netdev - Re: [PATCH net-next 5/5] net: lan743x: Add Clause-45 MDIO access

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YfMbZC8PIZ/8vwGJ@lunn.ch>
Date:   Thu, 27 Jan 2022 23:23:32 +0100
From:   Andrew Lunn <andrew@...n.ch>
To:     Raju Lakkaraju <Raju.Lakkaraju@...rochip.com>
Cc:     netdev@...r.kernel.org, davem@...emloft.net, kuba@...nel.org,
        UNGLinuxDriver@...rochip.com
Subject: Re: [PATCH net-next 5/5] net: lan743x: Add Clause-45 MDIO access

On Thu, Jan 27, 2022 at 11:00:55PM +0530, Raju Lakkaraju wrote:
> PCI1A011/PCI1A041 chip support the MDIO Clause-45 access
> 
> Signed-off-by: Raju Lakkaraju <Raju.Lakkaraju@...rochip.com>
> ---
>  drivers/net/ethernet/microchip/lan743x_main.c | 110 +++++++++++++++++-
>  drivers/net/ethernet/microchip/lan743x_main.h |  16 +++
>  2 files changed, 123 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/ethernet/microchip/lan743x_main.c b/drivers/net/ethernet/microchip/lan743x_main.c
> index 6f6655eb6438..98d346aaf627 100644
> --- a/drivers/net/ethernet/microchip/lan743x_main.c
> +++ b/drivers/net/ethernet/microchip/lan743x_main.c
> @@ -793,6 +793,95 @@ static int lan743x_mdiobus_write(struct mii_bus *bus,
>  	return ret;
>  }
>  
> +static u32 lan743x_mac_mmd_access(int id, int index, int op, u8 freq)
> +{
> +	u16 dev_addr;
> +	u32 ret;
> +
> +	dev_addr = (index >> 16) & 0x1f;
> +	ret = (id << MAC_MII_ACC_PHY_ADDR_SHIFT_) &
> +		MAC_MII_ACC_PHY_ADDR_MASK_;
> +	ret |= (dev_addr << MAC_MII_ACC_MIIMMD_SHIFT_) &
> +		MAC_MII_ACC_MIIMMD_MASK_;
> +	if (freq)
> +		ret |= (freq << MAC_MII_ACC_MDC_CYCLE_SHIFT_) &
> +			MAC_MII_ACC_MDC_CYCLE_MASK_;

All callers of this function appear to pass freq as 0. So you can
remove this.

> +	if (op == 1)
> +		ret |= MAC_MII_ACC_MIICMD_WRITE_;
> +	else if (op == 2)
> +		ret |= MAC_MII_ACC_MIICMD_READ_;
> +	else if (op == 3)
> +		ret |= MAC_MII_ACC_MIICMD_READ_INC_;
> +	else
> +		ret |= MAC_MII_ACC_MIICMD_ADDR_;

> +		mmd_access = lan743x_mac_mmd_access(phy_id, index, 0, 0);

It is pretty opaque what the 0 means here. How about you actually pass
MAC_MII_ACC_MIICMD_ values?

lan743x_mac_mmd_access(phy_id, index, );

> +		if (adapter->mdiobus->probe_capabilities == MDIOBUS_C45)
> +			phydev->c45_ids.devices_in_package &= ~BIT(0);
>  	}

A MAC driver should not be modifying the phydev structure.

>  	/* MAC doesn't support 1000T Half */
> @@ -2822,12 +2914,14 @@ static int lan743x_mdiobus_init(struct lan743x_adapter *adapter)
>  			sgmii_ctl &= ~SGMII_CTL_SGMII_POWER_DN_;
>  			lan743x_csr_write(adapter, SGMII_CTL, sgmii_ctl);
>  			netif_info(adapter, drv, adapter->netdev, "SGMII operation\n");
> +			adapter->mdiobus->probe_capabilities = MDIOBUS_C22_C45;
>  		} else {
>  			sgmii_ctl = lan743x_csr_read(adapter, SGMII_CTL);
>  			sgmii_ctl &= ~SGMII_CTL_SGMII_ENABLE_;
>  			sgmii_ctl |= SGMII_CTL_SGMII_POWER_DN_;
>  			lan743x_csr_write(adapter, SGMII_CTL, sgmii_ctl);
>  			netif_info(adapter, drv, adapter->netdev, "GMII operation\n");
> +			adapter->mdiobus->probe_capabilities = MDIOBUS_C22;
>  		}
>  	} else {
>  		chip_ver = lan743x_csr_read(adapter, STRAP_READ);
> @@ -2839,19 +2933,29 @@ static int lan743x_mdiobus_init(struct lan743x_adapter *adapter)
>  			sgmii_ctl &= ~SGMII_CTL_SGMII_POWER_DN_;
>  			lan743x_csr_write(adapter, SGMII_CTL, sgmii_ctl);
>  			netif_info(adapter, drv, adapter->netdev, "SGMII operation\n");
> +			adapter->mdiobus->probe_capabilities = MDIOBUS_C22_C45;
>  		} else {
>  			sgmii_ctl = lan743x_csr_read(adapter, SGMII_CTL);
>  			sgmii_ctl &= ~SGMII_CTL_SGMII_ENABLE_;
>  			sgmii_ctl |= SGMII_CTL_SGMII_POWER_DN_;
>  			lan743x_csr_write(adapter, SGMII_CTL, sgmii_ctl);
>  			netif_info(adapter, drv, adapter->netdev, "GMII operation\n");
> +			adapter->mdiobus->probe_capabilities = MDIOBUS_C22;

This manipulation of adapter->mdiobus->probe_capabilities based on
SGMII vs RGMII makes no sense. It should be set based on what the bus
master can actually do. I assume the PCI1A011/PCI1A041 can do both C22
and C45. So it should look at the reg value and either do a C45
transaction, or a C22 transaction. Do the older chips not support C45?
In that case, return -EOPNOTSUPP if asked to do a C45 transaction.

	Andrew