Segmentation Fault when using GPIOChip FB

[mayrthom]

When I deploy a Program that is using the GPIOChip FB and I want to redeploy the program I got one of these errors:

ERROR: T#1768990656523022822: [GPIOChipController::runLoop] Could not write output data to GPIO chip 0 line 17
INFO: T#1768990656523159932: [IOMapper] Disconnected TestApp.DI1 (lost observer)
DEBUG: T#1768990656523222969: [IOMapper] Deregister observer TestApp.DI1
Segmentation fault

ERROR: T#1769007599050000910: [GPIOChipController::runLoop] Could not write output data to GPIO chip 0 line 17
INFO: T#1769007599050074576: [IOMapper] Disconnected TestApp.DI1 (lost observer)
DEBUG: T#1769007599050114336: [IOMapper] Deregister observer TestApp.DI1
WARNING: T#1769007599133773875: [IODeviceController] No receiver for notification is available. Notification is dropped.
DEBUG: T#1769007599133828542: [IOMapper] Deregister handle TestApp.DI1
INFO: T#1769007599133944689: [GPIOChipController::deInit] GPIO Chip 0 line 17 shutdown
pure virtual method called
terminate called without an active exception
Aborted

I also got a segmentation fault when I quit forte with Ctrl + C.
While the program is running everything works fine (reading and writing IOs with the GPIOChip FB).

[gravemalte]

I think this issue is maybe due to the not existing destructor in IOHandleBit. But I'm unsure. I have to debug this further.

[gravemalte]

The SIGSEGV is happening while call of the IODeviceController::dropHandles() function.

There we get the IOMapper::getInstance(); singleton. Then we call deregisterHandle with the concrete inputHandle.

But once in IOMapper::deregisterHandle the map where we store the id and the handle (THandleMap mHandles;), the id is just random garbage, and the mHandles.erase(...); will fail. I'm not sure when the id is getting ruined but the insertion works correctly.

[gravemalte]

I found the error. Or more precise, I found multiple errors.

But the overall error of the pure virtual method called is in the file io_configFB_controller.cpp.

There the STATUS() is tried to be set in the stopped function, but the getDO(1) function call is in fact virtual and the whole object hierarchy is not existing anymore. This results into the above error message. But also the IOMapper does some shenanigans.

[cochicde]

althought I don't know the details, #362 might have had related causes