stmmac/RTL8211F/Meson GXBB: TX throughput problems

[peppe.cavallaro@st.com (Giuseppe CAVALLARO)]

Hi Jerome

We met, on some platform, problems that, in a glance,
look similar to yours and those were related to EEE and
LPI interrupts.
We had a subset of patches to disable EEE but, indeed,
if you disable from PHY, the STMMAC will never act
the protocol. We can provide you these patches... I will
ask Alex to decide how to proceed.
There is a issue on lpi timer and I have another patch to
share but I am not sure this can help.

In sum, if EEE is a suspected problem we can share and review
some patches.
In the meantime, I will read again the thread just to see if
there is something I am missing.

Regards
Peppe

On 11/3/2016 5:57 PM, Jerome Brunet wrote:
> On Mon, 2016-10-31 at 11:25 +0100, Andr? Roth wrote:
>> Hi all,
>>
>>>
>>> on my device this results in:
>>> [0xc9410018] = 0x2000000
>>> [0xc9410030] = 0x0
>>> [0xc941003c] = 0x0
>>> [0xc9411000] = 0x1100802
>>> [0xc9411018] = 0x2202006
>>> [0xc9411028] = 0x0
>>>
>>> maybe someone else could try the command from above on his device
>>> (running the original Amlogic kernel).
>>
>> those registers have the same value on an original image from
>> hardkernel:
>>
>> Linux odroid64 3.14.65-65 #1 SMP PREEMPT Sat May 28
>> 02:50:51 BRT 2016 aarch64 aarch64 aarch64 GNU/Linux
>>
>>>
>>> please also state if ethernet is working properly on the original
>>> kernel (and preferably which device/board this is).
>>
>> yes, the ethernet works flawless in 100 and 1000 Mbit/s on the 3.14
>> kernel.
>
> Andre, the 3.14 kernel you are talking, is it this one ? :
> https://github.com/hardkernel/linux/tree/odroidc2-3.14.y
>
> Because in drivers/net/phy/realtek.c, they disable EEE, but
> also 1000Base-T Full Duplex advertisement ?
>
> +	/* disable 1000m adv*/
> +	val = phy_read(phydev, 0x9);
> +	phy_write(phydev, 0x9, val&(~(1<<9)));
>
> If this is the kernel you are running, you should not be able to have
> ethernet at 1000MB/s ? Or is it in half duplex mode ?
>
>>
>> I can now confirm that both 100 and 1000 Mbit/s do not work properly
>> on the 4.8/integ branch. the network connection is interrupted after
>> some outbound traffic. it can be recovered by running a ifdown/ifup
>> which restarts dhclient, which I think is able to somehow reset the
>> interface.
>>
>> Anything I can help to debug the issue further ?
>>
>> Regards,
>>
>>  Andr?
>>
>>
>>
>> _______________________________________________
>> linux-amlogic mailing list
>> linux-amlogic at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-amlogic
>