Re: [PATCH 01/11] fs_enet: Add support for MPC512x to fs_enet driver

[Scott Wood <scottwood@freescale.com>]

Anatolij Gustschin wrote:
> +struct reg_tbl {
> +	void __iomem *fec_ievent;
> +	void __iomem *fec_imask;
> +	void __iomem *fec_r_des_active;
> +	void __iomem *fec_x_des_active;
> +	void __iomem *fec_r_des_start;
> +	void __iomem *fec_x_des_start;
> +	void __iomem *fec_r_cntrl;
> +	void __iomem *fec_ecntrl;
> +	void __iomem *fec_ivec;
> +	void __iomem *fec_mii_speed;
> +	void __iomem *fec_addr_low;
> +	void __iomem *fec_addr_high;
> +	void __iomem *fec_hash_table_high;
> +	void __iomem *fec_hash_table_low;
> +	void __iomem *fec_r_buff_size;
> +	void __iomem *fec_r_bound;
> +	void __iomem *fec_r_fstart;
> +	void __iomem *fec_x_fstart;
> +	void __iomem *fec_fun_code;
> +	void __iomem *fec_r_hash;
> +	void __iomem *fec_x_cntrl;
> +	void __iomem *fec_dma_control;
> +};

Why void and not the specific type?

>  static void set_promiscuous_mode(struct net_device *dev)
>  {
>  	struct fs_enet_private *fep = netdev_priv(dev);
> -	fec_t __iomem *fecp = fep->fec.fecp;
> +	struct reg_tbl *fecp = fep->fec.rtbl;

Hmm, having something called "fecp" that is a different type than other 
"fecp"s could be confusing.

> @@ -134,6 +143,20 @@ static int __devinit fs_enet_mdio_probe(struct of_device *ofdev,
>  	if (!fec->fecp)
>  		goto out_fec;
>  
> +	if (of_device_is_compatible(ofdev->node, "fsl,mpc5121-fec-mdio")) {

You can put a data pointer in the of_platform match struct, instead of 
re-checking the compatible.

-Scott