Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.c++ > #6782

Re: Threading question - read-read coherency almost, sort of, with seq_cst fences

Show all headers | View raw

Joshua Maurice <joshuamaurice@gmail.com> writes:

> Consider the following code.
>
> /*** code 1 ***/
> // Initially
> atomic<int> x(0), y(0);
>
> // Thread 1:
> y.store(1, memory_order_release);
> atomic_thread_fence(memory_order_seq_cst); //fence 1
> r1 = x.load(memory_order_acquire);
>
> // Thread 2:
> x.store(1, memory_order_release);
>
> // Thread 3:
> r2 = x.load(memory_order_acquire);
> atomic_thread_fence(memory_order_seq_cst); //fence 3
> r3 = y.load(memory_order_acquire);
> /***************/
>
> Q1. First, am I right in concluding that in the above code, all of the
> loads and stores might as well be memory_order_relaxed, right?

Yes.

> Q2. Second, is the following end condition possible?
>   r1 == 0 && r2 == 1 && r3 == 0

Yes.

> Anthony Williams in the comp.programming.threads (who is apparently on
> the concurrency working group), initially had a different
> interpretation. I just reread it several times, and I'm not quite sure
> what he's trying to say. He initially thought that you can use it
> directly to prove that NOT (r1 == 0 && r2 == 1).

But I now believe I'm mistaken. After checking my notes, we discussed
requiring that seq_cst fences ordered loads as in this example, but
decided against it. 29.3p7 is intended to apply only to writes.

Anthony
-- 
Author of C++ Concurrency in Action     http://www.stdthread.co.uk/book/
just::thread C++0x thread library             http://www.stdthread.co.uk
Just Software Solutions Ltd       http://www.justsoftwaresolutions.co.uk
15 Carrallack Mews, St Just, Cornwall, TR19 7UL, UK. Company No. 5478976

Back to comp.lang.c++ | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Threading question - read-read coherency almost, sort of, with seq_cst fences Joshua Maurice <joshuamaurice@gmail.com> - 2011-06-15 16:56 -0700
  Re: Threading question - read-read coherency almost, sort of, with seq_cst fences "Chris M. Thomasson" <cristom@charter.net> - 2011-06-16 00:28 -0700
    Re: Threading question - read-read coherency almost, sort of, with seq_cst fences "Chris M. Thomasson" <cristom@charter.net> - 2011-06-16 00:41 -0700
      Re: Threading question - read-read coherency almost, sort of, with seq_cst fences Joshua Maurice <joshuamaurice@gmail.com> - 2011-06-16 02:47 -0700
        Re: Threading question - read-read coherency almost, sort of, with seq_cst fences Anthony Williams <anthony.ajw@gmail.com> - 2011-06-16 11:45 +0100
          Re: Threading question - read-read coherency almost, sort of, with seq_cst fences "Chris M. Thomasson" <cristom@charter.net> - 2011-06-16 14:32 -0700
  Re: Threading question - read-read coherency almost, sort of, with seq_cst fences Anthony Williams <anthony.ajw@gmail.com> - 2011-06-16 09:42 +0100
    Re: Threading question - read-read coherency almost, sort of, with seq_cst fences Joshua Maurice <joshuamaurice@gmail.com> - 2011-06-16 02:50 -0700
      Re: Threading question - read-read coherency almost, sort of, with seq_cst fences Joshua Maurice <joshuamaurice@gmail.com> - 2011-06-16 02:53 -0700

csiph-web