Need confirmation on whether high-risk lines produce deadlock

[amab8901]

These lines were identified (#5479) as having high probability of producing deadlocks (as in #5429):

• task/atomic_waker.rs
  • 204
• broadcast.rs
  • 915
• notify.rs
  • 771
  • 835
  • 874
• oneshot.rs
  • 427

I think we need to create some kind of tests to verify whether they actually can cause deadlocks.

[Darksonn]

I took a look at them:

1. task/atomic_waker.rs - this one already handles panics correctly, and there's no mutex so it cannot deadlock
2. broadcast.rs - This assignment only happens if the field is None, so it cannot run the destructor of a waker.
3. notify.rs - will be fixed by #5471
4. oneshot.rs - This uses ptr.write, which doesn't run the destructor of old values, so it isn't vulnerable.