CPU 乱序执行验证

受到一个问题启发：linux kernel 为什么加 smp_mb，怎么引起的，添加又有怎么解决的问题？

最近遇到CPU的乱序问题，总是理论性的东西和讨论，找了个时间写了测试程序，验证CPU乱序执行的存在，理论知识网上已经很多，不再啰嗦，理论加实践，才能理解的更加深刻

/*
        this program test CPU out of order execute.

        ENV: CPU need more than two core thread
        gcc -Wall -O3 out_of_order.c -o out_of_order -lpthread
*/

#include <stdio.h>
#include <pthread.h>

static long a = 0, b = 0;
static long x = 0, y = 0;

#define mb()    __asm__ __volatile__("mfence":::"memory")
#define rmb()   __asm__ __volatile__("lfence":::"memory")
#define wmb()   __asm__ __volatile__("sfence":::"memory")
#define barrier() __asm__ __volatile__("": : :"memory")

void* func1(void *arg)
{
        a = 1;
        barrier();
        // mb();
        x = b;

        return NULL;
}

void* func2(void *arg)
{
        b = 1;
        barrier();
        // mb();
        y = a;

        return NULL;
}

int main()
{
        int i = 0;
        pthread_t  pit1, pit2;

        while (1) {
                i++;
                a = 0; b = 0;
                x = 0; y = 0;
                pthread_create(&pit1, NULL, &func1, NULL);
                pthread_create(&pit2, NULL, &func2, NULL);
                // wait for pthread stop
                pthread_join(pit1, NULL);
                pthread_join(pit2, NULL);
                
                // we don't think, but ...
                if (x == 0 && y == 0)
                        printf("i = %d\n", i);
        }
}

执行结果如下，会时不时出现 x==0 && y == 0的情况 

i = 20
i = 24
i = 26
i = 28
i = 31
i = 34
i = 38
i = 42
i = 43
i = 45
i = 47
i = 49
i = 51
i = 53
i = 57
i = 59
i = 63
i = 71
.....