Re: [PATCH v2 net-next 2/2] r8169: support RTL8168G

[Joe Perches <joe@perches.com>]

On Thu, 2012-07-05 at 00:04 +0200, Francois Romieu wrote:
> Hayes Wang <hayeswang@realtek.com> :
> > Support the new chip RTL8168G.

> - save Joe P. some work

Thanks.  Just a trivial thing below:

[]

> diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
[]
> +#define	RTL_LOOP_MAX	10000
> +
> +static void rtl_mcu_wait_list_ready(void __iomem *ioaddr)
> +{
> +	int i;
> +
> +	for (i = 0; i < RTL_LOOP_MAX; i++) {
> +		if (RTL_R8(MCU) & LINK_LIST_RDY)
> +			return;
> +		udelay(100);
> +	}
> +}

This pattern is used a couple more times.
There's no failure handling either.

Maybe use a macro with RTL_R8/32, register and test?
Maybe the delays could be tuned out a bit better.
Maybe a continuous read or a less frequent read
might be better.

>  static void __devinit rtl_hw_init_8168g(struct rtl8169_private *tp)
>  {
[]
> +	for (i = 0; i < RTL_LOOP_MAX; i++) {
> +		if (RTL_R32(TxConfig) & TXCFG_EMPTY)
> +			break;
>  		udelay(100);
> +	}

pattern

> +	for (i = 0; i < RTL_LOOP_MAX; i++) {
> +		if ((RTL_R8(MCU) & RXTX_EMPTY) == RXTX_EMPTY)
> +			break;
>  		udelay(100);
> +	}

pattern