Device gone forever after Ctrl-C or kill

• bleak version: 0.8.0
• Python version: 3.7.3
• Operating System: Linux-4.19.0-10-amd64-x86_64-with-debian-10.5
• BlueZ version (bluetoothctl -v) in case of Linux: 5.50

Description

If a bleak process with a active connection is terminated with Ctrl-C or kill SIGINT, the BLE device disappears forever. I am unable to reconnect and device do not show up in scan/discovery. sudo systemctl restart bluetooth resolves the problem, so it is likely a issue with a “zombie connection” or something in bluez/dbus.

What I Did

This reproduces the problem:

import asyncio
from bleak import BleakClient
import sys
import os

async def main():
    address = sys.argv[1]
    pid = os.getpid()
    async with BleakClient(address) as client:
        while 1:
            print("press ctrl-C or enter `kill -2 {}` in another shell!".format(str(pid)))
            await asyncio.sleep(1)

loop = asyncio.get_event_loop()
loop.run_until_complete(main())

Traceback when pressing ctrl-C:

Traceback (most recent call last):
  File "connect_and_exit.py", line 72, in <module>
    loop.run_until_complete(main())
  File "/usr/lib/python3.7/asyncio/base_events.py", line 571, in run_until_complete
    self.run_forever()
  File "/usr/lib/python3.7/asyncio/base_events.py", line 539, in run_forever
    self._run_once()
  File "/usr/lib/python3.7/asyncio/base_events.py", line 1739, in _run_once
    event_list = self._selector.select(timeout)
  File "/usr/lib/python3.7/selectors.py", line 468, in select
    fd_event_list = self._selector.poll(timeout, max_ev)

The following examples works as expected (i.e. BLE device show up in scan). It might seem they should have the same behavior, but apparently not.

async def test_Exception(address):
    async with BleakClient(address) as client:
        await asyncio.sleep(0.5)
        raise Exception

async def test_KeyboardInterrupt(address):
    async with BleakClient(address) as client:
        await asyncio.sleep(0.5)
        raise KeyboardInterrupt

async def test_SIGINT(address):
    pid = os.getpid()
    async with BleakClient(address) as client:
        await asyncio.sleep(0.5)
        os.kill(pid, signal.SIGINT)

Traceback when raising KeyboardInterrupt. Note that it differs from the one above!

Traceback (most recent call last):
  File "connect_and_exit.py", line 66, in <module>
    loop.run_until_complete(main())
  File "/usr/lib/python3.7/asyncio/base_events.py", line 571, in run_until_complete
    self.run_forever()
  File "/usr/lib/python3.7/asyncio/base_events.py", line 539, in run_forever
    self._run_once()
  File "/usr/lib/python3.7/asyncio/base_events.py", line 1775, in _run_once
    handle._run()
  File "/usr/lib/python3.7/asyncio/events.py", line 88, in _run
    self._context.run(self._callback, *self._args)
  File "connect_and_exit.py", line 50, in main
    await test_KeyboardInterrupt(address)
  File "connect_and_exit.py", line 17, in test_KeyboardInterrupt
    raise KeyboardInterrupt
KeyboardInterrupt

I believe this problem was recently introduced as I have not seen it earlier (might have missed it).

Can anyone reproduce it on another system? Ideas?

[dlech]

Yes, or turn it into a helper method.

[dlech]

Maybe we should switch to asyncio.run in examples?

We still support Python 3.6, so we would have to do something like this.