RunLoop之事件循环机制、线程

事件循环机制

分析UIApplicationMain如何启动主线程的RunLoop

打断点，通过lldb指令bt查看函数调用栈如下：

在UIApplicationMain函数中调用了Core Foundation框架下的CFRunLoopRunSpecific函数

CFRunLoopRunSpecific函数实现：RunLoop的入口

源码如下:

SInt32 CFRunLoopRunSpecific(CFRunLoopRef rl, CFStringRef modeName, CFTimeInterval seconds, Boolean returnAfterSourceHandled) {     /* DOES CALLOUT */
    CHECK_FOR_FORK();
    if (__CFRunLoopIsDeallocating(rl)) return kCFRunLoopRunFinished;
    __CFRunLoopLock(rl);
    // 根据 modeName 找到本次运行的 mode
    CFRunLoopModeRef currentMode = __CFRunLoopFindMode(rl, modeName, false);
    // 如果没找到 || mode 中没有注册任何事件，则就此停止，不进入循环
    if (NULL == currentMode || __CFRunLoopModeIsEmpty(rl, currentMode, rl->_currentMode)) {
	Boolean did = false;
	if (currentMode) __CFRunLoopModeUnlock(currentMode);
	__CFRunLoopUnlock(rl);
	return did ? kCFRunLoopRunHandledSource : kCFRunLoopRunFinished;
    }
    volatile _per_run_data *previousPerRun = __CFRunLoopPushPerRunData(rl);
    CFRunLoopModeRef previousMode = rl->_currentMode;
    rl->_currentMode = currentMode;
    int32_t result = kCFRunLoopRunFinished;

    // 通知 Observers：即将进入 RunLoop
	if (currentMode->_observerMask & kCFRunLoopEntry ) __CFRunLoopDoObservers(rl, currentMode, kCFRunLoopEntry);
    // RunLoop 具体要做的事情
	result = __CFRunLoopRun(rl, currentMode, seconds, returnAfterSourceHandled, previousMode);
    // 通知 Observers：即将退出 RunLoop
	if (currentMode->_observerMask & kCFRunLoopExit ) __CFRunLoopDoObservers(rl, currentMode, kCFRunLoopExit);

        __CFRunLoopModeUnlock(currentMode);
        __CFRunLoopPopPerRunData(rl, previousPerRun);
	rl->_currentMode = previousMode;
    __CFRunLoopUnlock(rl);
    return result;
}

删掉不重要的细节：

SInt32 CFRunLoopRunSpecific(CFRunLoopRef rl, CFStringRef modeName, CFTimeInterval seconds, Boolean returnAfterSourceHandled) {     /* DOES CALLOUT */

    // 通知 Observers：即将进入 RunLoop
    __CFRunLoopDoObservers(rl, currentMode, kCFRunLoopEntry);
    // RunLoop 具体要做的事情
	result = __CFRunLoopRun(rl, currentMode, seconds, returnAfterSourceHandled, previousMode);
    // 通知 Observers：即将退出 RunLoop
    __CFRunLoopDoObservers(rl, currentMode, kCFRunLoopExit);

    return result;
}

可知，RunLoop事件循环的实现机制体现在__CFRunLoopRun函数中。

__CFRunLoopRun函数实现：事件循环的实现机制

精简后的代码如下:

/**
 *  __CFRunLoopRun
 *
 *  @param rl              运行的 RunLoop 对象
 *  @param rlm             运行的 mode
 *  @param seconds         loop 超时时间
 *  @param stopAfterHandle true: RunLoop 处理完事件就退出  false:一直运行直到超时或者被手动终止
 *  @param previousMode    上一次运行的 mode
 *
 *  @return 返回 4 种状态
 */
static int32_t __CFRunLoopRun(CFRunLoopRef rl, CFRunLoopModeRef rlm, CFTimeInterval seconds, Boolean stopAfterHandle, CFRunLoopModeRef previousMode) 
{
    int32_t retVal = 0;
    do {
        // 通知 Observers：即将处理 Timers
        __CFRunLoopDoObservers(rl, rlm, kCFRunLoopBeforeTimers);
        // 通知 Observers：即将处理 Sources
        __CFRunLoopDoObservers(rl, rlm, kCFRunLoopBeforeSources);
        // 处理 Blocks
    	__CFRunLoopDoBlocks(rl, rlm);
        // 处理 Sources0
        if (__CFRunLoopDoSources0(rl, rlm, stopAfterHandle)) {
            // 处理 Blocks
            __CFRunLoopDoBlocks(rl, rlm);
	    }
        // 判断有无 Source1
        if (__CFRunLoopServiceMachPort(dispatchPort, &msg, sizeof(msg_buffer), &livePort, 0, &voucherState, NULL)) {
            // 如果有 Source1，就跳转到 handle_msg
            goto handle_msg;
        }
        // 通知 Observers：即将进入休眠
	    __CFRunLoopDoObservers(rl, rlm, kCFRunLoopBeforeWaiting);
    	__CFRunLoopSetSleeping(rl);
        // ⚠️休眠，等待消息来唤醒线程
        __CFRunLoopServiceMachPort(waitSet, &msg, sizeof(msg_buffer), &livePort, poll ? 0 : TIMEOUT_INFINITY, &voucherState, &voucherCopy);
    	__CFRunLoopUnsetSleeping(rl);
        // 通知 Observers：刚从休眠中唤醒
	    __CFRunLoopDoObservers(rl, rlm, kCFRunLoopAfterWaiting);

handle_msg:
        if (被 Timer 唤醒) {
            // 处理 Timer
            __CFRunLoopDoTimers(rl, rlm, mach_absolute_time())
        } else if (被 GCD 唤醒) {
            // 处理 GCD 
            __CFRUNLOOP_IS_SERVICING_THE_MAIN_DISPATCH_QUEUE__(msg);
        } else {  // 被 Source1 唤醒   
            // 处理 Source1
            __CFRunLoopDoSource1(rl, rlm, rls, msg, msg->msgh_size, &reply) || sourceHandledThisLoop;  
        }

        // 处理 Blocks
        __CFRunLoopDoBlocks(rl, rlm);
        
        /* 设置返回值 */
        // 进入 loop 时参数为处理完事件就返回
        if (sourceHandledThisLoop && stopAfterHandle) {  
            retVal = kCFRunLoopRunHandledSource;
        // 超出传入参数标记的超时时间
        } else if (timeout_context->termTSR < mach_absolute_time()) {  
            retVal = kCFRunLoopRunTimedOut;
        // 被外部调用者强制停止
        } else if (__CFRunLoopIsStopped(rl)) {  
            __CFRunLoopUnsetStopped(rl);
            retVal = kCFRunLoopRunStopped;
        // 自动停止
        } else if (rlm->_stopped) {  
            rlm->_stopped = false;
            retVal = kCFRunLoopRunStopped;
        // mode 中没有任何的 Source0/Source1/Timer/Observer
        } else if (__CFRunLoopModeIsEmpty(rl, rlm, previousMode)) {  
            retVal = kCFRunLoopRunFinished;
        }
    
    } while (0 == retVal);

    return retVal;
}

从该函数实现中可以得知RunLoop主要就做以下几件事情：

• __CFRunLoopDoObservers：通知Observers接下来要做什么
• __CFRunLoopDoBlocks：处理Blocks
• __CFRunLoopDoSources0：处理Sources0
• __CFRunLoopDoSources1：处理Sources1
• __CFRunLoopDoTimers：处理Timers
• 处理 GCD 相关：dispatch_async(dispatch_get_main_queue(), ^{ });
• __CFRunLoopSetSleeping/__CFRunLoopUnsetSleeping：休眠等待/结束休眠
• __CFRunLoopServiceMachPort -> mach-msg()：转移当前线程的控制权

__CFRunLoopServiceMachPort函数实现：RunLoop休眠的实现原理

在__CFRunLoopRun函数中，会调用__CFRunLoopServiceMachPort函数，该函数中调用了mach_msg()函数来转移当前线程的控制权给内核态/用户态。

• 没有消息需要处理时，休眠线程以避免资源占用。调用mach_msg()从用户态切换到内核态，等待消息；
• 有消息需要处理时，立刻唤醒线程，调用mach_msg()回到用户态处理消息。

这就是RunLoop休眠的实现原理，也是RunLoop与简单的do...while循环区别：

• RunLoop：休眠的时候，当前线程不会做任何事，CPU 不会再分配资源；
• 简单的do…while循环：当前线程并没有休息，一直占用 CPU 资源。

static Boolean __CFRunLoopServiceMachPort(mach_port_name_t port, mach_msg_header_t **buffer, size_t buffer_size, mach_port_t *livePort, mach_msg_timeout_t timeout, voucher_mach_msg_state_t *voucherState, voucher_t *voucherCopy) {
    Boolean originalBuffer = true;
    kern_return_t ret = KERN_SUCCESS;
    for (;;) {		/* In that sleep of death what nightmares may come ... */
        mach_msg_header_t *msg = (mach_msg_header_t *)*buffer;
        msg->msgh_bits = 0;
        msg->msgh_local_port = port;
        msg->msgh_remote_port = MACH_PORT_NULL;
        msg->msgh_size = buffer_size;
        msg->msgh_id = 0;
        if (TIMEOUT_INFINITY == timeout) { CFRUNLOOP_SLEEP(); } else { CFRUNLOOP_POLL(); }

        // ⚠️⚠️⚠️
        ret = mach_msg(msg, MACH_RCV_MSG|(voucherState ? MACH_RCV_VOUCHER : 0)|MACH_RCV_LARGE|((TIMEOUT_INFINITY != timeout) ? MACH_RCV_TIMEOUT : 0)|MACH_RCV_TRAILER_TYPE(MACH_MSG_TRAILER_FORMAT_0)|MACH_RCV_TRAILER_ELEMENTS(MACH_RCV_TRAILER_AV), 0, msg->msgh_size, port, timeout, MACH_PORT_NULL);

        // Take care of all voucher-related work right after mach_msg.
        // If we don't release the previous voucher we're going to leak it.
        voucher_mach_msg_revert(*voucherState);
        
        // Someone will be responsible for calling voucher_mach_msg_revert. This call makes the received voucher the current one.
        *voucherState = voucher_mach_msg_adopt(msg);
        
        if (voucherCopy) {
            if (*voucherState != VOUCHER_MACH_MSG_STATE_UNCHANGED) {
                // Caller requested a copy of the voucher at this point. By doing this right next to mach_msg we make sure that no voucher has been set in between the return of mach_msg and the use of the voucher copy.
                // CFMachPortBoost uses the voucher to drop importance explicitly. However, we want to make sure we only drop importance for a new voucher (not unchanged), so we only set the TSD when the voucher is not state_unchanged.
                *voucherCopy = voucher_copy();
            } else {
                *voucherCopy = NULL;
            }
        }

        CFRUNLOOP_WAKEUP(ret);
        if (MACH_MSG_SUCCESS == ret) {
            *livePort = msg ? msg->msgh_local_port : MACH_PORT_NULL;
            return true;
        }
        if (MACH_RCV_TIMED_OUT == ret) {
            if (!originalBuffer) free(msg);
            *buffer = NULL;
            *livePort = MACH_PORT_NULL;
            return false;
        }
        if (MACH_RCV_TOO_LARGE != ret) break;
        buffer_size = round_msg(msg->msgh_size + MAX_TRAILER_SIZE);
        if (originalBuffer) *buffer = NULL;
        originalBuffer = false;
        *buffer = realloc(*buffer, buffer_size);
    }
    HALT;
    return false;
}

RunLoop与线程

• RunLoop对象和线程是一一对应关系；
• RunLoop保存在一个全局的Dictionary里，线程作为key，RunLoop作为value
• 如果没有RunLoop，线程执行完任务就会退出；如果没有RunLoop，主线程执行完main()函数就会退出，程序就不能处于运行状态；
• RunLoop创建时机：线程刚创建时并没有RunLoop对象，RunLoop会在第一次获取它时创建
• RunLoop销毁时机：RunLoop会在线程结束时销毁；
• 主线程的RunLoop已经自动获取（创建），子线程默认没有开启RunLoop
• 主线程的RunLoop对象是在UIApplicationMain中通过[NSRunLoop currentRunLoop]获取，一旦发现它不存在，就会创建RunLoop对象。

未启动RunLoop的子线程

创建一个NSThread的子类HTThread并重写了dealloc方法来观察线程的状态。执行以下代码，发现线程执行完一次test任务就退出销毁了，没有再执行test任务，原因就是没有启动该线程的RunLoop

- (void)viewDidLoad {
    [super viewDidLoad];

    HTThread *thread = [[HTThread alloc] initWithTarget:self selector:@selector(test) object:nil];
    [thread start];
    [self performSelector:@selector(test) onThread:thread withObject:nil waitUntilDone:NO];
}

- (void)test {
    NSLog(@"test on %@", [NSThread currentThread]);
}
// test on <HTThread: 0x600003cb52c0>{number = 7, name = (null)}
// HTThread dealloc

开启子线程的RunLoop的过程

获取RunLoop对象

// Foundation
[NSRunLoop mainRunLoop];     // 获取主线程的 RunLoop 对象
[NSRunLoop currentRunLoop];  // 获取当前线程的 RunLoop 对象
// Core Foundation
CFRunLoopGetMain();     // 获取主线程的 RunLoop 对象
CFRunLoopGetCurrent();  // 获取当前线程的 RunLoop 对象

来看一下CFRunLoopGetCurrent()函数是怎么获取RunLoop对象的：

CFRunLoopRef CFRunLoopGetCurrent(void) {
    CHECK_FOR_FORK();
    CFRunLoopRef rl = (CFRunLoopRef)_CFGetTSD(__CFTSDKeyRunLoop);
    if (rl) return rl;
    return _CFRunLoopGet0(pthread_self());  // 调用 _CFRunLoopGet0 函数并传入当前线程
}

// should only be called by Foundation
// t==0 is a synonym for "main thread" that always works
CF_EXPORT CFRunLoopRef _CFRunLoopGet0(pthread_t t) {
    if (pthread_equal(t, kNilPthreadT)) {
	t = pthread_main_thread_np();
    }
    __CFLock(&loopsLock);
    if (!__CFRunLoops) {
        __CFUnlock(&loopsLock);
	CFMutableDictionaryRef dict = CFDictionaryCreateMutable(kCFAllocatorSystemDefault, 0, NULL, &kCFTypeDictionaryValueCallBacks);
	CFRunLoopRef mainLoop = __CFRunLoopCreate(pthread_main_thread_np());
	CFDictionarySetValue(dict, pthreadPointer(pthread_main_thread_np()), mainLoop);
	if (!OSAtomicCompareAndSwapPtrBarrier(NULL, dict, (void * volatile *)&__CFRunLoops)) {
	    CFRelease(dict);
	}
	CFRelease(mainLoop);
        __CFLock(&loopsLock);
    }
    // ️当前线程作为 Key，从 __CFRunLoops 字典中获取 RunLoop 
    CFRunLoopRef loop = (CFRunLoopRef)CFDictionaryGetValue(__CFRunLoops, pthreadPointer(t));
    __CFUnlock(&loopsLock);
    if (!loop) {  // ️如果字典中不存在
	    CFRunLoopRef newLoop = __CFRunLoopCreate(t);  // 创建当前线程的 RunLoop
        __CFLock(&loopsLock);
	    loop = (CFRunLoopRef)CFDictionaryGetValue(__CFRunLoops, pthreadPointer(t));
    	if (!loop) {
	        CFDictionarySetValue(__CFRunLoops, pthreadPointer(t), newLoop);  // 保存到字典中
	        loop = newLoop;
    	}
        // don't release run loops inside the loopsLock, because CFRunLoopDeallocate may end up taking it
        __CFUnlock(&loopsLock);
	    CFRelease(newLoop);
    }
    if (pthread_equal(t, pthread_self())) {
        _CFSetTSD(__CFTSDKeyRunLoop, (void *)loop, NULL);
        if (0 == _CFGetTSD(__CFTSDKeyRunLoopCntr)) {
            _CFSetTSD(__CFTSDKeyRunLoopCntr, (void *)(PTHREAD_DESTRUCTOR_ITERATIONS-1), (void (*)(void *))__CFFinalizeRunLoop);
        }
    }
    return loop;
}

启动子线程的RunLoop

// Foundation
[[NSRunLoop currentRunLoop] run];
[[NSRunLoop currentRunLoop] runMode:NSDefaultRunLoopMode beforeDate:[NSDate distantFuture]];
// Core Foundation
CFRunLoopRun();
CFRunLoopRunInMode(kCFRunLoopDefaultMode, 1.0e10, false);  // 第3个参数：设置为 true，代表执行完 Source/Port 后就会退出当前 loop

来看一下CFRunLoopRun()/CFRunLoopRunInMode()函数是怎么启动RunLoop的:

void CFRunLoopRun(void) {
    int32_t result;
    do {
        result = CFRunLoopRunSpecific(CFRunLoopGetCurrent(), kCFRunLoopDefaultMode, 1.0e10, false);
        CHECK_FOR_FORK();
    } while (kCFRunLoopRunStopped != result && kCFRunLoopRunFinished != result);
}

SInt32 CFRunLoopRunInMode(CFStringRef modeName, CFTimeInterval seconds, Boolean returnAfterSourceHandled) {   
    CHECK_FOR_FORK();
    return CFRunLoopRunSpecific(CFRunLoopGetCurrent(), modeName, seconds, returnAfterSourceHandled);
}

可以看到它通过调用CFRunLoopRunSpecific()函数来启动Runloop

实现一个常驻线程

• 好处：经常用到子线程的时候，不用一直创建销毁，提高性能；
• 条件：该任务需是串行的，而非并发；
• 步骤：
  ① 获取/创建当前线程的RunLoop；
  ② 向该RunLoop中添加一个Source/Port等来维持RunLoop的事件循环（如果 Mode 里没有任何Source0/Source1/Timer/Observer，RunLoop会立马退出）；
  ③ 启动该RunLoop。

// ViewController.m
#import "ViewController.h"
#import "HTThread.h"

@interface ViewController ()
@property (nonatomic, strong) HTThread *thread;
@property (nonatomic, assign, getter=isStoped) BOOL stopped;
@end

@implementation ViewController

- (void)viewDidLoad {
    [super viewDidLoad];
    
    __weak typeof(self) weakSelf = self;
    
    self.stopped = NO;
    self.thread = [[HTThread alloc] initWithBlock:^{
        NSLog(@"begin-----%@", [NSThread currentThread]);
        
        // ① 获取/创建当前线程的 RunLoop 
        // ② 向该 RunLoop 中添加一个 Source/Port 等来维持 RunLoop 的事件循环
        [[NSRunLoop currentRunLoop] addPort:[[NSPort alloc] init] forMode:NSDefaultRunLoopMode];
    
        while (weakSelf && !weakSelf.isStoped) {
            // ③ 启动该 RunLoop
            /* 
              [[NSRunLoop currentRunLoop] run]
              如果调用 RunLoop 的 run 方法，则会开启一个永不销毁的线程
              因为 run 方法会通过反复调用 runMode:beforeDate: 方法，以运行在 NSDefaultRunLoopMode 模式下
              换句话说，该方法有效地开启了一个无限的循环，处理来自 RunLoop 的输入源 Sources 和 Timers 的数据
            */ 
            [[NSRunLoop currentRunLoop] runMode:NSDefaultRunLoopMode beforeDate:[NSDate distantFuture]];
        }
        NSLog(@"end-----%@", [NSThread currentThread]);    
    }];
    [self.thread start];
}

- (void)touchesBegan:(NSSet<UITouch *> *)touches withEvent:(UIEvent *)event
{
    if (!self.thread) return;
    [self performSelector:@selector(test) onThread:self.thread withObject:nil waitUntilDone:NO];
}

// 子线程需要执行的任务
- (void)test
{
    NSLog(@"%s-----%@", __func__, [NSThread currentThread]);
}

// 停止子线程的 RunLoop
- (void)stopThread
{
    // 设置标记为 YES
    self.stopped = YES;   
    // 停止 RunLoop
    CFRunLoopStop(CFRunLoopGetCurrent());    
    NSLog(@"%s-----%@", __func__, [NSThread currentThread]);
    // 清空线程
    self.thread = nil;
}

- (void)dealloc
{
    NSLog(@"%s", __func__);
    
    if (!self.thread) return;
    // 在子线程调用（waitUntilDone设置为YES，代表子线程的代码执行完毕后，当前方法才会继续往下执行）
    [self performSelector:@selector(stopThread) onThread:self.thread withObject:nil waitUntilDone:YES];
}

@end


// HTThread.h
#import <Foundation/Foundation.h>
@interface HTThread : NSThread
@end

// HTThread.m
#import "HTThread.h"
@implementation HTThread
- (void)dealloc
{
    NSLog(@"%s", __func__);
}
@end

点击view，接着退出当前ViewController。输出如下:

begin-----<HTThread: 0x600002b71240>{number = 6, name = (null)}
-[ViewController test]-----<HTThread: 0x600002b71240>{number = 6, name = (null)}
-[ViewController dealloc]
-[ViewController stopThread]-----<HTThread: 0x600002b71240>{number = 6, name = (null)}
end-----<HTThread: 0x600002b71240>{number = 6, name = (null)}
-[HTThread dealloc]

参考

深入浅出 RunLoop