Skip to contentSkip to conversation

NCSI eth0 (ftgmac100): transmit queue 0 timed out error

DISCUSSION Евгений 2 participants 5 messages Nov 14, 2019 – Nov 24, 2019 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Hello Xiuzhi,

I have the similar problem using NCSI. Could you explain more clearly how you have solved one?

I use the Neptune Alfa OpenBMC kit from Portwell. It has an Intel I210 network interface chip. A host is connected to it through PCIe. The host network is working well. BMC has a dedicated interface (MAC0 + RTL8211), it works well. I want to create a BMC interface via NCSI + VLAN, but it does not work: see dmesg output below.
Please help me solve this problem.

:~# dmesg|grep net
[ 2.592506] ftgmac100 1e660000.ethernet: Error applying setting, reverse things back
[ 2.600628] ftgmac100 1e660000.ethernet: Generated random MAC address 7e:b5:bb:94:a2:d1
[ 2.621340] RTL8211E Gigabit Ethernet 1e660000.ethernet--1:00: attached PHY driver [RTL8211E Gigabit Ethernet] (mii_bus:phy_addr=1e660000.ethernet--1:00, irq=POLL)
[ 2.637074] ftgmac100 1e660000.ethernet eth0: irq 19, mapped at dfa05e77
[ 2.644477] ftgmac100 1e680000.ethernet: Generated random MAC address e6:cb:4e:cb:8c:06

*[ 2.652518] ftgmac100 1e680000.ethernet: Using NCSI interface*

[ 2.659567] ftgmac100 1e680000.ethernet eth1: irq 20, mapped at b403fb3e
[ 7.612280] Driver for 1-wire Dallas network protocol.
[ 7.682085] printk: console [netcon0] enabled
[ 7.686571] netconsole: network logging started
[ 21.697159] systemd[1]: /lib/systemd/system/[email protected]:3 <mailto:/lib/systemd/system/[email protected]:3>: Invalid interface name, ignoring: sys-subsystem-net-devices-%i.device
[ 22.535048] systemd[1]: Created slice system-phosphor\x2dipmi\x2dnet.slice.
[ 36.643642] ftgmac100 1e660000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx

*[ 49.219280] ftgmac100 1e680000.ethernet eth1: NCSI: No channel found to configure!*

*[ 50.242922] ftgmac100 1e680000.ethernet eth1: Wrong NCSI state 0x100 in workqueue*

[ 998.080088] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:447 dev_watchdog+0x268/0x288

/openbmc/meta-aspeed/recipes-kernel/linux/linux-aspeed/aspeed-g5/defconfig /CONFIG_VLAN_8021Q=y//
/CONFIG_NET_NCSI=y/
/CONFIG_FTGMAC100=y/
/CONFIG_NET_VENDOR_INTEL=y/
/CONFIG_REALTEK_PHY=y//

mac config from aspeed-bmc-facebook-tiogapas2.dts file:
/&mac0 {//
/status = "okay";/
/pinctrl-names = "default";/
/pinctrl-0 = <&pinctrl_rmii1_default>;/
/};//

/&mac1 {//
/status = "okay";/
/pinctrl-names = "default";/
/pinctrl-0 = <&pinctrl_rmii2_default>;/
/use-ncsi;/
/};//

#1 lore raw
Quoted 10 lines · Евгений

Hello Xiuzhi, I have the similar problem using NCSI. Could you explain
more clearly how you have solved one?
I use the Neptune Alfa OpenBMC kit from Portwell. It has an Intel I210 network interface chip. A host is connected to it through PCIe. The host network is working well. BMC has a dedicated interface (MAC0 + RTL8211), it works well.
I want to create a BMC interface via NCSI + VLAN, but it does not
work: see dmesg output below.
Please help me solve this problem.

Quoted 4 lines
:~# dmesg|grep net
[ 2.592506] ftgmac100 1e660000.ethernet: Error applying setting, 

reverse things back

This suggests a clash in the MAC pinctrl request - some other device has already claimed a pin need by the MAC.

Can you provide or link to the devicetree for the the system?

It's odd as some of the subsequent messages suggest the RMII link is working in some instances, but you still get failures in others.

Andrew

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw

Hi Evgeniy,

Please make sure to Reply-All or Cc [email protected] on your OpenBMC queries, that way others can learn from the discussion as well.

Quoted 11 lines · Евгений

Hello Andrew,

It seems that I sent you not quite what you asked. Therefore, I send in the attachment the final

my_neptune_from_dtb.dts

file made from the dtb file with the dtc utility, as well as the dtb itself.

Maybe you will find pin conflicts in it. Thanks for the help anyway.

Had a quick look but it would be better if you could send me the actual source rather than the DTS decompiled from a DTB. Do you have access to it?

Regardless, can you please build and boot a kernel with CONFIG_DEBUG_PINCTRL=y and send me the full dmesg output of the booted system that shows your symptom?

Cheers,

Andrew

#4 lore raw

Hello Andrew,

It turned out that the i210 firmware was not loaded by the manufacturer. Maybe this is normal, but how then did they test the board?

Anyway, thanks for the help.

Sincerely,

Evgeniy

22.11.2019 3:50, Andrew Jeffery пишет:

Quoted 28 lines

Hi Evgeniy,

Please make sure to Reply-All or Cc [email protected] on your OpenBMC queries, that way others can learn from the discussion as well.

Quoted 11 lines · Евгений

Hello Andrew,

It seems that I sent you not quite what you asked. Therefore, I send in the attachment the final

my_neptune_from_dtb.dts

file made from the dtb file with the dtc utility, as well as the dtb itself.

Maybe you will find pin conflicts in it. Thanks for the help anyway.

Had a quick look but it would be better if you could send me the actual source rather than the DTS decompiled from a DTB. Do you have access to it?

Regardless, can you please build and boot a kernel with CONFIG_DEBUG_PINCTRL=y and send me the full dmesg output of the booted system that shows your symptom?

Cheers,

Andrew

#5 lore raw
Quoted 4 lines · Евгений

Hello Andrew,

It turned out that the i210 firmware was not loaded by the manufacturer. Maybe this is normal, but how then did they test the board?

No idea. Glad you solved it. The pinmux issue is a bit disconcerting still.

Andrew