pthread库学习(2): 线程的同步，使用信号量

先看下面这段程序，主线程创建了三个线程，每个线程中均有一个打印语句。

#include <stdio.h>

#include <pthread.h>

#include <semaphore.h>

 void *thread_a(void *in)

{

printf("I am thread_a/n");

pthread_exit((void *)0);

}

 

void *thread_b(void *in)

{

printf("I am thread_b/n");

pthread_exit((void *)0);

}

 

void *thread_c(void *in)

{

    printf("I am thread_c/n");

pthread_exit((void *)0);

}

 

int main()

{

    pthread_t a,b,c;  /* thread id a, b, c*/

    int val;         /* used for function return result */

 

    /* create thread a, b, c */

    pthread_create(&a, NULL, thread_a, (void *)0);

    pthread_create(&b, NULL, thread_b, (void *)0);

    pthread_create(&c, NULL, thread_c, (void *)0);

 

    /* main thread waits for termination of a,b,c */

    pthread_join(a, (void **)0);

    pthread_join(b, (void **)0);

    pthread_join(c, (void **)0);

 

    printf("Main thread is over/n");

    return 0;

}

在Linux下进行编译: gcc -o My_Thread My_Thread.c ./MyThread 
得到的输出:
 I am thread_a 
I am thread_b 
I am thread_c 
Main thread is over 

现在我们希望线程C的最先打印，然后线程B打印，最后线程A打印。即三个线程之间的打印有一定的先后关系。看下面程序：

#include <stdio.h>#include <pthread.h>#include <semaphore.h>sem_t sem1;sem_t sem2;void *thread_a(void *in){sem_wait(&sem1);       /* wait for sem1 */printf("I am thread_a/n");pthread_exit((void *)0);return NULL;}void *thread_b(void *in){sem_wait(&sem2);       /* wait for sem2 */printf("I am thread_b/n");sem_post(&sem1);       /* increase sem1 by 1, make thread_a run*/pthread_exit((void *)0);return NULL;}void *thread_c(void *in){printf("I am thread_c/n");sem_post(&sem2);      /* increase sem2 by 1, make thread_b run*/pthread_exit((void *)0);return NULL;}int main(){pthread_t a,b,c;  /* thread id a, b, c*//* init sem1 sem2 to 0 , any thread waits for it will be blocked*/sem_init(&sem1, 0, 0);sem_init(&sem2, 0, 0);/* create thread a, b, c */pthread_create(&a, NULL, thread_a, (void *)0);pthread_create(&b, NULL, thread_b, (void *)0);pthread_create(&c, NULL, thread_c, (void *)0);/* main thread waits for termination of a,b,c */pthread_join(a, (void **)0);pthread_join(b, (void **)0);pthread_join(c, (void **)0);/* destroy sem1 sem2 */sem_destroy(&sem1);sem_destroy(&sem2);printf("Main thread is over/n");return 0;}

gcc -o My_thread1 My_thread1.c -lpthread
./My_thread1
I am thread_c
I am thread_b
I am thread_a
Main thread is over

可以见到，线程的执行顺序已经改变。

转自 http://blog.csdn.net/direstrait/article/details/566716