为dispatch_source_t添加std::atomic编译报错，求解决方法

此前代码编译正常，仅修改为给成员变量添加std::atomic以执行atomic_compare_exchange操作时出现编译错误，代码如下：

std::atomic<dispatch_source_t> msource;

报错信息：

Xcode-beta.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/atomic:963:37: error: _Atomic cannot be applied to qualified type 'NSObject<OS_dispatch_source> *__strong' 
  _LIBCPP_DISABLE_EXTENSION_WARNING _Atomic(_Tp) __a_value;
                                    ^

Xcode-beta.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/atomic:1501:35: note: in instantiation of template class 'std::__cxx_atomic_base_impl<NSObject<OS_dispatch_source> *>' requested here
struct __cxx_atomic_impl : public _Base {
                                  ^

Xcode-beta.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/atomic:1616:36: note: in instantiation of template class 'std::__cxx_atomic_impl<NSObject<OS_dispatch_source> *>' requested here
    mutable __cxx_atomic_impl<_Tp> __a_;
                                   ^

Xcode-beta.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/atomic:1848:14: note: in instantiation of template class 'std::__atomic_base<NSObject<OS_dispatch_source> *, false>' requested here
    : public __atomic_base<_Tp*>
             ^

note: in instantiation of template class 'std::atomic<NSObject<OS_dispatch_source> *>' requested here
    std::atomic<dispatch_source_t> msource;
                                   ^

编译失败的核心原因是：dispatch_source_t本质是带ARC所有权修饰符__strong的Objective-C对象指针，C++标准库的std::atomic不支持直接处理带有Objective-C限定符的指针类型，编译器报错明确指出_Atomic无法应用于NSObject<OS_dispatch_source> *__strong这种类型。

以下是两种可行的修复方案：

• 方案一：用std::atomic<void*>替代，手动转换类型
  将原子变量声明为无类型指针的原子封装，使用时手动转换为dispatch_source_t：

  std::atomic<void*> msource{nullptr};
  
  // 赋值操作
  dispatch_source_t source = dispatch_source_create(DISPATCH_SOURCE_TYPE_TIMER, 0, 0, dispatch_get_main_queue());
  msource.store(static_cast<void*>(source), std::memory_order_relaxed);
  
  // 读取操作
  dispatch_source_t currentSource = static_cast<dispatch_source_t>(msource.load(std::memory_order_acquire));
  
  // 原子比较交换操作
  dispatch_source_t expected = currentSource;
  dispatch_source_t newSource = dispatch_source_create(DISPATCH_SOURCE_TYPE_TIMER, 0, 0, dispatch_get_main_queue());
  bool exchanged = msource.compare_exchange_weak(
      reinterpret_cast<void*&>(expected),
      static_cast<void*>(newSource),
      std::memory_order_release,
      std::memory_order_acquire
  );
  

  注意：这种方式需要自行管理ARC内存所有权，确保对象的 retain/release 操作正确，避免内存泄漏或野指针问题。

• 方案二：用Objective-C原生机制保证线程安全
  放弃std::atomic，改用Apple平台惯用的线程安全方案：

  1. 原子属性：直接将变量声明为Objective-C原子属性，系统会自动处理线程安全：

     @property (atomic, strong) dispatch_source_t msource;
     

  2. GCD串行队列保护：用串行队列包裹所有对变量的读写操作，确保同一时间只有一个线程访问：

     // 类成员变量
     dispatch_queue_t _sourceAccessQueue = dispatch_queue_create("com.yourapp.source.access", DISPATCH_QUEUE_SERIAL);
     dispatch_source_t _msource;
     
     // 读取操作
     dispatch_source_t getSource() {
         __block dispatch_source_t result;
         dispatch_sync(_sourceAccessQueue, ^{
             result = _msource;
         });
         return result;
     }
     
     // 比较交换逻辑
     bool compareAndSetSource(dispatch_source_t expected, dispatch_source_t newSource) {
         __block bool success = false;
         dispatch_sync(_sourceAccessQueue, ^{
             if (_msource == expected) {
                 _msource = newSource;
                 success = true;
             }
         });
         return success;
     }
     

内容的提问来源于stack exchange，提问作者Titus