diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
@@ -714,8 +714,10 @@ ExprResult Sema::DefaultArgumentPromotion(Expr *E) {
     return ExprError();
   E = Res.get();
 
-  // If this is a 'float' or '__fp16' (CVR qualified or typedef) promote to
-  // double.
+  // If this is a 'float'  or '__fp16' (CVR qualified or typedef)
+  // promote to double.
+  // Note that default argument promotion applies only to float (and
+  // half/fp16); it does not apply to _Float16.
   const BuiltinType *BTy = Ty->getAs<BuiltinType>();
   if (BTy && (BTy->getKind() == BuiltinType::Half ||
               BTy->getKind() == BuiltinType::Float)) {
@@ -3321,6 +3323,8 @@ ExprResult Sema::ActOnNumericConstant(const Token &Tok, Scope *UDLScope) {
       Ty = Context.FloatTy;
     else if (Literal.isLong)
       Ty = Context.LongDoubleTy;
+    else if (Literal.isFloat16)
+      Ty = Context.Float16Ty;
     else if (Literal.isFloat128)
       Ty = Context.Float128Ty;
     else

diff --git a/clang/lib/Sema/SemaTemplateVariadic.cpp b/clang/lib/Sema/SemaTemplateVariadic.cpp
@@ -819,6 +819,7 @@ bool Sema::containsUnexpandedParameterPacks(Declarator &D) {
   case TST_half:
   case TST_float:
   case TST_double:
+  case TST_Float16:
   case TST_float128:
   case TST_bool:
   case TST_decimal32:

diff --git a/clang/lib/Sema/SemaType.cpp b/clang/lib/Sema/SemaType.cpp
@@ -1392,8 +1392,9 @@ static QualType ConvertDeclSpecToType(TypeProcessingState &state) {
     else
       Result = Context.Int128Ty;
     break;
-  case DeclSpec::TST_half: Result = Context.HalfTy; break;
-  case DeclSpec::TST_float: Result = Context.FloatTy; break;
+  case DeclSpec::TST_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)
       Result = Context.LongDoubleTy;

diff --git a/clang/lib/Serialization/ASTCommon.cpp b/clang/lib/Serialization/ASTCommon.cpp
@@ -91,6 +91,9 @@ serialization::TypeIdxFromBuiltin(const BuiltinType *BT) {
   case BuiltinType::LongDouble:
     ID = PREDEF_TYPE_LONGDOUBLE_ID;
     break;
+  case BuiltinType::Float16:
+    ID = PREDEF_TYPE_FLOAT16_ID;
+    break;
   case BuiltinType::Float128:
     ID = PREDEF_TYPE_FLOAT128_ID;
     break;

diff --git a/clang/lib/Serialization/ASTReader.cpp b/clang/lib/Serialization/ASTReader.cpp
@@ -6683,6 +6683,9 @@ QualType ASTReader::GetType(TypeID ID) {
     case PREDEF_TYPE_LONGDOUBLE_ID:
       T = Context.LongDoubleTy;
       break;
+    case PREDEF_TYPE_FLOAT16_ID:
+      T = Context.Float16Ty;
+      break;
     case PREDEF_TYPE_FLOAT128_ID:
       T = Context.Float128Ty;
       break;

diff --git a/clang/test/CodeGenCXX/float16-declarations.cpp b/clang/test/CodeGenCXX/float16-declarations.cpp
@@ -0,0 +1,156 @@
+// RUN: %clang -std=c++11 --target=aarch64-arm--eabi -S -emit-llvm %s -o - | FileCheck %s  --check-prefix=CHECK --check-prefix=CHECK-AARCH64
+// RUN: %clang -std=c++11 --target=x86_64 -S -emit-llvm %s -o - | FileCheck %s  --check-prefix=CHECK --check-prefix=CHECK-X86
+
+/*  Various contexts where type _Float16 can appear. */
+
+
+/*  Namespace */
+
+namespace {
+  _Float16 f1n;
+// CHECK-DAG: @_ZN12_GLOBAL__N_13f1nE = internal global half 0xH0000, align 2
+
+  _Float16 f2n = 33.f16;
+// CHECK-AARCH64-DAG: @_ZN12_GLOBAL__N_13f2nE = internal global half 0xH5020, align 2
+// CHECK-X86-DAG:     @_ZN12_GLOBAL__N_13f2nE = internal global i16 20512, align 2
+
+  _Float16 arr1n[10];
+// CHECK-AARCH64-DAG: @_ZN12_GLOBAL__N_15arr1nE = internal global [10 x half] zeroinitializer, align 2
+// CHECK-X86-DAG:     @_ZN12_GLOBAL__N_15arr1nE = internal global [10 x half] zeroinitializer, align 16
+
+  _Float16 arr2n[] = { 1.2, 3.0, 3.e4 };
+// CHECK-AARCH64-DAG: @_ZN12_GLOBAL__N_15arr2nE = internal global [3 x half] [half 0xH3CCD, half 0xH4200, half 0xH7753], align 2
+// CHECK-X86-DAG:     @_ZN12_GLOBAL__N_15arr2nE = internal global [3 x i16] [i16 15565, i16 16896, i16 30547], align 2
+
+  const volatile _Float16 func1n(const _Float16 &arg) {
+    return arg + f2n + arr1n[4] - arr2n[1];
+  }
+}
+
+
+/* File */
+
+_Float16 f1f;
+// CHECK-AARCH64-DAG: @f1f = global half 0xH0000, align 2
+// CHECK-X86-DAG: @f1f = global half 0xH0000, align 2
+
+_Float16 f2f = 32.4;
+// CHECK-AARCH64-DAG: @f2f = global half 0xH500D, align 2
+// CHECK-X86-DAG: @f2f = global i16 20493, align 2
+
+_Float16 arr1f[10];
+// CHECK-AARCH64-DAG: @arr1f = global [10 x half] zeroinitializer, align 2
+// CHECK-X86-DAG: @arr1f = global [10 x half] zeroinitializer, align 16
+
+_Float16 arr2f[] = { -1.2, -3.0, -3.e4 };
+// CHECK-AARCH64-DAG: @arr2f = global [3 x half] [half 0xHBCCD, half 0xHC200, half 0xHF753], align 2
+// CHECK-X86-DAG: @arr2f = global [3 x i16] [i16 -17203, i16 -15872, i16 -2221], align 2
+
+_Float16 func1f(_Float16 arg);
+
+
+/* Class */
+
+class C1 {
+  _Float16 f1c;
+
+  static const _Float16 f2c;
+// CHECK-DAG: @_ZN2C13f2cE = external constant half, align 2
+
+  volatile _Float16 f3c;
+
+public:
+  C1(_Float16 arg) : f1c(arg), f3c(arg) { }
+// Check that we mangle _Float16 to DF16_
+// CHECK-DAG: define linkonce_odr void @_ZN2C1C2EDF16_(%class.C1* %this, half %arg)
+
+  _Float16 func1c(_Float16 arg ) {
+    return f1c + arg;
+  }
+// CHECK-DAG: define linkonce_odr half @_ZN2C16func1cEDF16_(%class.C1* %this, half %arg)
+
+  static _Float16 func2c(_Float16 arg) {
+    return arg * C1::f2c;
+  }
+// CHECK-DAG: define linkonce_odr half @_ZN2C16func2cEDF16_(half %arg)
+};
+
+/*  Template */
+
+template <class C> C func1t(C arg) {
+  return arg * 2.f16;
+}
+// CHECK-DAG: define linkonce_odr half @_Z6func1tIDF16_ET_S0_(half %arg)
+
+template <class C> struct S1 {
+  C mem1;
+};
+
+template <> struct S1<_Float16> {
+  _Float16 mem2;
+};
+
+
+/* Local */
+
+extern int printf (const char *__restrict __format, ...);
+
+int main(void) {
+  _Float16 f1l = 1e3f16;
+// CHECK-DAG: store half 0xH63D0, half* %f1l, align 2
+
+  _Float16 f2l = -0.f16;
+// CHECK-DAG: store half 0xH8000, half* %f2l, align 2
+
+  _Float16 f3l = 1.000976562;
+// CHECK-DAG: store half 0xH3C01, half* %f3l, align 2
+
+  C1 c1(f1l);
+// CHECK-DAG:  [[F1L:%[a-z0-9]+]] = load half, half* %f1l, align 2
+// CHECK-DAG:  call void @_ZN2C1C2EDF16_(%class.C1* %c1, half [[F1L]])
+
+  S1<_Float16> s1 = { 132.f16 };
+// CHECK-AARCH64-DAG: @_ZZ4mainE2s1 = private unnamed_addr constant %struct.S1 { half 0xH5820 }, align 2
+// CHECK-X86-DAG:     @_ZZ4mainE2s1 = private unnamed_addr constant { i16 } { i16 22560 }, align 2
+// CHECK-DAG:  [[S1:%[0-9]+]] = bitcast %struct.S1* %s1 to i8*
+// CHECK-AARCH64-DAG: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[S1]], i8* bitcast (%struct.S1* @_ZZ4mainE2s1 to i8*), i64 2, i32 2, i1 false)
+// CHECK-X86-DAG:     call void @llvm.memcpy.p0i8.p0i8.i64(i8* %1, i8* bitcast ({ i16 }* @_ZZ4mainE2s1 to i8*), i64 2, i32 2, i1 false)
+
+  _Float16 f4l = func1n(f1l)  + func1f(f2l) + c1.func1c(f3l) + c1.func2c(f1l) +
+    func1t(f1l) + s1.mem2 - f1n + f2n;
+
+  auto f5l = -1.f16, *f6l = &f2l, f7l = func1t(f3l);
+// CHECK-DAG:  store half 0xHBC00, half* %f5l, align 2
+// CHECK-DAG:  store half* %f2l, half** %f6l, align 8
+
+  _Float16 f8l = f4l++;
+// CHECK-DAG:  [[F4L:%[a-z0-9]+]] = load half, half* %f4l, align 2
+// CHECK-DAG:  [[INC:%inc[0-9]*]] = fadd half [[F4L]], 0xH3C00
+// CHECK-DAG:  store half [[INC]], half* %f4l, align 2
+
+  _Float16 arr1l[] = { -1.f16, -0.f16, -11.f16 };
+// CHECK-AARCH64-DAG: @_ZZ4mainE5arr1l = private unnamed_addr constant [3 x half] [half 0xHBC00, half 0xH8000, half 0xHC980], align 2
+// CHECK-X86-DAG:     @_ZZ4mainE5arr1l = private unnamed_addr constant [3 x i16] [i16 -17408, i16 -32768, i16 -13952], align 2
+
+  float cvtf = f2n;
+//CHECK-DAG: [[H2F:%conv[0-9]*]] = fpext half {{%[0-9]+}} to float
+//CHECK-DAG:  store float [[H2F]], float* %cvtf, align 4
+
+  double cvtd = f2n;
+//CHECK-DAG: [[H2D:%conv[0-9]*]] = fpext half {{%[0-9]+}} to double
+//CHECK-DAG: store double [[H2D]], double* %cvtd, align 8
+
+
+  long double cvtld = f2n;
+//CHECK-AARCh64-DAG: [[H2LD:%conv[0-9]*]] = fpext half {{%[0-9]+}} to fp128
+//CHECK-AARCh64-DAG: store fp128 [[H2LD]], fp128* %cvtld, align 16
+//CHECK-X86-DAG:     [[H2LD:%conv[0-9]+]] = fpext half {{%[0-9]+}} to x86_fp80
+//CHECK-X86-DAG:     store x86_fp80 [[H2LD]], x86_fp80* %cvtld, align 16
+
+  _Float16 f2h = 42.0f;
+//CHECK-DAG: store half 0xH5140, half* %f2h, align 2
+  _Float16 d2h = 42.0;
+//CHECK-DAG: store half 0xH5140, half* %d2h, align 2
+  _Float16 ld2h = 42.0l;
+//CHECK-DAG:store half 0xH5140, half* %ld2h, align 2
+}
diff --git a/clang/test/Frontend/float16.cpp b/clang/test/Frontend/float16.cpp
diff --git a/clang/test/Lexer/half-literal.cpp b/clang/test/Lexer/half-literal.cpp
@@ -1,3 +1,6 @@
 // RUN: %clang_cc1 -fsyntax-only -verify -pedantic %s
 float a = 1.0h; // expected-error{{invalid suffix 'h' on floating constant}}
 float b = 1.0H; // expected-error{{invalid suffix 'H' on floating constant}}
+
+_Float16 c = 1.f166; // expected-error{{invalid suffix 'f166' on floating constant}}
+_Float16 d = 1.f1;   // expected-error{{invalid suffix 'f1' on floating constant}}
diff --git a/clang/tools/libclang/CXType.cpp b/clang/tools/libclang/CXType.cpp
@@ -53,6 +53,7 @@ static CXTypeKind GetBuiltinTypeKind(const BuiltinType *BT) {
     BTCASE(Float);
     BTCASE(Double);
     BTCASE(LongDouble);
+    BTCASE(Float16);
     BTCASE(Float128);
     BTCASE(NullPtr);
     BTCASE(Overload);
@@ -520,7 +521,7 @@ CXString clang_getTypeKindSpelling(enum CXTypeKind K) {
     TKIND(Char_U);
     TKIND(UChar);
     TKIND(Char16);
-    TKIND(Char32);  
+    TKIND(Char32);
     TKIND(UShort);
     TKIND(UInt);
     TKIND(ULong);
@@ -538,6 +539,7 @@ CXString clang_getTypeKindSpelling(enum CXTypeKind K) {
     TKIND(Float);
     TKIND(Double);
     TKIND(LongDouble);
+    TKIND(Float16);
     TKIND(Float128);
     TKIND(NullPtr);
     TKIND(Overload);