[PATCH] Update to new generic semaphore algorithm v2

[Florian Weimer]

* Andreas Schwab:

> This has the effect that a semaphore that is shared between a x86-64 and
> a x86-32 process no longer yields the same value, because the latter
> does not define __HAVE_64B_ATOMICS.

Hasn't this been reported before?  I think we decided back then that it
wasn't a bug.  Other process-shared synchronization objects (mutexes,
condition variables) had the same problem from the start.

Do you have new information that would call for a re-evaluation of this
decision?

Thanks,
Florian