多线程读写安全

多读单写

要想实现多线程的读写安全，必须满足“读读”并发，“读写”、“写写”互斥。即：

同一时间，只能有1个线程进行写的操作
同一时间，允许有多个线程进行读的操作
同一时间，不允许既有写的操作，又有读的操作

iOS中的实现方案：

（1）pthread_rwlock：读写锁
（2）dispatch_barrier_async：异步栅栏调用

pthread_rwlock

等待锁的线程会进入休眠

//初始化
int pthread_rwlock_init(pthread_rwlock_t * __restrict, 
const pthread_rwlockattr_t * _Nullable __restrict);
//销毁
int pthread_rwlock_destroy(pthread_rwlock_t * );
//读-加锁
int pthread_rwlock_rdlock(pthread_rwlock_t *);
//读-尝试加锁
int pthread_rwlock_tryrdlock(pthread_rwlock_t *);
//写-加锁
int pthread_rwlock_trywrlock(pthread_rwlock_t *);
//写-尝试加锁
int pthread_rwlock_wrlock(pthread_rwlock_t *);
//解锁
int pthread_rwlock_unlock(pthread_rwlock_t *);

#include 

- (void)test {
    // 初始化锁
   pthread_rwlock_init(&_lock, NULL);

   dispatch_queue_t queue = dispatch_get_global_queue(0, 0);
    for(int i = 1; i <= 5; i ++) {
        dispatch_async(queue, ^{
            [self write:i];
        });
        dispatch_async(queue, ^{
            [self read:i];
        });
    }
}

- (void)write:(NSInteger)i {
    pthread_rwlock_wrlock(&rwLock); //写锁
    sleep(5);
    NSLog(@"写入：%d", i);
    pthread_rwlock_unlock(&rwLock);//解锁
}

- (void)read:(NSInteger)i {
    pthread_rwlock_rdlock(&rwLock); //读锁
    sleep(5);
    NSLog(@"读取：%d", i);
    pthread_rwlock_unlock(&rwLock);//解锁
}

- (void)dealloc {
    //销毁
    pthread_rwlock_destroy(&rwLock);
}

“读读”并发，“读写”、“写写”互斥

dispatch_barrier_async 异步栅栏

注意：
（1）这个函数传入的并发队列必须是通过dispatch_queue_cretate自定义的。如果传入的是一个串行队列或是一个全局的并发队列，那这个函数便等同于dispatch_async函数的效果。
（2）读取的时候使用dispatch_sync，之所以取值时用同步操作，是因为我们明确返回一个值后才去执行后面的方法。

@interface ViewController ()
@property (nonatomic, strong) dispatch_queue_t queue;
@end

@implementation ViewController

- (void)viewDidLoad {
    [super viewDidLoad];
    self.queue = dispatch_queue_create("rw_queue", DISPATCH_QUEUE_CONCURRENT);
       
    for (int i = 0; i < 10; i++) {
        dispatch_async(self.queue, ^{
            [self read:@"a"];
        });
        
        dispatch_async(self.queue, ^{
            [self read:@"b"];
        });
        
        dispatch_async(self.queue, ^{
            [self read:@"c"];
        });
        
        // 这个函数传入的并发队列必须通过dispatch_queue_cretate自定义
        dispatch_barrier_async(self.queue, ^{
            [self write];
        });
    }
}

- (void)read:(NSString *)str {
    sleep(1);
    NSLog(@"read:%@",str);
}

- (void)write {
    sleep(1);
    NSLog(@"write");
}

@end