阻塞队列是后台开发中多线程异步架构的基本数据结构,像python, java 都提供线程安全的阻塞队列,c++ 可能需要自己实现一个模板。
从性能考虑,自己没有使用STL的queue作为基本数据结构,而是使用循环数组作为基本数据结构,性能应该比queue高,省去了动态内存分配和回收。
确定就是,队列大小不可动态扩展,当时实际开发中,可以通过压力测试,配置合适的队列大小。
/******************************************** function: thread safe blocking queue. author: liuyi date: 2014.11.13 version: 2.0 ********************************************/ #ifndef BLOCK_QUEUE_H #define BLOCK_QUEUE_H #include <iostream> #include <pthread.h> #include <sys/time.h> using namespace std; template<class T> class block_queue { public: block_queue(int max_size = 1000) { if(max_size <= 0) { exit(-1); } m_max_size = max_size; m_array = new T[max_size]; m_size = 0; m_front = -1; m_back = -1; m_mutex = new pthread_mutex_t; m_cond = new pthread_cond_t; pthread_mutex_init(m_mutex, NULL); pthread_cond_init(m_cond, NULL); } ~block_queue() { pthread_mutex_lock(m_mutex); if(m_array != NULL) delete m_array; pthread_mutex_unlock(m_mutex); pthread_mutex_destroy(m_mutex); pthread_cond_destroy(m_cond); delete m_mutex; delete m_cond; } bool full()const { pthread_mutex_lock(m_mutex); if(m_size >= m_max_size) { pthread_mutex_unlock(m_mutex); return true; } pthread_mutex_unlock(m_mutex); return false; } bool empty()const { pthread_mutex_lock(m_mutex); if(0 == m_size) { pthread_mutex_unlock(m_mutex); return true; } pthread_mutex_unlock(m_mutex); return false; } T front()const { T tmp; pthread_mutex_lock(m_mutex); tmp = m_array[m_front]; pthread_mutex_unlock(m_mutex); return tmp; } T back()const { T tmp; pthread_mutex_lock(m_mutex); tmp = m_array[m_back]; pthread_mutex_unlock(m_mutex); return tmp; } int size()const { int tmp = 0; pthread_mutex_lock(m_mutex); tmp = m_size; pthread_mutex_unlock(m_mutex); return tmp; } int max_size()const { int tmp = 0; pthread_mutex_lock(m_mutex); tmp = m_max_size; pthread_mutex_unlock(m_mutex); return tmp; } bool push(const T& item) { pthread_mutex_lock(m_mutex); if(m_size >= m_max_size) { pthread_cond_broadcast(m_cond); pthread_mutex_unlock(m_mutex); return false; } m_back = (m_back + 1) % m_max_size; m_array[m_back] = item; m_size++; pthread_cond_broadcast(m_cond); pthread_mutex_unlock(m_mutex); return true; } bool pop(T& item) { pthread_mutex_lock(m_mutex); while(m_size <= 0) { if(0 != pthread_cond_wait(m_cond, m_mutex)) { pthread_mutex_unlock(m_mutex); return false; } } m_front = (m_front + 1) % m_max_size; item = m_array[m_front]; m_size--; pthread_mutex_unlock(m_mutex); return true; } bool pop(T& item, int ms_timeout) { struct timespec t = {0,0}; struct timeval now = {0,0}; gettimeofday(&now, NULL); pthread_mutex_lock(m_mutex); if(m_size <= 0) { t.tv_sec = now.tv_sec + ms_timeout/1000; t.tv_nsec = (ms_timeout % 1000)*1000; if(0 != pthread_cond_timedwait(m_cond, m_mutex, &t)) { pthread_mutex_unlock(m_mutex); return false; } } m_front = (m_front + 1) % m_max_size; item = m_array[m_front]; m_size--; pthread_mutex_unlock(m_mutex); return true; } private: pthread_mutex_t *m_mutex; pthread_cond_t *m_cond; T *m_array; int m_size; int m_max_size; int m_front; int m_back; }; #endif
#include<iostream> #include"block_queue.h" using namespace std; block_queue<int> g_queue(100); void *p(void *args) { sleep(1); int data = 0; for(int i = 0; i < 100; i++) { g_queue.push(data++); } return NULL; } void *c(void* args) { while(true) { int t = 0; if(!g_queue.pop(t,1000)) { cout<<"timeout"<<endl; continue; } else { cout<<t<<endl; } g_queue.pop(t); cout<<"block="<<t<<endl; } return NULL; } int main() { pthread_t id; pthread_create(&id, NULL, p, NULL); //pthread_create(&id, NULL, p, NULL); //pthread_create(&id, NULL, c, NULL); pthread_create(&id, NULL, c, NULL); for(;;)sleep(1); return 0; }
原文地址:http://blog.csdn.net/coder_yi_liu/article/details/41093453