Hi @tonyxu ,
After analyzing customer's project, we reproduced this issue on S32K312 EVB, and found something.
It is the wrong memory allocation which caused abnormal code execution.
In "libAvas.c" file, here are 3 huge arrays, seems they were allocated at section DTCM:
However, there is no ".DTCM_data" section in linker file. Then the compiled result showed a wrong section was inserted in ".text" section, this accident may affect the code execution:
The wrong operation allocated the arrays into SRAM instead of DTCM, which deviate from customer's original intention. We still have no idea about why there is no issue/warning reported by GCC compiler, however, glad that we have found it finally!
Therefore, please kindly modified the attribute section name to the correct name, which can be found in linker file:
After tweaking the allocation section names, the issue disappeared.
By the way, please be careful about the memory usage especially the SRAM. The compile result of .bss in customer's project is getting close to the max size. You can try to allocate different data/code into other memory area, or try to improve the compile optimization level from "-O0" to "-Os", etc.
Another potential problem would be the "Power" component configuration in .mex file, in Which the MCU peripheral gate clock was not enabled totally. If customer trying to operate the non-gate clock enabled peripherals, it will generate a bus error, and then jump into hardfault:
Therefore, please kindly check the peripherals will be used, and check the corresponding peripheral clock enabled.
FYI.
Best Regards,
Shuailin Li
