[PATCH v6 08/16] MIPS: lantiq: Convert the fpi bus driver to a platform_driver

[Andy Shevchenko]

On Fri, Jun 30, 2017 at 12:39 AM, Hauke Mehrtens <hauke at hauke-m.de> wrote:
> Instead of hacking the configuration of the FPI bus into the arch code
> add an own bus driver for this internal bus. The FPI bus is the main
> bus of the SoC. This bus driver makes sure the bus is configured
> correctly before the child drivers are getting initialized. This driver
> will probably also be used on different SoC later.

> +       /* RCU configuration is optional */
> +       rcu_regmap = syscon_regmap_lookup_by_phandle(np, "lantiq,rcu");
> +       if (!IS_ERR_OR_NULL(rcu_regmap)) {

It's still concerning me.

If the API is correctly designed then it should look something like

if (IS_ERR(...))
 return PTR_ERR(...);

if (!...) {
 ...stuff for optional case...
}

> +               ret = device_property_read_u32(dev, "lantiq,offset-endianness",
> +                                          &rcu_ahb_endianness_reg_offset);
> +               if (ret) {
> +                       dev_err(&pdev->dev, "Failed to get RCU reg offset\n");
> +                       return -EINVAL;
> +               }
> +
> +               if (regmap_update_bits(rcu_regmap,
> +                                       rcu_ahb_endianness_reg_offset,
> +                                       RCU_VR9_BE_AHB1S,
> +                                       RCU_VR9_BE_AHB1S))
> +                       dev_warn(&pdev->dev,
> +                               "Failed to configure RCU AHB endianness\n");
> +       }
> +
> +       /* disable fpi burst */
> +       ltq_w32_mask(XBAR_FPI_BURST_EN, 0, xbar_membase + XBAR_ALWAYS_LAST);
> +
> +       return of_platform_populate(dev->of_node, NULL, NULL, dev);
> +}

-- 
With Best Regards,
Andy Shevchenko