diff mbox series

[4/4] amd-xgbe: Fix network fluctuations when using 1G BELFUSE SFP

Message ID 20210212180010.221129-5-Shyam-sundar.S-k@amd.com
State New
Headers show
Series Bug fixes to amd-xgbe driver | expand

Commit Message

Shyam Sundar S K Feb. 12, 2021, 6 p.m. UTC 
Frequent link up/down events can happen when a Bel Fuse SFP part is
connected to the amd-xgbe device. Try to avoid the frequent link
issues by resetting the PHY as documented in Bel Fuse SFP datasheets.

Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
---
 drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 6 ++++++
 1 file changed, 6 insertions(+)

Comments

Florian Fainelli Feb. 12, 2021, 6:56 p.m. UTC | #1 
On 2/12/2021 10:00 AM, Shyam Sundar S K wrote:
> Frequent link up/down events can happen when a Bel Fuse SFP part is
> connected to the amd-xgbe device. Try to avoid the frequent link
> issues by resetting the PHY as documented in Bel Fuse SFP datasheets.
> 
> Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> ---
>  drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> index 1bb468ac9635..e328fd9bd294 100644
> --- a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> @@ -922,6 +922,12 @@ static bool xgbe_phy_belfuse_phy_quirks(struct xgbe_prv_data *pdata)
>  	if ((phy_id & 0xfffffff0) != 0x03625d10)
>  		return false;
>  
> +	/* Reset PHY - wait for self-clearing reset bit to clear */
> +	reg = phy_read(phy_data->phydev, 0x00);
> +	phy_write(phy_data->phydev, 0x00, reg | 0x8000);
> +	read_poll_timeout(phy_read, reg, !(reg & 0x8000) || reg < 0,
> +			  10000, 50000, true, phy_data->phydev, 0x0);

Can you use the standard register definitions from include/linux/mii.h
here? You are doing a software reset of the PHY through the BMCR.RESET
register, so you might as well make that clear.

> +
>  	/* Disable RGMII mode */
>  	phy_write(phy_data->phydev, 0x18, 0x7007);
>  	reg = phy_read(phy_data->phydev, 0x18);
>
Tom Lendacky Feb. 12, 2021, 7:05 p.m. UTC | #2 
On 2/12/21 12:00 PM, Shyam Sundar S K wrote:
> Frequent link up/down events can happen when a Bel Fuse SFP part is
> connected to the amd-xgbe device. Try to avoid the frequent link
> issues by resetting the PHY as documented in Bel Fuse SFP datasheets.
> 
> Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>

Same comment about Co-developed-by: tag as previous patch.

With that addressed,

Acked-by: Tom Lendacky <thomas.lendacky@amd.com>

> ---
>   drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 6 ++++++
>   1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> index 1bb468ac9635..e328fd9bd294 100644
> --- a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> @@ -922,6 +922,12 @@ static bool xgbe_phy_belfuse_phy_quirks(struct xgbe_prv_data *pdata)
>   	if ((phy_id & 0xfffffff0) != 0x03625d10)
>   		return false;
>   
> +	/* Reset PHY - wait for self-clearing reset bit to clear */
> +	reg = phy_read(phy_data->phydev, 0x00);
> +	phy_write(phy_data->phydev, 0x00, reg | 0x8000);
> +	read_poll_timeout(phy_read, reg, !(reg & 0x8000) || reg < 0,
> +			  10000, 50000, true, phy_data->phydev, 0x0);
> +
>   	/* Disable RGMII mode */
>   	phy_write(phy_data->phydev, 0x18, 0x7007);
>   	reg = phy_read(phy_data->phydev, 0x18);
>
Heiner Kallweit Feb. 12, 2021, 9:14 p.m. UTC | #3 
On 12.02.2021 19:00, Shyam Sundar S K wrote:
> Frequent link up/down events can happen when a Bel Fuse SFP part is
> connected to the amd-xgbe device. Try to avoid the frequent link
> issues by resetting the PHY as documented in Bel Fuse SFP datasheets.
> 
> Signed-off-by: Sudheesh Mavila <sudheesh.mavila@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> ---
>  drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> index 1bb468ac9635..e328fd9bd294 100644
> --- a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
> @@ -922,6 +922,12 @@ static bool xgbe_phy_belfuse_phy_quirks(struct xgbe_prv_data *pdata)
>  	if ((phy_id & 0xfffffff0) != 0x03625d10)
>  		return false;
>  
> +	/* Reset PHY - wait for self-clearing reset bit to clear */
> +	reg = phy_read(phy_data->phydev, 0x00);
> +	phy_write(phy_data->phydev, 0x00, reg | 0x8000);
> +	read_poll_timeout(phy_read, reg, !(reg & 0x8000) || reg < 0,
> +			  10000, 50000, true, phy_data->phydev, 0x0);
> +

Why don't you simply use genphy_soft_reset() ?
Also it's not too nice to use magic register and bit numbers,
there are constants available, e.g. 0x00 = MII_BMCR

>  	/* Disable RGMII mode */
>  	phy_write(phy_data->phydev, 0x18, 0x7007);
>  	reg = phy_read(phy_data->phydev, 0x18);
>
diff mbox series

Patch

diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
index 1bb468ac9635..e328fd9bd294 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
@@ -922,6 +922,12 @@  static bool xgbe_phy_belfuse_phy_quirks(struct xgbe_prv_data *pdata)
 	if ((phy_id & 0xfffffff0) != 0x03625d10)
 		return false;
 
+	/* Reset PHY - wait for self-clearing reset bit to clear */
+	reg = phy_read(phy_data->phydev, 0x00);
+	phy_write(phy_data->phydev, 0x00, reg | 0x8000);
+	read_poll_timeout(phy_read, reg, !(reg & 0x8000) || reg < 0,
+			  10000, 50000, true, phy_data->phydev, 0x0);
+
 	/* Disable RGMII mode */
 	phy_write(phy_data->phydev, 0x18, 0x7007);
 	reg = phy_read(phy_data->phydev, 0x18);