继续浏览精彩内容
慕课网APP
程序员的梦工厂
打开
继续
感谢您的支持,我会继续努力的
赞赏金额会直接到老师账户
将二维码发送给自己后长按识别
微信支付
支付宝支付

深度解析Java线程池的异常处理机制 #3

关注TA
已关注
手记
粉丝
获赞

前言

今天小伙伴遇到个小问题,线程池提交的任务如果没有catch异常,那么会抛到哪里去,之前倒是没研究过,本着实事求是的原则,看了一下代码。

正文

小问题

考虑下面这段代码,有什么区别呢?你可以猜猜会不会有异常打出呢?如果打出来的话是在哪里?:

        ExecutorService threadPool = Executors.newFixedThreadPool(1);        threadPool.submit(() -> {            Object obj = null;            System.out.println(obj.toString());        });        threadPool.execute(() -> {            Object obj = null;            System.out.println(obj.toString());        });

源码解析

我们下面就来看下代码, 其实就是将我们提交过去的Runnable包装成一个Future

    public Future<?> submit(Runnable task) {        if (task == null) throw new NullPointerException();        RunnableFuture<Void> ftask = newTaskFor(task, null);        execute(ftask);        return ftask;    }    protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {        return new FutureTask<T>(runnable, value);    }    public FutureTask(Runnable runnable, V result) {        this.callable = Executors.callable(runnable, result);        this.state = NEW;       // volatile修饰,保证多线程下的可见性,可以看看Java内存模型    }    public static <T> Callable<T> callable(Runnable task, T result) {        if (task == null)            throw new NullPointerException();        return new RunnableAdapter<T>(task, result);    }    static final class RunnableAdapter<T> implements Callable<T> {        final Runnable task;        final T result;        RunnableAdapter(Runnable task, T result) {            this.task = task;            this.result = result;        }        public T call() {            task.run();            return result;        }    }

接下来就会实际提交到队列中交给线程池调度处理:

    /**    * 代码还是很清爽的,一个很典型的生产者/消费者模型,    * 这里暂不纠结这些细节,那么如果提交到workQueue成功的话,消费者是谁呢?    * 明显在这个newWorker里搞的鬼,同样细节有兴趣可以自己再去研究,这里我们会发现    * 核心就是Worker这个内部类    */    public void execute(Runnable command) {        if (command == null)            throw new NullPointerException();        int c = ctl.get();        if (workerCountOf(c) < corePoolSize) {            if (addWorker(command, true))                return;            c = ctl.get();        }        if (isRunning(c) && workQueue.offer(command)) {            int recheck = ctl.get();            if (! isRunning(recheck) && remove(command))                reject(command);            else if (workerCountOf(recheck) == 0)                addWorker(null, false);        }        else if (!addWorker(command, false))            reject(command);    }

那么接下来看看线程池核心的流程:

private final class Worker        extends AbstractQueuedSynchronizer        implements Runnable{          /** Delegates main run loop to outer runWorker  */        public void run() {            runWorker(this);        }}final void runWorker(Worker w) {        Thread wt = Thread.currentThread();        Runnable task = w.firstTask;        w.firstTask = null;        w.unlock(); // allow interrupts        boolean completedAbruptly = true;        try {            //getTask()方法会尝试从队列中抓取数据            while (task != null || (task = getTask()) != null) {                w.lock();                if ((runStateAtLeast(ctl.get(), STOP) ||                     (Thread.interrupted() &&                      runStateAtLeast(ctl.get(), STOP))) &&                    !wt.isInterrupted())                    wt.interrupt();                try {                    //可覆写此方法打日志埋点之类的                    beforeExecute(wt, task);                    Throwable thrown = null;                    try {                        //简单明了,直接调用run方法                        task.run();                    } catch (RuntimeException x) {                        thrown = x; throw x;                    } catch (Error x) {                        thrown = x; throw x;                    } catch (Throwable x) {                        thrown = x; throw new Error(x);                    } finally {                        afterExecute(task, thrown);                    }                } finally {                    task = null;                    w.completedTasks++;                    w.unlock();                }            }            completedAbruptly = false;        } finally {            processWorkerExit(w, completedAbruptly);        }    }

submit的方式

那么我们可以这里是直接调用的run方法,先看submit的方式,我们知道最终传递过去的是一个FutureTask,也就是说会调用这里的run方法,我们看看实现:

    public void run() {        if (state != NEW ||            !UNSAFE.compareAndSwapObject(this, runnerOffset,                                         null, Thread.currentThread()))            return;        try {            Callable<V> c = callable;            if (c != null && state == NEW) {                V result;                boolean ran;                try {                    result = c.call();                    ran = true;                } catch (Throwable ex) {                    result = null;                    ran = false;                    //。。。                    setException(ex);                }                if (ran)                    set(result);            }        } finally {          //省略    }      protected void setException(Throwable t) {        if (UNSAFE.compareAndSwapInt(this, stateOffset, NEW, COMPLETING)) {            outcome = t; //赋给了这个变量            UNSAFE.putOrderedInt(this, stateOffset, EXCEPTIONAL); // final state            finishCompletion();        }    }

可以看到其实类似于直接吞掉了,这样的话我们调用get()方法的时候会拿到, 比如我们可以重写afterExecute方法,从而可以得到实际的异常:

protected void afterExecute(Runnable r, Throwable t) {          super.afterExecute(r, t);          if (t == null && r instanceof Future<?>) {            try {              //get这里会首先检查任务的状态,然后将上面的异常包装成ExecutionException              Object result = ((Future<?>) r).get();            } catch (CancellationException ce) {                t = ce;            } catch (ExecutionException ee) {                t = ee.getCause();            } catch (InterruptedException ie) {                Thread.currentThread().interrupt(); // ignore/reset            }          }          if (t != null){            //异常处理            t.printStackTrace();          }        }

execute的方式

那么如果是直接exeture的方式有啥不同呢?这样的话传递过去的就直接是Runnable,因此就会直接抛出:

    try {        task.run();    } catch (RuntimeException x) {        thrown = x; throw x;    } catch (Error x) {        thrown = x; throw x;    } catch (Throwable x) {        thrown = x; throw new Error(x);    } finally {        afterExecute(task, thrown);    }

那么这里的异常到底会抛出到哪里呢, 我们看看JVM具体是怎么处理的:

if (!destroy_vm || JDK_Version::is_jdk12x_version()) {    // JSR-166: change call from from ThreadGroup.uncaughtException to    // java.lang.Thread.dispatchUncaughtException    if (uncaught_exception.not_null()) {      //如果有未捕获的异常      Handle group(this, java_lang_Thread::threadGroup(threadObj()));      {        KlassHandle recvrKlass(THREAD, threadObj->klass());        CallInfo callinfo;        KlassHandle thread_klass(THREAD, SystemDictionary::Thread_klass());        /*              这里类似一个方法表,实际就会去调用Thread#dispatchUncaughtException方法            template(dispatchUncaughtException_name,            "dispatchUncaughtException")                        */        LinkResolver::resolve_virtual_call(callinfo, threadObj, recvrKlass, thread_klass,                                           vmSymbols::dispatchUncaughtException_name(),                                           vmSymbols::throwable_void_signature(),                                           KlassHandle(), false, false, THREAD);        CLEAR_PENDING_EXCEPTION;        methodHandle method = callinfo.selected_method();        if (method.not_null()) {          JavaValue result(T_VOID);          JavaCalls::call_virtual(&result,                                  threadObj, thread_klass,                                  vmSymbols::dispatchUncaughtException_name(),                                  vmSymbols::throwable_void_signature(),                                  uncaught_exception,                                  THREAD);        } else {          KlassHandle thread_group(THREAD, SystemDictionary::ThreadGroup_klass());          JavaValue result(T_VOID);          JavaCalls::call_virtual(&result,                                  group, thread_group,                                  vmSymbols::uncaughtException_name(),                                  vmSymbols::thread_throwable_void_signature(),                                  threadObj,           // Arg 1                                  uncaught_exception,  // Arg 2                                  THREAD);        }        if (HAS_PENDING_EXCEPTION) {          ResourceMark rm(this);          jio_fprintf(defaultStream::error_stream(),                "\nException: %s thrown from the UncaughtExceptionHandler"                " in thread \"%s\"\n",                pending_exception()->klass()->external_name(),                get_thread_name());          CLEAR_PENDING_EXCEPTION;        }      }    }

可以看到这里最终会去调用Thread#dispatchUncaughtException方法:

    private void dispatchUncaughtException(Throwable e) {        //默认会调用ThreadGroup的实现        getUncaughtExceptionHandler().uncaughtException(this, e);    }


    public void uncaughtException(Thread t, Throwable e) {        if (parent != null) {            parent.uncaughtException(t, e);        } else {            Thread.UncaughtExceptionHandler ueh =                Thread.getDefaultUncaughtExceptionHandler();            if (ueh != null) {                ueh.uncaughtException(t, e);            } else if (!(e instanceof ThreadDeath)) {                //可以看到会打到System.err里面                System.err.print("Exception in thread \""                                 + t.getName() + "\" ");                e.printStackTrace(System.err);            }        }    }

这里如果环境是tomcat的话最终会打到catalina.out:


总结

对于线程池、包括线程的异常处理推荐一下方式:

1    直接try/catch,个人 基本都是用这种方式

2    线程直接重写整个方法:

          Thread t = new Thread();          t.setUncaughtExceptionHandler(new Thread.UncaughtExceptionHandler() {              public void uncaughtException(Thread t, Throwable e) {                 LOGGER.error(t + " throws exception: " + e);              }           });        //如果是线程池的模式:           ExecutorService threadPool = Executors.newFixedThreadPool(1, r -> {               Thread t = new Thread(r);               t.setUncaughtExceptionHandler(                   (t1, e) -> LOGGER.error(t1 + " throws exception: " + e));               return t;           });

3    也可以直接重写protected void afterExecute(Runnable r, Throwable t) { }方法

打开App,阅读手记
0人推荐
发表评论
随时随地看视频慕课网APP