Is there any way to get a .iso version of the Jetson Xavier Nx? I would like to have a customized version using Cubic, but it only downloads a sd-blob.img, which won’t work. Could someone help me? Or provide the steps to create a customization image? Thank you!
No, there is no iso image.
We provide tools for creating customized image in the driver package but again, it is not iso image.
I don’t know Cubic or what it expects (I can’t even find a good link for it) but you can create an ISO of the NX root filesystem pretty easily, well, on Linux anyway.
Follow the instructions at Tegra Linux Driver Package Development Guide - Setting Up Your Filesystem to create a root filesystem, then run mkisofs or genisoimage to create an ISO file.
Be aware though that the ISO will only be the root filesystem. You can’t use that alone to commission a new NX. If you can describe what Cubic does or better yet, post a link to it, maybe I can help further.
Sounds good! Could you redirect me or provide the steps to utilize the tools in the driver package in order to create the customized image? Thank you!
Thank you so much for replying to my post and explaining the ISO limitations with the nx. I will run mkisofs, anyways and explore those limitations. However, I do need to have a customized image with my own installed packages to be able to use in my other Xaviers.
I was trying to use Cubic: Cubic in Launchpad
as it works very well on a PC.
Ah, ‘Customized Ubuntu…’ . I’m not a ubuntu guy so that explains why I didn’t recognize it. :)
How do you plan on running your NXs? From the SDcard, nvme or from USB storage? Or are these production modules with the eMMC storage on board? How about partition layout? Are you going to partition them like a desktop where the ‘/boot’ and ‘/’ and maybe ‘/home’ partitions are separate or are you going to use one partition for all? Finally, which Linux for Tegra or SDK Manager/Jetpack version are you using?
Please check jetson-disk-image-creator.sh in the L4T driver package.
I ran the
sudo ./jetson-disk-image-creator.sh -o customNX.img -b jetson-xavier-nx-devkit command and it gives me this error:
However, I looked in the bootloader directory and both the tegraflash.py and tegrasign_v2 files are present.
Would this be a path error somewhere? Thank you!
Yes, sorry I forgot to mention that in the original post! :) I plan to run my customized .img from a USB storage device and use one partition for all. I am using the image creator provided in the Tegra186_Linux_R32.5.1 tools. I posted my update ^ . Thank you so much for helping me with this, I really appreciate it!
What is your environment to run this tool? And what is your method to prepare the driver package?
Hello, I’m using Ubuntu 18.04 bionic. I’m sorry I’m not sure about the ‘method to prepare the driver package’. I am new to this, I am trying to have a live installer for 50 Xavier NX boards for UNM students and we want to just be able to provide the .img, similar to the JetPack installer since it would be easier to provide with our own customization. The sizes for the sd cards will be different too, so getting a customized .img version will be great.
The error log you posted looks like there are some binaries missing.
Thus, I would like to know where did you get this Linux_For_Tegra package.
Thank you. Yes, it tells me the tegrasign_v2 is missing although I checked and it is present in the bootloader directory. I got the Linux_For_Tegra package from the Nvidia L4T Driver Package for Xavier NX 32.5.1 here: L4T Driver Package (BSP)
I am just wondering one thing here. Is the ubuntu 18.04 here just a jetson NX itself?
I flashed my Xavier NX with the
Jetson-nx-jp451-sd-card-image.img, then installed some modules as I want it customized, and now I’m trying to get the new custom image using the image-creator.sh :)
In fact, the driver package is for x64 host machine to run. Not arm64 device like jetson.
This is how jetson series works… we prepare the image on the host and then flash it to the device side.
Ah, got it. Does that mean that I have to connect my Xavier NX board to a PC, then download the driver package Linux_For_Tegra and use the image-creator.sh from there?
Yes, actually that is the common way for other users here who are making their own products.
If you heard sdkmanager before, that tool is same as the driver package. Need to run on a x64 18.04 host.
As for SDcard image, it is just a entry level tool for newbies.
Sounds good, thank you very much! I have copied the root files from the customized image to the rootfs directory and I was almost successful at creating the new image. However, now it is not finding this other file:
Do you know why it might be missing?
It might be that your root filesystem is bigger than the expected 32G.
It would help if you could capture the entire output instead of just a screen capture.
If you copy it from your terminal, you can then click the
</> button in the post toolbar then just paste the text.