This always resulted in the device booting from the emmc.
I always thought you decide which device is booted into by setting the argument in the flash command.
But now I see that during flash, the argument is set to mmcblk0p1 despite it being set to nvme0n1 in the command.
Could it be that I do not influence the boot order by any parameter I append to the flash.sh script?
To me it now seems like the mmcblk0p1 always is preferred to the nvme?
Thus Iād have to erase the esp partition from my emmc to actually boot from the NVMe?
I quite expected that the flashing process would handle this?
Afaik the production modules will ship with a preinstalled emmc image.
So if we flash the qspi and the NVMe in factory we have to attach every single board with serial console and enter the UEFI to change the boot order?
Use l4t_initrd_flash.sh to flash nvme only according to workflow #1
ā The system boots from emmc
Test 2
Use flash.sh to flash qspi and emmc
Use l4t_initrd_flash.sh to flash qspi and NVMe according to workflow #4
ā The system boots from emmc
After both, go into UEFI, change the boot options to have NVMe on top.
ā The system boots from NVMe
For Test 1 this is expected.
For Test 2 I would expect the flash tool to preconfigure the UEFI boot options so that modules can be used out of the box after flashing without the need to attach them to a serial console.