Index: clang/include/clang/Sema/Sema.h
===================================================================
--- clang/include/clang/Sema/Sema.h
+++ clang/include/clang/Sema/Sema.h
@@ -9364,6 +9364,7 @@
   bool inferObjCARCLifetime(ValueDecl *decl);
 
   void deduceOpenCLAddressSpace(ValueDecl *decl);
+  QualType deduceOpenCLPointeeAddrSpace(QualType PointeeType);
 
   ExprResult
   HandleExprPropertyRefExpr(const ObjCObjectPointerType *OPT,
Index: clang/lib/Sema/SemaTemplateDeduction.cpp
===================================================================
--- clang/lib/Sema/SemaTemplateDeduction.cpp
+++ clang/lib/Sema/SemaTemplateDeduction.cpp
@@ -3806,8 +3806,11 @@
     //   If P is a forwarding reference and the argument is an lvalue, the type
     //   "lvalue reference to A" is used in place of A for type deduction.
     if (isForwardingReference(QualType(ParamRefType, 0), FirstInnerIndex) &&
-        Arg->isLValue())
+        Arg->isLValue()) {
+      if(S.getLangOpts().OpenCL)
+        ArgType = S.deduceOpenCLPointeeAddrSpace(ArgType);
       ArgType = S.Context.getLValueReferenceType(ArgType);
+    }
   } else {
     // C++ [temp.deduct.call]p2:
     //   If P is not a reference type:
Index: clang/lib/Sema/SemaType.cpp
===================================================================
--- clang/lib/Sema/SemaType.cpp
+++ clang/lib/Sema/SemaType.cpp
@@ -2036,13 +2036,13 @@
 }
 
 // Helper to deduce addr space of a pointee type in OpenCL mode.
-static QualType deduceOpenCLPointeeAddrSpace(Sema &S, QualType PointeeType) {
+QualType Sema::deduceOpenCLPointeeAddrSpace(QualType PointeeType) {
   if (!PointeeType->isUndeducedAutoType() && !PointeeType->isDependentType() &&
       !PointeeType->isSamplerT() &&
       !PointeeType.hasAddressSpace())
-    PointeeType = S.getASTContext().getAddrSpaceQualType(
+    PointeeType = getASTContext().getAddrSpaceQualType(
         PointeeType,
-        S.getLangOpts().OpenCLCPlusPlus || S.getLangOpts().OpenCLVersion == 200
+        getLangOpts().OpenCLCPlusPlus || getLangOpts().OpenCLVersion == 200
             ? LangAS::opencl_generic
             : LangAS::opencl_private);
   return PointeeType;
@@ -2085,7 +2085,7 @@
     T = inferARCLifetimeForPointee(*this, T, Loc, /*reference*/ false);
 
   if (getLangOpts().OpenCL)
-    T = deduceOpenCLPointeeAddrSpace(*this, T);
+    T = deduceOpenCLPointeeAddrSpace(T);
 
   // Build the pointer type.
   return Context.getPointerType(T);
@@ -2148,7 +2148,7 @@
     T = inferARCLifetimeForPointee(*this, T, Loc, /*reference*/ true);
 
   if (getLangOpts().OpenCL)
-    T = deduceOpenCLPointeeAddrSpace(*this, T);
+    T = deduceOpenCLPointeeAddrSpace(T);
 
   // Handle restrict on references.
   if (LValueRef)
@@ -2862,7 +2862,7 @@
     return QualType();
 
   if (getLangOpts().OpenCL)
-    T = deduceOpenCLPointeeAddrSpace(*this, T);
+    T = deduceOpenCLPointeeAddrSpace(T);
 
   return Context.getBlockPointerType(T);
 }
Index: clang/test/SemaOpenCLCXX/address-space-templates.cl
===================================================================
--- clang/test/SemaOpenCLCXX/address-space-templates.cl
+++ clang/test/SemaOpenCLCXX/address-space-templates.cl
@@ -22,10 +22,28 @@
   __private T ii; // expected-error{{conflicting address space qualifiers are provided between types '__private T' and '__global int'}}
 }
 
+template <class _Tp> struct remove_reference         { typedef _Tp type; };
+template <class _Tp> struct remove_reference<_Tp &>  { typedef _Tp type; };
+template <class _Tp> struct as_pointer {
+    typedef typename remove_reference<_Tp>::type* type;
+};
+
+struct rep {
+	// CHECK |-CXXConstructorDecl {{.*}} rep 'void (const __generic rep &__private) __generic'
+	template<class U, class = typename as_pointer<U>::type>
+	rep(U&& v) {}
+};
+
+struct rep_outer : private rep {
+	rep_outer()
+		: rep(0) {}
+};
+
 void bar() {
   S<const __global int> sintgl; // expected-note{{in instantiation of template class 'S<const __global int>' requested here}}
 
   foo1<__local int>(1); // expected-error{{no matching function for call to 'foo1'}}
   foo2<__global int>(0);
   foo3<__global int>(); // expected-note{{in instantiation of function template specialization 'foo3<__global int>' requested here}}
+  rep_outer r;
 }