【整理】-- C++ 多线程

os:ubuntu   c++

1.创建线程

#include <iostream> 

#include <pthread.h> //多线程相关操作头文件,可移植众多平台 

using namespace std; 

#define NUM_THREADS 5 //线程数 

void* say_hello( void* args ) 

    cout << "hello..." << endl; 

} //函数返回的是函数指针,便于后面作为参数 

int main() 

    pthread_t tids[NUM_THREADS]; //线程id 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        int ret = pthread_create( &tids[i], NULL, say_hello, NULL ); //参数:创建的线程id,线程参数,线程运行函数的起始地址,运行函数的参数 

        if( ret != 0 ) //创建线程成功返回

        { 

            cout << "pthread_create error:error_code=" << ret << endl; 

        } 

    } 

    pthread_exit( NULL ); //等待各个线程退出后,进程才结束,否则进程强制结束,线程处于未终止的状态 

输入命令:g++ -o muti_thread_test_1 muti_thread_test_1.cpp -lpthread

 

注意:

1)此为c++程序,故用g++来编译生成可执行文件,并且要调用处理多线程操作相关的静态链接库文件pthread

2-lpthread 编译选项到位置可任意,如g++ -lpthread -o muti_thread_test_1 muti_thread_test_1.cpp

3)注意gccg++的区别,转到此文:点击打开链接

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_1 

hello...hello... 

hello... 

hello... 

hello... 

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_1 

hello...hello...hello... 

hello... 

hello... 

可知,两次运行的结果会有差别,这不是多线程的特点吧?这显然没有同步?还有待进一步探索...

多线程的运行是混乱的,混乱就是正常?

2.线程调用到函数在一个类中,那必须将该函数声明为静态函数函数

因为静态成员函数属于静态全局区,线程可以共享这个区域,故可以各自调用。

[html] view plaincopy

#include <iostream> 

#include <pthread.h> 

using namespace std; 

#define NUM_THREADS 5 

class Hello 

public: 

    static void* say_hello( void* args ) 

    { 

        cout << "hello..." << endl; 

    } 

}; 

int main() 

    pthread_t tids[NUM_THREADS]; 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        int ret = pthread_create( &tids[i], NULL, Hello::say_hello, NULL ); 

        if( ret != 0 ) 

        { 

            cout << "pthread_create error:error_code" << ret << endl; 

        } 

    } 

    pthread_exit( NULL ); 

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_2 

hello... 

hello... 

hello... 

hello... 

hello... 

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_2 

hello...hello...hello... 

hello... 

hello... 

3.如何在线程调用函数时传入参数呢?

先看下面修改的代码,传入线程编号作为参数:

[html] view plaincopy

#include <iostream> 

#include <pthread.h> //多线程相关操作头文件,可移植众多平台 

using namespace std; 

#define NUM_THREADS 5 //线程数 

void* say_hello( void* args ) 

    int i = *( (int*)args ); //对传入的参数进行强制类型转换,由无类型指针转变为整形指针,再用*读取其指向到内容 

    cout << "hello in " << i <<  endl; 

} //函数返回的是函数指针,便于后面作为参数 

int main() 

    pthread_t tids[NUM_THREADS]; //线程id 

    cout << "hello in main.." << endl; 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        int ret = pthread_create( &tids[i], NULL, say_hello, (void*)&i ); //传入到参数必须强转为void*类型,即无类型指针,&i表示取i的地址,即指向i的指针 

        cout << "Current pthread id = " << tids[i] << endl; //tids数组打印创建的进程id信息 

        if( ret != 0 ) //创建线程成功返回

        { 

            cout << "pthread_create error:error_code=" << ret << endl; 

        } 

    } 

    pthread_exit( NULL ); //等待各个线程退出后,进程才结束,否则进程强制结束,线程处于未终止的状态 

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_3 

hello in main.. 

Current pthread id = 3078458224 

Current pthread id = 3070065520 

hello in hello in 2 

Current pthread id = hello in 2 

3061672816 

Current pthread id = 3053280112 

hello in 4 

Current pthread id = hello in 4 

3044887408 

显然不是想要的结果,调用顺序很乱,这是为什么呢?

这是因为多线程到缘故,主进程还没开始对i赋值,线程已经开始跑了...?

修改代码如下:

[html] view plaincopy

#include <iostream> 

#include <pthread.h> //多线程相关操作头文件,可移植众多平台 

using namespace std; 

#define NUM_THREADS 5 //线程数 

void* say_hello( void* args ) 

    cout << "hello in thread " << *( (int *)args ) <<  endl; 

} //函数返回的是函数指针,便于后面作为参数 

int main() 

    pthread_t tids[NUM_THREADS]; //线程id 

    int indexes[NUM_THREADS]; //用来保存i的值避免被修改 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        indexes[i] = i; 

        int ret = pthread_create( &tids[i], NULL, say_hello, (void*)&(indexes[i]) ); 

        if( ret != 0 ) //创建线程成功返回

        { 

            cout << "pthread_create error:error_code=" << ret << endl; 

        } 

    } 

    for( int i = 0; i < NUM_THREADS; ++i ) 

        pthread_join( tids[i], NULL ); //pthread_join用来等待一个线程的结束,是一个线程阻塞的函数 

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_3 

hello in thread hello in thread hello in thread hello in thread hello in thread 30124 

这是正常的吗?感觉还是有问题...待续

代码中如果没有pthread_join主线程会很快结束从而使整个进程结束,从而使创建的线程没有机会开始执行就结束了。加入pthread_join后,主线程会一直等待直到等待的线程结束自己才结束,使创建的线程有机会执行。

4.线程创建时属性参数的设置pthread_attr_tjoin功能的使用

线程的属性由结构体pthread_attr_t进行管理。

typedef struct

{

    int                           detachstate;     线程的分离状态

    int                          schedpolicy;   线程调度策略

    struct sched_param      schedparam;   线程的调度参数

    int inheritsched; 线程的继承性

    int scope; 线程的作用域

    size_t guardsize; 线程栈末尾的警戒缓冲区大小

    int stackaddr_set; void * stackaddr; 线程栈的位置

    size_t stacksize; 线程栈的大小

}pthread_attr_t;

[html] view plaincopy

#include <iostream> 

#include <pthread.h> 

using namespace std; 

#define NUM_THREADS 5 

void* say_hello( void* args ) 

    cout << "hello in thread " << *(( int * )args) << endl; 

    int status = 10 + *(( int * )args); //线程退出时添加退出的信息,status供主程序提取该线程的结束信息 

    pthread_exit( ( void* )status );  

int main() 

    pthread_t tids[NUM_THREADS]; 

    int indexes[NUM_THREADS]; 

     

    pthread_attr_t attr; //线程属性结构体,创建线程时加入的参数 

    pthread_attr_init( &attr ); //初始化 

    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE ); //是设置你想要指定线程属性参数,这个参数表明这个线程是可以join连接的,join功能表示主程序可以等线程结束后再去做某事,实现了主程序和线程同步功能 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        indexes[i] = i; 

        int ret = pthread_create( &tids[i], &attr, say_hello, ( void* )&( indexes[i] ) ); 

        if( ret != 0 ) 

        { 

        cout << "pthread_create error:error_code=" << ret << endl; 

    } 

    }  

    pthread_attr_destroy( &attr ); //释放内存  

    void *status; 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

    int ret = pthread_join( tids[i], &status ); //主程序join每个线程后取得每个线程的退出信息status 

    if( ret != 0 ) 

    { 

        cout << "pthread_join error:error_code=" << ret << endl; 

    } 

    else 

    { 

        cout << "pthread_join get status:" << (long)status << endl; 

    } 

    } 

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_4 

hello in thread hello in thread hello in thread hello in thread 0hello in thread 321 

pthread_join get status:10 

pthread_join get status:11 

pthread_join get status:12 

pthread_join get status:13 

pthread_join get status:14 

5.互斥锁的实现

互斥锁是实现线程同步的一种机制,只要在临界区前后对资源加锁就能阻塞其他进程的访问。

[html] view plaincopy

#include <iostream> 

#include <pthread.h> 

using namespace std; 

#define NUM_THREADS 5 

int sum = 0; //定义全局变量,让所有线程同时写,这样就需要锁机制 

pthread_mutex_t sum_mutex; //互斥锁 

void* say_hello( void* args ) 

    cout << "hello in thread " << *(( int * )args) << endl; 

    pthread_mutex_lock( &sum_mutex ); //先加锁,再修改sum的值,锁被占用就阻塞,直到拿到锁再修改sum; 

    cout << "before sum is " << sum << " in thread " << *( ( int* )args ) << endl; 

    sum += *( ( int* )args ); 

    cout << "after sum is " << sum << " in thread " << *( ( int* )args ) << endl; 

    pthread_mutex_unlock( &sum_mutex ); //释放锁,供其他线程使用 

    pthread_exit( 0 );  

int main() 

    pthread_t tids[NUM_THREADS]; 

    int indexes[NUM_THREADS]; 

     

    pthread_attr_t attr; //线程属性结构体,创建线程时加入的参数 

    pthread_attr_init( &attr ); //初始化 

    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE ); //是设置你想要指定线程属性参数,这个参数表明这个线程是可以join连接的,join功能表示主程序可以等线程结束后再去做某事,实现了主程序和线程同步功能 

    pthread_mutex_init( &sum_mutex, NULL ); //对锁进行初始化     

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

        indexes[i] = i; 

        int ret = pthread_create( &tids[i], &attr, say_hello, ( void* )&( indexes[i] ) ); //5个进程同时去修改sum 

        if( ret != 0 ) 

        { 

        cout << "pthread_create error:error_code=" << ret << endl; 

    } 

    }  

    pthread_attr_destroy( &attr ); //释放内存  

    void *status; 

    for( int i = 0; i < NUM_THREADS; ++i ) 

    { 

    int ret = pthread_join( tids[i], &status ); //主程序join每个线程后取得每个线程的退出信息status 

    if( ret != 0 ) 

    { 

        cout << "pthread_join error:error_code=" << ret << endl; 

    } 

    } 

    cout << "finally sum is " << sum << endl; 

    pthread_mutex_destroy( &sum_mutex ); //注销锁 

测试结果:

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_5 

hello in thread hello in thread hello in thread 410 

before sum is hello in thread 0 in thread 4 

after sum is 4 in thread 4hello in thread  

before sum is 4 in thread 1 

after sum is 5 in thread 1 

before sum is 5 in thread 0 

after sum is 5 in thread 0 

before sum is 5 in thread 2 

after sum is 7 in thread 2 

before sum is 7 in thread 3 

after sum is 10 in thread 3 

finally sum is 10 

可知,sum的访问和修改顺序是正常的,这就达到了多线程的目的了,但是线程的运行顺序是混乱的,混乱就是正常?

6.信号量的实现

信号量是线程同步的另一种实现机制,信号量的操作有signalwait,本例子采用条件信号变量pthread_cond_t tasks_cond;

信号量的实现也要给予锁机制。

[html] view plaincopy

#include <iostream> 

#include <pthread.h> 

#include <stdio.h> 

using namespace std; 

#define BOUNDARY 5 

int tasks = 10; 

pthread_mutex_t tasks_mutex; //互斥锁 

pthread_cond_t tasks_cond; //条件信号变量,处理两个线程间的条件关系,当task>5hello2处理,反之hello1处理,直到task减为

void* say_hello2( void* args ) 

    pthread_t pid = pthread_self(); //获取当前线程id 

    cout << "[" << pid << "] hello in thread " <<  *( ( int* )args ) << endl; 

     

    bool is_signaled = false; //sign 

    while(1) 

    { 

    pthread_mutex_lock( &tasks_mutex ); //加锁 

    if( tasks > BOUNDARY ) 

    { 

        cout << "[" << pid << "] take task: " << tasks << " in thread " << *( (int*)args ) << endl; 

        --tasks; //modify 

    } 

    else if( !is_signaled ) 

    { 

        cout << "[" << pid << "] pthread_cond_signal in thread " << *( ( int* )args ) << endl; 

        pthread_cond_signal( &tasks_cond ); //signal:hello1发送信号,表明已经>5 

        is_signaled = true; //表明信号已发送,退出此线程 

    } 

    pthread_mutex_unlock( &tasks_mutex ); //解锁 

    if( tasks == 0 ) 

        break; 

    }     

void* say_hello1( void* args ) 

    pthread_t pid = pthread_self(); //获取当前线程id 

    cout << "[" << pid << "] hello in thread " <<  *( ( int* )args ) << endl; 

    while(1) 

    { 

        pthread_mutex_lock( &tasks_mutex ); //加锁 

        if( tasks > BOUNDARY ) 

        { 

        cout << "[" << pid << "] pthread_cond_signal in thread " << *( ( int* )args ) << endl; 

        pthread_cond_wait( &tasks_cond, &tasks_mutex ); //wait:等待信号量生效,接收到信号,向hello2发出信号,跳出wait,执行后续  

        } 

        else 

        { 

        cout << "[" << pid << "] take task: " << tasks << " in thread " << *( (int*)args ) << endl; 

            --tasks; 

    } 

        pthread_mutex_unlock( &tasks_mutex ); //解锁 

        if( tasks == 0 ) 

            break; 

    }  

int main() 

    pthread_attr_t attr; //线程属性结构体,创建线程时加入的参数 

    pthread_attr_init( &attr ); //初始化 

    pthread_attr_setdetachstate( &attr, PTHREAD_CREATE_JOINABLE ); //是设置你想要指定线程属性参数,这个参数表明这个线程是可以join连接的,join功能表示主程序可以等线程结束后再去做某事,实现了主程序和线程同步功能 

    pthread_cond_init( &tasks_cond, NULL ); //初始化条件信号量 

    pthread_mutex_init( &tasks_mutex, NULL ); //初始化互斥量 

    pthread_t tid1, tid2; //保存两个线程id 

    int index1 = 1; 

    int ret = pthread_create( &tid1, &attr, say_hello1, ( void* )&index1 ); 

    if( ret != 0 ) 

    { 

        cout << "pthread_create error:error_code=" << ret << endl; 

    } 

    int index2 = 2; 

    ret = pthread_create( &tid2, &attr, say_hello2, ( void* )&index2 ); 

    if( ret != 0 ) 

    { 

        cout << "pthread_create error:error_code=" << ret << endl; 

    } 

    pthread_join( tid1, NULL ); //连接两个线程 

    pthread_join( tid2, NULL );  

    pthread_attr_destroy( &attr ); //释放内存  

    pthread_mutex_destroy( &tasks_mutex ); //注销锁 

    pthread_cond_destroy( &tasks_cond ); //正常退出 

测试结果:

先在线程2中执行say_hello2,再跳转到线程1中执行say_hello1,直到tasks减到0为止。

[html] view plaincopy

wq@wq-desktop:~/coding/muti_thread$ ./muti_thread_test_6 

[3069823856] hello in thread 2 

[3078216560] hello in thread 1[3069823856] take task: 10 in thread 2 

[3069823856] take task: 9 in thread 2 

[3069823856] take task: 8 in thread 2 

[3069823856] take task: 7 in thread 2 

[3069823856] take task: 6 in thread 2 

[3069823856] pthread_cond_signal in thread 2 

[3078216560] take task: 5 in thread 1 

[3078216560] take task: 4 in thread 1 

[3078216560] take task: 3 in thread 1 

[3078216560] take task: 2 in thread 1 

[3078216560] take task: 1 in thread 1 

到此,对多线程编程有了一个初步的了解,当然还有其他实现线程同步的机制,有待进一步探索

原文地址:https://www.cnblogs.com/apolloenterprise/p/4341275.html