================ @@ -0,0 +1,9 @@ +// RUN: %clang_cc1 -triple amdgcn -cl-std=clc++ -verify %s + +// expected-no-diagnostics + +#define fold(x) (__builtin_constant_p(x) ? (x) : (x)) +static_assert(nullptr != fold(reinterpret_cast<private int*>(0))); + +static_assert(nullptr == (private int *)0); + ---------------- zygoloid wrote:
Yeah, it's weird, but that's just how C++ is: converting a null pointer constant to pointer type is a completely different operation from reinterpreting an integer value (that happens to be 0) as a pointer value. Similar weirdness would happen in C for this target: ```c private int *p = (private int*)0; // null pointer. intptr_t n = 0; private int *q = (private int*)n; // not a null pointer. intptr_t m = -1; private int *r = (private int*)m; // null pointer. ``` https://github.com/llvm/llvm-project/pull/150164 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
