[cfe-dev] Const-ness of atomic_load parameter in C

[JF Bastien via cfe-dev]

Hi cfe-dev,

C++11 onwards specs the non-member functions atomic_load and atomic_load_explicit as taking the atomic<T> by const (potentially volatile) pointer. C11, in its infinite wisdom, decided to drop the const, and C17 will fix this with DR459 <http://www.open-std.org/jtc1/sc22/wg14/www/docs/summary.htm#dr_459> (the current draft forgot to fix B.16, but that’s not the normative part).

I suggest we allow passing const _Atomic(T)* in all versions of C. It seems silly to special-case C11’s semantics because they allow strictly less code to work. I’d usually just post these patches, but the not-quite-correct C11 semantics made me think twice.

This doesn’t affect C++ (unless you try to use stdatomic.h in C++ which… you shouldn’t…).

I would change the following (independently, they don’t need to synchronize-with):
clang’s lib/Headers/stdatomic.h implements these as #define to the __c11_* equivalent, which are builtins with custom typecheck. Fix the typecheck.
Additionally, libc++ defines the __c11_* functions in include/atomic for the sake of GCC compatibility. Fix their declaration.

It seems GCC already doesn’t require const: https://godbolt.org/g/4Q2vre <https://godbolt.org/g/4Q2vre>

Thoughts?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20180529/a2ca92ef/attachment.html>