socket CAN switching between Xavier A & B

Hi,
we’re currently seeing issues with the socket CAN routing of CAN 2 & 6. (on DriveOS 9.0) Sometimes those get routed to Xavier B (verified using

candump can0/can1

), however after rebooting, this somehow got routed back to Xavier A. At some point we ran

ip link set can0 type can bitrate 500000

, not sure if that had a influence on that, but we restarted the device several times since then.

Thanks in advance,
Sebastian

Hi sebastiand3vrm,

According to below table (from https://developer.nvidia.com/DRIVE/secure/docs/DRIVE_Mech_Inst_Guide_DI-08957-001_v07.pdf ), CAN 2 is accessible to either Xavier A or B.

Did you mean not running candump on xavier B you still observed candump on xavier A not getting data?

yes, exactly that’s what we’re seeing!

Could you provide the steps of reproducing the issue? If not issuing “ip link set can0 type can bitrate 500000”, is it still reproducible?

Currently, after switching it on, without issuing the “ip link …”, we observe the following:

  1. “candump can0” @ Xavier A --> No output, exiting process
  2. “candump can1” @ Xavier A --> No output, exiting process
  3. “candump can0” @ Xavier B --> CAN traffic received, exiting process
  4. “candump can1” @ Xavier B --> CAN traffic received, exiting process

We have two CAN buses connected. It previously worked though, only starting today, we’re seeing this issue.

After “sudo reboot” on Xavier A, candump works from both Xaviers.

Rebooting though makes the GMSL cameras (Sekonix) attached not accessible.

For some reason, this issue only popped up today, eventually earlier noticed it but linked it to cabling issues.

“exiting process” here is normal or abnormal?

Did you mean CAN2 and CAN6 are connected together or they are connected to different CAN devices?

I mean manually exiting candump with Ctrl-c

I mean we’re using both CAN ports (2&6) that are routed to socket CAN (can0 and can1)

I don’t really have any ideas on this. It sounds candump at Xavier A after cold boot isn’t working but after reboot is working.

any clue from kernel log of cold boot and reboot?

It’s fixed now, we accidentally connected an USB CAN device which were assigned to CAN 0 & 1 and thus didn’t receive any CAN traffic. Thanks for help anyway!

Best,
Sebastian

Good to hear your issue was clarified!