C++ 多线程2 beginthread 启动线程知识 20131021
Reference: http://blog.csdn.net/laoyang360/article/details/7720656
前言:
之前曾经使用过WINAPI创建线程并且启动的知识,还有线程之间同步的知识,这里讲解一下使用__beginthread的原理和机制。
1._beginthread 简单的介绍
如果我们编写的是C++的程序,不应该使用CreateThread WINAPI接口去创建线程,而是应该使用Visual C++运行库函数_beginthread,退出线程的时候应该使用_endthread。因为_beginthreaex和_endthreadex是CRT的线程函数,所以必须注意编译选项runtimelibrary,使用的是MT或者是MTD(MultiThreaded,debug Multithread);_beginthread函数的参数列表和CreateThread的参数列表完全相同,但是参数名称和类型不是完全的相同。因为Microsoft的C/C++运行库的开的小组认为C/C++运行期函数不应该对Windows数据类型有任何的依赖。
每一个线程都会获得由C/C++运行期库的堆栈分配自己的tiddata内存结构;传递给_beginthreadex的线程函数的地址保存在tiddata内存数据块,传递给该函数的参数也保存在该数据块中;_beginthreadex确实从内部调用CreateThread函数,因为这是创建线程早OS层面上的唯一的方法;当调用CreateThread函数的时候,他会被告知通过调用_threadstartex而不是pfnStartAddr来启动执行新的线程,还有传递给线程函数的从哪回溯是tiddata结构而不是pvParam的地址;如果一切顺利的话,就会像CreateThread那样返回线程句柄,任何操作失败就会返回NULL。
总的来说就是_beginthreadex在内部调用CreateThread函数,在调用之前_beginthreadex做了很多的工作,从而比CreateThread更加安全.
#include <iostream>
#include <string>
#include <process.h>
#include <Windows.h>
using namespace std;
class ThreadX{
private:
int loopStart;
int loopEnd;
int dispFrequency;
public:
string threadName;
ThreadX(int startVal, int endVal, int frequency){
this->loopStart = startVal;
this->loopEnd = endVal;
this->dispFrequency = frequency;
}
static unsigned __stdcall ThreadStaticEntryPoint(void * pThis){
ThreadX* pThX = (ThreadX*) pThis;
pThX->ThreadEntryPoint();
return 1;
}
void ThreadEntryPoint(){
for (int i = loopStart; i <= loopEnd; i++){
cout << threadName << " i = " << i << endl;
Sleep(100);
}
}
};
int main(){
ThreadX * pThX = new ThreadX(0, 10, 2000);
HANDLE hth1 = NULL;
unsigned uiThread1ID;
hth1 =(HANDLE) _beginthreadex(NULL,
0,
ThreadX::ThreadStaticEntryPoint,
pThX,
CREATE_SUSPENDED,
&uiThread1ID);
if (hth1 == NULL){
cout << "failed to create thread 1" << endl;
}
DWORD dwExitCode;
GetExitCodeThread(hth1, &dwExitCode); // shoule be STILL_ACTIVE
cout << "init thread 1 exit code id = " << dwExitCode << endl;
pThX->threadName = "yang1";
ThreadX* pThX2 = new ThreadX(0, 10, 2000);
HANDLE hth2 = NULL;
unsigned uiThread2ID;
hth2 = (HANDLE) _beginthreadex(NULL,
0,
ThreadX::ThreadStaticEntryPoint,
pThX2,
CREATE_SUSPENDED,
&uiThread2ID);
if (hth2 == NULL){
cout << "create thread 2 failed" << endl;
}
GetExitCodeThread(hth2, &dwExitCode);
cout << "init thread 2 exit code id = " << dwExitCode << endl;
pThX2->threadName = "yang2";
ResumeThread(hth1);
ResumeThread(hth2);
WaitForSingleObject(hth1, INFINITE);
WaitForSingleObject(hth2, INFINITE);
GetExitCodeThread(hth1, &dwExitCode);
cout << "thread 1 exited with exit code " << dwExitCode << endl;
GetExitCodeThread(hth2, &dwExitCode);
cout << "thread 2 exited with exit code " << dwExitCode << endl;
CloseHandle(hth1);
CloseHandle(hth2);
delete pThX;
delete pThX2;
pThX = NULL;
pThX2 = NULL;
cout << "end program " << endl;
return 0;
}
2.线程临界区
线程之间共享的对象,当修改的时候,只能够一个线程修改,同时不存在其他的线程访问该资源,这就是线程之间的同步问题.在上一篇总结中使用的是CreateMutex函数创建一个HANDLE,这样的话创建一个Mutex,当线程访问资源的时候,使用WaitForSimpleObject(handl, INFINITE);获得该对象的锁,然后再使用完的时候,释放对象锁,ReleaseMutex(handle);
在线程中有一个概念是临界区,在C++中使用这一种实现的,就是用CRITICAL_SECTION 声明一个临界区变量,然后再程序开始的时候,使用InitializeCriticalSelection(CRITICAL_SELECTION*) 初始化临界变量,然后当线程进入临界区的时候,使用EneterCriticalSelection(CRITICAL_SELECTION*);当访问完成的时候使用LeaveCriticalSelection(CRITICAL_SELECTION);
#include <iostream>
#include<Windows.h>
using namespace std;
bool g_bContinue = true;
int g_count1 = 0;
int g_count2 = 0;
DWORD WINAPI ThreadProc(LPVOID lpParam){
for (int i = 0; i < 20; i++)
{
EnterCriticalSection(&g_cs);
g_count1++;
g_count2++;
LeaveCriticalSection(&g_cs);
}
return 0;
}
int main(){
HANDLE g_hThr[2];
InitializeCriticalSection(&g_cs);
g_hThr[0] = CreateThread(NULL, 0, ThreadProc, NULL, 0, NULL);
g_hThr[1] = CreateThread(NULL, 0, ThreadProc, NULL, 0, NULL);
WaitForSingleObject( g_hThr[0], INFINITE);
WaitForSingleObject( g_hThr[1], INFINITE);
g_bContinue = false;
CloseHandle(g_hThr[0]);
CloseHandle(g_hThr[1]);
DeleteCriticalSection(&g_cs);
Sleep(1000);
cout << "g_count1 = " << g_count1 << " g_count2 = " << g_count2 << endl;
}
追梦的飞飞
于广州中山大学 20131022
HomePage: http://yangtengfei.duapp.com