GCD中提供了两个函数，可以实现此功能，分别是dispatch_barrier_(a)sync和dispatch_group_(a)sync。
简单写一下两种函数的使用方式。

1.dispatch_barrier_(a)sync
dispatch_queue_t queue = dispatch_queue_create("concurrent", DISPATCH_QUEUE_CONCURRENT);
    dispatch_async(queue, ^{
        NSLog(@"111");
    })；
    dispatch_sync(queue, ^{
        NSLog(@“222”);
    });
    dispatch_barrier_async(queue, ^{
        NSLog(@"barrier finished");
    });
    NSLog(@"main tread");
    dispatch_async(queue, ^{
        NSLog(@“333”);
    });
    dispatch_async(queue, ^{
        NSLog(@“444”);
    });

2017-12-20 14:07:11.774484+0800 GCD[33826:2126185] 111
2017-12-20 14:07:11.774517+0800 GCD[33826:2126192] 222
2017-12-20 14:07:11.774540+0800 GCD[33826:2126069] main thread
2017-12-20 14:07:11.774715+0800 GCD[33826:2126192] barrier finished
2017-12-20 14:07:11.774836+0800 GCD[33826:2126185] 333
2017-12-20 14:07:11.774837+0800 GCD[33826:2126192] 444

dispatch_barrier_async换成dispatch_barrier_sync之后log如下

2017-12-20 14:07:51.093250+0800 GCD[33878:2127341] 222
2017-12-20 14:07:51.093250+0800 GCD[33878:2127340] 111
2017-12-20 14:07:51.093466+0800 GCD[33878:2127248] barrier finished
2017-12-20 14:07:51.093604+0800 GCD[33878:2127248] main thread
2017-12-20 14:07:51.093742+0800 GCD[33878:2127341] 333
2017-12-20 14:07:51.093786+0800 GCD[33878:2127338] 444

dispatch_barrier_sync的作用是阻塞当前线程，把queue中的所有任务执行完毕之后，再去执行barrier的block
dispatch_barrier_async不阻塞当前线程，调用之后，立即返回当前线程，继续执行后面的代码。
总之，dispatch_barrier_(a)sync的执行顺序就是先执行barrier block之前的所有任务，然后执行barrier block中的任务，最后执行barrier block之后提交的任务。

2.dispatch_group

dispatch_queue_t queue = dispatch_queue_create("concurrent", DISPATCH_QUEUE_CONCURRENT);
    dispatch_group_t group = dispatch_group_create();
    dispatch_group_async(group, queue, ^{
        NSLog(@"group 1");
    });
    dispatch_group_async(group, queue, ^{
        NSLog(@"group 2");
    });
    dispatch_group_async(group, queue, ^{
        NSLog(@"group 3");
    });
    
    dispatch_group_wait(group, DISPATCH_TIME_FOREVER);
    dispatch_group_notify(group, dispatch_get_main_queue(), ^{
        NSLog(@"group finished");
    });
    NSLog(@"666");

2017-12-20 14:18:53.984247+0800 GCD[34045:2139050] group 2
2017-12-20 14:18:53.984247+0800 GCD[34045:2139052] group 1
2017-12-20 14:18:53.984247+0800 GCD[34045:2139051] group 3
2017-12-20 14:18:53.984775+0800 GCD[34045:2138995] 666
2017-12-20 14:18:53.990947+0800 GCD[34045:2138995] group finished

与dispatch_barrier_(a)sync不同的是加入到group中的queue可以是不同的queue。

如果我们执行异步请求的时候如何保证在异步任务完成后，再去执行某段代码呢？这时候就要用到dispatch_group_enter和dispatch_group_leave了
dispatch_group_enter和dispatch_group_leave是成对出现的，出现数量必须保持一致。

dispatch_group_t group = dispatch_group_create();
    dispatch_group_async(group, queue, ^{
        dispatch_group_enter(group);
        dispatch_async(dispatch_get_global_queue(0, 0), ^{
            [NSThread sleepForTimeInterval:2];//发送一个网络请求
            NSLog(@"1111");
            dispatch_group_leave(group);
        });
    });
    dispatch_group_async(group, queue, ^{
        NSLog(@"group 2");
    });
    dispatch_group_async(group, queue, ^{
        NSLog(@"group 3");
    });
    dispatch_group_wait(group, DISPATCH_TIME_FOREVER);
    dispatch_group_notify(group, dispatch_get_main_queue(), ^{
        NSLog(@"group finished");
    });
    NSLog(@"666");

2017-12-20 14:24:17.069952+0800 GCD[34215:2145287] group 2
2017-12-20 14:24:17.069952+0800 GCD[34215:2145285] group 3
2017-12-20 14:24:19.070292+0800 GCD[34215:2145288] 1111
2017-12-20 14:24:19.070640+0800 GCD[34215:2145119] 666
2017-12-20 14:24:19.076618+0800 GCD[34215:2145119] group finished

这样就可以用来处理多个异步任务，任务完成后统一处理任务逻辑了。