As mentioned in this PR, we could use waitpid first and then fallback to the current implementation of reading the process status from /proc/<pid>/stat if the waitpid fails for some reason. However, we currently have a timeout (defaulting to 100ms) waiting for the state to change, before going ahead with the process dumping, meaning if we wanted to keep a timeout/best effort stopping, we'd need to do something similar to this to be able to timeout the waitpid if it takes too long. In most cases I'd assume the waitpid would succeed/fail quickly and the timeout wouldn't matter, but that's not guaranteed, and the current polling solution, while less efficient, is quite simple.