Re: [PATCH v4] net: macb: Fix several edge cases in validate

From: Sean Anderson <sean.anderson@seco.com>
To: Jakub Kicinski <kuba@kernel.org>
Cc: netdev@vger.kernel.org, "David S . Miller" <davem@davemloft.net>,
	Nicolas Ferre <nicolas.ferre@microchip.com>,
	Claudiu Beznea <claudiu.beznea@microchip.com>,
	Antoine Tenart <atenart@kernel.org>,
	Russell King <linux@armlinux.org.uk>
Subject: Re: [PATCH v4] net: macb: Fix several edge cases in validate
Date: Tue, 26 Oct 2021 11:30:08 -0400	[thread overview]
Message-ID: <4e430fbb-0908-fd3b-bb6e-ec316ea8d66a@seco.com> (raw)
In-Reply-To: <20211025174401.1de5e95d@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>

Hi Jakub,

On 10/25/21 8:44 PM, Jakub Kicinski wrote:
> On Mon, 25 Oct 2021 13:24:05 -0400 Sean Anderson wrote:
>> There were several cases where validate() would return bogus supported
>> modes with unusual combinations of interfaces and capabilities. For
>> example, if state->interface was 10GBASER and the macb had HIGH_SPEED
>> and PCS but not GIGABIT MODE, then 10/100 modes would be set anyway. In
>> another case, SGMII could be enabled even if the mac was not a GEM
>> (despite this being checked for later on in mac_config()). These
>> inconsistencies make it difficult to refactor this function cleanly.
>
> Since you're respinning anyway (AFAIU) would you mind clarifying
> the fix vs refactoring question? Sounds like it could be a fix for
> the right (wrong?) PHY/MAC combination, but I don't think you're
> intending it to be treated as a fix.
>
> If it's a fix it needs [PATCH net] in the subject and a Fixes tag,
> if it's not a fix it needs [PATCH net-next] in the subject.
>
> This will make the lifes of maintainers and backporters easier,
> thanks :)

I don't know if it's a "fix" per se. The current logic isn't wrong,
since I believe that the configurations where the above patch would make
a difference do not exist. However, as noted in the commit message, this
makes refactoring difficult. For example, one might want to implement
supported_interfaces like

        if (bp->caps & MACB_CAPS_HIGH_SPEED &&
            bp->caps & MACB_CAPS_PCS)
                __set_bit(PHY_INTERFACE_MODE_10GBASER, supported);
        if (macb_is_gem(bp) && bp->caps & MACB_CAPS_GIGABIT_MODE_AVAILABLE) {
                __set_bit(PHY_INTERFACE_MODE_GMII, supported);
		phy_interface_set_rgmii(supported);
                if (bp->caps & MACB_CAPS_PCS)
                        __set_bit(PHY_INTERFACE_MODE_SGMII, supported);
        }
        __set_bit(PHY_INTERFACE_MODE_MII, supported);
        __set_bit(PHY_INTERFACE_MODE_RMII, supported);

but then you still need to check for GIGABIT_MODE in validate to
determine whether 10GBASER should "support" 10/100. See [1] for more
discussion.

If you think this fixes a bug, then the appropriate tag is

Fixes: 7897b071ac3b ("net: macb: convert to phylink")

--Sean

[1] https://lore.kernel.org/netdev/YXaIWFB8Kx9rm%2Fj9@shell.armlinux.org.uk/