Target often fails the rcmboot

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin NX
1

Host: Ubuntu VM 20.04

Target: Jetson Orin NX 16GB

So I have tried many ways of flashing until now, be it by NVIDIA SDK manager GUI, by the script nvsdkm_flash.sh or by the script l4t_initrd_flash.sh. all taken on the same setup, but when flashing by script alone I notice that gradually the rate in which my target failed to reboot into initrd mode (to start flashing) increases.

When I first saw the issue it was just 1 out of 3-4 trials, but currently the fail rate has been as high as 70-80% but sometimes I still have 2-3 consecutive successful flashes.

Here are the logs for a typical failed trial.

flashing_rcmboot_failed.txt (11.1 KB)

putty_rcmboot_failed.txt (24.8 KB)

When it failed, the device usb will obviously not reconnected as “Linux for Tegra”, but just show error like this screenshot

USB_status
USB_status609×876 83.1 KB

Please support to check. Thank you very much

5

Hi,

We recommend using a native Ubuntu environment for flashing, since many users have reported USB detection issues when using a virtual machine.

Also help us to focus on whether the issue exist.

Thank you.

6

Dear,

I know that VM cannot compare to Native Ubuntu regarding stability. However I’m currently not given a choice to use native Ubuntu, hence trying to make it work with VM.

And from what I’ve experienced that my setup was pretty stable 1-2 months ago, where I can achieve 95-100% success rate (even when running in a Docker container situated inside the host VM), but it’s just keep decreasing ever since without any clear reason. That’s why in this topic what I want to achieve is not a way to make it 100% again, but to at least know where in the log may show some hint about what’s gone wrong.

Can you have a look into the logs and have some verdicts? I would be very appreciated.

Thank you.

7

Hi,

Thank you for the information.
From the log, it still appears to be an issue with VM detection. Could you try using VMware instead?

Some users share their successful experience about it.

Thanks.

8

Dear,

Thank you for your reply

Yes I also had many tests on VMWare Workstation (on a different PC) as well, and the same issue still persist, originally VMWare Workstation achieves 100% successful rate, but it also keeps declining.

Actually I’m aware about how VMWare is more preferred than VirtualBox thanks to this post: Let's make flashing from virtual machines reliable - #4 by cyato , but VMWare only handles the USB disconnection-reconnection process smoother, whereas in this case the issue is that the device was not even be able to reboot into initrd (hence there’s no new USB device to be detected by the VM)

Do you think somewhere in the script may I need to add/extend some delayed period when the instruction to rcmboot starts?

I’ll try looking into the links in your shared topic to find more hints