linux-kernel - Re: [PATCH net-next v5 08/16] net: ethtool: Add a command to expose current time stamping layer

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: <20231013090020.34e9f125@kernel.org>
Date:   Fri, 13 Oct 2023 09:00:20 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Köry Maincent <kory.maincent@...tlin.com>
Cc:     Florian Fainelli <florian.fainelli@...adcom.com>,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-doc@...r.kernel.org,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        "David S . Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Paolo Abeni <pabeni@...hat.com>,
        Jonathan Corbet <corbet@....net>,
        Jay Vosburgh <j.vosburgh@...il.com>,
        Andy Gospodarek <andy@...yhouse.net>,
        Nicolas Ferre <nicolas.ferre@...rochip.com>,
        Claudiu Beznea <claudiu.beznea@...on.dev>,
        Horatiu Vultur <horatiu.vultur@...rochip.com>,
        UNGLinuxDriver@...rochip.com,
        Broadcom internal kernel review list 
        <bcm-kernel-feedback-list@...adcom.com>,
        Andrew Lunn <andrew@...n.ch>,
        Heiner Kallweit <hkallweit1@...il.com>,
        Russell King <linux@...linux.org.uk>,
        Richard Cochran <richardcochran@...il.com>,
        Radu Pirea <radu-nicolae.pirea@....nxp.com>,
        Willem de Bruijn <willemdebruijn.kernel@...il.com>,
        Vladimir Oltean <vladimir.oltean@....com>,
        Michael Walle <michael@...le.cc>,
        Jacob Keller <jacob.e.keller@...el.com>,
        Maxime Chevallier <maxime.chevallier@...tlin.com>
Subject: Re: [PATCH net-next v5 08/16] net: ethtool: Add a command to expose
 current time stamping layer

On Tue, 10 Oct 2023 10:23:43 +0200 Köry Maincent wrote:
> > > +/*
> > > + * Hardware layer of the TIMESTAMPING provider
> > > + * New description layer should have the NETDEV_TIMESTAMPING or
> > > + * PHYLIB_TIMESTAMPING bit set to know which API to use for timestamping.    
> > 
> > If we are talking about hardware layers, then we shall use either 
> > PHY_TIMESTAMPING or MAC_TIMESTAMPING. PHYLIB is the sub-subsystem to 
> > deal with Ethernet PHYs, and netdev is the object through which we 
> > represent network devices, so they are not even quite describing similar 
> > things. If you go with the {PHY,MAC}_TIMESTAMPING suggestion, then I 
> > could see how we could somewhat easily add PCS_TIMESTAMPING for instance.  
> 
> I am indeed talking about hardware layers but I updated the name to use NETDEV
> and PHYLIB timestamping for a reason. It is indeed only PHY or MAC timestamping
> for now but it may be expanded in the future to theoretically to 7 layers of
> timestamps possible. Also there may be several possible timestamp within a MAC
> device precision vs volume.
> See the thread of my last version that talk about it:
> https://lore.kernel.org/netdev/20230511203646.ihljeknxni77uu5j@skbuf/
> 
> All these possibles timestamps go through exclusively the netdev API or the
> phylib API. Even the software timestamping is done in the netdev driver,
> therefore it goes through the netdev API and then should have the
> NETDEV_TIMESTAMPING bit set.

Netdev vs phylib is an implementation detail of Linux.
I'm also surprised that you changed this.

> > > + */
> > > +enum {
> > > +	NO_TIMESTAMPING = 0,
> > > +	NETDEV_TIMESTAMPING = (1 << 0),
> > > +	PHYLIB_TIMESTAMPING = (1 << 1),
> > > +	SOFTWARE_TIMESTAMPING = (1 << 2) | (1 << 0),    
> > 
> > Why do we have to set NETDEV_TIMESTAMPING here, or is this a round-about 
> > way of enumerating 0, 1, 2 and 3?  
> 
> I answered you above the software timestamping should have the
> NETDEV_TIMESTAMPING bit set as it is done from the net device driver.
> 
> What I was thinking is that all the new timestamping should have
> NETDEV_TIMESTAMPING or PHYLIB_TIMESTAMPING set to know which API to pass
> through.
> Like we could add these in the future:
> MAC_DMA_TIMESTAMPING = (2 << 2) | (1 >> 0),
> MAC_PRECISION_TIMESTAMPING = (3 << 2) | (1 >> 0),
> ...
> PHY_SFP_TIMESTAMPING = (2 << 2) | (1 << 1),
> ...

What is "PRECISION"? DMA is a separate block like MAC and PHY.