[PATCH net-next 00/14] net: stmmac: SerDes, PCS, BASE-X, and inband goodies

[PATCH net-next 00/14] net: stmmac: SerDes, PCS, BASE-X, and inband goodies

[Russell King (Oracle)]

This is the v1 submission: if it doesn't get tested but review goes
well, it'll end up in net-next and mainline without testing on the
affected hardware!

Mentioned previously, I've been trying to sort out the PCS support in
stmmac, and this series represents the current state of play.

Previous posted patches centred around merely getting autonegotiation
to be configured correctly, to a point where the manual configuration
can be removed from the qcom-ethqos driver. The qcom-ethqos driver
uses both SGMII and 2500BASE-X, manually configuring the dwmac's
integrated PCS appropriately.

This *untested* series attempts to take this further. The patches:

- clean up qcom-ethqos only-written mac_base member.
- convert qcom-ethqos to use the set_clk_tx_rate() method for setting
  the link clock rate.
- add support for phy_set_mode_ext() to the qcom "SGMII" ethernet
  SerDes driver (which is really only what it needs. Note that
  phy_set_mode_ext() is an expected call to be made, where as
  phy_set_speed() is optional and not. See PHY documentation.)
- add platform-glue independent SerDes support to the stmmac core
  driver. Currently, only qcom-ethqos will make use of this, and
  I suspect as we haven't had this, it's going to be difficult to
  convert other platform glue to use this - but had this existed
  earlier, we could've pushed people to use PHY to abstract some
  of the platform glue differences. Adding it now makes it available
  for future platform glue.
- convert qcom-ethqos to use this core SerDes support.
- arrange for stmmac_pcs.c to supply the phy_intf_sel field value
  if the integrated PCS will be used. (PHY_INTF_SEL_SGMII requires
  the integrated PCS rather than an external PCS.)
- add BASE-X support to the integrated PCS driver, and use it for
  BASE-X modes. This fully supports in-band mode, including reading
  the link partner advertisement.
- add in-band support for SGMII, reading the state from the RGSMII
  status field.

As we leave qcom-ethqos' manual configuration of the PCS in place at
the moment, the last patch adds reporting of any changes in its
configuration that the qcom-ethqos driver does beyond what phylink
requested, thus providing a path to debug and eventually remove
qcom-ethqos' manual configuration.

One patch is not included in this set - which adds a phy_intf_sel
value for external PCS (using PHY_INTF_SEL_GMII_MII). I believe all
external PCS use this mode when connected to a MAC capable of up to
2.5G. However, no platform glue that provides the mac_select_pcs()
method also provide the set_phy_intf_sel() method, so we can safely
ignore this for now.

I would like to get this into net-next before the next merge window,
so testing would be appreciated. If there are issues with these patches
applied, please check whether the issue exists without these patches
and only report regressions caused by this patch set. For example,
I'm aware that qcom-ethqos has issues with 10Mbps mode due to an AQR
PHY being insanely provisioned to use SGMII in 1000M mode but with
rate matching with 10M media. This is not an issue that is relevant
to this patch series, but a problem with the PHY provisioning.

rfc->v1:
 - fix SGMII link status
 - avoid calling phy_get_mode() if PHY is null

 drivers/net/ethernet/stmicro/stmmac/Makefile       |   2 +-
 drivers/net/ethernet/stmicro/stmmac/common.h       |   1 -
 .../ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c    |  74 ++-----
 .../net/ethernet/stmicro/stmmac/dwmac1000_core.c   |   9 +-
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c  |   8 +-
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c  |  69 +++++--
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c   | 220 +++++++++++++++++++--
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h   |  53 ++---
 .../net/ethernet/stmicro/stmmac/stmmac_serdes.c    | 111 +++++++++++
 .../net/ethernet/stmicro/stmmac/stmmac_serdes.h    |  16 ++
 drivers/phy/qualcomm/phy-qcom-sgmii-eth.c          |  43 ++++
 include/linux/stmmac.h                             |   2 +
 12 files changed, 483 insertions(+), 125 deletions(-)
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

[PATCH net-next 01/14] net: stmmac: qcom-ethqos: remove mac_base

[Russell King (Oracle)]

In commit 9b443e58a896 ("net: stmmac: qcom-ethqos: remove MAC_CTRL_REG
modification"), ethqos->mac_base is only written, never read. Let's
remove it.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
index 0826a7bd32ff..869f924f3cde 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
@@ -100,7 +100,6 @@ struct ethqos_emac_driver_data {
 struct qcom_ethqos {
 	struct platform_device *pdev;
 	void __iomem *rgmii_base;
-	void __iomem *mac_base;
 	int (*configure_func)(struct qcom_ethqos *ethqos, int speed);
 
 	unsigned int link_clk_rate;
@@ -772,8 +771,6 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
 		return dev_err_probe(dev, PTR_ERR(ethqos->rgmii_base),
 				     "Failed to map rgmii resource\n");
 
-	ethqos->mac_base = stmmac_res.addr;
-
 	data = of_device_get_match_data(dev);
 	ethqos->por = data->por;
 	ethqos->num_por = data->num_por;
-- 
2.47.3

Re: [PATCH net-next 01/14] net: stmmac: qcom-ethqos: remove mac_base

[Bartosz Golaszewski]

On Mon, 19 Jan 2026 13:33:41 +0100, "Russell King (Oracle)"
<rmk+kernel@armlinux.org.uk> said:
> In commit 9b443e58a896 ("net: stmmac: qcom-ethqos: remove MAC_CTRL_REG
> modification"), ethqos->mac_base is only written, never read. Let's
> remove it.
>
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> ---
>  drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c | 3 ---
>  1 file changed, 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> index 0826a7bd32ff..869f924f3cde 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> @@ -100,7 +100,6 @@ struct ethqos_emac_driver_data {
>  struct qcom_ethqos {
>  	struct platform_device *pdev;
>  	void __iomem *rgmii_base;
> -	void __iomem *mac_base;
>  	int (*configure_func)(struct qcom_ethqos *ethqos, int speed);
>
>  	unsigned int link_clk_rate;
> @@ -772,8 +771,6 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
>  		return dev_err_probe(dev, PTR_ERR(ethqos->rgmii_base),
>  				     "Failed to map rgmii resource\n");
>
> -	ethqos->mac_base = stmmac_res.addr;
> -
>  	data = of_device_get_match_data(dev);
>  	ethqos->por = data->por;
>  	ethqos->num_por = data->num_por;
> --
> 2.47.3
>
>
>

Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

[PATCH net-next 02/14] net: stmmac: qcom-ethqos: convert to set_clk_tx_rate() method

[Russell King (Oracle)]

Set the RGMII link clock using the set_clk_tx_rate() method rather than
coding it into the .fix_mac_speed() method. This simplifies ethqos's
ethqos_fix_mac_speed().

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../stmicro/stmmac/dwmac-qcom-ethqos.c        | 21 ++++++++++---------
 1 file changed, 11 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
index 869f924f3cde..80ea69fc8ee5 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
@@ -102,7 +102,6 @@ struct qcom_ethqos {
 	void __iomem *rgmii_base;
 	int (*configure_func)(struct qcom_ethqos *ethqos, int speed);
 
-	unsigned int link_clk_rate;
 	struct clk *link_clk;
 	struct phy *serdes_phy;
 	int serdes_speed;
@@ -174,19 +173,20 @@ static void rgmii_dump(void *priv)
 		rgmii_readl(ethqos, EMAC_SYSTEM_LOW_POWER_DEBUG));
 }
 
-static void
-ethqos_update_link_clk(struct qcom_ethqos *ethqos, int speed)
+static int ethqos_set_clk_tx_rate(void *bsp_priv, struct clk *clk_tx_i,
+				  phy_interface_t interface, int speed)
 {
+	struct qcom_ethqos *ethqos = bsp_priv;
 	long rate;
 
-	if (!phy_interface_mode_is_rgmii(ethqos->phy_mode))
-		return;
+	if (!phy_interface_mode_is_rgmii(interface))
+		return 0;
 
 	rate = rgmii_clock(speed);
-	if (rate > 0)
-		ethqos->link_clk_rate = rate * 2;
+	if (rate < 0)
+		return rate;
 
-	clk_set_rate(ethqos->link_clk, ethqos->link_clk_rate);
+	return clk_set_rate(ethqos->link_clk, rate * 2);
 }
 
 static void
@@ -645,7 +645,6 @@ static void ethqos_fix_mac_speed(void *priv, int speed, unsigned int mode)
 	struct qcom_ethqos *ethqos = priv;
 
 	qcom_ethqos_set_sgmii_loopback(ethqos, false);
-	ethqos_update_link_clk(ethqos, speed);
 	ethqos_configure(ethqos, speed);
 }
 
@@ -797,10 +796,12 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
 				     "Failed to get serdes phy\n");
 
 	ethqos->serdes_speed = SPEED_1000;
-	ethqos_update_link_clk(ethqos, SPEED_1000);
+	ethqos_set_clk_tx_rate(ethqos, NULL, plat_dat->phy_interface,
+			       SPEED_1000);
 	ethqos_set_func_clk_en(ethqos);
 
 	plat_dat->bsp_priv = ethqos;
+	plat_dat->set_clk_tx_rate = ethqos_set_clk_tx_rate;
 	plat_dat->fix_mac_speed = ethqos_fix_mac_speed;
 	plat_dat->dump_debug_regs = rgmii_dump;
 	plat_dat->ptp_clk_freq_config = ethqos_ptp_clk_freq_config;
-- 
2.47.3

Re: [PATCH net-next 02/14] net: stmmac: qcom-ethqos: convert to set_clk_tx_rate() method

[Bartosz Golaszewski]

On Mon, 19 Jan 2026 13:33:46 +0100, "Russell King (Oracle)"
<rmk+kernel@armlinux.org.uk> said:
> Set the RGMII link clock using the set_clk_tx_rate() method rather than
> coding it into the .fix_mac_speed() method. This simplifies ethqos's
> ethqos_fix_mac_speed().
>
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> ---

Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

[PATCH net-next 03/14] phy: qcom-sgmii-eth: add .set_mode() and .validate() methods

[Russell King (Oracle)]

qcom-sgmii-eth is an Ethernet SerDes supporting only Ethernet mode
using SGMII, 1000BASE-X and 2500BASE-X.

Add an implementation of the .set_mode() method, which can be used
instead of or as well as the .set_speed() method. The Ethernet
interface modes mentioned above all have a fixed data rate, so
setting the mode is sufficient to fully specify the operating
parameters.

Add an implementation of the .validate() method, which will be
necessary to allow discovery of the SerDes capabilities for platform
independent SerDes support in the stmmac network driver.

Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/phy/qualcomm/phy-qcom-sgmii-eth.c | 43 +++++++++++++++++++++++
 1 file changed, 43 insertions(+)

diff --git a/drivers/phy/qualcomm/phy-qcom-sgmii-eth.c b/drivers/phy/qualcomm/phy-qcom-sgmii-eth.c
index 5b1c82459c12..4ea3dce7719f 100644
--- a/drivers/phy/qualcomm/phy-qcom-sgmii-eth.c
+++ b/drivers/phy/qualcomm/phy-qcom-sgmii-eth.c
@@ -7,6 +7,7 @@
 #include <linux/ethtool.h>
 #include <linux/module.h>
 #include <linux/of.h>
+#include <linux/phy.h>
 #include <linux/phy/phy.h>
 #include <linux/platform_device.h>
 #include <linux/regmap.h>
@@ -286,6 +287,37 @@ static int qcom_dwmac_sgmii_phy_power_off(struct phy *phy)
 	return 0;
 }
 
+static int qcom_dwmac_sgmii_phy_speed(enum phy_mode mode, int submode)
+{
+	if (mode != PHY_MODE_ETHERNET)
+		return -EINVAL;
+
+	if (submode == PHY_INTERFACE_MODE_SGMII ||
+	    submode == PHY_INTERFACE_MODE_1000BASEX)
+		return SPEED_1000;
+
+	if (submode == PHY_INTERFACE_MODE_2500BASEX)
+		return SPEED_2500;
+
+	return -EINVAL;
+}
+
+static int qcom_dwmac_sgmii_phy_set_mode(struct phy *phy, enum phy_mode mode,
+					 int submode)
+{
+	struct qcom_dwmac_sgmii_phy_data *data = phy_get_drvdata(phy);
+	int speed;
+
+	speed = qcom_dwmac_sgmii_phy_speed(mode, submode);
+	if (speed < 0)
+		return speed;
+
+	if (speed != data->speed)
+		data->speed = speed;
+
+	return qcom_dwmac_sgmii_phy_calibrate(phy);
+}
+
 static int qcom_dwmac_sgmii_phy_set_speed(struct phy *phy, int speed)
 {
 	struct qcom_dwmac_sgmii_phy_data *data = phy_get_drvdata(phy);
@@ -296,10 +328,21 @@ static int qcom_dwmac_sgmii_phy_set_speed(struct phy *phy, int speed)
 	return qcom_dwmac_sgmii_phy_calibrate(phy);
 }
 
+static int qcom_dwmac_sgmii_phy_validate(struct phy *phy, enum phy_mode mode,
+					 int submode,
+					 union phy_configure_opts *opts)
+{
+	int ret = qcom_dwmac_sgmii_phy_speed(mode, submode);
+
+	return ret < 0 ? ret : 0;
+}
+
 static const struct phy_ops qcom_dwmac_sgmii_phy_ops = {
 	.power_on	= qcom_dwmac_sgmii_phy_power_on,
 	.power_off	= qcom_dwmac_sgmii_phy_power_off,
+	.set_mode	= qcom_dwmac_sgmii_phy_set_mode,
 	.set_speed	= qcom_dwmac_sgmii_phy_set_speed,
+	.validate	= qcom_dwmac_sgmii_phy_validate,
 	.calibrate	= qcom_dwmac_sgmii_phy_calibrate,
 	.owner		= THIS_MODULE,
 };
-- 
2.47.3

Re: [PATCH net-next 03/14] phy: qcom-sgmii-eth: add .set_mode() and .validate() methods

[Bartosz Golaszewski]

On Mon, 19 Jan 2026 13:33:51 +0100, "Russell King (Oracle)"
<rmk+kernel@armlinux.org.uk> said:
> qcom-sgmii-eth is an Ethernet SerDes supporting only Ethernet mode
> using SGMII, 1000BASE-X and 2500BASE-X.
>
> Add an implementation of the .set_mode() method, which can be used
> instead of or as well as the .set_speed() method. The Ethernet
> interface modes mentioned above all have a fixed data rate, so
> setting the mode is sufficient to fully specify the operating
> parameters.
>
> Add an implementation of the .validate() method, which will be
> necessary to allow discovery of the SerDes capabilities for platform
> independent SerDes support in the stmmac network driver.
>
> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> ---

Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

Re: [PATCH net-next 03/14] phy: qcom-sgmii-eth: add .set_mode() and .validate() methods

[Vinod Koul]

On 19-01-26, 12:33, Russell King (Oracle) wrote:
> qcom-sgmii-eth is an Ethernet SerDes supporting only Ethernet mode
> using SGMII, 1000BASE-X and 2500BASE-X.
> 
> Add an implementation of the .set_mode() method, which can be used
> instead of or as well as the .set_speed() method. The Ethernet
> interface modes mentioned above all have a fixed data rate, so
> setting the mode is sufficient to fully specify the operating
> parameters.
> 
> Add an implementation of the .validate() method, which will be
> necessary to allow discovery of the SerDes capabilities for platform
> independent SerDes support in the stmmac network driver.

Acked-by: Vinod Koul <vkoul@kernel.org>

-- 
~Vinod

[PATCH net-next 04/14] net: stmmac: wrap phylink's rx_clk_stop functions

[Russell King (Oracle)]

With generic SerDes support, stmmac will need to do more work to ensure
that clk_rx_i is running in all configurations. Rather than turn each
site that calls phylink_rx_clk_stop_xxx() into a list of functions,
move these to their own pair of functions so that they can be
augmented at a single location.

Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../net/ethernet/stmicro/stmmac/stmmac_main.c | 32 ++++++++++++-------
 1 file changed, 21 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index c2589f02ff7e..24a2555ca329 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -3547,6 +3547,16 @@ static void stmmac_safety_feat_configuration(struct stmmac_priv *priv)
 	}
 }
 
+static void stmmac_clk_rx_i_require(struct stmmac_priv *priv)
+{
+	phylink_rx_clk_stop_block(priv->phylink);
+}
+
+static void stmmac_clk_rx_i_release(struct stmmac_priv *priv)
+{
+	phylink_rx_clk_stop_unblock(priv->phylink);
+}
+
 /**
  * stmmac_hw_setup - setup mac in a usable state.
  *  @dev : pointer to the device structure.
@@ -3578,12 +3588,12 @@ static int stmmac_hw_setup(struct net_device *dev)
 	 * Block the receive clock stop for LPI mode at the PHY in case
 	 * the link is established with EEE mode active.
 	 */
-	phylink_rx_clk_stop_block(priv->phylink);
+	stmmac_clk_rx_i_require(priv);
 
 	/* DMA initialization and SW reset */
 	ret = stmmac_init_dma_engine(priv);
 	if (ret < 0) {
-		phylink_rx_clk_stop_unblock(priv->phylink);
+		stmmac_clk_rx_i_release(priv);
 		netdev_err(priv->dev, "%s: DMA engine initialization failed\n",
 			   __func__);
 		return ret;
@@ -3591,7 +3601,7 @@ static int stmmac_hw_setup(struct net_device *dev)
 
 	/* Copy the MAC addr into the HW  */
 	stmmac_set_umac_addr(priv, priv->hw, dev->dev_addr, 0);
-	phylink_rx_clk_stop_unblock(priv->phylink);
+	stmmac_clk_rx_i_release(priv);
 
 	/* Initialize the MAC Core */
 	stmmac_core_init(priv, priv->hw, dev);
@@ -3670,9 +3680,9 @@ static int stmmac_hw_setup(struct net_device *dev)
 	/* Start the ball rolling... */
 	stmmac_start_all_dma(priv);
 
-	phylink_rx_clk_stop_block(priv->phylink);
+	stmmac_clk_rx_i_require(priv);
 	stmmac_set_hw_vlan_mode(priv, priv->hw);
-	phylink_rx_clk_stop_unblock(priv->phylink);
+	stmmac_clk_rx_i_release(priv);
 
 	return 0;
 }
@@ -6107,9 +6117,9 @@ static int stmmac_set_features(struct net_device *netdev,
 	else
 		priv->hw->hw_vlan_en = false;
 
-	phylink_rx_clk_stop_block(priv->phylink);
+	stmmac_clk_rx_i_require(priv);
 	stmmac_set_hw_vlan_mode(priv, priv->hw);
-	phylink_rx_clk_stop_unblock(priv->phylink);
+	stmmac_clk_rx_i_release(priv);
 
 	return 0;
 }
@@ -6378,9 +6388,9 @@ static int stmmac_set_mac_address(struct net_device *ndev, void *addr)
 	if (ret)
 		goto set_mac_error;
 
-	phylink_rx_clk_stop_block(priv->phylink);
+	stmmac_clk_rx_i_require(priv);
 	stmmac_set_umac_addr(priv, priv->hw, ndev->dev_addr, 0);
-	phylink_rx_clk_stop_unblock(priv->phylink);
+	stmmac_clk_rx_i_release(priv);
 
 set_mac_error:
 	pm_runtime_put(priv->device);
@@ -8192,11 +8202,11 @@ int stmmac_resume(struct device *dev)
 	stmmac_init_timestamping(priv);
 
 	stmmac_init_coalesce(priv);
-	phylink_rx_clk_stop_block(priv->phylink);
+	stmmac_clk_rx_i_require(priv);
 	stmmac_set_rx_mode(ndev);
 
 	stmmac_restore_hw_vlan_rx_fltr(priv, ndev, priv->hw);
-	phylink_rx_clk_stop_unblock(priv->phylink);
+	stmmac_clk_rx_i_release(priv);
 
 	stmmac_enable_all_queues(priv);
 	stmmac_enable_all_dma_irq(priv);
-- 
2.47.3

[PATCH net-next 05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

Rather than having platform glue implement SerDes PHY support, add it
to the core driver, specifically to the stmmac integrated PCS driver
as the SerDes is connected to the integrated PCS.

Platforms using external PCS can also populate plat->serdes, and the
core driver will call phy_init() and phy_exit() when the administrative
state of the interface changes, but the other phy methods will not be
called.

Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
--
rfc->v1: avoid calling phy_get_mode() with NULL serdes PHY
---
 drivers/net/ethernet/stmicro/stmmac/Makefile  |   2 +-
 .../net/ethernet/stmicro/stmmac/stmmac_main.c |  14 ++-
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  |  38 +++++-
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.h  |   1 +
 .../ethernet/stmicro/stmmac/stmmac_serdes.c   | 111 ++++++++++++++++++
 .../ethernet/stmicro/stmmac/stmmac_serdes.h   |  16 +++
 include/linux/stmmac.h                        |   2 +
 7 files changed, 180 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h

diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index c9263987ef8d..a3c2cd5d0c91 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -7,7 +7,7 @@ stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o	\
 	      dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \
 	      stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \
 	      stmmac_xdp.o stmmac_est.o stmmac_fpe.o stmmac_vlan.o \
-	      stmmac_pcs.o $(stmmac-y)
+	      stmmac_pcs.o stmmac_serdes.o $(stmmac-y)
 
 stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 24a2555ca329..6c515f9efbe7 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -48,6 +48,7 @@
 #include "stmmac_fpe.h"
 #include "stmmac.h"
 #include "stmmac_pcs.h"
+#include "stmmac_serdes.h"
 #include "stmmac_xdp.h"
 #include <linux/reset.h>
 #include <linux/of_mdio.h>
@@ -3549,12 +3550,16 @@ static void stmmac_safety_feat_configuration(struct stmmac_priv *priv)
 
 static void stmmac_clk_rx_i_require(struct stmmac_priv *priv)
 {
+	dwmac_serdes_power_on(priv);
+	/* Only sets the SerDes mode if it wasn't already configured. */
+	dwmac_serdes_init_mode(priv, priv->plat->phy_interface);
 	phylink_rx_clk_stop_block(priv->phylink);
 }
 
 static void stmmac_clk_rx_i_release(struct stmmac_priv *priv)
 {
 	phylink_rx_clk_stop_unblock(priv->phylink);
+	dwmac_serdes_power_off(priv);
 }
 
 /**
@@ -4152,10 +4157,14 @@ static int stmmac_open(struct net_device *dev)
 	if (ret)
 		goto err_runtime_pm;
 
-	ret = __stmmac_open(dev, dma_conf);
+	ret = dwmac_serdes_init(priv);
 	if (ret)
 		goto err_disconnect_phy;
 
+	ret = __stmmac_open(dev, dma_conf);
+	if (ret)
+		goto err_serdes;
+
 	kfree(dma_conf);
 
 	/* We may have called phylink_speed_down before */
@@ -4163,6 +4172,8 @@ static int stmmac_open(struct net_device *dev)
 
 	return ret;
 
+err_serdes:
+	dwmac_serdes_exit(priv);
 err_disconnect_phy:
 	phylink_disconnect_phy(priv->phylink);
 err_runtime_pm:
@@ -4226,6 +4237,7 @@ static int stmmac_release(struct net_device *dev)
 
 	__stmmac_release(dev);
 
+	dwmac_serdes_exit(priv);
 	phylink_disconnect_phy(priv->phylink);
 	pm_runtime_put(priv->device);
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 2f826fe7229b..4d1902f3a58f 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -1,12 +1,25 @@
 // SPDX-License-Identifier: GPL-2.0-only
 #include "stmmac.h"
 #include "stmmac_pcs.h"
+#include "stmmac_serdes.h"
 
 static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	struct stmmac_priv *priv = spcs->priv;
+	int ret;
 
-	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
+	ret = dwmac_serdes_power_on(priv);
+	if (ret)
+		return ret;
+
+	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
+		ret = dwmac_serdes_set_mode(priv, spcs->interface);
+		if (ret)
+			return ret;
+	}
+
+	stmmac_mac_irq_modify(priv, 0, spcs->int_mask);
 
 	return 0;
 }
@@ -14,8 +27,11 @@ static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 static void dwmac_integrated_pcs_disable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	struct stmmac_priv *priv = spcs->priv;
 
-	stmmac_mac_irq_modify(spcs->priv, spcs->int_mask, 0);
+	stmmac_mac_irq_modify(priv, spcs->int_mask, 0);
+
+	dwmac_serdes_power_off(priv);
 }
 
 static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
@@ -32,6 +48,15 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 				       bool permit_pause_to_mac)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	int ret;
+
+	if (spcs->interface != interface) {
+		ret = dwmac_serdes_set_mode(spcs->priv, interface);
+		if (ret)
+			return ret;
+
+		spcs->interface = interface;
+	}
 
 	dwmac_ctrl_ane(spcs->base, 0, 1, spcs->priv->hw->reverse_sgmii_enable);
 
@@ -71,6 +96,7 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 			       u32 int_mask)
 {
 	struct stmmac_pcs *spcs;
+	int ret;
 
 	spcs = devm_kzalloc(priv->device, sizeof(*spcs), GFP_KERNEL);
 	if (!spcs)
@@ -81,6 +107,14 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 	spcs->int_mask = int_mask;
 	spcs->pcs.ops = &dwmac_integrated_pcs_ops;
 
+	if (priv->plat->serdes) {
+		ret = dwmac_serdes_validate(priv, PHY_INTERFACE_MODE_SGMII);
+		if (ret)
+			dev_warn(priv->device,
+				 "serdes does not support SGMII: %pe\n",
+				 ERR_PTR(ret));
+	}
+
 	__set_bit(PHY_INTERFACE_MODE_SGMII, spcs->pcs.supported_interfaces);
 
 	priv->integrated_pcs = spcs;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index c4e6b242d390..36bf75fdf478 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -53,6 +53,7 @@ struct stmmac_pcs {
 	struct stmmac_priv *priv;
 	void __iomem *base;
 	u32 int_mask;
+	phy_interface_t interface;
 	struct phylink_pcs pcs;
 };
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
new file mode 100644
index 000000000000..d46a071bc383
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
@@ -0,0 +1,111 @@
+#include <linux/phy/phy.h>
+
+#include "stmmac.h"
+#include "stmmac_serdes.h"
+
+static phy_interface_t dwmac_serdes_phy_modes[] = {
+	PHY_INTERFACE_MODE_SGMII,
+	PHY_INTERFACE_MODE_1000BASEX,
+	PHY_INTERFACE_MODE_2500BASEX
+};
+
+int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	return phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET, interface,
+			    NULL);
+}
+
+int dwmac_serdes_init(struct stmmac_priv *priv)
+{
+	size_t i;
+	int ret;
+
+	if (!priv->plat->serdes)
+		return 0;
+
+	/* Encourage good implementation of the SerDes PHY driver, so that
+	 * we can discover which Ethernet modes the SerDes supports.
+	 * Unfortunately, some implementations are noisy (bad), others
+	 * require phy_set_speed() to select the correct speed first
+	 * (which then reprograms the SerDes, negating the whole point of
+	 * phy_validate().) Weed out these incompatible implementations.
+	 */
+	for (i = 0; i < ARRAY_SIZE(dwmac_serdes_phy_modes); i++) {
+		ret = phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET,
+				   dwmac_serdes_phy_modes[i], NULL);
+		if (ret == 0 || ret == -EOPNOTSUPP)
+			break;
+	}
+
+	if (ret == -EOPNOTSUPP)
+		dev_warn(priv->device,
+			 "SerDes driver does not implement phy_validate()\n");
+	if (ret) {
+		/* The SerDes PHY failed validation, refuse to use it. */
+		dev_warn(priv->device,
+			 "SerDes driver fails to validate SGMII, 1000BASE-X nor 2500BASE-X\n");
+		return -EINVAL;
+	}
+
+	ret = phy_init(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to initialize SerDes: %pe\n",
+			ERR_PTR(ret));
+
+	return ret;
+}
+
+int dwmac_serdes_power_on(struct stmmac_priv *priv)
+{
+	int ret;
+
+	ret = phy_power_on(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to power on SerDes: %pe\n",
+			ERR_PTR(ret));
+
+	return ret;
+}
+
+int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	struct phy *serdes = priv->plat->serdes;
+
+	if (!serdes || phy_get_mode(serdes) == PHY_MODE_ETHERNET)
+		return 0;
+
+	return dwmac_serdes_set_mode(priv, interface);
+}
+
+int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	struct phy *serdes = priv->plat->serdes;
+	int ret;
+
+	ret = phy_set_mode_ext(serdes, PHY_MODE_ETHERNET, interface);
+	if (ret)
+		dev_err(priv->device,
+			"failed to set SerDes mode %s: %pe\n",
+			phy_modes(interface), ERR_PTR(ret));
+
+	return ret;
+}
+
+void dwmac_serdes_power_off(struct stmmac_priv *priv)
+{
+	int ret;
+
+	ret = phy_power_off(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to power off SerDes: %pe\n",
+			ERR_PTR(ret));
+}
+
+void dwmac_serdes_exit(struct stmmac_priv *priv)
+{
+	int ret = phy_exit(priv->plat->serdes);
+
+	if (ret)
+		dev_err(priv->device, "failed to shutdown SerDes: %pe\n",
+			ERR_PTR(ret));
+}
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
new file mode 100644
index 000000000000..a31e6c9e0570
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
@@ -0,0 +1,16 @@
+#ifndef STMMAC_SERDES_H
+#define STMMAC_SERDES_H
+
+#include <linux/phy.h>
+
+struct stmmac_priv;
+
+int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface);
+int dwmac_serdes_init(struct stmmac_priv *priv);
+int dwmac_serdes_power_on(struct stmmac_priv *priv);
+int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface);
+int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface);
+void dwmac_serdes_power_off(struct stmmac_priv *priv);
+void dwmac_serdes_exit(struct stmmac_priv *priv);
+
+#endif
diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
index f1054b9c2d8a..4db506e5cf13 100644
--- a/include/linux/stmmac.h
+++ b/include/linux/stmmac.h
@@ -193,6 +193,7 @@ enum dwmac_core_type {
 #define STMMAC_FLAG_HWTSTAMP_CORRECT_LATENCY	BIT(13)
 
 struct mac_device_info;
+struct phy;
 
 struct plat_stmmacenet_data {
 	enum dwmac_core_type core_type;
@@ -222,6 +223,7 @@ struct plat_stmmacenet_data {
 	 * that phylink uses.
 	 */
 	phy_interface_t phy_interface;
+	struct phy *serdes;
 	struct stmmac_mdio_bus_data *mdio_bus_data;
 	struct device_node *phy_node;
 	struct fwnode_handle *port_node;
-- 
2.47.3

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Jakub Kicinski]

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net: stmmac: add stmmac core serdes support

This commit adds SerDes PHY support to the stmmac core driver, moving it
from platform glue into the integrated PCS driver. The SerDes is managed
through phy_init/phy_exit for lifecycle and phy_power_on/phy_power_off
for the power state.

> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> index 2f826fe7229b..4d1902f3a58f 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c

[ ... ]

> static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
> {
> 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> +	struct stmmac_priv *priv = spcs->priv;
> +	int ret;
>
> -	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
> +	ret = dwmac_serdes_power_on(priv);
> +	if (ret)
> +		return ret;
> +
> +	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
> +		ret = dwmac_serdes_set_mode(priv, spcs->interface);
> +		if (ret)
> +			return ret;
                        ^^^^^^^^^^

If dwmac_serdes_set_mode() fails here, doesn't this leave the SerDes in
a powered-on state? dwmac_serdes_power_on() succeeded at this point but
there's no corresponding dwmac_serdes_power_off() call before returning
the error.

Looking at phylink_major_config(), it appears to discard the return
value from phylink_pcs_enable(), so the caller won't know to call
pcs_disable() to clean up the power state.

Should this be something like:

    if (spcs->interface != PHY_INTERFACE_MODE_NA) {
        ret = dwmac_serdes_set_mode(priv, spcs->interface);
        if (ret) {
            dwmac_serdes_power_off(priv);
            return ret;
        }
    }

> +	}
> +
> +	stmmac_mac_irq_modify(priv, 0, spcs->int_mask);
>
> 	return 0;
> }

[ ... ]

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

On Mon, Jan 19, 2026 at 11:21:25AM -0800, Jakub Kicinski wrote:
> This is an AI-generated review of your patch. The human sending this
> email has considered the AI review valid, or at least plausible.
> 
> For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
> ---
> net: stmmac: add stmmac core serdes support
> 
> This commit adds SerDes PHY support to the stmmac core driver, moving it
> from platform glue into the integrated PCS driver. The SerDes is managed
> through phy_init/phy_exit for lifecycle and phy_power_on/phy_power_off
> for the power state.
> 
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > index 2f826fe7229b..4d1902f3a58f 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> 
> [ ... ]
> 
> > static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
> > {
> > 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> > +	struct stmmac_priv *priv = spcs->priv;
> > +	int ret;
> >
> > -	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
> > +	ret = dwmac_serdes_power_on(priv);
> > +	if (ret)
> > +		return ret;
> > +
> > +	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
> > +		ret = dwmac_serdes_set_mode(priv, spcs->interface);
> > +		if (ret)
> > +			return ret;
>                         ^^^^^^^^^^
> 
> If dwmac_serdes_set_mode() fails here, doesn't this leave the SerDes in
> a powered-on state? dwmac_serdes_power_on() succeeded at this point but
> there's no corresponding dwmac_serdes_power_off() call before returning
> the error.
> 
> Looking at phylink_major_config(), it appears to discard the return
> value from phylink_pcs_enable(), so the caller won't know to call
> pcs_disable() to clean up the power state.

This AI analysis is incorrect.

By the time phylink_pcs_enable() has been called, the PCS is already
plumbed in to phylink. It _will_ have phylink_pcs_disable() called on
it at some point in the future, either by having the PCS displaced
by another in a subsequent phylink_major_config(), or by a driver
calling phylink_stop().

If we clean up here, then we will call dwmac_serdes_power_off() twice.

Yes, it's not "nice" but that's the way phylink is right now, and
without reworking phylink to record that pcs_enable() has failed
to avoid a subsequent pcs_disable(), and to stop the major config
(which then potentially causes a whole bunch of other issues). I
don't even want to think about that horrid scenario at the moment.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Vladimir Oltean]

On Tue, Jan 20, 2026 at 05:04:53AM +0000, Russell King (Oracle) wrote:
> On Mon, Jan 19, 2026 at 11:21:25AM -0800, Jakub Kicinski wrote:
> > This is an AI-generated review of your patch. The human sending this
> > email has considered the AI review valid, or at least plausible.
> > 
> > For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
> > ---
> > net: stmmac: add stmmac core serdes support
> > 
> > This commit adds SerDes PHY support to the stmmac core driver, moving it
> > from platform glue into the integrated PCS driver. The SerDes is managed
> > through phy_init/phy_exit for lifecycle and phy_power_on/phy_power_off
> > for the power state.
> > 
> > > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > > index 2f826fe7229b..4d1902f3a58f 100644
> > > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > 
> > [ ... ]
> > 
> > > static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
> > > {
> > > 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> > > +	struct stmmac_priv *priv = spcs->priv;
> > > +	int ret;
> > >
> > > -	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
> > > +	ret = dwmac_serdes_power_on(priv);
> > > +	if (ret)
> > > +		return ret;
> > > +
> > > +	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
> > > +		ret = dwmac_serdes_set_mode(priv, spcs->interface);
> > > +		if (ret)
> > > +			return ret;
> >                         ^^^^^^^^^^
> > 
> > If dwmac_serdes_set_mode() fails here, doesn't this leave the SerDes in
> > a powered-on state? dwmac_serdes_power_on() succeeded at this point but
> > there's no corresponding dwmac_serdes_power_off() call before returning
> > the error.
> > 
> > Looking at phylink_major_config(), it appears to discard the return
> > value from phylink_pcs_enable(), so the caller won't know to call
> > pcs_disable() to clean up the power state.
> 
> This AI analysis is incorrect.
> 
> By the time phylink_pcs_enable() has been called, the PCS is already
> plumbed in to phylink. It _will_ have phylink_pcs_disable() called on
> it at some point in the future, either by having the PCS displaced
> by another in a subsequent phylink_major_config(), or by a driver
> calling phylink_stop().
> 
> If we clean up here, then we will call dwmac_serdes_power_off() twice.
> 
> Yes, it's not "nice" but that's the way phylink is right now, and
> without reworking phylink to record that pcs_enable() has failed
> to avoid a subsequent pcs_disable(), and to stop the major config
> (which then potentially causes a whole bunch of other issues). I
> don't even want to think about that horrid scenario at the moment.

Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
after calling pcs_disable(), though?

I had to deal with the same issue when preparing patches that integrate
SerDes support into the Lynx PCS.

I had these patches (please pardon the unadapted commit messages for the
present situation):

-- >8 --
Subject: [PATCH] net: phylink: handle return code from phylink_pcs_enable()

I am trying to make phylink_pcs_ops :: pcs_enable() something that is
handled sufficiently carefully by phylink, such that we can expect that
when we return an error code here, no other phylink_pcs_ops call is
being made. This way, the API can be considered sufficiently reliable to
allocate memory in pcs_enable() which is freed in pcs_disable().

Currently this does not take place. The pcs_enable() method has an int
return code, which is ignored. If the PCS returns an error, the
initialization of the phylink instance is not stopped, but continues on
like a train, most likely triggering faults somewhere else.

Like this:

$ ip link set endpmac2 up
fsl_dpaa2_eth dpni.1 endpmac2: configuring for c73/10gbase-kr link mode
fsl_dpaa2_eth dpni.1 endpmac2: pcs_enable() failed: -ENOMEM // added by me
Unable to handle kernel paging request at virtual address fffffffffffffff4
Call trace:
 mtip_backplane_get_state+0x34/0x2b4
 lynx_pcs_get_state+0x30/0x180
 phylink_resolve+0x2c0/0x764
 process_scheduled_works+0x228/0x330
 worker_thread+0x28c/0x450

Do a minimal handling of the error by clearing pl->pcs, so that we lose
access to its ops, and thus are unable to call anything else (which
would be invalid anyway).

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
 drivers/net/phy/phylink.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c
index 32ffa4f9e5b2..a8459116b701 100644
--- a/drivers/net/phy/phylink.c
+++ b/drivers/net/phy/phylink.c
@@ -1315,8 +1315,15 @@ static void phylink_major_config(struct phylink *pl, bool restart,
 		}
 	}
 
-	if (pl->pcs_state == PCS_STATE_STARTING || pcs_changed)
-		phylink_pcs_enable(pl->pcs);
+	if (pl->pcs_state == PCS_STATE_STARTING || pcs_changed) {
+		err = phylink_pcs_enable(pl->pcs);
+		if (err < 0) {
+			phylink_err(pl, "pcs_enable() failed: %pe\n",
+				    ERR_PTR(err));
+			pl->pcs = NULL;
+			return;
+		}
+	}
 
 	err = phylink_pcs_config(pl->pcs, pl->pcs_neg_mode, state,
 				 !!(pl->link_config.pause & MLO_PAUSE_AN));
-- >8 --

-- >8 --
Subject: [PATCH] net: phylink: suppress pcs->ops->pcs_get_state() calls after
 phylink_stop()

I am attempting to make phylink_pcs_ops :: pcs_disable() treated
sufficiently carefully by phylink so as to be able to free memory
allocations from this PCS callback, and do not suffer from faults
attempting to access that memory later from other phylink_pcs callbacks.

Currently, nothing prevents this situation from happening:

$ ip link set endpmac2 up
$ ip link set endpmac2 down
$ ethtool endpmac2
Unable to handle kernel paging request at virtual address 0000100000000034
Call trace:
 __mutex_lock+0xb8/0x574
 __mutex_lock_slowpath+0x14/0x20
 mutex_lock+0x24/0x58
 mtip_backplane_get_state+0x44/0x24c
 lynx_pcs_get_state+0x30/0x180
 phylink_ethtool_ksettings_get+0x178/0x218
 dpaa2_eth_get_link_ksettings+0x54/0xa4
 __ethtool_get_link_ksettings+0x68/0xa8
 linkmodes_prepare_data+0x44/0xc4
 ethnl_default_doit+0x118/0x39c
 genl_rcv_msg+0x29c/0x314
 netlink_rcv_skb+0x11c/0x134
 genl_rcv+0x34/0x4c

However, the case where "ethtool endpmac2" is executed as the first
thing (before the interface is brought up) does not crash. What's
different is that second situation is that phylink_major_config() did
not run yet, so pl->pcs is still NULL inside phylink_mac_pcs_get_state().
In plain English, "as long as the PCS is disabled, the link is naturally
down, no need to ask".

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
 drivers/net/phy/phylink.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c
index a8459116b701..f78d0e0f7cfb 100644
--- a/drivers/net/phy/phylink.c
+++ b/drivers/net/phy/phylink.c
@@ -2527,6 +2527,7 @@ void phylink_stop(struct phylink *pl)
 	pl->pcs_state = PCS_STATE_DOWN;

 	phylink_pcs_disable(pl->pcs);
+	pl->pcs = NULL;
 }
 EXPORT_SYMBOL_GPL(phylink_stop);

-- >8 --

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

First, I'll say I'm on a very short fuse today; no dinner last night,
at the hospital up until 5:30am, and a fucking cold caller rang the door
bell at 10am this morning. Just fucking our luck.

On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> after calling pcs_disable(), though?

No. We've already called mac_prepare(), pcs_pre_config(),
pcs_post_config() by this time, we're past the point of being able to
unwind.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Vladimir Oltean]

On Tue, Jan 20, 2026 at 10:12:46AM +0000, Russell King (Oracle) wrote:
> First, I'll say I'm on a very short fuse today; no dinner last night,
> at the hospital up until 5:30am, and a fucking cold caller rang the door
> bell at 10am this morning. Just fucking our luck.

Sorry to hear that.

> On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> > Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> > after calling pcs_disable(), though?
>
> No. We've already called mac_prepare(), pcs_pre_config(),
> pcs_post_config() by this time, we're past the point of being able to
> unwind.

I'm set out to resolve a much smaller problem.

Calling it a full "unwind" is perhaps a bit much, because pcs_pre_config()
and pcs_post_config() don't have unwinding equivalents, unlike how
pcs_enable() has pcs_disable(). I don't see what API convention would be
violated if phylink decided to drop a PCS whose enable() returned an error.

Similarly, the fact we don't have to whom to report an error code
doesn't make it pointless to offer the guarantee that pcs_disable() will
be called only when pcs_enable() has succeeded.  It is only the latter
that seems necessary in order to develop reliable complexity on top of
these.

If SerDes PHY integration in phylink_pcs drivers is a model to follow
for other drivers, I think the way in which balanced calls can be made
from pcs_enable()/pcs_disable() needs to be given more attention.
And I think it's a bit worse than "doesn't matter, the port is dead
anyway".  For example, we can have QSGMII where 4 PCSes share a single
SerDes lane, so one single malfunctioning PCS instance can affect all
the others through the lane's phy->power_count.

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

On Tue, Jan 20, 2026 at 02:11:14PM +0200, Vladimir Oltean wrote:
> On Tue, Jan 20, 2026 at 10:12:46AM +0000, Russell King (Oracle) wrote:
> > First, I'll say I'm on a very short fuse today; no dinner last night,
> > at the hospital up until 5:30am, and a fucking cold caller rang the door
> > bell at 10am this morning. Just fucking our luck.
> 
> Sorry to hear that.
> 
> > On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> > > Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> > > after calling pcs_disable(), though?
> >
> > No. We've already called mac_prepare(), pcs_pre_config(),
> > pcs_post_config() by this time, we're past the point of being able to
> > unwind.
> 
> I'm set out to resolve a much smaller problem.
> 
> Calling it a full "unwind" is perhaps a bit much, because pcs_pre_config()
> and pcs_post_config() don't have unwinding equivalents, unlike how
> pcs_enable() has pcs_disable(). I don't see what API convention would be
> violated if phylink decided to drop a PCS whose enable() returned an error.

While pcs_pre_config() and pcs_post_config() do not have unwinding
equivalents (what would they be?) the issue here is that these could
have changed any state that isn't simply undone by calling
pcs_disable().

For example, pcs_pre_config() could have reprogrammed signal routing,
clocking, or power supplies to blocks.

This already applies to Marvell DSA pcs-639x.c, where the pre/post
config hooks change the power state of the PCS block (for errata
handling), and the only way that gets undone is via a call to
pcs_disable() which explicitly disables IRQs and power for the PCS. Its
pcs_disable() isn't a strict reversal of pcs_enable(), it does more.

We already declare the interface to be dead on pcs_post_config()
failure, but we don't do that for pcs_enable() failure.

Maybe I need to explicitly state that pcs_disable() does not directly
balance pcs_enable(), but that _and_ the effects of pcs_pre_config()
and pcs_post_config(). However, that itself will add to the problems.
What if pcs_pre_config() and pcs_post_config() succeed but not
pcs_enable()? pcs-639x needs pcs_disable() to be called, but if we
require pcs_disable() to be balanced with a successful call to
pcs_enable(), that messes up that driver, and pretty much makes it
impossible to work around the errata.

If you feel strongly about this, then the only thing I can think of
doing is to move this SerDes support out of stmmac and into phylink
(which is a point I already raised in the cover message) so that
its failure can be dealt with at the higher level, where we can
ensure that phy_power_off() is balaced with phy_power_on(). However,
that means pushing even more of the stmmac specific "we need the
clocks running to access registers XYZ or reset" weirdness into
phylink.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Vladimir Oltean]

On Wed, Jan 21, 2026 at 02:46:42PM +0000, Russell King (Oracle) wrote:
> On Tue, Jan 20, 2026 at 02:11:14PM +0200, Vladimir Oltean wrote:
> > On Tue, Jan 20, 2026 at 10:12:46AM +0000, Russell King (Oracle) wrote:
> > > First, I'll say I'm on a very short fuse today; no dinner last night,
> > > at the hospital up until 5:30am, and a fucking cold caller rang the door
> > > bell at 10am this morning. Just fucking our luck.
> > 
> > Sorry to hear that.
> > 
> > > On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> > > > Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> > > > after calling pcs_disable(), though?
> > >
> > > No. We've already called mac_prepare(), pcs_pre_config(),
> > > pcs_post_config() by this time, we're past the point of being able to
> > > unwind.
> > 
> > I'm set out to resolve a much smaller problem.
> > 
> > Calling it a full "unwind" is perhaps a bit much, because pcs_pre_config()
> > and pcs_post_config() don't have unwinding equivalents, unlike how
> > pcs_enable() has pcs_disable(). I don't see what API convention would be
> > violated if phylink decided to drop a PCS whose enable() returned an error.
> 
> While pcs_pre_config() and pcs_post_config() do not have unwinding
> equivalents (what would they be?) the issue here is that these could
> have changed any state that isn't simply undone by calling
> pcs_disable().
> 
> For example, pcs_pre_config() could have reprogrammed signal routing,
> clocking, or power supplies to blocks.
> 
> This already applies to Marvell DSA pcs-639x.c, where the pre/post
> config hooks change the power state of the PCS block (for errata
> handling), and the only way that gets undone is via a call to
> pcs_disable() which explicitly disables IRQs and power for the PCS. Its
> pcs_disable() isn't a strict reversal of pcs_enable(), it does more.
> 
> We already declare the interface to be dead on pcs_post_config()
> failure, but we don't do that for pcs_enable() failure.
> 
> Maybe I need to explicitly state that pcs_disable() does not directly
> balance pcs_enable(), but that _and_ the effects of pcs_pre_config()
> and pcs_post_config(). However, that itself will add to the problems.
> What if pcs_pre_config() and pcs_post_config() succeed but not
> pcs_enable()? pcs-639x needs pcs_disable() to be called, but if we
> require pcs_disable() to be balanced with a successful call to
> pcs_enable(), that messes up that driver, and pretty much makes it
> impossible to work around the errata.

What if we reordered phylink_major_config() such that phylink_pcs_enable()
comes first, followed by phylink_pcs_pre_config() -> phylink_mac_config() ->
phylink_pcs_post_config()? Superficially looking at pcs-639x, I don't
think it would break.

If we did that, we'd effectively have to also call pcs_disable() when
pcs_post_config() fails, and that is semantically compatible with saying
that pcs_disable() is balanced with pcs_enable(). It also gives the
ability for drivers such as pcs-639x to unwind in pcs_disable() any
actions done in pcs_enable(), pcs_pre_config() or pcs_post_config().

Plus, it's more natural/useful from an API perspective to say
"the PCS has to be enabled in order for anything to be done with it",
rather than the current "first mac_config cycle runs with the PCS not
enabled; subsequent mac_config cycles run with the PCS enabled".

> If you feel strongly about this, then the only thing I can think of
> doing is to move this SerDes support out of stmmac and into phylink
> (which is a point I already raised in the cover message) so that
> its failure can be dealt with at the higher level, where we can
> ensure that phy_power_off() is balaced with phy_power_on(). However,
> that means pushing even more of the stmmac specific "we need the
> clocks running to access registers XYZ or reset" weirdness into
> phylink.

I think core phylink support for generic PHYs eventually makes sense,
but at this stage it's perhaps too early, there's too much we don't yet
know. I would wait at least until it's clear, with an upstream example,
that multiple generic PHYs per phylink instance are needed: 1 SerDes PHY
per lane (for 40GBase-R etc), plus 1 retimer PHY per lane direction.
Also how do those retimers differ from SerDes PHYs. At the very least,
the phy_validate() of SerDes PHYs should be additive w.r.t.
supported_interfaces, whereas the phy_validate() of retimers should be
subtractive.

Also, moving SerDes PHY into phylink only avoids the problem, but if the
PCS driver needs to allocate memory, it will return. I have downstream
patches for a software backplane AN/LT state machine in phylink_pcs,
which is allocated in pcs_enable() and freed in pcs_disable().

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

On Wed, Jan 21, 2026 at 06:23:45PM +0200, Vladimir Oltean wrote:
> On Wed, Jan 21, 2026 at 02:46:42PM +0000, Russell King (Oracle) wrote:
> > On Tue, Jan 20, 2026 at 02:11:14PM +0200, Vladimir Oltean wrote:
> > > On Tue, Jan 20, 2026 at 10:12:46AM +0000, Russell King (Oracle) wrote:
> > > > First, I'll say I'm on a very short fuse today; no dinner last night,
> > > > at the hospital up until 5:30am, and a fucking cold caller rang the door
> > > > bell at 10am this morning. Just fucking our luck.
> > > 
> > > Sorry to hear that.
> > > 
> > > > On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> > > > > Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> > > > > after calling pcs_disable(), though?
> > > >
> > > > No. We've already called mac_prepare(), pcs_pre_config(),
> > > > pcs_post_config() by this time, we're past the point of being able to
> > > > unwind.
> > > 
> > > I'm set out to resolve a much smaller problem.
> > > 
> > > Calling it a full "unwind" is perhaps a bit much, because pcs_pre_config()
> > > and pcs_post_config() don't have unwinding equivalents, unlike how
> > > pcs_enable() has pcs_disable(). I don't see what API convention would be
> > > violated if phylink decided to drop a PCS whose enable() returned an error.
> > 
> > While pcs_pre_config() and pcs_post_config() do not have unwinding
> > equivalents (what would they be?) the issue here is that these could
> > have changed any state that isn't simply undone by calling
> > pcs_disable().
> > 
> > For example, pcs_pre_config() could have reprogrammed signal routing,
> > clocking, or power supplies to blocks.
> > 
> > This already applies to Marvell DSA pcs-639x.c, where the pre/post
> > config hooks change the power state of the PCS block (for errata
> > handling), and the only way that gets undone is via a call to
> > pcs_disable() which explicitly disables IRQs and power for the PCS. Its
> > pcs_disable() isn't a strict reversal of pcs_enable(), it does more.
> > 
> > We already declare the interface to be dead on pcs_post_config()
> > failure, but we don't do that for pcs_enable() failure.
> > 
> > Maybe I need to explicitly state that pcs_disable() does not directly
> > balance pcs_enable(), but that _and_ the effects of pcs_pre_config()
> > and pcs_post_config(). However, that itself will add to the problems.
> > What if pcs_pre_config() and pcs_post_config() succeed but not
> > pcs_enable()? pcs-639x needs pcs_disable() to be called, but if we
> > require pcs_disable() to be balanced with a successful call to
> > pcs_enable(), that messes up that driver, and pretty much makes it
> > impossible to work around the errata.
> 
> What if we reordered phylink_major_config() such that phylink_pcs_enable()
> comes first, followed by phylink_pcs_pre_config() -> phylink_mac_config() ->
> phylink_pcs_post_config()? Superficially looking at pcs-639x, I don't
> think it would break.

I'm sorry, but I don't have time to continue this discussion today. I
woke late, we're trying to cram in the meals (in the middle of delayed
lunch-time dinner right now), work wants a quick call to discuss a
project that I missed the meeting for yesterday (which I haven't yet
had time for...)

Sorry, but while you may wish to get this sorted, for me this is a very
low priority issue that can be addressed later. Don't think I will have
time to review anything you send - and that's not a personal attack,
it's because I'm barely managing to hold everything together at my
end, and I don't have the time.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Vladimir Oltean]

On Wed, Jan 21, 2026 at 05:33:28PM +0000, Russell King (Oracle) wrote:
> On Wed, Jan 21, 2026 at 06:23:45PM +0200, Vladimir Oltean wrote:
> > On Wed, Jan 21, 2026 at 02:46:42PM +0000, Russell King (Oracle) wrote:
> > > On Tue, Jan 20, 2026 at 02:11:14PM +0200, Vladimir Oltean wrote:
> > > > On Tue, Jan 20, 2026 at 10:12:46AM +0000, Russell King (Oracle) wrote:
> > > > > First, I'll say I'm on a very short fuse today; no dinner last night,
> > > > > at the hospital up until 5:30am, and a fucking cold caller rang the door
> > > > > bell at 10am this morning. Just fucking our luck.
> > > > 
> > > > Sorry to hear that.
> > > > 
> > > > > On Tue, Jan 20, 2026 at 10:18:44AM +0200, Vladimir Oltean wrote:
> > > > > > Isn't it sufficient to set pl->pcs to NULL when pcs_enable() fails and
> > > > > > after calling pcs_disable(), though?
> > > > >
> > > > > No. We've already called mac_prepare(), pcs_pre_config(),
> > > > > pcs_post_config() by this time, we're past the point of being able to
> > > > > unwind.
> > > > 
> > > > I'm set out to resolve a much smaller problem.
> > > > 
> > > > Calling it a full "unwind" is perhaps a bit much, because pcs_pre_config()
> > > > and pcs_post_config() don't have unwinding equivalents, unlike how
> > > > pcs_enable() has pcs_disable(). I don't see what API convention would be
> > > > violated if phylink decided to drop a PCS whose enable() returned an error.
> > > 
> > > While pcs_pre_config() and pcs_post_config() do not have unwinding
> > > equivalents (what would they be?) the issue here is that these could
> > > have changed any state that isn't simply undone by calling
> > > pcs_disable().
> > > 
> > > For example, pcs_pre_config() could have reprogrammed signal routing,
> > > clocking, or power supplies to blocks.
> > > 
> > > This already applies to Marvell DSA pcs-639x.c, where the pre/post
> > > config hooks change the power state of the PCS block (for errata
> > > handling), and the only way that gets undone is via a call to
> > > pcs_disable() which explicitly disables IRQs and power for the PCS. Its
> > > pcs_disable() isn't a strict reversal of pcs_enable(), it does more.
> > > 
> > > We already declare the interface to be dead on pcs_post_config()
> > > failure, but we don't do that for pcs_enable() failure.
> > > 
> > > Maybe I need to explicitly state that pcs_disable() does not directly
> > > balance pcs_enable(), but that _and_ the effects of pcs_pre_config()
> > > and pcs_post_config(). However, that itself will add to the problems.
> > > What if pcs_pre_config() and pcs_post_config() succeed but not
> > > pcs_enable()? pcs-639x needs pcs_disable() to be called, but if we
> > > require pcs_disable() to be balanced with a successful call to
> > > pcs_enable(), that messes up that driver, and pretty much makes it
> > > impossible to work around the errata.
> > 
> > What if we reordered phylink_major_config() such that phylink_pcs_enable()
> > comes first, followed by phylink_pcs_pre_config() -> phylink_mac_config() ->
> > phylink_pcs_post_config()? Superficially looking at pcs-639x, I don't
> > think it would break.
> 
> I'm sorry, but I don't have time to continue this discussion today. I
> woke late, we're trying to cram in the meals (in the middle of delayed
> lunch-time dinner right now), work wants a quick call to discuss a
> project that I missed the meeting for yesterday (which I haven't yet
> had time for...)
> 
> Sorry, but while you may wish to get this sorted, for me this is a very
> low priority issue that can be addressed later. Don't think I will have
> time to review anything you send - and that's not a personal attack,
> it's because I'm barely managing to hold everything together at my
> end, and I don't have the time.

Thanks, this was a good talk, I understood a bit more about the
challenges that need to be overcome. I'll do some testing on the
Turris MOX with a 6390 switch. From my side this shouldn't block the
stmmac integrated PCS from being integrated with the SerDes, but I do
agree that leaving a comment explaining the current phylink_pcs calling
convention, as Jakub requested, would be very useful.

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Vladimir Oltean]

On Tue, Jan 20, 2026 at 05:04:53AM +0000, Russell King (Oracle) wrote:
> On Mon, Jan 19, 2026 at 11:21:25AM -0800, Jakub Kicinski wrote:
> > This is an AI-generated review of your patch. The human sending this
> > email has considered the AI review valid, or at least plausible.
> > 
> > For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
> > ---
> > net: stmmac: add stmmac core serdes support
> > 
> > This commit adds SerDes PHY support to the stmmac core driver, moving it
> > from platform glue into the integrated PCS driver. The SerDes is managed
> > through phy_init/phy_exit for lifecycle and phy_power_on/phy_power_off
> > for the power state.
> > 
> > > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > > index 2f826fe7229b..4d1902f3a58f 100644
> > > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> > 
> > [ ... ]
> > 
> > > static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
> > > {
> > > 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> > > +	struct stmmac_priv *priv = spcs->priv;
> > > +	int ret;
> > >
> > > -	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
> > > +	ret = dwmac_serdes_power_on(priv);
> > > +	if (ret)
> > > +		return ret;
> > > +
> > > +	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
> > > +		ret = dwmac_serdes_set_mode(priv, spcs->interface);
> > > +		if (ret)
> > > +			return ret;
> >                         ^^^^^^^^^^
> > 
> > If dwmac_serdes_set_mode() fails here, doesn't this leave the SerDes in
> > a powered-on state? dwmac_serdes_power_on() succeeded at this point but
> > there's no corresponding dwmac_serdes_power_off() call before returning
> > the error.
> > 
> > Looking at phylink_major_config(), it appears to discard the return
> > value from phylink_pcs_enable(), so the caller won't know to call
> > pcs_disable() to clean up the power state.
> 
> This AI analysis is incorrect.
> 
> By the time phylink_pcs_enable() has been called, the PCS is already
> plumbed in to phylink. It _will_ have phylink_pcs_disable() called on
> it at some point in the future, either by having the PCS displaced
> by another in a subsequent phylink_major_config(), or by a driver
> calling phylink_stop().
> 
> If we clean up here, then we will call dwmac_serdes_power_off() twice.
> 
> Yes, it's not "nice" but that's the way phylink is right now, and
> without reworking phylink to record that pcs_enable() has failed
> to avoid a subsequent pcs_disable(), and to stop the major config
> (which then potentially causes a whole bunch of other issues). I
> don't even want to think about that horrid scenario at the moment.

More to the point, if dwmac_integrated_pcs_enable() fails at
dwmac_serdes_power_on() (thus, the SerDes is _not_ powered on), by your
own admission of this PCS calling convention, sooner or later
dwmac_integrated_pcs_disable() -> dwmac_serdes_power_off() will still be
called, leading to a negative phy->power_count.

That is to say, if the model is "irrespective of whether pcs_enable()
succeeds or fails mid way, pcs_disable is called anyway()", then these
methods are not prepared to handle that reliably.

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

On Tue, Jan 20, 2026 at 10:42:27AM +0200, Vladimir Oltean wrote:
> More to the point, if dwmac_integrated_pcs_enable() fails at
> dwmac_serdes_power_on() (thus, the SerDes is _not_ powered on), by your
> own admission of this PCS calling convention, sooner or later
> dwmac_integrated_pcs_disable() -> dwmac_serdes_power_off() will still be
> called, leading to a negative phy->power_count.
> 
> That is to say, if the model is "irrespective of whether pcs_enable()
> succeeds or fails mid way, pcs_disable is called anyway()", then these
> methods are not prepared to handle that reliably.

That's the way it currently is, and it's been this way in the
major_config path for a very long time. If anything fails in that
path, we can't report the error back up to anyone, and the netdev
is effectively dead.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [net-next,05/14] net: stmmac: add stmmac core serdes support

[Jakub Kicinski]

On Tue, 20 Jan 2026 05:04:53 +0000 Russell King (Oracle) wrote:
> By the time phylink_pcs_enable() has been called, the PCS is already
> plumbed in to phylink. It _will_ have phylink_pcs_disable() called on
> it at some point in the future, either by having the PCS displaced
> by another in a subsequent phylink_major_config(), or by a driver
> calling phylink_stop().
> 
> If we clean up here, then we will call dwmac_serdes_power_off() twice.
> 
> Yes, it's not "nice" but that's the way phylink is right now, and
> without reworking phylink to record that pcs_enable() has failed
> to avoid a subsequent pcs_disable(), and to stop the major config
> (which then potentially causes a whole bunch of other issues). I
> don't even want to think about that horrid scenario at the moment.

Would you mind adding a note to this effect to the commit message 
to shut up the bot?

Unless the comment on patch 12 is also incorrect in which case I'll
restore the v1 into patchwork.

[PATCH net-next 06/14] net: stmmac: qcom-ethqos: convert to dwmac generic SerDes support

[Russell King (Oracle)]

Convert qcom-ethqos to use the dwmac core's generic SerDes support,
which will handle SerDes initialisation, powering, and mode setting.

Note that generic support requires the SerDes to support phy_validate()
in order to probe which PHY interface modes are supported, and
phy_set_mode_ext() to configure the appropriate PHY interface mode
(and thus the speed.)

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../stmicro/stmmac/dwmac-qcom-ethqos.c        | 50 ++-----------------
 1 file changed, 3 insertions(+), 47 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
index 80ea69fc8ee5..a0b893d3fbd4 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
@@ -103,8 +103,6 @@ struct qcom_ethqos {
 	int (*configure_func)(struct qcom_ethqos *ethqos, int speed);
 
 	struct clk *link_clk;
-	struct phy *serdes_phy;
-	int serdes_speed;
 	phy_interface_t phy_mode;
 
 	const struct ethqos_emac_por *por;
@@ -584,14 +582,6 @@ static int ethqos_configure_rgmii(struct qcom_ethqos *ethqos, int speed)
 	return 0;
 }
 
-static void ethqos_set_serdes_speed(struct qcom_ethqos *ethqos, int speed)
-{
-	if (ethqos->serdes_speed != speed) {
-		phy_set_speed(ethqos->serdes_phy, speed);
-		ethqos->serdes_speed = speed;
-	}
-}
-
 static void ethqos_pcs_set_inband(struct stmmac_priv *priv, bool enable)
 {
 	stmmac_pcs_ctrl_ane(priv, enable, 0);
@@ -609,17 +599,14 @@ static int ethqos_configure_sgmii(struct qcom_ethqos *ethqos, int speed)
 	case SPEED_2500:
 		rgmii_setmask(ethqos, RGMII_CONFIG2_RGMII_CLK_SEL_CFG,
 			      RGMII_IO_MACRO_CONFIG2);
-		ethqos_set_serdes_speed(ethqos, SPEED_2500);
 		ethqos_pcs_set_inband(priv, false);
 		break;
 	case SPEED_1000:
 		rgmii_setmask(ethqos, RGMII_CONFIG2_RGMII_CLK_SEL_CFG,
 			      RGMII_IO_MACRO_CONFIG2);
-		ethqos_set_serdes_speed(ethqos, SPEED_1000);
 		ethqos_pcs_set_inband(priv, true);
 		break;
 	case SPEED_100:
-		ethqos_set_serdes_speed(ethqos, SPEED_1000);
 		ethqos_pcs_set_inband(priv, true);
 		break;
 	case SPEED_10:
@@ -627,7 +614,6 @@ static int ethqos_configure_sgmii(struct qcom_ethqos *ethqos, int speed)
 			      FIELD_PREP(RGMII_CONFIG_SGMII_CLK_DVDR,
 					 SGMII_10M_RX_CLK_DVDR),
 			      RGMII_IO_MACRO_CONFIG);
-		ethqos_set_serdes_speed(ethqos, SPEED_1000);
 		ethqos_pcs_set_inband(priv, true);
 		break;
 	}
@@ -648,30 +634,6 @@ static void ethqos_fix_mac_speed(void *priv, int speed, unsigned int mode)
 	ethqos_configure(ethqos, speed);
 }
 
-static int qcom_ethqos_serdes_powerup(struct net_device *ndev, void *priv)
-{
-	struct qcom_ethqos *ethqos = priv;
-	int ret;
-
-	ret = phy_init(ethqos->serdes_phy);
-	if (ret)
-		return ret;
-
-	ret = phy_power_on(ethqos->serdes_phy);
-	if (ret)
-		return ret;
-
-	return phy_set_speed(ethqos->serdes_phy, ethqos->serdes_speed);
-}
-
-static void qcom_ethqos_serdes_powerdown(struct net_device *ndev, void *priv)
-{
-	struct qcom_ethqos *ethqos = priv;
-
-	phy_power_off(ethqos->serdes_phy);
-	phy_exit(ethqos->serdes_phy);
-}
-
 static int ethqos_clks_config(void *priv, bool enabled)
 {
 	struct qcom_ethqos *ethqos = priv;
@@ -790,12 +752,11 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	ethqos->serdes_phy = devm_phy_optional_get(dev, "serdes");
-	if (IS_ERR(ethqos->serdes_phy))
-		return dev_err_probe(dev, PTR_ERR(ethqos->serdes_phy),
+	plat_dat->serdes = devm_phy_optional_get(dev, "serdes");
+	if (IS_ERR(plat_dat->serdes))
+		return dev_err_probe(dev, PTR_ERR(plat_dat->serdes),
 				     "Failed to get serdes phy\n");
 
-	ethqos->serdes_speed = SPEED_1000;
 	ethqos_set_clk_tx_rate(ethqos, NULL, plat_dat->phy_interface,
 			       SPEED_1000);
 	ethqos_set_func_clk_en(ethqos);
@@ -816,11 +777,6 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
 	if (data->dma_addr_width)
 		plat_dat->host_dma_width = data->dma_addr_width;
 
-	if (ethqos->serdes_phy) {
-		plat_dat->serdes_powerup = qcom_ethqos_serdes_powerup;
-		plat_dat->serdes_powerdown  = qcom_ethqos_serdes_powerdown;
-	}
-
 	/* Enable TSO on queue0 and enable TBS on rest of the queues */
 	for (i = 1; i < plat_dat->tx_queues_to_use; i++)
 		plat_dat->tx_queues_cfg[i].tbs_en = 1;
-- 
2.47.3

[PATCH net-next 07/14] net: stmmac: move most PCS register definitions to stmmac_pcs.c

[Russell King (Oracle)]

Move most of the PCS register offset definitions to stmmac_pcs.c.
Since stmmac_pcs.c only ever passes zero into the register offset
macros, remove that ability, making them simple constant integer
definitions.

Add appropriate descriptions of the registers, pointing out their
similarity with their IEEE 802.3 counterparts. Make use of the
BMSR definitions for the GMAC_AN_STATUS register and remove the
driver private versions.

Note that BMSR_LSTATUS is non-low-latching, unlike it's 802.3z
counterpart.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  | 22 +++++++++++++++----
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.h  | 22 -------------------
 2 files changed, 18 insertions(+), 26 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 4d1902f3a58f..718e5360fca3 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -3,6 +3,20 @@
 #include "stmmac_pcs.h"
 #include "stmmac_serdes.h"
 
+/*
+ * GMAC_AN_STATUS is equivalent to MII_BMSR
+ * GMAC_ANE_ADV is equivalent to 802.3z MII_ADVERTISE
+ * GMAC_ANE_LPA is equivalent to 802.3z MII_LPA
+ * GMAC_ANE_EXP is equivalent to MII_EXPANSION
+ * GMAC_TBI is equivalent to MII_ESTATUS
+ *
+ * ADV, LPA and EXP are only available for the TBI and RTBI modes.
+ */
+#define GMAC_AN_STATUS	0x04	/* AN status */
+#define GMAC_ANE_ADV	0x08	/* ANE Advertisement */
+#define GMAC_ANE_LPA	0x0c	/* ANE link partener ability */
+#define GMAC_TBI	0x14	/* TBI extend status */
+
 static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
@@ -74,11 +88,11 @@ void stmmac_integrated_pcs_irq(struct stmmac_priv *priv, u32 status,
 			       struct stmmac_extra_stats *x)
 {
 	struct stmmac_pcs *spcs = priv->integrated_pcs;
-	u32 val = readl(spcs->base + GMAC_AN_STATUS(0));
+	u32 val = readl(spcs->base + GMAC_AN_STATUS);
 
 	if (status & PCS_ANE_IRQ) {
 		x->irq_pcs_ane_n++;
-		if (val & GMAC_AN_STATUS_ANC)
+		if (val & BMSR_ANEGCOMPLETE)
 			dev_info(priv->device,
 				 "PCS ANE process completed\n");
 	}
@@ -86,9 +100,9 @@ void stmmac_integrated_pcs_irq(struct stmmac_priv *priv, u32 status,
 	if (status & PCS_LINK_IRQ) {
 		x->irq_pcs_link_n++;
 		dev_info(priv->device, "PCS Link %s\n",
-			 val & GMAC_AN_STATUS_LS ? "Up" : "Down");
+			 val & BMSR_LSTATUS ? "Up" : "Down");
 
-		phylink_pcs_change(&spcs->pcs, val & GMAC_AN_STATUS_LS);
+		phylink_pcs_change(&spcs->pcs, val & BMSR_LSTATUS);
 	}
 }
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index 36bf75fdf478..887c4ff302aa 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -16,13 +16,6 @@
 
 /* PCS registers (AN/TBI/SGMII/RGMII) offsets */
 #define GMAC_AN_CTRL(x)		(x)		/* AN control */
-#define GMAC_AN_STATUS(x)	(x + 0x4)	/* AN status */
-
-/* ADV, LPA and EXP are only available for the TBI and RTBI interfaces */
-#define GMAC_ANE_ADV(x)		(x + 0x8)	/* ANE Advertisement */
-#define GMAC_ANE_LPA(x)		(x + 0xc)	/* ANE link partener ability */
-#define GMAC_ANE_EXP(x)		(x + 0x10)	/* ANE expansion */
-#define GMAC_TBI(x)		(x + 0x14)	/* TBI extend status */
 
 /* AN Configuration defines */
 #define GMAC_AN_CTRL_RAN	BIT_U32(9)	/* Restart Auto-Negotiation */
@@ -32,21 +25,6 @@
 #define GMAC_AN_CTRL_LR		BIT_U32(17)	/* Lock to Reference */
 #define GMAC_AN_CTRL_SGMRAL	BIT_U32(18)	/* SGMII RAL Control */
 
-/* AN Status defines */
-#define GMAC_AN_STATUS_LS	BIT_U32(2)	/* Link Status 0:down 1:up */
-#define GMAC_AN_STATUS_ANA	BIT_U32(3)	/* Auto-Negotiation Ability */
-#define GMAC_AN_STATUS_ANC	BIT_U32(5)	/* Auto-Negotiation Complete */
-#define GMAC_AN_STATUS_ES	BIT_U32(8)	/* Extended Status */
-
-/* ADV and LPA defines */
-#define GMAC_ANE_FD		BIT_U32(5)
-#define GMAC_ANE_HD		BIT_U32(6)
-#define GMAC_ANE_PSE		GENMASK_U32(8, 7)
-#define GMAC_ANE_PSE_SHIFT	7
-#define GMAC_ANE_RFE		GENMASK_U32(13, 12)
-#define GMAC_ANE_RFE_SHIFT	12
-#define GMAC_ANE_ACK		BIT_U32(14)
-
 struct stmmac_priv;
 
 struct stmmac_pcs {
-- 
2.47.3

[PATCH net-next 08/14] net: stmmac: handle integrated PCS phy_intf_sel separately

[Russell King (Oracle)]

The dwmac core has no support for SGMII without using its integrated
PCS. Thus, PHY_INTF_SEL_SGMII is only supported when this block is
present, and it makes no sense for stmmac_get_phy_intf_sel() to decode
this.

None of the platform glue users that use stmmac_get_phy_intf_sel()
directly accept PHY_INTF_SEL_SGMII as a valid mode.

Check whether a PCS will be used by the driver for the interface mode,
and if it is the integrated PCS, query the integrated PCS for the
phy_intf_sel_i value to use.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 15 ++++++++++++---
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c  |  9 +++++++++
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h  |  2 ++
 3 files changed, 23 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 6c515f9efbe7..5254d9d19ffe 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -3135,8 +3135,6 @@ int stmmac_get_phy_intf_sel(phy_interface_t interface)
 		phy_intf_sel = PHY_INTF_SEL_GMII_MII;
 	else if (phy_interface_mode_is_rgmii(interface))
 		phy_intf_sel = PHY_INTF_SEL_RGMII;
-	else if (interface == PHY_INTERFACE_MODE_SGMII)
-		phy_intf_sel = PHY_INTF_SEL_SGMII;
 	else if (interface == PHY_INTERFACE_MODE_RMII)
 		phy_intf_sel = PHY_INTF_SEL_RMII;
 	else if (interface == PHY_INTERFACE_MODE_REVMII)
@@ -3150,13 +3148,24 @@ static int stmmac_prereset_configure(struct stmmac_priv *priv)
 {
 	struct plat_stmmacenet_data *plat_dat = priv->plat;
 	phy_interface_t interface;
+	struct phylink_pcs *pcs;
 	int phy_intf_sel, ret;
 
 	if (!plat_dat->set_phy_intf_sel)
 		return 0;
 
 	interface = plat_dat->phy_interface;
-	phy_intf_sel = stmmac_get_phy_intf_sel(interface);
+
+	/* Check whether this mode uses a PCS */
+	pcs = stmmac_mac_select_pcs(&priv->phylink_config, interface);
+	if (priv->integrated_pcs && pcs == &priv->integrated_pcs->pcs) {
+		/* Request the phy_intf_sel from the integrated PCS */
+		phy_intf_sel = stmmac_integrated_pcs_get_phy_intf_sel(priv,
+								    interface);
+	} else {
+		phy_intf_sel = stmmac_get_phy_intf_sel(interface);
+	}
+
 	if (phy_intf_sel < 0) {
 		netdev_err(priv->dev,
 			   "failed to get phy_intf_sel for %s: %pe\n",
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 718e5360fca3..cf7337e9ed3e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -106,6 +106,15 @@ void stmmac_integrated_pcs_irq(struct stmmac_priv *priv, u32 status,
 	}
 }
 
+int stmmac_integrated_pcs_get_phy_intf_sel(struct stmmac_priv *priv,
+					   phy_interface_t interface)
+{
+	if (interface == PHY_INTERFACE_MODE_SGMII)
+		return PHY_INTF_SEL_SGMII;
+
+	return -EINVAL;
+}
+
 int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 			       u32 int_mask)
 {
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index 887c4ff302aa..845bcad9d0f7 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -43,6 +43,8 @@ phylink_pcs_to_stmmac_pcs(struct phylink_pcs *pcs)
 
 void stmmac_integrated_pcs_irq(struct stmmac_priv *priv, u32 status,
 			       struct stmmac_extra_stats *x);
+int stmmac_integrated_pcs_get_phy_intf_sel(struct stmmac_priv *priv,
+					   phy_interface_t interface);
 int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 			       u32 int_mask);
 
-- 
2.47.3

[PATCH net-next 09/14] net: stmmac: add BASE-X support to integrated PCS

[Russell King (Oracle)]

The integrated PCS supports 802.3z (BASE-X) modes when the Synopsys
IP is coupled with an appropriate SerDes to provide the electrical
interface. The PCS presents a TBI interface to the SerDes for this.
Thus, the BASE-X related registers are only present when TBI mode is
supported.

dwmac-qcom-ethqos added support for using 2.5G with the integrated PCS
by calling dwmac_ctrl_ane() directly.

Add support for 1000BASE-X mode to the integrated PCS support if the
PCS supports TBI, and 2500BASE-X if we have a SerDes that supports
this mode.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  | 95 ++++++++++++++++++-
 1 file changed, 92 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index cf7337e9ed3e..edcf36083806 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -17,6 +17,50 @@
 #define GMAC_ANE_LPA	0x0c	/* ANE link partener ability */
 #define GMAC_TBI	0x14	/* TBI extend status */
 
+static enum ethtool_link_mode_bit_indices dwmac_hd_mode_bits[] = {
+	ETHTOOL_LINK_MODE_10baseT_Half_BIT,
+	ETHTOOL_LINK_MODE_100baseT_Half_BIT,
+	ETHTOOL_LINK_MODE_1000baseT_Half_BIT,
+	ETHTOOL_LINK_MODE_100baseFX_Half_BIT,
+	ETHTOOL_LINK_MODE_10baseT1S_Half_BIT,
+	ETHTOOL_LINK_MODE_10baseT1S_P2MP_Half_BIT,
+};
+
+static int dwmac_integrated_pcs_validate(struct phylink_pcs *pcs,
+					 unsigned long *supported,
+					 const struct phylink_link_state *state)
+{
+	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	size_t i;
+	u32 val;
+
+	if (phy_interface_mode_is_8023z(state->interface)) {
+		/* ESTATUS_1000_XFULL is always set, so full duplex is
+		 * supported. ESTATUS_1000_XHALF depends on core configuration.
+		 */
+		val = readl(spcs->base + GMAC_TBI);
+		if (~val & ESTATUS_1000_XHALF)
+			for (i = 0; i < ARRAY_SIZE(dwmac_hd_mode_bits); i++)
+				linkmode_clear_bit(dwmac_hd_mode_bits[i],
+						   supported);
+
+		return 0;
+	} else if (state->interface == PHY_INTERFACE_MODE_SGMII) {
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static unsigned int dwmac_integrated_pcs_inband_caps(struct phylink_pcs *pcs,
+						     phy_interface_t interface)
+{
+	if (phy_interface_mode_is_8023z(interface))
+		return LINK_INBAND_ENABLE | LINK_INBAND_DISABLE;
+
+	return 0;
+}
+
 static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
@@ -52,7 +96,23 @@ static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
 					   unsigned int neg_mode,
 					   struct phylink_link_state *state)
 {
-	state->link = false;
+	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	u32 status, lpa;
+
+	status = readl(spcs->base + GMAC_AN_STATUS);
+
+	if (phy_interface_mode_is_8023z(state->interface)) {
+		/* For 802.3z modes, the PCS block supports the advertisement
+		 * and link partner advertisement registers using standard
+		 * 802.3 format. The status register also has the link status
+		 * and AN complete bits in the same bit location.
+		 */
+		lpa = readl(spcs->base + GMAC_ANE_LPA);
+
+		phylink_mii_c22_pcs_decode_state(state, neg_mode, status, lpa);
+	} else {
+		state->link = false;
+	}
 }
 
 static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
@@ -62,6 +122,8 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 				       bool permit_pause_to_mac)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	bool changed = false, ane = true;
+	u32 adv;
 	int ret;
 
 	if (spcs->interface != interface) {
@@ -72,12 +134,25 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 		spcs->interface = interface;
 	}
 
-	dwmac_ctrl_ane(spcs->base, 0, 1, spcs->priv->hw->reverse_sgmii_enable);
+	if (phy_interface_mode_is_8023z(interface)) {
+		adv = phylink_mii_c22_pcs_encode_advertisement(interface,
+							       advertising);
+		if (readl(spcs->base + GMAC_ANE_ADV) != adv)
+			changed = true;
+		writel(adv, spcs->base + GMAC_ANE_ADV);
 
-	return 0;
+		ane = neg_mode == PHYLINK_PCS_NEG_INBAND_ENABLED;
+	}
+
+	dwmac_ctrl_ane(spcs->base, 0, ane,
+		       spcs->priv->hw->reverse_sgmii_enable);
+
+	return changed;
 }
 
 static const struct phylink_pcs_ops dwmac_integrated_pcs_ops = {
+	.pcs_validate = dwmac_integrated_pcs_validate,
+	.pcs_inband_caps = dwmac_integrated_pcs_inband_caps,
 	.pcs_enable = dwmac_integrated_pcs_enable,
 	.pcs_disable = dwmac_integrated_pcs_disable,
 	.pcs_get_state = dwmac_integrated_pcs_get_state,
@@ -112,6 +187,9 @@ int stmmac_integrated_pcs_get_phy_intf_sel(struct stmmac_priv *priv,
 	if (interface == PHY_INTERFACE_MODE_SGMII)
 		return PHY_INTF_SEL_SGMII;
 
+	if (phy_interface_mode_is_8023z(interface))
+		return PHY_INTF_SEL_TBI;
+
 	return -EINVAL;
 }
 
@@ -140,6 +218,17 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 
 	__set_bit(PHY_INTERFACE_MODE_SGMII, spcs->pcs.supported_interfaces);
 
+	if (readl(spcs->base + GMAC_AN_STATUS) & BMSR_ESTATEN) {
+		__set_bit(PHY_INTERFACE_MODE_1000BASEX,
+			  spcs->pcs.supported_interfaces);
+
+		/* Only allow 2500Base-X if the SerDes has support. */
+		ret = dwmac_serdes_validate(priv, PHY_INTERFACE_MODE_2500BASEX);
+		if (ret == 0)
+			__set_bit(PHY_INTERFACE_MODE_2500BASEX,
+				  spcs->pcs.supported_interfaces);
+	}
+
 	priv->integrated_pcs = spcs;
 
 	return 0;
-- 
2.47.3

[PATCH net-next 10/14] net: stmmac: use integrated PCS for BASE-X modes

[Russell King (Oracle)]

dwmac-qcom-ethqos supports SGMII and 2500BASE-X using the integrated
PCS, so we need to expand the PCS support to include support for
BASE-X modes.

Add support to the prereset configuration to detect 2500BASE-X, and
arrange for stmmac_mac_select_pcs() to return the integrated PCS if
its supported_interfaces bitmap reports support for the interface mode.

This results in priv->hw->pcs now being write-only, so remove it.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/common.h      | 1 -
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 8 ++------
 2 files changed, 2 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index 49df46be3669..8ef54f6e78f6 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -629,7 +629,6 @@ struct mac_device_info {
 	unsigned int unicast_filter_entries;
 	unsigned int mcast_bits_log2;
 	unsigned int rx_csum;
-	unsigned int pcs;
 	unsigned int xlgmac;
 	unsigned int num_vlan;
 	u32 vlan_filter[32];
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 5254d9d19ffe..a63ae6c4bc8a 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -911,11 +911,8 @@ static struct phylink_pcs *stmmac_mac_select_pcs(struct phylink_config *config,
 			return pcs;
 	}
 
-	/* The PCS control register is only relevant for SGMII, TBI and RTBI
-	 * modes. We no longer support TBI or RTBI, so only configure this
-	 * register when operating in SGMII mode with the integrated PCS.
-	 */
-	if (priv->hw->pcs & STMMAC_PCS_SGMII && priv->integrated_pcs)
+	if (priv->integrated_pcs &&
+	    test_bit(interface, priv->integrated_pcs->pcs.supported_interfaces))
 		return &priv->integrated_pcs->pcs;
 
 	return NULL;
@@ -1173,7 +1170,6 @@ static void stmmac_check_pcs_mode(struct stmmac_priv *priv)
 
 	if (priv->dma_cap.pcs && interface == PHY_INTERFACE_MODE_SGMII) {
 		netdev_dbg(priv->dev, "PCS SGMII support enabled\n");
-		priv->hw->pcs = STMMAC_PCS_SGMII;
 
 		switch (speed) {
 		case SPEED_10:
-- 
2.47.3

Re: [PATCH net-next 10/14] net: stmmac: use integrated PCS for BASE-X modes

[Maxime Chevallier]

On 19/01/2026 13:34, Russell King (Oracle) wrote:
> dwmac-qcom-ethqos supports SGMII and 2500BASE-X using the integrated
> PCS, so we need to expand the PCS support to include support for
> BASE-X modes.
> 
> Add support to the prereset configuration to detect 2500BASE-X, and
> arrange for stmmac_mac_select_pcs() to return the integrated PCS if
> its supported_interfaces bitmap reports support for the interface mode.
> 
> This results in priv->hw->pcs now being write-only, so remove it.
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>

Maxime

[PATCH net-next 11/14] net: stmmac: add struct stmmac_pcs_info

[Russell King (Oracle)]

We need to describe one more register (offset and field bitmask) to
the PCS code. Move the existing PCS offset and interrupt enable bits
to a new struct and pass that in to stmmac_integrated_pcs_init().

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c | 9 ++++++---
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c    | 8 ++++++--
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c     | 8 ++++----
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h     | 9 +++++++--
 4 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c b/drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c
index af566636fad9..a3ef237de1b8 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c
@@ -22,14 +22,17 @@
 #include "stmmac_ptp.h"
 #include "dwmac1000.h"
 
+static const struct stmmac_pcs_info dwmac1000_pcs_info = {
+	.pcs_offset = GMAC_PCS_BASE,
+	.int_mask = GMAC_INT_DISABLE_PCSLINK | GMAC_INT_DISABLE_PCSAN,
+};
+
 static int dwmac1000_pcs_init(struct stmmac_priv *priv)
 {
 	if (!priv->dma_cap.pcs)
 		return 0;
 
-	return stmmac_integrated_pcs_init(priv, GMAC_PCS_BASE,
-					  GMAC_INT_DISABLE_PCSLINK |
-					  GMAC_INT_DISABLE_PCSAN);
+	return stmmac_integrated_pcs_init(priv, &dwmac1000_pcs_info);
 }
 
 static void dwmac1000_core_init(struct mac_device_info *hw,
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c b/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
index 623868afe93d..7f4949229288 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
@@ -22,13 +22,17 @@
 #include "dwmac4.h"
 #include "dwmac5.h"
 
+static const struct stmmac_pcs_info dwmac4_pcs_info = {
+	.pcs_offset = GMAC_PCS_BASE,
+	.int_mask = GMAC_INT_PCS_LINK | GMAC_INT_PCS_ANE,
+};
+
 static int dwmac4_pcs_init(struct stmmac_priv *priv)
 {
 	if (!priv->dma_cap.pcs)
 		return 0;
 
-	return stmmac_integrated_pcs_init(priv, GMAC_PCS_BASE,
-					  GMAC_INT_PCS_LINK | GMAC_INT_PCS_ANE);
+	return stmmac_integrated_pcs_init(priv, &dwmac4_pcs_info);
 }
 
 static void dwmac4_core_init(struct mac_device_info *hw,
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index edcf36083806..73fc56ce5e55 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -193,8 +193,8 @@ int stmmac_integrated_pcs_get_phy_intf_sel(struct stmmac_priv *priv,
 	return -EINVAL;
 }
 
-int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
-			       u32 int_mask)
+int stmmac_integrated_pcs_init(struct stmmac_priv *priv,
+			       const struct stmmac_pcs_info *pcs_info)
 {
 	struct stmmac_pcs *spcs;
 	int ret;
@@ -204,8 +204,8 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 		return -ENOMEM;
 
 	spcs->priv = priv;
-	spcs->base = priv->ioaddr + offset;
-	spcs->int_mask = int_mask;
+	spcs->base = priv->ioaddr + pcs_info->pcs_offset;
+	spcs->int_mask = pcs_info->int_mask;
 	spcs->pcs.ops = &dwmac_integrated_pcs_ops;
 
 	if (priv->plat->serdes) {
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index 845bcad9d0f7..a7c71f40f952 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -27,6 +27,11 @@
 
 struct stmmac_priv;
 
+struct stmmac_pcs_info {
+	unsigned int pcs_offset;
+	u32 int_mask;
+};
+
 struct stmmac_pcs {
 	struct stmmac_priv *priv;
 	void __iomem *base;
@@ -45,8 +50,8 @@ void stmmac_integrated_pcs_irq(struct stmmac_priv *priv, u32 status,
 			       struct stmmac_extra_stats *x);
 int stmmac_integrated_pcs_get_phy_intf_sel(struct stmmac_priv *priv,
 					   phy_interface_t interface);
-int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
-			       u32 int_mask);
+int stmmac_integrated_pcs_init(struct stmmac_priv *priv,
+			       const struct stmmac_pcs_info *pcs_info);
 
 /**
  * dwmac_ctrl_ane - To program the AN Control Register.
-- 
2.47.3

Re: [PATCH net-next 11/14] net: stmmac: add struct stmmac_pcs_info

[Maxime Chevallier]

On 19/01/2026 13:34, Russell King (Oracle) wrote:
> We need to describe one more register (offset and field bitmask) to
> the PCS code. Move the existing PCS offset and interrupt enable bits
> to a new struct and pass that in to stmmac_integrated_pcs_init().
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>

Maxime

[PATCH net-next 12/14] net: stmmac: add support for reading inband SGMII status

[Russell King (Oracle)]

Report the link, speed and duplex for SGMII links, read from the
SGMII, RGMII and SMII status and control register.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
rfc->v1: fix setting SGMII's link status - depend on both link status.
---
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  | 46 ++++++++++++++++++-
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.h  |  4 ++
 2 files changed, 48 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 73fc56ce5e55..9dd7e78cfbc4 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -17,6 +17,16 @@
 #define GMAC_ANE_LPA	0x0c	/* ANE link partener ability */
 #define GMAC_TBI	0x14	/* TBI extend status */
 
+/*
+ * RGSMII status bitfield definitions.
+ */
+#define GMAC_RGSMIII_LNKMOD		BIT(0)
+#define GMAC_RGSMIII_SPEED_MASK		GENMASK(2, 1)
+#define GMAC_RGSMIII_SPEED_125		2
+#define GMAC_RGSMIII_SPEED_25		1
+#define GMAC_RGSMIII_SPEED_2_5		0
+#define GMAC_RGSMIII_LNKSTS		BIT(3)
+
 static enum ethtool_link_mode_bit_indices dwmac_hd_mode_bits[] = {
 	ETHTOOL_LINK_MODE_10baseT_Half_BIT,
 	ETHTOOL_LINK_MODE_100baseT_Half_BIT,
@@ -97,7 +107,7 @@ static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
 					   struct phylink_link_state *state)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
-	u32 status, lpa;
+	u32 status, lpa, rgsmii;
 
 	status = readl(spcs->base + GMAC_AN_STATUS);
 
@@ -111,7 +121,37 @@ static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
 
 		phylink_mii_c22_pcs_decode_state(state, neg_mode, status, lpa);
 	} else {
-		state->link = false;
+		rgsmii = field_get(spcs->rgsmii_status_mask,
+				   readl(spcs->rgsmii));
+
+		state->link = status & BMSR_LSTATUS &&
+			      rgsmii & GMAC_RGSMIII_LNKSTS;
+
+		if (state->link && neg_mode == PHYLINK_PCS_NEG_INBAND_ENABLED) {
+			/* FIXME: fill in speed and duplex. This requires the
+			 * contents of the dwmac1000 GMAC_RGSMIIIS or dwmac4
+			 * GMAC_PHYIF_CONTROL_STATUS register.
+			 */
+			state->duplex = rgsmii & GMAC_RGSMIII_LNKMOD ?
+					DUPLEX_FULL : DUPLEX_HALF;
+			switch (FIELD_GET(GMAC_RGSMIII_SPEED_MASK, rgsmii)) {
+			case GMAC_RGSMIII_SPEED_2_5:
+				state->speed = SPEED_10;
+				break;
+
+			case GMAC_RGSMIII_SPEED_25:
+				state->speed = SPEED_100;
+				break;
+
+			case GMAC_RGSMIII_SPEED_125:
+				state->speed = SPEED_1000;
+				break;
+
+			default:
+				state->link = false;
+				break;
+			}
+		}
 	}
 }
 
@@ -205,6 +245,8 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv,
 
 	spcs->priv = priv;
 	spcs->base = priv->ioaddr + pcs_info->pcs_offset;
+	spcs->rgsmii = priv->ioaddr + pcs_info->rgsmii_offset;
+	spcs->rgsmii_status_mask = pcs_info->rgsmii_status_mask;
 	spcs->int_mask = pcs_info->int_mask;
 	spcs->pcs.ops = &dwmac_integrated_pcs_ops;
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index a7c71f40f952..f9e7a7ed840b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -29,12 +29,16 @@ struct stmmac_priv;
 
 struct stmmac_pcs_info {
 	unsigned int pcs_offset;
+	unsigned int rgsmii_offset;
+	u32 rgsmii_status_mask;
 	u32 int_mask;
 };
 
 struct stmmac_pcs {
 	struct stmmac_priv *priv;
 	void __iomem *base;
+	void __iomem *rgsmii;
+	u32 rgsmii_status_mask;
 	u32 int_mask;
 	phy_interface_t interface;
 	struct phylink_pcs pcs;
-- 
2.47.3

Re: [net-next,12/14] net: stmmac: add support for reading inband SGMII status

[Jakub Kicinski]

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net: stmmac: add support for reading inband SGMII status

This patch adds support for reading link status, speed, and duplex for
SGMII links from the RGSMII status register. New fields are added to
stmmac_pcs_info and stmmac_pcs structures for the register offset and mask.

> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> index 73fc56ce5e55..9dd7e78cfbc4 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c

[ ... ]

> @@ -111,7 +121,37 @@ static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
>
>  		phylink_mii_c22_pcs_decode_state(state, neg_mode, status, lpa);
>  	} else {
> -		state->link = false;
> +		rgsmii = field_get(spcs->rgsmii_status_mask,
> +				   readl(spcs->rgsmii));
                                   ^^^^

When dwmac1000_pcs_init() or dwmac4_pcs_init() is used, the pcs_info
structures do not initialize the new rgsmii_offset and rgsmii_status_mask
fields. For example, dwmac4_pcs_info in dwmac4_core.c only sets:

    static const struct stmmac_pcs_info dwmac4_pcs_info = {
        .pcs_offset = GMAC_PCS_BASE,
        .int_mask = GMAC_INT_PCS_LINK | GMAC_INT_PCS_ANE,
    };

This leaves rgsmii_status_mask implicitly zero. Calling field_get() with a
mask of zero violates its API contract (mask must be non-zero) and results
in undefined behavior from __ffs(0).

Should the existing dwmac1000_pcs_info and dwmac4_pcs_info structures be
updated to provide proper values for the new fields? Or should there be a
guard check before calling field_get() when rgsmii_status_mask is zero?

> +
> +		state->link = status & BMSR_LSTATUS &&
> +			      rgsmii & GMAC_RGSMIII_LNKSTS;

[ ... ]

> @@ -205,6 +245,8 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv,
>
>  	spcs->priv = priv;
>  	spcs->base = priv->ioaddr + pcs_info->pcs_offset;
> +	spcs->rgsmii = priv->ioaddr + pcs_info->rgsmii_offset;
> +	spcs->rgsmii_status_mask = pcs_info->rgsmii_status_mask;

The values are copied from pcs_info here, but dwmac1000_pcs_info and
dwmac4_pcs_info do not initialize these fields.

> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> index a7c71f40f952..f9e7a7ed840b 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> @@ -29,12 +29,16 @@ struct stmmac_priv;
>
>  struct stmmac_pcs_info {
>  	unsigned int pcs_offset;
> +	unsigned int rgsmii_offset;
> +	u32 rgsmii_status_mask;
>  	u32 int_mask;
>  };

These new fields need to be initialized by all users of stmmac_pcs_info.

[PATCH net-next 13/14] net: stmmac: configure SGMII AN control according to phylink

[Russell King (Oracle)]

Provide phylink with the integrated PCS autonegotiation capabilities,
and configure the PCS's AN settings according to phylink's requested
requirements.

This may cause regressions.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  | 20 ++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 9dd7e78cfbc4..0426f608ebdf 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -65,7 +65,8 @@ static int dwmac_integrated_pcs_validate(struct phylink_pcs *pcs,
 static unsigned int dwmac_integrated_pcs_inband_caps(struct phylink_pcs *pcs,
 						     phy_interface_t interface)
 {
-	if (phy_interface_mode_is_8023z(interface))
+	if (phy_interface_mode_is_8023z(interface) ||
+	    interface == PHY_INTERFACE_MODE_SGMII)
 		return LINK_INBAND_ENABLE | LINK_INBAND_DISABLE;
 
 	return 0;
@@ -162,8 +163,9 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 				       bool permit_pause_to_mac)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
-	bool changed = false, ane = true;
-	u32 adv;
+	void __iomem *an_control = spcs->base + GMAC_AN_CTRL(0);
+	bool changed = false;
+	u32 adv, ctrl;
 	int ret;
 
 	if (spcs->interface != interface) {
@@ -180,12 +182,16 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 		if (readl(spcs->base + GMAC_ANE_ADV) != adv)
 			changed = true;
 		writel(adv, spcs->base + GMAC_ANE_ADV);
-
-		ane = neg_mode == PHYLINK_PCS_NEG_INBAND_ENABLED;
 	}
 
-	dwmac_ctrl_ane(spcs->base, 0, ane,
-		       spcs->priv->hw->reverse_sgmii_enable);
+	ctrl = readl(an_control) & ~(GMAC_AN_CTRL_ANE | GMAC_AN_CTRL_SGMRAL);
+	if (spcs->priv->hw->reverse_sgmii_enable)
+		ctrl |= GMAC_AN_CTRL_SGMRAL | GMAC_AN_CTRL_ANE;
+	else if (neg_mode == PHYLINK_PCS_NEG_INBAND_ENABLED)
+		ctrl |= GMAC_AN_CTRL_ANE;
+	else
+		ctrl |= GMAC_AN_CTRL_SGMRAL;
+	writel(ctrl, an_control);
 
 	return changed;
 }
-- 
2.47.3

[PATCH net-next 14/14] net: stmmac: report PCS configuration changes

[Russell King (Oracle)]

Report if/when qcom-ethqos changes the PCS configuration. With phylink
now setting the PCS configuration, there should be no need for drivers
to change this.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index f9e7a7ed840b..6a1e30b10740 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -71,6 +71,7 @@ static inline void dwmac_ctrl_ane(void __iomem *ioaddr, u32 reg, bool ane,
 				  bool srgmi_ral)
 {
 	u32 value = readl(ioaddr + GMAC_AN_CTRL(reg));
+	u32 old = value, diff;
 
 	/* Enable and restart the Auto-Negotiation */
 	if (ane)
@@ -84,6 +85,20 @@ static inline void dwmac_ctrl_ane(void __iomem *ioaddr, u32 reg, bool ane,
 	if (srgmi_ral)
 		value |= GMAC_AN_CTRL_SGMRAL;
 
+	diff = old ^ value;
+	if (diff & ~GMAC_AN_CTRL_RAN) {
+		pr_warn("dwmac: PCS configuration changed from phylink by glue, please report: 0x%08x -> 0x%08x\n",
+			old & ~GMAC_AN_CTRL_RAN, value & ~GMAC_AN_CTRL_RAN);
+#define REPORT_BIT(x) \
+		if (diff & GMAC_AN_CTRL_##x) \
+			pr_warn("dwmac: %8s %u -> %u\n", #x, \
+				!!(old & GMAC_AN_CTRL_##x), \
+				!!(value & GMAC_AN_CTRL_##x))
+		REPORT_BIT(ANE);
+		REPORT_BIT(SGMRAL);
+#undef REPORT_BIT
+	}
+
 	writel(value, ioaddr + GMAC_AN_CTRL(reg));
 }
 #endif /* __STMMAC_PCS_H__ */
-- 
2.47.3

Re: [PATCH net-next 14/14] net: stmmac: report PCS configuration changes

[Russell King (Oracle)]

On Mon, Jan 19, 2026 at 12:34:47PM +0000, Russell King (Oracle) wrote:
> Report if/when qcom-ethqos changes the PCS configuration. With phylink
> now setting the PCS configuration, there should be no need for drivers
> to change this.
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> ---
>  drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
> 
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> index f9e7a7ed840b..6a1e30b10740 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> @@ -71,6 +71,7 @@ static inline void dwmac_ctrl_ane(void __iomem *ioaddr, u32 reg, bool ane,
>  				  bool srgmi_ral)
>  {
>  	u32 value = readl(ioaddr + GMAC_AN_CTRL(reg));
> +	u32 old = value, diff;
>  
>  	/* Enable and restart the Auto-Negotiation */
>  	if (ane)
> @@ -84,6 +85,20 @@ static inline void dwmac_ctrl_ane(void __iomem *ioaddr, u32 reg, bool ane,
>  	if (srgmi_ral)
>  		value |= GMAC_AN_CTRL_SGMRAL;
>  
> +	diff = old ^ value;
> +	if (diff & ~GMAC_AN_CTRL_RAN) {
> +		pr_warn("dwmac: PCS configuration changed from phylink by glue, please report: 0x%08x -> 0x%08x\n",
> +			old & ~GMAC_AN_CTRL_RAN, value & ~GMAC_AN_CTRL_RAN);
> +#define REPORT_BIT(x) \
> +		if (diff & GMAC_AN_CTRL_##x) \
> +			pr_warn("dwmac: %8s %u -> %u\n", #x, \
> +				!!(old & GMAC_AN_CTRL_##x), \
> +				!!(value & GMAC_AN_CTRL_##x))
> +		REPORT_BIT(ANE);
> +		REPORT_BIT(SGMRAL);
> +#undef REPORT_BIT

I notice in patchwork, checkpatch doesn't like the "if" starting the in
the macro definition. While that would be true of definitions in kernel
header files which can be used anwhere, for such a simple limited usage
here, wrapping the thing in do { } while(0) is needless baggage, over-
complicates the code, and makes it less readable. So, IMHO, checkpatch
is wrong.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

[PATCH RFC net-next 00/14] net: stmmac: SerDes, PCS, BASE-X, and inband goodies

[Russell King (Oracle)]

Mentioned previously, I've been trying to sort out the PCS support in
stmmac, and this series represents the current state of play.

Previous posted patches centred around merely getting autonegotiation
to be configured correctly, to a point where the manual configuration
can be removed from the qcom-ethqos driver. The qcom-ethqos driver
uses both SGMII and 2500BASE-X, manually configuring the dwmac's
integrated PCS appropriately.

This *untested* series attempts to take this further. The patches:

- clean up qcom-ethqos only-written mac_base member.
- convert qcom-ethqos to use the set_clk_tx_rate() method for setting
  the link clock rate.
- add support for phy_set_mode_ext() to the qcom "SGMII" ethernet
  SerDes driver (which is really only what it needs. Note that
  phy_set_mode_ext() is an expected call to be made, where as
  phy_set_speed() is optional and not. See PHY documentation.)
- add platform-glue independent SerDes support to the stmmac core
  driver. Currently, only qcom-ethqos will make use of this, and
  I suspect as we haven't had this, it's going to be difficult to
  convert other platform glue to use this - but had this existed
  earlier, we could've pushed people to use PHY to abstract some
  of the platform glue differences. Adding it now makes it available
  for future platform glue.
- convert qcom-ethqos to use this core SerDes support.
- arrange for stmmac_pcs.c to supply the phy_intf_sel field value
  if the integrated PCS will be used. (PHY_INTF_SEL_SGMII requires
  the integrated PCS rather than an external PCS.)
- add BASE-X support to the integrated PCS driver, and use it for
  BASE-X modes. This fully supports in-band mode, including reading
  the link partner advertisement.
- add in-band support for SGMII, reading the state from the RGSMII
  status field.

As we leave qcom-ethqos' manual configuration of the PCS in place at
the moment, the last patch adds reporting of any changes in its
configuration that the qcom-ethqos driver does beyond what phylink
requested, thus providing a path to debug and eventually remove
qcom-ethqos' manual configuration.

One patch is not included in this set - which adds a phy_intf_sel
value for external PCS (using PHY_INTF_SEL_GMII_MII). I believe all
external PCS use this mode when connected to a MAC capable of up to
2.5G. However, no platform glue that provides the mac_select_pcs()
method also provide the set_phy_intf_sel() method, so we can safely
ignore this for now.

I would like to get this into net-next before the next merge window,
so testing would be appreciated. If there are issues with these patches
applied, please check whether the issue exists without these patches
and only report regressions caused by this patch set. For example,
I'm aware that qcom-ethqos has issues with 10Mbps mode due to an AQR
PHY being insanely provisioned to use SGMII in 1000M mode but with
rate matching with 10M media. This is not an issue that is relevant
to this patch series, but a problem with the PHY provisioning.

 drivers/net/ethernet/stmicro/stmmac/Makefile       |   2 +-
 drivers/net/ethernet/stmicro/stmmac/common.h       |   1 -
 .../ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c    |  74 ++-----
 .../net/ethernet/stmicro/stmmac/dwmac1000_core.c   |   9 +-
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c  |   8 +-
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c  |  69 +++++--
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c   | 218 +++++++++++++++++++--
 drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h   |  53 ++---
 .../net/ethernet/stmicro/stmmac/stmmac_serdes.c    | 111 +++++++++++
 .../net/ethernet/stmicro/stmmac/stmmac_serdes.h    |  16 ++
 drivers/phy/qualcomm/phy-qcom-sgmii-eth.c          |  42 ++++
 include/linux/stmmac.h                             |   2 +
 12 files changed, 480 insertions(+), 125 deletions(-)

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

[PATCH net-next 05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

Rather than having platform glue implement SerDes PHY support, add it
to the core driver, specifically to the stmmac integrated PCS driver
as the SerDes is connected to the integrated PCS.

Platforms using external PCS can also populate plat->serdes, and the
core driver will call phy_init() and phy_exit() when the administrative
state of the interface changes, but the other phy methods will not be
called.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/Makefile  |   2 +-
 .../net/ethernet/stmicro/stmmac/stmmac_main.c |  14 ++-
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  |  38 +++++-
 .../net/ethernet/stmicro/stmmac/stmmac_pcs.h  |   1 +
 .../ethernet/stmicro/stmmac/stmmac_serdes.c   | 111 ++++++++++++++++++
 .../ethernet/stmicro/stmmac/stmmac_serdes.h   |  16 +++
 include/linux/stmmac.h                        |   2 +
 7 files changed, 180 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h

diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index c9263987ef8d..a3c2cd5d0c91 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -7,7 +7,7 @@ stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o	\
 	      dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \
 	      stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \
 	      stmmac_xdp.o stmmac_est.o stmmac_fpe.o stmmac_vlan.o \
-	      stmmac_pcs.o $(stmmac-y)
+	      stmmac_pcs.o stmmac_serdes.o $(stmmac-y)
 
 stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 24a2555ca329..6c515f9efbe7 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -48,6 +48,7 @@
 #include "stmmac_fpe.h"
 #include "stmmac.h"
 #include "stmmac_pcs.h"
+#include "stmmac_serdes.h"
 #include "stmmac_xdp.h"
 #include <linux/reset.h>
 #include <linux/of_mdio.h>
@@ -3549,12 +3550,16 @@ static void stmmac_safety_feat_configuration(struct stmmac_priv *priv)
 
 static void stmmac_clk_rx_i_require(struct stmmac_priv *priv)
 {
+	dwmac_serdes_power_on(priv);
+	/* Only sets the SerDes mode if it wasn't already configured. */
+	dwmac_serdes_init_mode(priv, priv->plat->phy_interface);
 	phylink_rx_clk_stop_block(priv->phylink);
 }
 
 static void stmmac_clk_rx_i_release(struct stmmac_priv *priv)
 {
 	phylink_rx_clk_stop_unblock(priv->phylink);
+	dwmac_serdes_power_off(priv);
 }
 
 /**
@@ -4152,10 +4157,14 @@ static int stmmac_open(struct net_device *dev)
 	if (ret)
 		goto err_runtime_pm;
 
-	ret = __stmmac_open(dev, dma_conf);
+	ret = dwmac_serdes_init(priv);
 	if (ret)
 		goto err_disconnect_phy;
 
+	ret = __stmmac_open(dev, dma_conf);
+	if (ret)
+		goto err_serdes;
+
 	kfree(dma_conf);
 
 	/* We may have called phylink_speed_down before */
@@ -4163,6 +4172,8 @@ static int stmmac_open(struct net_device *dev)
 
 	return ret;
 
+err_serdes:
+	dwmac_serdes_exit(priv);
 err_disconnect_phy:
 	phylink_disconnect_phy(priv->phylink);
 err_runtime_pm:
@@ -4226,6 +4237,7 @@ static int stmmac_release(struct net_device *dev)
 
 	__stmmac_release(dev);
 
+	dwmac_serdes_exit(priv);
 	phylink_disconnect_phy(priv->phylink);
 	pm_runtime_put(priv->device);
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
index 2f826fe7229b..4d1902f3a58f 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
@@ -1,12 +1,25 @@
 // SPDX-License-Identifier: GPL-2.0-only
 #include "stmmac.h"
 #include "stmmac_pcs.h"
+#include "stmmac_serdes.h"
 
 static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	struct stmmac_priv *priv = spcs->priv;
+	int ret;
 
-	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
+	ret = dwmac_serdes_power_on(priv);
+	if (ret)
+		return ret;
+
+	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
+		ret = dwmac_serdes_set_mode(priv, spcs->interface);
+		if (ret)
+			return ret;
+	}
+
+	stmmac_mac_irq_modify(priv, 0, spcs->int_mask);
 
 	return 0;
 }
@@ -14,8 +27,11 @@ static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
 static void dwmac_integrated_pcs_disable(struct phylink_pcs *pcs)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	struct stmmac_priv *priv = spcs->priv;
 
-	stmmac_mac_irq_modify(spcs->priv, spcs->int_mask, 0);
+	stmmac_mac_irq_modify(priv, spcs->int_mask, 0);
+
+	dwmac_serdes_power_off(priv);
 }
 
 static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
@@ -32,6 +48,15 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
 				       bool permit_pause_to_mac)
 {
 	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
+	int ret;
+
+	if (spcs->interface != interface) {
+		ret = dwmac_serdes_set_mode(spcs->priv, interface);
+		if (ret)
+			return ret;
+
+		spcs->interface = interface;
+	}
 
 	dwmac_ctrl_ane(spcs->base, 0, 1, spcs->priv->hw->reverse_sgmii_enable);
 
@@ -71,6 +96,7 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 			       u32 int_mask)
 {
 	struct stmmac_pcs *spcs;
+	int ret;
 
 	spcs = devm_kzalloc(priv->device, sizeof(*spcs), GFP_KERNEL);
 	if (!spcs)
@@ -81,6 +107,14 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
 	spcs->int_mask = int_mask;
 	spcs->pcs.ops = &dwmac_integrated_pcs_ops;
 
+	if (priv->plat->serdes) {
+		ret = dwmac_serdes_validate(priv, PHY_INTERFACE_MODE_SGMII);
+		if (ret)
+			dev_warn(priv->device,
+				 "serdes does not support SGMII: %pe\n",
+				 ERR_PTR(ret));
+	}
+
 	__set_bit(PHY_INTERFACE_MODE_SGMII, spcs->pcs.supported_interfaces);
 
 	priv->integrated_pcs = spcs;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
index c4e6b242d390..36bf75fdf478 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
@@ -53,6 +53,7 @@ struct stmmac_pcs {
 	struct stmmac_priv *priv;
 	void __iomem *base;
 	u32 int_mask;
+	phy_interface_t interface;
 	struct phylink_pcs pcs;
 };
 
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
new file mode 100644
index 000000000000..3003e1ae38d2
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
@@ -0,0 +1,111 @@
+#include <linux/phy/phy.h>
+
+#include "stmmac.h"
+#include "stmmac_serdes.h"
+
+static phy_interface_t dwmac_serdes_phy_modes[] = {
+	PHY_INTERFACE_MODE_SGMII,
+	PHY_INTERFACE_MODE_1000BASEX,
+	PHY_INTERFACE_MODE_2500BASEX
+};
+
+int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	return phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET, interface,
+			    NULL);
+}
+
+int dwmac_serdes_init(struct stmmac_priv *priv)
+{
+	size_t i;
+	int ret;
+
+	if (!priv->plat->serdes)
+		return 0;
+
+	/* Encourage good implementation of the SerDes PHY driver, so that
+	 * we can discover which Ethernet modes the SerDes supports.
+	 * Unfortunately, some implementations are noisy (bad), others
+	 * require phy_set_speed() to select the correct speed first
+	 * (which then reprograms the SerDes, negating the whole point of
+	 * phy_validate().) Weed out these incompatible implementations.
+	 */
+	for (i = 0; i < ARRAY_SIZE(dwmac_serdes_phy_modes); i++) {
+		ret = phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET,
+				   dwmac_serdes_phy_modes[i], NULL);
+		if (ret == 0 || ret == -EOPNOTSUPP)
+			break;
+	}
+
+	if (ret == -EOPNOTSUPP)
+		dev_warn(priv->device,
+			 "SerDes driver does not implement phy_validate()\n");
+	if (ret) {
+		/* The SerDes PHY failed validation, refuse to use it. */
+		dev_warn(priv->device,
+			 "SerDes driver fails to validate SGMII, 1000BASE-X nor 2500BASE-X\n");
+		return -EINVAL;
+	}
+
+	ret = phy_init(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to initialize SerDes: %pe\n",
+			ERR_PTR(ret));
+
+	return ret;
+}
+
+int dwmac_serdes_power_on(struct stmmac_priv *priv)
+{
+	int ret;
+
+	ret = phy_power_on(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to power on SerDes: %pe\n",
+			ERR_PTR(ret));
+
+	return ret;
+}
+
+int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	struct phy *serdes = priv->plat->serdes;
+
+	if (phy_get_mode(serdes) == PHY_MODE_ETHERNET)
+		return 0;
+
+	return dwmac_serdes_set_mode(priv, interface);
+}
+
+int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface)
+{
+	struct phy *serdes = priv->plat->serdes;
+	int ret;
+
+	ret = phy_set_mode_ext(serdes, PHY_MODE_ETHERNET, interface);
+	if (ret)
+		dev_err(priv->device,
+			"failed to set SerDes mode %s: %pe\n",
+			phy_modes(interface), ERR_PTR(ret));
+
+	return ret;
+}
+
+void dwmac_serdes_power_off(struct stmmac_priv *priv)
+{
+	int ret;
+
+	ret = phy_power_off(priv->plat->serdes);
+	if (ret)
+		dev_err(priv->device, "failed to power off SerDes: %pe\n",
+			ERR_PTR(ret));
+}
+
+void dwmac_serdes_exit(struct stmmac_priv *priv)
+{
+	int ret = phy_exit(priv->plat->serdes);
+
+	if (ret)
+		dev_err(priv->device, "failed to shutdown SerDes: %pe\n",
+			ERR_PTR(ret));
+}
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
new file mode 100644
index 000000000000..a31e6c9e0570
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
@@ -0,0 +1,16 @@
+#ifndef STMMAC_SERDES_H
+#define STMMAC_SERDES_H
+
+#include <linux/phy.h>
+
+struct stmmac_priv;
+
+int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface);
+int dwmac_serdes_init(struct stmmac_priv *priv);
+int dwmac_serdes_power_on(struct stmmac_priv *priv);
+int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface);
+int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface);
+void dwmac_serdes_power_off(struct stmmac_priv *priv);
+void dwmac_serdes_exit(struct stmmac_priv *priv);
+
+#endif
diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
index f1054b9c2d8a..4db506e5cf13 100644
--- a/include/linux/stmmac.h
+++ b/include/linux/stmmac.h
@@ -193,6 +193,7 @@ enum dwmac_core_type {
 #define STMMAC_FLAG_HWTSTAMP_CORRECT_LATENCY	BIT(13)
 
 struct mac_device_info;
+struct phy;
 
 struct plat_stmmacenet_data {
 	enum dwmac_core_type core_type;
@@ -222,6 +223,7 @@ struct plat_stmmacenet_data {
 	 * that phylink uses.
 	 */
 	phy_interface_t phy_interface;
+	struct phy *serdes;
 	struct stmmac_mdio_bus_data *mdio_bus_data;
 	struct device_node *phy_node;
 	struct fwnode_handle *port_node;
-- 
2.47.3

Re: [PATCH net-next 05/14] net: stmmac: add stmmac core serdes support

[Maxime Chevallier]

Hi Russell,

On 14/01/2026 18:45, Russell King (Oracle) wrote:
> Rather than having platform glue implement SerDes PHY support, add it
> to the core driver, specifically to the stmmac integrated PCS driver
> as the SerDes is connected to the integrated PCS.
> 
> Platforms using external PCS can also populate plat->serdes, and the
> core driver will call phy_init() and phy_exit() when the administrative
> state of the interface changes, but the other phy methods will not be
> called.
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Unfortunately I have no way to test. But still,

Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>

Maxime

> ---
>  drivers/net/ethernet/stmicro/stmmac/Makefile  |   2 +-
>  .../net/ethernet/stmicro/stmmac/stmmac_main.c |  14 ++-
>  .../net/ethernet/stmicro/stmmac/stmmac_pcs.c  |  38 +++++-
>  .../net/ethernet/stmicro/stmmac/stmmac_pcs.h  |   1 +
>  .../ethernet/stmicro/stmmac/stmmac_serdes.c   | 111 ++++++++++++++++++
>  .../ethernet/stmicro/stmmac/stmmac_serdes.h   |  16 +++
>  include/linux/stmmac.h                        |   2 +
>  7 files changed, 180 insertions(+), 4 deletions(-)
>  create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
>  create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
> 
> diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
> index c9263987ef8d..a3c2cd5d0c91 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/Makefile
> +++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
> @@ -7,7 +7,7 @@ stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o	\
>  	      dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \
>  	      stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \
>  	      stmmac_xdp.o stmmac_est.o stmmac_fpe.o stmmac_vlan.o \
> -	      stmmac_pcs.o $(stmmac-y)
> +	      stmmac_pcs.o stmmac_serdes.o $(stmmac-y)
>  
>  stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o
>  
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index 24a2555ca329..6c515f9efbe7 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -48,6 +48,7 @@
>  #include "stmmac_fpe.h"
>  #include "stmmac.h"
>  #include "stmmac_pcs.h"
> +#include "stmmac_serdes.h"
>  #include "stmmac_xdp.h"
>  #include <linux/reset.h>
>  #include <linux/of_mdio.h>
> @@ -3549,12 +3550,16 @@ static void stmmac_safety_feat_configuration(struct stmmac_priv *priv)
>  
>  static void stmmac_clk_rx_i_require(struct stmmac_priv *priv)
>  {
> +	dwmac_serdes_power_on(priv);
> +	/* Only sets the SerDes mode if it wasn't already configured. */
> +	dwmac_serdes_init_mode(priv, priv->plat->phy_interface);
>  	phylink_rx_clk_stop_block(priv->phylink);
>  }
>  
>  static void stmmac_clk_rx_i_release(struct stmmac_priv *priv)
>  {
>  	phylink_rx_clk_stop_unblock(priv->phylink);
> +	dwmac_serdes_power_off(priv);
>  }
>  
>  /**
> @@ -4152,10 +4157,14 @@ static int stmmac_open(struct net_device *dev)
>  	if (ret)
>  		goto err_runtime_pm;
>  
> -	ret = __stmmac_open(dev, dma_conf);
> +	ret = dwmac_serdes_init(priv);
>  	if (ret)
>  		goto err_disconnect_phy;
>  
> +	ret = __stmmac_open(dev, dma_conf);
> +	if (ret)
> +		goto err_serdes;
> +
>  	kfree(dma_conf);
>  
>  	/* We may have called phylink_speed_down before */
> @@ -4163,6 +4172,8 @@ static int stmmac_open(struct net_device *dev)
>  
>  	return ret;
>  
> +err_serdes:
> +	dwmac_serdes_exit(priv);
>  err_disconnect_phy:
>  	phylink_disconnect_phy(priv->phylink);
>  err_runtime_pm:
> @@ -4226,6 +4237,7 @@ static int stmmac_release(struct net_device *dev)
>  
>  	__stmmac_release(dev);
>  
> +	dwmac_serdes_exit(priv);
>  	phylink_disconnect_phy(priv->phylink);
>  	pm_runtime_put(priv->device);
>  
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> index 2f826fe7229b..4d1902f3a58f 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.c
> @@ -1,12 +1,25 @@
>  // SPDX-License-Identifier: GPL-2.0-only
>  #include "stmmac.h"
>  #include "stmmac_pcs.h"
> +#include "stmmac_serdes.h"
>  
>  static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
>  {
>  	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> +	struct stmmac_priv *priv = spcs->priv;
> +	int ret;
>  
> -	stmmac_mac_irq_modify(spcs->priv, 0, spcs->int_mask);
> +	ret = dwmac_serdes_power_on(priv);
> +	if (ret)
> +		return ret;
> +
> +	if (spcs->interface != PHY_INTERFACE_MODE_NA) {
> +		ret = dwmac_serdes_set_mode(priv, spcs->interface);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	stmmac_mac_irq_modify(priv, 0, spcs->int_mask);
>  
>  	return 0;
>  }
> @@ -14,8 +27,11 @@ static int dwmac_integrated_pcs_enable(struct phylink_pcs *pcs)
>  static void dwmac_integrated_pcs_disable(struct phylink_pcs *pcs)
>  {
>  	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> +	struct stmmac_priv *priv = spcs->priv;
>  
> -	stmmac_mac_irq_modify(spcs->priv, spcs->int_mask, 0);
> +	stmmac_mac_irq_modify(priv, spcs->int_mask, 0);
> +
> +	dwmac_serdes_power_off(priv);
>  }
>  
>  static void dwmac_integrated_pcs_get_state(struct phylink_pcs *pcs,
> @@ -32,6 +48,15 @@ static int dwmac_integrated_pcs_config(struct phylink_pcs *pcs,
>  				       bool permit_pause_to_mac)
>  {
>  	struct stmmac_pcs *spcs = phylink_pcs_to_stmmac_pcs(pcs);
> +	int ret;
> +
> +	if (spcs->interface != interface) {
> +		ret = dwmac_serdes_set_mode(spcs->priv, interface);
> +		if (ret)
> +			return ret;
> +
> +		spcs->interface = interface;
> +	}
>  
>  	dwmac_ctrl_ane(spcs->base, 0, 1, spcs->priv->hw->reverse_sgmii_enable);
>  
> @@ -71,6 +96,7 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
>  			       u32 int_mask)
>  {
>  	struct stmmac_pcs *spcs;
> +	int ret;
>  
>  	spcs = devm_kzalloc(priv->device, sizeof(*spcs), GFP_KERNEL);
>  	if (!spcs)
> @@ -81,6 +107,14 @@ int stmmac_integrated_pcs_init(struct stmmac_priv *priv, unsigned int offset,
>  	spcs->int_mask = int_mask;
>  	spcs->pcs.ops = &dwmac_integrated_pcs_ops;
>  
> +	if (priv->plat->serdes) {
> +		ret = dwmac_serdes_validate(priv, PHY_INTERFACE_MODE_SGMII);
> +		if (ret)
> +			dev_warn(priv->device,
> +				 "serdes does not support SGMII: %pe\n",
> +				 ERR_PTR(ret));
> +	}
> +
>  	__set_bit(PHY_INTERFACE_MODE_SGMII, spcs->pcs.supported_interfaces);
>  
>  	priv->integrated_pcs = spcs;
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> index c4e6b242d390..36bf75fdf478 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_pcs.h
> @@ -53,6 +53,7 @@ struct stmmac_pcs {
>  	struct stmmac_priv *priv;
>  	void __iomem *base;
>  	u32 int_mask;
> +	phy_interface_t interface;
>  	struct phylink_pcs pcs;
>  };
>  
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
> new file mode 100644
> index 000000000000..3003e1ae38d2
> --- /dev/null
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.c
> @@ -0,0 +1,111 @@
> +#include <linux/phy/phy.h>
> +
> +#include "stmmac.h"
> +#include "stmmac_serdes.h"
> +
> +static phy_interface_t dwmac_serdes_phy_modes[] = {
> +	PHY_INTERFACE_MODE_SGMII,
> +	PHY_INTERFACE_MODE_1000BASEX,
> +	PHY_INTERFACE_MODE_2500BASEX
> +};
> +
> +int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface)
> +{
> +	return phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET, interface,
> +			    NULL);
> +}
> +
> +int dwmac_serdes_init(struct stmmac_priv *priv)
> +{
> +	size_t i;
> +	int ret;
> +
> +	if (!priv->plat->serdes)
> +		return 0;
> +
> +	/* Encourage good implementation of the SerDes PHY driver, so that
> +	 * we can discover which Ethernet modes the SerDes supports.
> +	 * Unfortunately, some implementations are noisy (bad), others
> +	 * require phy_set_speed() to select the correct speed first
> +	 * (which then reprograms the SerDes, negating the whole point of
> +	 * phy_validate().) Weed out these incompatible implementations.
> +	 */
> +	for (i = 0; i < ARRAY_SIZE(dwmac_serdes_phy_modes); i++) {
> +		ret = phy_validate(priv->plat->serdes, PHY_MODE_ETHERNET,
> +				   dwmac_serdes_phy_modes[i], NULL);
> +		if (ret == 0 || ret == -EOPNOTSUPP)
> +			break;
> +	}
> +
> +	if (ret == -EOPNOTSUPP)
> +		dev_warn(priv->device,
> +			 "SerDes driver does not implement phy_validate()\n");
> +	if (ret) {
> +		/* The SerDes PHY failed validation, refuse to use it. */
> +		dev_warn(priv->device,
> +			 "SerDes driver fails to validate SGMII, 1000BASE-X nor 2500BASE-X\n");
> +		return -EINVAL;
> +	}
> +
> +	ret = phy_init(priv->plat->serdes);
> +	if (ret)
> +		dev_err(priv->device, "failed to initialize SerDes: %pe\n",
> +			ERR_PTR(ret));
> +
> +	return ret;
> +}
> +
> +int dwmac_serdes_power_on(struct stmmac_priv *priv)
> +{
> +	int ret;
> +
> +	ret = phy_power_on(priv->plat->serdes);
> +	if (ret)
> +		dev_err(priv->device, "failed to power on SerDes: %pe\n",
> +			ERR_PTR(ret));
> +
> +	return ret;
> +}
> +
> +int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface)
> +{
> +	struct phy *serdes = priv->plat->serdes;
> +
> +	if (phy_get_mode(serdes) == PHY_MODE_ETHERNET)
> +		return 0;
> +
> +	return dwmac_serdes_set_mode(priv, interface);
> +}
> +
> +int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface)
> +{
> +	struct phy *serdes = priv->plat->serdes;
> +	int ret;
> +
> +	ret = phy_set_mode_ext(serdes, PHY_MODE_ETHERNET, interface);
> +	if (ret)
> +		dev_err(priv->device,
> +			"failed to set SerDes mode %s: %pe\n",
> +			phy_modes(interface), ERR_PTR(ret));
> +
> +	return ret;
> +}
> +
> +void dwmac_serdes_power_off(struct stmmac_priv *priv)
> +{
> +	int ret;
> +
> +	ret = phy_power_off(priv->plat->serdes);
> +	if (ret)
> +		dev_err(priv->device, "failed to power off SerDes: %pe\n",
> +			ERR_PTR(ret));
> +}
> +
> +void dwmac_serdes_exit(struct stmmac_priv *priv)
> +{
> +	int ret = phy_exit(priv->plat->serdes);
> +
> +	if (ret)
> +		dev_err(priv->device, "failed to shutdown SerDes: %pe\n",
> +			ERR_PTR(ret));
> +}
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
> new file mode 100644
> index 000000000000..a31e6c9e0570
> --- /dev/null
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_serdes.h
> @@ -0,0 +1,16 @@
> +#ifndef STMMAC_SERDES_H
> +#define STMMAC_SERDES_H
> +
> +#include <linux/phy.h>
> +
> +struct stmmac_priv;
> +
> +int dwmac_serdes_validate(struct stmmac_priv *priv, phy_interface_t interface);
> +int dwmac_serdes_init(struct stmmac_priv *priv);
> +int dwmac_serdes_power_on(struct stmmac_priv *priv);
> +int dwmac_serdes_init_mode(struct stmmac_priv *priv, phy_interface_t interface);
> +int dwmac_serdes_set_mode(struct stmmac_priv *priv, phy_interface_t interface);
> +void dwmac_serdes_power_off(struct stmmac_priv *priv);
> +void dwmac_serdes_exit(struct stmmac_priv *priv);
> +
> +#endif
> diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
> index f1054b9c2d8a..4db506e5cf13 100644
> --- a/include/linux/stmmac.h
> +++ b/include/linux/stmmac.h
> @@ -193,6 +193,7 @@ enum dwmac_core_type {
>  #define STMMAC_FLAG_HWTSTAMP_CORRECT_LATENCY	BIT(13)
>  
>  struct mac_device_info;
> +struct phy;
>  
>  struct plat_stmmacenet_data {
>  	enum dwmac_core_type core_type;
> @@ -222,6 +223,7 @@ struct plat_stmmacenet_data {
>  	 * that phylink uses.
>  	 */
>  	phy_interface_t phy_interface;
> +	struct phy *serdes;
>  	struct stmmac_mdio_bus_data *mdio_bus_data;
>  	struct device_node *phy_node;
>  	struct fwnode_handle *port_node;

Re: [PATCH net-next 05/14] net: stmmac: add stmmac core serdes support

[Russell King (Oracle)]

On Thu, Jan 15, 2026 at 03:48:40PM +0100, Maxime Chevallier wrote:
> Hi Russell,
> 
> On 14/01/2026 18:45, Russell King (Oracle) wrote:
> > Rather than having platform glue implement SerDes PHY support, add it
> > to the core driver, specifically to the stmmac integrated PCS driver
> > as the SerDes is connected to the integrated PCS.
> > 
> > Platforms using external PCS can also populate plat->serdes, and the
> > core driver will call phy_init() and phy_exit() when the administrative
> > state of the interface changes, but the other phy methods will not be
> > called.
> > 
> > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
> 
> Unfortunately I have no way to test. But still,

I am hoping Mohd Ayaan Anwar will be able to do at least some testing
on the qcom-ethqos hardware.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!