zthread学习 实例六 访问控制

就是对共享资源的同步访问，以免造成不确定的状态。

Zhread库是基于Mutex临界区来组建同步机制的。

---

#include "stdafx.h"#include <iostream>#include <fstream>#include "zthread/Guard.h"#include "zthread/Mutex.h"#include "zthread/ThreadedExecutor.h"#include "zthread/Cancelable.h"#include "zthread/Runnable.h"#include "zthread/Thread.h"using namespace ZThread;using namespace std;//用于共享资源的虚基类，class Generator : public Cancelable{public:Generator(): canceled(false){}virtual int nextValue() = 0;void cancel() {canceled = true;}bool isCanceled(){return canceled;}private:bool canceled;};//实现基类的虚函数nextValue（）class EvenGenerator : public Generator{public:EvenGenerator(){currentEvenValue = 0;}~EvenGenerator(){cout << "~EvenGenerator" << endl;}int nextValue(){Guard<Mutex> g(Lock);//同步控制currentEvenValue++;Thread::yield();//放大因线程调度问题产生的不一致currentEvenValue++;return currentEvenValue;}private:unsigned int currentEvenValue;MutexLock;};//任务类，重点run()函数class EvenChecker : public Runnable{public:EvenChecker(const CountedPtr<Generator>& g, int idn): pGenerator(g), id(idn){}~EvenChecker(){cout << "~EvenChecker" << id <<endl;}void run(){while (!pGenerator->isCanceled())//检查任务共享资源是否已退出，通过控制共享资源来控制所有创建的任务是否需要返回{int val = pGenerator->nextValue();if (val % 2 != 0){cout << val << " not even!" <<endl;pGenerator->cancel();//共享资源退出}else{cout <<"ID: "<<id << ", value = " << val<<endl;Thread::sleep(200);}}}template<typename GenType> static void test(int n = 5){try{ThreadedExecutor executor;CountedPtr<Generator> pGp(new GenType);for (int i = 0; i < n; i++){executor.execute(new EvenChecker(pGp, i));//引用同一个Generator，多个任务共享}}catch (Synchronization_Exception& e){cerr << e.what() << endl;}}private:CountedPtr<Generator> pGenerator;int id;};int _tmain(int argc, _TCHAR* argv[]){try{EvenChecker::test<EvenGenerator>();}catch (Synchronization_Exception& e){cerr << e.what() <<endl;}cin.get();return 0;}

Guard<>模板很方便的定义了同步机制，它在创建时用acquire()来获得一个Lockable对象，被销毁时用release()来释放这个锁。Guard对象的创建很好的利用了变量作用域概念。例如：

 

void fun()

{

      ... ...

      {    //特意添加的用于控制Guard作用域的括号

            Guard<Mutex> g(Lock);

            ... ...//重点临界区代码

       }

}

 

Zhread库中给Guard<>提供了4种类型的锁策略：

1、CompoundScope

     Note: Locking policy that aggregates two policies that share a target.It is not appropriate to use with any type of OverlappedScope

 

2、LockedScope（默认）

     Note: Locking policy for Lockable objects. This policy acquire()s a Lockable when the protection scope is created, and it release()s a Lockable when the scope is destroyed.

 

3、UnlockedScope（解锁）

     Note: Locking policy for Lockable objects. This policy release()s a Lockable when the protection scope is created, and it acquire()s a Lockable when the scope is destroyed.

4、TimedLockedScope（带时间）

      Note:  Locking policy that attempts to enterScope some resource in a certain amount of time using an tryEnterScope-relase protocol.

 

5、OverlappedScope

    Note:  Locking policy allows the effective scope of two locks to overlap by releasing and disabling one lock before its Guard does so.