LibThreading: Adjust ThreadPoolLooper m_busy_count sections

The `ThreadPoolLooper` should increment `m_busy_count` before attempting
to access the global queue. Otherwise, there exists a possible race
condition where `wait_for_all` checks the exit conditions before the
looper increments `m_busy_count` but after it empties the `ThreadPool`
queue.

Next, incrementing / decrementing `m_busy_count` is moved to be the
responsibility of `ThreadPoolLooper`. Otherwise, it is possible that
decrementing `m_busy_count` in the caller of `Looper::next` causes
`m_busy_count` to underflow if the call to `Looper::next` returns
before incrementing `m_busy_count`.
This commit is contained in:
Braydn 2024-08-01 22:26:04 -04:00 committed by Ali Mohammad Pur
parent a0fd7cf371
commit 8d336d2a25
Notes: github-actions[bot] 2024-08-19 01:09:04 +00:00

View file

@ -22,6 +22,7 @@ struct ThreadPoolLooper {
{
Optional<typename Pool::Work> entry;
while (true) {
pool.m_busy_count++;
entry = pool.m_work_queue.with_locked([&](auto& queue) -> Optional<typename Pool::Work> {
if (queue.is_empty())
return {};
@ -29,6 +30,8 @@ struct ThreadPoolLooper {
});
if (entry.has_value())
break;
pool.m_busy_count--;
if (pool.m_should_exit)
return IterationDecision::Break;
@ -47,8 +50,9 @@ struct ThreadPoolLooper {
pool.m_mutex.unlock();
}
pool.m_busy_count++;
pool.m_handler(entry.release_value());
pool.m_busy_count--;
pool.m_work_done.signal();
return IterationDecision::Continue;
}
};
@ -124,8 +128,6 @@ private:
Looper<ThreadPool> thread_looper;
for (; !m_should_exit;) {
auto result = thread_looper.next(*this, true);
m_busy_count--;
m_work_done.signal();
if (result == IterationDecision::Break)
break;
}