netdev - Re: [PATCH net-next] net: ethtool: allow MAC drivers to override ethtool get_ts

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: <20210121040451.GB14465@hoboy.vegasvil.org>
Date:   Wed, 20 Jan 2021 20:04:51 -0800
From:   Richard Cochran <richardcochran@...il.com>
To:     Russell King - ARM Linux admin <linux@...linux.org.uk>
Cc:     Andrew Lunn <andrew@...n.ch>,
        Heiner Kallweit <hkallweit1@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: ethtool: allow MAC drivers to override
 ethtool get_ts_info

On Thu, Jan 14, 2021 at 10:38:00PM +0000, Russell King - ARM Linux admin wrote:

> So, I think the only way to prevent a regression with the code as
> it is today is that we _never_ support PTP on Marvell PHYs - because
> doing so _will_ break the existing MVPP2 driver's implementation and
> cause a regression.

The situation isn't as bad as it seems.

For one thing, mvpp2 incorrectly selects NETWORK_PHY_TIMESTAMPING.
It really shouldn't.  I'm submitting a fix soon.

As long as the new PHY driver (or at least the PTP bit) depends on
NETWORK_PHY_TIMESTAMPING, then that allows users who _really_ want
that to enable the option at compile time.  This option adds extra
checks into the networking hot path, and almost everyone should avoid
enabling it.

> Right now, there is no option: if a PHY supports PTP, then the only
> option is to use the PHYs PTP. Which is utterly rediculous.
> 
> Unless you can see a way around it. Because I can't.

I still think the original and best method is to hide the two (and
with your new driver, three) esoteric PHY time stamping drivers behind
a Kconfig option, and structure the code to favor PHY devices.

The idea to favor the MACs, from back in July, doesn't really change
the fundamental limitations that

- MAC and PHY time stamping cannot work simultaneously, and

- Users of PHY devices (representing a tiny minority) must enable the
  otherwise uninteresting NETWORK_PHY_TIMESTAMPING option at compile
  time.

Thanks,
Richard