#ifndef cwMpScQueue_h #define cwMpScQueue_h namespace cw { template class MpScQueue { public: typedef struct node_str { struct node_str* next = nullptr; T* payload = nullptr; } node_t; MpScQueue() { node_t* stub = mem::allocZ(); _head = stub; // last-in _tail = stub; // first-out mutex::create( _mH ); } virtual ~MpScQueue() { mem::free(_tail); mutex::destroy( _mH ); } MpScQueue( const MpScQueue& ) = delete; MpScQueue( const MpScQueue&& ) = delete; MpScQueue& operator=(const MpScQueue& ) = delete; MpScQueue& operator=(const MpScQueue&& ) = delete; void push( T* payload ) { node_t* new_node = mem::allocZ(1); new_node->payload = payload; new_node->next = nullptr; if( mutex::lock(_mH) == kOkRC ) { // Note that the elements of the queue are only accessed from the end of the queue (tail). // New nodes can therefore safely be updated in two steps: node_t* prev = _head; _head = new_node; // 1. Atomically set _head to the new node and return 'old-head' //node_t* prev = _head.exchange(new_node,std::memory_order_acq_rel); // Note that at this point only the new node may have the 'old-head' as it's predecssor. // Other threads may therefore safely interrupt at this point. // 2. Set the old-head next pointer to the new node (thereby adding the new node to the list) prev->next = new_node; // RELEASE 'next' to consumer mutex::unlock(_mH); } } T* pop() { T* payload = nullptr; node_t* t = _tail; if( mutex::lock(_mH) == kOkRC ) { node_t* next = t->next; // ACQUIRE 'next' from producer if( next != nullptr ) { _tail = next; payload = next->payload; mem::free(t); } mutex::unlock(_mH); } return payload; } bool isempty() const { return _tail->next == nullptr; // ACQUIRE 'next' from producer } private: node_t* _stub; node_t* _tail; node_t* _head; mutex::handle_t _mH; }; } #endif