netdev - Re: [PATCH net-next 4/5] tsnep: Add TSN endpoint Ethernet MAC driver

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: <YP8l6cWaQU/2NoIA@lunn.ch>
Date:   Mon, 26 Jul 2021 23:15:21 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     Gerhard Engleder <gerhard@...leder-embedded.com>
Cc:     davem@...emloft.net, kuba@...nel.org, robh+dt@...nel.org,
        michal.simek@...inx.com, netdev@...r.kernel.org,
        devicetree@...r.kernel.org
Subject: Re: [PATCH net-next 4/5] tsnep: Add TSN endpoint Ethernet MAC driver

> +static int tsnep_mdiobus_read(struct mii_bus *bus, int addr, int regnum)
> +{
> +	struct tsnep_adapter *adapter = bus->priv;
> +	u16 data;
> +	int retval;
> +
> +	if (adapter->loopback)
> +		return 0;
> +
> +	retval = tsnep_read_md(adapter, addr, regnum, &data);
> +	if (retval != 0)
> +		return retval;

It appears your MDIO bus can only do C22. Please add a test for C45 and return -EOPNOTSUPP.

> +static void tsnep_phy_link_status_change(struct net_device *netdev)
> +{
> +	struct tsnep_adapter *adapter = netdev_priv(netdev);
> +	struct phy_device *phydev = netdev->phydev;
> +
> +	if (adapter->loopback)
> +		return;
> +
> +	if (adapter->gmii2rgmii) {
> +		u16 val;
> +
> +		if (phydev->link && phydev->speed == 1000)
> +			val = BMCR_SPEED1000;
> +		else
> +			val = BMCR_SPEED100;
> +		tsnep_write_md(adapter, ECM_GMII2RGMII_ADDR,
> +			       ECM_GMII2RGMII_BMCR, val);
> +	}

I _think_ this is wrong. They way the PHYs are chained means you
should not need to do this, the xgmiitorgmii_read_status() does it.
Maybe you have the chaining setup wrong?

> +static int tsnep_phy_open(struct tsnep_adapter *adapter)
> +{
> +	__ETHTOOL_DECLARE_LINK_MODE_MASK(mask);
> +	struct ethtool_eee ethtool_eee;
> +	int retval;
> +
> +	retval = phy_connect_direct(adapter->netdev, adapter->phydev,
> +				    tsnep_phy_link_status_change,
> +				    adapter->phy_mode);
> +	if (retval)
> +		return -EIO;

phy_connect_direct() returns an error code. Use it, rather than
changing it to something else. This applies everywhere. You must have
a good reason to change error codes, and then it is wise to put a
comment why you change it.

> +
> +	/* MAC supports only 100Mbps|1000Mbps full duplex
> +	 * SPE (Single Pair Ethernet) is also an option but not implemented yet
> +	 */
> +	linkmode_zero(mask);
> +	linkmode_set_bit(ETHTOOL_LINK_MODE_Autoneg_BIT, mask);
> +	linkmode_set_bit(ETHTOOL_LINK_MODE_100baseT_Full_BIT, mask);
> +	linkmode_set_bit(ETHTOOL_LINK_MODE_1000baseT_Full_BIT, mask);
> +	linkmode_and(mask, adapter->phydev->supported, mask);
> +	linkmode_copy(adapter->phydev->supported, mask);
> +	linkmode_copy(adapter->phydev->advertising, mask);

You should not be accessing the phydev directly. Use
phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_1000baseT_Half_BIT),
etc.

> +static int tsnep_phy_init(struct tsnep_adapter *adapter)
> +{
> +	struct device_node *dn;
> +	u16 val;
> +	u32 id;
> +	int retval;
> +
> +	retval = of_get_phy_mode(adapter->pdev->dev.of_node,
> +				 &adapter->phy_mode);
> +	if (retval)
> +		adapter->phy_mode = PHY_INTERFACE_MODE_GMII;
> +
> +	dn = of_parse_phandle(adapter->pdev->dev.of_node, "phy-handle", 0);
> +	adapter->phydev = of_phy_find_device(dn);
> +	if (!adapter->phydev)
> +		adapter->phydev = phy_find_first(adapter->mdiobus);
> +	if (!adapter->phydev)
> +		return -EIO;
> +
> +	/* detect optional GMII2RGMII */
> +	retval = tsnep_read_md(adapter, ECM_GMII2RGMII_ADDR, MII_PHYSID1, &val);
> +	if (retval)
> +		return retval;
> +	id = val << 16;
> +	retval = tsnep_read_md(adapter, ECM_GMII2RGMII_ADDR, MII_PHYSID2, &val);
> +	if (retval)
> +		return retval;
> +	id |= val;
> +	if (id == 0)
> +		adapter->gmii2rgmii = true;

This is where i think GMII2RGMII goes wrong. MAC phy-handle should
point to the GMII2RGMII device in DT. The GMII2RGMII should have a
phy-handle which points to the PHY.

> +	/* reset PHY */
> +	retval = tsnep_write_md(adapter, adapter->phydev->mdio.addr, MII_BMCR,
> +				BMCR_RESET);
> +	if (retval)
> +		return retval;
> +
> +	/* reset GMII2RGMII */
> +	if (adapter->gmii2rgmii) {
> +		retval = tsnep_write_md(adapter, ECM_GMII2RGMII_ADDR,
> +					ECM_GMII2RGMII_BMCR, BMCR_RESET);
> +		if (retval)
> +			return retval;
> +		retval = tsnep_write_md(adapter, ECM_GMII2RGMII_ADDR,
> +					ECM_GMII2RGMII_BMCR, BMCR_SPEED100);
> +		if (retval)
> +			return retval;
> +	}

The PHY driver is in control of the PHY, not the MAC. Please remove.

    Andrew