ios线程解决方法 ios线程组

转载

mob64ca13ffd0f1 2023-08-04 07:22:32

文章标签 ios线程解决方法 iOS 多线程 GCD 串行队列 文章分类 iOS 移动开发

GCD中涉及的基本名词：

线程：程序执行任务的最小调度单位

任务：需要在主线程或者子线程中执行的代码，在GCD中显示为Block中

队列：一种特殊的线性表，采取FIFO(先进先出)，可以看做是用来存储任务的数组

异步：开辟新线程，同时执行多个任务

同步：在单一线程中执行，只能按顺序从前往后执行

并行队列：队列中的任务同时进行，可以开启多个线程，并发功能只有在异步情况有效

串行队列：队列中的任务逐个进行，只有单一线程

GCD的创建：

概述的讲，就是创建一个队列，然后将任务添加到队列中去

（1）创建队列

// ?????????串行队列
dispatch_queue_t queue = dispatch_queue_create("com.app", DISPATCH_QUEUE_SERIAL);
// ?????????并发队列
dispatch_queue_t queue = dispatch_queue_create("com.app", DISPATCH_QUEUE_CONCURRENT);

第一个参数是队列的标志符，可以为空，第二个参数表示是串行队列还是并发队列

GCD中有简便的方法获取队列，一个是获取主队列（串行队列），还有一个是获取全局并发队列（并发队列）

//主队列
dispatch_queue_t queue = dispatch_get_main_queue();
//全局并发队列，第一个参数表示优先级，第二个参数是保留参数，暂时无用
dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DE FAULT, 0);

（2）创建任务

dispatch_sync(queue, ^{ 
    // 同步
});
???????dispatch_async(queue, ^{
    // 异步
});

GCD的使用：

- (void)gcdTest {
    NSLog(@"currentThread---%@",[NSThread currentThread]);
    NSLog(@"---begin");
    dispatch_queue_t queue = dispatch_queue_create(nil, DISPATCH_QUEUE_CONCURRENT);
    dispatch_sync(queue, ^{
        //任务一
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"1---%@",[NSThread currentThread]);
        } });
    dispatch_sync(queue, ^{
        //任务二
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"2---%@",[NSThread currentThread]);
        } });
    dispatch_sync(queue, ^{
        //任务三
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"3---%@",[NSThread currentThread]);
        } });
    NSLog(@"---end");
}

（1）同步+并发，同步+串行

2019-02-15 20:30:13.134436+0800 GCD[11324:226430] currentThread---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:13.134613+0800 GCD[11324:226430]---begin
2019-02-15 20:30:15.135160+0800 GCD[11324:226430] 1---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:17.136650+0800 GCD[11324:226430] 1---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:19.138189+0800 GCD[11324:226430] 2---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:21.139706+0800 GCD[11324:226430] 2---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:23.141219+0800 GCD[11324:226430] 3---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:25.142714+0800 GCD[11324:226430] 3---<NSThread: 0x6000019dafc0>{number = 1, name = main}
2019-02-15 20:30:25.142980+0800 GCD[11324:226430]---end

这两种结果都一样，虽然并发队列有执行多线程的能力，但是由于是同步执行，只会在单一线程中执行，所以会在同一线程中依次执行完任务后再添加下一个任务

（3）异步+串行

2019-02-15 20:42:52.282373+0800 GCD[11506:241315] currentThread---<NSThread: 0x600003b93640>{number = 1, name = main}
2019-02-15 20:42:52.282539+0800 GCD[11506:241315] ---begin
2019-02-15 20:42:52.282674+0800 GCD[11506:241315] ---end
2019-02-15 20:42:54.286775+0800 GCD[11506:241365] 1---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}
2019-02-15 20:42:56.287246+0800 GCD[11506:241365] 1---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}
2019-02-15 20:42:58.292730+0800 GCD[11506:241365] 2---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}
2019-02-15 20:43:00.293612+0800 GCD[11506:241365] 2---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}
2019-02-15 20:43:02.299173+0800 GCD[11506:241365] 3---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}
2019-02-15 20:43:04.300262+0800 GCD[11506:241365] 3---<NSThread: 0x600003bf1f00>{number = 3, name = (null)}

当遇到异步执行时，会创建一个新的线程，主线程继续向下执行，新线程执行添加的任务，新线程是串行队列，也是按顺执行

（4）异步+并行

2019-02-15 20:47:46.005210+0800 GCD[11582:247831] currentThread---<NSThread: 0x600000071400>{number = 1, name = main}
2019-02-15 20:47:46.005361+0800 GCD[11582:247831] ---begin
2019-02-15 20:47:46.005466+0800 GCD[11582:247831] ---end
2019-02-15 20:47:48.005804+0800 GCD[11582:247876] 2---<NSThread: 0x600000012940>{number = 4, name = (null)}
2019-02-15 20:47:48.005805+0800 GCD[11582:247877] 1---<NSThread: 0x60000001c0c0>{number = 3, name = (null)}
2019-02-15 20:47:48.005804+0800 GCD[11582:247879] 3---<NSThread: 0x60000001c100>{number = 5, name = (null)}
2019-02-15 20:47:50.008365+0800 GCD[11582:247876] 2---<NSThread: 0x600000012940>{number = 4, name = (null)}
2019-02-15 20:47:50.008365+0800 GCD[11582:247879] 3---<NSThread: 0x60000001c100>{number = 5, name = (null)}
2019-02-15 20:47:50.008387+0800 GCD[11582:247877] 1---<NSThread: 0x60000001c0c0>{number = 3, name = (null)}

这种情况就是主线程在执行，同时其他开辟多个线程同时执行

（5）同步+主队列

这种情况下，如果调用的函数在其他线程执行，和同步+串行情况相同，但是在主线程下执行，打印完begin之后会死锁，程序崩溃。因为函数在主队列，他会执行完函数再调用添加进来的任务，但是任务不执行完就没法走完函数，有点像循环引用，A引用B，B引用A，双方都等对方执行完再执行，这种情况要避免。

GCD的其他方法

（1）栅栏方法：?????dispatch_barrier_async

NSLog(@"currentThread---%@",[NSThread currentThread]);
    NSLog(@"---begin");
    dispatch_queue_t queue = dispatch_queue_create(nil, DISPATCH_QUEUE_CONCURRENT);
    dispatch_async(queue, ^{
        //任务一
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"1---%@",[NSThread currentThread]);
        } });
    dispatch_async(queue, ^{
        //任务二
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"2---%@",[NSThread currentThread]); 
        } });
    dispatch_barrier_async(queue, ^{
        NSLog(@"barrier---%@",[NSThread currentThread]);
    });
    
    dispatch_async(queue, ^{
        //任务三
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"3---%@",[NSThread currentThread]);
        } });
    NSLog(@"---end");

结果

2019-02-16 00:02:50.327233+0800 GCD[13192:337727] currentThread---<NSThread: 0x6000010e2900>{number = 1, name = main}
2019-02-16 00:02:50.327496+0800 GCD[13192:337727] ---begin
2019-02-16 00:02:50.327658+0800 GCD[13192:337727] ---end
2019-02-16 00:02:52.333127+0800 GCD[13192:337760] 1---<NSThread: 0x600001087480>{number = 3, name = (null)}
2019-02-16 00:02:52.333147+0800 GCD[13192:337762] 2---<NSThread: 0x600001087500>{number = 4, name = (null)}
2019-02-16 00:02:54.335308+0800 GCD[13192:337760] 1---<NSThread: 0x600001087480>{number = 3, name = (null)}
2019-02-16 00:02:54.335321+0800 GCD[13192:337762] 2---<NSThread: 0x600001087500>{number = 4, name = (null)}
2019-02-16 00:02:54.335729+0800 GCD[13192:337762] barrier---<NSThread: 0x600001087500>{number = 4, name = (null)}
2019-02-16 00:02:56.336141+0800 GCD[13192:337762] 3---<NSThread: 0x600001087500>{number = 4, name = (null)}
2019-02-16 00:02:58.340345+0800 GCD[13192:337762] 3---<NSThread: 0x600001087500>{number = 4, name = (null)}

可以看出，栅栏函数能隔开加载到队列里的任务，未加载到队列里的任务无效。在栅栏函数前面的任务全部执行完后，执行栅栏函数，然后执行栅栏函数后的任务。

（2）延时执行方法：dispatch_after

dispatch_after(dispatch_time(DISPATCH_TIME_NOW, (int64_t)(3 * NSEC_PER_SEC)), queue, ^{
        NSLog(@"after---%@",[NSThread currentThread]);
    });

顾名思义，就是延迟执行，上面代码是延迟3s，注意，这个延迟是3s后把任务加入队列，不是3s后就执行，如果你加入的是串行队列，3s后加入队列，但是前面的任务需要10s才执行完，那你得到结果的时间是13s。

（3）只执行一次：dispatch_once

static dispatch_once_t onceToken;
      dispatch_once(&onceToken, ^{
       //任务  
});

该函数能保证代码在程序运行过程中只执行一次，并且即使在多线程的环境下，也能保证里面的代码线程安全。

（4）快速迭代方法：dispatch_apply

NSLog(@"---begin");
    dispatch_queue_t queue = dispatch_queue_create(nil, DISPATCH_QUEUE_CONCURRENT);
    dispatch_apply(10, queue, ^(size_t index) {
        NSLog(@"%zd---%@",index, [NSThread currentThread]);
    });
    NSLog(@"---end");

2019-02-16 00:32:36.142181+0800 GCD[13675:371444] ---begin
2019-02-16 00:32:36.142501+0800 GCD[13675:371494] 2---<NSThread: 0x6000019e0b00>{number = 4, name = (null)}
2019-02-16 00:32:36.142526+0800 GCD[13675:371495] 1---<NSThread: 0x6000019dfec0>{number = 3, name = (null)}
2019-02-16 00:32:36.142529+0800 GCD[13675:371444] 0---<NSThread: 0x600001985400>{number = 1, name = main}
2019-02-16 00:32:36.142537+0800 GCD[13675:371493] 3---<NSThread: 0x6000019e0a80>{number = 5, name = (null)}
2019-02-16 00:32:36.142644+0800 GCD[13675:371494] 4---<NSThread: 0x6000019e0b00>{number = 4, name = (null)}
2019-02-16 00:32:36.142678+0800 GCD[13675:371495] 5---<NSThread: 0x6000019dfec0>{number = 3, name = (null)}
2019-02-16 00:32:36.142693+0800 GCD[13675:371444] 6---<NSThread: 0x600001985400>{number = 1, name = main}
2019-02-16 00:32:36.142735+0800 GCD[13675:371493] 7---<NSThread: 0x6000019e0a80>{number = 5, name = (null)}
2019-02-16 00:32:36.142786+0800 GCD[13675:371494] 8---<NSThread: 0x6000019e0b00>{number = 4, name = (null)}
2019-02-16 00:32:36.142807+0800 GCD[13675:371495] 9---<NSThread: 0x6000019dfec0>{number = 3, name = (null)}
2019-02-16 00:32:36.143608+0800 GCD[13675:371444] ---end

dispatch_apply有点像for循环，如果是串行队列，和for差不多，依次N次任务加入到队列中，顺序执行，如果是并行队列，则是并发执行任务，从结果中看出，这个函数会调用主线程，不知道会不会在某些情况下造成阻塞。这个函数本身是同步的，所以会执行完全部任务，再执行下面的任务。

（5）dispatch_group

1、group基础

NSLog(@"currentThread---%@",[NSThread currentThread]);
    NSLog(@"group---begin");
    dispatch_group_t group =  dispatch_group_create();
    dispatch_group_async(group, dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"1---%@",[NSThread currentThread]);
        };
    });
    
    dispatch_group_async(group, dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"2---%@",[NSThread currentThread]);
        };
    });
    dispatch_group_notify(group, dispatch_get_main_queue(), ^{
        for (int i = 0; i < 2; ++i) {
            [NSThread sleepForTimeInterval:2];
            NSLog(@"3---%@",[NSThread currentThread]);
        };
        NSLog(@"group---end");
    });

2019-02-16 01:16:04.203743+0800 GCD[14254:412235] currentThread---<NSThread: 0x600001b9a900>{number = 1, name = main}
2019-02-16 01:16:04.203910+0800 GCD[14254:412235] group---begin
2019-02-16 01:16:06.205824+0800 GCD[14254:412362] 1---<NSThread: 0x600001bf5000>{number = 3, name = (null)}
2019-02-16 01:16:06.205845+0800 GCD[14254:412365] 2---<NSThread: 0x600001bf5040>{number = 4, name = (null)}
2019-02-16 01:16:08.211355+0800 GCD[14254:412362] 1---<NSThread: 0x600001bf5000>{number = 3, name = (null)}
2019-02-16 01:16:08.211356+0800 GCD[14254:412365] 2---<NSThread: 0x600001bf5040>{number = 4, name = (null)}
2019-02-16 01:16:10.212927+0800 GCD[14254:412235] 3---<NSThread: 0x600001b9a900>{number = 1, name = main}
2019-02-16 01:16:12.214396+0800 GCD[14254:412235] 3---<NSThread: 0x600001b9a900>{number = 1, name = main}
2019-02-16 01:16:12.214740+0800 GCD[14254:412235] group---end

任务先加入到队列，然后队列加入到队列组，当队列组中的任务执行完成后，会执行dispatch_group_notify函数任务

2、wait

dispatch_group_wait(group, DISPATCH_TIME_FOREVER);

阻塞当前的线程，等到指定group中的任务执行完毕后再继续执行。

3、group_leave,group_enter

dispatch_group_enter ?????????标志着一个任务追加到group?,相当于group中未执行完毕的任务数+1

dispatch_group_leave ?????????标志着一个任务离开了group?,相当于group中未执行完毕的任务数-1

当group中未完成数为0的时候，才会执行notify和wait函数，这里有点像是引用计数。

（6）信号量dispatch_semaphore

dispatch_semaphore_create?：创建一个信号量

dispatch_semaphore_signal?：发送一个信号量，信号量+1

dispatch_semaphore_wait?：当线程信号量为0时阻塞线程，直到有新的信号量添加，当通过这个方法时，信号量的值-1

应用：保证线程安全，为线程加锁，或者保持线程同步，将异步执行任务转换成同步

dispatch_semaphore_t semaphore = dispatch_semaphore_create(0);
dispatch_time_t time = dispatch_time(DISPATCH_TIME_NOW, 2*NSEC_PER_SEC);
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0),                 ^{
    [NSThread sleepForTimeInterval:2];
    NSLog(@"2---%@",[NSThread currentThread]);
    dispatch_semaphore_signal(semaphore);
});
NSLog(@"1---%@",[NSThread currentThread]);
dispatch_semaphore_wait(semaphore,time);//DISPATCH_TIME_FOREVER没有废弃时间
NSLog(@"3---%@",[NSThread currentThread]);

2019-02-16 15:35:58.372359+0800 GCD[17462:642975] 1---<NSThread: 0x600002a193c0>{number = 1, name = main}
2019-02-16 15:36:00.375787+0800 GCD[17462:643022] 2---<NSThread: 0x600002a718c0>{number = 3, name = (null)}
2019-02-16 15:36:00.376150+0800 GCD[17462:642975] 3---<NSThread: 0x600002a193c0>{number = 1, name = main}

创建一个信号量，由于线程任务是异步执行，所以开辟了一个新线程去执行，主线程继续执行下面的打印任务，当遇到dispatch_semaphore_wait时，信号量为0，阻塞主线程。新线程由于有耗时操作，2s后执行完任务，打印并且使信号量+1，这时主线程中信号量为1，所以通过了wait方法，继续执行下面的打印。这里说下dispatch_semaphore_wait的第二个参数，是指等待时间，如果超出这个时间，也会继续向下运行，但是这个信号量就已经废弃了，下面再遇到dispatch_semaphore_wait方法也不会起作用。

本文章为转载内容，我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题，欢迎原作者联系我们进行内容更正或删除文章。