Tegra B sometimes hangs/locks when rebooting Tegra A

We have a few DPX2’s. They all run Drive 5.0.5.0bL, which is the latest release to this date. They all have a strange problem: sometimes, when we reboot Tegra A (sudo reboot), it will hang/lock Tegra B. We cannot ssh into it in any way. Not via the external ip (e.g. 10.10.172.72 in one setup), nor via the internal vlan ip (10.42.0.29). I believe it sometimes also happens the other way around: rebooting Tegra B can hang Tegra A. But of the latter I’m not 100% sure, because we mostly use Tegra A.

Our systems are often remote, so turning it off/on physically is not an option.

Of course I’m interested in understanding and solving the root cause. But a workaround would also be very welcome. One workaround is to plug the host-to-host usb A to usb A cable in Tegra A usb and connect it to the USB2 debug hub. Then we can connect via minicom to the aurix shell and reset tegra B. A rather ugly approach, as it needs the USB A to A cable to be permanently connected between Tegra A and the debug hub. Also, if rebooting B hangs A, we have no solution, as we cannot go into aurix shell from B, only from A.

Some other ideas:

  • is it possible to log into the aurix shell from Tegra A or B without the USB A to USB A cable?
  • is there some other way to reset tegra B from tegra A? e.g. using the TACP protocol? I tried searching documentation, but I couldn’t find API documentation of libtacp.so.

Dear c.j.lekkerkerker,

We will look into this symptom and update. Thanks.

Dear SteveNV,

Thanks, I hope you’ll find a solution. If I could supply you with any logs to support you, please let me know.

A workaround will be very helpful in the mean time. Could you look at these suggestions:

  • is it possible to log into the aurix shell from Tegra A or B without the USB A to USB A cable?
  • is there some other way to reset tegra B from tegra A? e.g. using the TACP protocol? I tried searching documentation, but I couldn’t find API documentation of libtacp.so.

Dear c.j.lekkerkerker,

We’ve posted new SDK on NVIDIA DRIVE Solutions for Autonomous Vehicles | NVIDIA Developer.
Could you please check if this symptom is reproduced after updating with new PDK and DriveWorks?
I couldn’t reproduce this symptom after updating new PDK. Thanks.