Received-SPF: pass (google.com: domain of linux-kernel+bounces-18532-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240104212439.3276458-1-maheshb@google.com> <ZZczNlXzM8lrZgH5@hoboy.vegasvil.org>
 <CAF2d9jga9oc4OST6PMU=C9rz_NDrURCcLGx-1tP31U00z63vbA@mail.gmail.com> <ZZjdUlaYyHZSiwSM@hoboy.vegasvil.org>
In-Reply-To: <ZZjdUlaYyHZSiwSM@hoboy.vegasvil.org>
From: =?UTF-8?B?TWFoZXNoIEJhbmRld2FyICjgpK7gpLngpYfgpLYg4KSs4KSC4KSh4KWH4KS14KS+4KSwKQ==?= <maheshb@google.com>
Date: Sat, 6 Jan 2024 00:08:57 -0800
Message-ID: <CAF2d9jhnsubL-sw792ZviSXrFB826G-U8OktdEMN1NCe5zuj0Q@mail.gmail.com>
Subject: Re: [PATCHv3 net-next 2/3] ptp: add ioctl interface for ptp_gettimex64any()
To: Richard Cochran <richardcochran@gmail.com>
Cc: Netdev <netdev@vger.kernel.org>, Linux <linux-kernel@vger.kernel.org>, 
	David Miller <davem@davemloft.net>, Jakub Kicinski <kuba@kernel.org>, 
	Eric Dumazet <edumazet@google.com>, Paolo Abeni <pabeni@redhat.com>, 
	Jonathan Corbet <corbet@lwn.net>, John Stultz <jstultz@google.com>, Don Hatchett <hatch@google.com>, 
	Yuliang Li <yuliangli@google.com>, Mahesh Bandewar <mahesh@bandewar.net>, 
	Willem de Bruijn <willemb@google.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Jan 5, 2024 at 8:55=E2=80=AFPM Richard Cochran <richardcochran@gmai=
l.com> wrote:
>
> On Fri, Jan 05, 2024 at 09:51:40AM -0800, Mahesh Bandewar (=E0=A4=AE=E0=
=A4=B9=E0=A5=87=E0=A4=B6 =E0=A4=AC=E0=A4=82=E0=A4=A1=E0=A5=87=E0=A4=B5=E0=
=A4=BE=E0=A4=B0) wrote:
>
> > POSIX clocks are employed in this series for syscall width
> > measurement, potentially leading to misunderstandings about
> > overlapping functionality. However, their roles are distinct and serve
> > different purposes.
>
> I don't see any difference in purposes.  The multi_clock_gettime call
> is a more general solution.  Thus it will obviate the need for any new
> PTP ioctls.
>

I disagree! NICs inherently benefit from bundled PTP devices due to
their superior low-latency, low-overhead, and precise TX/RX
timestamping capabilities. For demanding systems requiring increased
capacity, multiple NICs from various vendors are often deployed.
However, disciplining these diverse PTP devices across the host
demands a flexible approach; a general purpose syscall is not an
answer. The current PHC implementation using ioctls through exported
ptp devices (/dev/ptpX) provides a solid foundation that is per device
(/per NIC).

This series is providing another piece in an existing suite of methods
used for disciplining / precision tuning (along with adjfine, adjtime,
gettime etc.) This addition is to take that precision even further.

Having a general solution for posix timers is a nice addition.
However, expecting a general purpose syscall to eliminate need for
device ioctl is an unreasonable expectation.

Thanks,
--mahesh..

> Thanks,
> Richard