https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52291
Martin Sebor <msebor at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2016-01-19
CC| |msebor at gcc dot gnu.org
Assignee|unassigned at gcc dot gnu.org |msebor at gcc dot
gnu.org
Ever confirmed|0 |1
--- Comment #4 from Martin Sebor <msebor at gcc dot gnu.org> ---
I suspect this is not going to change for the __sync_xxx builtins that have
been made obsolete by the newer __atomic_xxx builtins. Those, unfortunately,
have the same semantics as far as pointers as the former builtins. Worse yet,
these semantics are exposed in the C11 atomic_xxx() generic functions, making
the latter incorrect. For instance, atomic_fetch_add(p, N) is expected to be
equivalent to p += N for any object pointer p, but in GCC it only holds for
pointers to objects 1 byte wide (see bug 64843).
I agree that the __sync_xxx semantics should be clarified in the manual to
avoid any surprises. Let me put something together.
