[locale.stdcvt]

git-svn-id: https://llvm.org/svn/llvm-project/libcxx/trunk@105174 91177308-0d34-0410-b5e6-96231b3b80d8

include/__locale

@@ -1047,92 +1047,6 @@ protected:
     virtual int do_max_length() const throw();
 };
 
-// template <> class codecvt<char32_t, char, mbstate_t>
-
-template <>
-class codecvt<char32_t, char16_t, mbstate_t>
-    : public locale::facet,
-      public codecvt_base
-{
-public:
-    typedef char32_t  intern_type;
-    typedef char16_t  extern_type;
-    typedef mbstate_t state_type;
-
-    _LIBCPP_ALWAYS_INLINE
-    explicit codecvt(size_t __refs = 0)
-        : locale::facet(__refs) {}
-
-    _LIBCPP_ALWAYS_INLINE
-    result out(state_type& __st,
-               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
-               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const
-    {
-        return do_out(__st, __frm, __frm_end, __frm_nxt, __to, __to_end, __to_nxt);
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    result unshift(state_type& __st,
-                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const
-    {
-        return do_unshift(__st, __to, __to_end, __to_nxt);
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    result in(state_type& __st,
-              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
-              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const
-    {
-        return do_in(__st, __frm, __frm_end, __frm_nxt, __to, __to_end, __to_nxt);
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    int encoding() const throw()
-    {
-        return do_encoding();
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    bool always_noconv() const throw()
-    {
-        return do_always_noconv();
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    int length(state_type& __st, const extern_type* __frm, const extern_type* __end, size_t __mx) const
-    {
-        return do_length(__st, __frm, __end, __mx);
-    }
-
-    _LIBCPP_ALWAYS_INLINE
-    int max_length() const throw()
-    {
-        return do_max_length();
-    }
-
-    static locale::id id;
-
-protected:
-    _LIBCPP_ALWAYS_INLINE
-    explicit codecvt(const char*, size_t __refs = 0)
-        : locale::facet(__refs) {}
-
-    ~codecvt();
-
-    virtual result do_out(state_type& __st,
-                          const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
-                          extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
-    virtual result do_in(state_type& __st,
-                         const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
-                         intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
-    virtual result do_unshift(state_type& __st,
-                              extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
-    virtual int do_encoding() const throw();
-    virtual bool do_always_noconv() const throw();
-    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end, size_t __mx) const;
-    virtual int do_max_length() const throw();
-};
-
 // template <class _InternT, class _ExternT, class _StateT> class codecvt_byname
 
 template <class _InternT, class _ExternT, class _StateT>

include/codecvt

@@ -0,0 +1,542 @@
+// -*- C++ -*-
+//===-------------------------- codecvt -----------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP_CODECVT
+#define _LIBCPP_CODECVT
+
+/*
+    codecvt synopsis
+
+namespace std
+{
+
+enum codecvt_mode
+{
+    consume_header = 4,
+    generate_header = 2,
+    little_endian = 1
+};
+
+template <class Elem, unsigned long Maxcode = 0x10ffff,
+          codecvt_mode Mode = (codecvt_mode)0>
+class codecvt_utf8
+    : public codecvt<Elem, char, mbstate_t>
+{
+    // unspecified
+};
+
+template <class Elem, unsigned long Maxcode = 0x10ffff,
+          codecvt_mode Mode = (codecvt_mode)0>
+class codecvt_utf16
+    : public codecvt<Elem, char, mbstate_t>
+{
+    // unspecified
+};
+
+template <class Elem, unsigned long Maxcode = 0x10ffff,
+          codecvt_mode Mode = (codecvt_mode)0>
+class codecvt_utf8_utf16
+    : public codecvt<Elem, char, mbstate_t>
+{
+    // unspecified
+};
+
+}  // std
+
+*/
+
+#include <__config>
+#include <__locale>
+
+#pragma GCC system_header
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+enum codecvt_mode
+{
+    consume_header = 4,
+    generate_header = 2,
+    little_endian = 1
+};
+
+// codecvt_utf8
+
+template <class _Elem> class __codecvt_utf8;
+
+template <>
+class __codecvt_utf8<wchar_t>
+    : public codecvt<wchar_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef wchar_t   intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<wchar_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf8<char16_t>
+    : public codecvt<char16_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char16_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char16_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf8<char32_t>
+    : public codecvt<char32_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char32_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char32_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <class _Elem, unsigned long _Maxcode = 0x10ffff,
+          codecvt_mode _Mode = (codecvt_mode)0>
+class codecvt_utf8
+    : public __codecvt_utf8<_Elem>
+{
+public:
+    _LIBCPP_ALWAYS_INLINE
+    explicit codecvt_utf8(size_t __refs = 0)
+        : __codecvt_utf8<_Elem>(__refs, _Maxcode, _Mode) {}
+
+    ~codecvt_utf8() {}
+};
+
+// codecvt_utf16
+
+template <class _Elem, bool _LittleEndian> class __codecvt_utf16;
+
+template <>
+class __codecvt_utf16<wchar_t, false>
+    : public codecvt<wchar_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef wchar_t   intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<wchar_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf16<wchar_t, true>
+    : public codecvt<wchar_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef wchar_t   intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<wchar_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf16<char16_t, false>
+    : public codecvt<char16_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char16_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char16_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf16<char16_t, true>
+    : public codecvt<char16_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char16_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char16_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf16<char32_t, false>
+    : public codecvt<char32_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char32_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char32_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf16<char32_t, true>
+    : public codecvt<char32_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char32_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char32_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <class _Elem, unsigned long _Maxcode = 0x10ffff,
+          codecvt_mode _Mode = (codecvt_mode)0>
+class codecvt_utf16
+    : public __codecvt_utf16<_Elem, _Mode & little_endian>
+{
+public:
+    _LIBCPP_ALWAYS_INLINE
+    explicit codecvt_utf16(size_t __refs = 0)
+        : __codecvt_utf16<_Elem, _Mode & little_endian>(__refs, _Maxcode, _Mode) {}
+
+    ~codecvt_utf16() {}
+};
+
+// codecvt_utf8_utf16
+
+template <class _Elem> class __codecvt_utf8_utf16;
+
+template <>
+class __codecvt_utf8_utf16<wchar_t>
+    : public codecvt<wchar_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef wchar_t   intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<wchar_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf8_utf16<char32_t>
+    : public codecvt<char32_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char32_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char32_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <>
+class __codecvt_utf8_utf16<char16_t>
+    : public codecvt<char16_t, char, mbstate_t>
+{
+    unsigned long _Maxcode_;
+    codecvt_mode _Mode_;
+public:
+    typedef char16_t  intern_type;
+    typedef char      extern_type;
+    typedef mbstate_t state_type;
+
+    _LIBCPP_ALWAYS_INLINE
+    explicit __codecvt_utf8_utf16(size_t __refs, unsigned long _Maxcode,
+                            codecvt_mode _Mode)
+        : codecvt<char16_t, char, mbstate_t>(__refs), _Maxcode_(_Maxcode),
+          _Mode_(_Mode) {}
+protected:
+    virtual result
+        do_out(state_type& __st,
+               const intern_type* __frm, const intern_type* __frm_end, const intern_type*& __frm_nxt,
+               extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual result
+        do_in(state_type& __st,
+              const extern_type* __frm, const extern_type* __frm_end, const extern_type*& __frm_nxt,
+              intern_type* __to, intern_type* __to_end, intern_type*& __to_nxt) const;
+    virtual result
+        do_unshift(state_type& __st,
+                   extern_type* __to, extern_type* __to_end, extern_type*& __to_nxt) const;
+    virtual int do_encoding() const throw();
+    virtual bool do_always_noconv() const throw();
+    virtual int do_length(state_type&, const extern_type* __frm, const extern_type* __end,
+                          size_t __mx) const;
+    virtual int do_max_length() const throw();
+};
+
+template <class _Elem, unsigned long _Maxcode = 0x10ffff,
+          codecvt_mode _Mode = (codecvt_mode)0>
+class codecvt_utf8_utf16
+    : public __codecvt_utf8_utf16<_Elem>
+{
+public:
+    _LIBCPP_ALWAYS_INLINE
+    explicit codecvt_utf8_utf16(size_t __refs = 0)
+        : __codecvt_utf8_utf16<_Elem>(__refs, _Maxcode, _Mode) {}
+
+    ~codecvt_utf8_utf16() {}
+};
+
+_LIBCPP_END_NAMESPACE_STD
+
+#endif  // _LIBCPP_CODECVT

test/localization/locale.categories/category.ctype/locale.codecvt/locale.codecvt.members/utf_sanity_check.pass.cpp

@@ -16,6 +16,7 @@
 // sanity check
 
 #include <locale>
+#include <codecvt>
 #include <cassert>
 
 #include <stdio.h>
@@ -24,11 +25,11 @@ int main()
 {
     typedef std::codecvt<char32_t, char, std::mbstate_t> F32_8;
     typedef std::codecvt<char16_t, char, std::mbstate_t> F16_8;
-    typedef std::codecvt<char32_t, char16_t, std::mbstate_t> F32_16;
+    typedef std::codecvt_utf16<char32_t> F32_16;
     std::locale l = std::locale(std::locale::classic(), new F32_16);
-    const F32_8& f32_8 = std::use_facet<F32_8>(l);
+    const F32_8& f32_8 = std::use_facet<F32_8>(std::locale::classic());
     const F32_16& f32_16 = std::use_facet<F32_16>(l);
-    const F16_8& f16_8 = std::use_facet<F16_8>(l);
+    const F16_8& f16_8 = std::use_facet<F16_8>(std::locale::classic());
     std::mbstate_t mbs = {0};
     F32_8::intern_type* c32p;
     F16_8::intern_type* c16p;
@@ -38,23 +39,27 @@ int main()
     const F32_8::extern_type* c_c8p;
     F32_8::intern_type c32;
     F16_8::intern_type c16[2];
+    char c16c[4];
+    char* c16cp;
     F32_8::extern_type c8[4];
     for (F32_8::intern_type c32x = 0; c32x < 0x110003; ++c32x)
     {
         if (0xD800 <= c32x && c32x < 0xE000 || c32x >= 0x110000)
         {
-            assert(f32_16.out(mbs, &c32x, &c32x+1, c_c32p, c16+0, c16+2, c16p) == F32_8::error);
+            assert(f32_16.out(mbs, &c32x, &c32x+1, c_c32p, c16c+0, c16c+4, c16cp) == F32_8::error);
             assert(f32_8.out(mbs, &c32x, &c32x+1, c_c32p, c8, c8+4, c8p) == F32_8::error);
         }
         else
         {
-            assert(f32_16.out(mbs, &c32x, &c32x+1, c_c32p, c16, c16+2, c16p) == F32_8::ok);
+            assert(f32_16.out(mbs, &c32x, &c32x+1, c_c32p, c16c, c16c+4, c16cp) == F32_8::ok);
             assert(c_c32p-&c32x == 1);
             if (c32x < 0x10000)
-                assert(c16p-c16 == 1);
+                assert(c16cp-c16c == 2);
             else
-                assert(c16p-c16 == 2);
+                assert(c16cp-c16c == 4);
-            c_c16p = c16p;
+            for (int i = 0; i < (c16cp - c16c) / 2; ++i)
+                c16[i] = (unsigned char)c16c[2*i] << 8 | (unsigned char)c16c[2*i+1];
+            c_c16p = c16 + (c16cp - c16c) / 2;
             assert(f16_8.out(mbs, c16, c_c16p, c_c16p, c8, c8+4, c8p) == F32_8::ok);
             if (c32x < 0x10000)
                 assert(c_c16p-c16 == 1);
@@ -104,12 +109,17 @@ int main()
                 assert(c16p-c16 == 1);
             else
                 assert(c16p-c16 == 2);
-            c_c16p = c16p;
+            for (int i = 0; i < c16p-c16; ++i)
-            assert(f32_16.in(mbs, c16, c_c16p, c_c16p, &c32, &c32+1, c32p) == F32_8::ok);
+            {
+                c16c[2*i] = static_cast<char>(c16[i] >> 8);
+                c16c[2*i+1] = static_cast<char>(c16[i]);
+            }
+            const char* c_c16cp = c16c + (c16p-c16)*2;
+            assert(f32_16.in(mbs, c16c, c_c16cp, c_c16cp, &c32, &c32+1, c32p) == F32_8::ok);
             if (c32x < 0x10000)
-                assert(c_c16p-c16 == 1);
+                assert(c_c16cp-c16c == 2);
             else
-                assert(c_c16p-c16 == 2);
+                assert(c_c16cp-c16c == 4);
             assert(c32p-&c32 == 1);
             assert(c32 == c32x);
         }

test/localization/locale.categories/category.time/locale.time.put/locale.time.put.members/put2.pass.cpp

@@ -33,7 +33,7 @@ int main()
     const my_facet f(1);
     char str[200];
     output_iterator<char*> iter;
-    tm t;
+    tm t = {0};
     t.tm_sec = 6;
     t.tm_min = 3;
     t.tm_hour = 13;

test/localization/locale.stdcvt/codecvt_mode.pass.cpp

@@ -0,0 +1,29 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// enum codecvt_mode
+// {
+//     consume_header = 4,
+//     generate_header = 2,
+//     little_endian = 1
+// };
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    assert(std::consume_header == 4);
+    assert(std::generate_header == 2);
+    assert(std::little_endian == 1);
+    std::codecvt_mode e = std::consume_header;
+    assert(e == 4);
+}

test/localization/locale.stdcvt/codecvt_utf16.pass.cpp

@@ -0,0 +1,56 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// Not a portable test
+
+#include <codecvt>
+#include <cassert>
+
+int outstanding_news = 0;
+
+void* operator new(std::size_t s) throw(std::bad_alloc)
+{
+    ++outstanding_news;
+    return std::malloc(s);
+}
+
+void  operator delete(void* p) throw()
+{
+    if (p)
+    {
+        --outstanding_news;
+        std::free(p);
+    }
+}
+
+int main()
+{
+    assert(outstanding_news == 0);
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        assert(outstanding_news == 0);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        std::locale loc(std::locale::classic(), new C);
+        assert(outstanding_news != 0);
+    }
+    assert(outstanding_news == 0);
+}

test/localization/locale.stdcvt/codecvt_utf16_always_noconv.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// bool always_noconv() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_encoding.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int encoding() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_in.pass.cpp

@@ -0,0 +1,739 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     in(stateT& state,
+//        const externT* from, const externT* from_end, const externT*& from_next,
+//        internT* to, internT* to_end, internT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::consume_header> C;
+        C c;
+        wchar_t w = 0;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(w == 0x40003);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::little_endian> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000, std::little_endian> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        wchar_t w = 0;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(w == 0x40003);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x1000> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char32_t w = 0;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(w == 0x40003);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::little_endian> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x1000, std::little_endian> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        char32_t w = 0;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(w == 0x40003);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        char16_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x1000> C;
+        C c;
+        char16_t w = 0;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char16_t w = 0;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n+2);
+        assert(w == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::little_endian> C;
+        C c;
+        char16_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x1000, std::little_endian> C;
+        C c;
+        char16_t w = 0;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        char16_t w = 0;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n+2);
+        assert(w == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x1005);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x56);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_length.pass.cpp

@@ -0,0 +1,449 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int length(stateT& state, const externT* from, const externT* from_end, 
+//            size_t max) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 6);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 6);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x1000> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 6);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x1000, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 6);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x1000> C;
+        C c;
+        char n[4] = {0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[6] = {0xFE, 0xFF, 0xD8, 0xC0, 0xDC, 0x03};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 2);
+
+        n[0] = char(0x10);
+        n[1] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x04);
+        n[1] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x00);
+        n[1] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x1000, std::little_endian> C;
+        C c;
+        char n[4] = {0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 0);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0x10ffff, std::codecvt_mode(
+                                                         std::consume_header |
+                                                         std::little_endian)> C;
+        C c;
+        char n[6] = {0xFF, 0xFE, 0xC0, 0xD8, 0x03, 0xDC};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+6, 2);
+        assert(r == 2);
+
+        n[1] = char(0x10);
+        n[0] = char(0x05);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x04);
+        n[0] = char(0x53);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[1] = char(0x00);
+        n[0] = char(0x56);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_max_length.pass.cpp

@@ -0,0 +1,63 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int max_length() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 6);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 2);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 6);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_out.pass.cpp

@@ -0,0 +1,331 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+// out(stateT& state,
+//     const internT* from, const internT* from_end, const internT*& from_next,
+//     externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xD8));
+        assert(n[1] == char(0xC0));
+        assert(n[2] == char(0xDC));
+        assert(n[3] == char(0x03));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0x10));
+        assert(n[1] == char(0x05));
+        assert(n[2] == char(0xDC));
+        assert(n[3] == char(0x03));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0x04));
+        assert(n[1] == char(0x53));
+        assert(n[2] == char(0xDC));
+        assert(n[3] == char(0x03));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0x00));
+        assert(n[1] == char(0x56));
+        assert(n[2] == char(0xDC));
+        assert(n[3] == char(0x03));
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0x04));
+        assert(n[1] == char(0x53));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0x00));
+        assert(n[1] == char(0x56));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::generate_header> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[6] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xFE));
+        assert(n[1] == char(0xFF));
+        assert(n[2] == char(0xD8));
+        assert(n[3] == char(0xC0));
+        assert(n[4] == char(0xDC));
+        assert(n[5] == char(0x03));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xFE));
+        assert(n[1] == char(0xFF));
+        assert(n[2] == char(0x10));
+        assert(n[3] == char(0x05));
+        assert(n[4] == char(0xDC));
+        assert(n[5] == char(0x03));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xFE));
+        assert(n[1] == char(0xFF));
+        assert(n[2] == char(0x04));
+        assert(n[3] == char(0x53));
+        assert(n[4] == char(0xDC));
+        assert(n[5] == char(0x03));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xFE));
+        assert(n[1] == char(0xFF));
+        assert(n[2] == char(0x00));
+        assert(n[3] == char(0x56));
+        assert(n[4] == char(0xDC));
+        assert(n[5] == char(0x03));
+    }
+
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10FFFF, std::little_endian> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[1] == char(0xD8));
+        assert(n[0] == char(0xC0));
+        assert(n[3] == char(0xDC));
+        assert(n[2] == char(0x03));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[1] == char(0x10));
+        assert(n[0] == char(0x05));
+        assert(n[3] == char(0xDC));
+        assert(n[2] == char(0x03));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[1] == char(0x04));
+        assert(n[0] == char(0x53));
+        assert(n[3] == char(0xDC));
+        assert(n[2] == char(0x03));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[1] == char(0x00));
+        assert(n[0] == char(0x56));
+        assert(n[3] == char(0xDC));
+        assert(n[2] == char(0x03));
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x1000, std::little_endian> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[1] == char(0));
+        assert(n[0] == char(0));
+        assert(n[3] == char(0));
+        assert(n[2] == char(0));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[1] == char(0));
+        assert(n[0] == char(0));
+        assert(n[3] == char(0));
+        assert(n[2] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[1] == char(0x04));
+        assert(n[0] == char(0x53));
+        assert(n[3] == char(0));
+        assert(n[2] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[1] == char(0x00));
+        assert(n[0] == char(0x56));
+        assert(n[3] == char(0));
+        assert(n[2] == char(0));
+    }
+    {
+        typedef std::codecvt_utf16<wchar_t, 0x10ffff, std::codecvt_mode(
+                                                         std::generate_header |
+                                                         std::little_endian)> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[6] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(n[1] == char(0xFE));
+        assert(n[0] == char(0xFF));
+        assert(n[3] == char(0xD8));
+        assert(n[2] == char(0xC0));
+        assert(n[5] == char(0xDC));
+        assert(n[4] == char(0x03));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[1] == char(0xFE));
+        assert(n[0] == char(0xFF));
+        assert(n[3] == char(0x10));
+        assert(n[2] == char(0x05));
+        assert(n[5] == char(0xDC));
+        assert(n[4] == char(0x03));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[1] == char(0xFE));
+        assert(n[0] == char(0xFF));
+        assert(n[3] == char(0x04));
+        assert(n[2] == char(0x53));
+        assert(n[5] == char(0xDC));
+        assert(n[4] == char(0x03));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+6, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[1] == char(0xFE));
+        assert(n[0] == char(0xFF));
+        assert(n[3] == char(0x00));
+        assert(n[2] == char(0x56));
+        assert(n[5] == char(0xDC));
+        assert(n[4] == char(0x03));
+    }
+}

test/localization/locale.stdcvt/codecvt_utf16_unshift.pass.cpp

@@ -0,0 +1,56 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     unshift(stateT& state, 
+//             externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf16<wchar_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf16<char16_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf16<char32_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8.pass.cpp

@@ -0,0 +1,56 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// Not a portable test
+
+#include <codecvt>
+#include <cassert>
+
+int outstanding_news = 0;
+
+void* operator new(std::size_t s) throw(std::bad_alloc)
+{
+    ++outstanding_news;
+    return std::malloc(s);
+}
+
+void  operator delete(void* p) throw()
+{
+    if (p)
+    {
+        --outstanding_news;
+        std::free(p);
+    }
+}
+
+int main()
+{
+    assert(outstanding_news == 0);
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        assert(outstanding_news == 0);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        std::locale loc(std::locale::classic(), new C);
+        assert(outstanding_news != 0);
+    }
+    assert(outstanding_news == 0);
+}

test/localization/locale.stdcvt/codecvt_utf8_always_noconv.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// bool always_noconv() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_encoding.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int encoding() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_in.pass.cpp

@@ -0,0 +1,360 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     in(stateT& state,
+//        const externT* from, const externT* from_end, const externT*& from_next,
+//        internT* to, internT* to_end, internT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(w == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w = 0;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        wchar_t w = 0;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+7, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+7);
+        assert(w == 0x40003);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(w == 0x1005);
+
+        n[0] = char(0xEF);
+        n[1] = char(0xBB);
+        n[2] = char(0xBF);
+        n[3] = char(0xD1);
+        n[4] = char(0x93);
+        r = c.in(m, n, n+5, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+5);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(w == 0x40003);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(w == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0x1000> C;
+        C c;
+        char32_t w = 0;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        char32_t w = 0;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+7, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+7);
+        assert(w == 0x40003);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(w == 0x1005);
+
+        n[0] = char(0xEF);
+        n[1] = char(0xBB);
+        n[2] = char(0xBF);
+        n[3] = char(0xD1);
+        n[4] = char(0x93);
+        r = c.in(m, n, n+5, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+5);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        char16_t w = 0;
+        char n[3] = {0xE1, 0x80, 0x85};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(w == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0x1000> C;
+        C c;
+        char16_t w = 0;
+        char n[3] = {0xE1, 0x80, 0x85};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+3, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(w == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        char16_t w = 0;
+        char n[6] = {0xEF, 0xBB, 0xBF, 0xE1, 0x80, 0x85};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+6, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(w == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(w == 0x453);
+
+        w = 0x56;
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, &w, &w+1, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(w == 0x56);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_length.pass.cpp

@@ -0,0 +1,244 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int length(stateT& state, const externT* from, const externT* from_end, 
+//            size_t max) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 4);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 1);
+        assert(r == 7);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xEF);
+        n[1] = char(0xBB);
+        n[2] = char(0xBF);
+        n[3] = char(0xD1);
+        n[4] = char(0x93);
+        r = c.length(m, n, n+5, 3);
+        assert(r == 5);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 4);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 1);
+        assert(r == 7);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xEF);
+        n[1] = char(0xBB);
+        n[2] = char(0xBF);
+        n[3] = char(0xD1);
+        n[4] = char(0x93);
+        r = c.length(m, n, n+5, 3);
+        assert(r == 5);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 1);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 3);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 1);
+        assert(r == 3);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xEF);
+        n[1] = char(0xBB);
+        n[2] = char(0xBF);
+        n[3] = char(0xD1);
+        n[4] = char(0x93);
+        r = c.length(m, n, n+5, 3);
+        assert(r == 5);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 3);
+        assert(r == 1);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_max_length.pass.cpp

@@ -0,0 +1,63 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int max_length() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 7);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 3);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 6);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 7);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_out.pass.cpp

@@ -0,0 +1,456 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+// out(stateT& state,
+//     const internT* from, const internT* from_end, const internT*& from_next,
+//     externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xF1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x80));
+        assert(n[3] == char(0x83));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+    }
+    {
+        typedef std::codecvt_utf8<wchar_t, 0xFFFFFFFF, std::generate_header> C;
+        C c;
+        wchar_t w = 0x40003;
+        char n[7] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+7);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xF1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x80));
+        assert(n[6] == char(0x83));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        char32_t w = 0x40003;
+        char n[4] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xF1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x80));
+        assert(n[3] == char(0x83));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0x83));
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0x1000> C;
+        C c;
+        char32_t w = 0x40003;
+        char n[4] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+    }
+    {
+        typedef std::codecvt_utf8<char32_t, 0xFFFFFFFF, std::generate_header> C;
+        C c;
+        char32_t w = 0x40003;
+        char n[7] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+7);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xF1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x80));
+        assert(n[6] == char(0x83));
+
+        w = 0x1005;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0x83));
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        char16_t w = 0x1005;
+        char n[4] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0x85));
+        assert(n[3] == char(0));
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0x1000> C;
+        C c;
+        char16_t w = 0x1005;
+        char n[4] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == &w);
+        assert(np == n);
+        assert(n[0] == char(0));
+        assert(n[1] == char(0));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+        assert(n[1] == char(0x93));
+        assert(n[2] == char(0));
+        assert(n[3] == char(0));
+    }
+    {
+        typedef std::codecvt_utf8<char16_t, 0xFFFFFFFF, std::generate_header> C;
+        C c;
+        char16_t w = 0x1005;
+        char n[7] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0));
+
+        w = 0x453;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0));
+
+        w = 0x56;
+        r = c.out(m, &w, &w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == &w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+        assert(n[4] == char(0x93));
+        assert(n[5] == char(0x85));
+        assert(n[6] == char(0));
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_unshift.pass.cpp

@@ -0,0 +1,56 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     unshift(stateT& state, 
+//             externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8<wchar_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf8<char16_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf8<char32_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_always_noconv.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// bool always_noconv() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        bool r = c.always_noconv();
+        assert(r == false);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_encoding.pass.cpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int encoding() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        int r = c.encoding();
+        assert(r == 0);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_in.pass.cpp

@@ -0,0 +1,372 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     in(stateT& state,
+//        const externT* from, const externT* from_end, const externT*& from_next,
+//        internT* to, internT* to_end, internT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        wchar_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x10ffff, std::consume_header> C;
+        C c;
+        wchar_t w[2] = {0};
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+7, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        char32_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x1000> C;
+        C c;
+        char32_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char32_t w[2] = {0};
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        char32_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+7, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        char16_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x1000> C;
+        C c;
+        char16_t w[2] = {0};
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+4, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char16_t w[2] = {0};
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        char16_t* wp = nullptr;
+        std::mbstate_t m;
+        const char* np = nullptr;
+        std::codecvt_base::result r = c.in(m, n, n+7, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(w[0] == 0xD8C0);
+        assert(w[1] == 0xDC03);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.in(m, n, n+3, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(w[0] == 0x1005);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.in(m, n, n+2, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(w[0] == 0x0453);
+
+        n[0] = char(0x56);
+        r = c.in(m, n, n+1, np, w, w+2, wp);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(w[0] == 0x0056);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_length.pass.cpp

@@ -0,0 +1,235 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int length(stateT& state, const externT* from, const externT* from_end, 
+//            size_t max) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 2);
+        assert(r == 7);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 2);
+        assert(r == 7);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 4);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x1000> C;
+        C c;
+        char n[4] = {0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+4, 2);
+        assert(r == 0);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 0);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x10ffff, std::consume_header> C;
+        C c;
+        char n[7] = {0xEF, 0xBB, 0xBF, 0xF1, 0x80, 0x80, 0x83};
+        std::mbstate_t m;
+        int r = c.length(m, n, n+7, 2);
+        assert(r == 7);
+
+        n[0] = char(0xE1);
+        n[1] = char(0x80);
+        n[2] = char(0x85);
+        r = c.length(m, n, n+3, 2);
+        assert(r == 3);
+
+        n[0] = char(0xD1);
+        n[1] = char(0x93);
+        r = c.length(m, n, n+2, 2);
+        assert(r == 2);
+
+        n[0] = char(0x56);
+        r = c.length(m, n, n+1, 2);
+        assert(r == 1);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_max_length.pass.cpp

@@ -0,0 +1,63 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// int max_length() const throw();
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 7);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 7);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 4);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0xFFFFFFFF, std::consume_header> C;
+        C c;
+        int r = c.max_length();
+        assert(r == 7);
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_out.pass.cpp

@@ -0,0 +1,415 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+// out(stateT& state,
+//     const internT* from, const internT* from_end, const internT*& from_next,
+//     externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        wchar_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(n[0] == char(0xF1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x80));
+        assert(n[3] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x1000> C;
+        C c;
+        wchar_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t, 0x10ffff, std::generate_header> C;
+        C c;
+        wchar_t w[2] = {0xD8C0, 0xDC03};
+        char n[7] = {0};
+        const wchar_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xF1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x80));
+        assert(n[6] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        char32_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(n[0] == char(0xF1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x80));
+        assert(n[3] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x1000> C;
+        C c;
+        char32_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t, 0x10ffff, std::generate_header> C;
+        C c;
+        char32_t w[2] = {0xD8C0, 0xDC03};
+        char n[7] = {0};
+        const char32_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xF1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x80));
+        assert(n[6] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+    }
+
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        char16_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+4);
+        assert(n[0] == char(0xF1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x80));
+        assert(n[3] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+3);
+        assert(n[0] == char(0xE1));
+        assert(n[1] == char(0x80));
+        assert(n[2] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x1000> C;
+        C c;
+        char16_t w[2] = {0xD8C0, 0xDC03};
+        char n[4] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::error);
+        assert(wp == w);
+        assert(np == n);
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+2);
+        assert(n[0] == char(0xD1));
+        assert(n[1] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+4, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+1);
+        assert(n[0] == char(0x56));
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t, 0x10ffff, std::generate_header> C;
+        C c;
+        char16_t w[2] = {0xD8C0, 0xDC03};
+        char n[7] = {0};
+        const char16_t* wp = nullptr;
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.out(m, w, w+2, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+2);
+        assert(np == n+7);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xF1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x80));
+        assert(n[6] == char(0x83));
+
+        w[0] = 0x1005;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+6);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xE1));
+        assert(n[4] == char(0x80));
+        assert(n[5] == char(0x85));
+
+        w[0] = 0x453;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+5);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0xD1));
+        assert(n[4] == char(0x93));
+
+        w[0] = 0x56;
+        r = c.out(m, w, w+1, wp, n, n+7, np);
+        assert(r == std::codecvt_base::ok);
+        assert(wp == w+1);
+        assert(np == n+4);
+        assert(n[0] == char(0xEF));
+        assert(n[1] == char(0xBB));
+        assert(n[2] == char(0xBF));
+        assert(n[3] == char(0x56));
+    }
+}

test/localization/locale.stdcvt/codecvt_utf8_utf16_unshift.pass.cpp

@@ -0,0 +1,56 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+// template <class Elem, unsigned long Maxcode = 0x10ffff,
+//           codecvt_mode Mode = (codecvt_mode)0>
+// class codecvt_utf8_utf16
+//     : public codecvt<Elem, char, mbstate_t>
+// {
+//     // unspecified
+// };
+
+// result
+//     unshift(stateT& state, 
+//             externT* to, externT* to_end, externT*& to_next) const;
+
+#include <codecvt>
+#include <cassert>
+
+int main()
+{
+    {
+        typedef std::codecvt_utf8_utf16<wchar_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char16_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+    {
+        typedef std::codecvt_utf8_utf16<char32_t> C;
+        C c;
+        char n[4] = {0};
+        std::mbstate_t m;
+        char* np = nullptr;
+        std::codecvt_base::result r = c.unshift(m, n, n+4, np);
+        assert(r == std::codecvt_base::noconv);
+    }
+}

test/localization/locale.stdcvt/version.pass.cpp

@@ -0,0 +1,20 @@
+//===----------------------------------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// <codecvt>
+
+#include <codecvt>
+
+#ifndef _LIBCPP_VERSION
+#error _LIBCPP_VERSION not defined
+#endif
+
+int main()
+{
+}