[CUDA][HIP] Do not diagnose use of _Float16

[yaxunl]

Summary

r352221 caused regressions in CUDA/HIP since device function may use _Float16 whereas host does not support it.
In this case host compilation should not diagnose usage of _Float16 in device functions or variables.

For now just do not diagnose _Float16 for CUDA/HIP. In the future we should have more precise check.

[rjmccall]

Okay.

lib/Lex/LiteralSupport.cpp

case 'H':
       isHalf = true;
       continue;  // Success.
     case 'f':      // FP Suffix for "float"
     case 'F':
       if (!isFPConstant) break;  // Error for integer constant.
       if (isHalf || isFloat || isLong || isFloat128)
         break; // HF, FF, LF, QF invalid.
 
-      if (PP.getTargetInfo().hasFloat16Type() && s + 2 < ThisTokEnd &&
-          s[1] == '1' && s[2] == '6') {
+      // CUDA host and device may have different _Float16 support, therefore
+      // allows f16 literals to avoid false alarm.
+      // ToDo: more precise check for CUDA.
+      if ((PP.getTargetInfo().hasFloat16Type() || PP.getLangOpts().CUDA) &&
+          s + 2 < ThisTokEnd && s[1] == '1' && s[2] == '6') {
         s += 2; // success, eat up 2 characters.
         isFloat16 = true;
         continue;
       }
 
       isFloat = true;
       continue;  // Success.
     case 'q':    // FP Suffix for "__float128"

lib/Sema/SemaType.cpp

if (!S.Context.getTargetInfo().hasInt128Type())
       S.Diag(DS.getTypeSpecTypeLoc(), diag::err_type_unsupported)
         << "__int128";
     if (DS.getTypeSpecSign() == DeclSpec::TSS_unsigned)
       Result = Context.UnsignedInt128Ty;
     else
       Result = Context.Int128Ty;
     break;
   case DeclSpec::TST_float16:
-    if (!S.Context.getTargetInfo().hasFloat16Type())
+    // CUDA host and device may have different _Float16 support, therefore
+    // do not diagnose _Float16 usage to avoid false alarm.
+    // ToDo: more precise diagnostics for CUDA.
+    if (!S.Context.getTargetInfo().hasFloat16Type() && !S.getLangOpts().CUDA)
       S.Diag(DS.getTypeSpecTypeLoc(), diag::err_type_unsupported)
         << "_Float16";
     Result = Context.Float16Ty;
     break;
   case DeclSpec::TST_half:    Result = Context.HalfTy; break;
   case DeclSpec::TST_float:   Result = Context.FloatTy; break;
   case DeclSpec::TST_double:
     if (DS.getTypeSpecWidth() == DeclSpec::TSW_long)

test/SemaCUDA/float16.cu

+// RUN: %clang_cc1 -fsyntax-only -triple x86_64 -aux-triple amdgcn -verify %s
+// expected-no-diagnostics
+#include "Inputs/cuda.h"
+
+__device__ void f(_Float16 x);
+
+__device__ _Float16 x = 1.0f16;