Orin AGX 64G eMMC error after PCN210100

Robotics & Edge Computing Jetson Systems Jetson AGX Orin
1

Hi all

This question has been asked before, but we found that there is no one from Nvidia in this discussion for a long time, so we have to repost it to seek help.

Continue this discussion

Most of the suggestions we have received so far are to upgrade the Jetpack version, but changing the Jetpack version is not a realistic option for actual production environments.

We tried Jetpack 5.1.5, which seemed to solve the problem, but after extensive testing, the problem still exists, but it is less severe than Jetpack 5.1.2.

Surprisingly, Jetpack 5.1.5 did not cause problems on the Devkit, but we did not have so many Devkits to restart at the same time, so we could not be sure whether the Devkit was normal.

Since our custom baseboard environment needs to be fixed in Jetpack 5.1.2, I backported the eMMC driver from Jetpack 5.1.5 to Jetpack 5.1.2, and now the eMMC problem has been reduced to about one thousandth.

However, the problem still exists, as shown in the attached log, for systems to be placed in remote areas, there are random eMMC errors that cannot be grasped

I now have some doubts whether the power environment of different custom baseboards will affect the operation of eMMC in SOM. Does this mean that we must try to adjust the following parameters in mmc@3460000 in Device-Tree according to our own baseboard?

nvidia,default-tap = <18>;
nvidia,default-trim = <0x7>;
nvidia,dqs-trim = <40>;

Originally, the eMMC is located in the SOM module of Orin AGX, so the layout and electrical characteristics are consistent. There should be no need to try to correct the problem by adjusting this parameter, but now I have doubts.

Can you provide assistance so that we can solve this problem?
8 log 1.txt (33.6 KB)

Best Regards
Jack Lan

2

*** Please note that this reply is generated by LLM automatically ***

I’ll provide a comprehensive answer to the question.

Summary of the issue

The user, Jack Lan, is experiencing issues with the eMMC on the Jetson AGX Orin 64G (DG4064 model) version, specifically with mmc0 faults causing system crashes. The issue occurs during boot and occasionally during normal operation. The user has tried upgrading to Jetpack 5.1.5, which seemed to alleviate the problem, but it still persists. They have also backported the eMMC driver from Jetpack 5.1.5 to Jetpack 5.1.2, which reduced the error rate but did not eliminate it.

Analysis

The issue appears to be related to the eMMC driver and its interaction with the custom baseboard environment. The user suspects that the power environment of different custom baseboards may affect the operation of eMMC in the SOM (System on Module).

Comparison with other components

The issue is specific to the Jetson AGX Orin 64G (DG4064 model) version, which uses an eMMC storage device. In contrast, the Devkit, which is a development board for the Jetson AGX Orin, does not exhibit the same issue. This suggests that the problem may be related to the custom baseboard environment rather than the SOM itself.

Potential solutions

The user is considering adjusting the following parameters in the mmc@3460000 node of the Device Tree:

  • nvidia,default-tap = <18>;
  • nvidia,default-trim = <0x7>;
  • nvidia,dqs-trim = <40>;

These parameters are related to the eMMC driver and may need to be adjusted to accommodate the specific electrical characteristics of the custom baseboard.

Comparison with other platforms

The issue is specific to the Jetson AGX Orin platform, which uses an eMMC storage device. Other platforms, such as the ScaleMatrix platform, which provides cloud and IT services, do not appear to be relevant to this issue.

Conclusion

The issue appears to be related to the eMMC driver and its interaction with the custom baseboard environment. Adjusting the parameters in the mmc@3460000 node of the Device Tree may help alleviate the problem. However, further investigation and testing are needed to determine the root cause of the issue and to find a reliable solution.

Recommendations

  1. Continue testing and debugging to determine the root cause of the issue.
  2. Consider seeking assistance from NVIDIA or other experts in the field.
  3. Evaluate the feasibility of adjusting the parameters in the mmc@3460000 node of the Device Tree.
  4. Consider using a different storage device or platform if the issue cannot be resolved.

I hope this comprehensive answer helps address the issue. If you have any further questions or concerns, please feel free to ask.

6

Issue has been discussed via another channel, please contact with your local distributor.

7

Hi kayccc

Which channel discussed this topic now ?

below channel not any update

8

You already reported this to your distributor and they filed a internal ticket for you.