Thanks,
Richard