[libcxx-commits] [libcxx] [libc++] fix `counting_semaphore` lost wakeups (PR #79265)

[Jan Kokemüller via libcxx-commits]

================
@@ -94,20 +95,25 @@ public:
     auto __old = __a_.fetch_add(__update, memory_order_release);
     _LIBCPP_ASSERT_ARGUMENT_WITHIN_DOMAIN(
         __update <= _LIBCPP_SEMAPHORE_MAX - __old, "update is greater than the expected value");
-
-    if (__old > 0) {
-      // Nothing to do
-    } else if (__update > 1)
+    (void)__old;
+    if (__update > 1) {
       __a_.notify_all();
-    else
+    } else {
       __a_.notify_one();
+    }
   }
   _LIBCPP_AVAILABILITY_SYNC _LIBCPP_HIDE_FROM_ABI void acquire() {
     auto const __test_fn = [this]() -> bool {
       auto __old = __a_.load(memory_order_relaxed);
       return (__old != 0) && __a_.compare_exchange_strong(__old, __old - 1, memory_order_acquire, memory_order_relaxed);
     };
-    __cxx_atomic_wait(&__a_.__a_, __test_fn);
+    auto const __test_with_old = [this](__cxx_contention_t& __monitor) -> bool {
----------------
jiixyj wrote:

> You just want to take your `try_wait()` and your eventcount abstraction (i.e. `__cxx_atomic_wait()`), compose them, and get your `wait()` for free.

For `semaphore` this could look like this:

```c++
#include <atomic>
#include <functional>
#include <iostream>

template <typename A, typename Fn>
void cxx_atomic_wait(A *, Fn &&fn, std::memory_order mo);

struct semaphore {
    std::atomic<std::ptrdiff_t> a_{};

    bool try_acquire_impl(std::ptrdiff_t &value) {
        do {
            if (value == 0) {
                return false;
            }
        } while (!a_.compare_exchange_weak(value, value - 1,
                                           std::memory_order_acquire,
                                           std::memory_order_relaxed));
        return true;
    }

    static constexpr auto acquire_initial_load_mo = std::memory_order_relaxed;

    bool try_acquire() {
        auto old = a_.load(acquire_initial_load_mo);
        return try_acquire_impl(old);
    }

    void acquire() {
        using namespace std::placeholders;
        cxx_atomic_wait(&a_, std::bind(&semaphore::try_acquire_impl, this, _1),
                        acquire_initial_load_mo);
    }
};
```

Just for fun I could even use `std::bind` in `acquire()` to really make it clear that no custom (lambda) logic is needed there.

Also, the initial load can be `relaxed`, and `compare_exchange_weak` can be used instead of `compare_exchange_strong` since it's called in a loop anyway.

https://github.com/llvm/llvm-project/pull/79265