BlockingPool num_idle_threads is wrongly double-increased when shutting down

[liuq19]

Version
Tokio v1.37.0

Description
BlockingPool num_idle_threads is doubly increased, when shutting down before the condvar wait_timeout.

The whole code is:

tokio/tokio/src/runtime/blocking/pool.rs

Lines 519 to 564 in 9c337ca

	self.metrics.inc_num_idle_threads();
	while !shared.shutdown {
	let lock_result = self.condvar.wait_timeout(shared, self.keep_alive).unwrap();
	shared = lock_result.0;
	let timeout_result = lock_result.1;
	if shared.num_notify != 0 {
	// We have received a legitimate wakeup,
	// acknowledge it by decrementing the counter
	// and transition to the BUSY state.
	shared.num_notify -= 1;
	break;
	}
	// Even if the condvar "timed out", if the pool is entering the
	// shutdown phase, we want to perform the cleanup logic.
	if !shared.shutdown && timeout_result.timed_out() {
	// We'll join the prior timed-out thread's JoinHandle after dropping the lock.
	// This isn't done when shutting down, because the thread calling shutdown will
	// handle joining everything.
	let my_handle = shared.worker_threads.remove(&worker_thread_id);
	join_on_thread = std::mem::replace(&mut shared.last_exiting_thread, my_handle);
	break 'main;
	}
	// Spurious wakeup detected, go back to sleep.
	}
	if shared.shutdown {
	// Drain the queue
	while let Some(task) = shared.queue.pop_front() {
	self.metrics.dec_queue_depth();
	drop(shared);
	task.shutdown_or_run_if_mandatory();
	shared = self.shared.lock();
	}
	// Work was produced, and we "took" it (by decrementing num_notify).
	// This means that num_idle was decremented once for our wakeup.
	// But, since we are exiting, we need to "undo" that, as we'll stay idle.
	self.metrics.inc_num_idle_threads();

When BlockingPool is shutting down,
the num_idle_threads will be increased at first in

tokio/tokio/src/runtime/blocking/pool.rs

Line 519 in 9c337ca

self.metrics.inc_num_idle_threads();

and then it will doubly be increased in this

tokio/tokio/src/runtime/blocking/pool.rs

Line 564 in 9c337ca

self.metrics.inc_num_idle_threads();