Xavier Industrial PCIe Errors

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Xavier
,
1

Hi,
We have developed a custom carrier board for the Xavier Industrial Module. Three of the PCIe ports are used. One of the PCIe port(the one which has x4 lanes), is connected to a PCIE Switch(PI7C9X2G608GPBNJEX). There are four x1 lane PCIe Endpoints at the other side of the switch.

When I add a PCIe-Ethernet Adaptor to one of the endpoints, the system starts to give some error messages on the serial console but rarely. However when I attach the ethernet cable to the endpoint, it starts to give lots of error messages on serial console. Also USB ports are being disabled randomly. In most of the cases the module becomes locked. And sometimes it restarts.

Note: The endpoint is listed in the list when I check it with lspci command.

I have attached one of the opening log output.
xavier_pcie_error_log.txt (583.3 KB)

JetPack version is 5.1.2.

What can be the reason of this problem?

Best Regards,

3

Hi,
I did some search on the internet about the following error messages .

  • BadTLP
  • BadDLLP
  • AER: Multiple Corrected error received: 0000:02:01.0
  • PCIe Bus Error: severity=Corrected, type=Data Link Layer, (Receiver ID)

There are a lot of cases where platforms and devices are totally different from our case but the common things are as following.

  • The architecture is aarch64,
  • The operating system is Ubuntu,
  • The problem is related to the NVIDIA drivers,

And the common solution suggested was to set following kernel options.

  • pci=nommconf (It didn’t solve when I only added this option to the extlinux.conf file and reboot)
  • pci=noaer ( This only disables AER, but I don’t try it, because I am not sure if it is true to just disable error reports)
  • pcie_aspm=off (I have searched and concluded that ASPM is Disabled default in JetPack 5.1.2)

I am still trying to solve this issue. I will be appreciated if someone helps me about this issue.

Best Regards,

5

Please check if you can have pcie analyzer and dump the trace.

7

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.