Can't flash Drive OS 7.0.3 to Drive AGX Thor

DRIVE OS Version: Drive OS : 7.0.3

**Software:**Docker Image, nvcr.io/drive/driveos-sdk/drive-agx-linux-nsr-aarch64-sdk-build-x86:7.0.3.0-0010

**Hardware Platform:**Drive AGX Thor (940-63960-0010-000)

Issue Description:

To initialize the Drive AGX Thor, I attempted to write Drive OS following the procedure in the link below:
https://developer.nvidia.com/docs/drive/drive-os/7.0.3/public/drive-os-linux-installation/install-driveos/install-driveos-developer.html

However, although the writing using bootburn.py succeeded, the Drive OS was not initialized properly as the previous account and other data remained.
Therefore, I tried writing again using the --init_persistent_partitions option of bootburn.py.
Then, due to some error, the writing was interrupted. (Unfortunately, I failed to capture the logs or traces at that time.)
When I tried again with bootburn.py using the --init_persistent_partitions option along with tracing options (-D -T),
an error occurred during the transfer of a specific file via the ADB command, causing the writing to be interrupted as shown below.
From the same trace, it is confirmed that files other than the specific one were successfully transferred via ADB.

trace:

[bootburnTegra-A]: [executeShellCommand(170)] : shell command – /drive/drive-foundation/tools/flashtools/flash/adb -s 2U1TS82I8000005010C0 push /drive/drive-foundation/tools/flashtools/bootburn_t264_py/__temp_t264/Tegra-A-d24c5ca948b/flash-images/a80b8d0000.ufshci:0_179 /tmp failed
[bootburnTegra-A]: [executeShellCommand(170)] : shell command – /drive/drive-foundation/tools/flashtools/flash/adb -s 2U1TS82I8000005010C0 push A_17_igbfw_gb10b_gsc_package_prod_zerosign.bin /tmp failed

Even when writing without using the --init_persistent_partitions option, the same file causes an error during writing and the process is interrupted.

trace:

[bootburnTegra-A]: [executeShellCommand(170)] : shell command – /drive/drive-foundation/tools/flashtools/flash/adb -s 2U1TS82I8000005010C0 push /drive/drive-foundation/tools/flashtools/bootburn_t264_py/__temp_t264/Tegra-A-516976f6b33/flash-images/a80b8d0000.ufshci:0_51 /tmp failed
[bootburnTegra-A]: [executeShellCommand(162)] :
[bootburnTegra-A]: [AdbShell(315)] :
[bootburnTegra-A]: [executeShellCommand(172)] : Output: adb: error: connect failed: closed

How should I proceed to be able to write Drive OS again with bootburn.py?

Dear @suguru.sanichi.cj ,
Is tegra console accessible now? Does it has DOS 7.0.3?

Could you restart host and target(issue poweroff/on from MCU console) and check flashing again. Please share complete logs and commands.

Thank you for your response.
I am not sure what the Tegra console is, but I am using the terminal on a host PC running Ubuntu 22.04.
I plan to reboot both the host PC and Thor next week and try the writing process again.
When performing the writing again, is it sufficient to collect trace logs for both bind_partition and bootburn.py?
Lastly, I will share the logs I currently have. These logs are from when I used bootburn.py with the --init_persistent_partitions -D -T option.

bootburn_init_persistent.log (2.2 MB)

Dear @SivaRamaKrishnaNV

I executed bind_partitions and bootburn.py, but I got the same result as before.
I will share the trace logs obtained this time.

I followed the steps below.:

①lsusb→detected “NVIDIA Corp. Tegra Device” and “NVIDIA Corp. Tegra On-Platform Operator”
②docker run -it –privilleged –net=host -v /drive_flassing:/drive_flashing -v /dev:/dev nvcr.io/drive/driveos-sdk/drive-agx-linux-nsr-aarch64-sdk-build-x86:7.0.3-0010

③cd $NV_WORKSPACE/drive-foundation

④./bind_partitions -v -b p3960-10-sw03 drive_av.linux -p dev_nsr > /drive_flashing/bind_paritions_verbose.log

→see “bind_partitions_verbose_stdout.log” , “bind_partitons_verbose.log”

⑤cd $NV_WORKSPACE/drive-foundation/tools/flashtools/bootburn

⑥./bootburn.py -b p3960-10-sw03 -x /dev/ttyACM2 -D -T > /drive_flashing/bootburn_trace.log

→I can’t find bootburn/log_trace*.txt. see “bootburn_trace.log” and “bootburn_trace_stdout.log”.

Please check the log files.

bootburn_trace_stdout.log (4.9 KB)

bootburn_trace.log (1.9 MB)

bind_partitons_verbose.log (1.1 MB)

bind_partitions_verbose_stdout.log (986.3 KB)

You can access Tegra console using sudo minicom -D /dev/ttyACM0 from host. Could you verify it?

bootburn logs has repeated ‘device not found’ and connection failures during the ADB push phase led to checksum mismatches and flashing aborts. This suggests issues with USB connections, device power stability . Could you please check changing USB port on host or connection wires to see if it fixes. Also, please share showvoltages output on aurix console

I attempted to run sudo minicom -D /dev/ttyACM0~4.
I was able to connect via ttyACM2. However, when I executed tegrareset x1 on ttyACM2, several errors were detected.

ttyACM0, 1, and 3 did not accept input and could not connect, and although ttyACM4 accepted input, all commands were treated as invalid commands.

sudo minicom -D /dev/ttyACM4.

# showvoltages

Invalid command!

#

I will share the results of the showvoltages command executed on ttyACM2. From the output of the showvoltages command, it was confirmed that some voltages are out of range.

Could the additional M.2 SSD connected to the Drive AGX Thor be related? However, since I had previously been able to run bootburn.py (without the --init_persistent_partitions option) while the M.2 SSD was connected to the Drive AGX Thor, I think it is unlikely to be related.
The USB cables used were two USB Type A-C cables included with the Drive AGX Thor, and the lsusb command detected “NVIDIA Corp. Tegra Device” and “NVIDIA Corp. Tegra On-Platform Operator.”

Is the host connected only to a single Thor devkit or any other devkits also connected?
Generally when a single thor devkit is connected. Tegra and MCU console can be accessed like below
Tegra console : sudo minicom -D /dev/ttyACM0
MCU console : sudo minicom -D /dev/ttyACM2

What do you notice when trying to access tegra console? Check if minicom has configuration settings like in
Configuring Minicom — NVIDIA DriveOS 7.0.3 Linux SDK Developer Guide for tegra console. Please share the noticed settings on minicom when tegra console is opened.

Please perform poweroff and poweron on Aurix console and see if tegra can be accessed.

I will check and get back to you if any of these can cause flashing failure.

I understand that board is able to enter recovery mode and detected. As I see connection failure related error while flashing, It is better to verify with other connection ports, cables . We notice changing cables/port/host machine resulted resolving flashing issues in the past.

Hello.
As @SivaRamaKrishnaNV mentioned, I changed the USB port and cable and then ran bootburn.
As a result, bootburn completed without outputting any errors.
However, when Drive AGX Thor boots up, the accounts and environment created before initialization are still available.
For now, I will give up on initializing the Drive AGX Thor and move on to another issue.

Thank you very much for your support despite my poor English.