A stdlib alternative to this whole mess would be to avoid acquiring the logging locks before fork() as we currently do and just blindly re-initialize all of them afterwards under the assumption that they "can't" be protecting anything in a newly forked child process.  Handlers that need specific resource synchronization around fork would then be required to deal with their own os.register_at_fork() calls.  (ex: to avoid multiple processes writing to the same file or fd at once)