Jetson Nano 2G First Boot - Partition TBC not found

I’m hitting a wall with the Jetson Nano 2G dev board early in TegraBoot with all available stock jetson nano images available to download from Nvidia. I also can’t seem to put the board in Recovery Mode, and thus can’t use the sdkmanager to correct the flash state. While I’m new to the Jetson products, this feels entirely too broken for a product less than a year old. At this point, I’m suspecting the board was flashed incorrectly at the factory or was otherwise corrupted. Please see the serial dumps below:

  1. sdcard with jp45, jp451,jp46 (2G images) as well as r32.1 stock image downloads. All images fail to detect partition layout early in boot with the attached serial log: jetson-nano-tbc_error.log (2.2 KB)

  2. as a comparison, the boot without an sdcard: jetson-no-sdcard.log (17.2 KB)

  3. Jumping ‘FC REC’ with ground before powering on the board results in no serial output to upload, although LED is on. The nano is not detected over the usb port on any of my 3 machines, tested with two separate usb cables… No output or usb events happen even when removing the jumper from ‘FC REC’ after power-on (LED stays on).

I originally wanted to suspect the 2018 version TegraBoot on a 2020 board, but I’ve seen a similar version string on other serial logs posted here. I unfortunately don’t know what a successful boot should look like for this board to compare against.

I attempted to dig into how to make a bootable “bootloader update package” (bup) image, but I’m blocked at how to create a bootable image after generating the bup blob – the documentation seems to assume you already have a bootable linux sdcard to place the bup blobs into…? Please advise.

Please directly flash your board with sdkmanager.

  1. Recovery mode does not care about what software is pre-installed on your board. Thus, we don’t need to study the date or version of the software there. It is just broken.

  2. Generally, the most common problem to new users are “they don’t know how to put recovery mode (RCM) and don’t know how to tell whether a board is in RCM”.

What is the method that you use to tell it is “not detected” ? Also, are these 3 machines able to flash other jetson devices? If they can, but only fails on this Nano 2gb, then I would suspect this is hardware defect.

Entering recovery mode is just a pure hardware triggered event. Every jetson devkit should have such capability. If it cannot, then it would be better RMA this device. However, you should tell us how you checked the device “not detected” first before doing the RMA. Otherwise you next jetson may suffer the same issue and turns out the previous RMA is just in vain.

Thanks for the reply. I am following the instructions from here:
https://docs.nvidia.com/jetson/archives/l4t-archived/l4t-325/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/quick_start.html#wwpID0E0XC0HA

Which line up with the pins described by sdkmanager at the fashing stage of the utility.

Since this is the 2G model, I’m using a jumper between “FC REC” and GND on J12 (carrier board under the SOM), to be explicit. I then power the board and look for any new usb devices (through dmesg and lsusb).

While the behavior is different from normal boot:

  1. I receive no serial output while powering on with “FC REC” and GND jumped (not sure if that’s normal). LED is solid on.
  2. Watching dmesg and checking lsusb shows no sign of the nvidia device (or any new/unknown usb devices) with the host machine connected to the microusb port on J13.

Tested over a dozen times with the primary workstation. While the other 2 linux machines did not have skdmanager installed, I was just looking for any new/unknown usb devices while in recovery mode (to rule out the primary workstation having a usb compatibility issue).

This is under expectation. In RCM, basically the board is just a brick, it won’t have anything from serial console or even the monitor. The log will start to print only after the flash process starts.

Just want to confirm: are all these 3 linux machine ubntu18.04 and not a VM?

Thanks for confirming no serial output is expected in RCM mode.

Only the primary workstation is 18.04. The other two machines are not 18.04. However, I would expect usb device events in dmesg irrespective of ubuntu version, is that incorrect?

Does the jetson nano need to be kicked in some way after entering RCM mode? All documentation suggests that sdkmanager looks for a usb device with the correct identifier. Is there another side channel mechanism in play that could be broken?

If the board is really in recovery mode, then it will be detected as a usb device on your host.

And this is totally a hardware triggered event, thus, if you tried lots of times and different cables all cannot bring it up, I would suggest you can RMA this device.

Thanks for the quick reply. Will RMA.