Re: [PATCH 2/2] rtl8180: don't use weird trick to access "far" registers

[Dave Kilroy <kilroyd@googlemail.com>]

On 28/03/2014 17:14, Andrea Merello wrote:
> @@ -340,9 +334,9 @@ struct rtl818x_csr {
>    * I don't like to introduce a ton of "reserved"..
>    * They are for RTL8187SE
>    */
> -#define REG_ADDR1(addr)	((u8 __iomem *)(&priv->map->offset1[(addr)]))
> -#define REG_ADDR2(addr)	((__le16 __iomem *)(&priv->map->offset2[((addr) >> 1)]))
> -#define REG_ADDR4(addr)	((__le32 __iomem *)(&priv->map->offset4[((addr) >> 2)]))
> +#define REG_ADDR1(addr)	((u8 __iomem *)priv->map + addr)
> +#define REG_ADDR2(addr)	((__le16 __iomem *)priv->map + (addr >> 1))
> +#define REG_ADDR4(addr)	((__le32 __iomem *)priv->map + (addr >> 2))
>   
>   #define FEMR_SE		REG_ADDR2(0x1D4)
>   #define ARFR		REG_ADDR2(0x1E0)
I suggest parenthesizing the use of addr in your macros (as the original 
code does), to avoid any issues with operator precedence wrt >>.

If the removal was intentional and you've verified there's no issues, 
you should probably mention it in the commit message.


Dave.