netdev - Re: [PATCH net-next 0/9] mlxsw: Add PTP support for Spectrum-2 and newer ASICs

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: <YuI0tWAXDBE5joAl@shredder>
Date:   Thu, 28 Jul 2022 10:03:17 +0300
From:   Ido Schimmel <idosch@...dia.com>
To:     Richard Cochran <richardcochran@...il.com>
Cc:     netdev@...r.kernel.org, davem@...emloft.net, kuba@...nel.org,
        pabeni@...hat.com, edumazet@...gle.com, petrm@...dia.com,
        amcohen@...dia.com, danieller@...dia.com, mlxsw@...dia.com
Subject: Re: [PATCH net-next 0/9] mlxsw: Add PTP support for Spectrum-2 and
 newer ASICs

On Wed, Jul 27, 2022 at 08:09:47PM -0700, Richard Cochran wrote:
> On Wed, Jul 27, 2022 at 05:44:49PM +0300, Ido Schimmel wrote:
> 
> > Right. The hardware can support a TC operation, but I did not find any
> > Linux interfaces to configure it (did I miss something?) nor got any
> > requirements to support it at the moment.
> 
> linuxptp can operate as a TC, but it sounds like your HW does E2E TC
> without any software support.

Yes. I believe linuxptp only implements a two-step transparent clock,
whereas the hardware supports a one-step transparent clock. Like I said,
I'm not aware of any requirements for that at the moment, but I imagine
we would need a per-netdev interface to instruct the kernel/hardware to
adjust the correction field at ingress/egress for specific message
types.

> 
> P2P TC does need SW support, because the ports must generate peer
> delay requests.

Right. The hardware has the ability to add/reduce a number of
nanoseconds (that it gets from software) from the computed latency on a
per-port and ingress/egress basis. 

> 
> Thanks,
> Richard