Re: [PATCH net-next 4/7] net: phy: marvell10g: Add support for 2.5GBASET and 5GBASET

From: Maxime Chevallier <maxime.chevallier@bootlin•com>
To: Andrew Lunn <andrew@lunn•ch>
Cc: Florian Fainelli <f.fainelli@gmail•com>,
	Antoine Tenart <antoine.tenart@bootlin•com>,
	netdev@vger•kernel.org, gregory.clement@bootlin•com,
	Russell King <linux@armlinux•org.uk>,
	linux-kernel@vger•kernel.org, nadavh@marvell•com,
	thomas.petazzoni@bootlin•com, miquel.raynal@bootlin•com,
	stefanc@marvell•com, mw@semihalf•com, davem@davemloft•net,
	linux-arm-kernel@lists•infradead.org,
	Heiner Kallweit <hkallweit1@gmail•com>
Subject: Re: [PATCH net-next 4/7] net: phy: marvell10g: Add support for 2.5GBASET and 5GBASET
Date: Tue, 22 Jan 2019 11:08:51 +0100	[thread overview]
Message-ID: <20190122110851.35c316d1@bootlin.com> (raw)
In-Reply-To: <20190121201715.GA20277@lunn.ch>

Hello Andrew, Russell,

On Mon, 21 Jan 2019 21:17:15 +0100
Andrew Lunn <andrew@lunn•ch> wrote:

>> @@ -264,8 +265,10 @@ static int mv3310_config_init(struct phy_device *phydev)
>>  	if (ret)
>>  		return ret;
>>  
>> -	linkmode_and(phydev->advertising, phydev->advertising,
>> -		     phydev->supported);
>> +	/* Make sure we advertise all the supported modes, and not just the
>> +	 * default one specified in the driver's .features.
>> +	 */
>> +	linkmode_copy(phydev->advertising, phydev->supported);  
>
>So by doing a copy of supported into advertising, you can stomping
>over any restrictions applied via of_set_phy_supported(),
>of_set_phy_eee_broken(phydev), and any pause control settings which
>might of happened.

Thanks for the explanations, this is indeed clearly not a good solution.

>What might make sense here is that a PHY driver can replace its
>.features member at run time, in its config_init() call. The core then
>needs to perform these evaluations. So i'm guessing we need to split
>this code out of probe() and move it into phy_init_hw()?

So the .features won't be read-only anymore ? We could also simply make
a helper that would add a mode to both the supported and advertising
modes list, that would be used in the 'genphy_c45_pma_read_abilities'
and config_init ?

I lack the big picture of the PHY init sequence, there seems to be a
lot of quirks and complex cases that we need to take into account, so
I'll let you decide :)

>Heiner, you know this code better than anybody. What do you think?
>
>	Andrew

Thanks for the feedback,

Maxime

-- 
Maxime Chevallier, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists•infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel