[libc++] Simplify when the sized global deallocations overloads are available

[philnik777]

There doesn't seem to be much benefit in always providing declarations for the sized deallocations from C++14 onwards if the user explicitly passed -fno-sized-deallocation to disable them. This patch simplifies the declarations to be available exactly when the compiler expects sized deallocation functions to be available.

[llvmbot]

@llvm/pr-subscribers-libcxx

Author: Nikolas Klauser (philnik777)

Changes

There doesn't seem to be much benefit in always providing declarations for the sized deallocations from C++14 onwards if the user explicitly passed -fno-sized-deallocation to disable them. This patch simplifies the declarations to be available exactly when the compiler expects sized deallocation functions to be available.

---

Full diff: https://github.com/llvm/llvm-project/pull/114667.diff

1 Files Affected:

• (modified) libcxx/include/new (+4-16)

diff --git a/libcxx/include/new b/libcxx/include/new
index 290ad9e97f8ded..c7678b64c7d494 100644
--- a/libcxx/include/new
+++ b/libcxx/include/new
@@ -104,18 +104,6 @@ void  operator delete[](void* ptr, void*) noexcept;
 #endif
 
 #if defined(__cpp_sized_deallocation) && __cpp_sized_deallocation >= 201309L
-#  define _LIBCPP_HAS_LANGUAGE_SIZED_DEALLOCATION 1
-#else
-#  define _LIBCPP_HAS_LANGUAGE_SIZED_DEALLOCATION 0
-#endif
-
-#if _LIBCPP_STD_VER >= 14 || _LIBCPP_HAS_LANGUAGE_SIZED_DEALLOCATION
-#  define _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION 1
-#else
-#  define _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION 0
-#endif
-
-#if _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION && _LIBCPP_HAS_LANGUAGE_SIZED_DEALLOCATION
 #  define _LIBCPP_HAS_SIZED_DEALLOCATION 1
 #else
 #  define _LIBCPP_HAS_SIZED_DEALLOCATION 0
@@ -214,7 +202,7 @@ inline constexpr destroying_delete_t destroying_delete{};
     _LIBCPP_NOALIAS;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p) _NOEXCEPT;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, const std::nothrow_t&) _NOEXCEPT;
-#  if _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION
+#  if _LIBCPP_HAS_SIZED_DEALLOCATION
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
 #  endif
 
@@ -223,7 +211,7 @@ _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, std::size_t __sz) _
     _LIBCPP_NOALIAS;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p) _NOEXCEPT;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, const std::nothrow_t&) _NOEXCEPT;
-#  if _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION
+#  if _LIBCPP_HAS_SIZED_DEALLOCATION
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, std::size_t __sz) _NOEXCEPT;
 #  endif
 
@@ -233,7 +221,7 @@ _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, std::size_t __sz)
 operator new(std::size_t __sz, std::align_val_t, const std::nothrow_t&) _NOEXCEPT _LIBCPP_NOALIAS;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, std::align_val_t) _NOEXCEPT;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, std::align_val_t, const std::nothrow_t&) _NOEXCEPT;
-#    if _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION
+#    if _LIBCPP_HAS_SIZED_DEALLOCATION
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete(void* __p, std::size_t __sz, std::align_val_t) _NOEXCEPT;
 #    endif
 
@@ -243,7 +231,7 @@ operator new[](std::size_t __sz, std::align_val_t) _THROW_BAD_ALLOC;
 operator new[](std::size_t __sz, std::align_val_t, const std::nothrow_t&) _NOEXCEPT _LIBCPP_NOALIAS;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, std::align_val_t) _NOEXCEPT;
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, std::align_val_t, const std::nothrow_t&) _NOEXCEPT;
-#    if _LIBCPP_HAS_LIBRARY_SIZED_DEALLOCATION
+#    if _LIBCPP_HAS_SIZED_DEALLOCATION
 _LIBCPP_OVERRIDABLE_FUNC_VIS void operator delete[](void* __p, std::size_t __sz, std::align_val_t) _NOEXCEPT;
 #    endif
 #  endif

[ldionne]

This makes sense to me, and I think the same should apply to aligned allocation as well (we should do that separately).

However, I'd like to get @EricWF 's thoughts on this since he was the one to introduce this 2-level dance for aligned allocation if I remember correctly.

Also, some CI failures are not flukes and should be addressed.