Re: [net-next 3/3] net: ftgmac100: Support for AST2700

[Simon Horman]

On Thu, Nov 07, 2024 at 07:15:00PM +0800, Jacky Chou wrote:
> The AST2700 is the 7th generation SoC from Aspeed, featuring three GPIO
> controllers that are support 64-bit DMA capability.
> Adding features is shown in the following list.
> 1.Support 64-bit DMA
>   Add the high address (63:32) registers for description address and the
>   description field for packet buffer with high address part.
>   These registers and fields in legacy Aspeed SoC are reserved.
>   This 64-bit DMA changing has verified on legacy Aspeed Soc, like
>   AST2600.
> 2.Set RMII pin strap in AST2700 compitable
>   Use bit 20 of MAC 0x50 to represent the pin strap of AST2700 RMII and
>   RGMII. Set to 1 is RMII pin, otherwise is RGMII.
>   This bis is also reserved in legacy Aspeed SoC.
> 
> Signed-off-by: Jacky Chou <jacky_chou@xxxxxxxxxxxxxx>

...

> @@ -1965,16 +1980,27 @@ static int ftgmac100_probe(struct platform_device *pdev)
>  			dev_err(priv->dev, "MII probe failed!\n");
>  			goto err_ncsi_dev;
>  		}
> -
>  	}
>  
>  	if (priv->is_aspeed) {
> +		struct reset_control *rst;
> +
>  		err = ftgmac100_setup_clk(priv);
>  		if (err)
>  			goto err_phy_connect;
>  
> -		/* Disable ast2600 problematic HW arbitration */
> -		if (of_device_is_compatible(np, "aspeed,ast2600-mac"))
> +		rst = devm_reset_control_get_optional(priv->dev, NULL);
> +		if (IS_ERR(rst))

Hi Jacky,

Should err be set to ERR_PTR(rst) here so that value is returned by
the function?

> +			goto err_register_netdev;
> +
> +		priv->rst = rst;
> +		err = reset_control_assert(priv->rst);
> +		mdelay(10);
> +		err = reset_control_deassert(priv->rst);
> +
> +		/* Disable some aspeed platform problematic HW arbitration */
> +		if (of_device_is_compatible(np, "aspeed,ast2600-mac") ||
> +		    of_device_is_compatible(np, "aspeed,ast2700-mac"))
>  			iowrite32(FTGMAC100_TM_DEFAULT,
>  				  priv->base + FTGMAC100_OFFSET_TM);
>  	}