Jetson AGX Xavier - clone issue. Black screen

Hello everybody,
I would like to consult with you guys regarding the clone-restore process of Jetson AGX Xavier unit.
I use Ubuntu 18.04 machine with SDK Manager.
We have working Jetson AGX Xavier unit with Linux 18.04 and our application suite installed.
I am trying to create the image from the working device and extract it to another Jetson AGX Xavier unit.

Some info:
I succeeded to preform the fresh installation by flash.sh utility exactly as explained in NVIDIA site.
I am able to create an image (by dd command as well as by flash.sh utility) of our working unit and restore it on the new one. All phases are completed successfully without any errors.

Problem:
A new cloned unit freezes at startup. The last row of the startup screen is:
[ OK ] Started Process error reports when automatic reporting is enabled.
By the way, all lines of the startup screen are started with [ OK ].

Some lines from Xorg.0.log file:
(EE) NVIDIA(GPU-0): Failed to initialize the NVIDIA graphics device!
(EE) NVIDIA(0): Failing initialization of X screen 0
(EE) Screens found, but none have a usable configuration
(EE)
Fatal server error:
(EE) no screens found(EE)

I use the same Monitor and HDMI cable for both the source and the target units.
I use the same SDK Manager and flash.sh version for clone/restore.

It looks like I miss some pre- or post-installation steps.

Could you guys help me to figure the issue out? Thank you very much in advance.

AGX is an eMMC model, and this has many partitions. Those non-rootfs (non-APP) partitions are used for booting, and much of this is eventually inherited by the Linux kernel once past the bootloader stages. If that other content does not match in the new AGX, then what the rootfs image expects may not work with that other content.

Note that using the “flash.sh” “-r” option to reuse an image flashes everything, but it uses your rootfs clone for the rootfs/APP. If for some reason you added flash options to just flash the rootfs, and the other partitions are not an exact match to the old system, then failure is expected.

To see the L4T release you can use “head -n 1 /etc/nv_tegra_release” on the working unit which is the donor of the rootfs.

Here is a URL for the list of various L4T releases:
https://developer.nvidia.com/linux-tegra

Here is a URL for the list JetPack/SDK Manager releases (which flash a particular L4T release):
https://developer.nvidia.com/embedded/jetpack-archive

Note that you can see older releases in JetPack/SDKM if you start SDKM like this (you can uncheck all options but L4T flash, and then say to download only, then use the command line flash.sh):
sdkmanager --archivedversions

Use the exact L4T release from the donor to completely flash the recipient, and only the “-r” option to reuse a donor clone rootfs.

You can try to attach the full xorg log instead of partial log.

linuxdev, thank you very much for the detailed explanation. I will try as soon as I get access to the board.

WayneWWW, thank you so much for the reply.
Full xorg log is attached.Xorg.0.log (17.8 KB)

I am not sure if you ever read the log you just attached.

In the first comment, you said there is

“(EE) NVIDIA(GPU-0): Failed to initialize the NVIDIA graphics device!
(EE) NVIDIA(0): Failing initialization of X screen 0”

But now I don’t see these logs.

Please make sure you are talking about the same error and the log is under same scenario first.

Xorg.0.log (7.0 KB)

I apologize for uploading the wrong log file.
Thank you very much again.

Do you have dmesg to share?

Unfortunately I will not have access to the unit for 3 or even more days.

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

Is this still an issue to support? Any result can be shared?

Thanks

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