The previous implementation would perform a load that might be part of a
data race. The value read would be used only when race did not occur.
This would be well defined in a memory model where a load that is a part
of race merely returns an undefined value, the Rust memory model on the
other hand defines it to be undefined behaviour.
Perform read conditionally to avoid data race.
Covered by existing loom tests after changing casualty check to be
immediate rather than deferred.
Fixes: #2087
Previously acquire operations reading a value written by a successful
CAS in `drop_join_handle_fast` did not synchronize with it. The CAS
wasn't guaranteed to happen before the task deallocation, and so
created a data race between the two.
Use release success ordering to ensure synchronization.
The `Waker::will_wake` compares both a data pointer and a vtable to
decide if wakers are equivalent. To avoid false negatives during
comparison, use the same vtable for a waker stored in `WakerRef`.
Tomasz Miąsko