为高效的访问，读取处理追加到Concurrent Dispatch Queue，写入处理在任何一个读取操作都没有执行的状态下，追加到Serial Dispatch Queue。类似这样的需求，我们可以使用dispatch_barrier_async。
dispatch_queue_t queue = dispatch_queue_create("com.example.gdc.ForBarrier", DISPATCH_QUEUE_CONCURRENT);
dispatch_async(queue, ^{
NSLog(@"reading1");
});
dispatch_async(queue, ^{
NSLog(@"reading2");
});
dispatch_async(queue, ^{
NSLog(@"reading3");
});
dispatch_barrier_async(queue, ^{
NSLog(@"writing");
});
dispatch_async(queue, ^{
NSLog(@"reading4");
});
dispatch_async(queue, ^{
NSLog(@"reading5");
});
输出：
2018-07-23 22:38:16.498997+0800 Demo[9100:359307] reading3
2018-07-23 22:38:16.498997+0800 Demo[9100:359308] reading1
2018-07-23 22:38:16.498997+0800 Demo[9100:359311] reading2
2018-07-23 22:38:16.500972+0800 Demo[9100:359308] writing
2018-07-23 22:38:16.501812+0800 Demo[9100:359308] reading4
2018-07-23 22:38:16.501831+0800 Demo[9100:359311] reading5
说明：
dispatch_barrier_async函数会等待该函数之前追加到Concurrent Dispatch Queue上的并行执行的处理全部结束后，再将指定的处理追加到该Concurrent Dispatch Queue中，dispatch_barrier_async函数追加的处理执行完毕后，该函数之后追加到Concurrent Dispatch Queue的处理才又开始并行处理。
