malloc() causing HardFault

[NewtoM]

Hi,

I work on a project for the nRF52832 with Keil µVision using ARM Compiler V5.06 update 2 and nRF5 SDK 11.

In a module I allocate quite some memory using malloc() for signal processing (SP). Upon user request, this mode ends and with that I free() up all the memory allocated previously. Upon another user request, I enter the SP mode again and do my malloc() calls again.

This is when it happens that one of the malloc() calls causes a hard fault. It does not return a null pointer (as it should if heap was exhausted), but simply fails. (Some of the last assembler steps in the debugger are attached below.)

I already tried to change heap and also stack sizes (based on a comment in https://devzone.nordicsemi.com/question/39385/checking-for-memory-usage-on-runtime/), but no success.

What can be the cause and how do I help the situation?

Thanks, Tamas

[WestCoastDaz]

Have you tired enabling the hardfault handler to see what was executing during the hardfault?

Cheers, Darren

[NewtoM]

Can you please point me to some resource on how to do that?

[NewtoM]

OK, I looked in SDK...\components\libraries\hardfault\ and learned a bit. Thanks!

[NewtoM]

Hi,

once more I needed to conclude that my old programming teacher was right: "Computers do not make mistakes; programmers do."

The cause of the strange behavior of malloc() was my fault, a logical error in the program. I called free() twice with the same pointer, which corrupted memory so that the next malloc() call caused the hard fault.

Sorry for taking your time.

Tamas

[endnode]

(sorry, cannot resist the note: haven't your teacher also mentioned "don't use malloc in embedded systems - and possibly don't use it anywhere"?:)