On Tue, Dec 15, 2020 at 07:33:40PM +0200, Andy Shevchenko wrote:

On Tue, Dec 15, 2020 at 6:44 PM Calvin Johnson
[off-list ref] wrote:

quoted

Introduce fwnode_mdiobus_register_phy() to register PHYs on the
mdiobus. From the compatible string, identify whether the PHY is
c45 and based on this create a PHY device instance which is
registered on the mdiobus.

...

quoted

+int fwnode_mdiobus_register_phy(struct mii_bus *bus,
+                               struct fwnode_handle *child, u32 addr)
+{
+       struct mii_timestamper *mii_ts;
+       struct phy_device *phy;
+       const char *cp;
+       bool is_c45;
+       u32 phy_id;
+       int rc;

quoted

+       if (is_of_node(child)) {
+               mii_ts = of_find_mii_timestamper(to_of_node(child));
+               if (IS_ERR(mii_ts))
+                       return PTR_ERR(mii_ts);
+       }

Perhaps

               mii_ts = of_find_mii_timestamper(to_of_node(child));

quoted

+
+       rc = fwnode_property_read_string(child, "compatible", &cp);
+       is_c45 = !(rc || strcmp(cp, "ethernet-phy-ieee802.3-c45"));
+
+       if (is_c45 || fwnode_get_phy_id(child, &phy_id))
+               phy = get_phy_device(bus, addr, is_c45);
+       else
+               phy = phy_device_create(bus, addr, phy_id, 0, NULL);
+       if (IS_ERR(phy)) {

quoted

+               if (mii_ts && is_of_node(child))
+                       unregister_mii_timestamper(mii_ts);

if (!IS_ERR_OR_NULL(mii_ts))
 ...

However it points to the question why unregister() doesn't handle the
above cases.
I would expect unconditional call to unregister() here.

This is following the logic defined in of_mdiobus_register_phy().
https://elixir.bootlin.com/linux/latest/source/drivers/net/mdio/of_mdio.c#L107

	mii_ts = of_find_mii_timestamper(child);
	if (IS_ERR(mii_ts))
		return PTR_ERR(mii_ts);

I think the logic above is correct because this function returns only if
of_find_mii_timestamper() returns error. On NULL return, it proceeds further.

quoted

+               return PTR_ERR(phy);
+       }
+
+       if (is_acpi_node(child)) {
+               phy->irq = bus->irq[addr];
+
+               /* Associate the fwnode with the device structure so it
+                * can be looked up later.
+                */
+               phy->mdio.dev.fwnode = child;
+
+               /* All data is now stored in the phy struct, so register it */
+               rc = phy_device_register(phy);
+               if (rc) {
+                       phy_device_free(phy);
+                       fwnode_handle_put(phy->mdio.dev.fwnode);
+                       return rc;
+               }
+
+               dev_dbg(&bus->dev, "registered phy at address %i\n", addr);
+       } else if (is_of_node(child)) {
+               rc = of_mdiobus_phy_device_register(bus, phy, to_of_node(child), addr);
+               if (rc) {

quoted

+                       if (mii_ts)
+                               unregister_mii_timestamper(mii_ts);

Ditto.

quoted

+                       phy_device_free(phy);
+                       return rc;
+               }
+
+               /* phy->mii_ts may already be defined by the PHY driver. A
+                * mii_timestamper probed via the device tree will still have
+                * precedence.
+                */

quoted

+               if (mii_ts)
+                       phy->mii_ts = mii_ts;

How is that defined? Do you need to do something with an old pointer?

As the comment says, I think PHY drivers which got invoked before calling
of_mdiobus_register_phy() may have defined phy->mii_ts.

quoted

+       }
+       return 0;
+}

-- 
With Best Regards,
Andy Shevchenko