Date:   Tue, 5 May 2020 15:35:59 +0100
From:   Russell King - ARM Linux admin <linux@...linux.org.uk>
To:     Calvin Johnson <calvin.johnson@....nxp.com>
Cc:     "Rafael J . Wysocki" <rafael@...nel.org>, linux.cj@...il.com,
        Jeremy Linton <jeremy.linton@....com>,
        Andrew Lunn <andrew@...n.ch>,
        Andy Shevchenko <andy.shevchenko@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Cristi Sovaiala <cristian.sovaiala@....com>,
        Florin Laurentiu Chiculita <florinlaurentiu.chiculita@....com>,
        Ioana Ciornei <ioana.ciornei@....com>,
        Madalin Bucur <madalin.bucur@....nxp.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Heikki Krogerus <heikki.krogerus@...ux.intel.com>,
        Varun Sethi <V.Sethi@....com>,
        "Rajesh V . Bikkina" <rajesh.bikkina@....com>,
        linux-acpi@...r.kernel.org, linux-kernel@...r.kernel.org,
        Diana Madalina Craciun <diana.craciun@....com>,
        netdev@...r.kernel.org, Marcin Wojtas <mw@...ihalf.com>,
        Laurentiu Tudor <laurentiu.tudor@....com>,
        Makarand Pawagi <makarand.pawagi@....com>,
        linux-arm-kernel@...ts.infradead.org,
        Pankaj Bansal <pankaj.bansal@....com>,
        "David S. Miller" <davem@...emloft.net>,
        Heiner Kallweit <hkallweit1@...il.com>
Subject: Re: [net-next PATCH v3 3/5] phylink: Introduce
 phylink_fwnode_phy_connect()

On Tue, May 05, 2020 at 06:59:03PM +0530, Calvin Johnson wrote:
> Define phylink_fwnode_phy_connect() to connect phy specified by
> a fwnode to a phylink instance.
> 
> Signed-off-by: Calvin Johnson <calvin.johnson@....nxp.com>
> ---
> 
> Changes in v3:
>   remove NULL return check as it is invalid
>   remove unused phylink_device_phy_connect()
> 
> Changes in v2:
>   replace of_ and acpi_ code with generic fwnode to get phy-handle.
> 
>  drivers/net/phy/phylink.c | 48 +++++++++++++++++++++++++++++++++++++++
>  include/linux/phylink.h   |  3 +++
>  2 files changed, 51 insertions(+)
> 
> diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c
> index 0f23bec431c1..560d1069426c 100644
> --- a/drivers/net/phy/phylink.c
> +++ b/drivers/net/phy/phylink.c
> @@ -961,6 +961,54 @@ int phylink_connect_phy(struct phylink *pl, struct phy_device *phy)
>  }
>  EXPORT_SYMBOL_GPL(phylink_connect_phy);
>  
> +/**
> + * phylink_fwnode_phy_connect() - connect the PHY specified in the fwnode.
> + * @pl: a pointer to a &struct phylink returned from phylink_create()
> + * @fwnode: a pointer to a &struct fwnode_handle.
> + * @flags: PHY-specific flags to communicate to the PHY device driver
> + *
> + * Connect the phy specified @fwnode to the phylink instance specified
> + * by @pl. Actions specified in phylink_connect_phy() will be
> + * performed.
> + *
> + * Returns 0 on success or a negative errno.
> + */
> +int phylink_fwnode_phy_connect(struct phylink *pl,
> +			       struct fwnode_handle *fwnode,
> +			       u32 flags)
> +{
> +	struct fwnode_handle *phy_fwnode;
> +	struct phy_device *phy_dev;
> +	int ret = 0;
> +
> +	/* Fixed links and 802.3z are handled without needing a PHY */
> +	if (pl->cfg_link_an_mode == MLO_AN_FIXED ||
> +	    (pl->cfg_link_an_mode == MLO_AN_INBAND &&
> +	     phy_interface_mode_is_8023z(pl->link_interface)))
> +		return 0;
> +
> +	phy_fwnode = fwnode_get_phy_node(fwnode);
> +	if ((IS_ERR(phy_fwnode)) && pl->cfg_link_an_mode == MLO_AN_PHY)
> +		return -ENODEV;

This doesn't reflect the behaviour of phylink_of_phy_connect() - it is
*not* a cleanup of what is there, which is:

                if (!phy_node) {
                        if (pl->cfg_link_an_mode == MLO_AN_PHY)
                                return -ENODEV;
                        return 0;
                }

which does:

- if there is a PHY node, find the PHY and connect it.
- if there is no PHY node, then:
   + if we are expecting a PHY to be present, return an error.
   + otherwise, it is not a problem, continue.

That is very important behaviour - it allows drivers to call
phylink_*_phy_connect() without knowing whether there should or should
not be a PHY - and keeps that knowledge within phylink.  It means
network drivers don't have to parse the firmware to find out if there's
a fixed link or SFP cage attached, and decide whether to call these
functions.

> +
> +	phy_dev = fwnode_phy_find_device(phy_fwnode);
> +	fwnode_handle_put(phy_fwnode);
> +	if (!phy_dev)
> +		return -ENODEV;
> +
> +	ret = phy_attach_direct(pl->netdev, phy_dev, flags,
> +				pl->link_interface);
> +	if (ret)
> +		return ret;
> +
> +	ret = phylink_bringup_phy(pl, phy_dev, pl->link_config.interface);
> +	if (ret)
> +		phy_detach(phy_dev);
> +
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(phylink_fwnode_phy_connect);
> +

I think we need to go further with this, and we need to have
phylink_fwnode_phy_connect() functionally identical to
phylink_of_phy_connect() for DT-based fwnodes.  Doing so will avoid
introducing errors such as the one you've added above.

The only difference between these two is that DT has a number of
legacy properties - these can be omitted if the fwnode is not a DT
node.

Remember that fwnode is compatible with DT, so fwnode_phy_find_device()
can internally decide whether to look for the ACPI property or one of
the three DT properties.

It also means that phylink_of_phy_connect() can become:

int phylink_of_phy_connect(struct phylink *pl, struct device_node *dn,
                           u32 flags)
{
        return phylink_fwnode_phy_connect(pl, of_fwnode_handle(dn), flags);
}

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 10.2Mbps down 587kbps up

Powered by blists - more mailing lists