Subj : Re: Memory Barriers, Compiler Optimizations, etc.
To   : comp.programming.threads
From : Alexander Terekhov
Date : Thu Feb 03 2005 04:33 pm


Markus Schaaf wrote:
[...]
> > http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vclang/html/vclrf_readwritebarrier.asp.)
> 
> It seems worth noting, that acquire and release barriers are just a variety
> of memory visibility models. If you are concerned with broader aspects of
> memory visibility, say, from perspective of a general purpose programming
> language, it might be useful to widen the focus.

Right. Idiotic "global memory" optimization blocker aside for a moment, MS 
folks actually meant:

  int G;
  int i;

  atomic<int> ReleaseF(0), WaitF(0);

  void f(void *p)
  {
     G = 1;

     WaitF.store(1, msync::ssb);
     while (ReleaseF.load(msync::hsb) == 0);

     G = 2;   
  }

  int main()
  {
     _beginthread(f, 0, NULL);   // New thread
  
     while (WaitF.load(msync::hlb) == 0)   
        Sleep(1);

     if (G == 1)
        puts("G is equal to 1, as expected.");
     else
        puts("G is NOT equal to 1!");

     ReleaseF.store(1, msync::slb);
  }

regards,
alexander.

.