Re: 5.4 Regression in r8169 with jumbo frames - packet loss/delays

[Alan J. Wylie]

at 22:37 on Sat 30-Nov-2019 Heiner Kallweit (hkallweit1@xxxxxxxxx) wrote:

> Thanks for the report. A jumbo fix for one chip version may have
> revealed an issue with another chip version. Could you please try
> the following?
> I checked the vendor driver r8168 and there's no special sequence
> to configure jumbo mode.
> 
> What would be interesting:
> Do you set the (jumbo) MTU before bringing the device up?
> 
> diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c
> index 0b47db2ff..38d212686 100644
> --- a/drivers/net/ethernet/realtek/r8169_main.c
> +++ b/drivers/net/ethernet/realtek/r8169_main.c
> @@ -3873,7 +3873,7 @@ static void rtl_hw_jumbo_enable(struct rtl8169_private *tp)
>  	case RTL_GIGA_MAC_VER_27 ... RTL_GIGA_MAC_VER_28:
>  		r8168dp_hw_jumbo_enable(tp);
>  		break;
> -	case RTL_GIGA_MAC_VER_31 ... RTL_GIGA_MAC_VER_34:
> +	case RTL_GIGA_MAC_VER_31 ... RTL_GIGA_MAC_VER_33:
>  		r8168e_hw_jumbo_enable(tp);
>  		break;
>  	default:
> -- 
> 2.24.0

That patch fixes the issue for me.

Thanks

Alan

-- 
Alan J. Wylie                                          https://www.wylie.me.uk/

Dance like no-one's watching. / Encrypt like everyone is.
Security is inversely proportional to convenience