Trouble flashing the Jetson TX2 (EMMC model) after we forgot the password

Hello, I forgot the password to my jetson board. I am trying to flash it again so that it can reboot and I can create an account again. I used [this website](https://wiki.seeedstudio.com/reComputer_J1020_A206_Flas


h_JetPack/#flashing-jetpack-os-via-command-line) for reference . When I try to unzip the package files and execute the script (./apply_binaries.sh) to install Tegra binaries, I am getting an error which says that the script is invalid and that I should add arguments. I tried doing that and even tried using environmental variables but it is showing the same error. Can anyone tell me what is wrong and what I should be doing? I am attaching the error I am getting below.

Do you have hardware from Seeed? If so, then that is the correct software (I assume you also downloaded from Seeed).

Some explanation might clear up confusion…

During the normal flash a default image is created. That comes from the “sample root filesystem” package, which is purely Ubuntu. When running “sudo ./apply_binaries.sh”, it overlays NVIDIA drivers on top of that. You’d only run apply_binaries.sh once for a given sample rootfs. Running it a second time would attempt to install content which is already there.

Seeed might have modified the sample root filesystem. I don’t know. If they did, then possibly they also modified what is installed from apply_binaries.sh. Don’t know. If Seeed did modify that, then it seems logical they would have also modified arguments (normally one would not use arguments to apply_binaries.sh, but that’s on a purely NVIDIA dev kit).

The default install location of flash software is to produce a directory “Linux_for_Tegra/”. If this was from the NVIDIA SDKM, then you’d have:
~/nvidia/nvidia_sdk/JetPack...version.../Linux_for_Tegra/

Possibly Seeed has this elsewhere? Don’t know.

The apply_binaries.sh file itself is normally always at “Linux_for_Tegra/apply_binaries.sh. If you have this file somewhere else, then I’d expect an error and to have the script tell you about its options. This would be true regardless of whether it is a stock NVIDIA version or a Seeed version (which might be the same, I don’t know). Is apply_binaries.sh in the “Linux_for_Tegra/” subdirectory?

Incidentally, this is applied to content in “Linux_for_Tegra/rootfs/”. That content would need to exist prior to running apply_binaries.sh.

All of the above just prepares the rootfs for flash. There is a separate script for setting up first boot login, and it is at:
Linux_for_Tegra/tools/l4t_create_default_user.sh
(you’d execute that from “Linux_for_Tegra/” via “sudo ./tools/l4t_create_default_user.sh”)

Note that I don’t know what modifications Seeed makes, so this might not be 100% correct for you.

Hello, I rechecked your suggestions and I think it could be an issue with the apply_binaries.sh. It has a function ShowUsage which is running in a while loop that can cause errors. It probably occurred after some modification like you suggested.

Thank you so much!

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