I’ve found that the current method of flashing using l4t_initrd_flash.sh is quite time-consuming, especially during the development phase. It seems necessary to sift through extensive documentation and forum posts to understand the partition layout and the flashing sequence. In my search, I encountered several forum posts suggesting the use of dd for customized flashing. However, while consulting the documentation, I often come across information irrelevant to the Orin Nano, realizing its inapplicability only after reaching the end or encountering instruction failures.
To facilitate a more efficient process, I would greatly appreciate some guidance on the following points:
The configuration of the partition layout for the Orin Nano (production version, with NVMe as storage).
Methods to specifically target and flash individual partitions, rather than re-flashing the entire storage each time.
Additionally, I find the flashing instructions in the web documentation somewhat unclear. I am uncertain whether the existing flashing scripts support these functionalities.
Any pointers or clarifications would be highly beneficial.
The purpose is to understand the current layout and customize it according to our needs.
You want to know what each individual partitions stores?
Yes, I came across the documentation, but it’s not clear whether it’s the same for all models or not. The documentation does not specifically refer to the Orin Nano. ( this doc )
I don’t think this is supported on initrd flash.
Noted. What is the recommended approach for improving the flashing process for targeted flashing?
There are two partition layout files require to flash Orin Nano with NVMe drive.
Linux_for_Tegra/bootloader/t186ref/cfg/flash_t234_qspi.xml for flashing bootloader stuff to QSPI image. Users should not make any changes to this file, so there is no document on it.
Linux_for_Tegra/tools/kernel_flash/flash_l4t_external.xml for flashing the real OS to external devices. For this one, you are supposed to add new partition, instead of deleting any existing ones. Just make sure they are put after the APP partition and before the secondary_gpt partition.
For speeding up the flashing process, once the flashing image is generated, and the same model of Jetson device is used, you can run just