[net-next PATCH v2 08/14] net: mdiobus: Introduce fwnode_mdiobus_register()

Calvin Johnson calvin.johnson at oss.nxp.com
Fri Dec 18 00:40:44 EST 2020


On Tue, Dec 15, 2020 at 07:53:26PM +0200, Andy Shevchenko wrote:
> On Tue, Dec 15, 2020 at 6:44 PM Calvin Johnson
> <calvin.johnson at oss.nxp.com> wrote:
> >
> > Introduce fwnode_mdiobus_register() to register PHYs on the  mdiobus.
> > If the fwnode is DT node, then call of_mdiobus_register().
> > If it is an ACPI node, then:
> >         - disable auto probing of mdiobus
> >         - register mdiobus
> >         - save fwnode to mdio structure
> >         - loop over child nodes & register a phy_device for each PHY
> 
> ...
> 
> > +/**
> > + * fwnode_mdiobus_register - Register mii_bus and create PHYs from fwnode
> > + * @mdio: pointer to mii_bus structure
> > + * @fwnode: pointer to fwnode of MDIO bus.
> > + *
> > + * This function registers the mii_bus structure and registers a phy_device
> > + * for each child node of @fwnode.
> > + */
> > +int fwnode_mdiobus_register(struct mii_bus *mdio, struct fwnode_handle *fwnode)
> > +{
> > +       struct fwnode_handle *child;
> > +       unsigned long long addr;
> > +       acpi_status status;
> > +       int ret;
> > +
> > +       if (is_of_node(fwnode)) {
> > +               return of_mdiobus_register(mdio, to_of_node(fwnode));
> > +       } else if (is_acpi_node(fwnode)) {
> 
> I would rather see this as simple as
> 
>      if (is_of_node(fwnode))
>                return of_mdiobus_register(mdio, to_of_node(fwnode));
>      if (is_acpi_node(fwnode))
>                return acpi_mdiobus_register(mdio, fwnode);
> 
> where the latter one is defined somewhere in drivers/acpi/.
Makes sense. I'll do it. But I think it will be better to place
acpi_mdiobus_register() here itself in the network subsystem, maybe
/drivers/net/mdio/acpi_mdio.c.
> 
> > +               /* Mask out all PHYs from auto probing. */
> > +               mdio->phy_mask = ~0;
> > +               ret = mdiobus_register(mdio);
> > +               if (ret)
> > +                       return ret;
> > +
> > +               mdio->dev.fwnode = fwnode;
> > +       /* Loop over the child nodes and register a phy_device for each PHY */
> > +               fwnode_for_each_child_node(fwnode, child) {
> 
> > +                       status = acpi_evaluate_integer(ACPI_HANDLE_FWNODE(child),
> > +                                                      "_ADR", NULL, &addr);
> > +                       if (ACPI_FAILURE(status)) {
> 
> Isn't it fwnode_get_id() now?
Yes. Will change it.
> 
> > +                               pr_debug("_ADR returned %d\n", status);
> > +                               continue;
> > +                       }
> 
> > +                       if (addr < 0 || addr >= PHY_MAX_ADDR)
> > +                               continue;
> 
> addr can't be less than 0.
Yes. will update in v3.
> 
> > +                       ret = fwnode_mdiobus_register_phy(mdio, child, addr);
> > +                       if (ret == -ENODEV)
> > +                               dev_err(&mdio->dev,
> > +                                       "MDIO device at address %lld is missing.\n",
> > +                                       addr);
> > +               }
> > +               return 0;
> > +       }
> > +       return -EINVAL;
> > +}
> 
> -- 
> With Best Regards,
> Andy Shevchenko



More information about the linux-arm-kernel mailing list