The mii_bus API conversion to read_c45() and write_c45() did not cover
the mdio-mux driver before read() and write() were made C22-only.

This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
of_mdiobus_register() and this makes the mdio-mux driver fail to probe
the entire child buses, not just the PHYs that cause access errors.

Fix the regression by introducing special c45 read and write accessors
to mdio-mux which forward the operation to the parent MDIO bus.

Fixes: db1a63aed89c ("net: phy: Remove fallback to old C45 method")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
v1->v2: apply Russell's suggestion to assign the function pointers
        conditionally based on their presence in the parent bus

 drivers/net/mdio/mdio-mux.c | 47 +++++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/drivers/net/mdio/mdio-mux.c b/drivers/net/mdio/mdio-mux.c
index a881e3523328..bef4cce71287 100644
--- a/drivers/net/mdio/mdio-mux.c
+++ b/drivers/net/mdio/mdio-mux.c
@@ -55,6 +55,27 @@ static int mdio_mux_read(struct mii_bus *bus, int phy_id, int regnum)
 	return r;
 }
 
+static int mdio_mux_read_c45(struct mii_bus *bus, int phy_id, int dev_addr,
+			     int regnum)
+{
+	struct mdio_mux_child_bus *cb = bus->priv;
+	struct mdio_mux_parent_bus *pb = cb->parent;
+	int r;
+
+	mutex_lock_nested(&pb->mii_bus->mdio_lock, MDIO_MUTEX_MUX);
+	r = pb->switch_fn(pb->current_child, cb->bus_number, pb->switch_data);
+	if (r)
+		goto out;
+
+	pb->current_child = cb->bus_number;
+
+	r = pb->mii_bus->read_c45(pb->mii_bus, phy_id, dev_addr, regnum);
+out:
+	mutex_unlock(&pb->mii_bus->mdio_lock);
+
+	return r;
+}
+
 /*
  * The parent bus' lock is used to order access to the switch_fn.
  */
@@ -80,6 +101,28 @@ static int mdio_mux_write(struct mii_bus *bus, int phy_id,
 	return r;
 }
 
+static int mdio_mux_write_c45(struct mii_bus *bus, int phy_id, int dev_addr,
+			      int regnum, u16 val)
+{
+	struct mdio_mux_child_bus *cb = bus->priv;
+	struct mdio_mux_parent_bus *pb = cb->parent;
+
+	int r;
+
+	mutex_lock_nested(&pb->mii_bus->mdio_lock, MDIO_MUTEX_MUX);
+	r = pb->switch_fn(pb->current_child, cb->bus_number, pb->switch_data);
+	if (r)
+		goto out;
+
+	pb->current_child = cb->bus_number;
+
+	r = pb->mii_bus->write_c45(pb->mii_bus, phy_id, dev_addr, regnum, val);
+out:
+	mutex_unlock(&pb->mii_bus->mdio_lock);
+
+	return r;
+}
+
 static int parent_count;
 
 static void mdio_mux_uninit_children(struct mdio_mux_parent_bus *pb)
@@ -173,6 +216,10 @@ int mdio_mux_init(struct device *dev,
 		cb->mii_bus->parent = dev;
 		cb->mii_bus->read = mdio_mux_read;
 		cb->mii_bus->write = mdio_mux_write;
+		if (parent_bus->read_c45)
+			cb->mii_bus->read_c45 = mdio_mux_read_c45;
+		if (parent_bus->write_c45)
+			cb->mii_bus->write_c45 = mdio_mux_write_c45;
 		r = of_mdiobus_register(cb->mii_bus, child_bus_node);
 		if (r) {
 			mdiobus_free(cb->mii_bus);
-- 
2.34.1

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 17 Oct 2023 17:31:44 +0300 you wrote:
> The mii_bus API conversion to read_c45() and write_c45() did not cover
> the mdio-mux driver before read() and write() were made C22-only.
> 
> This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
> The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
> get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
> of_mdiobus_register() and this makes the mdio-mux driver fail to probe
> the entire child buses, not just the PHYs that cause access errors.
> 
> [...]

Here is the summary with links:
  - [v2,net] net: mdio-mux: fix C45 access returning -EIO after API change
    https://git.kernel.org/netdev/net/c/1f9f2143f24e

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

On Tue, Oct 17, 2023 at 05:31:44PM +0300, Vladimir Oltean wrote:
> The mii_bus API conversion to read_c45() and write_c45() did not cover
> the mdio-mux driver before read() and write() were made C22-only.
> 
> This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
> The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
> get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
> of_mdiobus_register() and this makes the mdio-mux driver fail to probe
> the entire child buses, not just the PHYs that cause access errors.
> 
> Fix the regression by introducing special c45 read and write accessors
> to mdio-mux which forward the operation to the parent MDIO bus.
> 
> Fixes: db1a63aed89c ("net: phy: Remove fallback to old C45 method")
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Thanks!

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

On Tue, Oct 17, 2023 at 05:31:44PM +0300, Vladimir Oltean wrote:
> The mii_bus API conversion to read_c45() and write_c45() did not cover
> the mdio-mux driver before read() and write() were made C22-only.
> 
> This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
> The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
> get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
> of_mdiobus_register() and this makes the mdio-mux driver fail to probe
> the entire child buses, not just the PHYs that cause access errors.
> 
> Fix the regression by introducing special c45 read and write accessors
> to mdio-mux which forward the operation to the parent MDIO bus.
> 
> Fixes: db1a63aed89c ("net: phy: Remove fallback to old C45 method")
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Andrew Lunn <andrew@lunn.ch>

    Andrew

On 10/17/23 07:31, Vladimir Oltean wrote:
> The mii_bus API conversion to read_c45() and write_c45() did not cover
> the mdio-mux driver before read() and write() were made C22-only.
> 
> This broke arch/arm64/boot/dts/freescale/fsl-ls1028a-qds-13bb.dtso.
> The -EOPNOTSUPP from mdiobus_c45_read() is transformed by
> get_phy_c45_devs_in_pkg() into -EIO, is further propagated to
> of_mdiobus_register() and this makes the mdio-mux driver fail to probe
> the entire child buses, not just the PHYs that cause access errors.
> 
> Fix the regression by introducing special c45 read and write accessors
> to mdio-mux which forward the operation to the parent MDIO bus.
> 
> Fixes: db1a63aed89c ("net: phy: Remove fallback to old C45 method")
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>

Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
-- 
Florian