[RFT PATCH net] net: phy: aquantia: only poll GLOBAL_CFG registers on aqr113c and aqr115c

[RFT PATCH net] net: phy: aquantia: only poll GLOBAL_CFG registers on aqr113c and aqr115c

[Bartosz Golaszewski]

From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>

Commit 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to
start returning real values") introduced a workaround for an issue
observed on aqr115c. However there were never any reports of it
happening on other models and the workaround has been reported to cause
and issue on aqr113c (and it may cause the same on any other model not
supporting 10M mode).

Let's limit the impact of the workaround to aqr113c and aqr115c and poll
the 100M GLOBAL_CFG register instead as both models are known to support
it correctly.

Reported-by: Jon Hunter <jonathanh@nvidia.com>
Closes: https://lore.kernel.org/lkml/7c0140be-4325-4005-9068-7e0fc5ff344d@nvidia.com/
Fixes: 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to start returning real values")
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
---
 drivers/net/phy/aquantia/aquantia_main.c | 29 +++++++++++++++++-------
 1 file changed, 21 insertions(+), 8 deletions(-)

diff --git a/drivers/net/phy/aquantia/aquantia_main.c b/drivers/net/phy/aquantia/aquantia_main.c
index d12e35374231..6e3e0fc6ea27 100644
--- a/drivers/net/phy/aquantia/aquantia_main.c
+++ b/drivers/net/phy/aquantia/aquantia_main.c
@@ -653,13 +653,7 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
 	unsigned long *possible = phydev->possible_interfaces;
 	unsigned int serdes_mode, rate_adapt;
 	phy_interface_t interface;
-	int i, val, ret;
-
-	ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
-					VEND1_GLOBAL_CFG_10M, val, val != 0,
-					1000, 100000, false);
-	if (ret)
-		return ret;
+	int i, val;
 
 	/* Walk the media-speed configuration registers to determine which
 	 * host-side serdes modes may be used by the PHY depending on the
@@ -708,6 +702,25 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
 	return 0;
 }
 
+static int aqr113c_fill_interface_modes(struct phy_device *phydev)
+{
+	int val, ret;
+
+	/* It's been observed on some models that - when coming out of suspend
+	 * - the FW signals that the PHY is ready but the GLOBAL_CFG registers
+	 * continue on returning zeroes for some time. Let's poll the 10M
+	 * register until it returns a real value as both 113c and 115c support
+	 * this mode.
+	 */
+	ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
+					VEND1_GLOBAL_CFG_100M, val, val != 0,
+					1000, 100000, false);
+	if (ret)
+		return ret;
+
+	return aqr107_fill_interface_modes(phydev);
+}
+
 static int aqr113c_config_init(struct phy_device *phydev)
 {
 	int ret;
@@ -725,7 +738,7 @@ static int aqr113c_config_init(struct phy_device *phydev)
 	if (ret)
 		return ret;
 
-	return aqr107_fill_interface_modes(phydev);
+	return aqr113c_fill_interface_modes(phydev);
 }
 
 static int aqr107_probe(struct phy_device *phydev)
-- 
2.43.0

Re: [RFT PATCH net] net: phy: aquantia: only poll GLOBAL_CFG registers on aqr113c and aqr115c

[Jon Hunter]

On 18/07/2024 15:57, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Commit 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to
> start returning real values") introduced a workaround for an issue
> observed on aqr115c. However there were never any reports of it
> happening on other models and the workaround has been reported to cause
> and issue on aqr113c (and it may cause the same on any other model not
> supporting 10M mode).
> 
> Let's limit the impact of the workaround to aqr113c and aqr115c and poll
> the 100M GLOBAL_CFG register instead as both models are known to support
> it correctly.
> 
> Reported-by: Jon Hunter <jonathanh@nvidia.com>
> Closes: https://lore.kernel.org/lkml/7c0140be-4325-4005-9068-7e0fc5ff344d@nvidia.com/
> Fixes: 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to start returning real values")
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> ---
>   drivers/net/phy/aquantia/aquantia_main.c | 29 +++++++++++++++++-------
>   1 file changed, 21 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/net/phy/aquantia/aquantia_main.c b/drivers/net/phy/aquantia/aquantia_main.c
> index d12e35374231..6e3e0fc6ea27 100644
> --- a/drivers/net/phy/aquantia/aquantia_main.c
> +++ b/drivers/net/phy/aquantia/aquantia_main.c
> @@ -653,13 +653,7 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
>   	unsigned long *possible = phydev->possible_interfaces;
>   	unsigned int serdes_mode, rate_adapt;
>   	phy_interface_t interface;
> -	int i, val, ret;
> -
> -	ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
> -					VEND1_GLOBAL_CFG_10M, val, val != 0,
> -					1000, 100000, false);
> -	if (ret)
> -		return ret;
> +	int i, val;
>   
>   	/* Walk the media-speed configuration registers to determine which
>   	 * host-side serdes modes may be used by the PHY depending on the
> @@ -708,6 +702,25 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
>   	return 0;
>   }
>   
> +static int aqr113c_fill_interface_modes(struct phy_device *phydev)
> +{
> +	int val, ret;
> +
> +	/* It's been observed on some models that - when coming out of suspend
> +	 * - the FW signals that the PHY is ready but the GLOBAL_CFG registers
> +	 * continue on returning zeroes for some time. Let's poll the 10M
> +	 * register until it returns a real value as both 113c and 115c support
> +	 * this mode.
> +	 */
> +	ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
> +					VEND1_GLOBAL_CFG_100M, val, val != 0,
> +					1000, 100000, false);
> +	if (ret)
> +		return ret;
> +
> +	return aqr107_fill_interface_modes(phydev);
> +}
> +
>   static int aqr113c_config_init(struct phy_device *phydev)
>   {
>   	int ret;
> @@ -725,7 +738,7 @@ static int aqr113c_config_init(struct phy_device *phydev)
>   	if (ret)
>   		return ret;
>   
> -	return aqr107_fill_interface_modes(phydev);
> +	return aqr113c_fill_interface_modes(phydev);
>   }
>   
>   static int aqr107_probe(struct phy_device *phydev)


This works for Tegra ...

Tested-by: Jon Hunter <jonathanh@nvidia.com>

Thanks!
Jon

-- 
nvpublic

Re: [RFT PATCH net] net: phy: aquantia: only poll GLOBAL_CFG registers on aqr113c and aqr115c

[Antoine Tenart]

Quoting Bartosz Golaszewski (2024-07-18 16:57:47)
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Commit 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to
> start returning real values") introduced a workaround for an issue
> observed on aqr115c. However there were never any reports of it
> happening on other models and the workaround has been reported to cause
> and issue on aqr113c (and it may cause the same on any other model not
> supporting 10M mode).
> 
> Let's limit the impact of the workaround to aqr113c and aqr115c and poll
> the 100M GLOBAL_CFG register instead as both models are known to support
> it correctly.
> 
> Reported-by: Jon Hunter <jonathanh@nvidia.com>
> Closes: https://lore.kernel.org/lkml/7c0140be-4325-4005-9068-7e0fc5ff344d@nvidia.com/
> Fixes: 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to start returning real values")
> Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> ---
>  drivers/net/phy/aquantia/aquantia_main.c | 29 +++++++++++++++++-------
>  1 file changed, 21 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/net/phy/aquantia/aquantia_main.c b/drivers/net/phy/aquantia/aquantia_main.c
> index d12e35374231..6e3e0fc6ea27 100644
> --- a/drivers/net/phy/aquantia/aquantia_main.c
> +++ b/drivers/net/phy/aquantia/aquantia_main.c
> @@ -653,13 +653,7 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
>         unsigned long *possible = phydev->possible_interfaces;
>         unsigned int serdes_mode, rate_adapt;
>         phy_interface_t interface;
> -       int i, val, ret;
> -
> -       ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
> -                                       VEND1_GLOBAL_CFG_10M, val, val != 0,
> -                                       1000, 100000, false);
> -       if (ret)
> -               return ret;
> +       int i, val;
>  
>         /* Walk the media-speed configuration registers to determine which
>          * host-side serdes modes may be used by the PHY depending on the
> @@ -708,6 +702,25 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
>         return 0;
>  }
>  
> +static int aqr113c_fill_interface_modes(struct phy_device *phydev)
> +{
> +       int val, ret;
> +
> +       /* It's been observed on some models that - when coming out of suspend
> +        * - the FW signals that the PHY is ready but the GLOBAL_CFG registers
> +        * continue on returning zeroes for some time. Let's poll the 10M

nit: 100M?

> +        * register until it returns a real value as both 113c and 115c support
> +        * this mode.
> +        */
> +       ret = phy_read_mmd_poll_timeout(phydev, MDIO_MMD_VEND1,
> +                                       VEND1_GLOBAL_CFG_100M, val, val != 0,
> +                                       1000, 100000, false);
> +       if (ret)
> +               return ret;
> +
> +       return aqr107_fill_interface_modes(phydev);
> +}
> +
>  static int aqr113c_config_init(struct phy_device *phydev)
>  {
>         int ret;
> @@ -725,7 +738,7 @@ static int aqr113c_config_init(struct phy_device *phydev)
>         if (ret)
>                 return ret;
>  
> -       return aqr107_fill_interface_modes(phydev);
> +       return aqr113c_fill_interface_modes(phydev);
>  }
>  
>  static int aqr107_probe(struct phy_device *phydev)
> -- 
> 2.43.0
> 
>

Re: [RFT PATCH net] net: phy: aquantia: only poll GLOBAL_CFG registers on aqr113c and aqr115c

[Paolo Abeni]

On 7/18/24 16:57, Bartosz Golaszewski wrote:
> From: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
> 
> Commit 708405f3e56e ("net: phy: aquantia: wait for the GLOBAL_CFG to
> start returning real values") introduced a workaround for an issue
> observed on aqr115c. However there were never any reports of it
> happening on other models and the workaround has been reported to cause
> and issue on aqr113c (and it may cause the same on any other model not
> supporting 10M mode).
> 
> Let's limit the impact of the workaround to aqr113c and aqr115c and poll

AFAICS this patch also affect aqr113, I guess the commit message should 
be updated accordingly.

[...]
> @@ -708,6 +702,25 @@ static int aqr107_fill_interface_modes(struct phy_device *phydev)
>   	return 0;
>   }
>   
> +static int aqr113c_fill_interface_modes(struct phy_device *phydev)
> +{
> +	int val, ret;
> +
> +	/* It's been observed on some models that - when coming out of suspend
> +	 * - the FW signals that the PHY is ready but the GLOBAL_CFG registers
> +	 * continue on returning zeroes for some time. Let's poll the 10M

Please address the typo above as noted by Antoine.

Thanks!

Paolo