SARBE - iMXS6X-SDB: ethernet - Link is Down

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

SARBE - iMXS6X-SDB: ethernet - Link is Down

2,262 Views
pranavmondhe14
Contributor I

Hi Team,

i am working on iMXS6X-SDB board with our custom linux kernel image version 3.14.28. I have some python reboot tests which gives the reboot command via ssh, waits for some time and this cycle repeats.

While running those tests, when serial port was connected(A9), i saw below log trace on the console before eth1 dropped off the network:

libphy:2188000.ethernet:02 - Link is Down

and the board looks completely in hang state with no ethernet LEDS glowing (orange and Red color -nearby ethernet connector).

Also, serial port looks down - since also i cannot login inside the board to see more information inside dmesg and/or /var/log/messages.

Please let me know if you need any further details. I am not sure why eth1 link drops off after its power cycled multiple times.

This is seen when its power cycled over several hundreds times (though it may fall at any random attempt).

 

Note- that Sabre board is connected to gigabit ethernet switch and from python (running on PC desktop) is connected to gigabit ethernet switch via USB-gigabit adaptor in between creating a local network.

pastedImage_2.png

0 Kudos
9 Replies

1,700 Views
pranavmondhe14
Contributor I

Hi kunalkotecha

I am working on Sabre iM6X Solo board where i have customised linux kernel and not on any EVK.

I cannot find any file named - "arch/arm/boot/dts/imx6sx-sdb.dts" inside my Kernel source tree.

How i can find this ? I have customised kernel image given to us by a vendor.

 

But yes, eth1 keeps on dropping  after several power reboots.

Best Regards

Pranav Mondhe

0 Kudos

1,700 Views
kunalkotecha1
Senior Contributor II

Hi pranavmondhe14@gmail.com,

In this case I would suggest to contact to your vendor for further help.As on EVK the same is not reproducible and you have  customised images.

Regards,

Kunal

0 Kudos

1,700 Views
pranavmondhe14
Contributor I

Also, when the eth0 interface is down, we see below log trace on serial port - (and later on serial port is not accessible)

Poky (Yocto Project Reference Distro) 1.7 imx6sxsabresd /dev/ttymxc0

imx6sxsabresd login: hrtimer: interrupt took 181667 ns

I always gets above message on serial port when eth1 interface is down after reboot multiple times.

0 Kudos

1,700 Views
pranavmondhe14
Contributor I

Hi igorpadykov‌ 

Thanks for your time for going through my question.

My senior engineer at my company tried as per your suggestion, below is the replied i got from him:

A couple of things:

  • The capacitor I added doesn’t extend the output reset pulse, it delays detection of the reset input signal!  I now have to hold the reset button for 2.7seconds before a reset is generated – I’ll take the cap off again.  I don’t think I can extend the reset duration as the reset controller chip has a fixed reset pulse duration of 210ms (140ms – 280ms).
  • The Watchdog output from the i.MX6 IS high following a power on, then goes low when a reboot command is issued.  It then stays low and never returns high.  This means that a 210ms reset pulse is generated the first time a reboot command is issues after a power on, but isn’t issued subsequently.

Can you please have a look at above two points made by him.

Can you please suggest how we look into this further?

0 Kudos

1,700 Views
kunalkotecha1
Senior Contributor II

Hi pranavmondhe14@gmail.com,

Can you please send the dts node information related to watchdog? You can try with kernel version to 4.9.88_2.0.0_ga and check if this is produced in latest version or not. As it has fix related to the watchdog. 

Regards,

Kunal

0 Kudos

1,700 Views
pranavmondhe14
Contributor I

HI kunalkotecha

Thanks for your time looking my issue mentioned.

Can you please tell me how do we obtained the dts node information related to watchdog?

We are using Linux version 3.14.28-1.0.0_ga+g91cf351. Do you think is worth upgrading to kernel version to 4.9.88_2.0.0_ga ?

We still are having issues with eth1/eth0 random link interface going down.

Best Regards

Pranav Mondhe

0 Kudos

1,700 Views
kunalkotecha1
Senior Contributor II

Hi pranavmondhe14@gmail.com,

Are you using custom board or EVK?As I have rebooted the EVK few times and able to get the IP address for both the ethernet ports everytime. If you are using EVK then you should give a quick try to kernel version 4.9.88_2.0.0.

Also you would be able to see the dts node information related to watchdog in kernel source "arch/arm/boot/dts/imx6sx-sdb.dts".(If not in this file, look into include files used in this file)

Regards,

Kunal

0 Kudos

1,700 Views
igorpadykov
NXP Employee
NXP Employee

Hi Pranav

one can try to increase C383 so MR of STM6779YWB6F was hold longer time.

Best regards
igor

0 Kudos

1,700 Views
igorpadykov
NXP Employee
NXP Employee

Hi Pranav

one can look at reboot issue comments on i.MX6SX Sabre SD SPF-27962 schematic,

in particular one can try to tweak RC delays of STM6779YWB6F component, so

time when board was powered off was increased. This will allow to fully discharge all board

capacitors and may improve reboot stability.

IMX6SOLOX-SABRESDB-DESIGNFILES

Design files, including hardware schematics, Gerbers, and OrCAD files.

Best regards
igor
-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

0 Kudos