Re: [Boost-users] shared_ptr race condition

From: David Abrahams (dave_at_[hidden])
Date: 2008-08-16 05:37:53

on Sat Aug 16 2008, "Eduardo Panisset" <eduardo.panisset-AT-gmail.com> wrote:

> Excuse me for insisting on my question,
>  
> But I would like to ensure that I'm understanding shared_ptr correctly.

Okay, but please use standard quoting so threading is preserved:

> > Not without additional synchronization (e.g. via a mutex with all
> > threads cooperating by locking before accessing the shared_ptr).  But do
> > you really need to do that?  Can you not give each thread that needs
> > access a separate copy of the shared_ptr?
>  
> Exactly. I give each thread a instance of shared_ptr, but the point is: the
> critical region of release member function does not avoid race conditions
> because It does not protect de code right below:
>  
>         if( new_use_count == 0 )
>         {
>             dispose();
>             weak_release();
>         }
>  
> How Can I ensure that the release member function wouldn't dispose a pointee
> more than one time ? (release executing concurrently from different
> shared_ptr instances) 

Because the reference count was 1, there can be only one shared_ptr
instance that owns the pointee.  No other threads can be referencing it.

[In fact it's very likely that no mutex is used on any given platform --
atomic reference counting is used where available.]

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com