netdev - Re: [PATCH net-next v3] net: mvpp2: tai: warn once if we fail to update our timestamp

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] [day] [month] [year] [list]

Message-ID: <Z3eq2B7vg1RbQBdJ@shell.armlinux.org.uk>
Date: Fri, 3 Jan 2025 09:16:08 +0000
From: "Russell King (Oracle)" <linux@...linux.org.uk>
To: Richard Cochran <richardcochran@...il.com>
Cc: Andrew Lunn <andrew@...n.ch>, Heiner Kallweit <hkallweit1@...il.com>,
	Marcin Wojtas <marcin.s.wojtas@...il.com>,
	Andrew Lunn <andrew+netdev@...n.ch>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	netdev@...r.kernel.org
Subject: Re: [PATCH net-next v3] net: mvpp2: tai: warn once if we fail to
 update our timestamp

On Fri, Jan 03, 2025 at 09:08:22AM +0000, Russell King (Oracle) wrote:
> On Fri, Jan 03, 2025 at 12:43:56AM -0800, Richard Cochran wrote:
> > On Thu, Jan 02, 2025 at 04:26:04PM +0000, Russell King (Oracle) wrote:
> > 
> > > If we fail to read the clock, that will be because the hardware didn't
> > > respond to our request to read it, which means the hardware broke in
> > > some way. We could make mvpp22_tai_tstamp() fail and not provide
> > > timestamps until we have successfully read the HW clock, but we would
> > > still want to print a warning to explain why HW timestamps vanish.
> > 
> > Sure, keep the warning, but also block time stamp delivery.
> > 
> > > This is to catch a spurious failure that may only affects an occasoinal
> > > attempt to read the HW PTP time. Currently, we would never know,
> > > because the kernel is currently completely silent if that were to ever
> > > happen.
> > 
> > Is the failure spurious, or is the hardware broken and won't recover?
> 
> I have absolutely no idea. I've never seen it happen.
> 
> That's why I think going further than I have (and as you are suggesting)
> is totally overkill.

However, I should point out that I don't use PTP as a general rule for
multiple reasons:

1. PTP is not as easy to deploy as NTP
2. Not everything has PTP support, whereas anything can support NTP.
3. PHY-based PTP support (where we need to read system time, read the
   PHY's PTP clock, re-read the system time and interpolate) seems to
   lead to a lot of noise making the stability of PHY based PTP
   inferior to NTP.

Given this, I don't run PTP except for testing.

Therefore, I think I'll drop this patch - it's clearly an approach that
isn't wanted, and we'll just have silent *and* *wrong* hardware
timestamps if this occurs. It seems it's better that users live in
ignorance that their system has a problem.

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!