Jetson Orin Nano boot failure with temperature dependency

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
,
1

Hi,

I have encountered an issue with boot process of Jetson Orin Nano. noticed that if the Tj temperature is slightingly lower, the boot fails on “E> tos: digest on binary did not mat”.

I tried experiment at 5C ambient Vs. 50C ambient. tried to boot the same unit at different ambient temperature. I had 15 fails at 5C ambient and 15 successes at 50C ambient.

Some facts:

  1. I have a custom carrier board
  2. I have checked power up sequence + 5V - all looks fine
  3. we are flashing Jetpack6.2 with the following modifications:
    1 - DTB and PINMUX for our carrier board.
    2 - Uefi secure boot
    3 - Disk encryption (full encryption of the NVMe)
    4 - we are flashing the ODM fuses for production (disabling the jtag
    and set the SBK and PKC as well as OEM1)
    5 - creating a EKB partition before the flashing process
    6 - using chroot to modify the rootfs to match NoTraffic needs

Also, attached a log of working vs. not working boot log for your reference.
I will appreciate your support in this case.

Thanks!
Elad

working.txt (32.9 KB)
Not working boot.txt (62.4 KB)

1 Like
3

Is this issue able to reproduce with NV devkit?

4

Not yet, but I’ll try it

1 Like
5

Tried to check it on the dev. kit and the same issue occurs

6

Just to clarify. Are you still able to reproduce this situation on all the modules you got there?

Also, after this issue happened, do you need reflash to make it back to work? or just change the temperature would make it back to live again?

7

Yes, I reproduced the issue on the dev. kit with the particular Jetson + NVME.
To make it work I just need to rise temperature and then restart or unplug DC and plug again.

8

How many modules got this kind of issue here? Could you also share the serial number of them?

9

I have 1 module with this specific issue.
SN: 1421524215281,3C6D6603B5AC,699-13767-0003-300,161-0472-90X

10

Hi,

Just to make the situation more simplified.

For NV devkit, I believe you should only need secure boot setting (keys) to make it flash and boot. Could you confirm that is what you did on NV devkit before you reproduce this issue?

11

I just moved the Jetson + NVME from my carrier board (customized) to the nvidia dev. kit.
then I saw the same issue, in cold temperature the Jetson couldn’t boot and when I slightly increased the temperature, the Jetson boot successfully (on the nvidia dev. kit)
I shouldn’t flash it again because I already seen that the temperature rise make it works on the dev. kit

12

Hi,

Just to clarify. This is a debug phase now. I don’t see any reason to say “I shouldn’t flash it”.

You should reflash it and if this is really a bug, then it would be repeatable even after your reflash. That is what we should confirm now.

13

OK, I will try to reflash it again

14

Hi Wayne,

We tried to reflash the Jetson but still having the same issue.
Do you have any suggestion?

Thanks,

15

Hi,

Could you share your steps to flash the NV devkit for us to review?

16

Hi @elad6

And do you have other modules that are not with ODM fused? Sounds like you don’t have any?

17 
sudo systemctl stop udisks2 sudo systemctl restart   nfs-kernel-server sudo ufw disable  ====== once on a blank Jetson ======== # Flash the QSPI first to allow fusing the jetson  sudo ./tools/kernel_flash/l4t_initrd_flash.sh  --showlogs -p "-c bootloader/generic/cfg/flash_t234_qspi.xml" jetson-orin-nano-nvme-nt internal  # Reboot the unit  # Fuse the unit - run once! sudo ./odmfuse.sh -i 0x23 -k rsa.pem -S sbk.key -X fuse.xml  --disable-jtag jetson-orin-nano-nvme-nt  ====== End of block to run once on a blank Jetson ========  # create the QSPI Images sudo ./tools/kernel_flash/l4t_initrd_flash.sh -u ./rsa.pem -v ./sbk.key --no-flash --showlogs -p "-c bootloader/generic/cfg/flash_t234_qspi.xml" jetson-orin-nano-nvme-nt internal   # Create the enc FS and other Nvidia Images sudo ROOTFS_ENC=1 EXT_NUM_SECTORS=468846000 ./tools/kernel_flash/l4t_initrd_flash.sh -S 221GiB  --showlogs --external-device nvme0n1p1 -c ./tools/kernel_flash/flash_l4t_t234_nvme_rootfs_enc.xml -p "-i sym2_t234.key -c bootloader/generic/cfg/flash_t234_qspi.xml" --append --no-flash --uefi-keys uefi_keys/uefi_keys.conf -i sym2_t234.key -u ./rsa.pem -v ./sbk.key  jetson-orin-nano-nvme-nt internal  sudo systemctl stop udisks2 sudo systemctl restart   nfs-kernel-server sudo ufw disable # Reboot the unit  # Actually flash sudo ./tools/kernel_flash/l4t_initrd_flash.sh --showlogs --flash-only
18

And do you have other modules that are not with ODM fused? Sounds like you don’t have any? - We don’t have

19

What is “jetson-orin-nano-nvme-nt” here? As we are requesting your steps on NV devkit, why does such board config come out?

20

This is our board based on jetson-orin-nano-nvme

21

As I already told in previous comment, could you do this test on NV developer kit?

We really don’t care about anything for your custom board for now. Only devkit result and configs matter.