线程同步（一）

上篇文章讲述了什么是线程，以及在Linux系统下线程的相关操作

 线程（Linux系统实现）_小梁今天敲代码了吗的博客-CSDN博客

本文将继续讲述线程的相关知识——线程同步

目录

1.线程同步的概念

2.线程不同步可能会发生什么

3.线程同步方式   

4.互斥锁

申请一个互斥锁

尝试获取互斥锁

互斥锁解锁

互斥锁示例

1.线程同步的概念

        线程同步是指在多线程编程中，为了保证多个线程按照某种特定的方式正确、有序地执行，需要进行线程间的协作与同步。在多线程编程中，当多个线程共享同一份资源时，由于线程的执行顺序是不确定的，因此会存在一些并发问题，如死锁、竞态条件、资源争用等问题。为了避免这些问题，需要对线程进行同步。线程同步实际上就是通过线程之间的协作，使得线程能够按照一定的顺序来访问共享资源，从而避免并发问题的发生。常用的线程同步机制有互斥锁、信号量、条件变量等。

2.线程不同步可能会发生什么

线程不同步可能会导致以下问题：

1. 竞态条件（Race Condition）：多个线程同时访问、修改一份共享资源，可能会导致资源的状态不确定，进而导致程序出现逻辑错误，甚至崩溃。

2. 死锁（Deadlock）：多个线程在等待对方释放锁，导致所有线程都无法继续执行，程序陷入死循环，最终可能会崩溃。

3. 饥饿（Starvation）：某些线程可能因为无法获取资源而一直等待，导致无法正常执行，进而影响整个程序的性能。

4. 资源争用（Resource Contention）：多个线程同时竞争同一份资源，导致资源的使用效率下降，总体性能降低。

示例：两个线程交替数数（每个线程数 50 个数，交替数到 100）

#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <string.h>
#include <pthread.h>

#define MAX 50
// 全局变量
int number;

// 线程处理函数
void* funcA_num(void* arg)
{
    for(int i=0; i<MAX; ++i)
    {
        int cur = number;
        cur++;
        usleep(10);
        number = cur;
        printf("Thread A, id = %lu, number = %d\n", pthread_self(), number);
    }

    return NULL;
}

void* funcB_num(void* arg)
{
    for(int i=0; i<MAX; ++i)
    {
        int cur = number;
        cur++;
        number = cur;
        printf("Thread B, id = %lu, number = %d\n", pthread_self(), number);
        usleep(5);
    }

    return NULL;
}

int main(int argc, const char* argv[])
{
    pthread_t p1, p2;

    // 创建两个子线程
    pthread_create(&p1, NULL, funcA_num, NULL);
    pthread_create(&p2, NULL, funcB_num, NULL);

    // 阻塞，资源回收
    pthread_join(p1, NULL);
    pthread_join(p2, NULL);

    return 0;
}

在linux系统中编译，运行结果：

[itliang@localhost ~]$ ./xctb
Thread B, id = 140388304152320, number = 1
Thread A, id = 140388312545024, number = 1
Thread B, id = 140388304152320, number = 2
Thread A, id = 140388312545024, number = 2
Thread B, id = 140388304152320, number = 3
Thread A, id = 140388312545024, number = 3
Thread B, id = 140388304152320, number = 4
Thread A, id = 140388312545024, number = 4
Thread B, id = 140388304152320, number = 5
Thread A, id = 140388312545024, number = 5
Thread B, id = 140388304152320, number = 6
Thread B, id = 140388304152320, number = 7
Thread A, id = 140388312545024, number = 6
Thread B, id = 140388304152320, number = 7
Thread A, id = 140388312545024, number = 7
Thread A, id = 140388312545024, number = 8
Thread B, id = 140388304152320, number = 9
Thread B, id = 140388304152320, number = 10
Thread A, id = 140388312545024, number = 9
Thread A, id = 140388312545024, number = 10
Thread B, id = 140388304152320, number = 11
Thread B, id = 140388304152320, number = 12
Thread A, id = 140388312545024, number = 11
Thread A, id = 140388312545024, number = 12
Thread B, id = 140388304152320, number = 13
Thread A, id = 140388312545024, number = 13
Thread B, id = 140388304152320, number = 14
Thread B, id = 140388304152320, number = 15
Thread A, id = 140388312545024, number = 14
Thread B, id = 140388304152320, number = 15
Thread A, id = 140388312545024, number = 15
Thread A, id = 140388312545024, number = 16
Thread B, id = 140388304152320, number = 17
Thread B, id = 140388304152320, number = 18
Thread A, id = 140388312545024, number = 17
Thread A, id = 140388312545024, number = 18
Thread B, id = 140388304152320, number = 19
Thread A, id = 140388312545024, number = 19
Thread B, id = 140388304152320, number = 20
Thread B, id = 140388304152320, number = 21
Thread A, id = 140388312545024, number = 20
Thread A, id = 140388312545024, number = 21
Thread B, id = 140388304152320, number = 22
Thread B, id = 140388304152320, number = 23
Thread A, id = 140388312545024, number = 22
Thread A, id = 140388312545024, number = 23
Thread B, id = 140388304152320, number = 24
Thread B, id = 140388304152320, number = 25
Thread A, id = 140388312545024, number = 24
Thread B, id = 140388304152320, number = 25
Thread A, id = 140388312545024, number = 25
Thread A, id = 140388312545024, number = 26
Thread B, id = 140388304152320, number = 27
Thread B, id = 140388304152320, number = 28
Thread A, id = 140388312545024, number = 27
Thread A, id = 140388312545024, number = 28
Thread B, id = 140388304152320, number = 29
Thread A, id = 140388312545024, number = 29
Thread B, id = 140388304152320, number = 30
Thread B, id = 140388304152320, number = 31
Thread A, id = 140388312545024, number = 30
Thread A, id = 140388312545024, number = 31
Thread B, id = 140388304152320, number = 32
Thread B, id = 140388304152320, number = 33
Thread A, id = 140388312545024, number = 32
Thread A, id = 140388312545024, number = 33
Thread B, id = 140388304152320, number = 34
Thread B, id = 140388304152320, number = 35
Thread A, id = 140388312545024, number = 34
Thread A, id = 140388312545024, number = 35
Thread B, id = 140388304152320, number = 36
Thread B, id = 140388304152320, number = 37
Thread A, id = 140388312545024, number = 36
Thread A, id = 140388312545024, number = 37
Thread B, id = 140388304152320, number = 38
Thread B, id = 140388304152320, number = 39
Thread A, id = 140388312545024, number = 38
Thread A, id = 140388312545024, number = 39
Thread B, id = 140388304152320, number = 40
Thread B, id = 140388304152320, number = 41
Thread A, id = 140388312545024, number = 40
Thread A, id = 140388312545024, number = 41
Thread B, id = 140388304152320, number = 42
Thread B, id = 140388304152320, number = 43
Thread A, id = 140388312545024, number = 42
Thread A, id = 140388312545024, number = 43
Thread B, id = 140388304152320, number = 44
Thread B, id = 140388304152320, number = 45
Thread A, id = 140388312545024, number = 44
Thread A, id = 140388312545024, number = 45
Thread B, id = 140388304152320, number = 46
Thread B, id = 140388304152320, number = 47
Thread A, id = 140388312545024, number = 46
Thread A, id = 140388312545024, number = 47
Thread B, id = 140388304152320, number = 48
Thread A, id = 140388312545024, number = 48
Thread B, id = 140388304152320, number = 49
Thread B, id = 140388304152320, number = 50
Thread A, id = 140388312545024, number = 49
Thread A, id = 140388312545024, number = 50

        通过对上面例子的测试，可以看出虽然每个线程内部循环了 50 次每次数一个数，但是最终没有数到 100，通过输出的结果可以看到，有些数字被重复数了多次，其原因就是没有对线程进行同步处理，造成了数据的混乱。

3.线程同步方式   

线程同步方式有以下几种：

1. 互斥锁（Mutex）：使用互斥锁来控制多个线程对共享资源的访问。只有获得锁的线程才能进入临界区进行操作，其他线程必须等待锁释放后才能进入。

2. 读写锁（Read-Write Lock）：允许多个线程同时读取共享资源，但只允许一个线程写入共享资源。读写锁可以提高程序的并发性能，减少资源争用问题。

3. 条件变量（Condition Variable）：用于等待特定条件的发生。当某个线程等待某个条件变量时，它会被阻塞，直到其他线程发出信号，通知条件变量已经满足。

4. 信号量（Semaphore）：使用信号量来控制多个线程对有限数量资源的访问。信号量表示资源的数量，每个线程在使用完资源后必须释放信号量，以便其他线程可以使用资源。

以上是常见的几种线程同步方式，每种方式都有其适用的场景和优缺点，根据具体的应用场景选择适合的同步方式是非常重要的，这里先介绍互斥锁。

4.互斥锁

        互斥锁是线程同步最常用的一种方式，通过互斥锁可以锁定一个代码块，被锁定的这个代码块，所有的线程只能顺序执行 (不能并行处理)，这样多线程访问共享资源数据混乱的问题就可以被解决了，需要付出的代价就是执行效率的降低，因为默认临界区多个线程是可以并行处理的，现在只能串行处理。

在 Linux 中互斥锁的类型为 pthread_mutex_t，创建一个这种类型的变量就得到了一把互斥锁：

pthread_mutex_t  mutex;

       在创建的锁对象中保存了当前这把锁的状态信息：锁定还是打开，如果是锁定状态还记录了给这把锁加锁的线程信息（线程 ID）。一个互斥锁变量只能被一个线程锁定，被锁定之后其他线程再对互斥锁变量加锁就会被阻塞，直到这把互斥锁被解锁，被阻塞的线程才能被解除阻塞。一般情况下，每一个共享资源对应一个把互斥锁，锁的个数和线程的个数无关。

Linux 提供的互斥锁操作函数如下，如果函数调用成功会返回 0，调用失败会返回相应的错误号：

// 初始化互斥锁
// restrict: 是一个关键字, 用来修饰指针, 只有这个关键字修饰的指针可以访问指向的内存地址, 其他指针是不行的
int pthread_mutex_init(pthread_mutex_t *restrict mutex,
           const pthread_mutexattr_t *restrict attr);
// 释放互斥锁资源            
int pthread_mutex_destroy(pthread_mutex_t *mutex);


//mutex: 互斥锁变量的地址
//attr: 互斥锁的属性，一般使用默认属性即可，这个参数指定为 NULL

申请一个互斥锁

修改互斥锁的状态, 将其设定为锁定状态, 这个状态被写入到参数 mutex 中

int pthread_mutex_lock(pthread_mutex_t *mutex);

这个函数被调用，首先会判断参数 mutex 互斥锁中的状态是不是锁定状态:

        没有被锁定，是打开的，这个线程可以加锁成功，这个这个锁中会记录是哪个线程加锁成功
如果被锁定了，其他线程加锁就失败了，这些线程都会阻塞在这把锁上
当这把锁被解开之后，这些阻塞在锁上的线程就解除阻塞了，并且这些线程是通过竞争的方式对这把锁加锁，没抢到锁的线程继续阻塞

尝试获取互斥锁

        下面这个函数是尝试获取一个互斥锁。如果当前的锁没有被其他线程占用，则该函数会立即返回，线程成功获取锁，并返回0。否则，该函数会立即返回，线程不会阻塞，直接返回EBUSY错误码。

// 尝试加锁
int pthread_mutex_trylock(pthread_mutex_t *mutex);

        使用pthread_mutex_trylock函数获取锁的好处是，在获取锁的过程中不会引起线程阻塞，避免了由于长时间阻塞而导致的程序性能下降或死锁问题。但是，需要注意的是，在使用pthread_mutex_trylock函数时，如果获取锁失败，则需要有相应的处理措施来处理获取锁失败的情况。

互斥锁解锁

        通过调用pthread_mutex_unlock()函数将占用的互斥锁释放，以便其他线程可以竞争和访问共享资源。如果未正确解锁互斥锁，则可能会导致死锁或其他线程无法访问共享资源。因此，正确使用互斥锁是确保多线程代码正确运行的关键之一。

// 对互斥锁解锁
int pthread_mutex_unlock(pthread_mutex_t *mutex);

        我们可以将上面多线程交替数数的例子修改一下，使用互斥锁进行线程同步。两个线程一共操作了同一个全局变量，因此需要添加一互斥锁，来控制这两个线程。

互斥锁示例：

#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <string.h>
#include <pthread.h>

#define MAX 100
// 全局变量
int number;

// 创建一把互斥锁
// 全局变量, 多个线程共享
pthread_mutex_t mutex;

// 线程处理函数
void* funcA_num(void* arg)
{
    for(int i=0; i<MAX; ++i)
    {
        // 如果线程A加锁成功, 不阻塞
        // 如果B加锁成功, 线程A阻塞
        pthread_mutex_lock(&mutex);
        int cur = number;
        cur++;
        usleep(10);
        number = cur;
        pthread_mutex_unlock(&mutex);
        printf("Thread A, id = %lu, number = %d\n", pthread_self(), number);
    }

    return NULL;
}

void* funcB_num(void* arg)
{
    for(int i=0; i<MAX; ++i)
    {
        // a加锁成功, b线程访问这把锁的时候是锁定的
        // 线程B先阻塞, a线程解锁之后阻塞解除
        // 线程B加锁成功了
        pthread_mutex_lock(&mutex);
        int cur = number;
        cur++;
        number = cur;
        pthread_mutex_unlock(&mutex);
        printf("Thread B, id = %lu, number = %d\n", pthread_self(), number);
        usleep(5);
    }

    return NULL;
}

int main(int argc, const char* argv[])
{
    pthread_t p1, p2;

    // 初始化互斥锁
    pthread_mutex_init(&mutex, NULL);

    // 创建两个子线程
    pthread_create(&p1, NULL, funcA_num, NULL);
    pthread_create(&p2, NULL, funcB_num, NULL);

    // 阻塞，资源回收
    pthread_join(p1, NULL);
    pthread_join(p2, NULL);

    // 销毁互斥锁
    // 线程销毁之后, 再去释放互斥锁
    pthread_mutex_destroy(&mutex);

    return 0;
}

linux系统运行结果：

[itliang@localhost ~]$ ./xctb
Thread A, id = 139912715360000, number = 1
Thread A, id = 139912715360000, number = 2
Thread A, id = 139912715360000, number = 3
Thread A, id = 139912715360000, number = 4
Thread A, id = 139912715360000, number = 5
Thread A, id = 139912715360000, number = 6
Thread A, id = 139912715360000, number = 7
Thread A, id = 139912715360000, number = 8
Thread A, id = 139912715360000, number = 9
Thread A, id = 139912715360000, number = 10
Thread A, id = 139912715360000, number = 11
Thread A, id = 139912715360000, number = 12
Thread A, id = 139912715360000, number = 13
Thread A, id = 139912715360000, number = 14
Thread A, id = 139912715360000, number = 15
Thread A, id = 139912715360000, number = 16
Thread A, id = 139912715360000, number = 17
Thread A, id = 139912715360000, number = 18
Thread A, id = 139912715360000, number = 19
Thread A, id = 139912715360000, number = 20
Thread A, id = 139912715360000, number = 21
Thread A, id = 139912715360000, number = 22
Thread A, id = 139912715360000, number = 23
Thread A, id = 139912715360000, number = 24
Thread A, id = 139912715360000, number = 25
Thread A, id = 139912715360000, number = 26
Thread A, id = 139912715360000, number = 27
Thread A, id = 139912715360000, number = 28
Thread A, id = 139912715360000, number = 29
Thread A, id = 139912715360000, number = 30
Thread A, id = 139912715360000, number = 31
Thread A, id = 139912715360000, number = 32
Thread A, id = 139912715360000, number = 33
Thread A, id = 139912715360000, number = 34
Thread A, id = 139912715360000, number = 35
Thread A, id = 139912715360000, number = 36
Thread A, id = 139912715360000, number = 37
Thread A, id = 139912715360000, number = 38
Thread A, id = 139912715360000, number = 39
Thread A, id = 139912715360000, number = 40
Thread A, id = 139912715360000, number = 41
Thread A, id = 139912715360000, number = 42
Thread A, id = 139912715360000, number = 43
Thread A, id = 139912715360000, number = 44
Thread A, id = 139912715360000, number = 45
Thread A, id = 139912715360000, number = 46
Thread A, id = 139912715360000, number = 47
Thread A, id = 139912715360000, number = 48
Thread A, id = 139912715360000, number = 49
Thread A, id = 139912715360000, number = 50
Thread B, id = 139912706967296, number = 51
Thread B, id = 139912706967296, number = 52
Thread B, id = 139912706967296, number = 53
Thread B, id = 139912706967296, number = 54
Thread B, id = 139912706967296, number = 55
Thread B, id = 139912706967296, number = 56
Thread B, id = 139912706967296, number = 57
Thread B, id = 139912706967296, number = 58
Thread B, id = 139912706967296, number = 59
Thread B, id = 139912706967296, number = 60
Thread B, id = 139912706967296, number = 61
Thread B, id = 139912706967296, number = 62
Thread B, id = 139912706967296, number = 63
Thread B, id = 139912706967296, number = 64
Thread B, id = 139912706967296, number = 65
Thread B, id = 139912706967296, number = 66
Thread B, id = 139912706967296, number = 67
Thread B, id = 139912706967296, number = 68
Thread B, id = 139912706967296, number = 69
Thread B, id = 139912706967296, number = 70
Thread B, id = 139912706967296, number = 71
Thread B, id = 139912706967296, number = 72
Thread B, id = 139912706967296, number = 73
Thread B, id = 139912706967296, number = 74
Thread B, id = 139912706967296, number = 75
Thread B, id = 139912706967296, number = 76
Thread B, id = 139912706967296, number = 77
Thread B, id = 139912706967296, number = 78
Thread B, id = 139912706967296, number = 79
Thread B, id = 139912706967296, number = 80
Thread B, id = 139912706967296, number = 81
Thread B, id = 139912706967296, number = 82
Thread B, id = 139912706967296, number = 83
Thread B, id = 139912706967296, number = 84
Thread B, id = 139912706967296, number = 85
Thread B, id = 139912706967296, number = 86
Thread B, id = 139912706967296, number = 87
Thread B, id = 139912706967296, number = 88
Thread B, id = 139912706967296, number = 89
Thread B, id = 139912706967296, number = 90
Thread B, id = 139912706967296, number = 91
Thread B, id = 139912706967296, number = 92
Thread B, id = 139912706967296, number = 93
Thread B, id = 139912706967296, number = 94
Thread B, id = 139912706967296, number = 95
Thread B, id = 139912706967296, number = 96
Thread B, id = 139912706967296, number = 97
Thread B, id = 139912706967296, number = 98
Thread B, id = 139912706967296, number = 99
Thread B, id = 139912706967296, number = 100