[clang][aarch64] Fix mangling of bfloat16 neon vectors

[c-rhodes]

Summary

The AAPCS64 specifies the internal type is used for c++ mangling. For
bfloat16 it was defined as BFloat16 when it should be Bfloat16, i.e.
lowercase 'f'.

For more information, see:

https://github.com/ARM-software/abi-aa/blob/master/aapcs64/aapcs64.rst#appendix-support-for-advanced-simd-extensions

[stuij]

Thanks, this one crept in during mass-renaming.

[stuij]

I'm not sure what the procedure is to get patches into LLVM 11, but I think there's still time. Could you try to get this in there as well @c-rhodes ?

[c-rhodes]

In D87463#2267613, @stuij wrote:

I'm not sure what the procedure is to get patches into LLVM 11, but I think there's still time. Could you try to get this in there as well @c-rhodes ?

Cherry-picked into LLVM 11, see: https://bugs.llvm.org/show_bug.cgi?id=47490

clang/lib/AST/ItaniumMangle.cpp

case BuiltinType::ULongLong:
     return "Uint64";
   case BuiltinType::Half:
     return "Float16";
   case BuiltinType::Float:
     return "Float32";
   case BuiltinType::Double:
     return "Float64";
   case BuiltinType::BFloat16:
-    return "BFloat16";
+    return "Bfloat16";
   default:
     llvm_unreachable("Unexpected vector element base type");
   }
 }
 
 // AArch64's ABI for Neon vector types specifies that they should be mangled as
 // the equivalent internal name. The vector type must be one of the special
 // types predefined by ARM.

clang/test/CodeGenCXX/mangle-neon-vectors.cpp

 // RUN: %clang_cc1 -triple armv7-apple-ios -target-feature +neon  %s -emit-llvm -o - | FileCheck %s
 // RUN: %clang_cc1 -triple arm64-apple-ios -target-feature +neon %s -emit-llvm -o - | FileCheck %s
 // RUN: %clang_cc1 -triple arm64-linux-gnu -target-feature +neon %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK-AARCH64
+// RUN: %clang_cc1 -triple arm64-linux-gnu -target-feature +neon -target-feature +bf16 %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK-AARCH64-BF16
 
 typedef float float32_t;
 typedef double float64_t;
 typedef __fp16 float16_t;
 #if defined(__aarch64__)
 typedef unsigned char poly8_t;
 typedef unsigned short poly16_t;
 #else
 typedef signed char poly8_t;
 typedef short poly16_t;
 #endif
 typedef unsigned __INT64_TYPE__ uint64_t;
 
+#if defined(__ARM_FEATURE_BF16)
+typedef __bf16 bfloat16_t;
+#endif
+
 typedef __attribute__((neon_vector_type(2))) int int32x2_t;
 typedef __attribute__((neon_vector_type(4))) int int32x4_t;
 typedef __attribute__((neon_vector_type(1))) uint64_t uint64x1_t;
 typedef __attribute__((neon_vector_type(2))) uint64_t uint64x2_t;
 typedef __attribute__((neon_vector_type(2))) float32_t float32x2_t;
 typedef __attribute__((neon_vector_type(4))) float32_t float32x4_t;
 typedef __attribute__((neon_vector_type(4))) float16_t float16x4_t;
 typedef __attribute__((neon_vector_type(8))) float16_t float16x8_t;
 #ifdef __aarch64__
 typedef __attribute__((neon_vector_type(2))) float64_t float64x2_t;
 #endif
 typedef __attribute__((neon_polyvector_type(16))) poly8_t  poly8x16_t;
 typedef __attribute__((neon_polyvector_type(8)))  poly16_t poly16x8_t;
 
+#if defined(__ARM_FEATURE_BF16)
+typedef __attribute__((neon_vector_type(4))) __bf16 bfloat16x4_t;
+#endif
+
 // CHECK: 16__simd64_int32_t
 // CHECK-AARCH64: 11__Int32x2_t
 void f1(int32x2_t v) { }
 
 // CHECK: 17__simd128_int32_t
 // CHECK-AARCH64: 11__Int32x4_t
 void f2(int32x4_t v) { }
 

 // CHECK: 18__simd128_poly16_t
 // CHECK-AARCH64: 12__Poly16x8_t
 void f10(poly16x8_t v) {}
 
 #ifdef __aarch64__
 // CHECK-AARCH64: 13__Float64x2_t
 void f11(float64x2_t v) { }
 #endif
+
+#if defined(__ARM_FEATURE_BF16)
+// CHECK-AARCH64-BF16: 14__Bfloat16x4_t
+void f12(bfloat16x4_t v) {}
+#endif