ios 利用RunLoop的原理去监控卡顿
一、卡顿问题的几种原因
- 复杂 UI 、图文混排的绘制量过大;
- 在主线程上做网络同步请求;
- 在主线程做大量的 IO 操作;
- 运算量过大,CPU 持续高占用;
- 死锁和主子线程抢锁。
二、监测卡顿的思路
- 监测FPS:FPS 是一秒显示的帧数,也就是一秒内画面变化数量。如果按照动画片来说,动画片的 FPS 就是 24,是达不到 60 满帧的。也就是说,对于动画片来说,24 帧时虽然没有 60 帧时流畅,但也已经是连贯的了,所以并不能说 24 帧时就算是卡住了。 由此可见,简单地通过监视 FPS 是很难确定是否会出现卡顿问题了,所以我就果断弃了通过监视 FPS 来监控卡顿的方案。
- RunLoop:通过监控 RunLoop 的状态来判断是否会出现卡顿。RunLoop原理这里就不再多说,主要说方法,首先明确loop的状态有六个
typedef CF_OPTIONS(CFOptionFlags, CFRunLoopActivity) { kCFRunLoopEntry , // 进入 loop kCFRunLoopBeforeTimers , // 触发 Timer 回调 kCFRunLoopBeforeSources , // 触发 Source0 回调 kCFRunLoopBeforeWaiting , // 等待 mach_port 消息 kCFRunLoopAfterWaiting ), // 接收 mach_port 消息 kCFRunLoopExit , // 退出 loop kCFRunLoopAllActivities // loop 所有状态改变 }
我们需要监测的状态有两个:RunLoop 在进入睡眠之前和唤醒后的两个 loop 状态定义的值,分别是 kCFRunLoopBeforeSources 和 kCFRunLoopAfterWaiting ,也就是要触发 Source0 回调和接收 mach_port 消息两个状态。
三、如何检查卡顿
先粗略说下步骤:
- 创建一个 CFRunLoopObserverContext 观察者;
- 将创建好的观察者 runLoopObserver 添加到主线程 RunLoop 的 common 模式下观察;
- 创建一个持续的子线程专门用来监控主线程的 RunLoop 状态;
- 一旦发现进入睡眠前的 kCFRunLoopBeforeSources 状态,或者唤醒后的状态 kCFRunLoopAfterWaiting,在设置的时间阈值内一直没有变化,即可判定为卡顿;
- dump 出堆栈的信息,从而进一步分析出具体是哪个方法的执行时间过长;
上代码:
// // SMLagMonitor.h // // Created by DaiMing on 16/3/28. // #import <Foundation/Foundation.h> @interface SMLagMonitor : NSObject + (instancetype)shareInstance; - (void)beginMonitor; //开始监视卡顿 - (void)endMonitor; //停止监视卡顿 @end
// // SMLagMonitor.m // // Created by DaiMing on 16/3/28. // #import "SMLagMonitor.h" #import "SMCallStack.h" #import "SMCPUMonitor.h" @interface SMLagMonitor() { int timeoutCount; CFRunLoopObserverRef runLoopObserver; @public dispatch_semaphore_t dispatchSemaphore; CFRunLoopActivity runLoopActivity; } @property (nonatomic, strong) NSTimer *cpuMonitorTimer; @end @implementation SMLagMonitor #pragma mark - Interface + (instancetype)shareInstance { static id instance = nil; static dispatch_once_t dispatchOnce; dispatch_once(&dispatchOnce, ^{ instance = [[self alloc] init]; }); return instance; } - (void)beginMonitor { //监测 CPU 消耗 self.cpuMonitorTimer = [NSTimer scheduledTimerWithTimeInterval:3 target:self selector:@selector(updateCPUInfo) userInfo:nil repeats:YES]; //监测卡顿 if (runLoopObserver) { return; } dispatchSemaphore = dispatch_semaphore_create(0); //Dispatch Semaphore保证同步 //创建一个观察者 CFRunLoopObserverContext context = {0,(__bridge void*)self,NULL,NULL}; runLoopObserver = CFRunLoopObserverCreate(kCFAllocatorDefault, kCFRunLoopAllActivities, YES, 0, &runLoopObserverCallBack, &context); //将观察者添加到主线程runloop的common模式下的观察中 CFRunLoopAddObserver(CFRunLoopGetMain(), runLoopObserver, kCFRunLoopCommonModes); //创建子线程监控 dispatch_async(dispatch_get_global_queue(0, 0), ^{ //子线程开启一个持续的loop用来进行监控 while (YES) { long semaphoreWait = dispatch_semaphore_wait(dispatchSemaphore, dispatch_time(DISPATCH_TIME_NOW, 20*NSEC_PER_MSEC)); if (semaphoreWait != 0) { if (!runLoopObserver) { timeoutCount = 0; dispatchSemaphore = 0; runLoopActivity = 0; return; } //两个runloop的状态,BeforeSources和AfterWaiting这两个状态区间时间能够检测到是否卡顿 if (runLoopActivity == kCFRunLoopBeforeSources || runLoopActivity == kCFRunLoopAfterWaiting) { // 将堆栈信息上报服务器的代码放到这里 //出现三次出结果 // if (++timeoutCount < 3) { // continue; // } NSLog(@"monitor trigger"); dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_HIGH, 0), ^{ // [SMCallStack callStackWithType:SMCallStackTypeAll]; }); } //end activity }// end semaphore wait timeoutCount = 0; }// end while }); } - (void)endMonitor { [self.cpuMonitorTimer invalidate]; if (!runLoopObserver) { return; } CFRunLoopRemoveObserver(CFRunLoopGetMain(), runLoopObserver, kCFRunLoopCommonModes); CFRelease(runLoopObserver); runLoopObserver = NULL; } #pragma mark - Private static void runLoopObserverCallBack(CFRunLoopObserverRef observer, CFRunLoopActivity activity, void *info){ SMLagMonitor *lagMonitor = (__bridge SMLagMonitor*)info; lagMonitor->runLoopActivity = activity; dispatch_semaphore_t semaphore = lagMonitor->dispatchSemaphore; dispatch_semaphore_signal(semaphore); } - (void)updateCPUInfo { thread_act_array_t threads; mach_msg_type_number_t threadCount = 0; const task_t thisTask = mach_task_self(); kern_return_t kr = task_threads(thisTask, &threads, &threadCount); if (kr != KERN_SUCCESS) { return; } for (int i = 0; i < threadCount; i++) { thread_info_data_t threadInfo; thread_basic_info_t threadBaseInfo; mach_msg_type_number_t threadInfoCount = THREAD_INFO_MAX; if (thread_info((thread_act_t)threads[i], THREAD_BASIC_INFO, (thread_info_t)threadInfo, &threadInfoCount) == KERN_SUCCESS) { threadBaseInfo = (thread_basic_info_t)threadInfo; if (!(threadBaseInfo->flags & TH_FLAGS_IDLE)) { integer_t cpuUsage = threadBaseInfo->cpu_usage / 10; if (cpuUsage > 70) { //cup 消耗大于 70 时打印和记录堆栈 NSString *reStr = smStackOfThread(threads[i]); //记录数据库中 // [[[SMLagDB shareInstance] increaseWithStackString:reStr] subscribeNext:^(id x) {}]; NSLog(@"CPU useage overload thread stack:\n%@",reStr); } } } } } @end
使用,直接在APP didFinishLaunchingWithOptions 方法里面这样写:
[[SMLagMonitor shareInstance] beginMonitor];
搞定!