Re: [PATCH RFC net-next 09/12] net: dsa: ocelot: convert to phylink_generic_validate()

From: "Russell King (Oracle)" <linux@armlinux.org.uk>
To: Vladimir Oltean <vladimir.oltean@nxp.com>
Cc: Andrew Lunn <andrew@lunn.ch>,
	Florian Fainelli <f.fainelli@gmail.com>,
	Vivien Didelot <vivien.didelot@gmail.com>,
	Vladimir Oltean <olteanv@gmail.com>,
	Alexandre Belloni <alexandre.belloni@bootlin.com>,
	Claudiu Manoil <claudiu.manoil@nxp.com>,
	George McCollister <george.mccollister@gmail.com>,
	Hauke Mehrtens <hauke@hauke-m.de>,
	Kurt Kanzenbach <kurt@linutronix.de>,
	Woojung Huh <woojung.huh@microchip.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"UNGLinuxDriver@microchip.com" <UNGLinuxDriver@microchip.com>
Subject: Re: [PATCH RFC net-next 09/12] net: dsa: ocelot: convert to phylink_generic_validate()
Date: Wed, 24 Nov 2021 21:21:13 +0000	[thread overview]
Message-ID: <YZ6syUqWKoXJiQGv@shell.armlinux.org.uk> (raw)
In-Reply-To: <20211124200748.mrjuzgwunnn4zjxf@skbuf>

On Wed, Nov 24, 2021 at 08:07:49PM +0000, Vladimir Oltean wrote:
> On Wed, Nov 24, 2021 at 05:53:09PM +0000, Russell King (Oracle) wrote:
> > Populate the supported interfaces and MAC capabilities for the Ocelot
> > DSA switches and remove the old validate implementation to allow DSA to
> > use phylink_generic_validate() for this switch driver.
> > 
> > The felix_vsc9959 and seville_vsc9953 sub-drivers only supports a
> > single interface mode, defined by ocelot_port->phy_mode, so we indicate
> > only this interface mode to phylink. Since phylink restricts the
> > ethtool link modes based on interface, we do not need to make the MAC
> > capabilities dependent on the interface mode.
> 
> Yes, and this driver cannot make use of phylink_generic_validate()
> unless something changes in phylink_get_linkmodes(). You've said a
> number of times that PHY rate adaptation via PAUSE frames is not
> something that is supported, yet it works with 2500base-x and the felix
> driver, and we use this functionality on LS1028A-QDS boards and the
> AQR412 PHY, and customer boards using LS1028A probably use it too. See
> this comment in ocelot_phylink_mac_link_up():

I'll drop this for now, since the issues around rate adaption should
not be handled by phylink_generic_validate(). The point of this
generic helper is to deal with the common case.

We don't get have a way of knowing that the PHY is using rate adaption,
and when rate adaption is in use, it changes the requirements for the
validation path quite substantially. So, we need:

1) phylib to be able to tell us that rate adaption is happening on the
   PHY.
2) change the way we restrict the PHY support/advertisements when
   rate adaption is in use.

I view this as an entirely separate change to this series; it needs to
change in phylink_bringup_phy(), where the restriction is applied to
the PHY, and not in phylink_validate() or below that function.

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!