is it a bug of /glibc/sysdeps/arm/atomicity.h ?

[Carlos O'Donell]

> > let us consider the following scenario. Suppose there are three thread
> > pathes. 
> > Init:   *p = 0
> > T1 compare_and_swap(p, 0, 1) ; 
> > T2 compare_and_swap(p,0,1); 
> > T3 compare_and_swap(p,1,2);
> 
> Yes, that function is not actually atomic.  If you search the mailing
> list archives for gcc@gcc.gnu.org you'll find several possible fixes;
> no one's done any of them.

Can you get away with a light-weight syscall?

c.