Error when flashing TX2 NX

[juan.tettamanti]

Hello everyone,
I’ve recently been working on flashing Jetson TX2 NX boards for a custom product.
So far I’ve had a few issues regarding the necessary files as well the command line sequence and solved most of them.

I’ve read most of the available documentation and checked flash.sh as reference, however I can’t seem to find a satisfying answer within the available information.
Since almost every partition is marked as required without a lengthy explanation and just one sample implementation it is hard to determine which ones are really required.

A simple example of this would be the case where I don’t require update redundancy.
In this context, one would expect all the required slot B partitions to be completely unnecessary.
Since the default implementation emphasizes redundancy and available documentation is really scattered, it is hard to determine if that’s the case.
Are the slot B partitions really necessary if redundancy is disabled?
This Update and Redundancy section makes me think they are not.

There are some other cases of required partitions I find confusing like RECNAME and BOOTCTRLNAME.
What if I don’t want a recovery image?
What’s the purpose of the so called “boot control data”?
Which bootloader/bootloaders use this data and for what purpose?
What happens if I don’t have this partitions?

Last but not least, the system is currently throwing the following message repeatedly:

C> ERROR: Highest Layer Module = 0x1d, Lowest Layer Module = 0x1d, Aux Info = 0x0, Reason = 0xd

It seems to come from one of the earliest stages of the boot process.
However, I don’t know which one, nor its meaning.

Maybe it’s a matter of partition ordering or an expected partition size.
Whatever it is, I’m hoping someone might be able to point me in the right direction here.

Best Regards,
Juan Pablo.

[WayneWWW]

I think what you should firstly clarify is what is the full error log during flash and why do think you think this is related to some partition layout?

Do you modify something in the partition layout so that you think they are related to your flash problem?

[juan.tettamanti]

Hi @WayneWWW,
I would expect that the error message would point to the possible error without the need for additional information (otherwise the message is totally useless).
I do have a different layout and have no desire to keep any partition that is not strictly necessary to my specific use case.
I’ve attached the xml file corresponding to said layout.

Best Regards,
Juan Pablo.

flash.xml (16.6 KB)

[juan.tettamanti]

Hi @WayneWWW,
I also wanted to point out that the system seems to be flashing everything to the board. It’s just not booting after that.

Best Regards,
Juan Pablo.
flash.log (19.2 KB)
serial.log (6.5 KB)

[WayneWWW]

I would firstly suggest you can flash your custom board with default partitions layout first and see if it can boot fine.

Also, what is your software release version?

[juan.tettamanti]

Hi @WayneWWW

At the beginning of this post I asked some very concrete questions and so far none of them have been answered.

I will place them again here so I can work from there.

1. What does the following error code mean?

C> ERROR: Highest Layer Module = 0x1d, Lowest Layer Module = 0x1d, Aux Info = 0x0, Reason = 0xd

2. What are the possible root causes for said error code?

3. What partitions are strictly necessary for a working system?

4. Why are those partitions required?

Best Regards,
Juan Pablo.

[juan.tettamanti]

Hi,
I still have to confirm my suspicion since right now I’m using a layout with an awful lot of partitions.

So first I got this close to Nvidia’s original layout and it would still not work.
Then I decided to add the optional partitions asociated with the ECC features.

As some documents mention, it would seem that the ECC is enabled by default.
I also understand from the same documents that most TX2 systems don’t use ECC.
This doesn’t look like a sane default to me, but whatever.

It’s likely that I’ll go back to removing most of the redundant partitions after I’m done with some other stuff.

Best Regards,
Juan Pablo.

[WayneWWW]

1. If even nv original layout cannot work, please review your hardware design.
   It sounds like you are using a custom board and you want to add all those your custom components to it altogether.
   My suggest is please verify them one by one instead of putting all of them into your board and cannot find out what is the cause of the crash.

2. You can remove all the _b partitions, but you can’t move partitions from the boot device to the user device freely.
   In your layout file, you moved mb2, mts-preboot, and other boot partitions to the user device, so it causes the boot-up failure at the beginning of the MB1. The MB1 can’t access the user device.