Xavier NX going into bootloader shell mode at power up

On our custom carrier board, we use a xavier nx booting off of an M.2 NVME. Sometimes at power up, the NX goes in to a bootloader shell mode instead of automatically booting off of the NVME.

While in this mode, we can connect to the UART and get the boot process so continue.

Some questions:

  1. What are the factors that would cause the module to go in to a bootloader shell mode?
  2. Is there a timing condition in which the NX module tried to boot off of the NVME but the NVME is not powered in time so the module gets stuck in a bootloader shell mode?

Not quite sure about your question. Could you share the log when you enter shell mode?

Which jetpack release is that?

You’d have to show the serial console log to know where it stops, but consider this: There are two places where, if you strike any key on serial console, the boot pauses. The first place is when you are offered a chance to enter the serial console command line bootloader environment (where you’d type the command “boot” to continue). The other is to pick which kernel image. If the voltage level presented to the serial console is such that it thinks there is a stray keystroke, then it’ll halt just like it is supposed to. Example: Leakage current or incorrect high/low default state. Then attaching a serial console would allow it to function normally and you’d never know that it was the actual serial console causing the stoppage.

We are using bootloader from 4.6.1, but we are using the 4.6 OS and file system.

We will work to try and get the logs, but from what we have seen, when the UART is connected to the device we do not enter the bootloader shell. We have observed this only when we are not connected to the UART interface.

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Could you attach a share screen here so that I can know what is “bootloader shell”?

I am not sure if that is what we saw from some jetson nano issue.

I’ll also suggest you put an oscilloscope on the UART TX and RX pins to see if it is the same voltage level during halt as when you do or don’t have the UART connected.