[LEDE-DEV] SUNXI: Network problem with kernel 4.9.51

[Hauke Mehrtens]

On 09/27/2017 03:35 AM, Lucian Cristian wrote:
> On 26.09.2017 12:25, Gerhard Bertelsmann wrote:
>> Hi Hauke,
> [snip]
>>>> root at LEDE:~# ping -qc 100 192.168.0.9
>>>> PING 192.168.0.9 (192.168.0.9): 56 data bytes
>>>>
>>>> --- 192.168.0.9 ping statistics ---
>>>> 100 packets transmitted, 100 packets received, 0% packet loss
>>>> round-trip min/avg/max = 0.363/0.514/0.813 ms
>>>>
>>>>
>>>> root at LEDE:~# iperf -c 192.168.0.9 -d
>>>> ------------------------------------------------------------
>>>> Server listening on TCP port 5001
>>>> TCP window size: 85.3 KByte (default)
>>>> ------------------------------------------------------------
>>>> ------------------------------------------------------------
>>>> Client connecting to 192.168.0.9, TCP port 5001
>>>> TCP window size: 87.5 KByte (default)
>>>> ------------------------------------------------------------
>>>> [  5] local 192.168.0.159 port 54946 connected with 192.168.0.9 port
>>>> 5001
>>>> [  4] local 192.168.0.159 port 5001 connected with 192.168.0.9 port
>>>> 52498
>>>> [ ID] Interval       Transfer     Bandwidth
>>>> [  5]  0.0-10.0 sec   695 MBytes   583 Mbits/sec
>>>> [  4]  0.0-10.0 sec   330 MBytes   276 Mbits/sec
>>>>
>>>> I'm not an Uboot expert, maybe somebody else could migrate
>>>> this parameter from the old Uboot to the new one.
>>>>
>>>> Regards
>>>>
>>>> Gerd
>>>>
>>>>
>>>> _______________________________________________
>>>> Lede-dev mailing list
>>>> Lede-dev at lists.infradead.org
>>>> http://lists.infradead.org/mailman/listinfo/lede-dev
> 
> ok, is easy, adding  +CONFIG_GMAC_TX_DELAY=1 to uboot defconfig improves
> somehow (at least on LIME2), now I have ~270mb on one upload in ~500mb
> on multiple streams, before this speed was random, most of the time
> worked bad, I'll ad a PR
> 
> but on reset, the speed will drop to about 50mb, removing the power and
> cold boot would make the board work to ~270mb again

For the Banana PI CONFIG_GMAC_TX_DELAY=3 is set in U-Boot since a long
time, but I do not know if this is now overwritten by some kernel
driver. Could you try to set it to 0, it could be that some other
component sets adds the delay now.

> regarding to stmmac backport, I applied only this
> https://git.lede-project.org/?p=lede/hauke/staging.git;a=commit;h=9a359a1c990207f74f3429db23f163d3370787b4
> and stmmac is crashing the kernel, I'll try with a clean tree to see if
> it happens again

Do you have a crash log?

I would like to add this to master soon, because it adds ssupport for
the mac found on some recent Allwinner SoCs.

Hauke