[LEDE-DEV] Transmit timeouts with mtk_eth_soc and MT7621

[Kristian Evensen]

Hello,

On Wed, Nov 8, 2017 at 10:56 PM, Kristian Evensen
<kristian.evensen at gmail.com> wrote:
> I have been trying to solve this myself for a couple of days, but I am
> starting to run out of idea. Could it be that there is some traffic
> destined for the client (via. the 2626) that gets stuck in the TX
> queue on the 3526? Any help, pointers on where to look or ideas for
> what could be wrong would be much appreciated.

I have been doing some more debugging during the day today, and it
seems that the problem is that the hardware for some reason stops
transmitting data. I dumped the content of the TX ring on every
timeout, and the output seems to indicate that the state of the ring
is sane. For example, with the following error:

[  325.650638] mtk_soc_eth 1e100000.ethernet eth0: transmit timed out
[  325.656857] mtk_soc_eth 1e100000.ethernet eth0: dma_cfg:80000067
[  325.662902] mtk_soc_eth 1e100000.ethernet eth0: tx_ring=0,
base=0ef58000, max=512, ctx=384, dtx=0, fdx=0, next=384
[  325.673234] mtk_soc_eth 1e100000.ethernet eth0: rx_ring=0,
base=0ef5e000, max=512, calc=5, drx=6

I see that the CPU txds [384-511] have DDONE set and no SKB, while
DDONE is not set for the DMA txds [0-383] and an skb is attached. I
also looked at the content of the skb, and as far as I can see it is
valid. Looking at the content of the SKB also shows that
fe_reset_pending() does its job. For every timeout, there is a new set
of packets on the ring. So new packets are put on the ring, but none
are sent.

-Kristian