[cpp-threads] Re: Thread API interface tweaks

[Alexander Terekhov]

On 9/1/06, Ion Gaztañaga <igaztanaga at gmail.com> wrote:
> > [...]
> > > >>     void unlock()
> > > >>     {
> > > >>         if (--count_ == 0)
> > > >>         {
> > > >>             id_ = thread_id();  // id_ = "not a thread"
> > > >
> > > > ... and here... data race?
> > >
> > > It is undefined for a thread for which id_ != self to call unlock().
> >
> > True, but think about thread id re-use. You need to delay re-use until
> > id_ = thread_id() becomes globally visible and there's nothing in your
> > code that ensures it, AFAICS.
>
> Sorry if I'm wrong, but does not thread creation/destruction involve
> memory synchronization? Hasn't pthread_create() release sematics and

http://www.opengroup.org/austin/mailarchives/ag-review/msg02017.html

> pthread_join() acquire semantics? So if a thread ends all previous
> operations shouldn't be completed (including writing an invalid
> thread_id)?

You seem to make assumptions. Suppose that a thread doing unlock() is
a POSIX detached thread so you just can't join it.

>
> And if this does not guarantee anything, isn't any way to make
> Howard's code correct adding some memory synchronization instruction?

I don't see how.

regards,
alexander.