On Thu, Dec 05, 2024 at 03:51:33PM +0100, Christian Marangi wrote:
> +static int an8855_efuse_read(void *context, unsigned int offset,
> + void *val, size_t bytes)
> +{
> + struct an8855_priv *priv = context;
> +
> + return regmap_bulk_read(priv->regmap, AN8855_EFUSE_DATA0 + offset,
> + val, bytes / sizeof(u32));
> +}
> +
> +static struct nvmem_config an8855_nvmem_config = {
> + .name = "an8855-efuse",
> + .size = AN8855_EFUSE_CELL * sizeof(u32),
> + .stride = sizeof(u32),
> + .word_size = sizeof(u32),
> + .reg_read = an8855_efuse_read,
> +};
> +
> +static int an8855_sw_register_nvmem(struct an8855_priv *priv)
> +{
> + struct nvmem_device *nvmem;
> +
> + an8855_nvmem_config.priv = priv;
> + an8855_nvmem_config.dev = priv->dev;
> + nvmem = devm_nvmem_register(priv->dev, &an8855_nvmem_config);
> + if (IS_ERR(nvmem))
> + return PTR_ERR(nvmem);
> +
> + return 0;
> +}
At some point we should enforce the rule that new drivers for switch
SoCs with complex peripherals should use MFD and move all non-networking
peripherals to drivers handled by their respective subsystems.
I don't have the expertise to review a nvmem driver, and the majority of
them are in drivers/nvmem, with a dedicated subsystem and maintainer.
In general I want to make sure it is clear that I don't encourage the
model where DSA owns the entire mdio_device.
What other peripherals are there on this SoC other than an MDIO bus and
an EFUSE? IRQCHIP, GPIOs, LED controller, sensors?
You can take a look at drivers/mfd/ocelot* and
Documentation/devicetree/bindings/mfd/mscc,ocelot.yaml for an example on
how to use mfd for the top-level MDIO device, and DSA as just the driver
for the Ethernet switch component (which will be represented as a
platform_device).
