Skip to contentSkip to conversation

net: phy: fix duplicate eth_phy binding

PATCH Superseded Pranav Tilak 2 participants 4 messages Apr 8 – Apr 21

Subsystems (2)

v2 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 net: phy: fix duplicate eth_phy binding

When both CONFIG_PHY_ETHERNET_ID and CONFIG_DM_ETH_PHY are enabled, eth_phy_binds_nodes() called from eth_post_bind() already binds the ethernet PHY node to eth_phy_generic_drv. However, phy_connect_phy_id() called via zynq_phy_init() -> phy_connect() cannot attach to the already bound PHY device and creates a second instance, resulting in duplicate entries in the dm tree:

Fix this by guarding the phy_connect_phy_id() call in phy_connect() with uclass_get_device_by_phandle() to skip re-binding if the PHY node is already registered in UCLASS_ETH_PHY.

Fixes 68a4d1506109 ("net: phy: Bind ETH_PHY uclass driver to each new PHY")

---
Changes in v2:
- Move duplicate binding check to caller phy_connect()
- Update commit description

 drivers/net/phy/phy.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/net/phy/phy.c +7 -2
index d7e0c4fe02d..cfca8328f1c 100644
@@ -938,8 +938,13 @@ struct phy_device *phy_connect(struct mii_dev *bus, int addr,
 #endif
 
 #ifdef CONFIG_PHY_ETHERNET_ID
-	if (!phydev)
-		phydev = phy_connect_phy_id(bus, dev, addr);
+	if (!phydev) {
+		struct udevice *phy_dev;
+
+		if (uclass_get_device_by_phandle(UCLASS_ETH_PHY, dev,
+						 "phy-handle", &phy_dev))
+			phydev = phy_connect_phy_id(bus, dev, addr);
+	}
 #endif
 
 #ifdef CONFIG_PHY_XILINX_GMII2RGMII
Signature · 1 line
2.34.1
#1 lore raw
Quoted 27 lines · Pranav Tilak

When both CONFIG_PHY_ETHERNET_ID and CONFIG_DM_ETH_PHY are enabled, eth_phy_binds_nodes() called from eth_post_bind() already binds the ethernet PHY node to eth_phy_generic_drv. However, phy_connect_phy_id() called via zynq_phy_init() -> phy_connect() cannot attach to the already bound PHY device and creates a second instance, resulting in duplicate entries in the dm tree:

Fix this by guarding the phy_connect_phy_id() call in phy_connect() with uclass_get_device_by_phandle() to skip re-binding if the PHY node is already registered in UCLASS_ETH_PHY.

Fixes 68a4d1506109 ("net: phy: Bind ETH_PHY uclass driver to each new PHY")

---
Changes in v2:
- Move duplicate binding check to caller phy_connect()
- Update commit description

  drivers/net/phy/phy.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)
drivers/net/phy/phy.c +0 -0
index d7e0c4fe02d..cfca8328f1c 100644
@@ -938,8 +938,13 @@ struct phy_device *phy_connect(struct mii_dev *bus, int addr,
  #endif

drivers/net/phy/phy.c struct phy_device *phy_connect(struct mii_dev *bus, int addr, · line 946

Quoted 10 lines
drivers/net/phy/phy.c +6 -2
 
  #ifdef CONFIG_PHY_ETHERNET_ID
-	if (!phydev)
-		phydev = phy_connect_phy_id(bus, dev, addr);
+	if (!phydev) {
+		struct udevice *phy_dev;
+
+		if (uclass_get_device_by_phandle(UCLASS_ETH_PHY, dev,
+						 "phy-handle", &phy_dev))
+			phydev = phy_connect_phy_id(bus, dev, addr);

Shouldn't this binding check be closer to the beginning of phy_connect() ? Basically check whether the PHY driver is already bound at the beginning, and if so, you are done. I don't think that test is specific to CONFIG_PHY_ETHERNET_ID, is it ?

#2 lore raw

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Marek,

Quoted 55 lines

-----Original Message-----
From: Marek Vasut <[email protected]>
Sent: Wednesday, April 8, 2026 10:02 PM
To: Vinay Tilak, Pranav <[email protected]>; [email protected]; Simek, Michal <[email protected]>
Cc: git (AMD-Xilinx) <[email protected]>; Begari, Padmarao <[email protected]>; Jerome Forissier <[email protected]>; Tom Rini <[email protected]>; Peng Fan <[email protected]>; Lucien.Jheng <[email protected]>; Yao Zi <[email protected]>; SkyLake.Huang <[email protected]>; Siddharth Vadapalli <[email protected]>; Marek Vasut <[email protected]>; Ramon Fried <[email protected]>
Subject: Re: [PATCH v2] net: phy: fix duplicate eth_phy binding

Quoted 1 line · Pranav Tilak

When both CONFIG_PHY_ETHERNET_ID and CONFIG_DM_ETH_PHY are

enabled,

drivers/net/phy/phy.c struct phy_device *phy_connect(struct mii_dev

Quoted 26 lines

eth_phy_binds_nodes() called from eth_post_bind() already binds the ethernet PHY node to eth_phy_generic_drv. However, phy_connect_phy_id() called via zynq_phy_init() -> phy_connect() cannot attach to the already bound PHY device and creates a second instance, resulting in duplicate entries in the dm tree:

Fix this by guarding the phy_connect_phy_id() call in phy_connect() with uclass_get_device_by_phandle() to skip re-binding if the PHY node is already registered in UCLASS_ETH_PHY.

Fixes 68a4d1506109 ("net: phy: Bind ETH_PHY uclass driver to each new

PHY")

---
Changes in v2:
- Move duplicate binding check to caller phy_connect()
- Update commit description

  drivers/net/phy/phy.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)
drivers/net/phy/phy.c index renamed from drivers/net/phy/phy.c +0 -0

d7e0c4fe02d..cfca8328f1c 100644

drivers/net/phy/phy.c +0 -0
@@ -938,8 +938,13 @@ struct phy_device *phy_connect(struct mii_dev

*bus, int addr,

drivers/net/phy/phy.c struct phy_device *phy_connect(struct mii_dev *bus, int addr, · line 946

Quoted 11 lines
drivers/net/phy/phy.c +6 -2
  #endif
 
  #ifdef CONFIG_PHY_ETHERNET_ID
-   if (!phydev)
-           phydev = phy_connect_phy_id(bus, dev, addr);
+   if (!phydev) {
+           struct udevice *phy_dev;
+
+           if (uclass_get_device_by_phandle(UCLASS_ETH_PHY, dev,
+                                            "phy-handle", &phy_dev))
+                   phydev = phy_connect_phy_id(bus, dev, addr);
Quoted 4 lines

Shouldn't this binding check be closer to the beginning of phy_connect() ? Basically check whether the PHY driver is already bound at the beginning, and if so, you are done. I don't think that test is specific to CONFIG_PHY_ETHERNET_ID, is it ?

The check is placed inside phy_connect_phy_id() because that is the only code path that calls device_bind_driver_to_node() on the PHY node, which creates the duplicate. The other paths (phy_connect_fixed, NCSI, phy_connect_gmii2rgmii) do not bind the PHY node and are not affected.

Thanks & Regards,
Pranav Tilak

#3 lore raw
Quoted 59 lines · Pranav

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Marek,

Quoted 55 lines

-----Original Message-----
From: Marek Vasut <[email protected]>
Sent: Wednesday, April 8, 2026 10:02 PM
To: Vinay Tilak, Pranav <[email protected]>; [email protected]; Simek, Michal <[email protected]>
Cc: git (AMD-Xilinx) <[email protected]>; Begari, Padmarao <[email protected]>; Jerome Forissier <[email protected]>; Tom Rini <[email protected]>; Peng Fan <[email protected]>; Lucien.Jheng <[email protected]>; Yao Zi <[email protected]>; SkyLake.Huang <[email protected]>; Siddharth Vadapalli <[email protected]>; Marek Vasut <[email protected]>; Ramon Fried <[email protected]>
Subject: Re: [PATCH v2] net: phy: fix duplicate eth_phy binding

Quoted 1 line · Pranav Tilak

When both CONFIG_PHY_ETHERNET_ID and CONFIG_DM_ETH_PHY are

enabled,

drivers/net/phy/phy.c struct phy_device *phy_connect(struct mii_dev

Quoted 26 lines

eth_phy_binds_nodes() called from eth_post_bind() already binds the ethernet PHY node to eth_phy_generic_drv. However, phy_connect_phy_id() called via zynq_phy_init() -> phy_connect() cannot attach to the already bound PHY device and creates a second instance, resulting in duplicate entries in the dm tree:

Fix this by guarding the phy_connect_phy_id() call in phy_connect() with uclass_get_device_by_phandle() to skip re-binding if the PHY node is already registered in UCLASS_ETH_PHY.

Fixes 68a4d1506109 ("net: phy: Bind ETH_PHY uclass driver to each new

PHY")

---
Changes in v2:
- Move duplicate binding check to caller phy_connect()
- Update commit description

   drivers/net/phy/phy.c | 9 +++++++--
   1 file changed, 7 insertions(+), 2 deletions(-)
drivers/net/phy/phy.c index renamed from drivers/net/phy/phy.c +0 -0

d7e0c4fe02d..cfca8328f1c 100644

drivers/net/phy/phy.c +0 -0
@@ -938,8 +938,13 @@ struct phy_device *phy_connect(struct mii_dev

*bus, int addr,

Quoted 11 lines
drivers/net/phy/phy.c +6 -2
   #endif
 
   #ifdef CONFIG_PHY_ETHERNET_ID
-   if (!phydev)
-           phydev = phy_connect_phy_id(bus, dev, addr);
+   if (!phydev) {
+           struct udevice *phy_dev;
+
+           if (uclass_get_device_by_phandle(UCLASS_ETH_PHY, dev,
+                                            "phy-handle", &phy_dev))
+                   phydev = phy_connect_phy_id(bus, dev, addr);
Quoted 6 lines
Quoted 4 lines

Shouldn't this binding check be closer to the beginning of phy_connect() ? Basically check whether the PHY driver is already bound at the beginning, and if so, you are done. I don't think that test is specific to CONFIG_PHY_ETHERNET_ID, is it ?

The check is placed inside phy_connect_phy_id() because that is the only code path that calls device_bind_driver_to_node() on the PHY node, which creates the duplicate.

The device_bind_driver_to_node() in phy_connect_phy_id() is called only under very specific conditions -- if IS_ENABLED(CONFIG_DM_ETH_PHY) and phy-handle property exists and points to a valid node. The DM_ETH_PHY test is not covered here.

But that is besides the point. I think phy_connect() should figure out whether any PHY is already connected, and if it is, simply exit early. This sounds to me like a more generic solution.

Quoted 1 line

The other paths (phy_connect_fixed, NCSI, phy_connect_gmii2rgmii) do not bind the PHY node and are not affected.

I worry this is only a matter of time, hence it would be good to have a generic solution which covers all the PHY options.