Great report Steven!

I was able to reproduce this with the attached patch (just adds some sleeps and prints) and this script:

from threading import current_thread
from concurrent.futures import ThreadPoolExecutor
from time import sleep

pool = ThreadPoolExecutor(100)

def f():
    print("I'm running in: ", current_thread().name)

def g():
    print("I'm running in: ", current_thread().name)
    for _ in range(100):
        pool.submit(f)
        sleep(0.1)

pool.submit(g)
sleep(1.5)

The output for me was:

👶 Creating new thread:  ThreadPoolExecutor-0_0
I'm running in:  ThreadPoolExecutor-0_0
Setting _shutdown
💀 Killing 1 workers 💀
👶 Creating new thread:  ThreadPoolExecutor-0_1
I'm running in:  ThreadPoolExecutor-0_1

So another thread was created *after* shutdown.

It seems like the most obvious way to fix this is by adding a lock for the global _shutdown variable.