How to unit test that os.walk triggers an error handler?

# my_module.py

def walk_error_handler(os_error_instance):
    logger.error(...)

for root, dirs, files in os.walk(path, onerror=walk_error_handler):
    ...

How to unit test that os.walk, should it encounter an OSError, calls the custom handler?

I tried to do this by creating a temporary file in a temporary directory, and then restricting the permissions of the directory. Unfortunately that crashed the test runner. Looking for a more elegant solution.