Failed to flash nx xavier

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

Hi, I have just flashed 40 nx xavier devices, where 39 of the 40 was flashed succesfully, but the last one is not flashing correctly. When running the flashing process in jetpack 5.1 (as docker), it terminates with the following errors:

debug: running command < true >
info: exec_command: true
error: shell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directory
info: command finished successfully
debug: using adapter to install NV_L4T_DRIVERS@JETSON_XAVIER_NX_TARGETS to /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS/
info: Found file /home/nvidia/Downloads/nvidia/sdkm_downloads/nv-tensorrt-local-repo-l4t-8.5.2-cuda-11.4_1.0-1_arm64.deb with correct checksum, skip downloading.
info: [host] [ Disk Avail on Partition overlay: 185.12 GB ]
info: Installing component ‘Drivers for Jetson’ (NV_L4T_DRIVERS)
info: NV_L4T_DRIVERS include 2 package
info: change working directory to /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS
info: [ Package PreInstall Started ]
info: current working directory is /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS
info: exec_command [host]:
info: **********************
info: sudo rm -rf /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS/Linux_for_Tegra/
info: **********************
info: exec_command: /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh
info: [ Package PreInstall Finished Successfully ]
info: [ Package Install Started ]
info: current working directory is /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS
info: exec_command [host]:
info: **********************
info: tar xf /home/nvidia/Downloads/nvidia/sdkm_downloads/Jetson_Linux_R35.2.1_aarch64.tbz2
info: **********************
info: exec_command: /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh
info: Found file /home/nvidia/Downloads/nvidia/sdkm_downloads/cudnn-local-tegra-repo-ubuntu2004-8.6.0.166_1.0-1_arm64.deb with correct checksum, skip downloading.
info: Found file /home/nvidia/Downloads/nvidia/sdkm_downloads/Tegra_Linux_Sample-Root-Filesystem_R35.2.1_aarch64.tbz2 with correct checksum, skip downloading.
info: Found file /home/nvidia/Downloads/nvidia/sdkm_downloads/cuda-repo-l4t-11-4-local_11.4.19-1_arm64.deb with correct checksum, skip downloading.
info: [ Package Install Finished Successfully ]
info: change working directory to /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS/Linux_for_Tegra/
info: [ Package Install Started ]
info: current working directory is /home/nvidia/nvidia/nvidia_sdk/JetPack_5.1_Linux_JETSON_XAVIER_NX_TARGETS/Linux_for_Tegra
info: exec_command [host]:
info: **********************
info: unzip -o /home/nvidia/Downloads/nvidia/sdkm_downloads/nvsdkmanager_flash.zip
info: **********************
info: exec_command: /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh
error: /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh: line 3: unzip: command not found
error: [exec_command]: /bin/bash -c /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh; [error]: /home/nvidia/.nv sdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh: line 3: unzip: command not found
info: [ Package Install Finished with Error ]
info: [host] [ 690.78 MB used. Disk Avail on Partition overlay: 184.44 GB ]
info: [ NV_L4T_DRIVERS Install took 38s ]
error: command error code: 11
error: command terminated with error
info: Installation failed.
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component
info: Depends on failed component

3

Hi kaare.steensig,

Are you using the devkit or custom board for Xavier NX?
Have you also tried with the latest JP5.1.4(R35.6.0)?

error: /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh: line 3: unzip: command not found
error: [exec_command]: /bin/bash -c /home/nvidia/.nvsdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh; [error]: /home/nvidia/.nv sdkm/replays/scripts/JetPack_5.1_Linux/NV_L4T_DRIVERS.sh: line 3: unzip: command not found
info: [ Package Install Finished with Error ]

From the error you shared, it seems there’s the issue from host rather than flashing.

Please share the full flash log as file here for further check.

4

I am using a custom board.

When I’m trying with latest JP5.1.4(R25.6.0), it continues further, but here it runs into problems when trying to locate the device. When checking with lsusb command, it appears as NVIDIA Corp. APX. (Ihave followed the instructions in order to boot in force flash mode)

How can I get the full flash log, when using the docker version of the sdk manager? (the --export-response-file argument does not do it)

5

SDK manager may not work for your custom carrier board since it use the default official BSP pakcage.
For the custom carrier board, please request the custom BSP package from your vendor first and use command to flash your board.

6

I misunderstood one of my colleagues. It is a development board. Sorry for the confusion

7

Do you confirm that it is an NVIDIA official Xavier NX devkit rather than custom carrier board from 3-rd party?

If you are using the devkit, please use SDK manager to flash the latest JP5.1.4(R35.6.0) and check if it could be flashed.
And share the full flash log as file here if you still have flash issue.

8

Yes, it is an NVIDIA official Xavier NX devkit.

When I tried to flash it using the docker version of SDK manager, it failed as earlier stated:

How do I get the full flash log from the docker version of the SDK manager? - the --export-response-file argument does not do it

9

Have you tried using standalone Ubuntu host PC with SDK manager to flash your board?
The log should be able to export from GUI.

Is your devkit with SD or eMMC module?

10

It is with SSD.

No, I have tried to use the docker version, since I am running ubuntu 22.04, but in order to get the right jetpack version, I need ubuntu 18.04 or 20.04, which I believe the docker versions is for?

12

I tried some more and eventually got it to flash from ubuntu 18.04 with Jetpack 5.1.4.

But I still can’t get it to flash the way, that I need to in my use case.
I have tried uninstalling the version of Jetpack, that I installed with SDK manager and then I tried to flash the way, I need to do in my use case (and as I have done for several others succesfully).
When reading the serial connecter, while flashing, I now get the error:

[ 31.452316] **************************************
[ 31.457121] CPU:0, Error: cbb-noc@2300000, irq=14
[ 31.462043] **************************************
[ 31.466757] CPU:0, Error:cbb-noc
[ 31.470014] Error Logger : 1
[ 31.473234] ErrLog0 : 0x80070100
[ 31.476816] Transaction Type : RD - Read, Incrementing
[ 31.481982] Error Code : DEC
[ 31.485652] Error Source : Initiator NIU
[ 31.489682] Error Description : Address decode error
[ 31.495190] Packet header Lock : 0
[ 31.498608] Packet header Len1 : 7
[ 31.502540] NOC protocol version : version >= 2.7
[ 31.507352] ErrLog1 : 0x12000c
[ 31.510419] ErrLog2 : 0x0
[ 31.513565] RouteId : 0x12000c
[ 31.516719] InitFlow : ape_p2ps/I/ape_p2ps
[ 31.521357] Targflow : axis_satellite_grout/T/axis_satellite_grout
[ 31.527875] TargSubRange : 0
[ 31.531039] SeqId : 0
[ 31.533697] ErrLog3 : 0xa00080
[ 31.537193] ErrLog4 : 0x0
[ 31.539741] Address accessed : 0xa00080
[ 31.544017] ErrLog5 : 0x358fcf0
[ 31.547171] Non-Modify : 0x1
[ 31.550842] AXI ID : 0x6
[ 31.553564] Master ID : APE
[ 31.556270] Security Group(GRPSEC): 0x3f
[ 31.560471] Cache : 0x0 – Device Non-Bufferable
[ 31.565978] Protection : 0x7 – Privileged, Non-Secure, Instruction Access
[ 31.573327] FALCONSEC : 0x0
[ 31.576390] Virtual Queuing Channel(VQC): 0x0
[ 31.580860] **************************************

This error seems similar to the one at https://forums.developer.nvidia.com/t/cpu-0-error-cbb-noc-errors-after-secureboot-enabled-xavier-nx/280507, except the address. Should I reinstall Jetpack 5.1.4 as I am now able to and then follow the suggestions, or is there another way, when trying to flash it from new?

13

SSD should be the external device.
What I mean is about the module. Is that Xavier NX module coming with internal eMMC or SD?

JP5 should also work with Ubuntu 22.04 and we don’t suggest using docker for SDK manager.

Please verify with the latest JP5.1.4(R35.6.0) and share the full flash log and serial console log if you still have flash issue.

14

It is with internal eMMC.

Following your documentation, you need either ubuntu 18.04 or 20.04 in order to flash JP5 (look at the table under the headline “Host Operating System Compatibility Matrix” at https://developer.nvidia.com/sdk-manager)
If I try to flash with SDK manager from ubuntu 22.04, I am only able to choose JP6 (because of this problem you have also made three docker-version of SDK Manager available, when looking at the download site).

I will return, when I have tried again.

15

You can also refer to Quick Start — NVIDIA Jetson Linux Developer Guide 1 documentation to download BSP package and sample rootfs to flash the board through commands.
For the custom carrier board, you have to refer to Jetson Module Adaptation and Bring-Up: Jetson Xavier NX Series since the default BSP package may not work on your custom board.

17

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