[Omp] OpenMP spec 2.5 seems to have incorrect flush example on page 12

[Greg Bronevetsky]

> Anyway I think that flushes are handy in the absence of any 
> point-to-point synchronization mechanisms in OpenMP.
> 
> So I think that an advice "you should not ..." is not helpful in those 
> circumstances.
> 
   I disagree. If you look at how little is actually guaranteed by the
OpenMP memory model, there are very few point-to-point synchronizations
are are sure to work. The only one that I'm aware of is one process
writing to a variable and another process polling this variable until it
sees a change (with flushes sprinkled liberally).

                             Greg Bronevetsky