[CUDA][NFCI] Use unqualified lookup for math functions

[jdoerfert]

Summary

The other macro uses a unqualified lookup already and the qualified one
will cause problems in the OpenMP overlay.

Depends on D77238.

[tra]

The other macro uses a unqualified lookup already and the qualified one
will cause problems in the OpenMP overlay.

There's a bit of inconsitency here. While __CUDA_CLANG_FN_INTEGER_OVERLOAD_2 indeed uses unqualified lookup, pretty much all other functions use qualified global scope lookups, which, I believe is correct, as we want to use the functions defined by CUDA headers there.

@jlebar -- WDYT?

[jdoerfert]

In D77239#1955720, @tra wrote:

The other macro uses a unqualified lookup already and the qualified one
will cause problems in the OpenMP overlay.

There's a bit of inconsitency here. While __CUDA_CLANG_FN_INTEGER_OVERLOAD_2 indeed uses unqualified lookup, pretty much all other functions use qualified global scope lookups, which, I believe is correct, as we want to use the functions defined by CUDA headers there.

@jlebar -- WDYT?

It turns out I can disable this entire code region for OpenMP (or have to for c++17 I tihnk), so I'll abandon this revision and you make it consistent in the way that fits your need :)

clang/lib/Headers/__clang_cuda_cmath.h

 // Defines an overload of __fn that accepts one integral argument, calls
 // __fn((double)x), and returns __retty.
 #define __CUDA_CLANG_FN_INTEGER_OVERLOAD_1(__retty, __fn)                      \
   template <typename __T>                                                      \
   __DEVICE__                                                                   \
       typename __clang_cuda_enable_if<std::numeric_limits<__T>::is_integer,    \
                                       __retty>::type                           \
       __fn(__T __x) {                                                          \
-    return ::__fn((double)__x);                                                \
+    return __fn((double)__x);                                                  \
   }
 
 // Defines an overload of __fn that accepts one two arithmetic arguments, calls
 // __fn((double)x, (double)y), and returns a double.
 //
 // Note this is different from OVERLOAD_1, which generates an overload that
 // accepts only *integral* arguments.
 #define __CUDA_CLANG_FN_INTEGER_OVERLOAD_2(__retty, __fn)                      \