Downgrade from Jetpack 6 to jetpack 5

Continuing the discussion from How do I downgrade from Jetpack 6DP to JetPack 5 on Jetson Orin Nano Dev Kit (SD Card):

Very recently purchased a new Jetson Orin Nano developer kit (about a week ago), and then as prompted followed the guide to get started. Unfortunately, the upgrade to Jetpack 6, also upgrades the Ubuntu version from 20.04 to 22.04, which I don’t require. I had tried for multiple hours since to downgrade using both the SDK and jetpack images, failing each time. Most likely due the the change in UEFI firmware change.

Just to let you know, the host PC running the SDK is Ubuntu 20.04.

As a result, I found the above linked discussion, and it seems to be a reasonable solution for me. However, i have a number of questions, to add to the discussion as I am still unable to downgrade to Jetpack 5:

  1. Do i need to purchase a USB to UART lead in order to carryout the ./tools/kernel_flash/l4t_initrd_flash.sh operation.

  2. I do not understand what packages i need to download i.e Jetson Release release package and sample file system from the linux-tegra webpage? Could you outline exactly what is I need to download please. I think i’ll require any 35.x.x for ubuntu 20.04

  3. Whilst I realise this might be unconventional, but I have been desperate to downgrade ASAP. I have tried using the l4t_initrd_flash.sh, that is provided as part of the SDK.
    Using the following command:
    sudo ./nvidia/nvidia_sdk/JetPack_5.1.3_Linux_JETSON_ORIN_NANO_TARGETS/Linux_for_Tegra/tools/kernel_flash/l4t_initrd_flash.sh --external-device sda1 -c tools/kernel_flash/flash_l4t_external.xml -p “-T 62333952 -c bootloader/t186ref/cfg/flash_t234_qspi.xml” --showlogs --network usb0 jetson-orin-nano-devkit internal

Whilst it carried out some part of the download. It also provided the following error:


  •                                    *
    
  • Step 2: Generate rcm boot commandline *
  •                                    *
    

ROOTFS_AB= ROOTFS_ENC= /home//nvidia/nvidia_sdk/JetPack_5.1.3_Linux_JETSON_ORIN_NANO_TARGETS/Linux_for_Tegra/flash.sh --no-flash --rcm-boot jetson-orin-nano-devkit mmcblk0p1
Error: Invalid target board - jetson-orin-nano-devkit.

Is this part of the reason I cannot downgrade?

  1. Is there any plan to simplify the downgrading from Jetpack 6 to Jetpack 5 in future SDK editions?

  2. Could the ‘Initial Setup guide - Jetson Orin Nano’ webpage be changed to identify that changing to Jetpack 6 in the method given, will result in changing to Ubuntu 22.04 also?

Many thanks in advance for any assistance you can provide me. I really have been at this for days.

1 Like

It’s just for debug purposes. So if you don’t get any error then it’s not required.

These two:
https://developer.nvidia.com/downloads/embedded/l4t/r35_release_v5.0/release/jetson_linux_r35.5.0_aarch64.tbz2
https://developer.nvidia.com/downloads/embedded/l4t/r35_release_v5.0/release/tegra_linux_sample-root-filesystem_r35.5.0_aarch64.tbz2

You run this command under Linux_for_Tegra so it does not report invalid target board

NO.
There is no downgrade at all. It’s the same as you flash any other BSP version.

This is stated very clearly:

It includes Linux Kernel 5.15, an Ubuntu 22.04 based root file system, a UEFI based bootloader, and OP-TEE as Trusted Execution Environment.

Thank you for your support. I will try it tonight

Many thanks, really appreciated.

OMG it worked wonderfully. Thank you ever so much.

Can I please ask one further question?

The reason for my purchase is I have a project running ROS Noetic, which uses Ubuntu 20.04, that’s the reason for my need to use Jetpack 5. Initially after the upgrade, I tried using a docker container to hold ros-noetic-desktop. Unfortunately, I could never expose the GPU to the container, as Jetpack 6 used (NVCC) version12.2, whilst the Ubuntu 20.04 container used NVCC 11.2. (I think the versions are correct). Anyway, I suppose due to the differences between the NVCC versions, the container never connected with the physical GPU.

Is it actually possible in Jetpack 6, with a Ubuntu 20.04 container to expose the GPU?

Sorry for the additional question.

But once again, thank you your suuport on my initial issue.

I think it can be better tracked in a separate topic.

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