On Mon, Sep 30, 2019 at 04:29:07PM +0200, Oleksij Rempel wrote:

On Mon, Sep 30, 2019 at 03:42:09PM +0200, Andrew Lunn wrote:

quoted

On Mon, Sep 30, 2019 at 11:33:10AM +0200, Oleksij Rempel wrote:

quoted

The driver was working with fixed phy without any noticeable issues. This bug
was uncovered by introducing dsa ar9331-switch driver.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
---
 drivers/net/ethernet/atheros/ag71xx.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/atheros/ag71xx.c b/drivers/net/ethernet/atheros/ag71xx.c
index 6703960c7cf5..d1101eea15c2 100644
--- a/drivers/net/ethernet/atheros/ag71xx.c
+++ b/drivers/net/ethernet/atheros/ag71xx.c

@@ -526,7 +526,7 @@ static int ag71xx_mdio_probe(struct ag71xx *ag)
 	struct device *dev = &ag->pdev->dev;
 	struct net_device *ndev = ag->ndev;
 	static struct mii_bus *mii_bus;
-	struct device_node *np;
+	struct device_node *np, *mnp;
 	int err;
 
 	np = dev->of_node;

@@ -571,7 +571,9 @@ static int ag71xx_mdio_probe(struct ag71xx *ag)
 		msleep(200);
 	}
 
-	err = of_mdiobus_register(mii_bus, np);
+	mnp = of_get_child_by_name(np, "mdio");
+	err = of_mdiobus_register(mii_bus, mnp);
+	of_node_put(mnp);
 	if (err)
 		goto mdio_err_put_clk;

Hi Oleksij

You need to keep backwards compatibility here. If you find an mdio
node, use it, but if not, you need to still register np.

This is also extending the driver binding, so you need to update the
binding documentation.

Hi Andrew,

Normally i would agree. But in this case:
- this driver is freshly added to the kernel and is different to OpenWrt
  implementation any way. No users from this side.
- Devicetree binding says:
  Documentation/devicetree/bindings/net/qca,ar71xx.txt
|Optional properties:
|- phy-handle : phandle to the PHY device connected to this device.
|- fixed-link : Assume a fixed link. See fixed-link.txt in the same directory.
|  Use instead of phy-handle.
|
|Optional subnodes:
|- mdio : specifies the mdio bus, used as a container for phy nodes
|  according to phy.txt in the same directory

So, it is driver bug ...ooOO (my personal bug :D)

Hi Oleksij

Ah, O.K. You should of explained that in the commit message.

Is the mdio support just in -rc, or is it older?

You need to add a Fixes: tag.

The patch subject should be [PATCH net] to indicate this is a fix to
the net tree

The patch should be against net, not next-next.

    Andrew