M4 Cortex firmware stops working after Mender integration on Apalis iMX8QM

Board Integrations
,
1

hello community i’ ve already posted the same problem in toradex community

i need your help to understand mender changes made on u-boot so i can i know where ‘s the conflict is coming from and why integrating this latter cause the corruption of the cortex m4 binary

2

Hi @rania,

Thanks for reaching out! The Mender integration definitely should not interfere with the M4 at all. From the linked post I gather that it works on BSP 6.7 (without Mender) and fails on BSP 6.2 (with Mender), correct?

So the two things to check:

  • does it work on vanilla BSP 6.2, without Mender? If no, then the BSP revision breaks/fixes something. If yes, then it is Mender :-/
  • after that, try bumping the Mender integration to BSP 6.7. That should be very straightforward, please ping here if you hit any problems. Then, repeat test.

Greetz,
Josef

3

Hi Josef,

To clarify the setup:
My board is actually running Toradex BSP 6.7 (Yocto Kirkstone).
However, since meta-mender-community only seemed to have configurations for BSP 6.0 and 6.2, I attempted to use the 6.2 integration logic on my 6.7 build.

This version mismatch might be why I am seeing these conflicts.

The specific behavior I am fighting is:

  1. Memory Collision: The default Toradex 6.7 setup loads M4 to 0x95400000. Mender loads the Linux Kernel to the same address. I tried moving M4 to 0x88000000 manually.
  2. Silent Failures: Even when I adjust the addresses manually in the U-Boot console, I struggle to get the M4 to boot reliably alongside Linux. I can see rpmsg communication , but mostly it fails silently .

Since I am on BSP 6.7, should I be using a different branch or specific configuration? Is there a known “safe” way to adapt the 6.2 configs to 6.7?

u have both cortex M4 and mender configurations can u help me find the u-boot conflicts whether address allocation or bootcmd