chrome源码里weakptr的使用

[小宇去读书]

最近在看chrome源码，对weakptr的使用来使task具有撤销特性大致有了一个了解，但是也有一个问题：
weakptr是使用弱引用来使原有引用对象析构时，自身可以通过查询得到引用对象已经失效，一般性使用为：

    
    template<class Method, class Params>
    class RunnableMethod : public CancelableTask
    {
    public:
        RunnableMethod(const base::WeakPtr<T>& obj,
            Method meth, const Params& params)
            : obj_(obj), meth_(meth), params_(params)
        {
            COMPILE_ASSERT((MethodUsesScopedRefptrCorrectly<Method, Params>::value),
                badscopedrunnablemethodparams);
        }

        virtual void Run()
        {
            if(obj_)    
            {
                DispatchToMethod(obj_.get(), meth_, params_);
            }
        }
        virtual void Cancel()
        {
            obj_.reset();
        }

    private:
        base::WeakPtr<T> obj_;
        Method meth_;
        Params params_;

        DISALLOW_COPY_AND_ASSIGN(RunnableMethod);
    };

为题就是在run函数这里：
virtual void Run()
{
if(obj_) //判断引用对象是否有效，weakptr等一系列函数的功劳
{
DispatchToMethod(obj_.get(), meth_, params_); //有效的话执行函数调用
}
}

有种情况就是run是在另一个线程中执行，执行时得知obj_有效，进入meth_调用，此时obj_对象的拥有线程执行了obj对象的析构，那么此时meth_调用中访问obj的成员变量就会发生不可预知的行为了，这种情况是如何避免的，欢迎大家指教。

[小宇去读书]

官方文档写明了这种方式是non threadsafe的，哎，翻遍了整个baidu都没有看到这个，还是官方的英文文档靠谱。。 base::WeakPtr and Cancellation [NOT THREAD SAFE] You can use a base::WeakPtr and base::WeakPtrFactory (in base/memory/weak_ptr.h) to ensure that any invokes can not outlive the object they are being invoked on, without using reference counting. The base::Bind mechanism has special understanding for base::WeakPtr that will disable the task's execution if the base::WeakPtr has been invalidated. The base::WeakPtrFactory object can be used to generate base::WeakPtr instances that know about the factory object. When the factory is destroyed, all the base::WeakPtr will have their internal "invalidated" flag set, which will make any tasks bound to them to not dispatch. By putting the factory as a member of the object being dispatched to, you can get automatic cancellation. NOTE: This only works when the task is posted to the same thread. Currently there is not a general solution that works for tasks posted to other threads. See the next section about CancelableTaskTracker for an alternative solution. class MyObject { public: MyObject() : weak_factory_(this) {} void DoSomething() { const int kDelayMS = 100; MessageLoop::current()->PostDelayedTask(FROM_HERE, base::Bind(&MyObject::DoSomethingLater, weak_factory_.GetWeakPtr()), kDelayMS); } void DoSomethingLater() { ... } private: base::WeakPtrFactory<MyObject> weak_factory_; };

[要追问请引用回复]

应该和std::weak_ptr std::tr1::weak_ptr boost::weak_ptr是类似的，是shared_ptr的补充

[misserwell]

楼主的意思 是 不是 NOTE: This only works when the task is posted to the same thread. 这句话

[ri_aje]

不知道 google 家的 weakptr 具体怎么写的。 标准的 weak_ptr lock 的时候会返回 shared_ptr，从而保证引用计数没有线程问题，不过多线程读写被管理对象仍然需要同步。

[sumos]

与c++的weak_ptr一样，weak_ptr是线程安全，但是task就不是了， NOTE: This only works when the task is posted to the same thread. 这句话是对的，必须保证你的对象 和 该task在同一线程 或者说保证task执行时对象不会被删除