soc: stm32: Disable null pointer exception detection by default

[erwango]

Null pointer exception detection systematically triggers an
assert when program is run with debug mode on, which is currently
systematic when flashed using openocd runner.
Until this is fixed in openocd, make sure this detection is not
enabled by default (which is currently the case in zephyr tests).

Fixes: #32867

Signed-off-by: Erwan Gouriou [email protected]

[galak]

Does this only apply to STM32? Seems like the issue could possibly impact other Cortex-M based SoCs

[erwango]

Does this only apply to STM32? Seems like the issue could possibly impact other Cortex-M based SoCs

Not sure, but this was dicussed in #31481.
@ioannisg had a preference for this solution (cf #32867 (comment)). He might be able to comment more.
Meanwhile a wide bunch of STM32 tests are failing, so if this could be merged, even as a temp patch, it would be nice.

[galak]

So this issue is related to openOCD and not STM32, so I think it best we try and find an OpenOCD solution. ie can we have the west runner for openocd set the mode correctly.

[ioannisg]

So this issue is related to openOCD and not STM32, so I think it best we try and find an OpenOCD solution. ie can we have the west runner for openocd set the mode correctly.

Fully agree, though, I wanted to unblock, temporarily, @erwango , that's why I +1'ed this.

[galak]

So this issue is related to openOCD and not STM32, so I think it best we try and find an OpenOCD solution. ie can we have the west runner for openocd set the mode correctly.

Fully agree, though, I wanted to unblock, temporarily, @erwango , that's why I +1'ed this.

If we want to unblock than we should disable the feature completely for now. Since non-STM32 OpenOCD users will run into the same issue.