Question: CPU abort instead of exception?

[rogerdbenson]

I've been developing some code, and I've had errors in said code. One behavior that I've seen that I don't understand is this; if a bad code pointer is fetched off of the stack into PC, then the framework issues a CPU abort. Why does this happen? This type of issue show up quite often in real devices, and the MCU issues an exception when it occurs. Why doesn't the Renode model issue a valid exception instead of a CPU abort? In a multi-machine situation, the abort causes the entire framework to become unusable, but an exception in a single model could be continued with a restart on the single machine with the exception. As a matter of fact, this is a valid test case, which can't be tested because of the CPU abort behavior.

Thanks,
Roger

[PiotrZierhoffer]

Hi @rogerdbenson

Your scenario is of course valid, but it's also quite useful to be able to detect these situations early (if your exception handler handles it, you might not even notice the problem easily).

We do, however, try to make it an option. Can you provide more details on your setup? Especially what architecture and which exception are you concerned about?