Regarding eth0 (ftgmac100): transmit queue 0 timed out

[Mohammed.Habeeb ISV]

-----Original Message-----
From: Joel Stanley <joel at jms.id.au> 
Sent: Thursday, January 14, 2021 4:00 PM
To: Mohammed.Habeeb ISV <mohammed.habeeb at inventec.com>
Cc: openbmc at lists.ozlabs.org
Subject: Re: Regarding eth0 (ftgmac100): transmit queue 0 timed out

On Thu, 14 Jan 2021 at 23:48, Mohammed.Habeeb ISV <mohammed.habeeb at inventec.com> wrote:
>
> Hi All,
>
>
>
> We see below backtrace on our runBmc board. And networking does not work on eth0.
>
> No DHCP and static IP address can work. Pings and ARP fails, no packets received seen running tcpdump.
>
> Using 2.7 version of openBmc. Please let me know if there is a known fix or patch for this.
>
>
>
> transformers:~# [  158.082251] ------------[ cut here ]------------
>
> [  158.087442] WARNING: CPU: 1 PID: 0 at net/sched/sch_generic.c:447 
> dev_watchdog+0x2d4/0x2f8
>
> [  158.096681] NETDEV WATCHDOG: eth0 (ftgmac100): transmit queue 0 
> timed out
>
> [  158.104279] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 
> 5.4.8-671f622-dirty-f5845fd #1

This is a symptom of networking not working; there's many possible root causes.

Is this an ast2500?
-It is an ast2600.

From your log, it appears you're using a 5.4 kernel:

[  158.104279] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.4.8-671f622-dirty-f5845fd #1

This is not maintained by anyone, and is probably missing patches. It would be best if you moved to the latest openbmc kernel tree, but if you can't do that then look at the history of the ethernet driver in
dev-5.8 and see which patches you're missing:

$ git log --no-merges --oneline v5.4...dev-5.8
drivers/net/ethernet/faraday/ftgmac100.*
9f55d7825b96 net: ftgmac100: Ensure tx descriptor updates are visible
8de0141c2524 net: ftgmac100: Fix Aspeed ast2600 TX hang issue e6b45ee790ec net/faraday: fix grammar in function
ftgmac100_setup_clk() in ftgmac100.c
86eeb97de710 ftgmac100: Remove redundant judgement
d560b733ed53 net/faraday: Delete driver version from the drivers
f1294617d2f3 net: convert suitable network drivers to use phy_do_ioctl
0290bd291cc0 netdev: pass the stuck queue to the timeout handler
0c65b2b90d13 net: of_get_phy_mode: Change API to solve int/unit warnings 9bce4b27f3ec net: ftgmac100: Ungate RCLK for RMII on ASPEED MACs
0ac624f47dd3 docs: fix some broken references

The top two patches "net: ftgmac100: Ensure tx descriptor updates are visible" and "net: ftgmac100: Fix Aspeed ast2600 TX hang issue" are
ast2600 specific.

-Applied all these patches to sync drivers/net/ethernet/faraday/* with latest dev-5.8, still I see the backtrace. 
>
>
>
> Few issues were reported but no resolution.
>
> https://github.com/openbmc/openbmc/issues/3192
>
> https://github.com/openbmc/openbmc/issues/2228

These two issues were specific to the broadcom network device used on that platform. I doubt they relate to your issue.

>
>
>
> Tried below patch , seems does not work.
>
> http://lkml.iu.edu/hypermail/linux/kernel/2010.1/05957.html
>
>
>
> FYI ..We tried openBmc 2.9 version. It can work fine.
>
>
>
> Attached is the complete backtrace.
>
>
>
> Thanks
>
>