Vendor import of clang trunk r306325:vendor/clang/clang-trunk-r306325

https://llvm.org/svn/llvm-project/cfe/trunk@306325

201 files changed, 7904 insertions, 2165 deletions

diff --git a/docs/Block-ABI-Apple.rst b/docs/Block-ABI-Apple.rst
index 628e6f3d90ba..7f49bbd40d71 100644
--- a/docs/Block-ABI-Apple.rst
+++ b/docs/Block-ABI-Apple.rst
@@ -856,15 +856,15 @@ mentioned above, call:
 
 .. code-block:: c
 
-     _Block_object_assign(&dst->target, src->target, BLOCK_FIELD_<appropo>);
+     _Block_object_assign(&dst->target, src->target, BLOCK_FIELD_<apropos>);
 
 in the copy helper and:
 
 .. code-block:: c
 
-    _Block_object_dispose(->target, BLOCK_FIELD_<appropo>);
+    _Block_object_dispose(->target, BLOCK_FIELD_<apropos>);
 
-in the dispose helper where ``<appropo>`` is:
+in the dispose helper where ``<apropos>`` is:
 
 .. code-block:: c
 
@@ -888,7 +888,7 @@ and functions are generated in the same manner.
 Under ObjC we allow ``__weak`` as an attribute on ``__block`` variables, and
 this causes the addition of ``BLOCK_FIELD_IS_WEAK`` orred onto the
 ``BLOCK_FIELD_IS_BYREF`` flag when copying the ``block_byref`` structure in the
-``Block`` copy helper, and onto the ``BLOCK_FIELD_<appropo>`` field within the
+``Block`` copy helper, and onto the ``BLOCK_FIELD_<apropos>`` field within the
 ``block_byref`` copy/dispose helper calls.
 
 The prototypes, and summary, of the helper functions are:
diff --git a/docs/ClangFormat.rst b/docs/ClangFormat.rst
index ed0e58e452d3..902afcd08ed5 100644
--- a/docs/ClangFormat.rst
+++ b/docs/ClangFormat.rst
@@ -120,6 +120,18 @@ entity.
 It operates on the current, potentially unsaved buffer and does not create
 or save any files. To revert a formatting, just undo.
 
+An alternative option is to format changes when saving a file and thus to
+have a zero-effort integration into the coding workflow. To do this, add this to
+your `.vimrc`:
+
+.. code-block:: vim
+
+  function! Formatonsave()
+    let l:formatdiff = 1
+    pyf ~/llvm/tools/clang/tools/clang-format/clang-format.py
+  endfunction
+  autocmd BufWritePre *.h,*.cc,*.cpp call Formatonsave()
+
 
 Emacs Integration
 =================
diff --git a/docs/ClangFormatStyleOptions.rst b/docs/ClangFormatStyleOptions.rst
index fb014241809c..6133ca9900c9 100644
--- a/docs/ClangFormatStyleOptions.rst
+++ b/docs/ClangFormatStyleOptions.rst
@@ -309,12 +309,28 @@ the configuration (without a prefix: ``Auto``).
   * ``SFS_None`` (in configuration: ``None``)
     Never merge functions into a single line.
 
+  * ``SFS_InlineOnly`` (in configuration: ``InlineOnly``)
+    Only merge functions defined inside a class. Same as "inline",
+    except it does not implies "empty": i.e. top level empty functions
+    are not merged either.
+
+    .. code-block:: c++
+
+      class Foo {
+        void f() { foo(); }
+      };
+      void f() {
+        foo();
+      }
+      void f() {
+      }
+
   * ``SFS_Empty`` (in configuration: ``Empty``)
     Only merge empty functions.
 
     .. code-block:: c++
 
-      void f() { bar(); }
+      void f() {}
       void f2() {
         bar2();
       }
@@ -327,6 +343,10 @@ the configuration (without a prefix: ``Auto``).
       class Foo {
         void f() { foo(); }
       };
+      void f() {
+        foo();
+      }
+      void f() {}
 
   * ``SFS_All`` (in configuration: ``All``)
     Merge all functions fitting on a single line.
@@ -518,147 +538,159 @@ the configuration (without a prefix: ``Auto``).
 
   * ``bool AfterClass`` Wrap class definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    class foo
-    {};
+      true:
+      class foo {};
 
-    false:
-    class foo {};
+      false:
+      class foo
+      {};
 
   * ``bool AfterControlStatement`` Wrap control statements (``if``/``for``/``while``/``switch``/..).
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    if (foo())
-    {
-    } else
-    {}
-    for (int i = 0; i < 10; ++i)
-    {}
+      true:
+      if (foo())
+      {
+      } else
+      {}
+      for (int i = 0; i < 10; ++i)
+      {}
 
-    false:
-    if (foo()) {
-    } else {
-    }
-    for (int i = 0; i < 10; ++i) {
-    }
+      false:
+      if (foo()) {
+      } else {
+      }
+      for (int i = 0; i < 10; ++i) {
+      }
 
   * ``bool AfterEnum`` Wrap enum definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    enum X : int
-    {
-      B
-    };
+      true:
+      enum X : int
+      {
+        B
+      };
 
-    false:
-    enum X : int { B };
+      false:
+      enum X : int { B };
 
   * ``bool AfterFunction`` Wrap function definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    void foo()
-    {
-      bar();
-      bar2();
-    }
+      true:
+      void foo()
+      {
+        bar();
+        bar2();
+      }
 
-    false:
-    void foo() {
-      bar();
-      bar2();
-    }
+      false:
+      void foo() {
+        bar();
+        bar2();
+      }
 
   * ``bool AfterNamespace`` Wrap namespace definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    namespace
-    {
-    int foo();
-    int bar();
-    }
+      true:
+      namespace
+      {
+      int foo();
+      int bar();
+      }
 
-    false:
-    namespace {
-    int foo();
-    int bar();
-    }
+      false:
+      namespace {
+      int foo();
+      int bar();
+      }
 
   * ``bool AfterObjCDeclaration`` Wrap ObjC definitions (``@autoreleasepool``, interfaces, ..).
 
   * ``bool AfterStruct`` Wrap struct definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    struct foo
-    {
-      int x;
-    };
+      true:
+      struct foo
+      {
+        int x;
+      };
 
-    false:
-    struct foo {
-      int x;
-    };
+      false:
+      struct foo {
+        int x;
+      };
 
   * ``bool AfterUnion`` Wrap union definitions.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    union foo
-    {
-      int x;
-    }
+      true:
+      union foo
+      {
+        int x;
+      }
 
-    false:
-    union foo {
-      int x;
-    }
+      false:
+      union foo {
+        int x;
+      }
 
   * ``bool BeforeCatch`` Wrap before ``catch``.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    try {
-      foo();
-    }
-    catch () {
-    }
+      true:
+      try {
+        foo();
+      }
+      catch () {
+      }
 
-    false:
-    try {
-      foo();
-    } catch () {
-    }
+      false:
+      try {
+        foo();
+      } catch () {
+      }
 
   * ``bool BeforeElse`` Wrap before ``else``.
 
-  .. code-block:: c++
+    .. code-block:: c++
 
-    true:
-    if (foo()) {
-    }
-    else {
-    }
+      true:
+      if (foo()) {
+      }
+      else {
+      }
 
-    false:
-    if (foo()) {
-    } else {
-    }
+      false:
+      if (foo()) {
+      } else {
+      }
 
   * ``bool IndentBraces`` Indent the wrapped braces themselves.
 
+  * ``bool SplitEmptyFunctionBody`` If ``false``, empty function body can be put on a single line.
+    This option is used only if the opening brace of the function has
+    already been wrapped, i.e. the `AfterFunction` brace wrapping mode is
+    set, and the function could/should not be put on a single line (as per
+    `AllowShortFunctionsOnASingleLine` and constructor formatting options).
+
+    .. code-block:: c++
+
+      int f()   vs.   inf f()
+      {}              {
+                      }
+
 
 **BreakAfterJavaFieldAnnotations** (``bool``)
   Break after each annotation on a field in Java files.
@@ -899,17 +931,40 @@ the configuration (without a prefix: ``Auto``).
          firstValue :
          SecondValueVeryVeryVeryVeryLong;
 
-**BreakConstructorInitializersBeforeComma** (``bool``)
-  Always break constructor initializers before commas and align
-  the commas with the colon.
+**BreakConstructorInitializers** (``BreakConstructorInitializersStyle``)
+  The constructor initializers style to use.
+
+  Possible values:
+
+  * ``BCIS_BeforeColon`` (in configuration: ``BeforeColon``)
+    Break constructor initializers before the colon and after the commas.
+
+    .. code-block:: c++
+
+    Constructor()
+        : initializer1(),
+          initializer2()
+
+  * ``BCIS_BeforeComma`` (in configuration: ``BeforeComma``)
+    Break constructor initializers before the colon and commas, and align
+    the commas with the colon.
+
+    .. code-block:: c++
+
+    Constructor()
+        : initializer1()
+        , initializer2()
+
+  * ``BCIS_AfterColon`` (in configuration: ``AfterColon``)
+    Break constructor initializers after the colon and commas.
+
+    .. code-block:: c++
+
+    Constructor() :
+        initializer1(),
+        initializer2()
 
-  .. code-block:: c++
 
-     true:                                  false:
-     SomeClass::Constructor()       vs.     SomeClass::Constructor() : a(a),
-         : a(a)                                                   b(b),
-         , b(b)                                                   c(c) {}
-         , c(c) {}
 
 **BreakStringLiterals** (``bool``)
   Allow breaking string literals when formatting.
@@ -931,6 +986,31 @@ the configuration (without a prefix: ``Auto``).
      // Will leave the following line unaffected
      #include <vector> // FOOBAR pragma: keep
 
+**CompactNamespaces** (``bool``)
+  If ``true``, consecutive namespace declarations will be on the same
+  line. If ``false``, each namespace is declared on a new line.
+
+  .. code-block:: c++
+
+    true:
+    namespace Foo { namespace Bar {
+    }}
+
+    false:
+    namespace Foo {
+    namespace Bar {
+    }
+    }
+
+  If it does not fit on a single line, the overflowing namespaces get
+  wrapped:
+
+  .. code-block:: c++
+
+    namespace Foo { namespace Bar {
+    namespace Extra {
+    }}}
+
 **ConstructorInitializerAllOnOneLineOrOnePerLine** (``bool``)
   If the constructor initializers don't fit on a line, put each
   initializer on its own line.
@@ -1321,6 +1401,9 @@ the configuration (without a prefix: ``Auto``).
   Add a space in front of an Objective-C protocol list, i.e. use
   ``Foo <Protocol>`` instead of ``Foo<Protocol>``.
 
+**PenaltyBreakAssignment** (``unsigned``)
+  The penalty for breaking around an assignment operator.
+
 **PenaltyBreakBeforeFirstCallParameter** (``unsigned``)
   The penalty for breaking a function call after ``call(``.
 
@@ -1392,6 +1475,15 @@ the configuration (without a prefix: ``Auto``).
      #include "b.h"                 vs.     #include "a.h"
      #include "a.h"                         #include "b.h"
 
+**SortUsingDeclarations** (``bool``)
+  If ``true``, clang-format will sort using declarations.
+
+  .. code-block:: c++
+
+     false:                                 true:
+     using std::cout;               vs.     using std::cin;
+     using std::cin;                        using std::cout;
+
 **SpaceAfterCStyleCast** (``bool``)
   If ``true``, a space is inserted after C style casts.
 
diff --git a/docs/LibFormat.rst b/docs/LibFormat.rst
index eacdc1614567..086a52827d8c 100644
--- a/docs/LibFormat.rst
+++ b/docs/LibFormat.rst
@@ -28,7 +28,9 @@ The core routine of LibFormat is ``reformat()``:
 
 This reads a token stream out of the lexer ``Lex`` and reformats all the code
 ranges in ``Ranges``. The ``FormatStyle`` controls basic decisions made during
-formatting. A list of options can be found under :ref:`style-options`. 
+formatting. A list of options can be found under :ref:`style-options`.
+
+The style options are described in :doc:`ClangFormatStyleOptions`.
 
 
 .. _style-options:
diff --git a/docs/MemorySanitizer.rst b/docs/MemorySanitizer.rst
index 8088ecdf561f..5bb19ed8a509 100644
--- a/docs/MemorySanitizer.rst
+++ b/docs/MemorySanitizer.rst
@@ -27,7 +27,7 @@ executable, so make sure to use ``clang`` (not ``ld``) for the final
 link step. When linking shared libraries, the MemorySanitizer run-time
 is not linked, so ``-Wl,-z,defs`` may cause link errors (don't use it
 with MemorySanitizer). To get a reasonable performance add ``-O1`` or
-higher. To get meaninful stack traces in error messages add
+higher. To get meaningful stack traces in error messages add
 ``-fno-omit-frame-pointer``. To get perfect stack traces you may need
 to disable inlining (just use ``-O1``) and tail call elimination
 (``-fno-optimize-sibling-calls``).
diff --git a/docs/SourceBasedCodeCoverage.rst b/docs/SourceBasedCodeCoverage.rst
index 474af30ae30f..805c98794804 100644
--- a/docs/SourceBasedCodeCoverage.rst
+++ b/docs/SourceBasedCodeCoverage.rst
@@ -274,6 +274,11 @@ To specify an alternate directory for raw profiles, use
 Drawbacks and limitations
 =========================
 
+* Prior to version 2.26, the GNU binutils BFD linker is not able link programs
+  compiled with ``-fcoverage-mapping`` in its ``--gc-sections`` mode.  Possible
+  workarounds include disabling ``--gc-sections``, upgrading to a newer version
+  of BFD, or using the Gold linker.
+
 * Code coverage does not handle unpredictable changes in control flow or stack
   unwinding in the presence of exceptions precisely. Consider the following
   function:
diff --git a/docs/ThinLTO.rst b/docs/ThinLTO.rst
index d417febda502..31fff51a61e9 100644
--- a/docs/ThinLTO.rst
+++ b/docs/ThinLTO.rst
@@ -126,6 +126,50 @@ which currently must be enabled through a linker option.
 - lld (as of LLVM r296702):
   ``-Wl,--thinlto-cache-dir=/path/to/cache``
 
+Cache Pruning
+-------------
+
+To help keep the size of the cache under control, ThinLTO supports cache
+pruning. Cache pruning is supported with ld64 and ELF lld, but currently only
+ELF lld allows you to control the policy with a policy string.  The cache
+policy must be specified with a linker option.
+
+- ELF lld (as of LLVM r298036):
+  ``-Wl,--thinlto-cache-policy,POLICY``
+
+A policy string is a series of key-value pairs separated by ``:`` characters.
+Possible key-value pairs are:
+
+- ``cache_size=X%``: The maximum size for the cache directory is ``X`` percent
+  of the available space on the the disk. Set to 100 to indicate no limit,
+  50 to indicate that the cache size will not be left over half the available
+  disk space. A value over 100 is invalid. A value of 0 disables the percentage
+  size-based pruning. The default is 75%.
+
+- ``cache_size_bytes=X``, ``cache_size_bytes=Xk``, ``cache_size_bytes=Xm``,
+  ``cache_size_bytes=Xg``:
+  Sets the maximum size for the cache directory to ``X`` bytes (or KB, MB,
+  GB respectively). A value over the amount of available space on the disk
+  will be reduced to the amount of available space. A value of 0 disables
+  the byte size-based pruning. The default is no byte size-based pruning.
+
+  Note that ThinLTO will apply both size-based pruning policies simultaneously,
+  and changing one does not affect the other. For example, a policy of
+  ``cache_size_bytes=1g`` on its own will cause both the 1GB and default 75%
+  policies to be applied unless the default ``cache_size`` is overridden.
+
+- ``prune_after=Xs``, ``prune_after=Xm``, ``prune_after=Xh``: Sets the
+  expiration time for cache files to ``X`` seconds (or minutes, hours
+  respectively).  When a file hasn't been accessed for ``prune_after`` seconds,
+  it is removed from the cache. A value of 0 disables the expiration-based
+  pruning. The default is 1 week.
+
+- ``prune_interval=Xs``, ``prune_interval=Xm``, ``prune_interval=Xh``:
+  Sets the pruning interval to ``X`` seconds (or minutes, hours
+  respectively). This is intended to be used to avoid scanning the directory
+  too often. It does not impact the decision of which files to prune. A
+  value of 0 forces the scan to occur. The default is every 20 minutes.
+
 Clang Bootstrap
 ---------------
 
diff --git a/docs/tools/dump_format_style.py b/docs/tools/dump_format_style.py
index 81a5af6ef42b..e2571f46448d 100644
--- a/docs/tools/dump_format_style.py
+++ b/docs/tools/dump_format_style.py
@@ -24,10 +24,10 @@ def doxygen2rst(text):
   text = re.sub(r'\\\w+ ', '', text)
   return text
 
-def indent(text, columns):
+def indent(text, columns, indent_first_line=True):
   indent = ' ' * columns
   s = re.sub(r'\n([^\n])', '\n' + indent + '\\1', text, flags=re.S)
-  if s.startswith('\n'):
+  if not indent_first_line or s.startswith('\n'):
     return s
   return indent + s
 
@@ -64,7 +64,9 @@ class NestedField:
     self.comment = comment.strip()
 
   def __str__(self):
-    return '\n* ``%s`` %s' % (self.name, doxygen2rst(self.comment))
+    return '\n* ``%s`` %s' % (
+        self.name,
+        doxygen2rst(indent(self.comment, 2, indent_first_line=False)))
 
 class Enum:
   def __init__(self, name, comment):
@@ -179,7 +181,7 @@ def read_options(header):
       if enums.has_key(option.type):
         option.enum = enums[option.type]
       elif nested_structs.has_key(option.type):
-        option.nested_struct = nested_structs[option.type];
+        option.nested_struct = nested_structs[option.type]
       else:
         raise Exception('Unknown type: %s' % option.type)
   return options
@@ -195,4 +197,3 @@ contents = substitute(contents, 'FORMAT_STYLE_OPTIONS', options_text)
 
 with open(DOC_FILE, 'wb') as output:
   output.write(contents)
-
diff --git a/include/clang-c/CXCompilationDatabase.h b/include/clang-c/CXCompilationDatabase.h
index 9359abfebfe0..29f89e52a6e3 100644
--- a/include/clang-c/CXCompilationDatabase.h
+++ b/include/clang-c/CXCompilationDatabase.h
@@ -7,7 +7,7 @@
 |*                                                                            *|
 |*===----------------------------------------------------------------------===*|
 |*                                                                            *|
-|* This header provides a public inferface to use CompilationDatabase without *|
+|* This header provides a public interface to use CompilationDatabase without *|
 |* the full Clang C++ API.                                                    *|
 |*                                                                            *|
 \*===----------------------------------------------------------------------===*/
diff --git a/include/clang-c/Index.h b/include/clang-c/Index.h
index 417ac9fd369a..4241e43a9697 100644
--- a/include/clang-c/Index.h
+++ b/include/clang-c/Index.h
@@ -7,7 +7,7 @@
 |*                                                                            *|
 |*===----------------------------------------------------------------------===*|
 |*                                                                            *|
-|* This header provides a public inferface to a Clang library for extracting  *|
+|* This header provides a public interface to a Clang library for extracting  *|
 |* high-level symbol information from source files without exposing the full  *|
 |* Clang C++ API.                                                             *|
 |*                                                                            *|
diff --git a/include/clang/AST/Decl.h b/include/clang/AST/Decl.h
index 9d49bac26a86..30552be9b381 100644
--- a/include/clang/AST/Decl.h
+++ b/include/clang/AST/Decl.h
@@ -1656,6 +1656,7 @@ private:
   unsigned HasImplicitReturnZero : 1;
   unsigned IsLateTemplateParsed : 1;
   unsigned IsConstexpr : 1;
+  unsigned InstantiationIsPending:1;
 
   /// \brief Indicates if the function uses __try.
   unsigned UsesSEHTry : 1;
@@ -1751,6 +1752,7 @@ protected:
         IsDeleted(false), IsTrivial(false), IsDefaulted(false),
         IsExplicitlyDefaulted(false), HasImplicitReturnZero(false),
         IsLateTemplateParsed(false), IsConstexpr(isConstexprSpecified),
+        InstantiationIsPending(false),
         UsesSEHTry(false), HasSkippedBody(false), WillHaveBody(false),
         EndRangeLoc(NameInfo.getEndLoc()), TemplateOrSpecialization(),
         DNLoc(NameInfo.getInfo()) {}
@@ -1872,7 +1874,7 @@ public:
   ///
   bool isThisDeclarationADefinition() const {
     return IsDeleted || IsDefaulted || Body || IsLateTemplateParsed ||
-      hasDefiningAttr();
+      WillHaveBody || hasDefiningAttr();
   }
 
   /// doesThisDeclarationHaveABody - Returns whether this specific
@@ -1943,6 +1945,15 @@ public:
   bool isConstexpr() const { return IsConstexpr; }
   void setConstexpr(bool IC) { IsConstexpr = IC; }
 
+  /// \brief Whether the instantiation of this function is pending.
+  /// This bit is set when the decision to instantiate this function is made
+  /// and unset if and when the function body is created. That leaves out
+  /// cases where instantiation did not happen because the template definition
+  /// was not seen in this TU. This bit remains set in those cases, under the
+  /// assumption that the instantiation will happen in some other TU.
+  bool instantiationIsPending() const { return InstantiationIsPending; }
+  void setInstantiationIsPending(bool IC) { InstantiationIsPending = IC; }
+
   /// \brief Indicates the function uses __try.
   bool usesSEHTry() const { return UsesSEHTry; }
   void setUsesSEHTry(bool UST) { UsesSEHTry = UST; }
diff --git a/include/clang/AST/DeclBase.h b/include/clang/AST/DeclBase.h
index c26e2d7bde95..0f1f481ae49b 100644
--- a/include/clang/AST/DeclBase.h
+++ b/include/clang/AST/DeclBase.h
@@ -202,26 +202,33 @@ public:
     OBJC_TQ_CSNullability = 0x40
   };
 
-protected:
-  // Enumeration values used in the bits stored in NextInContextAndBits.
-  enum {
-    /// \brief Whether this declaration is a top-level declaration (function,
-    /// global variable, etc.) that is lexically inside an objc container
-    /// definition.
-    TopLevelDeclInObjCContainerFlag = 0x01,
-    
-    /// \brief Whether this declaration is private to the module in which it was
-    /// defined.
-    ModulePrivateFlag = 0x02
+  /// The kind of ownership a declaration has, for visibility purposes.
+  /// This enumeration is designed such that higher values represent higher
+  /// levels of name hiding.
+  enum class ModuleOwnershipKind : unsigned {
+    /// This declaration is not owned by a module.
+    Unowned,
+    /// This declaration has an owning module, but is globally visible
+    /// (typically because its owning module is visible and we know that
+    /// modules cannot later become hidden in this compilation).
+    /// After serialization and deserialization, this will be converted
+    /// to VisibleWhenImported.
+    Visible,
+    /// This declaration has an owning module, and is visible when that
+    /// module is imported.
+    VisibleWhenImported,
+    /// This declaration has an owning module, but is only visible to
+    /// lookups that occur within that module.
+    ModulePrivate
   };
-  
+
+protected:
   /// \brief The next declaration within the same lexical
   /// DeclContext. These pointers form the linked list that is
   /// traversed via DeclContext's decls_begin()/decls_end().
   ///
-  /// The extra two bits are used for the TopLevelDeclInObjCContainer and
-  /// ModulePrivate bits.
-  llvm::PointerIntPair<Decl *, 2, unsigned> NextInContextAndBits;
+  /// The extra two bits are used for the ModuleOwnershipKind.
+  llvm::PointerIntPair<Decl *, 2, ModuleOwnershipKind> NextInContextAndBits;
 
 private:
   friend class DeclContext;
@@ -282,6 +289,11 @@ private:
   /// are regarded as "referenced" but not "used".
   unsigned Referenced : 1;
 
+  /// \brief Whether this declaration is a top-level declaration (function,
+  /// global variable, etc.) that is lexically inside an objc container
+  /// definition.
+  unsigned TopLevelDeclInObjCContainer : 1;
+  
   /// \brief Whether statistic collection is enabled.
   static bool StatisticsEnabled;
 
@@ -294,11 +306,6 @@ protected:
   /// \brief Whether this declaration was loaded from an AST file.
   unsigned FromASTFile : 1;
 
-  /// \brief Whether this declaration is hidden from normal name lookup, e.g.,
-  /// because it is was loaded from an AST file is either module-private or
-  /// because its submodule has not been made visible.
-  unsigned Hidden : 1;
-  
   /// IdentifierNamespace - This specifies what IDNS_* namespace this lives in.
   unsigned IdentifierNamespace : 13;
 
@@ -332,26 +339,38 @@ protected:
 private:
   bool AccessDeclContextSanity() const;
 
+  /// Get the module ownership kind to use for a local lexical child of \p DC,
+  /// which may be either a local or (rarely) an imported declaration.
+  static ModuleOwnershipKind getModuleOwnershipKindForChildOf(DeclContext *DC) {
+    if (DC) {
+      auto *D = cast<Decl>(DC);
+      auto MOK = D->getModuleOwnershipKind();
+      if (MOK != ModuleOwnershipKind::Unowned &&
+          (!D->isFromASTFile() || D->hasLocalOwningModuleStorage()))
+        return MOK;
+      // If D is not local and we have no local module storage, then we don't
+      // need to track module ownership at all.
+    }
+    return ModuleOwnershipKind::Unowned;
+  }
+
 protected:
   Decl(Kind DK, DeclContext *DC, SourceLocation L)
-      : NextInContextAndBits(), DeclCtx(DC), Loc(L), DeclKind(DK),
-        InvalidDecl(0), HasAttrs(false), Implicit(false), Used(false),
-        Referenced(false), Access(AS_none), FromASTFile(0),
-        Hidden(DC && cast<Decl>(DC)->Hidden &&
-               (!cast<Decl>(DC)->isFromASTFile() ||
-                hasLocalOwningModuleStorage())),
+      : NextInContextAndBits(nullptr, getModuleOwnershipKindForChildOf(DC)),
+        DeclCtx(DC), Loc(L), DeclKind(DK), InvalidDecl(0), HasAttrs(false),
+        Implicit(false), Used(false), Referenced(false),
+        TopLevelDeclInObjCContainer(false), Access(AS_none), FromASTFile(0),
         IdentifierNamespace(getIdentifierNamespaceForKind(DK)),
         CacheValidAndLinkage(0) {
     if (StatisticsEnabled) add(DK);
   }
 
   Decl(Kind DK, EmptyShell Empty)
-    : NextInContextAndBits(), DeclKind(DK), InvalidDecl(0),
-      HasAttrs(false), Implicit(false), Used(false), Referenced(false),
-      Access(AS_none), FromASTFile(0), Hidden(0),
-      IdentifierNamespace(getIdentifierNamespaceForKind(DK)),
-      CacheValidAndLinkage(0)
-  {
+      : NextInContextAndBits(), DeclKind(DK), InvalidDecl(0), HasAttrs(false),
+        Implicit(false), Used(false), Referenced(false),
+        TopLevelDeclInObjCContainer(false), Access(AS_none), FromASTFile(0),
+        IdentifierNamespace(getIdentifierNamespaceForKind(DK)),
+        CacheValidAndLinkage(0) {
     if (StatisticsEnabled) add(DK);
   }
 
@@ -551,16 +570,11 @@ public:
   /// global variable, etc.) that is lexically inside an objc container
   /// definition.
   bool isTopLevelDeclInObjCContainer() const {
-    return NextInContextAndBits.getInt() & TopLevelDeclInObjCContainerFlag;
+    return TopLevelDeclInObjCContainer;
   }
 
   void setTopLevelDeclInObjCContainer(bool V = true) {
-    unsigned Bits = NextInContextAndBits.getInt();
-    if (V)
-      Bits |= TopLevelDeclInObjCContainerFlag;
-    else
-      Bits &= ~TopLevelDeclInObjCContainerFlag;
-    NextInContextAndBits.setInt(Bits);
+    TopLevelDeclInObjCContainer = V;
   }
 
   /// \brief Looks on this and related declarations for an applicable
@@ -570,7 +584,7 @@ public:
   /// \brief Whether this declaration was marked as being private to the
   /// module in which it was defined.
   bool isModulePrivate() const {
-    return NextInContextAndBits.getInt() & ModulePrivateFlag;
+    return getModuleOwnershipKind() == ModuleOwnershipKind::ModulePrivate;
   }
 
   /// \brief Whether this declaration is exported (by virtue of being lexically
@@ -585,15 +599,14 @@ public:
   const Attr *getDefiningAttr() const;
 
 protected:
-  /// \brief Specify whether this declaration was marked as being private
+  /// \brief Specify that this declaration was marked as being private
   /// to the module in which it was defined.
-  void setModulePrivate(bool MP = true) {
-    unsigned Bits = NextInContextAndBits.getInt();
-    if (MP)
-      Bits |= ModulePrivateFlag;
-    else
-      Bits &= ~ModulePrivateFlag;
-    NextInContextAndBits.setInt(Bits);
+  void setModulePrivate() {
+    // The module-private specifier has no effect on unowned declarations.
+    // FIXME: We should track this in some way for source fidelity.
+    if (getModuleOwnershipKind() == ModuleOwnershipKind::Unowned)
+      return;
+    setModuleOwnershipKind(ModuleOwnershipKind::ModulePrivate);
   }
 
   /// \brief Set the owning module ID.
@@ -692,7 +705,7 @@ public:
   /// \brief Get the imported owning module, if this decl is from an imported
   /// (non-local) module.
   Module *getImportedOwningModule() const {
-    if (!isFromASTFile())
+    if (!isFromASTFile() || !hasOwningModule())
       return nullptr;
 
     return getOwningModuleSlow();
@@ -701,31 +714,57 @@ public:
   /// \brief Get the local owning module, if known. Returns nullptr if owner is
   /// not yet known or declaration is not from a module.
   Module *getLocalOwningModule() const {
-    if (isFromASTFile() || !Hidden)
+    if (isFromASTFile() || !hasOwningModule())
       return nullptr;
 
     assert(hasLocalOwningModuleStorage() &&
-           "hidden local decl but no local module storage");
+           "owned local decl but no local module storage");
     return reinterpret_cast<Module *const *>(this)[-1];
   }
   void setLocalOwningModule(Module *M) {
-    assert(!isFromASTFile() && Hidden && hasLocalOwningModuleStorage() &&
+    assert(!isFromASTFile() && hasOwningModule() &&
+           hasLocalOwningModuleStorage() &&
            "should not have a cached owning module");
     reinterpret_cast<Module **>(this)[-1] = M;
   }
 
+  /// Is this declaration owned by some module?
+  bool hasOwningModule() const {
+    return getModuleOwnershipKind() != ModuleOwnershipKind::Unowned;
+  }
+
+  /// Get the module that owns this declaration.
   Module *getOwningModule() const {
     return isFromASTFile() ? getImportedOwningModule() : getLocalOwningModule();
   }
 
-  /// \brief Determine whether this declaration is hidden from name lookup.
-  bool isHidden() const { return Hidden; }
+  /// \brief Determine whether this declaration might be hidden from name
+  /// lookup. Note that the declaration might be visible even if this returns
+  /// \c false, if the owning module is visible within the query context.
+  // FIXME: Rename this to make it clearer what it does.
+  bool isHidden() const {
+    return (int)getModuleOwnershipKind() > (int)ModuleOwnershipKind::Visible;
+  }
+
+  /// Set that this declaration is globally visible, even if it came from a
+  /// module that is not visible.
+  void setVisibleDespiteOwningModule() {
+    if (hasOwningModule())
+      setModuleOwnershipKind(ModuleOwnershipKind::Visible);
+  }
+
+  /// \brief Get the kind of module ownership for this declaration.
+  ModuleOwnershipKind getModuleOwnershipKind() const {
+    return NextInContextAndBits.getInt();
+  }
 
   /// \brief Set whether this declaration is hidden from name lookup.
-  void setHidden(bool Hide) {
-    assert((!Hide || isFromASTFile() || hasLocalOwningModuleStorage()) &&
-           "declaration with no owning module can't be hidden");
-    Hidden = Hide;
+  void setModuleOwnershipKind(ModuleOwnershipKind MOK) {
+    assert(!(getModuleOwnershipKind() == ModuleOwnershipKind::Unowned &&
+             MOK != ModuleOwnershipKind::Unowned && !isFromASTFile() &&
+             !hasLocalOwningModuleStorage()) &&
+           "no storage available for owning module for this declaration");
+    NextInContextAndBits.setInt(MOK);
   }
 
   unsigned getIdentifierNamespace() const {
diff --git a/include/clang/Analysis/CloneDetection.h b/include/clang/Analysis/CloneDetection.h
index 3b8173558408..1ca3514e69b0 100644
--- a/include/clang/Analysis/CloneDetection.h
+++ b/include/clang/Analysis/CloneDetection.h
@@ -17,6 +17,8 @@
 
 #include "clang/Basic/SourceLocation.h"
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/Regex.h"
 #include <vector>
 
 namespace clang {
@@ -319,6 +321,26 @@ struct OnlyLargestCloneConstraint {
   void constrain(std::vector<CloneDetector::CloneGroup> &Result);
 };
 
+struct FilenamePatternConstraint {
+  StringRef IgnoredFilesPattern;
+  std::shared_ptr<llvm::Regex> IgnoredFilesRegex;
+
+  FilenamePatternConstraint(StringRef IgnoredFilesPattern) 
+      : IgnoredFilesPattern(IgnoredFilesPattern) {
+    IgnoredFilesRegex = std::make_shared<llvm::Regex>("^(" +
+        IgnoredFilesPattern.str() + "$)");
+  }
+
+  bool isAutoGenerated(const CloneDetector::CloneGroup &Group);
+
+  void constrain(std::vector<CloneDetector::CloneGroup> &CloneGroups) {
+    CloneConstraint::filterGroups(
+        CloneGroups, [this](const CloneDetector::CloneGroup &Group) {
+          return isAutoGenerated(Group);
+        });
+  }
+};
+
 /// Analyzes the pattern of the referenced variables in a statement.
 class VariablePattern {
 
diff --git a/include/clang/Basic/Builtins.def b/include/clang/Basic/Builtins.def
index a9ec172422ab..75781dc7491d 100644
--- a/include/clang/Basic/Builtins.def
+++ b/include/clang/Basic/Builtins.def
@@ -52,6 +52,7 @@
 //  LL  -> long long
 //  LLL -> __int128_t (e.g. LLLi)
 //  W   -> int64_t
+//  N   -> 'int' size if target is LP64, 'L' otherwise.
 //  S   -> signed
 //  U   -> unsigned
 //  I   -> Required to constant fold to an integer constant expression.
@@ -718,11 +719,11 @@ BUILTIN(__builtin_rindex, "c*cC*i", "Fn")
 LANGBUILTIN(_alloca,          "v*z", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__assume,         "vb",  "n", ALL_MS_LANGUAGES)
 LIBBUILTIN(_byteswap_ushort, "UsUs",     "fnc", "stdlib.h", ALL_MS_LANGUAGES)
-LIBBUILTIN(_byteswap_ulong,  "ULiULi",   "fnc", "stdlib.h", ALL_MS_LANGUAGES)
+LIBBUILTIN(_byteswap_ulong,  "UNiUNi",   "fnc", "stdlib.h", ALL_MS_LANGUAGES)
 LIBBUILTIN(_byteswap_uint64, "ULLiULLi", "fnc", "stdlib.h", ALL_MS_LANGUAGES)
 LANGBUILTIN(__debugbreak,     "v",   "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(__exception_code, "ULi", "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_exception_code,  "ULi", "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(__exception_code, "UNi", "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_exception_code,  "UNi", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__exception_info, "v*",  "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_exception_info,  "v*",  "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__abnormal_termination, "i", "n", ALL_MS_LANGUAGES)
@@ -730,33 +731,33 @@ LANGBUILTIN(_abnormal_termination,  "i", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__GetExceptionInfo, "v*.", "ntu", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedAnd8,   "ccD*c",        "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedAnd16,  "ssD*s",        "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedAnd,    "LiLiD*Li",     "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedAnd,    "NiNiD*Ni",     "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedCompareExchange8,   "ccD*cc",         "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedCompareExchange16,  "ssD*ss",         "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedCompareExchange,    "LiLiD*LiLi",     "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedCompareExchange,    "NiNiD*NiNi",     "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedCompareExchange64,  "LLiLLiD*LLiLLi", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedCompareExchangePointer, "v*v*D*v*v*", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedDecrement16,        "ssD*",     "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedDecrement,          "LiLiD*",   "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedExchange,           "LiLiD*Li",     "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedDecrement,          "NiNiD*",   "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedExchange,           "NiNiD*Ni",     "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchange8,          "ccD*c",        "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchange16,         "ssD*s",        "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchangeAdd8,       "ccD*c",          "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchangeAdd16,      "ssD*s",          "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedExchangeAdd,        "LiLiD*Li",       "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedExchangeAdd,        "NiNiD*Ni",       "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchangePointer,    "v*v*D*v*",   "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchangeSub8,   "ccD*c",        "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedExchangeSub16,  "ssD*s",        "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedExchangeSub,    "LiLiD*Li",     "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedExchangeSub,    "NiNiD*Ni",     "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedIncrement16,        "ssD*",     "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedIncrement,          "LiLiD*",   "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedIncrement,          "NiNiD*",   "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedOr8,  "ccD*c",        "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedOr16, "ssD*s",        "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedOr,   "LiLiD*Li",     "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedOr,   "NiNiD*Ni",     "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedXor8,  "ccD*c",       "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_InterlockedXor16, "ssD*s",       "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_InterlockedXor,   "LiLiD*Li",    "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_interlockedbittestandset, "UcLiD*Li", "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_InterlockedXor,   "NiNiD*Ni",    "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_interlockedbittestandset, "UcNiD*Ni", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__noop,           "i.",  "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__popcnt16, "UsUs",     "nc", ALL_MS_LANGUAGES)
 LANGBUILTIN(__popcnt,   "UiUi",     "nc", ALL_MS_LANGUAGES)
@@ -765,12 +766,12 @@ LANGBUILTIN(_ReturnAddress, "v*", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotl8,  "UcUcUc",    "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotl16, "UsUsUc",    "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotl,   "UiUii",     "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_lrotl,  "ULiULii",   "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_lrotl,  "UNiUNii",   "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotl64, "ULLiULLii", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotr8,  "UcUcUc",    "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotr16, "UsUsUc",    "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotr,   "UiUii",     "n", ALL_MS_LANGUAGES)
-LANGBUILTIN(_lrotr,  "ULiULii",   "n", ALL_MS_LANGUAGES)
+LANGBUILTIN(_lrotr,  "UNiUNii",   "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(_rotr64, "ULLiULLii", "n", ALL_MS_LANGUAGES)
 LANGBUILTIN(__va_start,       "vc**.", "nt", ALL_MS_LANGUAGES)
 LANGBUILTIN(__fastfail, "vUi",    "nr", ALL_MS_LANGUAGES)
diff --git a/include/clang/Basic/BuiltinsARM.def b/include/clang/Basic/BuiltinsARM.def
index e8db347d4be5..4e277f8a5a6b 100644
--- a/include/clang/Basic/BuiltinsARM.def
+++ b/include/clang/Basic/BuiltinsARM.def
@@ -215,10 +215,10 @@ LANGBUILTIN(_MoveFromCoprocessor2, "UiIUiIUiIUiIUiIUi", "", ALL_MS_LANGUAGES)
 LANGBUILTIN(_MoveToCoprocessor, "vUiIUiIUiIUiIUiIUi", "", ALL_MS_LANGUAGES)
 LANGBUILTIN(_MoveToCoprocessor2, "vUiIUiIUiIUiIUiIUi", "", ALL_MS_LANGUAGES)
 
-TARGET_HEADER_BUILTIN(_BitScanForward, "UcULi*ULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(_BitScanReverse, "UcULi*ULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(_BitScanForward64, "UcULi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(_BitScanReverse64, "UcULi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanForward, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanReverse, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanForward64, "UcUNi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanReverse64, "UcUNi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
 TARGET_HEADER_BUILTIN(_InterlockedAnd64,         "LLiLLiD*LLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(_InterlockedDecrement64,   "LLiLLiD*",    "nh", "intrin.h", ALL_MS_LANGUAGES, "")
diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index a98c8f0a53db..4cd3f1d46473 100644
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -1822,8 +1822,8 @@ TARGET_BUILTIN(__builtin_ia32_mwaitx, "vUiUiUi", "", "mwaitx")
 TARGET_BUILTIN(__builtin_ia32_clzero, "vv*", "", "clzero")
 
 // MSVC
-TARGET_HEADER_BUILTIN(_BitScanForward, "UcULi*ULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(_BitScanReverse, "UcULi*ULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanForward, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanReverse, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
 TARGET_HEADER_BUILTIN(_ReadWriteBarrier, "v", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(_ReadBarrier,      "v", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
@@ -1838,15 +1838,15 @@ TARGET_HEADER_BUILTIN(__stosb, "vUc*Ucz", "nh", "intrin.h", ALL_MS_LANGUAGES, ""
 TARGET_HEADER_BUILTIN(__int2c, "v",       "nr", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(__ud2,   "v",       "nr", "intrin.h", ALL_MS_LANGUAGES, "")
 
-TARGET_HEADER_BUILTIN(__readfsbyte,  "UcULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readfsword,  "UsULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readfsdword, "ULiULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readfsqword, "ULLiULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readfsbyte,  "UcUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readfsword,  "UsUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readfsdword, "UNiUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readfsqword, "ULLiUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
-TARGET_HEADER_BUILTIN(__readgsbyte,  "UcULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readgsword,  "UsULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readgsdword, "ULiULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(__readgsqword, "ULLiULi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readgsbyte,  "UcUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readgsword,  "UsUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readgsdword, "UNiUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(__readgsqword, "ULLiUNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
 #undef BUILTIN
 #undef TARGET_BUILTIN
diff --git a/include/clang/Basic/BuiltinsX86_64.def b/include/clang/Basic/BuiltinsX86_64.def
index 2851184c2c84..4cde153d8372 100644
--- a/include/clang/Basic/BuiltinsX86_64.def
+++ b/include/clang/Basic/BuiltinsX86_64.def
@@ -22,8 +22,8 @@
 #  define TARGET_HEADER_BUILTIN(ID, TYPE, ATTRS, HEADER, LANG, FEATURE) BUILTIN(ID, TYPE, ATTRS)
 #endif
 
-TARGET_HEADER_BUILTIN(_BitScanForward64, "UcULi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
-TARGET_HEADER_BUILTIN(_BitScanReverse64, "UcULi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanForward64, "UcUNi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
+TARGET_HEADER_BUILTIN(_BitScanReverse64, "UcUNi*ULLi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
 TARGET_HEADER_BUILTIN(__mulh,  "LLiLLiLLi",    "nch", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(__umulh, "ULLiULLiULLi", "nch", "intrin.h", ALL_MS_LANGUAGES, "")
diff --git a/include/clang/Basic/DiagnosticDriverKinds.td b/include/clang/Basic/DiagnosticDriverKinds.td
index 3833f0f28f05..eb6dd37c148d 100644
--- a/include/clang/Basic/DiagnosticDriverKinds.td
+++ b/include/clang/Basic/DiagnosticDriverKinds.td
@@ -226,6 +226,9 @@ def warn_drv_enabling_rtti_with_exceptions : Warning<
 def warn_drv_disabling_vptr_no_rtti_default : Warning<
   "implicitly disabling vptr sanitizer because rtti wasn't enabled">,
   InGroup<DiagGroup<"auto-disable-vptr-sanitizer">>;
+def warn_drv_object_size_disabled_O0 : Warning<
+  "the object size sanitizer has no effect at -O0, but is explicitly enabled: %0">,
+  InGroup<InvalidCommandLineArgument>;
 
 def note_drv_command_failed_diag_msg : Note<
   "diagnostic msg: %0">;
diff --git a/include/clang/Basic/DiagnosticGroups.td b/include/clang/Basic/DiagnosticGroups.td
index cf404768472f..464c2425a1f1 100644
--- a/include/clang/Basic/DiagnosticGroups.td
+++ b/include/clang/Basic/DiagnosticGroups.td
@@ -98,7 +98,9 @@ def CXX11CompatDeprecatedWritableStr :
 def DeprecatedAttributes : DiagGroup<"deprecated-attributes">;
 def DeprecatedDeclarations : DiagGroup<"deprecated-declarations">;
 def UnavailableDeclarations : DiagGroup<"unavailable-declarations">;
-def UnguardedAvailability : DiagGroup<"unguarded-availability">;
+def UnguardedAvailabilityNew : DiagGroup<"unguarded-availability-new">;
+def UnguardedAvailability : DiagGroup<"unguarded-availability",
+                                      [UnguardedAvailabilityNew]>;
 // partial-availability is an alias of unguarded-availability.
 def : DiagGroup<"partial-availability", [UnguardedAvailability]>;
 def DeprecatedDynamicExceptionSpec
@@ -149,6 +151,7 @@ def GNUFlexibleArrayUnionMember : DiagGroup<"gnu-flexible-array-union-member">;
 def GNUFoldingConstant : DiagGroup<"gnu-folding-constant">;
 def FormatExtraArgs : DiagGroup<"format-extra-args">;
 def FormatZeroLength : DiagGroup<"format-zero-length">;
+def CXX1zCompatMangling : DiagGroup<"c++1z-compat-mangling">;
 
 // Warnings for C++1y code which is not compatible with prior C++ standards.
 def CXXPre14Compat : DiagGroup<"c++98-c++11-compat">;
@@ -211,7 +214,8 @@ def CXX14CompatPedantic : DiagGroup<"c++14-compat-pedantic",
                                     [CXXPre1zCompatPedantic]>;
 
 def CXX1zCompat : DiagGroup<"c++1z-compat", [DeprecatedRegister,
-                                             DeprecatedIncrementBool]>;
+                                             DeprecatedIncrementBool,
+                                             CXX1zCompatMangling]>;
 
 def ExitTimeDestructors : DiagGroup<"exit-time-destructors">;
 def FlexibleArrayExtensions : DiagGroup<"flexible-array-extensions">;
diff --git a/include/clang/Basic/DiagnosticSemaKinds.td b/include/clang/Basic/DiagnosticSemaKinds.td
index aa73a6934518..cba9b251215a 100644
--- a/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/include/clang/Basic/DiagnosticSemaKinds.td
@@ -507,7 +507,7 @@ def warn_deprecated_copy_operation : Warning<
   InGroup<Deprecated>, DefaultIgnore;
 def warn_cxx1z_compat_exception_spec_in_signature : Warning<
   "mangled name of %0 will change in C++17 due to non-throwing exception "
-  "specification in function signature">, InGroup<CXX1zCompat>;
+  "specification in function signature">, InGroup<CXX1zCompatMangling>;
 
 def warn_global_constructor : Warning<
   "declaration requires a global constructor">,
@@ -2870,8 +2870,13 @@ def note_protocol_method : Note<
 def warn_unguarded_availability :
   Warning<"%0 is only available on %1 %2 or newer">,
   InGroup<UnguardedAvailability>, DefaultIgnore;
+def warn_unguarded_availability_new :
+  Warning<warn_unguarded_availability.Text>,
+  InGroup<UnguardedAvailabilityNew>;
 def warn_partial_availability : Warning<"%0 is only available conditionally">,
     InGroup<UnguardedAvailability>, DefaultIgnore;
+def warn_partial_availability_new : Warning<warn_partial_availability.Text>,
+  InGroup<UnguardedAvailabilityNew>;
 def note_partial_availability_silence : Note<
   "explicitly redeclare %0 to silence this warning">;
 def note_unguarded_available_silence : Note<
@@ -2879,9 +2884,14 @@ def note_unguarded_available_silence : Note<
   " this warning">;
 def warn_partial_message : Warning<"%0 is partial: %1">,
     InGroup<UnguardedAvailability>, DefaultIgnore;
+def warn_partial_message_new : Warning<warn_partial_message.Text>,
+  InGroup<UnguardedAvailabilityNew>;
 def warn_partial_fwdclass_message : Warning<
     "%0 may be partial because the receiver type is unknown">,
     InGroup<UnguardedAvailability>, DefaultIgnore;
+def warn_partial_fwdclass_message_new :
+  Warning<warn_partial_fwdclass_message.Text>,
+  InGroup<UnguardedAvailabilityNew>;
 def warn_at_available_unchecked_use : Warning<
   "%select{@available|__builtin_available}0 does not guard availability here; "
   "use if (%select{@available|__builtin_available}0) instead">,
@@ -6341,6 +6351,15 @@ def err_exceptions_disabled : Error<
   "cannot use '%0' with exceptions disabled">;
 def err_objc_exceptions_disabled : Error<
   "cannot use '%0' with Objective-C exceptions disabled">;
+def warn_throw_in_noexcept_func 
+    : Warning<"%0 has a non-throwing exception specification but can still "
+      "throw, resulting in unexpected program termination">,
+      InGroup<Exceptions>;
+def note_throw_in_dtor 
+    : Note<"destructor or deallocator has a (possibly implicit) non-throwing "
+      "excepton specification">;
+def note_throw_in_function 
+    : Note<"non-throwing function declare here">;
 def err_seh_try_outside_functions : Error<
   "cannot use SEH '__try' in blocks, captured regions, or Obj-C method decls">;
 def err_mixing_cxx_try_seh_try : Error<
@@ -8311,6 +8330,9 @@ def err_opencl_ext_vector_component_invalid_length : Error<
   "vector component access has invalid length %0.  Supported: 1,2,3,4,8,16.">;
 def err_opencl_function_variable : Error<
   "%select{non-kernel function|function scope}0 variable cannot be declared in %1 address space">;
+def err_opencl_addrspace_scope : Error<
+  "variables in the %0 address space can only be declared in the outermost "
+  "scope of a kernel function">;
 def err_static_function_scope : Error<
   "variables in function scope cannot be declared static">;
 def err_opencl_bitfields : Error<
diff --git a/include/clang/Basic/DiagnosticSerializationKinds.td b/include/clang/Basic/DiagnosticSerializationKinds.td
index 0a59a633232c..3c64ebb9c7f4 100644
--- a/include/clang/Basic/DiagnosticSerializationKinds.td
+++ b/include/clang/Basic/DiagnosticSerializationKinds.td
@@ -112,8 +112,13 @@ def note_module_odr_violation_possible_decl : Note<
 def err_module_odr_violation_different_definitions : Error<
   "%q0 has different definitions in different modules; "
   "%select{definition in module '%2' is here|defined here}1">;
+def note_first_module_difference : Note<
+  "in first definition, possible difference is here">;
 def note_module_odr_violation_different_definitions : Note<
   "definition in module '%0' is here">;
+def note_second_module_difference : Note<
+  "in second definition, possible difference is here">;
+
 def err_module_odr_violation_different_instantiations : Error<
   "instantiation of %q0 is different in different modules">;
 
diff --git a/include/clang/Basic/Module.h b/include/clang/Basic/Module.h
index 1e52b29367b2..177175eae965 100644
--- a/include/clang/Basic/Module.h
+++ b/include/clang/Basic/Module.h
@@ -393,7 +393,9 @@ public:
 
   /// \brief Retrieve the full name of this module, including the path from
   /// its top-level module.
-  std::string getFullModuleName() const;
+  /// \param AllowStringLiterals If \c true, components that might not be
+  ///        lexically valid as identifiers will be emitted as string literals.
+  std::string getFullModuleName(bool AllowStringLiterals = false) const;
 
   /// \brief Whether the full name of this module is equal to joining
   /// \p nameParts with "."s.
diff --git a/include/clang/Basic/arm_neon.td b/include/clang/Basic/arm_neon.td
index ad8d679a1664..d5c16a91a34f 100644
--- a/include/clang/Basic/arm_neon.td
+++ b/include/clang/Basic/arm_neon.td
@@ -227,6 +227,7 @@ def OP_UNAVAILABLE : Operation {
 // u: unsigned integer (int/float args)
 // f: float (int args)
 // F: double (int args)
+// H: half (int args)
 // d: default
 // g: default, ignore 'Q' size modifier.
 // j: default, force 'Q' size modifier.
@@ -345,6 +346,7 @@ def OP_MLSLHi   : Op<(call "vmlsl", $p0, (call "vget_high", $p1),
                                          (call "vget_high", $p2))>;
 def OP_MLSLHi_N : Op<(call "vmlsl_n", $p0, (call "vget_high", $p1), $p2)>;
 def OP_MUL_N    : Op<(op "*", $p0, (dup $p1))>;
+def OP_MULX_N   : Op<(call "vmulx", $p0, (dup $p1))>;
 def OP_MLA_N    : Op<(op "+", $p0, (op "*", $p1, (dup $p2)))>;
 def OP_MLS_N    : Op<(op "-", $p0, (op "*", $p1, (dup $p2)))>;
 def OP_FMLA_N   : Op<(call "vfma", $p0, $p1, (dup $p2))>;
@@ -1661,3 +1663,186 @@ def SCALAR_SQRDMLSH_LANEQ : SOpInst<"vqrdmlsh_laneq", "sssji", "SsSi", OP_SCALAR
 def SCALAR_VDUP_LANE : IInst<"vdup_lane", "sdi", "ScSsSiSlSfSdSUcSUsSUiSUlSPcSPs">;
 def SCALAR_VDUP_LANEQ : IInst<"vdup_laneq", "sji", "ScSsSiSlSfSdSUcSUsSUiSUlSPcSPs">;
 }
+
+// ARMv8.2-A FP16 intrinsics.
+let ArchGuard = "defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(__aarch64__)" in {
+
+  // ARMv8.2-A FP16 one-operand vector intrinsics.
+
+  // Comparison
+  def CMEQH    : SInst<"vceqz", "ud", "hQh">;
+  def CMGEH    : SInst<"vcgez", "ud", "hQh">;
+  def CMGTH    : SInst<"vcgtz", "ud", "hQh">;
+  def CMLEH    : SInst<"vclez", "ud", "hQh">;
+  def CMLTH    : SInst<"vcltz", "ud", "hQh">;
+
+  // Vector conversion
+  def VCVT_F16     : SInst<"vcvt_f16", "Hd",  "sUsQsQUs">;
+  def VCVT_S16     : SInst<"vcvt_s16", "xd",  "hQh">;
+  def VCVT_U16     : SInst<"vcvt_u16", "ud",  "hQh">;
+  def VCVTA_S16    : SInst<"vcvta_s16", "xd", "hQh">;
+  def VCVTA_U16    : SInst<"vcvta_u16", "ud", "hQh">;
+  def VCVTM_S16    : SInst<"vcvtm_s16", "xd", "hQh">;
+  def VCVTM_U16    : SInst<"vcvtm_u16", "ud", "hQh">;
+  def VCVTN_S16    : SInst<"vcvtn_s16", "xd", "hQh">;
+  def VCVTN_U16    : SInst<"vcvtn_u16", "ud", "hQh">;
+  def VCVTP_S16    : SInst<"vcvtp_s16", "xd", "hQh">;
+  def VCVTP_U16    : SInst<"vcvtp_u16", "ud", "hQh">;
+
+  // Vector rounding
+  def FRINTZH      : SInst<"vrnd",  "dd", "hQh">;
+  def FRINTNH      : SInst<"vrndn", "dd", "hQh">;
+  def FRINTAH      : SInst<"vrnda", "dd", "hQh">;
+  def FRINTPH      : SInst<"vrndp", "dd", "hQh">;
+  def FRINTMH      : SInst<"vrndm", "dd", "hQh">;
+  def FRINTXH      : SInst<"vrndx", "dd", "hQh">;
+  def FRINTIH      : SInst<"vrndi", "dd", "hQh">;
+
+  // Misc.
+  def VABSH        : SInst<"vabs", "dd", "hQh">;
+  def VNEGH        : SOpInst<"vneg", "dd", "hQh", OP_NEG>;
+  def VRECPEH      : SInst<"vrecpe", "dd", "hQh">;
+  def FRSQRTEH     : SInst<"vrsqrte", "dd", "hQh">;
+  def FSQRTH       : SInst<"vsqrt", "dd", "hQh">;
+
+  // ARMv8.2-A FP16 two-operands vector intrinsics.
+
+  // Misc.
+  def VADDH        : SOpInst<"vadd", "ddd", "hQh", OP_ADD>;
+  def VABDH        : SInst<"vabd", "ddd",  "hQh">;
+  def VSUBH         : SOpInst<"vsub", "ddd", "hQh", OP_SUB>;
+
+  // Comparison
+  let InstName = "vacge" in {
+	  def VCAGEH     : SInst<"vcage", "udd", "hQh">;
+	  def VCALEH     : SInst<"vcale", "udd", "hQh">;
+  }
+  let InstName = "vacgt" in {
+    def VCAGTH     : SInst<"vcagt", "udd", "hQh">;
+	  def VCALTH     : SInst<"vcalt", "udd", "hQh">;
+  }
+  def VCEQH        : SOpInst<"vceq", "udd", "hQh", OP_EQ>;
+  def VCGEH        : SOpInst<"vcge", "udd", "hQh", OP_GE>;
+  def VCGTH        : SOpInst<"vcgt", "udd", "hQh", OP_GT>;
+  let InstName = "vcge" in
+    def VCLEH      : SOpInst<"vcle", "udd", "hQh", OP_LE>;
+  let InstName = "vcgt" in
+    def VCLTH      : SOpInst<"vclt", "udd", "hQh", OP_LT>;
+
+  // Vector conversion
+  let isVCVT_N = 1 in {
+    def VCVT_N_F16 : SInst<"vcvt_n_f16", "Hdi", "sUsQsQUs">;
+    def VCVT_N_S16 : SInst<"vcvt_n_s16", "xdi", "hQh">;
+    def VCVT_N_U16 : SInst<"vcvt_n_u16", "udi", "hQh">;
+  }
+
+  // Max/Min
+  def VMAXH         : SInst<"vmax", "ddd", "hQh">;
+  def VMINH         : SInst<"vmin", "ddd", "hQh">;
+  def FMAXNMH       : SInst<"vmaxnm", "ddd", "hQh">;
+  def FMINNMH       : SInst<"vminnm", "ddd", "hQh">;
+
+  // Multiplication/Division
+  def VMULH         : SOpInst<"vmul", "ddd", "hQh", OP_MUL>;
+  def MULXH         : SInst<"vmulx", "ddd", "hQh">;
+  def FDIVH         : IOpInst<"vdiv", "ddd",  "hQh", OP_DIV>;
+
+  // Pairwise addition
+  def VPADDH        : SInst<"vpadd", "ddd", "hQh">;
+
+  // Pairwise Max/Min
+  def VPMAXH        : SInst<"vpmax", "ddd", "hQh">;
+  def VPMINH        : SInst<"vpmin", "ddd", "hQh">;
+  // Pairwise MaxNum/MinNum
+  def FMAXNMPH      : SInst<"vpmaxnm", "ddd", "hQh">;
+  def FMINNMPH      : SInst<"vpminnm", "ddd", "hQh">;
+
+  // Reciprocal/Sqrt
+  def VRECPSH       : SInst<"vrecps", "ddd", "hQh">;
+  def VRSQRTSH      : SInst<"vrsqrts", "ddd", "hQh">;
+
+  // ARMv8.2-A FP16 three-operands vector intrinsics.
+
+  // Vector fused multiply-add operations
+  def VFMAH        : SInst<"vfma", "dddd", "hQh">;
+  def VFMSH        : SOpInst<"vfms", "dddd", "hQh", OP_FMLS>;
+
+  // ARMv8.2-A FP16 lane vector intrinsics.
+
+  // FMA lane
+  def VFMA_LANEH   : IInst<"vfma_lane", "dddgi", "hQh">;
+  def VFMA_LANEQH  : IInst<"vfma_laneq", "dddji", "hQh">;
+
+  // FMA lane with scalar argument
+  def FMLA_NH      : SOpInst<"vfma_n", "ddds", "hQh", OP_FMLA_N>;
+  // Scalar floating point fused multiply-add (scalar, by element)
+  def SCALAR_FMLA_LANEH  : IInst<"vfma_lane", "sssdi", "Sh">;
+  def SCALAR_FMLA_LANEQH : IInst<"vfma_laneq", "sssji", "Sh">;
+
+  // FMS lane
+  def VFMS_LANEH   : IOpInst<"vfms_lane", "dddgi", "hQh", OP_FMS_LN>;
+  def VFMS_LANEQH  : IOpInst<"vfms_laneq", "dddji", "hQh", OP_FMS_LNQ>;
+  // FMS lane with scalar argument
+  def FMLS_NH      : SOpInst<"vfms_n", "ddds", "hQh", OP_FMLS_N>;
+  // Scalar floating foint fused multiply-subtract (scalar, by element)
+  def SCALAR_FMLS_LANEH  : IOpInst<"vfms_lane", "sssdi", "Sh", OP_FMS_LN>;
+  def SCALAR_FMLS_LANEQH : IOpInst<"vfms_laneq", "sssji", "Sh", OP_FMS_LNQ>;
+
+  // Mul lane
+  def VMUL_LANEH    : IOpInst<"vmul_lane", "ddgi", "hQh", OP_MUL_LN>;
+  def VMUL_LANEQH   : IOpInst<"vmul_laneq", "ddji", "hQh", OP_MUL_LN>;
+  def VMUL_NH       : IOpInst<"vmul_n", "dds", "hQh", OP_MUL_N>;
+  // Scalar floating point  multiply (scalar, by element)
+  def SCALAR_FMUL_LANEH  : IOpInst<"vmul_lane", "ssdi", "Sh", OP_SCALAR_MUL_LN>;
+  def SCALAR_FMUL_LANEQH : IOpInst<"vmul_laneq", "ssji", "Sh", OP_SCALAR_MUL_LN>;
+
+  // Mulx lane
+  def VMULX_LANEH   : IOpInst<"vmulx_lane", "ddgi", "hQh", OP_MULX_LN>;
+  def VMULX_LANEQH  : IOpInst<"vmulx_laneq", "ddji", "hQh", OP_MULX_LN>;
+  def VMULX_NH      : IOpInst<"vmulx_n", "dds", "hQh", OP_MULX_N>;
+  // TODO: Scalar floating point multiply extended (scalar, by element)
+  // Below ones are commented out because they need vmulx_f16(float16_t, float16_t)
+  // which will be implemented later with fp16 scalar intrinsic (arm_fp16.h)
+  //def SCALAR_FMULX_LANEH : IOpInst<"vmulx_lane", "ssdi", "Sh", OP_SCALAR_MUL_LN>;
+  //def SCALAR_FMULX_LANEQH : IOpInst<"vmulx_laneq", "ssji", "Sh", OP_SCALAR_MUL_LN>;
+
+  // ARMv8.2-A FP16 reduction vector intrinsics.
+  def VMAXVH   : SInst<"vmaxv", "sd", "hQh">;
+  def VMINVH   : SInst<"vminv", "sd", "hQh">;
+  def FMAXNMVH : SInst<"vmaxnmv", "sd", "hQh">;
+  def FMINNMVH : SInst<"vminnmv", "sd", "hQh">;
+
+  // Data processing intrinsics - section 5
+
+  // Logical operations
+  let isHiddenLInst = 1 in
+  def VBSLH    : SInst<"vbsl", "dudd", "hQh">;
+
+  // Transposition operations
+  def VZIPH    : WInst<"vzip", "2dd", "hQh">;
+  def VUZPH    : WInst<"vuzp", "2dd", "hQh">;
+  def VTRNH    : WInst<"vtrn", "2dd", "hQh">;
+
+  // Set all lanes to same value.
+  /* Already implemented prior to ARMv8.2-A.
+  def VMOV_NH  : WOpInst<"vmov_n", "ds", "hQh", OP_DUP>;
+  def VDUP_NH  : WOpInst<"vdup_n", "ds", "hQh", OP_DUP>;
+  def VDUP_LANE1H : WOpInst<"vdup_lane", "dgi", "hQh", OP_DUP_LN>;*/
+
+  // Vector Extract
+  def VEXTH      : WInst<"vext", "dddi", "hQh">;
+
+  // Reverse vector elements
+  def VREV64H    : WOpInst<"vrev64", "dd", "hQh", OP_REV64>;
+
+  // Permutation
+  def VTRN1H     : SOpInst<"vtrn1", "ddd", "hQh", OP_TRN1>;
+  def VZIP1H     : SOpInst<"vzip1", "ddd", "hQh", OP_ZIP1>;
+  def VUZP1H     : SOpInst<"vuzp1", "ddd", "hQh", OP_UZP1>;
+  def VTRN2H     : SOpInst<"vtrn2", "ddd", "hQh", OP_TRN2>;
+  def VZIP2H     : SOpInst<"vzip2", "ddd", "hQh", OP_ZIP2>;
+  def VUZP2H     : SOpInst<"vuzp2", "ddd", "hQh", OP_UZP2>;
+
+  def SCALAR_VDUP_LANEH  : IInst<"vdup_lane", "sdi", "Sh">;
+  def SCALAR_VDUP_LANEQH : IInst<"vdup_laneq", "sji", "Sh">;
+}
diff --git a/include/clang/Driver/CC1Options.td b/include/clang/Driver/CC1Options.td
index 1132d1345f60..c478cfc7833e 100644
--- a/include/clang/Driver/CC1Options.td
+++ b/include/clang/Driver/CC1Options.td
@@ -134,7 +134,6 @@ def migrator_no_finalize_removal : Flag<["-"], "no-finalize-removal">,
 //===----------------------------------------------------------------------===//
 
 let Flags = [CC1Option, CC1AsOption, NoDriverOption] in {
-
 def debug_info_kind_EQ : Joined<["-"], "debug-info-kind=">;
 def debug_info_macro : Flag<["-"], "debug-info-macro">,
   HelpText<"Emit macro debug information">;
@@ -144,20 +143,22 @@ def fdebug_compilation_dir : Separate<["-"], "fdebug-compilation-dir">,
   HelpText<"The compilation directory to embed in the debug info.">;
 def dwarf_debug_flags : Separate<["-"], "dwarf-debug-flags">,
   HelpText<"The string to embed in the Dwarf debug flags record.">;
+def compress_debug_sections : Flag<["-", "--"], "compress-debug-sections">,
+    HelpText<"DWARF debug sections compression">;
+def compress_debug_sections_EQ : Flag<["-"], "compress-debug-sections=">,
+    HelpText<"DWARF debug sections compression type">;
 def mno_exec_stack : Flag<["-"], "mnoexecstack">,
   HelpText<"Mark the file as not needing an executable stack">;
 def massembler_fatal_warnings : Flag<["-"], "massembler-fatal-warnings">,
   HelpText<"Make assembler warnings fatal">;
 def mrelax_relocations : Flag<["--"], "mrelax-relocations">,
     HelpText<"Use relaxable elf relocations">;
-def compress_debug_sections : Flag<["-"], "compress-debug-sections">,
-    HelpText<"Compress DWARF debug sections using zlib">;
 def msave_temp_labels : Flag<["-"], "msave-temp-labels">,
   HelpText<"Save temporary labels in the symbol table. "
            "Note this may change .s semantics and shouldn't generally be used "
            "on compiler-generated code.">;
 def mrelocation_model : Separate<["-"], "mrelocation-model">,
-  HelpText<"The relocation model to use">;
+  HelpText<"The relocation model to use">, Values<"static,pic,ropi,rwpi,ropi-rwpi,dynamic-no-pic">;
 def fno_math_builtin : Flag<["-"], "fno-math-builtin">,
   HelpText<"Disable implicit builtin knowledge of math functions">;
 }
@@ -228,7 +229,7 @@ def no_struct_path_tbaa : Flag<["-"], "no-struct-path-tbaa">,
 def masm_verbose : Flag<["-"], "masm-verbose">,
   HelpText<"Generate verbose assembly output">;
 def mcode_model : Separate<["-"], "mcode-model">,
-  HelpText<"The code model to use">;
+  HelpText<"The code model to use">, Values<"small,kernel,medium,large">;
 def mdebug_pass : Separate<["-"], "mdebug-pass">,
   HelpText<"Enable additional debug output">;
 def mdisable_fp_elim : Flag<["-"], "mdisable-fp-elim">,
@@ -307,7 +308,7 @@ def fsanitize_coverage_no_prune
       HelpText<"Disable coverage pruning (i.e. instrument all blocks/edges)">;
 def fprofile_instrument_EQ : Joined<["-"], "fprofile-instrument=">,
     HelpText<"Enable PGO instrumentation. The accepted value is clang, llvm, "
-             "or none">;
+             "or none">, Values<"none,clang,llvm">;
 def fprofile_instrument_path_EQ : Joined<["-"], "fprofile-instrument-path=">,
     HelpText<"Generate instrumented code to collect execution counts into "
              "<file> (overridden by LLVM_PROFILE_FILE env var)">;
@@ -347,9 +348,9 @@ def diagnostic_serialized_file : Separate<["-"], "serialize-diagnostic-file">,
   HelpText<"File for serializing diagnostics in a binary format">;
 
 def fdiagnostics_format : Separate<["-"], "fdiagnostics-format">,
-  HelpText<"Change diagnostic formatting to match IDE and command line tools">;
+  HelpText<"Change diagnostic formatting to match IDE and command line tools">, Values<"clang,msvc,msvc-fallback,vi">;
 def fdiagnostics_show_category : Separate<["-"], "fdiagnostics-show-category">,
-  HelpText<"Print diagnostic category">;
+  HelpText<"Print diagnostic category">, Values<"none,id,name">;
 def fno_diagnostics_use_presumed_location : Flag<["-"], "fno-diagnostics-use-presumed-location">,
   HelpText<"Ignore #line directives when displaying diagnostic locations">;
 def ftabstop : Separate<["-"], "ftabstop">, MetaVarName<"<N>">,
@@ -594,11 +595,11 @@ def fconstant_string_class : Separate<["-"], "fconstant-string-class">,
   MetaVarName<"<class name>">,
   HelpText<"Specify the class to use for constant Objective-C string objects.">;
 def fobjc_arc_cxxlib_EQ : Joined<["-"], "fobjc-arc-cxxlib=">,
-  HelpText<"Objective-C++ Automatic Reference Counting standard library kind">;
+  HelpText<"Objective-C++ Automatic Reference Counting standard library kind">, Values<"libc++,libstdc++,none">;
 def fobjc_runtime_has_weak : Flag<["-"], "fobjc-runtime-has-weak">,
   HelpText<"The target Objective-C runtime supports ARC weak operations">;
 def fobjc_dispatch_method_EQ : Joined<["-"], "fobjc-dispatch-method=">,
-  HelpText<"Objective-C dispatch method to use">;
+  HelpText<"Objective-C dispatch method to use">, Values<"legacy,non-legacy,mixed">;
 def disable_objc_default_synthesize_properties : Flag<["-"], "disable-objc-default-synthesize-properties">,
   HelpText<"disable the default synthesis of Objective-C properties">;
 def fencode_extended_block_signature : Flag<["-"], "fencode-extended-block-signature">,
@@ -672,7 +673,7 @@ def fnative_half_arguments_and_returns : Flag<["-"], "fnative-half-arguments-and
 def fallow_half_arguments_and_returns : Flag<["-"], "fallow-half-arguments-and-returns">,
   HelpText<"Allow function arguments and returns of type half">;
 def fdefault_calling_conv_EQ : Joined<["-"], "fdefault-calling-conv=">,
-  HelpText<"Set default MS calling convention">;
+  HelpText<"Set default MS calling convention">, Values<"cdecl,fastcall,stdcall,vectorcall">;
 def finclude_default_header : Flag<["-"], "finclude-default-header">,
   HelpText<"Include the default header file for OpenCL">;
 def fpreserve_vec3_type : Flag<["-"], "fpreserve-vec3-type">,
diff --git a/include/clang/Driver/Options.h b/include/clang/Driver/Options.h
index 57e4452f3e8c..2da3cb4828c8 100644
--- a/include/clang/Driver/Options.h
+++ b/include/clang/Driver/Options.h
@@ -39,8 +39,9 @@ enum ClangFlags {
 
 enum ID {
     OPT_INVALID = 0, // This is not an option ID.
-#define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM, \
-               HELPTEXT, METAVAR) OPT_##ID,
+#define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM,  \
+               HELPTEXT, METAVAR, VALUES)                                      \
+  OPT_##ID,
 #include "clang/Driver/Options.inc"
     LastOption
 #undef OPTION
diff --git a/include/clang/Driver/Options.td b/include/clang/Driver/Options.td
index 6c51976e98fe..68b331fed2bd 100644
--- a/include/clang/Driver/Options.td
+++ b/include/clang/Driver/Options.td
@@ -493,7 +493,7 @@ def cl_mad_enable : Flag<["-"], "cl-mad-enable">, Group<opencl_Group>, Flags<[CC
 def cl_no_signed_zeros : Flag<["-"], "cl-no-signed-zeros">, Group<opencl_Group>, Flags<[CC1Option]>,
   HelpText<"OpenCL only. Allow use of less precise no signed zeros computations in the generated binary.">;
 def cl_std_EQ : Joined<["-"], "cl-std=">, Group<opencl_Group>, Flags<[CC1Option]>,
-  HelpText<"OpenCL language standard to compile for.">;
+  HelpText<"OpenCL language standard to compile for.">, Values<"cl,CL,cl1.1,CL1.1,cl1.2,CL1.2,cl2.0,CL2.0">;
 def cl_denorms_are_zero : Flag<["-"], "cl-denorms-are-zero">, Group<opencl_Group>, Flags<[CC1Option]>,
   HelpText<"OpenCL only. Allow denormals to be flushed to zero.">;
 def cl_fp32_correctly_rounded_divide_sqrt : Flag<["-"], "cl-fp32-correctly-rounded-divide-sqrt">, Group<opencl_Group>, Flags<[CC1Option]>,
@@ -804,7 +804,7 @@ def fno_sanitize_coverage
     : CommaJoined<["-"], "fno-sanitize-coverage=">,
       Group<f_clang_Group>, Flags<[CoreOption, DriverOption]>,
       HelpText<"Disable specified features of coverage instrumentation for "
-               "Sanitizers">;
+               "Sanitizers">, Values<"func,bb,edge,indirect-calls,trace-bb,trace-cmp,trace-div,trace-gep,8bit-counters,trace-pc,trace-pc-guard,no-prune,inline-8bit-counters">;
 def fsanitize_memory_track_origins_EQ : Joined<["-"], "fsanitize-memory-track-origins=">,
                                         Group<f_clang_Group>,
                                         HelpText<"Enable origins tracking in MemorySanitizer">;
@@ -923,7 +923,7 @@ def ftrapping_math : Flag<["-"], "ftrapping-math">, Group<f_Group>, Flags<[CC1Op
 def fno_trapping_math : Flag<["-"], "fno-trapping-math">, Group<f_Group>, Flags<[CC1Option]>;
 def ffp_contract : Joined<["-"], "ffp-contract=">, Group<f_Group>,
   Flags<[CC1Option]>, HelpText<"Form fused FP ops (e.g. FMAs): fast (everywhere)"
-  " | on (according to FP_CONTRACT pragma, default) | off (never fuse)">;
+  " | on (according to FP_CONTRACT pragma, default) | off (never fuse)">, Values<"fast,on,off">;
 
 def ffor_scope : Flag<["-"], "ffor-scope">, Group<f_Group>;
 def fno_for_scope : Flag<["-"], "fno-for-scope">, Group<f_Group>;
@@ -1000,7 +1000,7 @@ def flat__namespace : Flag<["-"], "flat_namespace">;
 def flax_vector_conversions : Flag<["-"], "flax-vector-conversions">, Group<f_Group>;
 def flimited_precision_EQ : Joined<["-"], "flimited-precision=">, Group<f_Group>;
 def flto_EQ : Joined<["-"], "flto=">, Flags<[CoreOption, CC1Option]>, Group<f_Group>,
-  HelpText<"Set LTO mode to either 'full' or 'thin'">;
+  HelpText<"Set LTO mode to either 'full' or 'thin'">, Values<"thin,full">;
 def flto : Flag<["-"], "flto">, Flags<[CoreOption, CC1Option]>, Group<f_Group>,
   HelpText<"Enable LTO in 'full' mode">;
 def fno_lto : Flag<["-"], "fno-lto">, Group<f_Group>,
@@ -1158,7 +1158,7 @@ def fno_experimental_new_pass_manager : Flag<["-"], "fno-experimental-new-pass-m
   Group<f_clang_Group>, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group<f_Group>, Flags<[CC1Option]>,
-    HelpText<"Use the given vector functions library">;
+    HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group<f_Group>,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group<f_Group>,
@@ -1342,7 +1342,7 @@ def fno_short_wchar : Flag<["-"], "fno-short-wchar">, Group<f_Group>, Flags<[CC1
   HelpText<"Force wchar_t to be an unsigned int">;
 def fshow_overloads_EQ : Joined<["-"], "fshow-overloads=">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Which overload candidates to show when overload resolution fails: "
-           "best|all; defaults to all">;
+           "best|all; defaults to all">, Values<"best,all">;
 def fshow_column : Flag<["-"], "fshow-column">, Group<f_Group>, Flags<[CC1Option]>;
 def fshow_source_location : Flag<["-"], "fshow-source-location">, Group<f_Group>;
 def fspell_checking : Flag<["-"], "fspell-checking">, Group<f_Group>;
@@ -1456,7 +1456,7 @@ def fuse_init_array : Flag<["-"], "fuse-init-array">, Group<f_Group>, Flags<[CC1
 def fno_var_tracking : Flag<["-"], "fno-var-tracking">, Group<clang_ignored_f_Group>;
 def fverbose_asm : Flag<["-"], "fverbose-asm">, Group<f_Group>;
 def fvisibility_EQ : Joined<["-"], "fvisibility=">, Group<f_Group>,
-  HelpText<"Set the default symbol visibility for all global declarations">;
+  HelpText<"Set the default symbol visibility for all global declarations">, Values<"hidden,default">;
 def fvisibility_inlines_hidden : Flag<["-"], "fvisibility-inlines-hidden">, Group<f_Group>,
   HelpText<"Give inline C++ member functions default visibility by default">,
   Flags<[CC1Option]>;
@@ -1563,6 +1563,10 @@ def gdwarf_aranges : Flag<["-"], "gdwarf-aranges">, Group<g_flags_Group>;
 def gmodules : Flag <["-"], "gmodules">, Group<gN_Group>,
   HelpText<"Generate debug info with external references to clang modules"
            " or precompiled headers">;
+def gz : Flag<["-"], "gz">, Group<g_flags_Group>,
+    HelpText<"DWARF debug sections compression type">;
+def gz_EQ : Joined<["-"], "gz=">, Group<g_flags_Group>,
+    HelpText<"DWARF debug sections compression type">;
 def headerpad__max__install__names : Joined<["-"], "headerpad_max_install_names">;
 def help : Flag<["-", "--"], "help">, Flags<[CC1Option,CC1AsOption]>,
   HelpText<"Display available options">;
@@ -1677,7 +1681,7 @@ def minline_all_stringops : Flag<["-"], "minline-all-stringops">, Group<clang_ig
 def mno_inline_all_stringops : Flag<["-"], "mno-inline-all-stringops">, Group<clang_ignored_m_Group>;
 def malign_double : Flag<["-"], "malign-double">, Group<m_Group>, Flags<[CC1Option]>,
   HelpText<"Align doubles to two words in structs (x86 only)">;
-def mfloat_abi_EQ : Joined<["-"], "mfloat-abi=">, Group<m_Group>;
+def mfloat_abi_EQ : Joined<["-"], "mfloat-abi=">, Group<m_Group>, Values<"soft,softfp,hard">;
 def mfpmath_EQ : Joined<["-"], "mfpmath=">, Group<m_Group>;
 def mfpu_EQ : Joined<["-"], "mfpu=">, Group<m_Group>;
 def mhwdiv_EQ : Joined<["-"], "mhwdiv=">, Group<m_Group>;
@@ -1709,9 +1713,9 @@ def mstack_alignment : Joined<["-"], "mstack-alignment=">, Group<m_Group>, Flags
 def mstack_probe_size : Joined<["-"], "mstack-probe-size=">, Group<m_Group>, Flags<[CC1Option]>,
   HelpText<"Set the stack probe size">;
 def mthread_model : Separate<["-"], "mthread-model">, Group<m_Group>, Flags<[CC1Option]>,
-  HelpText<"The thread model to use, e.g. posix, single (posix by default)">;
+  HelpText<"The thread model to use, e.g. posix, single (posix by default)">, Values<"posix,single">;
 def meabi : Separate<["-"], "meabi">, Group<m_Group>, Flags<[CC1Option]>,
-  HelpText<"Set EABI type, e.g. 4, 5 or gnu (default depends on triple)">;
+  HelpText<"Set EABI type, e.g. 4, 5 or gnu (default depends on triple)">, Values<"default,4,5,gnu">;
 
 def mmmx : Flag<["-"], "mmmx">, Group<m_x86_Features_Group>;
 def mno_3dnowa : Flag<["-"], "mno-3dnowa">, Group<m_x86_Features_Group>;
@@ -2205,7 +2209,7 @@ def std_default_EQ : Joined<["-"], "std-default=">;
 def std_EQ : Joined<["-", "--"], "std=">, Flags<[CC1Option]>,
   Group<CompileOnly_Group>, HelpText<"Language standard to compile for">;
 def stdlib_EQ : Joined<["-", "--"], "stdlib=">, Flags<[CC1Option]>,
-  HelpText<"C++ standard library to use">;
+  HelpText<"C++ standard library to use">, Values<"libc++,libstdc++,platform">;
 def sub__library : JoinedOrSeparate<["-"], "sub_library">;
 def sub__umbrella : JoinedOrSeparate<["-"], "sub_umbrella">;
 def system_header_prefix : Joined<["--"], "system-header-prefix=">,
diff --git a/include/clang/Edit/EditedSource.h b/include/clang/Edit/EditedSource.h
index 970791420734..d95a0c2be805 100644
--- a/include/clang/Edit/EditedSource.h
+++ b/include/clang/Edit/EditedSource.h
@@ -17,6 +17,7 @@
 #include "llvm/ADT/TinyPtrVector.h"
 #include "llvm/Support/Allocator.h"
 #include <map>
+#include <tuple>
 
 namespace clang {
   class LangOptions;
@@ -41,10 +42,20 @@ class EditedSource {
   typedef std::map<FileOffset, FileEdit> FileEditsTy;
   FileEditsTy FileEdits;
 
-  // Location of argument use inside the macro body 
-  typedef std::pair<IdentifierInfo*, SourceLocation> MacroArgUse;
-  llvm::DenseMap<unsigned, SmallVector<MacroArgUse, 2>>
-    ExpansionToArgMap;
+  struct MacroArgUse {
+    IdentifierInfo *Identifier;
+    SourceLocation ImmediateExpansionLoc;
+    // Location of argument use inside the top-level macro
+    SourceLocation UseLoc;
+
+    bool operator==(const MacroArgUse &Other) const {
+      return std::tie(Identifier, ImmediateExpansionLoc, UseLoc) ==
+             std::tie(Other.Identifier, Other.ImmediateExpansionLoc,
+                      Other.UseLoc);
+    }
+  };
+
+  llvm::DenseMap<unsigned, SmallVector<MacroArgUse, 2>> ExpansionToArgMap;
   SmallVector<std::pair<SourceLocation, MacroArgUse>, 2>
     CurrCommitMacroArgExps;
 
diff --git a/include/clang/Format/Format.h b/include/clang/Format/Format.h
index 1891c06eba51..c1b62477ed1c 100644
--- a/include/clang/Format/Format.h
+++ b/include/clang/Format/Format.h
@@ -184,9 +184,23 @@ struct FormatStyle {
   enum ShortFunctionStyle {
     /// \brief Never merge functions into a single line.
     SFS_None,
+    /// \brief Only merge functions defined inside a class. Same as "inline",
+    /// except it does not implies "empty": i.e. top level empty functions
+    /// are not merged either.
+    /// \code
+    ///   class Foo {
+    ///     void f() { foo(); }
+    ///   };
+    ///   void f() {
+    ///     foo();
+    ///   }
+    ///   void f() {
+    ///   }
+    /// \endcode
+    SFS_InlineOnly,
     /// \brief Only merge empty functions.
     /// \code
-    ///   void f() { bar(); }
+    ///   void f() {}
     ///   void f2() {
     ///     bar2();
     ///   }
@@ -197,6 +211,10 @@ struct FormatStyle {
     ///   class Foo {
     ///     void f() { foo(); }
     ///   };
+    ///   void f() {
+    ///     foo();
+    ///   }
+    ///   void f() {}
     /// \endcode
     SFS_Inline,
     /// \brief Merge all functions fitting on a single line.
@@ -634,12 +652,12 @@ struct FormatStyle {
     ///   struct foo
     ///   {
     ///     int x;
-    ///   }
+    ///   };
     ///
     ///   false:
     ///   struct foo {
     ///     int x;
-    ///   }
+    ///   };
     /// \endcode
     bool AfterStruct;
     /// \brief Wrap union definitions.
@@ -715,7 +733,7 @@ struct FormatStyle {
   ///        ? firstValue
   ///        : SecondValueVeryVeryVeryVeryLong;
   ///
-  ///    true:
+  ///    false:
   ///    veryVeryVeryVeryVeryVeryVeryVeryVeryVeryVeryLongDescription ?
   ///        firstValue :
   ///        SecondValueVeryVeryVeryVeryLong;
@@ -723,8 +741,7 @@ struct FormatStyle {
   bool BreakBeforeTernaryOperators;
 
   /// \brief Different ways to break initializers.
-  enum BreakConstructorInitializersStyle
-  {
+  enum BreakConstructorInitializersStyle {
     /// Break constructor initializers before the colon and after the commas.
     /// \code
     /// Constructor()
@@ -749,7 +766,7 @@ struct FormatStyle {
     BCIS_AfterColon
   };
 
-  /// \brief The constructor initializers style to use..
+  /// \brief The constructor initializers style to use.
   BreakConstructorInitializersStyle BreakConstructorInitializers;
 
   /// \brief Break after each annotation on a field in Java files.
@@ -1253,6 +1270,14 @@ struct FormatStyle {
   /// \endcode
   bool SortIncludes;
 
+  /// \brief If ``true``, clang-format will sort using declarations.
+  /// \code
+  ///    false:                                 true:
+  ///    using std::cout;               vs.     using std::cin;
+  ///    using std::cin;                        using std::cout;
+  /// \endcode
+  bool SortUsingDeclarations;
+
   /// \brief If ``true``, a space is inserted after C style casts.
   /// \code
   ///    true:                                  false:
@@ -1641,6 +1666,16 @@ tooling::Replacements fixNamespaceEndComments(const FormatStyle &Style,
                                               ArrayRef<tooling::Range> Ranges,
                                               StringRef FileName = "<stdin>");
 
+/// \brief Sort consecutive using declarations in the given \p Ranges in
+/// \p Code.
+///
+/// Returns the ``Replacements`` that sort the using declarations in all
+/// \p Ranges in \p Code.
+tooling::Replacements sortUsingDeclarations(const FormatStyle &Style,
+                                            StringRef Code,
+                                            ArrayRef<tooling::Range> Ranges,
+                                            StringRef FileName = "<stdin>");
+
 /// \brief Returns the ``LangOpts`` that the formatter expects you to set.
 ///
 /// \param Style determines specific settings for lexing mode.
diff --git a/include/clang/Frontend/ASTUnit.h b/include/clang/Frontend/ASTUnit.h
index ae54d4151415..2950c31c2d3d 100644
--- a/include/clang/Frontend/ASTUnit.h
+++ b/include/clang/Frontend/ASTUnit.h
@@ -25,6 +25,7 @@
 #include "clang/Lex/PreprocessingRecord.h"
 #include "clang/Sema/CodeCompleteConsumer.h"
 #include "clang/Serialization/ASTBitCodes.h"
+#include "clang/Frontend/PrecompiledPreamble.h"
 #include "llvm/ADT/IntrusiveRefCntPtr.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
@@ -199,103 +200,15 @@ private:
   /// of that loading. It must be cleared when preamble is recreated.
   llvm::StringMap<SourceLocation> PreambleSrcLocCache;
 
-public:
-  class PreambleData {
-    const FileEntry *File;
-    std::vector<char> Buffer;
-    mutable unsigned NumLines;
-    
-  public:
-    PreambleData() : File(nullptr), NumLines(0) { }
-    
-    void assign(const FileEntry *F, const char *begin, const char *end) {
-      File = F;
-      Buffer.assign(begin, end);
-      NumLines = 0;
-    }
-
-    void clear() { Buffer.clear(); File = nullptr; NumLines = 0; }
-
-    size_t size() const { return Buffer.size(); }
-    bool empty() const { return Buffer.empty(); }
-
-    const char *getBufferStart() const { return &Buffer[0]; }
-
-    unsigned getNumLines() const {
-      if (NumLines)
-        return NumLines;
-      countLines();
-      return NumLines;
-    }
-
-    SourceRange getSourceRange(const SourceManager &SM) const {
-      SourceLocation FileLoc = SM.getLocForStartOfFile(SM.getPreambleFileID());
-      return SourceRange(FileLoc, FileLoc.getLocWithOffset(size()-1));
-    }
-
-  private:
-    void countLines() const;
-  };
-
-  const PreambleData &getPreambleData() const {
-    return Preamble;
-  }
-
-  /// Data used to determine if a file used in the preamble has been changed.
-  struct PreambleFileHash {
-    /// All files have size set.
-    off_t Size;
-
-    /// Modification time is set for files that are on disk.  For memory
-    /// buffers it is zero.
-    time_t ModTime;
-
-    /// Memory buffers have MD5 instead of modification time.  We don't
-    /// compute MD5 for on-disk files because we hope that modification time is
-    /// enough to tell if the file was changed.
-    llvm::MD5::MD5Result MD5;
-
-    static PreambleFileHash createForFile(off_t Size, time_t ModTime);
-    static PreambleFileHash
-    createForMemoryBuffer(const llvm::MemoryBuffer *Buffer);
-
-    friend bool operator==(const PreambleFileHash &LHS,
-                           const PreambleFileHash &RHS);
-
-    friend bool operator!=(const PreambleFileHash &LHS,
-                           const PreambleFileHash &RHS) {
-      return !(LHS == RHS);
-    }
-  };
-
 private:
-  /// \brief The contents of the preamble that has been precompiled to
-  /// \c PreambleFile.
-  PreambleData Preamble;
-
-  /// \brief Whether the preamble ends at the start of a new line.
-  /// 
-  /// Used to inform the lexer as to whether it's starting at the beginning of
-  /// a line after skipping the preamble.
-  bool PreambleEndsAtStartOfLine;
-
-  /// \brief Keeps track of the files that were used when computing the 
-  /// preamble, with both their buffer size and their modification time.
-  ///
-  /// If any of the files have changed from one compile to the next,
-  /// the preamble must be thrown away.
-  llvm::StringMap<PreambleFileHash> FilesInPreamble;
+  /// The contents of the preamble.
+  llvm::Optional<PrecompiledPreamble> Preamble;
 
   /// \brief When non-NULL, this is the buffer used to store the contents of
   /// the main file when it has been padded for use with the precompiled
   /// preamble.
   std::unique_ptr<llvm::MemoryBuffer> SavedMainFileBuffer;
 
-  /// \brief When non-NULL, this is the buffer used to store the
-  /// contents of the preamble when it has been padded to build the
-  /// precompiled preamble.
-  std::unique_ptr<llvm::MemoryBuffer> PreambleBuffer;
-
   /// \brief The number of warnings that occurred while parsing the preamble.
   ///
   /// This value will be used to restore the state of the \c DiagnosticsEngine
@@ -438,21 +351,6 @@ private:
              std::unique_ptr<llvm::MemoryBuffer> OverrideMainBuffer,
              IntrusiveRefCntPtr<vfs::FileSystem> VFS);
 
-  struct ComputedPreamble {
-    llvm::MemoryBuffer *Buffer;
-    std::unique_ptr<llvm::MemoryBuffer> Owner;
-    unsigned Size;
-    bool PreambleEndsAtStartOfLine;
-    ComputedPreamble(llvm::MemoryBuffer *Buffer,
-                     std::unique_ptr<llvm::MemoryBuffer> Owner, unsigned Size,
-                     bool PreambleEndsAtStartOfLine)
-        : Buffer(Buffer), Owner(std::move(Owner)), Size(Size),
-          PreambleEndsAtStartOfLine(PreambleEndsAtStartOfLine) {}
-  };
-  ComputedPreamble ComputePreamble(CompilerInvocation &Invocation,
-                                   unsigned MaxLines,
-                                   IntrusiveRefCntPtr<vfs::FileSystem> VFS);
-
   std::unique_ptr<llvm::MemoryBuffer> getMainBufferWithPrecompiledPreamble(
       std::shared_ptr<PCHContainerOperations> PCHContainerOps,
       const CompilerInvocation &PreambleInvocationIn,
@@ -607,12 +505,6 @@ public:
   void findFileRegionDecls(FileID File, unsigned Offset, unsigned Length,
                            SmallVectorImpl<Decl *> &Decls);
 
-  /// \brief Add a new top-level declaration, identified by its ID in
-  /// the precompiled preamble.
-  void addTopLevelDeclFromPreamble(serialization::DeclID D) {
-    TopLevelDeclsInPreamble.push_back(D);
-  }
-
   /// \brief Retrieve a reference to the current top-level name hash value.
   ///
   /// Note: This is used internally by the top-level tracking action
diff --git a/include/clang/Frontend/PrecompiledPreamble.h b/include/clang/Frontend/PrecompiledPreamble.h
new file mode 100644
index 000000000000..8307392e7feb
--- /dev/null
+++ b/include/clang/Frontend/PrecompiledPreamble.h
@@ -0,0 +1,248 @@
+//===--- PrecompiledPreamble.h - Build precompiled preambles ----*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Helper class to build precompiled preamble.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_FRONTEND_PRECOMPILED_PREAMBLE_H
+#define LLVM_CLANG_FRONTEND_PRECOMPILED_PREAMBLE_H
+
+#include "clang/Lex/Lexer.h"
+#include "clang/Lex/Preprocessor.h"
+#include "llvm/ADT/IntrusiveRefCntPtr.h"
+#include "llvm/Support/MD5.h"
+#include <memory>
+#include <system_error>
+#include <type_traits>
+
+namespace llvm {
+class MemoryBuffer;
+}
+
+namespace clang {
+namespace vfs {
+class FileSystem;
+}
+
+class CompilerInstance;
+class CompilerInvocation;
+class DeclGroupRef;
+class PCHContainerOperations;
+
+/// A size of the preamble and a flag required by
+/// PreprocessorOptions::PrecompiledPreambleBytes.
+struct PreambleBounds {
+  PreambleBounds(unsigned Size, bool PreambleEndsAtStartOfLine)
+      : Size(Size), PreambleEndsAtStartOfLine(PreambleEndsAtStartOfLine) {}
+
+  /// \brief Size of the preamble in bytes.
+  unsigned Size;
+  /// \brief Whether the preamble ends at the start of a new line.
+  ///
+  /// Used to inform the lexer as to whether it's starting at the beginning of
+  /// a line after skipping the preamble.
+  bool PreambleEndsAtStartOfLine;
+};
+
+/// \brief Runs lexer to compute suggested preamble bounds.
+PreambleBounds ComputePreambleBounds(const LangOptions &LangOpts,
+                                     llvm::MemoryBuffer *Buffer,
+                                     unsigned MaxLines);
+
+class PreambleCallbacks;
+
+/// A class holding a PCH and all information to check whether it is valid to
+/// reuse the PCH for the subsequent runs. Use BuildPreamble to create PCH and
+/// CanReusePreamble + AddImplicitPreamble to make use of it.
+class PrecompiledPreamble {
+  class TempPCHFile;
+  struct PreambleFileHash;
+
+public:
+  /// \brief Try to build PrecompiledPreamble for \p Invocation. See
+  /// BuildPreambleError for possible error codes.
+  ///
+  /// \param Invocation Original CompilerInvocation with options to compile the
+  /// file.
+  ///
+  /// \param MainFileBuffer Buffer with the contents of the main file.
+  ///
+  /// \param Bounds Bounds of the preamble, result of calling
+  /// ComputePreambleBounds.
+  ///
+  /// \param Diagnostics Diagnostics engine to be used while building the
+  /// preamble.
+  ///
+  /// \param VFS An instance of vfs::FileSystem to be used for file
+  /// accesses.
+  ///
+  /// \param PCHContainerOps An instance of PCHContainerOperations.
+  ///
+  /// \param Callbacks A set of callbacks to be executed when building
+  /// the preamble.
+  static llvm::ErrorOr<PrecompiledPreamble>
+  Build(const CompilerInvocation &Invocation,
+        const llvm::MemoryBuffer *MainFileBuffer, PreambleBounds Bounds,
+        DiagnosticsEngine &Diagnostics, IntrusiveRefCntPtr<vfs::FileSystem> VFS,
+        std::shared_ptr<PCHContainerOperations> PCHContainerOps,
+        PreambleCallbacks &Callbacks);
+
+  PrecompiledPreamble(PrecompiledPreamble &&) = default;
+  PrecompiledPreamble &operator=(PrecompiledPreamble &&) = default;
+
+  /// PreambleBounds used to build the preamble
+  PreambleBounds getBounds() const;
+
+  /// Check whether PrecompiledPreamble can be reused for the new contents(\p
+  /// MainFileBuffer) of the main file.
+  bool CanReuse(const CompilerInvocation &Invocation,
+                const llvm::MemoryBuffer *MainFileBuffer, PreambleBounds Bounds,
+                vfs::FileSystem *VFS) const;
+
+  /// Changes options inside \p CI to use PCH from this preamble. Also remaps
+  /// main file to \p MainFileBuffer.
+  void AddImplicitPreamble(CompilerInvocation &CI,
+                           llvm::MemoryBuffer *MainFileBuffer) const;
+
+private:
+  PrecompiledPreamble(TempPCHFile PCHFile, std::vector<char> PreambleBytes,
+                      bool PreambleEndsAtStartOfLine,
+                      llvm::StringMap<PreambleFileHash> FilesInPreamble);
+
+  /// A temp file that would be deleted on destructor call. If destructor is not
+  /// called for any reason, the file will be deleted at static objects'
+  /// destruction.
+  /// An assertion will fire if two TempPCHFiles are created with the same name,
+  /// so it's not intended to be used outside preamble-handling.
+  class TempPCHFile {
+  public:
+    // A main method used to construct TempPCHFile.
+    static llvm::ErrorOr<TempPCHFile> CreateNewPreamblePCHFile();
+
+    /// Call llvm::sys::fs::createTemporaryFile to create a new temporary file.
+    static llvm::ErrorOr<TempPCHFile> createInSystemTempDir(const Twine &Prefix,
+                                                            StringRef Suffix);
+    /// Create a new instance of TemporaryFile for file at \p Path. Use with
+    /// extreme caution, there's an assertion checking that there's only a
+    /// single instance of TempPCHFile alive for each path.
+    static llvm::ErrorOr<TempPCHFile> createFromCustomPath(const Twine &Path);
+
+  private:
+    TempPCHFile(std::string FilePath);
+
+  public:
+    TempPCHFile(TempPCHFile &&Other);
+    TempPCHFile &operator=(TempPCHFile &&Other);
+
+    TempPCHFile(const TempPCHFile &) = delete;
+    ~TempPCHFile();
+
+    /// A path where temporary file is stored.
+    llvm::StringRef getFilePath() const;
+
+  private:
+    void RemoveFileIfPresent();
+
+  private:
+    llvm::Optional<std::string> FilePath;
+  };
+
+  /// Data used to determine if a file used in the preamble has been changed.
+  struct PreambleFileHash {
+    /// All files have size set.
+    off_t Size = 0;
+
+    /// Modification time is set for files that are on disk.  For memory
+    /// buffers it is zero.
+    time_t ModTime = 0;
+
+    /// Memory buffers have MD5 instead of modification time.  We don't
+    /// compute MD5 for on-disk files because we hope that modification time is
+    /// enough to tell if the file was changed.
+    llvm::MD5::MD5Result MD5 = {};
+
+    static PreambleFileHash createForFile(off_t Size, time_t ModTime);
+    static PreambleFileHash
+    createForMemoryBuffer(const llvm::MemoryBuffer *Buffer);
+
+    friend bool operator==(const PreambleFileHash &LHS,
+                           const PreambleFileHash &RHS) {
+      return LHS.Size == RHS.Size && LHS.ModTime == RHS.ModTime &&
+             LHS.MD5 == RHS.MD5;
+    }
+    friend bool operator!=(const PreambleFileHash &LHS,
+                           const PreambleFileHash &RHS) {
+      return !(LHS == RHS);
+    }
+  };
+
+  /// Manages the lifetime of temporary file that stores a PCH.
+  TempPCHFile PCHFile;
+  /// Keeps track of the files that were used when computing the
+  /// preamble, with both their buffer size and their modification time.
+  ///
+  /// If any of the files have changed from one compile to the next,
+  /// the preamble must be thrown away.
+  llvm::StringMap<PreambleFileHash> FilesInPreamble;
+  /// The contents of the file that was used to precompile the preamble. Only
+  /// contains first PreambleBounds::Size bytes. Used to compare if the relevant
+  /// part of the file has not changed, so that preamble can be reused.
+  std::vector<char> PreambleBytes;
+  /// See PreambleBounds::PreambleEndsAtStartOfLine
+  bool PreambleEndsAtStartOfLine;
+};
+
+/// A set of callbacks to gather useful information while building a preamble.
+class PreambleCallbacks {
+public:
+  virtual ~PreambleCallbacks() = default;
+
+  /// Called after FrontendAction::Execute(), but before
+  /// FrontendAction::EndSourceFile(). Can be used to transfer ownership of
+  /// various CompilerInstance fields before they are destroyed.
+  virtual void AfterExecute(CompilerInstance &CI);
+  /// Called after PCH has been emitted. \p Writer may be used to retrieve
+  /// information about AST, serialized in PCH.
+  virtual void AfterPCHEmitted(ASTWriter &Writer);
+  /// Called for each TopLevelDecl.
+  /// NOTE: To allow more flexibility a custom ASTConsumer could probably be
+  /// used instead, but having only this method allows a simpler API.
+  virtual void HandleTopLevelDecl(DeclGroupRef DG);
+  /// Called for each macro defined in the Preamble.
+  /// NOTE: To allow more flexibility a custom PPCallbacks could probably be
+  /// used instead, but having only this method allows a simpler API.
+  virtual void HandleMacroDefined(const Token &MacroNameTok,
+                                  const MacroDirective *MD);
+};
+
+enum class BuildPreambleError {
+  PreambleIsEmpty = 1,
+  CouldntCreateTempFile,
+  CouldntCreateTargetInfo,
+  CouldntCreateVFSOverlay,
+  BeginSourceFileFailed,
+  CouldntEmitPCH
+};
+
+class BuildPreambleErrorCategory final : public std::error_category {
+public:
+  const char *name() const noexcept override;
+  std::string message(int condition) const override;
+};
+
+std::error_code make_error_code(BuildPreambleError Error);
+} // namespace clang
+
+namespace std {
+template <>
+struct is_error_code_enum<clang::BuildPreambleError> : std::true_type {};
+} // namespace std
+
+#endif
diff --git a/include/clang/Lex/Preprocessor.h b/include/clang/Lex/Preprocessor.h
index 302d006ea16c..712e1ab9fbf5 100644
--- a/include/clang/Lex/Preprocessor.h
+++ b/include/clang/Lex/Preprocessor.h
@@ -1835,11 +1835,20 @@ private:
   /// \brief A fast PTH version of SkipExcludedConditionalBlock.
   void PTHSkipExcludedConditionalBlock();
 
+  /// Information about the result for evaluating an expression for a
+  /// preprocessor directive.
+  struct DirectiveEvalResult {
+    /// Whether the expression was evaluated as true or not.
+    bool Conditional;
+    /// True if the expression contained identifiers that were undefined.
+    bool IncludedUndefinedIds;
+  };
+
   /// \brief Evaluate an integer constant expression that may occur after a
-  /// \#if or \#elif directive and return it as a bool.
+  /// \#if or \#elif directive and return a \p DirectiveEvalResult object.
   ///
   /// If the expression is equivalent to "!defined(X)" return X in IfNDefMacro.
-  bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
+  DirectiveEvalResult EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
 
   /// \brief Install the standard preprocessor pragmas:
   /// \#pragma GCC poison/system_header/dependency and \#pragma once.
diff --git a/include/clang/Lex/PreprocessorOptions.h b/include/clang/Lex/PreprocessorOptions.h
index f536be8d8e69..d91c665cf1dd 100644
--- a/include/clang/Lex/PreprocessorOptions.h
+++ b/include/clang/Lex/PreprocessorOptions.h
@@ -96,8 +96,15 @@ public:
   std::string TokenCache;
 
   /// When enabled, preprocessor is in a mode for parsing a single file only.
+  ///
+  /// Disables #includes of other files and if there are unresolved identifiers
+  /// in preprocessor directive conditions it causes all blocks to be parsed so
+  /// that the client can get the maximum amount of information from the parser.
   bool SingleFileParseMode = false;
 
+  /// When enabled, the preprocessor will construct editor placeholder tokens.
+  bool LexEditorPlaceholders = true;
+
   /// \brief True if the SourceManager should report the original file name for
   /// contents of files that were remapped to other files. Defaults to true.
   bool RemappedFilesKeepOriginalName;
@@ -185,6 +192,7 @@ public:
     ImplicitPTHInclude.clear();
     TokenCache.clear();
     SingleFileParseMode = false;
+    LexEditorPlaceholders = true;
     RetainRemappedFileBuffers = true;
     PrecompiledPreambleBytes.first = 0;
     PrecompiledPreambleBytes.second = 0;
diff --git a/include/clang/Parse/Parser.h b/include/clang/Parse/Parser.h
index a51d3a51d435..21d699ec402e 100644
--- a/include/clang/Parse/Parser.h
+++ b/include/clang/Parse/Parser.h
@@ -604,7 +604,7 @@ public:
   }
 
   /// getTypeAnnotation - Read a parsed type out of an annotation token.
-  static ParsedType getTypeAnnotation(Token &Tok) {
+  static ParsedType getTypeAnnotation(const Token &Tok) {
     return ParsedType::getFromOpaquePtr(Tok.getAnnotationValue());
   }
 
@@ -615,7 +615,7 @@ private:
 
   /// \brief Read an already-translated primary expression out of an annotation
   /// token.
-  static ExprResult getExprAnnotation(Token &Tok) {
+  static ExprResult getExprAnnotation(const Token &Tok) {
     return ExprResult::getFromOpaquePointer(Tok.getAnnotationValue());
   }
 
@@ -1852,6 +1852,7 @@ private:
     DSC_trailing, // C++11 trailing-type-specifier in a trailing return type
     DSC_alias_declaration, // C++11 type-specifier-seq in an alias-declaration
     DSC_top_level, // top-level/namespace declaration context
+    DSC_template_param, // template parameter context
     DSC_template_type_arg, // template type argument context
     DSC_objc_method_result, // ObjC method result context, enables 'instancetype'
     DSC_condition // condition declaration context
@@ -1862,6 +1863,7 @@ private:
   static bool isTypeSpecifier(DeclSpecContext DSC) {
     switch (DSC) {
     case DSC_normal:
+    case DSC_template_param:
     case DSC_class:
     case DSC_top_level:
     case DSC_objc_method_result:
@@ -1882,6 +1884,7 @@ private:
   static bool isClassTemplateDeductionContext(DeclSpecContext DSC) {
     switch (DSC) {
     case DSC_normal:
+    case DSC_template_param:
     case DSC_class:
     case DSC_top_level:
     case DSC_condition:
diff --git a/include/clang/Sema/Sema.h b/include/clang/Sema/Sema.h
index 1dedf4ba24ce..02c133d1c4fc 100644
--- a/include/clang/Sema/Sema.h
+++ b/include/clang/Sema/Sema.h
@@ -2153,7 +2153,8 @@ public:
                  bool &OwnedDecl, bool &IsDependent,
                  SourceLocation ScopedEnumKWLoc,
                  bool ScopedEnumUsesClassTag, TypeResult UnderlyingType,
-                 bool IsTypeSpecifier, SkipBodyInfo *SkipBody = nullptr);
+                 bool IsTypeSpecifier, bool IsTemplateParamOrArg,
+                 SkipBodyInfo *SkipBody = nullptr);
 
   Decl *ActOnTemplatedFriendTag(Scope *S, SourceLocation FriendLoc,
                                 unsigned TagSpec, SourceLocation TagLoc,
diff --git a/include/clang/StaticAnalyzer/Core/PathSensitive/BasicValueFactory.h b/include/clang/StaticAnalyzer/Core/PathSensitive/BasicValueFactory.h
index fb427f618575..b8ec2aa6ae8d 100644
--- a/include/clang/StaticAnalyzer/Core/PathSensitive/BasicValueFactory.h
+++ b/include/clang/StaticAnalyzer/Core/PathSensitive/BasicValueFactory.h
@@ -180,6 +180,11 @@ public:
     return getValue(X);
   }
 
+  inline const llvm::APSInt& getZeroWithTypeSize(QualType T) {
+    assert(T->isScalarType());
+    return getValue(0, Ctx.getTypeSize(T), true);
+  }
+
   inline const llvm::APSInt& getZeroWithPtrWidth(bool isUnsigned = true) {
     return getValue(0, Ctx.getTypeSize(Ctx.VoidPtrTy), isUnsigned);
   }
diff --git a/include/clang/StaticAnalyzer/Core/PathSensitive/SValBuilder.h b/include/clang/StaticAnalyzer/Core/PathSensitive/SValBuilder.h
index 14aa3af37620..d58d0a690c88 100644
--- a/include/clang/StaticAnalyzer/Core/PathSensitive/SValBuilder.h
+++ b/include/clang/StaticAnalyzer/Core/PathSensitive/SValBuilder.h
@@ -315,6 +315,13 @@ public:
     return nonloc::ConcreteInt(BasicVals.getTruthValue(b));
   }
 
+  /// Create NULL pointer, with proper pointer bit-width for given address
+  /// space.
+  /// \param type pointer type.
+  Loc makeNullWithType(QualType type) {
+    return loc::ConcreteInt(BasicVals.getZeroWithTypeSize(type));
+  }
+
   Loc makeNull() {
     return loc::ConcreteInt(BasicVals.getZeroWithPtrWidth());
   }
diff --git a/lib/AST/ASTContext.cpp b/lib/AST/ASTContext.cpp
index 2300801c1a9c..fabfdc9ef7e5 100644
--- a/lib/AST/ASTContext.cpp
+++ b/lib/AST/ASTContext.cpp
@@ -894,7 +894,7 @@ void ASTContext::mergeDefinitionIntoModule(NamedDecl *ND, Module *M,
   if (getLangOpts().ModulesLocalVisibility)
     MergedDefModules[ND].push_back(M);
   else
-    ND->setHidden(false);
+    ND->setVisibleDespiteOwningModule();
 }
 
 void ASTContext::deduplicateMergedDefinitonsFor(NamedDecl *ND) {
@@ -8513,7 +8513,7 @@ static QualType DecodeTypeFromStr(const char *&Str, const ASTContext &Context,
   RequiresICE = false;
   
   // Read the prefixed modifiers first.
-  bool Done = false;
+  bool Done = false, IsSpecialLong = false;
   while (!Done) {
     switch (*Str++) {
     default: Done = true; --Str; break;
@@ -8531,12 +8531,24 @@ static QualType DecodeTypeFromStr(const char *&Str, const ASTContext &Context,
       Unsigned = true;
       break;
     case 'L':
+      assert(!IsSpecialLong && "Can't use 'L' with 'W' or 'N' modifiers");
       assert(HowLong <= 2 && "Can't have LLLL modifier");
       ++HowLong;
       break;
+    case 'N': {
+      // 'N' behaves like 'L' for all non LP64 targets and 'int' otherwise.
+      assert(!IsSpecialLong && "Can't use two 'N' or 'W' modifiers!");
+      assert(HowLong == 0 && "Can't use both 'L' and 'N' modifiers!");
+      IsSpecialLong = true;
+      if (Context.getTargetInfo().getLongWidth() == 32)
+        ++HowLong;
+      break;
+    }
     case 'W':
       // This modifier represents int64 type.
+      assert(!IsSpecialLong && "Can't use two 'N' or 'W' modifiers!");
       assert(HowLong == 0 && "Can't use both 'L' and 'W' modifiers!");
+      IsSpecialLong = true;
       switch (Context.getTargetInfo().getInt64Type()) {
       default:
         llvm_unreachable("Unexpected integer type");
diff --git a/lib/AST/ASTDumper.cpp b/lib/AST/ASTDumper.cpp
index d89be0d9e6fa..4758109fbcf7 100644
--- a/lib/AST/ASTDumper.cpp
+++ b/lib/AST/ASTDumper.cpp
@@ -1184,6 +1184,31 @@ void ASTDumper::VisitFunctionDecl(const FunctionDecl *D) {
          I != E; ++I)
       dumpCXXCtorInitializer(*I);
 
+  if (const CXXMethodDecl *MD = dyn_cast<CXXMethodDecl>(D))
+    if (MD->size_overridden_methods() != 0) {
+      auto dumpOverride =
+        [=](const CXXMethodDecl *D) {
+          SplitQualType T_split = D->getType().split();
+          OS << D << " " << D->getParent()->getName() << "::";
+          if (isa<CXXDestructorDecl>(D))
+            OS << "~" << D->getParent()->getName();
+          else
+            OS << D->getName();
+          OS << " '" << QualType::getAsString(T_split) << "'";
+        };
+
+      dumpChild([=] {
+        auto FirstOverrideItr = MD->begin_overridden_methods();
+        OS << "Overrides: [ ";
+        dumpOverride(*FirstOverrideItr);
+        for (const auto *Override :
+               llvm::make_range(FirstOverrideItr + 1,
+                                MD->end_overridden_methods()))
+          dumpOverride(Override);
+        OS << " ]";
+      });
+    }
+
   if (D->doesThisDeclarationHaveABody())
     dumpStmt(D->getBody());
 }
diff --git a/lib/AST/ASTImporter.cpp b/lib/AST/ASTImporter.cpp
index 493cb6df8b83..6e33b98d2f18 100644
--- a/lib/AST/ASTImporter.cpp
+++ b/lib/AST/ASTImporter.cpp
@@ -319,6 +319,9 @@ namespace clang {
     bool ImportArrayChecked(const InContainerTy &InContainer, OIter Obegin) {
       return ImportArrayChecked(InContainer.begin(), InContainer.end(), Obegin);
     }
+
+    // Importing overrides.
+    void ImportOverrides(CXXMethodDecl *ToMethod, CXXMethodDecl *FromMethod);
   };
 }
 
@@ -2025,6 +2028,9 @@ Decl *ASTNodeImporter::VisitFunctionDecl(FunctionDecl *D) {
   // Add this function to the lexical context.
   LexicalDC->addDeclInternal(ToFunction);
 
+  if (auto *FromCXXMethod = dyn_cast<CXXMethodDecl>(D))
+    ImportOverrides(cast<CXXMethodDecl>(ToFunction), FromCXXMethod);
+
   return ToFunction;
 }
 
@@ -5499,6 +5505,14 @@ Expr *ASTNodeImporter::VisitSubstNonTypeTemplateParmExpr(
         Replacement);
 }
 
+void ASTNodeImporter::ImportOverrides(CXXMethodDecl *ToMethod,
+                                      CXXMethodDecl *FromMethod) {
+  for (auto *FromOverriddenMethod : FromMethod->overridden_methods())
+    ToMethod->addOverriddenMethod(
+      cast<CXXMethodDecl>(Importer.Import(const_cast<CXXMethodDecl*>(
+                                            FromOverriddenMethod))));
+}
+
 ASTImporter::ASTImporter(ASTContext &ToContext, FileManager &ToFileManager,
                          ASTContext &FromContext, FileManager &FromFileManager,
                          bool MinimalImport)
diff --git a/lib/AST/DeclBase.cpp b/lib/AST/DeclBase.cpp
index 032a20afa834..a0594a020362 100644
--- a/lib/AST/DeclBase.cpp
+++ b/lib/AST/DeclBase.cpp
@@ -278,12 +278,12 @@ void Decl::setLexicalDeclContext(DeclContext *DC) {
   // FIXME: We shouldn't be changing the lexical context of declarations
   // imported from AST files.
   if (!isFromASTFile()) {
-    Hidden = cast<Decl>(DC)->Hidden && hasLocalOwningModuleStorage();
-    if (Hidden)
+    setModuleOwnershipKind(getModuleOwnershipKindForChildOf(DC));
+    if (hasOwningModule())
       setLocalOwningModule(cast<Decl>(DC)->getOwningModule());
   }
 
-  assert((!Hidden || getOwningModule()) &&
+  assert((!hasOwningModule() || getOwningModule()) &&
          "hidden declaration has no owning module");
 }
 
@@ -1352,7 +1352,7 @@ void DeclContext::removeDecl(Decl *D) {
     // Remove only decls that have a name
     if (!ND->getDeclName()) return;
 
-    auto *DC = this;
+    auto *DC = D->getDeclContext();
     do {
       StoredDeclsMap *Map = DC->getPrimaryContext()->LookupPtr;
       if (Map) {
diff --git a/lib/AST/ExternalASTMerger.cpp b/lib/AST/ExternalASTMerger.cpp
index 1dc472a5f753..b746edaf6439 100644
--- a/lib/AST/ExternalASTMerger.cpp
+++ b/lib/AST/ExternalASTMerger.cpp
@@ -41,6 +41,7 @@ public:
   Decl *Imported(Decl *From, Decl *To) override {
     if (auto ToTag = dyn_cast<TagDecl>(To)) {
       ToTag->setHasExternalLexicalStorage();
+      ToTag->setMustBuildLookupTable();
     } else if (auto ToNamespace = dyn_cast<NamespaceDecl>(To)) {
       ToNamespace->setHasExternalVisibleStorage();
     }
diff --git a/lib/Analysis/CloneDetection.cpp b/lib/Analysis/CloneDetection.cpp
index 5bbcbe4e5722..ee848ac711d6 100644
--- a/lib/Analysis/CloneDetection.cpp
+++ b/lib/Analysis/CloneDetection.cpp
@@ -18,9 +18,9 @@
 #include "clang/AST/Stmt.h"
 #include "clang/AST/StmtVisitor.h"
 #include "clang/Lex/Lexer.h"
-#include "llvm/ADT/StringRef.h"
 #include "llvm/Support/MD5.h"
 #include "llvm/Support/raw_ostream.h"
+#include "llvm/Support/Path.h"
 
 using namespace clang;
 
@@ -366,6 +366,23 @@ void OnlyLargestCloneConstraint::constrain(
   }
 }
 
+bool FilenamePatternConstraint::isAutoGenerated(const CloneDetector::CloneGroup &Group) {
+  std::string Error;
+  if (IgnoredFilesPattern.empty() || Group.empty() || 
+      !IgnoredFilesRegex->isValid(Error))
+    return false;
+
+  for (const StmtSequence &S : Group) {
+    const SourceManager &SM = S.getASTContext().getSourceManager();
+    StringRef Filename = llvm::sys::path::filename(SM.getFilename(
+        S.getContainingDecl()->getLocation()));
+    if (IgnoredFilesRegex->match(Filename))
+      return true;
+  }
+
+  return false;
+}
+
 static size_t createHash(llvm::MD5 &Hash) {
   size_t HashCode;
 
@@ -618,8 +635,7 @@ void CloneConstraint::splitCloneGroups(
         if (Indexes[j])
           continue;
 
-        // If a following StmtSequence belongs to our CloneGroup, we add it to
-        // it.
+        // If a following StmtSequence belongs to our CloneGroup, we add it.
         const StmtSequence &Candidate = HashGroup[j];
 
         if (!Compare(Prototype, Candidate))
diff --git a/lib/Basic/Module.cpp b/lib/Basic/Module.cpp
index 83c524877ab0..1d96afd476ef 100644
--- a/lib/Basic/Module.cpp
+++ b/lib/Basic/Module.cpp
@@ -13,6 +13,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "clang/Basic/Module.h"
+#include "clang/Basic/CharInfo.h"
 #include "clang/Basic/FileManager.h"
 #include "clang/Basic/LangOptions.h"
 #include "clang/Basic/TargetInfo.h"
@@ -125,7 +126,36 @@ const Module *Module::getTopLevelModule() const {
   return Result;
 }
 
-std::string Module::getFullModuleName() const {
+static StringRef getModuleNameFromComponent(
+    const std::pair<std::string, SourceLocation> &IdComponent) {
+  return IdComponent.first;
+}
+static StringRef getModuleNameFromComponent(StringRef R) { return R; }
+
+template<typename InputIter>
+static void printModuleId(raw_ostream &OS, InputIter Begin, InputIter End,
+                          bool AllowStringLiterals = true) {
+  for (InputIter It = Begin; It != End; ++It) {
+    if (It != Begin)
+      OS << ".";
+
+    StringRef Name = getModuleNameFromComponent(*It);
+    if (!AllowStringLiterals || isValidIdentifier(Name))
+      OS << Name;
+    else {
+      OS << '"';
+      OS.write_escaped(Name);
+      OS << '"';
+    }
+  }
+}
+
+template<typename Container>
+static void printModuleId(raw_ostream &OS, const Container &C) {
+  return printModuleId(OS, C.begin(), C.end());
+}
+
+std::string Module::getFullModuleName(bool AllowStringLiterals) const {
   SmallVector<StringRef, 2> Names;
   
   // Build up the set of module names (from innermost to outermost).
@@ -133,15 +163,11 @@ std::string Module::getFullModuleName() const {
     Names.push_back(M->Name);
   
   std::string Result;
-  for (SmallVectorImpl<StringRef>::reverse_iterator I = Names.rbegin(),
-                                                 IEnd = Names.rend();
-       I != IEnd; ++I) {
-    if (!Result.empty())
-      Result += '.';
-    
-    Result += *I;
-  }
-  
+
+  llvm::raw_string_ostream Out(Result);
+  printModuleId(Out, Names.rbegin(), Names.rend(), AllowStringLiterals);
+  Out.flush(); 
+
   return Result;
 }
 
@@ -240,14 +266,6 @@ Module *Module::findSubmodule(StringRef Name) const {
   return SubModules[Pos->getValue()];
 }
 
-static void printModuleId(raw_ostream &OS, const ModuleId &Id) {
-  for (unsigned I = 0, N = Id.size(); I != N; ++I) {
-    if (I)
-      OS << ".";
-    OS << Id[I].first;
-  }
-}
-
 void Module::getExportedModules(SmallVectorImpl<Module *> &Exported) const {
   // All non-explicit submodules are exported.
   for (std::vector<Module *>::const_iterator I = SubModules.begin(),
@@ -334,7 +352,8 @@ void Module::print(raw_ostream &OS, unsigned Indent) const {
     OS << "framework ";
   if (IsExplicit)
     OS << "explicit ";
-  OS << "module " << Name;
+  OS << "module ";
+  printModuleId(OS, &Name, &Name + 1);
 
   if (IsSystem || IsExternC) {
     OS.indent(Indent + 2);
@@ -434,7 +453,7 @@ void Module::print(raw_ostream &OS, unsigned Indent) const {
     OS.indent(Indent + 2);
     OS << "export ";
     if (Module *Restriction = Exports[I].getPointer()) {
-      OS << Restriction->getFullModuleName();
+      OS << Restriction->getFullModuleName(true);
       if (Exports[I].getInt())
         OS << ".*";
     } else {
@@ -455,7 +474,7 @@ void Module::print(raw_ostream &OS, unsigned Indent) const {
   for (unsigned I = 0, N = DirectUses.size(); I != N; ++I) {
     OS.indent(Indent + 2);
     OS << "use ";
-    OS << DirectUses[I]->getFullModuleName();
+    OS << DirectUses[I]->getFullModuleName(true);
     OS << "\n";
   }
 
@@ -488,7 +507,7 @@ void Module::print(raw_ostream &OS, unsigned Indent) const {
   for (unsigned I = 0, N = Conflicts.size(); I != N; ++I) {
     OS.indent(Indent + 2);
     OS << "conflict ";
-    OS << Conflicts[I].Other->getFullModuleName();
+    OS << Conflicts[I].Other->getFullModuleName(true);
     OS << ", \"";
     OS.write_escaped(Conflicts[I].Message);
     OS << "\"\n";
diff --git a/lib/Basic/SourceManager.cpp b/lib/Basic/SourceManager.cpp
index fc4c6d303801..3936afab21a4 100644
--- a/lib/Basic/SourceManager.cpp
+++ b/lib/Basic/SourceManager.cpp
@@ -359,15 +359,6 @@ void SourceManager::initializeForReplay(const SourceManager &Old) {
     return Clone;
   };
 
-  // Set up our main file ID as a copy of the old source manager's main file.
-  const SLocEntry &OldMainFile = Old.getSLocEntry(Old.getMainFileID());
-  assert(OldMainFile.isFile() && "main file is macro expansion?");
-  auto *MainCC = CloneContentCache(OldMainFile.getFile().getContentCache());
-  MemBufferInfos.push_back(MainCC);
-  setMainFileID(createFileID(MainCC, SourceLocation(),
-                             OldMainFile.getFile().getFileCharacteristic(),
-                             0, 0));
-
   // Ensure all SLocEntries are loaded from the external source.
   for (unsigned I = 0, N = Old.LoadedSLocEntryTable.size(); I != N; ++I)
     if (!Old.SLocEntryLoaded[I])
diff --git a/lib/Basic/TargetInfo.cpp b/lib/Basic/TargetInfo.cpp
index 8cfd8bde9cbb..4bcebadf458f 100644
--- a/lib/Basic/TargetInfo.cpp
+++ b/lib/Basic/TargetInfo.cpp
@@ -507,6 +507,11 @@ bool TargetInfo::validateOutputConstraint(ConstraintInfo &Info) const {
     case '?': // Disparage slightly code.
     case '!': // Disparage severely.
     case '*': // Ignore for choosing register preferences.
+    case 'i': // Ignore i,n,E,F as output constraints (match from the other
+              // chars)
+    case 'n':
+    case 'E':
+    case 'F':
       break;  // Pass them.
     }
 
diff --git a/lib/Basic/Targets.cpp b/lib/Basic/Targets.cpp
index a3b8330707b9..c5af99e4b6af 100644
--- a/lib/Basic/Targets.cpp
+++ b/lib/Basic/Targets.cpp
@@ -111,6 +111,21 @@ public:
       : OSTargetInfo<Target>(Triple, Opts) {}
 };
 
+// Ananas target
+template<typename Target>
+class AnanasTargetInfo : public OSTargetInfo<Target> {
+protected:
+  void getOSDefines(const LangOptions &Opts, const llvm::Triple &Triple,
+                    MacroBuilder &Builder) const override {
+    // Ananas defines
+    Builder.defineMacro("__Ananas__");
+    Builder.defineMacro("__ELF__");
+  }
+public:
+  AnanasTargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
+      : OSTargetInfo<Target>(Triple, Opts) {}
+};
+
 static void getDarwinDefines(MacroBuilder &Builder, const LangOptions &Opts,
                              const llvm::Triple &Triple,
                              StringRef &PlatformName,
@@ -6172,6 +6187,8 @@ class AArch64TargetInfo : public TargetInfo {
   unsigned Crypto;
   unsigned Unaligned;
   unsigned V8_1A;
+  unsigned V8_2A;
+  unsigned HasFullFP16;
 
   static const Builtin::Info BuiltinInfo[];
 
@@ -6303,6 +6320,8 @@ public:
 
     if (V8_1A)
       Builder.defineMacro("__ARM_FEATURE_QRDMX", "1");
+    if (V8_2A && FPU == NeonMode && HasFullFP16)
+      Builder.defineMacro("__ARM_FEATURE_FP16_VECTOR_ARITHMETIC", "1");
 
     // All of the __sync_(bool|val)_compare_and_swap_(1|2|4|8) builtins work.
     Builder.defineMacro("__GCC_HAVE_SYNC_COMPARE_AND_SWAP_1");
@@ -6330,6 +6349,8 @@ public:
     Crypto = 0;
     Unaligned = 1;
     V8_1A = 0;
+    V8_2A = 0;
+    HasFullFP16 = 0;
 
     for (const auto &Feature : Features) {
       if (Feature == "+neon")
@@ -6342,6 +6363,10 @@ public:
         Unaligned = 0;
       if (Feature == "+v8.1a")
         V8_1A = 1;
+      if (Feature == "+v8.2a")
+        V8_2A = 1;
+      if (Feature == "+fullfp16")
+        HasFullFP16 = 1;
     }
 
     setDataLayout();
@@ -7493,7 +7518,7 @@ public:
     IntPtrType = SignedInt;
     PtrDiffType = SignedInt;
     SigAtomicType = SignedLong;
-    resetDataLayout("e-m:e-p:16:16-i32:16:32-a:16-n8:16");
+    resetDataLayout("e-m:e-p:16:16-i32:16-i64:16-f32:16-f64:16-a:8-n8:16-S16");
   }
   void getTargetDefines(const LangOptions &Opts,
                         MacroBuilder &Builder) const override {
@@ -9520,6 +9545,8 @@ static TargetInfo *AllocateTarget(const llvm::Triple &Triple,
       return new DarwinI386TargetInfo(Triple, Opts);
 
     switch (os) {
+    case llvm::Triple::Ananas:
+      return new AnanasTargetInfo<X86_32TargetInfo>(Triple, Opts);
     case llvm::Triple::CloudABI:
       return new CloudABITargetInfo<X86_32TargetInfo>(Triple, Opts);
     case llvm::Triple::Linux: {
@@ -9575,6 +9602,8 @@ static TargetInfo *AllocateTarget(const llvm::Triple &Triple,
       return new DarwinX86_64TargetInfo(Triple, Opts);
 
     switch (os) {
+    case llvm::Triple::Ananas:
+      return new AnanasTargetInfo<X86_64TargetInfo>(Triple, Opts);
     case llvm::Triple::CloudABI:
       return new CloudABITargetInfo<X86_64TargetInfo>(Triple, Opts);
     case llvm::Triple::Linux: {
diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index 8f0c22d1f7ef..a6451b7fc3c1 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -2956,8 +2956,9 @@ static llvm::VectorType *GetNeonType(CodeGenFunction *CGF,
     return llvm::VectorType::get(CGF->Int8Ty, V1Ty ? 1 : (8 << IsQuad));
   case NeonTypeFlags::Int16:
   case NeonTypeFlags::Poly16:
-  case NeonTypeFlags::Float16:
     return llvm::VectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::Float16:
+    return llvm::VectorType::get(CGF->HalfTy, V1Ty ? 1 : (4 << IsQuad));
   case NeonTypeFlags::Int32:
     return llvm::VectorType::get(CGF->Int32Ty, V1Ty ? 1 : (2 << IsQuad));
   case NeonTypeFlags::Int64:
@@ -2980,6 +2981,8 @@ static llvm::VectorType *GetFloatNeonType(CodeGenFunction *CGF,
                                           NeonTypeFlags IntTypeFlags) {
   int IsQuad = IntTypeFlags.isQuad();
   switch (IntTypeFlags.getEltType()) {
+  case NeonTypeFlags::Int16:
+    return llvm::VectorType::get(CGF->HalfTy, (4 << IsQuad));
   case NeonTypeFlags::Int32:
     return llvm::VectorType::get(CGF->FloatTy, (2 << IsQuad));
   case NeonTypeFlags::Int64:
@@ -3127,55 +3130,80 @@ static const NeonIntrinsicInfo ARMSIMDIntrinsicMap [] = {
   NEONMAP1(vcvt_f16_f32, arm_neon_vcvtfp2hf, 0),
   NEONMAP1(vcvt_f32_f16, arm_neon_vcvthf2fp, 0),
   NEONMAP0(vcvt_f32_v),
+  NEONMAP2(vcvt_n_f16_v, arm_neon_vcvtfxu2fp, arm_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvt_n_f32_v, arm_neon_vcvtfxu2fp, arm_neon_vcvtfxs2fp, 0),
+  NEONMAP1(vcvt_n_s16_v, arm_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvt_n_s32_v, arm_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvt_n_s64_v, arm_neon_vcvtfp2fxs, 0),
+  NEONMAP1(vcvt_n_u16_v, arm_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvt_n_u32_v, arm_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvt_n_u64_v, arm_neon_vcvtfp2fxu, 0),
+  NEONMAP0(vcvt_s16_v),
   NEONMAP0(vcvt_s32_v),
   NEONMAP0(vcvt_s64_v),
+  NEONMAP0(vcvt_u16_v),
   NEONMAP0(vcvt_u32_v),
   NEONMAP0(vcvt_u64_v),
+  NEONMAP1(vcvta_s16_v, arm_neon_vcvtas, 0),
   NEONMAP1(vcvta_s32_v, arm_neon_vcvtas, 0),
   NEONMAP1(vcvta_s64_v, arm_neon_vcvtas, 0),
   NEONMAP1(vcvta_u32_v, arm_neon_vcvtau, 0),
   NEONMAP1(vcvta_u64_v, arm_neon_vcvtau, 0),
+  NEONMAP1(vcvtaq_s16_v, arm_neon_vcvtas, 0),
   NEONMAP1(vcvtaq_s32_v, arm_neon_vcvtas, 0),
   NEONMAP1(vcvtaq_s64_v, arm_neon_vcvtas, 0),
+  NEONMAP1(vcvtaq_u16_v, arm_neon_vcvtau, 0),
   NEONMAP1(vcvtaq_u32_v, arm_neon_vcvtau, 0),
   NEONMAP1(vcvtaq_u64_v, arm_neon_vcvtau, 0),
+  NEONMAP1(vcvtm_s16_v, arm_neon_vcvtms, 0),
   NEONMAP1(vcvtm_s32_v, arm_neon_vcvtms, 0),
   NEONMAP1(vcvtm_s64_v, arm_neon_vcvtms, 0),
+  NEONMAP1(vcvtm_u16_v, arm_neon_vcvtmu, 0),
   NEONMAP1(vcvtm_u32_v, arm_neon_vcvtmu, 0),
   NEONMAP1(vcvtm_u64_v, arm_neon_vcvtmu, 0),
+  NEONMAP1(vcvtmq_s16_v, arm_neon_vcvtms, 0),
   NEONMAP1(vcvtmq_s32_v, arm_neon_vcvtms, 0),
   NEONMAP1(vcvtmq_s64_v, arm_neon_vcvtms, 0),
+  NEONMAP1(vcvtmq_u16_v, arm_neon_vcvtmu, 0),
   NEONMAP1(vcvtmq_u32_v, arm_neon_vcvtmu, 0),
   NEONMAP1(vcvtmq_u64_v, arm_neon_vcvtmu, 0),
+  NEONMAP1(vcvtn_s16_v, arm_neon_vcvtns, 0),
   NEONMAP1(vcvtn_s32_v, arm_neon_vcvtns, 0),
   NEONMAP1(vcvtn_s64_v, arm_neon_vcvtns, 0),
+  NEONMAP1(vcvtn_u16_v, arm_neon_vcvtnu, 0),
   NEONMAP1(vcvtn_u32_v, arm_neon_vcvtnu, 0),
   NEONMAP1(vcvtn_u64_v, arm_neon_vcvtnu, 0),
+  NEONMAP1(vcvtnq_s16_v, arm_neon_vcvtns, 0),
   NEONMAP1(vcvtnq_s32_v, arm_neon_vcvtns, 0),
   NEONMAP1(vcvtnq_s64_v, arm_neon_vcvtns, 0),
+  NEONMAP1(vcvtnq_u16_v, arm_neon_vcvtnu, 0),
   NEONMAP1(vcvtnq_u32_v, arm_neon_vcvtnu, 0),
   NEONMAP1(vcvtnq_u64_v, arm_neon_vcvtnu, 0),
+  NEONMAP1(vcvtp_s16_v, arm_neon_vcvtps, 0),
   NEONMAP1(vcvtp_s32_v, arm_neon_vcvtps, 0),
   NEONMAP1(vcvtp_s64_v, arm_neon_vcvtps, 0),
+  NEONMAP1(vcvtp_u16_v, arm_neon_vcvtpu, 0),
   NEONMAP1(vcvtp_u32_v, arm_neon_vcvtpu, 0),
   NEONMAP1(vcvtp_u64_v, arm_neon_vcvtpu, 0),
+  NEONMAP1(vcvtpq_s16_v, arm_neon_vcvtps, 0),
   NEONMAP1(vcvtpq_s32_v, arm_neon_vcvtps, 0),
   NEONMAP1(vcvtpq_s64_v, arm_neon_vcvtps, 0),
+  NEONMAP1(vcvtpq_u16_v, arm_neon_vcvtpu, 0),
   NEONMAP1(vcvtpq_u32_v, arm_neon_vcvtpu, 0),
   NEONMAP1(vcvtpq_u64_v, arm_neon_vcvtpu, 0),
   NEONMAP0(vcvtq_f32_v),
+  NEONMAP2(vcvtq_n_f16_v, arm_neon_vcvtfxu2fp, arm_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvtq_n_f32_v, arm_neon_vcvtfxu2fp, arm_neon_vcvtfxs2fp, 0),
+  NEONMAP1(vcvtq_n_s16_v, arm_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvtq_n_s32_v, arm_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvtq_n_s64_v, arm_neon_vcvtfp2fxs, 0),
+  NEONMAP1(vcvtq_n_u16_v, arm_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvtq_n_u32_v, arm_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvtq_n_u64_v, arm_neon_vcvtfp2fxu, 0),
+  NEONMAP0(vcvtq_s16_v),
   NEONMAP0(vcvtq_s32_v),
   NEONMAP0(vcvtq_s64_v),
+  NEONMAP0(vcvtq_u16_v),
   NEONMAP0(vcvtq_u32_v),
   NEONMAP0(vcvtq_u64_v),
   NEONMAP0(vext_v),
@@ -3338,19 +3366,27 @@ static const NeonIntrinsicInfo AArch64SIMDIntrinsicMap[] = {
   NEONMAP1(vcnt_v, ctpop, Add1ArgType),
   NEONMAP1(vcntq_v, ctpop, Add1ArgType),
   NEONMAP1(vcvt_f16_f32, aarch64_neon_vcvtfp2hf, 0),
+  NEONMAP0(vcvt_f16_v),
   NEONMAP1(vcvt_f32_f16, aarch64_neon_vcvthf2fp, 0),
   NEONMAP0(vcvt_f32_v),
+  NEONMAP2(vcvt_n_f16_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvt_n_f32_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvt_n_f64_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
+  NEONMAP1(vcvt_n_s16_v, aarch64_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvt_n_s32_v, aarch64_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvt_n_s64_v, aarch64_neon_vcvtfp2fxs, 0),
+  NEONMAP1(vcvt_n_u16_v, aarch64_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvt_n_u32_v, aarch64_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvt_n_u64_v, aarch64_neon_vcvtfp2fxu, 0),
+  NEONMAP0(vcvtq_f16_v),
   NEONMAP0(vcvtq_f32_v),
+  NEONMAP2(vcvtq_n_f16_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvtq_n_f32_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
   NEONMAP2(vcvtq_n_f64_v, aarch64_neon_vcvtfxu2fp, aarch64_neon_vcvtfxs2fp, 0),
+  NEONMAP1(vcvtq_n_s16_v, aarch64_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvtq_n_s32_v, aarch64_neon_vcvtfp2fxs, 0),
   NEONMAP1(vcvtq_n_s64_v, aarch64_neon_vcvtfp2fxs, 0),
+  NEONMAP1(vcvtq_n_u16_v, aarch64_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvtq_n_u32_v, aarch64_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvtq_n_u64_v, aarch64_neon_vcvtfp2fxu, 0),
   NEONMAP1(vcvtx_f32_v, aarch64_neon_fcvtxn, AddRetType | Add1ArgType),
@@ -3819,9 +3855,20 @@ Value *CodeGenFunction::EmitCommonNeonBuiltinExpr(
   case NEON::BI__builtin_neon_vcageq_v:
   case NEON::BI__builtin_neon_vcagt_v:
   case NEON::BI__builtin_neon_vcagtq_v: {
-    llvm::Type *VecFlt = llvm::VectorType::get(
-        VTy->getScalarSizeInBits() == 32 ? FloatTy : DoubleTy,
-        VTy->getNumElements());
+    llvm::Type *Ty;
+    switch (VTy->getScalarSizeInBits()) {
+    default: llvm_unreachable("unexpected type");
+    case 32:
+      Ty = FloatTy;
+      break;
+    case 64:
+      Ty = DoubleTy;
+      break;
+    case 16:
+      Ty = HalfTy;
+      break;
+    }
+    llvm::Type *VecFlt = llvm::VectorType::get(Ty, VTy->getNumElements());
     llvm::Type *Tys[] = { VTy, VecFlt };
     Function *F = CGM.getIntrinsic(LLVMIntrinsic, Tys);
     return EmitNeonCall(F, Ops, NameHint);
@@ -3838,8 +3885,16 @@ Value *CodeGenFunction::EmitCommonNeonBuiltinExpr(
     Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float32, false, Quad));
     return Usgn ? Builder.CreateUIToFP(Ops[0], Ty, "vcvt")
                 : Builder.CreateSIToFP(Ops[0], Ty, "vcvt");
+  case NEON::BI__builtin_neon_vcvt_f16_v:
+  case NEON::BI__builtin_neon_vcvtq_f16_v:
+    Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
+    Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float16, false, Quad));
+    return Usgn ? Builder.CreateUIToFP(Ops[0], Ty, "vcvt")
+                : Builder.CreateSIToFP(Ops[0], Ty, "vcvt");
+  case NEON::BI__builtin_neon_vcvt_n_f16_v:
   case NEON::BI__builtin_neon_vcvt_n_f32_v:
   case NEON::BI__builtin_neon_vcvt_n_f64_v:
+  case NEON::BI__builtin_neon_vcvtq_n_f16_v:
   case NEON::BI__builtin_neon_vcvtq_n_f32_v:
   case NEON::BI__builtin_neon_vcvtq_n_f64_v: {
     llvm::Type *Tys[2] = { GetFloatNeonType(this, Type), Ty };
@@ -3847,11 +3902,15 @@ Value *CodeGenFunction::EmitCommonNeonBuiltinExpr(
     Function *F = CGM.getIntrinsic(Int, Tys);
     return EmitNeonCall(F, Ops, "vcvt_n");
   }
+  case NEON::BI__builtin_neon_vcvt_n_s16_v:
   case NEON::BI__builtin_neon_vcvt_n_s32_v:
+  case NEON::BI__builtin_neon_vcvt_n_u16_v:
   case NEON::BI__builtin_neon_vcvt_n_u32_v:
   case NEON::BI__builtin_neon_vcvt_n_s64_v:
   case NEON::BI__builtin_neon_vcvt_n_u64_v:
+  case NEON::BI__builtin_neon_vcvtq_n_s16_v:
   case NEON::BI__builtin_neon_vcvtq_n_s32_v:
+  case NEON::BI__builtin_neon_vcvtq_n_u16_v:
   case NEON::BI__builtin_neon_vcvtq_n_u32_v:
   case NEON::BI__builtin_neon_vcvtq_n_s64_v:
   case NEON::BI__builtin_neon_vcvtq_n_u64_v: {
@@ -3863,44 +3922,63 @@ Value *CodeGenFunction::EmitCommonNeonBuiltinExpr(
   case NEON::BI__builtin_neon_vcvt_u32_v:
   case NEON::BI__builtin_neon_vcvt_s64_v:
   case NEON::BI__builtin_neon_vcvt_u64_v:
+  case NEON::BI__builtin_neon_vcvt_s16_v:
+  case NEON::BI__builtin_neon_vcvt_u16_v:
   case NEON::BI__builtin_neon_vcvtq_s32_v:
   case NEON::BI__builtin_neon_vcvtq_u32_v:
   case NEON::BI__builtin_neon_vcvtq_s64_v:
-  case NEON::BI__builtin_neon_vcvtq_u64_v: {
+  case NEON::BI__builtin_neon_vcvtq_u64_v:
+  case NEON::BI__builtin_neon_vcvtq_s16_v:
+  case NEON::BI__builtin_neon_vcvtq_u16_v: {
     Ops[0] = Builder.CreateBitCast(Ops[0], GetFloatNeonType(this, Type));
     return Usgn ? Builder.CreateFPToUI(Ops[0], Ty, "vcvt")
                 : Builder.CreateFPToSI(Ops[0], Ty, "vcvt");
   }
+  case NEON::BI__builtin_neon_vcvta_s16_v:
   case NEON::BI__builtin_neon_vcvta_s32_v:
   case NEON::BI__builtin_neon_vcvta_s64_v:
   case NEON::BI__builtin_neon_vcvta_u32_v:
   case NEON::BI__builtin_neon_vcvta_u64_v:
+  case NEON::BI__builtin_neon_vcvtaq_s16_v:
   case NEON::BI__builtin_neon_vcvtaq_s32_v:
   case NEON::BI__builtin_neon_vcvtaq_s64_v:
+  case NEON::BI__builtin_neon_vcvtaq_u16_v:
   case NEON::BI__builtin_neon_vcvtaq_u32_v:
   case NEON::BI__builtin_neon_vcvtaq_u64_v:
+  case NEON::BI__builtin_neon_vcvtn_s16_v:
   case NEON::BI__builtin_neon_vcvtn_s32_v:
   case NEON::BI__builtin_neon_vcvtn_s64_v:
+  case NEON::BI__builtin_neon_vcvtn_u16_v:
   case NEON::BI__builtin_neon_vcvtn_u32_v:
   case NEON::BI__builtin_neon_vcvtn_u64_v:
+  case NEON::BI__builtin_neon_vcvtnq_s16_v:
   case NEON::BI__builtin_neon_vcvtnq_s32_v:
   case NEON::BI__builtin_neon_vcvtnq_s64_v:
+  case NEON::BI__builtin_neon_vcvtnq_u16_v:
   case NEON::BI__builtin_neon_vcvtnq_u32_v:
   case NEON::BI__builtin_neon_vcvtnq_u64_v:
+  case NEON::BI__builtin_neon_vcvtp_s16_v:
   case NEON::BI__builtin_neon_vcvtp_s32_v:
   case NEON::BI__builtin_neon_vcvtp_s64_v:
+  case NEON::BI__builtin_neon_vcvtp_u16_v:
   case NEON::BI__builtin_neon_vcvtp_u32_v:
   case NEON::BI__builtin_neon_vcvtp_u64_v:
+  case NEON::BI__builtin_neon_vcvtpq_s16_v:
   case NEON::BI__builtin_neon_vcvtpq_s32_v:
   case NEON::BI__builtin_neon_vcvtpq_s64_v:
+  case NEON::BI__builtin_neon_vcvtpq_u16_v:
   case NEON::BI__builtin_neon_vcvtpq_u32_v:
   case NEON::BI__builtin_neon_vcvtpq_u64_v:
+  case NEON::BI__builtin_neon_vcvtm_s16_v:
   case NEON::BI__builtin_neon_vcvtm_s32_v:
   case NEON::BI__builtin_neon_vcvtm_s64_v:
+  case NEON::BI__builtin_neon_vcvtm_u16_v:
   case NEON::BI__builtin_neon_vcvtm_u32_v:
   case NEON::BI__builtin_neon_vcvtm_u64_v:
+  case NEON::BI__builtin_neon_vcvtmq_s16_v:
   case NEON::BI__builtin_neon_vcvtmq_s32_v:
   case NEON::BI__builtin_neon_vcvtmq_s64_v:
+  case NEON::BI__builtin_neon_vcvtmq_u16_v:
   case NEON::BI__builtin_neon_vcvtmq_u32_v:
   case NEON::BI__builtin_neon_vcvtmq_u64_v: {
     llvm::Type *Tys[2] = { Ty, GetFloatNeonType(this, Type) };
@@ -6110,7 +6188,9 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     Ops[2] = EmitNeonSplat(Ops[2], cast<ConstantInt>(Ops[3]));
     return Builder.CreateCall(F, {Ops[2], Ops[1], Ops[0]});
   }
+  case NEON::BI__builtin_neon_vfmah_lane_f16:
   case NEON::BI__builtin_neon_vfmas_lane_f32:
+  case NEON::BI__builtin_neon_vfmah_laneq_f16:
   case NEON::BI__builtin_neon_vfmas_laneq_f32:
   case NEON::BI__builtin_neon_vfmad_lane_f64:
   case NEON::BI__builtin_neon_vfmad_laneq_f64: {
@@ -6285,18 +6365,25 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
   case NEON::BI__builtin_neon_vcvt_u32_v:
   case NEON::BI__builtin_neon_vcvt_s64_v:
   case NEON::BI__builtin_neon_vcvt_u64_v:
+	case NEON::BI__builtin_neon_vcvt_s16_v:
+	case NEON::BI__builtin_neon_vcvt_u16_v:
   case NEON::BI__builtin_neon_vcvtq_s32_v:
   case NEON::BI__builtin_neon_vcvtq_u32_v:
   case NEON::BI__builtin_neon_vcvtq_s64_v:
-  case NEON::BI__builtin_neon_vcvtq_u64_v: {
+  case NEON::BI__builtin_neon_vcvtq_u64_v:
+	case NEON::BI__builtin_neon_vcvtq_s16_v:
+	case NEON::BI__builtin_neon_vcvtq_u16_v: {
     Ops[0] = Builder.CreateBitCast(Ops[0], GetFloatNeonType(this, Type));
     if (usgn)
       return Builder.CreateFPToUI(Ops[0], Ty);
     return Builder.CreateFPToSI(Ops[0], Ty);
   }
+  case NEON::BI__builtin_neon_vcvta_s16_v:
   case NEON::BI__builtin_neon_vcvta_s32_v:
+  case NEON::BI__builtin_neon_vcvtaq_s16_v:
   case NEON::BI__builtin_neon_vcvtaq_s32_v:
   case NEON::BI__builtin_neon_vcvta_u32_v:
+  case NEON::BI__builtin_neon_vcvtaq_u16_v:
   case NEON::BI__builtin_neon_vcvtaq_u32_v:
   case NEON::BI__builtin_neon_vcvta_s64_v:
   case NEON::BI__builtin_neon_vcvtaq_s64_v:
@@ -6306,9 +6393,13 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     llvm::Type *Tys[2] = { Ty, GetFloatNeonType(this, Type) };
     return EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vcvta");
   }
+  case NEON::BI__builtin_neon_vcvtm_s16_v:
   case NEON::BI__builtin_neon_vcvtm_s32_v:
+  case NEON::BI__builtin_neon_vcvtmq_s16_v:
   case NEON::BI__builtin_neon_vcvtmq_s32_v:
+  case NEON::BI__builtin_neon_vcvtm_u16_v:
   case NEON::BI__builtin_neon_vcvtm_u32_v:
+  case NEON::BI__builtin_neon_vcvtmq_u16_v:
   case NEON::BI__builtin_neon_vcvtmq_u32_v:
   case NEON::BI__builtin_neon_vcvtm_s64_v:
   case NEON::BI__builtin_neon_vcvtmq_s64_v:
@@ -6318,9 +6409,13 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     llvm::Type *Tys[2] = { Ty, GetFloatNeonType(this, Type) };
     return EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vcvtm");
   }
+  case NEON::BI__builtin_neon_vcvtn_s16_v:
   case NEON::BI__builtin_neon_vcvtn_s32_v:
+  case NEON::BI__builtin_neon_vcvtnq_s16_v:
   case NEON::BI__builtin_neon_vcvtnq_s32_v:
+  case NEON::BI__builtin_neon_vcvtn_u16_v:
   case NEON::BI__builtin_neon_vcvtn_u32_v:
+  case NEON::BI__builtin_neon_vcvtnq_u16_v:
   case NEON::BI__builtin_neon_vcvtnq_u32_v:
   case NEON::BI__builtin_neon_vcvtn_s64_v:
   case NEON::BI__builtin_neon_vcvtnq_s64_v:
@@ -6330,9 +6425,13 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     llvm::Type *Tys[2] = { Ty, GetFloatNeonType(this, Type) };
     return EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vcvtn");
   }
+  case NEON::BI__builtin_neon_vcvtp_s16_v:
   case NEON::BI__builtin_neon_vcvtp_s32_v:
+  case NEON::BI__builtin_neon_vcvtpq_s16_v:
   case NEON::BI__builtin_neon_vcvtpq_s32_v:
+  case NEON::BI__builtin_neon_vcvtp_u16_v:
   case NEON::BI__builtin_neon_vcvtp_u32_v:
+  case NEON::BI__builtin_neon_vcvtpq_u16_v:
   case NEON::BI__builtin_neon_vcvtpq_u32_v:
   case NEON::BI__builtin_neon_vcvtp_s64_v:
   case NEON::BI__builtin_neon_vcvtpq_s64_v:
@@ -6505,6 +6604,24 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vmaxv");
     return Builder.CreateTrunc(Ops[0], Int16Ty);
   }
+  case NEON::BI__builtin_neon_vmaxv_f16: {
+    Int = Intrinsic::aarch64_neon_fmaxv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 4);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vmaxv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vmaxvq_f16: {
+    Int = Intrinsic::aarch64_neon_fmaxv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 8);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vmaxv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
   case NEON::BI__builtin_neon_vminv_u8: {
     Int = Intrinsic::aarch64_neon_uminv;
     Ty = Int32Ty;
@@ -6577,6 +6694,60 @@ Value *CodeGenFunction::EmitAArch64BuiltinExpr(unsigned BuiltinID,
     Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vminv");
     return Builder.CreateTrunc(Ops[0], Int16Ty);
   }
+  case NEON::BI__builtin_neon_vminv_f16: {
+    Int = Intrinsic::aarch64_neon_fminv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 4);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vminv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vminvq_f16: {
+    Int = Intrinsic::aarch64_neon_fminv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 8);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vminv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vmaxnmv_f16: {
+    Int = Intrinsic::aarch64_neon_fmaxnmv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 4);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vmaxnmv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vmaxnmvq_f16: {
+    Int = Intrinsic::aarch64_neon_fmaxnmv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 8);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vmaxnmv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vminnmv_f16: {
+    Int = Intrinsic::aarch64_neon_fminnmv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 4);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vminnmv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
+  case NEON::BI__builtin_neon_vminnmvq_f16: {
+    Int = Intrinsic::aarch64_neon_fminnmv;
+    Ty = HalfTy;
+    VTy = llvm::VectorType::get(HalfTy, 8);
+    llvm::Type *Tys[2] = { Ty, VTy };
+    Ops.push_back(EmitScalarExpr(E->getArg(0)));
+    Ops[0] = EmitNeonCall(CGM.getIntrinsic(Int, Tys), Ops, "vminnmv");
+    return Builder.CreateTrunc(Ops[0], HalfTy);
+  }
   case NEON::BI__builtin_neon_vmul_n_f64: {
     Ops[0] = Builder.CreateBitCast(Ops[0], DoubleTy);
     Value *RHS = Builder.CreateBitCast(EmitScalarExpr(E->getArg(1)), DoubleTy);
diff --git a/lib/CodeGen/CGCall.cpp b/lib/CodeGen/CGCall.cpp
index c65dc18be306..38d520a2beb0 100644
--- a/lib/CodeGen/CGCall.cpp
+++ b/lib/CodeGen/CGCall.cpp
@@ -2906,7 +2906,7 @@ void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
 
   llvm::Instruction *Ret;
   if (RV) {
-    EmitReturnValueCheck(RV, EndLoc);
+    EmitReturnValueCheck(RV);
     Ret = Builder.CreateRet(RV);
   } else {
     Ret = Builder.CreateRetVoid();
@@ -2916,8 +2916,7 @@ void CodeGenFunction::EmitFunctionEpilog(const CGFunctionInfo &FI,
     Ret->setDebugLoc(std::move(RetDbgLoc));
 }
 
-void CodeGenFunction::EmitReturnValueCheck(llvm::Value *RV,
-                                           SourceLocation EndLoc) {
+void CodeGenFunction::EmitReturnValueCheck(llvm::Value *RV) {
   // A current decl may not be available when emitting vtable thunks.
   if (!CurCodeDecl)
     return;
@@ -2950,27 +2949,30 @@ void CodeGenFunction::EmitReturnValueCheck(llvm::Value *RV,
 
   SanitizerScope SanScope(this);
 
-  llvm::BasicBlock *Check = nullptr;
-  llvm::BasicBlock *NoCheck = nullptr;
-  if (requiresReturnValueNullabilityCheck()) {
-    // Before doing the nullability check, make sure that the preconditions for
-    // the check are met.
-    Check = createBasicBlock("nullcheck");
-    NoCheck = createBasicBlock("no.nullcheck");
-    Builder.CreateCondBr(RetValNullabilityPrecondition, Check, NoCheck);
-    EmitBlock(Check);
-  }
+  // Make sure the "return" source location is valid. If we're checking a
+  // nullability annotation, make sure the preconditions for the check are met.
+  llvm::BasicBlock *Check = createBasicBlock("nullcheck");
+  llvm::BasicBlock *NoCheck = createBasicBlock("no.nullcheck");
+  llvm::Value *SLocPtr = Builder.CreateLoad(ReturnLocation, "return.sloc.load");
+  llvm::Value *CanNullCheck = Builder.CreateIsNotNull(SLocPtr);
+  if (requiresReturnValueNullabilityCheck())
+    CanNullCheck =
+        Builder.CreateAnd(CanNullCheck, RetValNullabilityPrecondition);
+  Builder.CreateCondBr(CanNullCheck, Check, NoCheck);
+  EmitBlock(Check);
 
-  // Now do the null check. If the returns_nonnull attribute is present, this
-  // is done unconditionally.
+  // Now do the null check.
   llvm::Value *Cond = Builder.CreateIsNotNull(RV);
-  llvm::Constant *StaticData[] = {
-      EmitCheckSourceLocation(EndLoc), EmitCheckSourceLocation(AttrLoc),
-  };
-  EmitCheck(std::make_pair(Cond, CheckKind), Handler, StaticData, None);
+  llvm::Constant *StaticData[] = {EmitCheckSourceLocation(AttrLoc)};
+  llvm::Value *DynamicData[] = {SLocPtr};
+  EmitCheck(std::make_pair(Cond, CheckKind), Handler, StaticData, DynamicData);
 
-  if (requiresReturnValueNullabilityCheck())
-    EmitBlock(NoCheck);
+  EmitBlock(NoCheck);
+
+#ifndef NDEBUG
+  // The return location should not be used after the check has been emitted.
+  ReturnLocation = Address::invalid();
+#endif
 }
 
 static bool isInAllocaArgument(CGCXXABI &ABI, QualType type) {
@@ -3813,7 +3815,8 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo,
       assert(NumIRArgs == 1);
       if (RV.isScalar() || RV.isComplex()) {
         // Make a temporary alloca to pass the argument.
-        Address Addr = CreateMemTemp(I->Ty, ArgInfo.getIndirectAlign());
+        Address Addr = CreateMemTemp(I->Ty, ArgInfo.getIndirectAlign(),
+                                     "indirect-arg-temp", false);
         IRCallArgs[FirstIRArg] = Addr.getPointer();
 
         LValue argLV = MakeAddrLValue(Addr, I->Ty);
@@ -3842,7 +3845,8 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo,
                < Align.getQuantity()) ||
             (ArgInfo.getIndirectByVal() && (RVAddrSpace != ArgAddrSpace))) {
           // Create an aligned temporary, and copy to it.
-          Address AI = CreateMemTemp(I->Ty, ArgInfo.getIndirectAlign());
+          Address AI = CreateMemTemp(I->Ty, ArgInfo.getIndirectAlign(),
+                                     "byval-temp", false);
           IRCallArgs[FirstIRArg] = AI.getPointer();
           EmitAggregateCopy(AI, Addr, I->Ty, RV.isVolatileQualified());
         } else {
diff --git a/lib/CodeGen/CGClass.cpp b/lib/CodeGen/CGClass.cpp
index 5b4dc5ff0ab3..127d7df348ee 100644
--- a/lib/CodeGen/CGClass.cpp
+++ b/lib/CodeGen/CGClass.cpp
@@ -2770,10 +2770,19 @@ CodeGenFunction::CanDevirtualizeMemberFunctionCall(const Expr *Base,
 
   // We can devirtualize calls on an object accessed by a class member access
   // expression, since by C++11 [basic.life]p6 we know that it can't refer to
-  // a derived class object constructed in the same location.
+  // a derived class object constructed in the same location. However, we avoid
+  // devirtualizing a call to a template function that we could instantiate
+  // implicitly, but have not decided to do so. This is needed because if this
+  // function does not get instantiated, the devirtualization will create a
+  // direct call to a function whose body may not exist. In contrast, calls to
+  // template functions that are not defined in this TU are allowed to be 
+  // devirtualized under assumption that it is user responsibility to
+  // instantiate them in some other TU.
   if (const MemberExpr *ME = dyn_cast<MemberExpr>(Base))
     if (const ValueDecl *VD = dyn_cast<ValueDecl>(ME->getMemberDecl()))
-      return VD->getType()->isRecordType();
+      return VD->getType()->isRecordType() && 
+             (MD->instantiationIsPending() || MD->isDefined() ||
+               !MD->isImplicitlyInstantiable());
 
   // Likewise for calls on an object accessed by a (non-reference) pointer to
   // member access.
diff --git a/lib/CodeGen/CGDecl.cpp b/lib/CodeGen/CGDecl.cpp
index 87bfa507a8c0..ccd3b8d513b1 100644
--- a/lib/CodeGen/CGDecl.cpp
+++ b/lib/CodeGen/CGDecl.cpp
@@ -954,6 +954,7 @@ void CodeGenFunction::EmitLifetimeEnd(llvm::Value *Size, llvm::Value *Addr) {
 CodeGenFunction::AutoVarEmission
 CodeGenFunction::EmitAutoVarAlloca(const VarDecl &D) {
   QualType Ty = D.getType();
+  assert(Ty.getAddressSpace() == LangAS::Default);
 
   AutoVarEmission emission(D);
 
@@ -1046,8 +1047,7 @@ CodeGenFunction::EmitAutoVarAlloca(const VarDecl &D) {
       // Create the alloca.  Note that we set the name separately from
       // building the instruction so that it's there even in no-asserts
       // builds.
-      address = CreateTempAlloca(allocaTy, allocaAlignment);
-      address.getPointer()->setName(D.getName());
+      address = CreateTempAlloca(allocaTy, allocaAlignment, D.getName());
 
       // Don't emit lifetime markers for MSVC catch parameters. The lifetime of
       // the catch parameter starts in the catchpad instruction, and we can't
@@ -1107,27 +1107,9 @@ CodeGenFunction::EmitAutoVarAlloca(const VarDecl &D) {
     llvm::Type *llvmTy = ConvertTypeForMem(elementType);
 
     // Allocate memory for the array.
-    llvm::AllocaInst *vla = Builder.CreateAlloca(llvmTy, elementCount, "vla");
-    vla->setAlignment(alignment.getQuantity());
-
-    address = Address(vla, alignment);
+    address = CreateTempAlloca(llvmTy, alignment, "vla", elementCount);
   }
 
-  // Alloca always returns a pointer in alloca address space, which may
-  // be different from the type defined by the language. For example,
-  // in C++ the auto variables are in the default address space. Therefore
-  // cast alloca to the expected address space when necessary.
-  auto T = D.getType();
-  assert(T.getAddressSpace() == LangAS::Default);
-  if (getASTAllocaAddressSpace() != LangAS::Default) {
-    auto *Addr = getTargetHooks().performAddrSpaceCast(
-        *this, address.getPointer(), getASTAllocaAddressSpace(),
-        T.getAddressSpace(),
-        address.getElementType()->getPointerTo(
-            getContext().getTargetAddressSpace(T.getAddressSpace())),
-        /*non-null*/ true);
-    address = Address(Addr, address.getAlignment());
-  }
   setAddrOfLocalVar(&D, address);
   emission.Addr = address;
 
diff --git a/lib/CodeGen/CGExpr.cpp b/lib/CodeGen/CGExpr.cpp
index 7359006677f4..2ee1c96a6619 100644
--- a/lib/CodeGen/CGExpr.cpp
+++ b/lib/CodeGen/CGExpr.cpp
@@ -61,18 +61,36 @@ llvm::Value *CodeGenFunction::EmitCastToVoidPtr(llvm::Value *value) {
 /// CreateTempAlloca - This creates a alloca and inserts it into the entry
 /// block.
 Address CodeGenFunction::CreateTempAlloca(llvm::Type *Ty, CharUnits Align,
-                                          const Twine &Name) {
-  auto Alloca = CreateTempAlloca(Ty, Name);
+                                          const Twine &Name,
+                                          llvm::Value *ArraySize,
+                                          bool CastToDefaultAddrSpace) {
+  auto Alloca = CreateTempAlloca(Ty, Name, ArraySize);
   Alloca->setAlignment(Align.getQuantity());
-  return Address(Alloca, Align);
+  llvm::Value *V = Alloca;
+  // Alloca always returns a pointer in alloca address space, which may
+  // be different from the type defined by the language. For example,
+  // in C++ the auto variables are in the default address space. Therefore
+  // cast alloca to the default address space when necessary.
+  if (CastToDefaultAddrSpace && getASTAllocaAddressSpace() != LangAS::Default) {
+    auto DestAddrSpace = getContext().getTargetAddressSpace(LangAS::Default);
+    V = getTargetHooks().performAddrSpaceCast(
+        *this, V, getASTAllocaAddressSpace(), LangAS::Default,
+        Ty->getPointerTo(DestAddrSpace), /*non-null*/ true);
+  }
+
+  return Address(V, Align);
 }
 
-/// CreateTempAlloca - This creates a alloca and inserts it into the entry
-/// block.
+/// CreateTempAlloca - This creates an alloca and inserts it into the entry
+/// block if \p ArraySize is nullptr, otherwise inserts it at the current
+/// insertion point of the builder.
 llvm::AllocaInst *CodeGenFunction::CreateTempAlloca(llvm::Type *Ty,
-                                                    const Twine &Name) {
+                                                    const Twine &Name,
+                                                    llvm::Value *ArraySize) {
+  if (ArraySize)
+    return Builder.CreateAlloca(Ty, ArraySize, Name);
   return new llvm::AllocaInst(Ty, CGM.getDataLayout().getAllocaAddrSpace(),
-                              nullptr, Name, AllocaInsertPt);
+                              ArraySize, Name, AllocaInsertPt);
 }
 
 /// CreateDefaultAlignTempAlloca - This creates an alloca with the
@@ -99,14 +117,18 @@ Address CodeGenFunction::CreateIRTemp(QualType Ty, const Twine &Name) {
   return CreateTempAlloca(ConvertType(Ty), Align, Name);
 }
 
-Address CodeGenFunction::CreateMemTemp(QualType Ty, const Twine &Name) {
+Address CodeGenFunction::CreateMemTemp(QualType Ty, const Twine &Name,
+                                       bool CastToDefaultAddrSpace) {
   // FIXME: Should we prefer the preferred type alignment here?
-  return CreateMemTemp(Ty, getContext().getTypeAlignInChars(Ty), Name);
+  return CreateMemTemp(Ty, getContext().getTypeAlignInChars(Ty), Name,
+                       CastToDefaultAddrSpace);
 }
 
 Address CodeGenFunction::CreateMemTemp(QualType Ty, CharUnits Align,
-                                       const Twine &Name) {
-  return CreateTempAlloca(ConvertTypeForMem(Ty), Align, Name);
+                                       const Twine &Name,
+                                       bool CastToDefaultAddrSpace) {
+  return CreateTempAlloca(ConvertTypeForMem(Ty), Align, Name, nullptr,
+                          CastToDefaultAddrSpace);
 }
 
 /// EvaluateExprAsBool - Perform the usual unary conversions on the specified
diff --git a/lib/CodeGen/CGObjCMac.cpp b/lib/CodeGen/CGObjCMac.cpp
index c78d3febd4cd..7976c53d9e98 100644
--- a/lib/CodeGen/CGObjCMac.cpp
+++ b/lib/CodeGen/CGObjCMac.cpp
@@ -308,7 +308,7 @@ public:
     SmallVector<CanQualType,5> Params;
     Params.push_back(Ctx.VoidPtrTy);
     Params.push_back(Ctx.VoidPtrTy);
-    Params.push_back(Ctx.LongTy);
+    Params.push_back(Ctx.getSizeType());
     Params.push_back(Ctx.BoolTy);
     Params.push_back(Ctx.BoolTy);
     llvm::FunctionType *FTy =
diff --git a/lib/CodeGen/CGStmt.cpp b/lib/CodeGen/CGStmt.cpp
index 683f366ebe45..a13c38646164 100644
--- a/lib/CodeGen/CGStmt.cpp
+++ b/lib/CodeGen/CGStmt.cpp
@@ -1024,6 +1024,18 @@ void CodeGenFunction::EmitReturnOfRValue(RValue RV, QualType Ty) {
 /// if the function returns void, or may be missing one if the function returns
 /// non-void.  Fun stuff :).
 void CodeGenFunction::EmitReturnStmt(const ReturnStmt &S) {
+  if (requiresReturnValueCheck()) {
+    llvm::Constant *SLoc = EmitCheckSourceLocation(S.getLocStart());
+    auto *SLocPtr =
+        new llvm::GlobalVariable(CGM.getModule(), SLoc->getType(), false,
+                                 llvm::GlobalVariable::PrivateLinkage, SLoc);
+    SLocPtr->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
+    CGM.getSanitizerMetadata()->disableSanitizerForGlobal(SLocPtr);
+    assert(ReturnLocation.isValid() && "No valid return location");
+    Builder.CreateStore(Builder.CreateBitCast(SLocPtr, Int8PtrTy),
+                        ReturnLocation);
+  }
+
   // Returning from an outlined SEH helper is UB, and we already warn on it.
   if (IsOutlinedSEHHelper) {
     Builder.CreateUnreachable();
diff --git a/lib/CodeGen/CodeGenFunction.cpp b/lib/CodeGen/CodeGenFunction.cpp
index ac1a1334f103..93a4a3866193 100644
--- a/lib/CodeGen/CodeGenFunction.cpp
+++ b/lib/CodeGen/CodeGenFunction.cpp
@@ -860,6 +860,13 @@ void CodeGenFunction::StartFunction(GlobalDecl GD,
 
   Builder.SetInsertPoint(EntryBB);
 
+  // If we're checking the return value, allocate space for a pointer to a
+  // precise source location of the checked return statement.
+  if (requiresReturnValueCheck()) {
+    ReturnLocation = CreateDefaultAlignTempAlloca(Int8PtrTy, "return.sloc.ptr");
+    InitTempAlloca(ReturnLocation, llvm::ConstantPointerNull::get(Int8PtrTy));
+  }
+
   // Emit subprogram debug descriptor.
   if (CGDebugInfo *DI = getDebugInfo()) {
     // Reconstruct the type from the argument list so that implicit parameters,
@@ -887,8 +894,10 @@ void CodeGenFunction::StartFunction(GlobalDecl GD,
   if (CGM.getCodeGenOpts().InstrumentForProfiling) {
     if (CGM.getCodeGenOpts().CallFEntry)
       Fn->addFnAttr("fentry-call", "true");
-    else
-      Fn->addFnAttr("counting-function", getTarget().getMCountName());
+    else {
+      if (!CurFuncDecl || !CurFuncDecl->hasAttr<NoInstrumentFunctionAttr>())
+        Fn->addFnAttr("counting-function", getTarget().getMCountName());
+    }
   }
 
   if (RetTy->isVoidType()) {
diff --git a/lib/CodeGen/CodeGenFunction.h b/lib/CodeGen/CodeGenFunction.h
index 831eedf9e478..6785111bd052 100644
--- a/lib/CodeGen/CodeGenFunction.h
+++ b/lib/CodeGen/CodeGenFunction.h
@@ -116,9 +116,9 @@ enum TypeEvaluationKind {
   SANITIZER_CHECK(MulOverflow, mul_overflow, 0)                                \
   SANITIZER_CHECK(NegateOverflow, negate_overflow, 0)                          \
   SANITIZER_CHECK(NullabilityArg, nullability_arg, 0)                          \
-  SANITIZER_CHECK(NullabilityReturn, nullability_return, 0)                    \
+  SANITIZER_CHECK(NullabilityReturn, nullability_return, 1)                    \
   SANITIZER_CHECK(NonnullArg, nonnull_arg, 0)                                  \
-  SANITIZER_CHECK(NonnullReturn, nonnull_return, 0)                            \
+  SANITIZER_CHECK(NonnullReturn, nonnull_return, 1)                            \
   SANITIZER_CHECK(OutOfBounds, out_of_bounds, 0)                               \
   SANITIZER_CHECK(PointerOverflow, pointer_overflow, 0)                        \
   SANITIZER_CHECK(ShiftOutOfBounds, shift_out_of_bounds, 0)                    \
@@ -1407,6 +1407,17 @@ private:
     return RetValNullabilityPrecondition;
   }
 
+  /// Used to store precise source locations for return statements by the
+  /// runtime return value checks.
+  Address ReturnLocation = Address::invalid();
+
+  /// Check if the return value of this function requires sanitization.
+  bool requiresReturnValueCheck() const {
+    return requiresReturnValueNullabilityCheck() ||
+           (SanOpts.has(SanitizerKind::ReturnsNonnullAttribute) &&
+            CurCodeDecl && CurCodeDecl->getAttr<ReturnsNonNullAttr>());
+  }
+
   llvm::BasicBlock *TerminateLandingPad;
   llvm::BasicBlock *TerminateHandler;
   llvm::BasicBlock *TrapBB;
@@ -1778,7 +1789,7 @@ public:
                           SourceLocation EndLoc);
 
   /// Emit a test that checks if the return value \p RV is nonnull.
-  void EmitReturnValueCheck(llvm::Value *RV, SourceLocation EndLoc);
+  void EmitReturnValueCheck(llvm::Value *RV);
 
   /// EmitStartEHSpec - Emit the start of the exception spec.
   void EmitStartEHSpec(const Decl *D);
@@ -1916,13 +1927,36 @@ public:
                             LValueBaseInfo *BaseInfo = nullptr);
   LValue EmitLoadOfPointerLValue(Address Ptr, const PointerType *PtrTy);
 
-  /// CreateTempAlloca - This creates a alloca and inserts it into the entry
-  /// block. The caller is responsible for setting an appropriate alignment on
+  /// CreateTempAlloca - This creates an alloca and inserts it into the entry
+  /// block if \p ArraySize is nullptr, otherwise inserts it at the current
+  /// insertion point of the builder. The caller is responsible for setting an
+  /// appropriate alignment on
   /// the alloca.
-  llvm::AllocaInst *CreateTempAlloca(llvm::Type *Ty,
-                                     const Twine &Name = "tmp");
+  ///
+  /// \p ArraySize is the number of array elements to be allocated if it
+  ///    is not nullptr.
+  ///
+  /// LangAS::Default is the address space of pointers to local variables and
+  /// temporaries, as exposed in the source language. In certain
+  /// configurations, this is not the same as the alloca address space, and a
+  /// cast is needed to lift the pointer from the alloca AS into
+  /// LangAS::Default. This can happen when the target uses a restricted
+  /// address space for the stack but the source language requires
+  /// LangAS::Default to be a generic address space. The latter condition is
+  /// common for most programming languages; OpenCL is an exception in that
+  /// LangAS::Default is the private address space, which naturally maps
+  /// to the stack.
+  ///
+  /// Because the address of a temporary is often exposed to the program in
+  /// various ways, this function will perform the cast by default. The cast
+  /// may be avoided by passing false as \p CastToDefaultAddrSpace; this is
+  /// more efficient if the caller knows that the address will not be exposed.
+  llvm::AllocaInst *CreateTempAlloca(llvm::Type *Ty, const Twine &Name = "tmp",
+                                     llvm::Value *ArraySize = nullptr);
   Address CreateTempAlloca(llvm::Type *Ty, CharUnits align,
-                           const Twine &Name = "tmp");
+                           const Twine &Name = "tmp",
+                           llvm::Value *ArraySize = nullptr,
+                           bool CastToDefaultAddrSpace = true);
 
   /// CreateDefaultAlignedTempAlloca - This creates an alloca with the
   /// default ABI alignment of the given LLVM type.
@@ -1957,9 +1991,12 @@ public:
   Address CreateIRTemp(QualType T, const Twine &Name = "tmp");
 
   /// CreateMemTemp - Create a temporary memory object of the given type, with
-  /// appropriate alignment.
-  Address CreateMemTemp(QualType T, const Twine &Name = "tmp");
-  Address CreateMemTemp(QualType T, CharUnits Align, const Twine &Name = "tmp");
+  /// appropriate alignment. Cast it to the default address space if
+  /// \p CastToDefaultAddrSpace is true.
+  Address CreateMemTemp(QualType T, const Twine &Name = "tmp",
+                        bool CastToDefaultAddrSpace = true);
+  Address CreateMemTemp(QualType T, CharUnits Align, const Twine &Name = "tmp",
+                        bool CastToDefaultAddrSpace = true);
 
   /// CreateAggTemp - Create a temporary memory object for the given
   /// aggregate type.
diff --git a/lib/CodeGen/CodeGenModule.cpp b/lib/CodeGen/CodeGenModule.cpp
index 19a055075604..5319ccec163f 100644
--- a/lib/CodeGen/CodeGenModule.cpp
+++ b/lib/CodeGen/CodeGenModule.cpp
@@ -98,6 +98,7 @@ CodeGenModule::CodeGenModule(ASTContext &C, const HeaderSearchOptions &HSO,
   Int16Ty = llvm::Type::getInt16Ty(LLVMContext);
   Int32Ty = llvm::Type::getInt32Ty(LLVMContext);
   Int64Ty = llvm::Type::getInt64Ty(LLVMContext);
+  HalfTy = llvm::Type::getHalfTy(LLVMContext);
   FloatTy = llvm::Type::getFloatTy(LLVMContext);
   DoubleTy = llvm::Type::getDoubleTy(LLVMContext);
   PointerWidthInBits = C.getTargetInfo().getPointerWidth(0);
@@ -506,6 +507,26 @@ void CodeGenModule::Release() {
                               LangOpts.CUDADeviceFlushDenormalsToZero ? 1 : 0);
   }
 
+  // Emit OpenCL specific module metadata: OpenCL/SPIR version.
+  if (LangOpts.OpenCL) {
+    EmitOpenCLMetadata();
+    // Emit SPIR version.
+    if (getTriple().getArch() == llvm::Triple::spir ||
+        getTriple().getArch() == llvm::Triple::spir64) {
+      // SPIR v2.0 s2.12 - The SPIR version used by the module is stored in the
+      // opencl.spir.version named metadata.
+      llvm::Metadata *SPIRVerElts[] = {
+          llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
+              Int32Ty, LangOpts.OpenCLVersion / 100)),
+          llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
+              Int32Ty, (LangOpts.OpenCLVersion / 100 > 1) ? 0 : 2))};
+      llvm::NamedMDNode *SPIRVerMD =
+          TheModule.getOrInsertNamedMetadata("opencl.spir.version");
+      llvm::LLVMContext &Ctx = TheModule.getContext();
+      SPIRVerMD->addOperand(llvm::MDNode::get(Ctx, SPIRVerElts));
+    }
+  }
+
   if (uint32_t PLevel = Context.getLangOpts().PICLevel) {
     assert(PLevel < 3 && "Invalid PIC Level");
     getModule().setPICLevel(static_cast<llvm::PICLevel::Level>(PLevel));
@@ -529,6 +550,20 @@ void CodeGenModule::Release() {
   EmitTargetMetadata();
 }
 
+void CodeGenModule::EmitOpenCLMetadata() {
+  // SPIR v2.0 s2.13 - The OpenCL version used by the module is stored in the
+  // opencl.ocl.version named metadata node.
+  llvm::Metadata *OCLVerElts[] = {
+      llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
+          Int32Ty, LangOpts.OpenCLVersion / 100)),
+      llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
+          Int32Ty, (LangOpts.OpenCLVersion % 100) / 10))};
+  llvm::NamedMDNode *OCLVerMD =
+      TheModule.getOrInsertNamedMetadata("opencl.ocl.version");
+  llvm::LLVMContext &Ctx = TheModule.getContext();
+  OCLVerMD->addOperand(llvm::MDNode::get(Ctx, OCLVerElts));
+}
+
 void CodeGenModule::UpdateCompletedType(const TagDecl *TD) {
   // Make sure that this type is translated.
   Types.UpdateCompletedType(TD);
diff --git a/lib/CodeGen/CodeGenModule.h b/lib/CodeGen/CodeGenModule.h
index 59e56a6ba194..c5f1a2b409ee 100644
--- a/lib/CodeGen/CodeGenModule.h
+++ b/lib/CodeGen/CodeGenModule.h
@@ -1321,6 +1321,9 @@ private:
   /// Emits target specific Metadata for global declarations.
   void EmitTargetMetadata();
 
+  /// Emits OpenCL specific Metadata e.g. OpenCL version.
+  void EmitOpenCLMetadata();
+
   /// Emit the llvm.gcov metadata used to tell LLVM where to emit the .gcno and
   /// .gcda files in a way that persists in .bc files.
   void EmitCoverageFile();
diff --git a/lib/CodeGen/CodeGenTypeCache.h b/lib/CodeGen/CodeGenTypeCache.h
index 450eab48a3b4..6910d36733dc 100644
--- a/lib/CodeGen/CodeGenTypeCache.h
+++ b/lib/CodeGen/CodeGenTypeCache.h
@@ -36,7 +36,7 @@ struct CodeGenTypeCache {
   /// i8, i16, i32, and i64
   llvm::IntegerType *Int8Ty, *Int16Ty, *Int32Ty, *Int64Ty;
   /// float, double
-  llvm::Type *FloatTy, *DoubleTy;
+  llvm::Type *HalfTy, *FloatTy, *DoubleTy;
 
   /// int
   llvm::IntegerType *IntTy;
diff --git a/lib/CodeGen/SwiftCallingConv.cpp b/lib/CodeGen/SwiftCallingConv.cpp
index 0bfe30a32c80..fc8e36d2c599 100644
--- a/lib/CodeGen/SwiftCallingConv.cpp
+++ b/lib/CodeGen/SwiftCallingConv.cpp
@@ -57,6 +57,10 @@ static CharUnits getTypeStoreSize(CodeGenModule &CGM, llvm::Type *type) {
   return CharUnits::fromQuantity(CGM.getDataLayout().getTypeStoreSize(type));
 }
 
+static CharUnits getTypeAllocSize(CodeGenModule &CGM, llvm::Type *type) {
+  return CharUnits::fromQuantity(CGM.getDataLayout().getTypeAllocSize(type));
+}
+
 void SwiftAggLowering::addTypedData(QualType type, CharUnits begin) {
   // Deal with various aggregate types as special cases:
 
@@ -542,7 +546,9 @@ SwiftAggLowering::getCoerceAndExpandTypes() const {
       packed = true;
 
     elts.push_back(entry.Type);
-    lastEnd = entry.End;
+
+    lastEnd = entry.Begin + getTypeAllocSize(CGM, entry.Type);
+    assert(entry.End <= lastEnd);
   }
 
   // We don't need to adjust 'packed' to deal with possible tail padding
diff --git a/lib/CodeGen/TargetInfo.cpp b/lib/CodeGen/TargetInfo.cpp
index 427ec06a2fff..8d00e055306d 100644
--- a/lib/CodeGen/TargetInfo.cpp
+++ b/lib/CodeGen/TargetInfo.cpp
@@ -951,8 +951,7 @@ class X86_32ABIInfo : public SwiftABIInfo {
   Class classify(QualType Ty) const;
   ABIArgInfo classifyReturnType(QualType RetTy, CCState &State) const;
   ABIArgInfo classifyArgumentType(QualType RetTy, CCState &State) const;
-  ABIArgInfo reclassifyHvaArgType(QualType RetTy, CCState &State, 
-                                  const ABIArgInfo& current) const;
+
   /// \brief Updates the number of available free registers, returns 
   /// true if any registers were allocated.
   bool updateFreeRegs(QualType Ty, CCState &State) const;
@@ -1536,27 +1535,6 @@ bool X86_32ABIInfo::shouldPrimitiveUseInReg(QualType Ty, CCState &State) const {
   return true;
 }
 
-ABIArgInfo
-X86_32ABIInfo::reclassifyHvaArgType(QualType Ty, CCState &State,
-                                    const ABIArgInfo &current) const {
-  // Assumes vectorCall calling convention.
-  const Type *Base = nullptr;
-  uint64_t NumElts = 0;
-
-  if (!Ty->isBuiltinType() && !Ty->isVectorType() &&
-      isHomogeneousAggregate(Ty, Base, NumElts)) {
-    if (State.FreeSSERegs >= NumElts) {
-      // HVA types get passed directly in registers if there is room.
-      State.FreeSSERegs -= NumElts;
-      return getDirectX86Hva();
-    }
-    // If there's no room, the HVA gets passed as normal indirect
-    // structure.
-    return getIndirectResult(Ty, /*ByVal=*/false, State);
-  } 
-  return current;
-}
-
 ABIArgInfo X86_32ABIInfo::classifyArgumentType(QualType Ty,
                                                CCState &State) const {
   // FIXME: Set alignment on indirect arguments.
@@ -1575,35 +1553,20 @@ ABIArgInfo X86_32ABIInfo::classifyArgumentType(QualType Ty,
     }
   }
 
-  // vectorcall adds the concept of a homogenous vector aggregate, similar
-  // to other targets, regcall uses some of the HVA rules.
+  // Regcall uses the concept of a homogenous vector aggregate, similar
+  // to other targets.
   const Type *Base = nullptr;
   uint64_t NumElts = 0;
-  if ((State.CC == llvm::CallingConv::X86_VectorCall ||
-       State.CC == llvm::CallingConv::X86_RegCall) &&
+  if (State.CC == llvm::CallingConv::X86_RegCall &&
       isHomogeneousAggregate(Ty, Base, NumElts)) {
 
-    if (State.CC == llvm::CallingConv::X86_RegCall) {
-      if (State.FreeSSERegs >= NumElts) {
-        State.FreeSSERegs -= NumElts;
-        if (Ty->isBuiltinType() || Ty->isVectorType())
-          return ABIArgInfo::getDirect();
-        return ABIArgInfo::getExpand();
-
-      }
-      return getIndirectResult(Ty, /*ByVal=*/false, State);
-    } else if (State.CC == llvm::CallingConv::X86_VectorCall) {
-      if (State.FreeSSERegs >= NumElts && (Ty->isBuiltinType() || Ty->isVectorType())) {
-        // Actual floating-point types get registers first time through if
-        // there is registers available
-        State.FreeSSERegs -= NumElts;
+    if (State.FreeSSERegs >= NumElts) {
+      State.FreeSSERegs -= NumElts;
+      if (Ty->isBuiltinType() || Ty->isVectorType())
         return ABIArgInfo::getDirect();
-      }  else if (!Ty->isBuiltinType() && !Ty->isVectorType()) {
-        // HVA Types only get registers after everything else has been
-        // set, so it gets set as indirect for now.
-        return ABIArgInfo::getIndirect(getContext().getTypeAlignInChars(Ty));
-      }
+      return ABIArgInfo::getExpand();
     }
+    return getIndirectResult(Ty, /*ByVal=*/false, State);
   }
 
   if (isAggregateTypeForABI(Ty)) {
@@ -1684,31 +1647,53 @@ ABIArgInfo X86_32ABIInfo::classifyArgumentType(QualType Ty,
 
 void X86_32ABIInfo::computeVectorCallArgs(CGFunctionInfo &FI, CCState &State,
                                           bool &UsedInAlloca) const {
-  // Vectorcall only allows the first 6 parameters to be passed in registers,
-  // and homogeneous vector aggregates are only put into registers as a second
-  // priority.
-  unsigned Count = 0;
-  CCState ZeroState = State;
-  ZeroState.FreeRegs = ZeroState.FreeSSERegs = 0;
-  // HVAs must be done as a second priority for registers, so the deferred
-  // items are dealt with by going through the pattern a second time.
+  // Vectorcall x86 works subtly different than in x64, so the format is
+  // a bit different than the x64 version.  First, all vector types (not HVAs)
+  // are assigned, with the first 6 ending up in the YMM0-5 or XMM0-5 registers.
+  // This differs from the x64 implementation, where the first 6 by INDEX get
+  // registers.
+  // After that, integers AND HVAs are assigned Left to Right in the same pass.
+  // Integers are passed as ECX/EDX if one is available (in order).  HVAs will
+  // first take up the remaining YMM/XMM registers. If insufficient registers
+  // remain but an integer register (ECX/EDX) is available, it will be passed
+  // in that, else, on the stack.
   for (auto &I : FI.arguments()) {
-    if (Count < VectorcallMaxParamNumAsReg)
-      I.info = classifyArgumentType(I.type, State);
-    else
-      // Parameters after the 6th cannot be passed in registers,
-      // so pretend there are no registers left for them.
-      I.info = classifyArgumentType(I.type, ZeroState);
-    UsedInAlloca |= (I.info.getKind() == ABIArgInfo::InAlloca);
-    ++Count;
+    // First pass do all the vector types.
+    const Type *Base = nullptr;
+    uint64_t NumElts = 0;
+    const QualType& Ty = I.type;
+    if ((Ty->isVectorType() || Ty->isBuiltinType()) &&
+        isHomogeneousAggregate(Ty, Base, NumElts)) {
+      if (State.FreeSSERegs >= NumElts) {
+        State.FreeSSERegs -= NumElts;
+        I.info = ABIArgInfo::getDirect();
+      } else {
+        I.info = classifyArgumentType(Ty, State);
+      }
+      UsedInAlloca |= (I.info.getKind() == ABIArgInfo::InAlloca);
+    }
   }
-  Count = 0;
-  // Go through the arguments a second time to get HVAs registers if there
-  // are still some available.
+
   for (auto &I : FI.arguments()) {
-    if (Count < VectorcallMaxParamNumAsReg)
-      I.info = reclassifyHvaArgType(I.type, State, I.info);
-    ++Count;
+    // Second pass, do the rest!
+    const Type *Base = nullptr;
+    uint64_t NumElts = 0;
+    const QualType& Ty = I.type;
+    bool IsHva = isHomogeneousAggregate(Ty, Base, NumElts);
+
+    if (IsHva && !Ty->isVectorType() && !Ty->isBuiltinType()) {
+      // Assign true HVAs (non vector/native FP types).
+      if (State.FreeSSERegs >= NumElts) {
+        State.FreeSSERegs -= NumElts;
+        I.info = getDirectX86Hva();
+      } else {
+        I.info = getIndirectResult(Ty, /*ByVal=*/false, State);
+      }
+    } else if (!IsHva) {
+      // Assign all Non-HVAs, so this will exclude Vector/FP args.
+      I.info = classifyArgumentType(Ty, State);
+      UsedInAlloca |= (I.info.getKind() == ABIArgInfo::InAlloca);
+    }
   }
 }
 
@@ -3901,6 +3886,8 @@ void WinX86_64ABIInfo::computeVectorCallArgs(CGFunctionInfo &FI,
                                              bool IsRegCall) const {
   unsigned Count = 0;
   for (auto &I : FI.arguments()) {
+    // Vectorcall in x64 only permits the first 6 arguments to be passed
+    // as XMM/YMM registers.
     if (Count < VectorcallMaxParamNumAsReg)
       I.info = classify(I.type, FreeSSERegs, false, IsVectorCall, IsRegCall);
     else {
@@ -3913,11 +3900,8 @@ void WinX86_64ABIInfo::computeVectorCallArgs(CGFunctionInfo &FI,
     ++Count;
   }
 
-  Count = 0;
   for (auto &I : FI.arguments()) {
-    if (Count < VectorcallMaxParamNumAsReg)
-      I.info = reclassifyHvaArgType(I.type, FreeSSERegs, I.info);
-    ++Count;
+    I.info = reclassifyHvaArgType(I.type, FreeSSERegs, I.info);
   }
 }
 
@@ -7344,8 +7328,6 @@ public:
 };
 }
 
-static void appendOpenCLVersionMD (CodeGen::CodeGenModule &CGM);
-
 void AMDGPUTargetCodeGenInfo::setTargetAttributes(
     const Decl *D,
     llvm::GlobalValue *GV,
@@ -7402,8 +7384,6 @@ void AMDGPUTargetCodeGenInfo::setTargetAttributes(
     if (NumVGPR != 0)
       F->addFnAttr("amdgpu-num-vgpr", llvm::utostr(NumVGPR));
   }
-
-  appendOpenCLVersionMD(M);
 }
 
 unsigned AMDGPUTargetCodeGenInfo::getOpenCLKernelCallingConv() const {
@@ -8074,8 +8054,6 @@ class SPIRTargetCodeGenInfo : public TargetCodeGenInfo {
 public:
   SPIRTargetCodeGenInfo(CodeGen::CodeGenTypes &CGT)
     : TargetCodeGenInfo(new DefaultABIInfo(CGT)) {}
-  void emitTargetMD(const Decl *D, llvm::GlobalValue *GV,
-                    CodeGen::CodeGenModule &M) const override;
   unsigned getOpenCLKernelCallingConv() const override;
 };
 
@@ -8090,41 +8068,6 @@ void computeSPIRKernelABIInfo(CodeGenModule &CGM, CGFunctionInfo &FI) {
 }
 }
 
-/// Emit SPIR specific metadata: OpenCL and SPIR version.
-void SPIRTargetCodeGenInfo::emitTargetMD(const Decl *D, llvm::GlobalValue *GV,
-                                         CodeGen::CodeGenModule &CGM) const {
-  llvm::LLVMContext &Ctx = CGM.getModule().getContext();
-  llvm::Type *Int32Ty = llvm::Type::getInt32Ty(Ctx);
-  llvm::Module &M = CGM.getModule();
-  // SPIR v2.0 s2.12 - The SPIR version used by the module is stored in the
-  // opencl.spir.version named metadata.
-  llvm::Metadata *SPIRVerElts[] = {
-      llvm::ConstantAsMetadata::get(
-          llvm::ConstantInt::get(Int32Ty, CGM.getLangOpts().OpenCLVersion / 100)),
-      llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
-          Int32Ty, (CGM.getLangOpts().OpenCLVersion / 100 > 1) ? 0 : 2))};
-  llvm::NamedMDNode *SPIRVerMD =
-      M.getOrInsertNamedMetadata("opencl.spir.version");
-  SPIRVerMD->addOperand(llvm::MDNode::get(Ctx, SPIRVerElts));
-  appendOpenCLVersionMD(CGM);
-}
-
-static void appendOpenCLVersionMD(CodeGen::CodeGenModule &CGM) {
-  llvm::LLVMContext &Ctx = CGM.getModule().getContext();
-  llvm::Type *Int32Ty = llvm::Type::getInt32Ty(Ctx);
-  llvm::Module &M = CGM.getModule();
-  // SPIR v2.0 s2.13 - The OpenCL version used by the module is stored in the
-  // opencl.ocl.version named metadata node.
-  llvm::Metadata *OCLVerElts[] = {
-      llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
-          Int32Ty, CGM.getLangOpts().OpenCLVersion / 100)),
-      llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
-          Int32Ty, (CGM.getLangOpts().OpenCLVersion % 100) / 10))};
-  llvm::NamedMDNode *OCLVerMD =
-      M.getOrInsertNamedMetadata("opencl.ocl.version");
-  OCLVerMD->addOperand(llvm::MDNode::get(Ctx, OCLVerElts));
-}
-
 unsigned SPIRTargetCodeGenInfo::getOpenCLKernelCallingConv() const {
   return llvm::CallingConv::SPIR_KERNEL;
 }
diff --git a/lib/Driver/CMakeLists.txt b/lib/Driver/CMakeLists.txt
index b639927a95b6..c7ca698d95a0 100644
--- a/lib/Driver/CMakeLists.txt
+++ b/lib/Driver/CMakeLists.txt
@@ -28,6 +28,7 @@ add_clang_library(clangDriver
   ToolChains/Arch/Sparc.cpp
   ToolChains/Arch/SystemZ.cpp
   ToolChains/Arch/X86.cpp
+  ToolChains/Ananas.cpp
   ToolChains/AMDGPU.cpp
   ToolChains/AVR.cpp
   ToolChains/Bitrig.cpp
diff --git a/lib/Driver/Driver.cpp b/lib/Driver/Driver.cpp
index eb504fd4e541..f23975829eaf 100644
--- a/lib/Driver/Driver.cpp
+++ b/lib/Driver/Driver.cpp
@@ -11,6 +11,7 @@
 #include "InputInfo.h"
 #include "ToolChains/AMDGPU.h"
 #include "ToolChains/AVR.h"
+#include "ToolChains/Ananas.h"
 #include "ToolChains/Bitrig.h"
 #include "ToolChains/Clang.h"
 #include "ToolChains/CloudABI.h"
@@ -1227,7 +1228,32 @@ bool Driver::HandleImmediateArgs(const Compilation &C) {
   if (Arg *A = C.getArgs().getLastArg(options::OPT_autocomplete)) {
     // Print out all options that start with a given argument. This is used for
     // shell autocompletion.
-    llvm::outs() << llvm::join(Opts->findByPrefix(A->getValue()), " ") << '\n';
+    StringRef PassedFlags = A->getValue();
+    std::vector<std::string> SuggestedCompletions;
+
+    if (PassedFlags.find(',') == StringRef::npos) {
+      // If the flag is in the form of "--autocomplete=-foo",
+      // we were requested to print out all option names that start with "-foo".
+      // For example, "--autocomplete=-fsyn" is expanded to "-fsyntax-only".
+      SuggestedCompletions = Opts->findByPrefix(PassedFlags);
+    } else {
+      // If the flag is in the form of "--autocomplete=foo,bar", we were
+      // requested to print out all option values for "-foo" that start with
+      // "bar". For example,
+      // "--autocomplete=-stdlib=,l" is expanded to "libc++" and "libstdc++".
+      StringRef Option, Arg;
+      std::tie(Option, Arg) = PassedFlags.split(',');
+      SuggestedCompletions = Opts->suggestValueCompletions(Option, Arg);
+    }
+
+    // Sort the autocomplete candidates so that shells print them out in a
+    // deterministic order. We could sort in any way, but we chose
+    // case-insensitive sorting for consistency with the -help option
+    // which prints out options in the case-insensitive alphabetical order.
+    std::sort(SuggestedCompletions.begin(), SuggestedCompletions.end(),
+              [](StringRef A, StringRef B) { return A.compare_lower(B) < 0; });
+
+    llvm::outs() << llvm::join(SuggestedCompletions, " ") << '\n';
     return false;
   }
 
@@ -3724,6 +3750,9 @@ const ToolChain &Driver::getToolChain(const ArgList &Args,
     case llvm::Triple::Haiku:
       TC = llvm::make_unique<toolchains::Haiku>(*this, Target, Args);
       break;
+    case llvm::Triple::Ananas:
+      TC = llvm::make_unique<toolchains::Ananas>(*this, Target, Args);
+      break;
     case llvm::Triple::CloudABI:
       TC = llvm::make_unique<toolchains::CloudABI>(*this, Target, Args);
       break;
diff --git a/lib/Driver/DriverOptions.cpp b/lib/Driver/DriverOptions.cpp
index 6a7410901d25..ac63b96cf96d 100644
--- a/lib/Driver/DriverOptions.cpp
+++ b/lib/Driver/DriverOptions.cpp
@@ -21,10 +21,10 @@ using namespace llvm::opt;
 #undef PREFIX
 
 static const OptTable::Info InfoTable[] = {
-#define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM, \
-               HELPTEXT, METAVAR)   \
-  { PREFIX, NAME, HELPTEXT, METAVAR, OPT_##ID, Option::KIND##Class, PARAM, \
-    FLAGS, OPT_##GROUP, OPT_##ALIAS, ALIASARGS },
+#define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM,  \
+               HELPTEXT, METAVAR, VALUES)                                      \
+  {PREFIX, NAME,  HELPTEXT,    METAVAR,     OPT_##ID,  Option::KIND##Class,    \
+   PARAM,  FLAGS, OPT_##GROUP, OPT_##ALIAS, ALIASARGS, VALUES},
 #include "clang/Driver/Options.inc"
 #undef OPTION
 };
diff --git a/lib/Driver/SanitizerArgs.cpp b/lib/Driver/SanitizerArgs.cpp
index d6a9c35eda78..3a30f2a289b0 100644
--- a/lib/Driver/SanitizerArgs.cpp
+++ b/lib/Driver/SanitizerArgs.cpp
@@ -208,12 +208,28 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
   SanitizerMask TrappingKinds = parseSanitizeTrapArgs(D, Args);
   SanitizerMask InvalidTrappingKinds = TrappingKinds & NotAllowedWithTrap;
 
+  // The object size sanitizer should not be enabled at -O0.
+  Arg *OptLevel = Args.getLastArg(options::OPT_O_Group);
+  bool RemoveObjectSizeAtO0 =
+      !OptLevel || OptLevel->getOption().matches(options::OPT_O0);
+
   for (ArgList::const_reverse_iterator I = Args.rbegin(), E = Args.rend();
        I != E; ++I) {
     const auto *Arg = *I;
     if (Arg->getOption().matches(options::OPT_fsanitize_EQ)) {
       Arg->claim();
-      SanitizerMask Add = parseArgValues(D, Arg, true);
+      SanitizerMask Add = parseArgValues(D, Arg, /*AllowGroups=*/true);
+
+      if (RemoveObjectSizeAtO0) {
+        AllRemove |= SanitizerKind::ObjectSize;
+
+        // The user explicitly enabled the object size sanitizer. Warn that
+        // that this does nothing at -O0.
+        if (Add & SanitizerKind::ObjectSize)
+          D.Diag(diag::warn_drv_object_size_disabled_O0)
+              << Arg->getAsString(Args);
+      }
+
       AllAddedKinds |= expandSanitizerGroups(Add);
 
       // Avoid diagnosing any sanitizer which is disabled later.
diff --git a/lib/Driver/ToolChains/Ananas.cpp b/lib/Driver/ToolChains/Ananas.cpp
new file mode 100644
index 000000000000..a67e1d2378f5
--- /dev/null
+++ b/lib/Driver/ToolChains/Ananas.cpp
@@ -0,0 +1,120 @@
+//===--- Ananas.cpp - Ananas ToolChain Implementations ------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "Ananas.h"
+#include "InputInfo.h"
+#include "CommonArgs.h"
+#include "clang/Config/config.h"
+#include "clang/Driver/Compilation.h"
+#include "clang/Driver/Driver.h"
+#include "clang/Driver/Options.h"
+#include "llvm/ADT/SmallString.h"
+#include "llvm/Option/ArgList.h"
+#include "llvm/Support/Path.h"
+
+using namespace clang::driver;
+using namespace clang::driver::tools;
+using namespace clang::driver::toolchains;
+using namespace clang;
+using namespace llvm::opt;
+
+void ananas::Assembler::ConstructJob(Compilation &C, const JobAction &JA,
+                                     const InputInfo &Output,
+                                     const InputInfoList &Inputs,
+                                     const ArgList &Args,
+                                     const char *LinkingOutput) const {
+  claimNoWarnArgs(Args);
+  ArgStringList CmdArgs;
+
+  Args.AddAllArgValues(CmdArgs, options::OPT_Wa_COMMA, options::OPT_Xassembler);
+
+  CmdArgs.push_back("-o");
+  CmdArgs.push_back(Output.getFilename());
+
+  for (const auto &II : Inputs)
+    CmdArgs.push_back(II.getFilename());
+
+  const char *Exec = Args.MakeArgString(getToolChain().GetProgramPath("as"));
+  C.addCommand(llvm::make_unique<Command>(JA, *this, Exec, CmdArgs, Inputs));
+}
+
+void ananas::Linker::ConstructJob(Compilation &C, const JobAction &JA,
+                                  const InputInfo &Output,
+                                  const InputInfoList &Inputs,
+                                  const ArgList &Args,
+                                  const char *LinkingOutput) const {
+  const ToolChain &ToolChain = getToolChain();
+  const Driver &D = ToolChain.getDriver();
+  ArgStringList CmdArgs;
+
+  // Silence warning for "clang -g foo.o -o foo"
+  Args.ClaimAllArgs(options::OPT_g_Group);
+  // and "clang -emit-llvm foo.o -o foo"
+  Args.ClaimAllArgs(options::OPT_emit_llvm);
+  // and for "clang -w foo.o -o foo". Other warning options are already
+  // handled somewhere else.
+  Args.ClaimAllArgs(options::OPT_w);
+
+  if (!D.SysRoot.empty())
+    CmdArgs.push_back(Args.MakeArgString("--sysroot=" + D.SysRoot));
+
+  // Ananas only supports static linkage for now.
+  CmdArgs.push_back("-Bstatic");
+
+  if (Output.isFilename()) {
+    CmdArgs.push_back("-o");
+    CmdArgs.push_back(Output.getFilename());
+  } else {
+    assert(Output.isNothing() && "Invalid output.");
+  }
+
+  if (!Args.hasArg(options::OPT_nostdlib, options::OPT_nostartfiles)) {
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("crt0.o")));
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("crti.o")));
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("crtbegin.o")));
+  }
+
+  Args.AddAllArgs(CmdArgs, options::OPT_L);
+  ToolChain.AddFilePathLibArgs(Args, CmdArgs);
+  Args.AddAllArgs(CmdArgs,
+                  {options::OPT_T_Group, options::OPT_e, options::OPT_s,
+                   options::OPT_t, options::OPT_Z_Flag, options::OPT_r});
+
+  if (D.isUsingLTO())
+    AddGoldPlugin(ToolChain, Args, CmdArgs, D.getLTOMode() == LTOK_Thin, D);
+
+  AddLinkerInputs(ToolChain, Inputs, Args, CmdArgs, JA);
+
+  if (!Args.hasArg(options::OPT_nostdlib, options::OPT_nodefaultlibs)) {
+    if (D.CCCIsCXX())
+      ToolChain.AddCXXStdlibLibArgs(Args, CmdArgs);
+    CmdArgs.push_back("-lc");
+  }
+
+  if (!Args.hasArg(options::OPT_nostdlib, options::OPT_nostartfiles)) {
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("crtend.o")));
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("crtn.o")));
+  }
+
+  const char *Exec = Args.MakeArgString(ToolChain.GetLinkerPath());
+  C.addCommand(llvm::make_unique<Command>(JA, *this, Exec, CmdArgs, Inputs));
+}
+
+// Ananas - Ananas tool chain which can call as(1) and ld(1) directly.
+
+Ananas::Ananas(const Driver &D, const llvm::Triple &Triple, const ArgList &Args)
+    : Generic_ELF(D, Triple, Args) {
+  getFilePaths().push_back(getDriver().SysRoot + "/usr/lib");
+}
+
+Tool *Ananas::buildAssembler() const {
+  return new tools::ananas::Assembler(*this);
+}
+
+Tool *Ananas::buildLinker() const { return new tools::ananas::Linker(*this); }
diff --git a/lib/Driver/ToolChains/Ananas.h b/lib/Driver/ToolChains/Ananas.h
new file mode 100644
index 000000000000..2563dd2d49a9
--- /dev/null
+++ b/lib/Driver/ToolChains/Ananas.h
@@ -0,0 +1,67 @@
+//===--- Ananas.h - Ananas ToolChain Implementations --------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_LIB_DRIVER_TOOLCHAINS_ANANAS_H
+#define LLVM_CLANG_LIB_DRIVER_TOOLCHAINS_ANANAS_H
+
+#include "Gnu.h"
+#include "clang/Driver/Tool.h"
+#include "clang/Driver/ToolChain.h"
+
+namespace clang {
+namespace driver {
+namespace tools {
+
+/// ananas -- Directly call GNU Binutils assembler and linker
+namespace ananas {
+class LLVM_LIBRARY_VISIBILITY Assembler : public GnuTool {
+public:
+  Assembler(const ToolChain &TC)
+      : GnuTool("ananas::Assembler", "assembler", TC) {}
+
+  bool hasIntegratedCPP() const override { return false; }
+
+  void ConstructJob(Compilation &C, const JobAction &JA,
+                    const InputInfo &Output, const InputInfoList &Inputs,
+                    const llvm::opt::ArgList &TCArgs,
+                    const char *LinkingOutput) const override;
+};
+
+class LLVM_LIBRARY_VISIBILITY Linker : public GnuTool {
+public:
+  Linker(const ToolChain &TC) : GnuTool("ananas::Linker", "linker", TC) {}
+
+  bool hasIntegratedCPP() const override { return false; }
+  bool isLinkJob() const override { return true; }
+
+  void ConstructJob(Compilation &C, const JobAction &JA,
+                    const InputInfo &Output, const InputInfoList &Inputs,
+                    const llvm::opt::ArgList &TCArgs,
+                    const char *LinkingOutput) const override;
+};
+} // end namespace ananas
+} // end namespace tools
+
+namespace toolchains {
+
+class LLVM_LIBRARY_VISIBILITY Ananas : public Generic_ELF {
+public:
+  Ananas(const Driver &D, const llvm::Triple &Triple,
+         const llvm::opt::ArgList &Args);
+
+protected:
+  Tool *buildAssembler() const override;
+  Tool *buildLinker() const override;
+};
+
+} // end namespace toolchains
+} // end namespace driver
+} // end namespace clang
+
+#endif // LLVM_CLANG_LIB_DRIVER_TOOLCHAINS_ANANAS_H
diff --git a/lib/Driver/ToolChains/Clang.cpp b/lib/Driver/ToolChains/Clang.cpp
index bd4e894d6504..292cf72b56ca 100644
--- a/lib/Driver/ToolChains/Clang.cpp
+++ b/lib/Driver/ToolChains/Clang.cpp
@@ -910,6 +910,37 @@ static void RenderDebugEnablingArgs(const ArgList &Args, ArgStringList &CmdArgs,
   }
 }
 
+static void RenderDebugInfoCompressionArgs(const ArgList &Args,
+                                           ArgStringList &CmdArgs,
+                                           const Driver &D) {
+  const Arg *A = Args.getLastArg(options::OPT_gz, options::OPT_gz_EQ);
+  if (!A)
+    return;
+
+  if (A->getOption().getID() == options::OPT_gz) {
+    if (llvm::zlib::isAvailable())
+      CmdArgs.push_back("-compress-debug-sections");
+    else
+      D.Diag(diag::warn_debug_compression_unavailable);
+    return;
+  }
+
+  StringRef Value = A->getValue();
+  if (Value == "none") {
+    CmdArgs.push_back("-compress-debug-sections=none");
+  } else if (Value == "zlib" || Value == "zlib-gnu") {
+    if (llvm::zlib::isAvailable()) {
+      CmdArgs.push_back(
+          Args.MakeArgString("-compress-debug-sections=" + Twine(Value)));
+    } else {
+      D.Diag(diag::warn_debug_compression_unavailable);
+    }
+  } else {
+    D.Diag(diag::err_drv_unsupported_option_argument)
+        << A->getOption().getName() << Value;
+  }
+}
+
 static const char *RelocationModelName(llvm::Reloc::Model Model) {
   switch (Model) {
   case llvm::Reloc::Static:
@@ -1747,10 +1778,6 @@ static void CollectArgsForIntegratedAssembler(Compilation &C,
   // arg after parsing the '-I' arg.
   bool TakeNextArg = false;
 
-  // When using an integrated assembler, translate -Wa, and -Xassembler
-  // options.
-  bool CompressDebugSections = false;
-
   bool UseRelaxRelocations = ENABLE_X86_RELAX_RELOCATIONS;
   const char *MipsTargetFeature = nullptr;
   for (const Arg *A :
@@ -1825,12 +1852,11 @@ static void CollectArgsForIntegratedAssembler(Compilation &C,
         CmdArgs.push_back("-massembler-fatal-warnings");
       } else if (Value == "--noexecstack") {
         CmdArgs.push_back("-mnoexecstack");
-      } else if (Value == "-compress-debug-sections" ||
-                 Value == "--compress-debug-sections") {
-        CompressDebugSections = true;
-      } else if (Value == "-nocompress-debug-sections" ||
+      } else if (Value.startswith("-compress-debug-sections") ||
+                 Value.startswith("--compress-debug-sections") ||
+                 Value == "-nocompress-debug-sections" ||
                  Value == "--nocompress-debug-sections") {
-        CompressDebugSections = false;
+        CmdArgs.push_back(Value.data());
       } else if (Value == "-mrelax-relocations=yes" ||
                  Value == "--mrelax-relocations=yes") {
         UseRelaxRelocations = true;
@@ -1883,12 +1909,6 @@ static void CollectArgsForIntegratedAssembler(Compilation &C,
       }
     }
   }
-  if (CompressDebugSections) {
-    if (llvm::zlib::isAvailable())
-      CmdArgs.push_back("-compress-debug-sections");
-    else
-      D.Diag(diag::warn_debug_compression_unavailable);
-  }
   if (UseRelaxRelocations)
     CmdArgs.push_back("--mrelax-relocations");
   if (MipsTargetFeature != nullptr) {
@@ -2824,6 +2844,8 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
     CmdArgs.push_back("-generate-type-units");
   }
 
+  RenderDebugInfoCompressionArgs(Args, CmdArgs, D);
+
   bool UseSeparateSections = isUseSeparateSections(Triple);
 
   if (Args.hasFlag(options::OPT_ffunction_sections,
@@ -4927,6 +4949,7 @@ void ClangAs::ConstructJob(Compilation &C, const JobAction &JA,
 
   const llvm::Triple &Triple = getToolChain().getEffectiveTriple();
   const std::string &TripleStr = Triple.getTriple();
+  const auto &D = getToolChain().getDriver();
 
   // Don't warn about "clang -w -c foo.s"
   Args.ClaimAllArgs(options::OPT_w);
@@ -5014,6 +5037,8 @@ void ClangAs::ConstructJob(Compilation &C, const JobAction &JA,
   }
   RenderDebugEnablingArgs(Args, CmdArgs, DebugInfoKind, DwarfVersion,
                           llvm::DebuggerKind::Default);
+  RenderDebugInfoCompressionArgs(Args, CmdArgs, D);
+
 
   // Handle -fPIC et al -- the relocation-model affects the assembler
   // for some targets.
diff --git a/lib/Driver/ToolChains/Gnu.cpp b/lib/Driver/ToolChains/Gnu.cpp
index d50f8e21f62f..bca5d3a3f28b 100644
--- a/lib/Driver/ToolChains/Gnu.cpp
+++ b/lib/Driver/ToolChains/Gnu.cpp
@@ -650,6 +650,8 @@ void tools::gnutools::Assembler::ConstructJob(Compilation &C,
                                               const InputInfoList &Inputs,
                                               const ArgList &Args,
                                               const char *LinkingOutput) const {
+  const auto &D = getToolChain().getDriver();
+
   claimNoWarnArgs(Args);
 
   ArgStringList CmdArgs;
@@ -660,6 +662,23 @@ void tools::gnutools::Assembler::ConstructJob(Compilation &C,
   std::tie(RelocationModel, PICLevel, IsPIE) =
       ParsePICArgs(getToolChain(), Args);
 
+  if (const Arg *A = Args.getLastArg(options::OPT_gz, options::OPT_gz_EQ)) {
+    if (A->getOption().getID() == options::OPT_gz) {
+      CmdArgs.push_back("-compress-debug-sections");
+    } else {
+      StringRef Value = A->getValue();
+      if (Value == "none") {
+        CmdArgs.push_back("-compress-debug-sections=none");
+      } else if (Value == "zlib" || Value == "zlib-gnu") {
+        CmdArgs.push_back(
+            Args.MakeArgString("-compress-debug-sections=" + Twine(Value)));
+      } else {
+        D.Diag(diag::err_drv_unsupported_option_argument)
+            << A->getOption().getName() << Value;
+      }
+    }
+  }
+
   switch (getToolChain().getArch()) {
   default:
     break;
@@ -2319,9 +2338,11 @@ bool Generic_GCC::IsIntegratedAssemblerDefault() const {
     return true;
   case llvm::Triple::mips64:
   case llvm::Triple::mips64el:
-    // Enabled for Debian mips64/mips64el only. Other targets are unable to
-    // distinguish N32 from N64.
-    if (getTriple().getEnvironment() == llvm::Triple::GNUABI64)
+    // Enabled for Debian and Android mips64/mipsel, as they can precisely
+    // identify the ABI in use (Debian) or only use N64 for MIPS64 (Android).
+    // Other targets are unable to distinguish N32 from N64.
+    if (getTriple().getEnvironment() == llvm::Triple::GNUABI64 ||
+        getTriple().isAndroid())
       return true;
     return false;
   default:
diff --git a/lib/Driver/ToolChains/WebAssembly.cpp b/lib/Driver/ToolChains/WebAssembly.cpp
index 3471569b6884..fcb6418b2517 100644
--- a/lib/Driver/ToolChains/WebAssembly.cpp
+++ b/lib/Driver/ToolChains/WebAssembly.cpp
@@ -83,6 +83,8 @@ void wasm::Linker::ConstructJob(Compilation &C, const JobAction &JA,
     if (Args.hasArg(options::OPT_pthread))
       CmdArgs.push_back("-lpthread");
 
+    CmdArgs.push_back("-allow-undefined-file");
+    CmdArgs.push_back(Args.MakeArgString(ToolChain.GetFilePath("wasm.syms")));
     CmdArgs.push_back("-lc");
     CmdArgs.push_back("-lcompiler_rt");
   }
@@ -104,8 +106,7 @@ WebAssembly::WebAssembly(const Driver &D, const llvm::Triple &Triple,
 
   getProgramPaths().push_back(getDriver().getInstalledDir());
 
-  getFilePaths().push_back(
-      getDriver().SysRoot + "/lib" + (Triple.isArch32Bit() ? "32" : "64"));
+  getFilePaths().push_back(getDriver().SysRoot + "/lib");
 }
 
 bool WebAssembly::IsMathErrnoDefault() const { return false; }
diff --git a/lib/Edit/EditedSource.cpp b/lib/Edit/EditedSource.cpp
index 03d45cf185c9..444d0393cccd 100644
--- a/lib/Edit/EditedSource.cpp
+++ b/lib/Edit/EditedSource.cpp
@@ -28,13 +28,18 @@ void EditedSource::deconstructMacroArgLoc(SourceLocation Loc,
                                           MacroArgUse &ArgUse) {
   assert(SourceMgr.isMacroArgExpansion(Loc));
   SourceLocation DefArgLoc = SourceMgr.getImmediateExpansionRange(Loc).first;
-  ExpansionLoc = SourceMgr.getImmediateExpansionRange(DefArgLoc).first;
+  SourceLocation ImmediateExpansionLoc =
+      SourceMgr.getImmediateExpansionRange(DefArgLoc).first;
+  ExpansionLoc = ImmediateExpansionLoc;
+  while (SourceMgr.isMacroBodyExpansion(ExpansionLoc))
+    ExpansionLoc = SourceMgr.getImmediateExpansionRange(ExpansionLoc).first;
   SmallString<20> Buf;
   StringRef ArgName = Lexer::getSpelling(SourceMgr.getSpellingLoc(DefArgLoc),
                                          Buf, SourceMgr, LangOpts);
-  ArgUse = {nullptr, SourceLocation()};
+  ArgUse = MacroArgUse{nullptr, SourceLocation(), SourceLocation()};
   if (!ArgName.empty())
-    ArgUse = {&IdentTable.get(ArgName), SourceMgr.getSpellingLoc(DefArgLoc)};
+    ArgUse = {&IdentTable.get(ArgName), ImmediateExpansionLoc,
+              SourceMgr.getSpellingLoc(DefArgLoc)};
 }
 
 void EditedSource::startingCommit() {}
@@ -69,10 +74,11 @@ bool EditedSource::canInsertInOffset(SourceLocation OrigLoc, FileOffset Offs) {
     deconstructMacroArgLoc(OrigLoc, ExpLoc, ArgUse);
     auto I = ExpansionToArgMap.find(ExpLoc.getRawEncoding());
     if (I != ExpansionToArgMap.end() &&
-        std::find_if(
-            I->second.begin(), I->second.end(), [&](const MacroArgUse &U) {
-              return ArgUse.first == U.first && ArgUse.second != U.second;
-            }) != I->second.end()) {
+        find_if(I->second, [&](const MacroArgUse &U) {
+          return ArgUse.Identifier == U.Identifier &&
+                 std::tie(ArgUse.ImmediateExpansionLoc, ArgUse.UseLoc) !=
+                     std::tie(U.ImmediateExpansionLoc, U.UseLoc);
+        }) != I->second.end()) {
       // Trying to write in a macro argument input that has already been
       // written by a previous commit for another expansion of the same macro
       // argument name. For example:
@@ -89,7 +95,6 @@ bool EditedSource::canInsertInOffset(SourceLocation OrigLoc, FileOffset Offs) {
       return false;
     }
   }
-
   return true;
 }
 
@@ -102,13 +107,13 @@ bool EditedSource::commitInsert(SourceLocation OrigLoc,
     return true;
 
   if (SourceMgr.isMacroArgExpansion(OrigLoc)) {
-    SourceLocation ExpLoc;
     MacroArgUse ArgUse;
+    SourceLocation ExpLoc;
     deconstructMacroArgLoc(OrigLoc, ExpLoc, ArgUse);
-    if (ArgUse.first)
+    if (ArgUse.Identifier)
       CurrCommitMacroArgExps.emplace_back(ExpLoc, ArgUse);
   }
-  
+
   FileEdit &FA = FileEdits[Offs];
   if (FA.Text.empty()) {
     FA.Text = copyString(text);
diff --git a/lib/Format/CMakeLists.txt b/lib/Format/CMakeLists.txt
index 0c7511c1bb07..42e6d53d9fe6 100644
--- a/lib/Format/CMakeLists.txt
+++ b/lib/Format/CMakeLists.txt
@@ -13,6 +13,7 @@ add_clang_library(clangFormat
   TokenAnnotator.cpp
   UnwrappedLineFormatter.cpp
   UnwrappedLineParser.cpp
+  UsingDeclarationsSorter.cpp
   WhitespaceManager.cpp
 
   LINK_LIBS
diff --git a/lib/Format/ContinuationIndenter.cpp b/lib/Format/ContinuationIndenter.cpp
index 387923031f85..cca773cfe3cb 100644
--- a/lib/Format/ContinuationIndenter.cpp
+++ b/lib/Format/ContinuationIndenter.cpp
@@ -453,7 +453,8 @@ void ContinuationIndenter::addTokenOnCurrentLine(LineState &State, bool DryRun,
   State.Column += Spaces;
   if (Current.isNot(tok::comment) && Previous.is(tok::l_paren) &&
       Previous.Previous &&
-      Previous.Previous->isOneOf(tok::kw_if, tok::kw_for)) {
+      (Previous.Previous->isOneOf(tok::kw_if, tok::kw_for) ||
+       Previous.Previous->endsSequence(tok::kw_constexpr, tok::kw_if))) {
     // Treat the condition inside an if as if it was a second function
     // parameter, i.e. let nested calls have a continuation indent.
     State.Stack.back().LastSpace = State.Column;
@@ -1049,7 +1050,9 @@ void ContinuationIndenter::moveStatePastScopeOpener(LineState &State,
         (Current.is(TT_ArrayInitializerLSquare) && EndsInComma) ||
         Current.is(TT_DictLiteral) ||
         Style.Language == FormatStyle::LK_Proto || !Style.BinPackArguments ||
-        (NextNoComment && NextNoComment->is(TT_DesignatedInitializerPeriod));
+        (NextNoComment &&
+         NextNoComment->isOneOf(TT_DesignatedInitializerPeriod,
+                                TT_DesignatedInitializerLSquare));
     if (Current.ParameterCount > 1)
       NestedBlockIndent = std::max(NestedBlockIndent, State.Column + 1);
   } else {
diff --git a/lib/Format/Format.cpp b/lib/Format/Format.cpp
index 39da87cf9988..7659d56adf25 100644
--- a/lib/Format/Format.cpp
+++ b/lib/Format/Format.cpp
@@ -23,6 +23,7 @@
 #include "TokenAnnotator.h"
 #include "UnwrappedLineFormatter.h"
 #include "UnwrappedLineParser.h"
+#include "UsingDeclarationsSorter.h"
 #include "WhitespaceManager.h"
 #include "clang/Basic/Diagnostic.h"
 #include "clang/Basic/DiagnosticOptions.h"
@@ -96,6 +97,7 @@ template <> struct ScalarEnumerationTraits<FormatStyle::ShortFunctionStyle> {
     IO.enumCase(Value, "All", FormatStyle::SFS_All);
     IO.enumCase(Value, "true", FormatStyle::SFS_All);
     IO.enumCase(Value, "Inline", FormatStyle::SFS_Inline);
+    IO.enumCase(Value, "InlineOnly", FormatStyle::SFS_InlineOnly);
     IO.enumCase(Value, "Empty", FormatStyle::SFS_Empty);
   }
 };
@@ -377,6 +379,7 @@ template <> struct MappingTraits<FormatStyle> {
     IO.mapOptional("PointerAlignment", Style.PointerAlignment);
     IO.mapOptional("ReflowComments", Style.ReflowComments);
     IO.mapOptional("SortIncludes", Style.SortIncludes);
+    IO.mapOptional("SortUsingDeclarations", Style.SortUsingDeclarations);
     IO.mapOptional("SpaceAfterCStyleCast", Style.SpaceAfterCStyleCast);
     IO.mapOptional("SpaceAfterTemplateKeyword", Style.SpaceAfterTemplateKeyword);
     IO.mapOptional("SpaceBeforeAssignmentOperators",
@@ -617,6 +620,7 @@ FormatStyle getLLVMStyle() {
 
   LLVMStyle.DisableFormat = false;
   LLVMStyle.SortIncludes = true;
+  LLVMStyle.SortUsingDeclarations = true;
 
   return LLVMStyle;
 }
@@ -771,6 +775,7 @@ FormatStyle getNoStyle() {
   FormatStyle NoStyle = getLLVMStyle();
   NoStyle.DisableFormat = true;
   NoStyle.SortIncludes = false;
+  NoStyle.SortUsingDeclarations = false;
   return NoStyle;
 }
 
@@ -1877,38 +1882,53 @@ tooling::Replacements reformat(const FormatStyle &Style, StringRef Code,
     return tooling::Replacements();
   if (Expanded.Language == FormatStyle::LK_JavaScript && isMpegTS(Code))
     return tooling::Replacements();
-  auto Env = Environment::CreateVirtualEnvironment(Code, FileName, Ranges);
-
-  auto reformatAfterApplying = [&] (TokenAnalyzer& Fixer) {
-    tooling::Replacements Fixes = Fixer.process();
-    if (!Fixes.empty()) {
-      auto NewCode = applyAllReplacements(Code, Fixes);
-      if (NewCode) {
-        auto NewEnv = Environment::CreateVirtualEnvironment(
-            *NewCode, FileName,
-            tooling::calculateRangesAfterReplacements(Fixes, Ranges));
-        Formatter Format(*NewEnv, Expanded, Status);
-        return Fixes.merge(Format.process());
-      }
-    }
-    Formatter Format(*Env, Expanded, Status);
-    return Format.process();
-  };
 
-  if (Style.Language == FormatStyle::LK_Cpp &&
-      Style.FixNamespaceComments) {
-    NamespaceEndCommentsFixer CommentsFixer(*Env, Expanded);
-    return reformatAfterApplying(CommentsFixer);
+  typedef std::function<tooling::Replacements(const Environment &)>
+      AnalyzerPass;
+  SmallVector<AnalyzerPass, 4> Passes;
+
+  if (Style.Language == FormatStyle::LK_Cpp) {
+    if (Style.FixNamespaceComments)
+      Passes.emplace_back([&](const Environment &Env) {
+        return NamespaceEndCommentsFixer(Env, Expanded).process();
+      });
+
+    if (Style.SortUsingDeclarations)
+      Passes.emplace_back([&](const Environment &Env) {
+        return UsingDeclarationsSorter(Env, Expanded).process();
+      });
   }
 
   if (Style.Language == FormatStyle::LK_JavaScript &&
-      Style.JavaScriptQuotes != FormatStyle::JSQS_Leave) {
-    JavaScriptRequoter Requoter(*Env, Expanded);
-    return reformatAfterApplying(Requoter);
+      Style.JavaScriptQuotes != FormatStyle::JSQS_Leave)
+    Passes.emplace_back([&](const Environment &Env) {
+      return JavaScriptRequoter(Env, Expanded).process();
+    });
+
+  Passes.emplace_back([&](const Environment &Env) {
+    return Formatter(Env, Expanded, Status).process();
+  });
+
+  std::unique_ptr<Environment> Env =
+      Environment::CreateVirtualEnvironment(Code, FileName, Ranges);
+  llvm::Optional<std::string> CurrentCode = None;
+  tooling::Replacements Fixes;
+  for (size_t I = 0, E = Passes.size(); I < E; ++I) {
+    tooling::Replacements PassFixes = Passes[I](*Env);
+    auto NewCode = applyAllReplacements(
+        CurrentCode ? StringRef(*CurrentCode) : Code, PassFixes);
+    if (NewCode) {
+      Fixes = Fixes.merge(PassFixes);
+      if (I + 1 < E) {
+        CurrentCode = std::move(*NewCode);
+        Env = Environment::CreateVirtualEnvironment(
+            *CurrentCode, FileName,
+            tooling::calculateRangesAfterReplacements(Fixes, Ranges));
+      }
+    }
   }
 
-  Formatter Format(*Env, Expanded, Status);
-  return Format.process();
+  return Fixes;
 }
 
 tooling::Replacements cleanup(const FormatStyle &Style, StringRef Code,
@@ -1943,6 +1963,16 @@ tooling::Replacements fixNamespaceEndComments(const FormatStyle &Style,
   return Fix.process();
 }
 
+tooling::Replacements sortUsingDeclarations(const FormatStyle &Style,
+                                            StringRef Code,
+                                            ArrayRef<tooling::Range> Ranges,
+                                            StringRef FileName) {
+  std::unique_ptr<Environment> Env =
+      Environment::CreateVirtualEnvironment(Code, FileName, Ranges);
+  UsingDeclarationsSorter Sorter(*Env, Style);
+  return Sorter.process();
+}
+
 LangOptions getFormattingLangOpts(const FormatStyle &Style) {
   LangOptions LangOpts;
   LangOpts.CPlusPlus = 1;
diff --git a/lib/Format/FormatToken.h b/lib/Format/FormatToken.h
index 0c5a5284627c..c5bf48cdcc06 100644
--- a/lib/Format/FormatToken.h
+++ b/lib/Format/FormatToken.h
@@ -39,6 +39,7 @@ namespace format {
   TYPE(ConflictStart) \
   TYPE(CtorInitializerColon) \
   TYPE(CtorInitializerComma) \
+  TYPE(DesignatedInitializerLSquare) \
   TYPE(DesignatedInitializerPeriod) \
   TYPE(DictLiteral) \
   TYPE(ForEachMacro) \
diff --git a/lib/Format/TokenAnnotator.cpp b/lib/Format/TokenAnnotator.cpp
index 7ce699cf14a1..505f42ec9a06 100644
--- a/lib/Format/TokenAnnotator.cpp
+++ b/lib/Format/TokenAnnotator.cpp
@@ -145,6 +145,7 @@ private:
         (Left->Previous->isOneOf(tok::kw_static_assert, tok::kw_decltype,
                                  tok::kw_if, tok::kw_while, tok::l_paren,
                                  tok::comma) ||
+         Left->Previous->endsSequence(tok::kw_constexpr, tok::kw_if) ||
          Left->Previous->is(TT_BinaryOperator))) {
       // static_assert, if and while usually contain expressions.
       Contexts.back().IsExpression = true;
@@ -339,6 +340,9 @@ private:
                  Contexts.back().ContextKind == tok::l_brace &&
                  Parent->isOneOf(tok::l_brace, tok::comma)) {
         Left->Type = TT_JsComputedPropertyName;
+      } else if (Style.isCpp() && Contexts.back().ContextKind == tok::l_brace &&
+                 Parent && Parent->isOneOf(tok::l_brace, tok::comma)) {
+        Left->Type = TT_DesignatedInitializerLSquare;
       } else if (CurrentToken->is(tok::r_square) && Parent &&
                  Parent->is(TT_TemplateCloser)) {
         Left->Type = TT_ArraySubscriptLSquare;
@@ -391,7 +395,8 @@ private:
       if (CurrentToken->isOneOf(tok::r_paren, tok::r_brace))
         return false;
       if (CurrentToken->is(tok::colon)) {
-        if (Left->is(TT_ArraySubscriptLSquare)) {
+        if (Left->isOneOf(TT_ArraySubscriptLSquare,
+                          TT_DesignatedInitializerLSquare)) {
           Left->Type = TT_ObjCMethodExpr;
           StartsObjCMethodExpr = true;
           Contexts.back().ColonIsObjCMethodExpr = true;
@@ -572,6 +577,8 @@ private:
       break;
     case tok::kw_if:
     case tok::kw_while:
+      if (Tok->is(tok::kw_if) && CurrentToken && CurrentToken->is(tok::kw_constexpr))
+        next();
       if (CurrentToken && CurrentToken->is(tok::l_paren)) {
         next();
         if (!parseParens(/*LookForDecls=*/true))
@@ -1972,7 +1979,8 @@ unsigned TokenAnnotator::splitPenalty(const AnnotatedLine &Line,
     if (Right.is(TT_LambdaLSquare) && Left.is(tok::equal))
       return 35;
     if (!Right.isOneOf(TT_ObjCMethodExpr, TT_LambdaLSquare,
-                       TT_ArrayInitializerLSquare))
+                       TT_ArrayInitializerLSquare,
+                       TT_DesignatedInitializerLSquare))
       return 500;
   }
 
@@ -2060,7 +2068,8 @@ unsigned TokenAnnotator::splitPenalty(const AnnotatedLine &Line,
       Style.AlignAfterOpenBracket != FormatStyle::BAS_DontAlign)
     return 100;
   if (Left.is(tok::l_paren) && Left.Previous &&
-      Left.Previous->isOneOf(tok::kw_if, tok::kw_for))
+      (Left.Previous->isOneOf(tok::kw_if, tok::kw_for)
+       || Left.Previous->endsSequence(tok::kw_constexpr, tok::kw_if)))
     return 1000;
   if (Left.is(tok::equal) && InFunctionDecl)
     return 110;
@@ -2192,7 +2201,8 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
             (Style.SpacesInSquareBrackets &&
              Right.MatchingParen->is(TT_ArraySubscriptLSquare)));
   if (Right.is(tok::l_square) &&
-      !Right.isOneOf(TT_ObjCMethodExpr, TT_LambdaLSquare) &&
+      !Right.isOneOf(TT_ObjCMethodExpr, TT_LambdaLSquare,
+                     TT_DesignatedInitializerLSquare) &&
       !Left.isOneOf(tok::numeric_constant, TT_DictLiteral))
     return false;
   if (Left.is(tok::l_brace) && Right.is(tok::r_brace))
@@ -2211,6 +2221,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
             (Left.isOneOf(tok::kw_if, tok::pp_elif, tok::kw_for, tok::kw_while,
                           tok::kw_switch, tok::kw_case, TT_ForEachMacro,
                           TT_ObjCForIn) ||
+             Left.endsSequence(tok::kw_constexpr, tok::kw_if) ||
              (Left.isOneOf(tok::kw_try, Keywords.kw___except, tok::kw_catch,
                            tok::kw_new, tok::kw_delete) &&
               (!Left.Previous || Left.Previous->isNot(tok::period))))) ||
@@ -2391,8 +2402,9 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
   if (Left.is(tok::greater) && Right.is(tok::greater))
     return Right.is(TT_TemplateCloser) && Left.is(TT_TemplateCloser) &&
            (Style.Standard != FormatStyle::LS_Cpp11 || Style.SpacesInAngles);
-  if (Right.isOneOf(tok::arrow, tok::period, tok::arrowstar, tok::periodstar) ||
-      Left.isOneOf(tok::arrow, tok::period, tok::arrowstar, tok::periodstar))
+  if (Right.isOneOf(tok::arrow, tok::arrowstar, tok::periodstar) ||
+      Left.isOneOf(tok::arrow, tok::period, tok::arrowstar, tok::periodstar) ||
+      (Right.is(tok::period) && Right.isNot(TT_DesignatedInitializerPeriod)))
     return false;
   if (!Style.SpaceBeforeAssignmentOperators &&
       Right.getPrecedence() == prec::Assignment)
@@ -2468,8 +2480,8 @@ bool TokenAnnotator::mustBreakBefore(const AnnotatedLine &Line,
       return Style.AllowShortFunctionsOnASingleLine == FormatStyle::SFS_None ||
              Style.AllowShortFunctionsOnASingleLine == FormatStyle::SFS_Empty ||
              (Left.NestingLevel == 0 && Line.Level == 0 &&
-              Style.AllowShortFunctionsOnASingleLine ==
-                  FormatStyle::SFS_Inline);
+              Style.AllowShortFunctionsOnASingleLine &
+                  FormatStyle::SFS_InlineOnly);
   } else if (Style.Language == FormatStyle::LK_Java) {
     if (Right.is(tok::plus) && Left.is(tok::string_literal) && Right.Next &&
         Right.Next->is(tok::string_literal))
diff --git a/lib/Format/UnwrappedLineFormatter.cpp b/lib/Format/UnwrappedLineFormatter.cpp
index 7f644651a6ab..8836c07cac71 100644
--- a/lib/Format/UnwrappedLineFormatter.cpp
+++ b/lib/Format/UnwrappedLineFormatter.cpp
@@ -226,7 +226,7 @@ private:
         Style.AllowShortFunctionsOnASingleLine == FormatStyle::SFS_All ||
         (Style.AllowShortFunctionsOnASingleLine >= FormatStyle::SFS_Empty &&
          I[1]->First->is(tok::r_brace)) ||
-        (Style.AllowShortFunctionsOnASingleLine == FormatStyle::SFS_Inline &&
+        (Style.AllowShortFunctionsOnASingleLine & FormatStyle::SFS_InlineOnly &&
          TheLine->Level != 0);
 
     if (Style.CompactNamespaces) {
@@ -434,8 +434,11 @@ private:
     } else if (Limit != 0 && !Line.startsWith(tok::kw_namespace) &&
                !startsExternCBlock(Line)) {
       // We don't merge short records.
-      if (Line.First->isOneOf(tok::kw_class, tok::kw_union, tok::kw_struct,
-                              Keywords.kw_interface))
+      FormatToken *RecordTok =
+          Line.First->is(tok::kw_typedef) ? Line.First->Next : Line.First;
+      if (RecordTok &&
+          RecordTok->isOneOf(tok::kw_class, tok::kw_union, tok::kw_struct,
+                             Keywords.kw_interface))
         return 0;
 
       // Check that we still have three lines and they fit into the limit.
diff --git a/lib/Format/UnwrappedLineParser.cpp b/lib/Format/UnwrappedLineParser.cpp
index 27436dda67a7..f7678bb6b2a5 100644
--- a/lib/Format/UnwrappedLineParser.cpp
+++ b/lib/Format/UnwrappedLineParser.cpp
@@ -1516,6 +1516,8 @@ void UnwrappedLineParser::parseSquare() {
 void UnwrappedLineParser::parseIfThenElse() {
   assert(FormatTok->Tok.is(tok::kw_if) && "'if' expected");
   nextToken();
+  if (FormatTok->Tok.is(tok::kw_constexpr))
+    nextToken();
   if (FormatTok->Tok.is(tok::l_paren))
     parseParens();
   bool NeedsUnwrappedLine = false;
diff --git a/lib/Format/UsingDeclarationsSorter.cpp b/lib/Format/UsingDeclarationsSorter.cpp
new file mode 100644
index 000000000000..fb4f59fbc9bc
--- /dev/null
+++ b/lib/Format/UsingDeclarationsSorter.cpp
@@ -0,0 +1,144 @@
+//===--- UsingDeclarationsSorter.cpp ----------------------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \brief This file implements UsingDeclarationsSorter, a TokenAnalyzer that
+/// sorts consecutive using declarations.
+///
+//===----------------------------------------------------------------------===//
+
+#include "UsingDeclarationsSorter.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/Regex.h"
+
+#include <algorithm>
+
+#define DEBUG_TYPE "using-declarations-sorter"
+
+namespace clang {
+namespace format {
+
+namespace {
+
+struct UsingDeclaration {
+  const AnnotatedLine *Line;
+  std::string Label;
+
+  UsingDeclaration(const AnnotatedLine *Line, const std::string &Label)
+      : Line(Line), Label(Label) {}
+
+  bool operator<(const UsingDeclaration &Other) const {
+    return Label < Other.Label;
+  }
+};
+
+/// Computes the label of a using declaration starting at tthe using token
+/// \p UsingTok.
+/// If \p UsingTok doesn't begin a using declaration, returns the empty string.
+/// Note that this detects specifically using declarations, as in:
+/// using A::B::C;
+/// and not type aliases, as in:
+/// using A = B::C;
+/// Type aliases are in general not safe to permute.
+std::string computeUsingDeclarationLabel(const FormatToken *UsingTok) {
+  assert(UsingTok && UsingTok->is(tok::kw_using) && "Expecting a using token");
+  std::string Label;
+  const FormatToken *Tok = UsingTok->Next;
+  if (Tok && Tok->is(tok::kw_typename)) {
+    Label.append("typename ");
+    Tok = Tok->Next;
+  }
+  if (Tok && Tok->is(tok::coloncolon)) {
+    Label.append("::");
+    Tok = Tok->Next;
+  }
+  bool HasIdentifier = false;
+  while (Tok && Tok->is(tok::identifier)) {
+    HasIdentifier = true;
+    Label.append(Tok->TokenText.str());
+    Tok = Tok->Next;
+    if (!Tok || Tok->isNot(tok::coloncolon))
+      break;
+    Label.append("::");
+    Tok = Tok->Next;
+  }
+  if (HasIdentifier && Tok && Tok->isOneOf(tok::semi, tok::comma))
+    return Label;
+  return "";
+}
+
+void endUsingDeclarationBlock(
+    SmallVectorImpl<UsingDeclaration> *UsingDeclarations,
+    const SourceManager &SourceMgr, tooling::Replacements *Fixes) {
+  SmallVector<UsingDeclaration, 4> SortedUsingDeclarations(
+      UsingDeclarations->begin(), UsingDeclarations->end());
+  std::sort(SortedUsingDeclarations.begin(), SortedUsingDeclarations.end());
+  for (size_t I = 0, E = UsingDeclarations->size(); I < E; ++I) {
+    if ((*UsingDeclarations)[I].Line == SortedUsingDeclarations[I].Line)
+      continue;
+    auto Begin = (*UsingDeclarations)[I].Line->First->Tok.getLocation();
+    auto End = (*UsingDeclarations)[I].Line->Last->Tok.getEndLoc();
+    auto SortedBegin =
+        SortedUsingDeclarations[I].Line->First->Tok.getLocation();
+    auto SortedEnd = SortedUsingDeclarations[I].Line->Last->Tok.getEndLoc();
+    StringRef Text(SourceMgr.getCharacterData(SortedBegin),
+                   SourceMgr.getCharacterData(SortedEnd) -
+                       SourceMgr.getCharacterData(SortedBegin));
+    DEBUG({
+      StringRef OldText(SourceMgr.getCharacterData(Begin),
+                        SourceMgr.getCharacterData(End) -
+                            SourceMgr.getCharacterData(Begin));
+      llvm::dbgs() << "Replacing '" << OldText << "' with '" << Text << "'\n";
+    });
+    auto Range = CharSourceRange::getCharRange(Begin, End);
+    auto Err = Fixes->add(tooling::Replacement(SourceMgr, Range, Text));
+    if (Err) {
+      llvm::errs() << "Error while sorting using declarations: "
+                   << llvm::toString(std::move(Err)) << "\n";
+    }
+  }
+  UsingDeclarations->clear();
+}
+
+} // namespace
+
+UsingDeclarationsSorter::UsingDeclarationsSorter(const Environment &Env,
+                                                 const FormatStyle &Style)
+    : TokenAnalyzer(Env, Style) {}
+
+tooling::Replacements UsingDeclarationsSorter::analyze(
+    TokenAnnotator &Annotator, SmallVectorImpl<AnnotatedLine *> &AnnotatedLines,
+    FormatTokenLexer &Tokens) {
+  const SourceManager &SourceMgr = Env.getSourceManager();
+  AffectedRangeMgr.computeAffectedLines(AnnotatedLines.begin(),
+                                        AnnotatedLines.end());
+  tooling::Replacements Fixes;
+  SmallVector<UsingDeclaration, 4> UsingDeclarations;
+  for (size_t I = 0, E = AnnotatedLines.size(); I != E; ++I) {
+    if (!AnnotatedLines[I]->Affected || AnnotatedLines[I]->InPPDirective ||
+        !AnnotatedLines[I]->startsWith(tok::kw_using) ||
+        AnnotatedLines[I]->First->Finalized) {
+      endUsingDeclarationBlock(&UsingDeclarations, SourceMgr, &Fixes);
+      continue;
+    }
+    if (AnnotatedLines[I]->First->NewlinesBefore > 1)
+      endUsingDeclarationBlock(&UsingDeclarations, SourceMgr, &Fixes);
+    std::string Label = computeUsingDeclarationLabel(AnnotatedLines[I]->First);
+    if (Label.empty()) {
+      endUsingDeclarationBlock(&UsingDeclarations, SourceMgr, &Fixes);
+      continue;
+    }
+    UsingDeclarations.push_back(UsingDeclaration(AnnotatedLines[I], Label));
+  }
+  endUsingDeclarationBlock(&UsingDeclarations, SourceMgr, &Fixes);
+  return Fixes;
+}
+
+} // namespace format
+} // namespace clang
diff --git a/lib/Format/UsingDeclarationsSorter.h b/lib/Format/UsingDeclarationsSorter.h
new file mode 100644
index 000000000000..f7d5f97e3a2a
--- /dev/null
+++ b/lib/Format/UsingDeclarationsSorter.h
@@ -0,0 +1,37 @@
+//===--- UsingDeclarationsSorter.h ------------------------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \brief This file declares UsingDeclarationsSorter, a TokenAnalyzer that
+/// sorts consecutive using declarations.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_LIB_FORMAT_USINGDECLARATIONSSORTER_H
+#define LLVM_CLANG_LIB_FORMAT_USINGDECLARATIONSSORTER_H
+
+#include "TokenAnalyzer.h"
+
+namespace clang {
+namespace format {
+
+class UsingDeclarationsSorter : public TokenAnalyzer {
+public:
+  UsingDeclarationsSorter(const Environment &Env, const FormatStyle &Style);
+
+  tooling::Replacements
+  analyze(TokenAnnotator &Annotator,
+          SmallVectorImpl<AnnotatedLine *> &AnnotatedLines,
+          FormatTokenLexer &Tokens) override;
+};
+
+} // end namespace format
+} // end namespace clang
+
+#endif
diff --git a/lib/Frontend/ASTUnit.cpp b/lib/Frontend/ASTUnit.cpp
index 8d139f2c0a6d..1f34f10f55af 100644
--- a/lib/Frontend/ASTUnit.cpp
+++ b/lib/Frontend/ASTUnit.cpp
@@ -79,17 +79,6 @@ namespace {
       }
     }
   };
-  
-  struct OnDiskData {
-    /// \brief The file in which the precompiled preamble is stored.
-    std::string PreambleFile;
-
-    /// \brief Erase the preamble file.
-    void CleanPreambleFile();
-
-    /// \brief Erase temporary files and the preamble file.
-    void Cleanup();
-  };
 
   template <class T>
   std::unique_ptr<T> valueOrNull(llvm::ErrorOr<std::unique_ptr<T>> Val) {
@@ -105,81 +94,68 @@ namespace {
     Output = std::move(*Val);
     return true;
   }
-}
 
-static llvm::sys::SmartMutex<false> &getOnDiskMutex() {
-  static llvm::sys::SmartMutex<false> M(/* recursive = */ true);
-  return M;
-}
+/// \brief Get a source buffer for \p MainFilePath, handling all file-to-file
+/// and file-to-buffer remappings inside \p Invocation.
+static std::unique_ptr<llvm::MemoryBuffer>
+getBufferForFileHandlingRemapping(const CompilerInvocation &Invocation,
+                                  vfs::FileSystem *VFS,
+                                  StringRef FilePath) {
+  const auto &PreprocessorOpts = Invocation.getPreprocessorOpts();
 
-static void cleanupOnDiskMapAtExit();
+  // Try to determine if the main file has been remapped, either from the
+  // command line (to another file) or directly through the compiler
+  // invocation (to a memory buffer).
+  llvm::MemoryBuffer *Buffer = nullptr;
+  std::unique_ptr<llvm::MemoryBuffer> BufferOwner;
+  auto FileStatus = VFS->status(FilePath);
+  if (FileStatus) {
+    llvm::sys::fs::UniqueID MainFileID = FileStatus->getUniqueID();
 
-typedef llvm::DenseMap<const ASTUnit *,
-                       std::unique_ptr<OnDiskData>> OnDiskDataMap;
-static OnDiskDataMap &getOnDiskDataMap() {
-  static OnDiskDataMap M;
-  static bool hasRegisteredAtExit = false;
-  if (!hasRegisteredAtExit) {
-    hasRegisteredAtExit = true;
-    atexit(cleanupOnDiskMapAtExit);
-  }
-  return M;
-}
+    // Check whether there is a file-file remapping of the main file
+    for (const auto &RF : PreprocessorOpts.RemappedFiles) {
+      std::string MPath(RF.first);
+      auto MPathStatus = VFS->status(MPath);
+      if (MPathStatus) {
+        llvm::sys::fs::UniqueID MID = MPathStatus->getUniqueID();
+        if (MainFileID == MID) {
+          // We found a remapping. Try to load the resulting, remapped source.
+          BufferOwner = valueOrNull(VFS->getBufferForFile(RF.second));
+          if (!BufferOwner)
+            return nullptr;
+        }
+      }
+    }
 
-static void cleanupOnDiskMapAtExit() {
-  // Use the mutex because there can be an alive thread destroying an ASTUnit.
-  llvm::MutexGuard Guard(getOnDiskMutex());
-  for (const auto &I : getOnDiskDataMap()) {
-    // We don't worry about freeing the memory associated with OnDiskDataMap.
-    // All we care about is erasing stale files.
-    I.second->Cleanup();
+    // Check whether there is a file-buffer remapping. It supercedes the
+    // file-file remapping.
+    for (const auto &RB : PreprocessorOpts.RemappedFileBuffers) {
+      std::string MPath(RB.first);
+      auto MPathStatus = VFS->status(MPath);
+      if (MPathStatus) {
+        llvm::sys::fs::UniqueID MID = MPathStatus->getUniqueID();
+        if (MainFileID == MID) {
+          // We found a remapping.
+          BufferOwner.reset();
+          Buffer = const_cast<llvm::MemoryBuffer *>(RB.second);
+        }
+      }
+    }
   }
-}
-
-static OnDiskData &getOnDiskData(const ASTUnit *AU) {
-  // We require the mutex since we are modifying the structure of the
-  // DenseMap.
-  llvm::MutexGuard Guard(getOnDiskMutex());
-  OnDiskDataMap &M = getOnDiskDataMap();
-  auto &D = M[AU];
-  if (!D)
-    D = llvm::make_unique<OnDiskData>();
-  return *D;
-}
-
-static void erasePreambleFile(const ASTUnit *AU) {
-  getOnDiskData(AU).CleanPreambleFile();
-}
 
-static void removeOnDiskEntry(const ASTUnit *AU) {
-  // We require the mutex since we are modifying the structure of the
-  // DenseMap.
-  llvm::MutexGuard Guard(getOnDiskMutex());
-  OnDiskDataMap &M = getOnDiskDataMap();
-  OnDiskDataMap::iterator I = M.find(AU);
-  if (I != M.end()) {
-    I->second->Cleanup();
-    M.erase(I);
+  // If the main source file was not remapped, load it now.
+  if (!Buffer && !BufferOwner) {
+    BufferOwner = valueOrNull(VFS->getBufferForFile(FilePath));
+    if (!BufferOwner)
+      return nullptr;
   }
-}
-
-static void setPreambleFile(const ASTUnit *AU, StringRef preambleFile) {
-  getOnDiskData(AU).PreambleFile = preambleFile;
-}
 
-static const std::string &getPreambleFile(const ASTUnit *AU) {
-  return getOnDiskData(AU).PreambleFile;  
+  if (BufferOwner)
+    return BufferOwner;
+  if (!Buffer)
+    return nullptr;
+  return llvm::MemoryBuffer::getMemBufferCopy(Buffer->getBuffer(), FilePath);
 }
-
-void OnDiskData::CleanPreambleFile() {
-  if (!PreambleFile.empty()) {
-    llvm::sys::fs::remove(PreambleFile);
-    PreambleFile.clear();
-  }
-}
-
-void OnDiskData::Cleanup() {
-  CleanPreambleFile();
 }
 
 struct ASTUnit::ASTWriterData {
@@ -233,9 +209,6 @@ ASTUnit::~ASTUnit() {
 
   clearFileLevelDecls();
 
-  // Clean up the temporary files and the preamble file.
-  removeOnDiskEntry(this);
-
   // Free the buffers associated with remapped files. We are required to
   // perform this operation here because we explicitly request that the
   // compiler instance *not* free these buffers for each invocation of the
@@ -575,16 +548,24 @@ private:
 
   /// \brief Diagnostic consumer that saves each diagnostic it is given.
 class StoredDiagnosticConsumer : public DiagnosticConsumer {
-  SmallVectorImpl<StoredDiagnostic> &StoredDiags;
+  SmallVectorImpl<StoredDiagnostic> *StoredDiags;
+  SmallVectorImpl<ASTUnit::StandaloneDiagnostic> *StandaloneDiags;
+  const LangOptions *LangOpts;
   SourceManager *SourceMgr;
 
 public:
-  explicit StoredDiagnosticConsumer(
-                          SmallVectorImpl<StoredDiagnostic> &StoredDiags)
-    : StoredDiags(StoredDiags), SourceMgr(nullptr) {}
+  StoredDiagnosticConsumer(
+      SmallVectorImpl<StoredDiagnostic> *StoredDiags,
+      SmallVectorImpl<ASTUnit::StandaloneDiagnostic> *StandaloneDiags)
+      : StoredDiags(StoredDiags), StandaloneDiags(StandaloneDiags),
+        LangOpts(nullptr), SourceMgr(nullptr) {
+    assert((StoredDiags || StandaloneDiags) &&
+           "No output collections were passed to StoredDiagnosticConsumer.");
+  }
 
   void BeginSourceFile(const LangOptions &LangOpts,
                        const Preprocessor *PP = nullptr) override {
+    this->LangOpts = &LangOpts;
     if (PP)
       SourceMgr = &PP->getSourceManager();
   }
@@ -603,8 +584,9 @@ class CaptureDroppedDiagnostics {
 
 public:
   CaptureDroppedDiagnostics(bool RequestCapture, DiagnosticsEngine &Diags,
-                          SmallVectorImpl<StoredDiagnostic> &StoredDiags)
-    : Diags(Diags), Client(StoredDiags), PreviousClient(nullptr)
+                            SmallVectorImpl<StoredDiagnostic> *StoredDiags,
+                            SmallVectorImpl<ASTUnit::StandaloneDiagnostic> *StandaloneDiags)
+      : Diags(Diags), Client(StoredDiags, StandaloneDiags), PreviousClient(nullptr)
   {
     if (RequestCapture || Diags.getClient() == nullptr) {
       OwningPreviousClient = Diags.takeClient();
@@ -621,16 +603,35 @@ public:
 
 } // anonymous namespace
 
+static ASTUnit::StandaloneDiagnostic
+makeStandaloneDiagnostic(const LangOptions &LangOpts,
+                         const StoredDiagnostic &InDiag);
+
 void StoredDiagnosticConsumer::HandleDiagnostic(DiagnosticsEngine::Level Level,
-                                              const Diagnostic &Info) {
+                                                const Diagnostic &Info) {
   // Default implementation (Warnings/errors count).
   DiagnosticConsumer::HandleDiagnostic(Level, Info);
 
   // Only record the diagnostic if it's part of the source manager we know
   // about. This effectively drops diagnostics from modules we're building.
   // FIXME: In the long run, ee don't want to drop source managers from modules.
-  if (!Info.hasSourceManager() || &Info.getSourceManager() == SourceMgr)
-    StoredDiags.emplace_back(Level, Info);
+  if (!Info.hasSourceManager() || &Info.getSourceManager() == SourceMgr) {
+    StoredDiagnostic *ResultDiag = nullptr;
+    if (StoredDiags) {
+      StoredDiags->emplace_back(Level, Info);
+      ResultDiag = &StoredDiags->back();
+    }
+
+    if (StandaloneDiags) {
+      llvm::Optional<StoredDiagnostic> StoredDiag = llvm::None;
+      if (!ResultDiag) {
+        StoredDiag.emplace(Level, Info);
+        ResultDiag = StoredDiag.getPointer();
+      }
+      StandaloneDiags->push_back(
+          makeStandaloneDiagnostic(*LangOpts, *ResultDiag));
+    }
+  }
 }
 
 IntrusiveRefCntPtr<ASTReader> ASTUnit::getASTReader() const {
@@ -665,7 +666,7 @@ void ASTUnit::ConfigureDiags(IntrusiveRefCntPtr<DiagnosticsEngine> Diags,
                              ASTUnit &AST, bool CaptureDiagnostics) {
   assert(Diags.get() && "no DiagnosticsEngine was provided");
   if (CaptureDiagnostics)
-    Diags->setClient(new StoredDiagnosticConsumer(AST.StoredDiagnostics));
+    Diags->setClient(new StoredDiagnosticConsumer(&AST.StoredDiagnostics, nullptr));
 }
 
 std::unique_ptr<ASTUnit> ASTUnit::LoadFromASTFile(
@@ -780,6 +781,11 @@ std::unique_ptr<ASTUnit> ASTUnit::LoadFromASTFile(
 
 namespace {
 
+/// \brief Add the given macro to the hash of all top-level entities.
+void AddDefinedMacroToHash(const Token &MacroNameTok, unsigned &Hash) {
+  Hash = llvm::HashString(MacroNameTok.getIdentifierInfo()->getName(), Hash);
+}
+
 /// \brief Preprocessor callback class that updates a hash value with the names 
 /// of all macros that have been defined by the translation unit.
 class MacroDefinitionTrackerPPCallbacks : public PPCallbacks {
@@ -790,7 +796,7 @@ public:
 
   void MacroDefined(const Token &MacroNameTok,
                     const MacroDirective *MD) override {
-    Hash = llvm::HashString(MacroNameTok.getIdentifierInfo()->getName(), Hash);
+    AddDefinedMacroToHash(MacroNameTok, Hash);
   }
 };
 
@@ -916,45 +922,27 @@ public:
   }
 };
 
-class PrecompilePreambleAction : public ASTFrontendAction {
-  ASTUnit &Unit;
-  bool HasEmittedPreamblePCH;
-
+class ASTUnitPreambleCallbacks : public PreambleCallbacks {
 public:
-  explicit PrecompilePreambleAction(ASTUnit &Unit)
-      : Unit(Unit), HasEmittedPreamblePCH(false) {}
+  unsigned getHash() const { return Hash; }
 
-  std::unique_ptr<ASTConsumer> CreateASTConsumer(CompilerInstance &CI,
-                                                 StringRef InFile) override;
-  bool hasEmittedPreamblePCH() const { return HasEmittedPreamblePCH; }
-  void setHasEmittedPreamblePCH() { HasEmittedPreamblePCH = true; }
-  bool shouldEraseOutputFiles() override { return !hasEmittedPreamblePCH(); }
-
-  bool hasCodeCompletionSupport() const override { return false; }
-  bool hasASTFileSupport() const override { return false; }
-  TranslationUnitKind getTranslationUnitKind() override { return TU_Prefix; }
-};
+  std::vector<Decl *> takeTopLevelDecls() { return std::move(TopLevelDecls); }
 
-class PrecompilePreambleConsumer : public PCHGenerator {
-  ASTUnit &Unit;
-  unsigned &Hash;
-  std::vector<Decl *> TopLevelDecls;
-  PrecompilePreambleAction *Action;
-  std::unique_ptr<raw_ostream> Out;
+  std::vector<serialization::DeclID> takeTopLevelDeclIDs() {
+    return std::move(TopLevelDeclIDs);
+  }
 
-public:
-  PrecompilePreambleConsumer(ASTUnit &Unit, PrecompilePreambleAction *Action,
-                             const Preprocessor &PP, StringRef isysroot,
-                             std::unique_ptr<raw_ostream> Out)
-      : PCHGenerator(PP, "", isysroot, std::make_shared<PCHBuffer>(),
-                     ArrayRef<std::shared_ptr<ModuleFileExtension>>(),
-                     /*AllowASTWithErrors=*/true),
-        Unit(Unit), Hash(Unit.getCurrentTopLevelHashValue()), Action(Action),
-        Out(std::move(Out)) {
-    Hash = 0;
+  void AfterPCHEmitted(ASTWriter &Writer) override {
+    TopLevelDeclIDs.reserve(TopLevelDecls.size());
+    for (Decl *D : TopLevelDecls) {
+      // Invalid top-level decls may not have been serialized.
+      if (D->isInvalidDecl())
+        continue;
+      TopLevelDeclIDs.push_back(Writer.getDeclID(D));
+    }
   }
 
-  bool HandleTopLevelDecl(DeclGroupRef DG) override {
+  void HandleTopLevelDecl(DeclGroupRef DG) override {
     for (Decl *D : DG) {
       // FIXME: Currently ObjC method declarations are incorrectly being
       // reported as top-level declarations, even though their DeclContext
@@ -965,59 +953,22 @@ public:
       AddTopLevelDeclarationToHash(D, Hash);
       TopLevelDecls.push_back(D);
     }
-    return true;
   }
 
-  void HandleTranslationUnit(ASTContext &Ctx) override {
-    PCHGenerator::HandleTranslationUnit(Ctx);
-    if (hasEmittedPCH()) {
-      // Write the generated bitstream to "Out".
-      *Out << getPCH();
-      // Make sure it hits disk now.
-      Out->flush();
-      // Free the buffer.
-      llvm::SmallVector<char, 0> Empty;
-      getPCH() = std::move(Empty);
-
-      // Translate the top-level declarations we captured during
-      // parsing into declaration IDs in the precompiled
-      // preamble. This will allow us to deserialize those top-level
-      // declarations when requested.
-      for (Decl *D : TopLevelDecls) {
-        // Invalid top-level decls may not have been serialized.
-        if (D->isInvalidDecl())
-          continue;
-        Unit.addTopLevelDeclFromPreamble(getWriter().getDeclID(D));
-      }
-
-      Action->setHasEmittedPreamblePCH();
-    }
+  void HandleMacroDefined(const Token &MacroNameTok,
+                          const MacroDirective *MD) override {
+    AddDefinedMacroToHash(MacroNameTok, Hash);
   }
+
+private:
+  unsigned Hash = 0;
+  std::vector<Decl *> TopLevelDecls;
+  std::vector<serialization::DeclID> TopLevelDeclIDs;
+  llvm::SmallVector<ASTUnit::StandaloneDiagnostic, 4> PreambleDiags;
 };
 
 } // anonymous namespace
 
-std::unique_ptr<ASTConsumer>
-PrecompilePreambleAction::CreateASTConsumer(CompilerInstance &CI,
-                                            StringRef InFile) {
-  std::string Sysroot;
-  std::string OutputFile;
-  std::unique_ptr<raw_ostream> OS =
-      GeneratePCHAction::ComputeASTConsumerArguments(CI, InFile, Sysroot,
-                                                     OutputFile);
-  if (!OS)
-    return nullptr;
-
-  if (!CI.getFrontendOpts().RelocatablePCH)
-    Sysroot.clear();
-
-  CI.getPreprocessor().addPPCallbacks(
-      llvm::make_unique<MacroDefinitionTrackerPPCallbacks>(
-                                           Unit.getCurrentTopLevelHashValue()));
-  return llvm::make_unique<PrecompilePreambleConsumer>(
-      Unit, this, CI.getPreprocessor(), Sysroot, std::move(OS));
-}
-
 static bool isNonDriverDiag(const StoredDiagnostic &StoredDiag) {
   return StoredDiag.getLocation().isValid();
 }
@@ -1125,15 +1076,9 @@ bool ASTUnit::Parse(std::shared_ptr<PCHContainerOperations> PCHContainerOps,
   
   // If the main file has been overridden due to the use of a preamble,
   // make that override happen and introduce the preamble.
-  PreprocessorOptions &PreprocessorOpts = Clang->getPreprocessorOpts();
   if (OverrideMainBuffer) {
-    PreprocessorOpts.addRemappedFile(OriginalSourceFile,
-                                     OverrideMainBuffer.get());
-    PreprocessorOpts.PrecompiledPreambleBytes.first = Preamble.size();
-    PreprocessorOpts.PrecompiledPreambleBytes.second
-                                                    = PreambleEndsAtStartOfLine;
-    PreprocessorOpts.ImplicitPCHInclude = getPreambleFile(this);
-    PreprocessorOpts.DisablePCHValidation = true;
+    assert(Preamble && "No preamble was built, but OverrideMainBuffer is not null");
+    Preamble->AddImplicitPreamble(Clang->getInvocation(), OverrideMainBuffer.get());
     
     // The stored diagnostic has the old source manager in it; update
     // the locations to refer into the new source manager. Since we've
@@ -1186,116 +1131,6 @@ error:
   return true;
 }
 
-/// \brief Simple function to retrieve a path for a preamble precompiled header.
-static std::string GetPreamblePCHPath() {
-  // FIXME: This is a hack so that we can override the preamble file during
-  // crash-recovery testing, which is the only case where the preamble files
-  // are not necessarily cleaned up.
-  const char *TmpFile = ::getenv("CINDEXTEST_PREAMBLE_FILE");
-  if (TmpFile)
-    return TmpFile;
-
-  SmallString<128> Path;
-  llvm::sys::fs::createTemporaryFile("preamble", "pch", Path);
-
-  return Path.str();
-}
-
-/// \brief Compute the preamble for the main file, providing the source buffer
-/// that corresponds to the main file along with a pair (bytes, start-of-line)
-/// that describes the preamble.
-ASTUnit::ComputedPreamble
-ASTUnit::ComputePreamble(CompilerInvocation &Invocation, unsigned MaxLines,
-                         IntrusiveRefCntPtr<vfs::FileSystem> VFS) {
-  FrontendOptions &FrontendOpts = Invocation.getFrontendOpts();
-  PreprocessorOptions &PreprocessorOpts = Invocation.getPreprocessorOpts();
-  
-  // Try to determine if the main file has been remapped, either from the 
-  // command line (to another file) or directly through the compiler invocation
-  // (to a memory buffer).
-  llvm::MemoryBuffer *Buffer = nullptr;
-  std::unique_ptr<llvm::MemoryBuffer> BufferOwner;
-  std::string MainFilePath(FrontendOpts.Inputs[0].getFile());
-  auto MainFileStatus = VFS->status(MainFilePath);
-  if (MainFileStatus) {
-    llvm::sys::fs::UniqueID MainFileID = MainFileStatus->getUniqueID();
-
-    // Check whether there is a file-file remapping of the main file
-    for (const auto &RF : PreprocessorOpts.RemappedFiles) {
-      std::string MPath(RF.first);
-      auto MPathStatus = VFS->status(MPath);
-      if (MPathStatus) {
-        llvm::sys::fs::UniqueID MID = MPathStatus->getUniqueID();
-        if (MainFileID == MID) {
-          // We found a remapping. Try to load the resulting, remapped source.
-          BufferOwner = valueOrNull(VFS->getBufferForFile(RF.second));
-          if (!BufferOwner)
-            return ComputedPreamble(nullptr, nullptr, 0, true);
-        }
-      }
-    }
-
-    // Check whether there is a file-buffer remapping. It supercedes the
-    // file-file remapping.
-    for (const auto &RB : PreprocessorOpts.RemappedFileBuffers) {
-      std::string MPath(RB.first);
-      auto MPathStatus = VFS->status(MPath);
-      if (MPathStatus) {
-        llvm::sys::fs::UniqueID MID = MPathStatus->getUniqueID();
-        if (MainFileID == MID) {
-          // We found a remapping.
-          BufferOwner.reset();
-          Buffer = const_cast<llvm::MemoryBuffer *>(RB.second);
-        }
-      }
-    }
-  }
-  
-  // If the main source file was not remapped, load it now.
-  if (!Buffer && !BufferOwner) {
-    BufferOwner = valueOrNull(VFS->getBufferForFile(FrontendOpts.Inputs[0].getFile()));
-    if (!BufferOwner)
-      return ComputedPreamble(nullptr, nullptr, 0, true);
-  }
-
-  if (!Buffer)
-    Buffer = BufferOwner.get();
-  auto Pre = Lexer::ComputePreamble(Buffer->getBuffer(),
-                                    *Invocation.getLangOpts(), MaxLines);
-  return ComputedPreamble(Buffer, std::move(BufferOwner), Pre.first,
-                          Pre.second);
-}
-
-ASTUnit::PreambleFileHash
-ASTUnit::PreambleFileHash::createForFile(off_t Size, time_t ModTime) {
-  PreambleFileHash Result;
-  Result.Size = Size;
-  Result.ModTime = ModTime;
-  Result.MD5 = {};
-  return Result;
-}
-
-ASTUnit::PreambleFileHash ASTUnit::PreambleFileHash::createForMemoryBuffer(
-    const llvm::MemoryBuffer *Buffer) {
-  PreambleFileHash Result;
-  Result.Size = Buffer->getBufferSize();
-  Result.ModTime = 0;
-
-  llvm::MD5 MD5Ctx;
-  MD5Ctx.update(Buffer->getBuffer().data());
-  MD5Ctx.final(Result.MD5);
-
-  return Result;
-}
-
-namespace clang {
-bool operator==(const ASTUnit::PreambleFileHash &LHS,
-                const ASTUnit::PreambleFileHash &RHS) {
-  return LHS.Size == RHS.Size && LHS.ModTime == RHS.ModTime &&
-         LHS.MD5 == RHS.MD5;
-}
-} // namespace clang
-
 static std::pair<unsigned, unsigned>
 makeStandaloneRange(CharSourceRange Range, const SourceManager &SM,
                     const LangOptions &LangOpts) {
@@ -1367,135 +1202,41 @@ ASTUnit::getMainBufferWithPrecompiledPreamble(
     const CompilerInvocation &PreambleInvocationIn,
     IntrusiveRefCntPtr<vfs::FileSystem> VFS, bool AllowRebuild,
     unsigned MaxLines) {
-  assert(VFS && "VFS is null");
-
-  auto PreambleInvocation =
-      std::make_shared<CompilerInvocation>(PreambleInvocationIn);
-  FrontendOptions &FrontendOpts = PreambleInvocation->getFrontendOpts();
-  PreprocessorOptions &PreprocessorOpts
-    = PreambleInvocation->getPreprocessorOpts();
-
-  ComputedPreamble NewPreamble =
-      ComputePreamble(*PreambleInvocation, MaxLines, VFS);
 
-  if (!NewPreamble.Size) {
-    // We couldn't find a preamble in the main source. Clear out the current
-    // preamble, if we have one. It's obviously no good any more.
-    Preamble.clear();
-    erasePreambleFile(this);
-
-    // The next time we actually see a preamble, precompile it.
-    PreambleRebuildCounter = 1;
+  auto MainFilePath =
+      PreambleInvocationIn.getFrontendOpts().Inputs[0].getFile();
+  std::unique_ptr<llvm::MemoryBuffer> MainFileBuffer =
+      getBufferForFileHandlingRemapping(PreambleInvocationIn, VFS.get(),
+                                        MainFilePath);
+  if (!MainFileBuffer)
     return nullptr;
-  }
-  
-  if (!Preamble.empty()) {
-    // We've previously computed a preamble. Check whether we have the same
-    // preamble now that we did before, and that there's enough space in
-    // the main-file buffer within the precompiled preamble to fit the
-    // new main file.
-    if (Preamble.size() == NewPreamble.Size &&
-        PreambleEndsAtStartOfLine == NewPreamble.PreambleEndsAtStartOfLine &&
-        memcmp(Preamble.getBufferStart(), NewPreamble.Buffer->getBufferStart(),
-               NewPreamble.Size) == 0) {
-      // The preamble has not changed. We may be able to re-use the precompiled
-      // preamble.
-
-      // Check that none of the files used by the preamble have changed.
-      bool AnyFileChanged = false;
-          
-      // First, make a record of those files that have been overridden via
-      // remapping or unsaved_files.
-      std::map<llvm::sys::fs::UniqueID, PreambleFileHash> OverriddenFiles;
-      for (const auto &R : PreprocessorOpts.RemappedFiles) {
-        if (AnyFileChanged)
-          break;
-
-        vfs::Status Status;
-        if (!moveOnNoError(VFS->status(R.second), Status)) {
-          // If we can't stat the file we're remapping to, assume that something
-          // horrible happened.
-          AnyFileChanged = true;
-          break;
-        }
 
-        OverriddenFiles[Status.getUniqueID()] = PreambleFileHash::createForFile(
-            Status.getSize(),
-            llvm::sys::toTimeT(Status.getLastModificationTime()));
-      }
-
-      for (const auto &RB : PreprocessorOpts.RemappedFileBuffers) {
-        if (AnyFileChanged)
-          break;
+  PreambleBounds Bounds =
+      ComputePreambleBounds(*PreambleInvocationIn.getLangOpts(),
+                            MainFileBuffer.get(), MaxLines);
+  if (!Bounds.Size)
+    return nullptr;
 
-        vfs::Status Status;
-        if (!moveOnNoError(VFS->status(RB.first), Status)) {
-          AnyFileChanged = true;
-          break;
-        }
+  if (Preamble) {
+    if (Preamble->CanReuse(PreambleInvocationIn, MainFileBuffer.get(), Bounds,
+                           VFS.get())) {
+      // Okay! We can re-use the precompiled preamble.
 
-        OverriddenFiles[Status.getUniqueID()] =
-            PreambleFileHash::createForMemoryBuffer(RB.second);
-      }
-       
-      // Check whether anything has changed.
-      for (llvm::StringMap<PreambleFileHash>::iterator
-             F = FilesInPreamble.begin(), FEnd = FilesInPreamble.end();
-           !AnyFileChanged && F != FEnd; 
-           ++F) {
-        vfs::Status Status;
-        if (!moveOnNoError(VFS->status(F->first()), Status)) {
-          // If we can't stat the file, assume that something horrible happened.
-          AnyFileChanged = true;
-          break;
-        }
+      // Set the state of the diagnostic object to mimic its state
+      // after parsing the preamble.
+      getDiagnostics().Reset();
+      ProcessWarningOptions(getDiagnostics(),
+                            PreambleInvocationIn.getDiagnosticOpts());
+      getDiagnostics().setNumWarnings(NumWarningsInPreamble);
 
-        std::map<llvm::sys::fs::UniqueID, PreambleFileHash>::iterator Overridden
-          = OverriddenFiles.find(Status.getUniqueID());
-        if (Overridden != OverriddenFiles.end()) {
-          // This file was remapped; check whether the newly-mapped file 
-          // matches up with the previous mapping.
-          if (Overridden->second != F->second)
-            AnyFileChanged = true;
-          continue;
-        }
-        
-        // The file was not remapped; check whether it has changed on disk.
-        if (Status.getSize() != uint64_t(F->second.Size) ||
-            llvm::sys::toTimeT(Status.getLastModificationTime()) !=
-                F->second.ModTime)
-          AnyFileChanged = true;
-      }
-          
-      if (!AnyFileChanged) {
-        // Okay! We can re-use the precompiled preamble.
-
-        // Set the state of the diagnostic object to mimic its state
-        // after parsing the preamble.
-        getDiagnostics().Reset();
-        ProcessWarningOptions(getDiagnostics(), 
-                              PreambleInvocation->getDiagnosticOpts());
-        getDiagnostics().setNumWarnings(NumWarningsInPreamble);
-
-        return llvm::MemoryBuffer::getMemBufferCopy(
-            NewPreamble.Buffer->getBuffer(), FrontendOpts.Inputs[0].getFile());
-      }
+      PreambleRebuildCounter = 1;
+      return MainFileBuffer;
+    } else {
+      Preamble.reset();
+      PreambleDiagnostics.clear();
+      TopLevelDeclsInPreamble.clear();
+      PreambleRebuildCounter = 1;
     }
-
-    // If we aren't allowed to rebuild the precompiled preamble, just
-    // return now.
-    if (!AllowRebuild)
-      return nullptr;
-
-    // We can't reuse the previously-computed preamble. Build a new one.
-    Preamble.clear();
-    PreambleDiagnostics.clear();
-    erasePreambleFile(this);
-    PreambleRebuildCounter = 1;
-  } else if (!AllowRebuild) {
-    // We aren't allowed to rebuild the precompiled preamble; just
-    // return now.
-    return nullptr;
   }
 
   // If the preamble rebuild counter > 1, it's because we previously
@@ -1506,164 +1247,61 @@ ASTUnit::getMainBufferWithPrecompiledPreamble(
     return nullptr;
   }
 
-  // Create a temporary file for the precompiled preamble. In rare 
-  // circumstances, this can fail.
-  std::string PreamblePCHPath = GetPreamblePCHPath();
-  if (PreamblePCHPath.empty()) {
-    // Try again next time.
-    PreambleRebuildCounter = 1;
+  assert(!Preamble && "No Preamble should be stored at that point");
+  // If we aren't allowed to rebuild the precompiled preamble, just
+  // return now.
+  if (!AllowRebuild)
     return nullptr;
-  }
-  
-  // We did not previously compute a preamble, or it can't be reused anyway.
-  SimpleTimer PreambleTimer(WantTiming);
-  PreambleTimer.setOutput("Precompiling preamble");
-
-  // Save the preamble text for later; we'll need to compare against it for
-  // subsequent reparses.
-  StringRef MainFilename = FrontendOpts.Inputs[0].getFile();
-  Preamble.assign(FileMgr->getFile(MainFilename),
-                  NewPreamble.Buffer->getBufferStart(),
-                  NewPreamble.Buffer->getBufferStart() + NewPreamble.Size);
-  PreambleEndsAtStartOfLine = NewPreamble.PreambleEndsAtStartOfLine;
-
-  PreambleBuffer = llvm::MemoryBuffer::getMemBufferCopy(
-      NewPreamble.Buffer->getBuffer().slice(0, Preamble.size()), MainFilename);
-
-  // Remap the main source file to the preamble buffer.
-  StringRef MainFilePath = FrontendOpts.Inputs[0].getFile();
-  PreprocessorOpts.addRemappedFile(MainFilePath, PreambleBuffer.get());
-
-  // Tell the compiler invocation to generate a temporary precompiled header.
-  FrontendOpts.ProgramAction = frontend::GeneratePCH;
-  // FIXME: Generate the precompiled header into memory?
-  FrontendOpts.OutputFile = PreamblePCHPath;
-  PreprocessorOpts.PrecompiledPreambleBytes.first = 0;
-  PreprocessorOpts.PrecompiledPreambleBytes.second = false;
-  
-  // Create the compiler instance to use for building the precompiled preamble.
-  std::unique_ptr<CompilerInstance> Clang(
-      new CompilerInstance(std::move(PCHContainerOps)));
 
-  // Recover resources if we crash before exiting this method.
-  llvm::CrashRecoveryContextCleanupRegistrar<CompilerInstance>
-    CICleanup(Clang.get());
-
-  Clang->setInvocation(std::move(PreambleInvocation));
-  OriginalSourceFile = Clang->getFrontendOpts().Inputs[0].getFile();
-  
-  // Set up diagnostics, capturing all of the diagnostics produced.
-  Clang->setDiagnostics(&getDiagnostics());
-  
-  // Create the target instance.
-  Clang->setTarget(TargetInfo::CreateTargetInfo(
-      Clang->getDiagnostics(), Clang->getInvocation().TargetOpts));
-  if (!Clang->hasTarget()) {
-    llvm::sys::fs::remove(FrontendOpts.OutputFile);
-    Preamble.clear();
-    PreambleRebuildCounter = DefaultPreambleRebuildInterval;
-    PreprocessorOpts.RemappedFileBuffers.pop_back();
-    return nullptr;
-  }
-  
-  // Inform the target of the language options.
-  //
-  // FIXME: We shouldn't need to do this, the target should be immutable once
-  // created. This complexity should be lifted elsewhere.
-  Clang->getTarget().adjust(Clang->getLangOpts());
-  
-  assert(Clang->getFrontendOpts().Inputs.size() == 1 &&
-         "Invocation must have exactly one source file!");
-  assert(Clang->getFrontendOpts().Inputs[0].getKind().getFormat() ==
-             InputKind::Source &&
-         "FIXME: AST inputs not yet supported here!");
-  assert(Clang->getFrontendOpts().Inputs[0].getKind().getLanguage() !=
-             InputKind::LLVM_IR &&
-         "IR inputs not support here!");
-  
-  // Clear out old caches and data.
-  getDiagnostics().Reset();
-  ProcessWarningOptions(getDiagnostics(), Clang->getDiagnosticOpts());
-  checkAndRemoveNonDriverDiags(StoredDiagnostics);
-  TopLevelDecls.clear();
-  TopLevelDeclsInPreamble.clear();
-  PreambleDiagnostics.clear();
-
-  VFS = createVFSFromCompilerInvocation(Clang->getInvocation(),
-                                        getDiagnostics(), VFS);
-  if (!VFS)
-    return nullptr;
-
-  // Create a file manager object to provide access to and cache the filesystem.
-  Clang->setFileManager(new FileManager(Clang->getFileSystemOpts(), VFS));
-
-  // Create the source manager.
-  Clang->setSourceManager(new SourceManager(getDiagnostics(),
-                                            Clang->getFileManager()));
-
-  auto PreambleDepCollector = std::make_shared<DependencyCollector>();
-  Clang->addDependencyCollector(PreambleDepCollector);
-
-  std::unique_ptr<PrecompilePreambleAction> Act;
-  Act.reset(new PrecompilePreambleAction(*this));
-  if (!Act->BeginSourceFile(*Clang.get(), Clang->getFrontendOpts().Inputs[0])) {
-    llvm::sys::fs::remove(FrontendOpts.OutputFile);
-    Preamble.clear();
-    PreambleRebuildCounter = DefaultPreambleRebuildInterval;
-    PreprocessorOpts.RemappedFileBuffers.pop_back();
-    return nullptr;
+  SmallVector<StandaloneDiagnostic, 4> NewPreambleDiagsStandalone;
+  SmallVector<StoredDiagnostic, 4> NewPreambleDiags;
+  ASTUnitPreambleCallbacks Callbacks;
+  {
+    llvm::Optional<CaptureDroppedDiagnostics> Capture;
+    if (CaptureDiagnostics)
+      Capture.emplace(/*RequestCapture=*/true, *Diagnostics, &NewPreambleDiags,
+                      &NewPreambleDiagsStandalone);
+
+    // We did not previously compute a preamble, or it can't be reused anyway.
+    SimpleTimer PreambleTimer(WantTiming);
+    PreambleTimer.setOutput("Precompiling preamble");
+
+    llvm::ErrorOr<PrecompiledPreamble> NewPreamble = PrecompiledPreamble::Build(
+        PreambleInvocationIn, MainFileBuffer.get(), Bounds, *Diagnostics, VFS,
+        PCHContainerOps, Callbacks);
+    if (NewPreamble) {
+      Preamble = std::move(*NewPreamble);
+      PreambleRebuildCounter = 1;
+    } else {
+      switch (static_cast<BuildPreambleError>(NewPreamble.getError().value())) {
+      case BuildPreambleError::CouldntCreateTempFile:
+      case BuildPreambleError::PreambleIsEmpty:
+        // Try again next time.
+        PreambleRebuildCounter = 1;
+        return nullptr;
+      case BuildPreambleError::CouldntCreateTargetInfo:
+      case BuildPreambleError::BeginSourceFileFailed:
+      case BuildPreambleError::CouldntEmitPCH:
+      case BuildPreambleError::CouldntCreateVFSOverlay:
+        // These erros are more likely to repeat, retry after some period.
+        PreambleRebuildCounter = DefaultPreambleRebuildInterval;
+        return nullptr;
+      }
+      llvm_unreachable("unexpected BuildPreambleError");
+    }
   }
-  
-  Act->Execute();
 
-  // Transfer any diagnostics generated when parsing the preamble into the set
-  // of preamble diagnostics.
-  for (stored_diag_iterator I = stored_diag_afterDriver_begin(),
-                            E = stored_diag_end();
-       I != E; ++I)
-    PreambleDiagnostics.push_back(
-        makeStandaloneDiagnostic(Clang->getLangOpts(), *I));
+  assert(Preamble && "Preamble wasn't built");
 
-  Act->EndSourceFile();
-
-  checkAndRemoveNonDriverDiags(StoredDiagnostics);
+  TopLevelDecls.clear();
+  TopLevelDeclsInPreamble = Callbacks.takeTopLevelDeclIDs();
+  PreambleTopLevelHashValue = Callbacks.getHash();
 
-  if (!Act->hasEmittedPreamblePCH()) {
-    // The preamble PCH failed (e.g. there was a module loading fatal error),
-    // so no precompiled header was generated. Forget that we even tried.
-    // FIXME: Should we leave a note for ourselves to try again?
-    llvm::sys::fs::remove(FrontendOpts.OutputFile);
-    Preamble.clear();
-    TopLevelDeclsInPreamble.clear();
-    PreambleRebuildCounter = DefaultPreambleRebuildInterval;
-    PreprocessorOpts.RemappedFileBuffers.pop_back();
-    return nullptr;
-  }
-  
-  // Keep track of the preamble we precompiled.
-  setPreambleFile(this, FrontendOpts.OutputFile);
   NumWarningsInPreamble = getDiagnostics().getNumWarnings();
-  
-  // Keep track of all of the files that the source manager knows about,
-  // so we can verify whether they have changed or not.
-  FilesInPreamble.clear();
-  SourceManager &SourceMgr = Clang->getSourceManager();
-  for (auto &Filename : PreambleDepCollector->getDependencies()) {
-    const FileEntry *File = Clang->getFileManager().getFile(Filename);
-    if (!File || File == SourceMgr.getFileEntryForID(SourceMgr.getMainFileID()))
-      continue;
-    if (time_t ModTime = File->getModificationTime()) {
-      FilesInPreamble[File->getName()] = PreambleFileHash::createForFile(
-          File->getSize(), ModTime);
-    } else {
-      llvm::MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File);
-      FilesInPreamble[File->getName()] =
-          PreambleFileHash::createForMemoryBuffer(Buffer);
-    }
-  }
 
-  PreambleRebuildCounter = 1;
-  PreprocessorOpts.RemappedFileBuffers.pop_back();
+  checkAndRemoveNonDriverDiags(NewPreambleDiags);
+  StoredDiagnostics = std::move(NewPreambleDiags);
+  PreambleDiagnostics = std::move(NewPreambleDiagsStandalone);
 
   // If the hash of top-level entities differs from the hash of the top-level
   // entities the last time we rebuilt the preamble, clear out the completion
@@ -1673,11 +1311,12 @@ ASTUnit::getMainBufferWithPrecompiledPreamble(
     PreambleTopLevelHashValue = CurrentTopLevelHashValue;
   }
 
-  return llvm::MemoryBuffer::getMemBufferCopy(NewPreamble.Buffer->getBuffer(),
-                                              MainFilename);
+  return MainFileBuffer;
 }
 
 void ASTUnit::RealizeTopLevelDeclsFromPreamble() {
+  assert(Preamble && "Should only be called when preamble was built");
+
   std::vector<Decl *> Resolved;
   Resolved.reserve(TopLevelDeclsInPreamble.size());
   ExternalASTSource &Source = *getASTContext().getExternalSource();
@@ -1995,8 +1634,8 @@ ASTUnit *ASTUnit::LoadFromCommandLine(
 
   {
 
-    CaptureDroppedDiagnostics Capture(CaptureDiagnostics, *Diags, 
-                                      StoredDiagnostics);
+    CaptureDroppedDiagnostics Capture(CaptureDiagnostics, *Diags,
+                                      &StoredDiagnostics, nullptr);
 
     CI = clang::createInvocationFromCommandLine(
         llvm::makeArrayRef(ArgBegin, ArgEnd), Diags);
@@ -2101,7 +1740,7 @@ bool ASTUnit::Reparse(std::shared_ptr<PCHContainerOperations> PCHContainerOps,
   // If we have a preamble file lying around, or if we might try to
   // build a precompiled preamble, do so now.
   std::unique_ptr<llvm::MemoryBuffer> OverrideMainBuffer;
-  if (!getPreambleFile(this).empty() || PreambleRebuildCounter > 0)
+  if (Preamble || PreambleRebuildCounter > 0)
     OverrideMainBuffer =
         getMainBufferWithPrecompiledPreamble(PCHContainerOps, *Invocation, VFS);
 
@@ -2435,7 +2074,7 @@ void ASTUnit::CodeComplete(
   Clang->setDiagnostics(&Diag);
   CaptureDroppedDiagnostics Capture(true, 
                                     Clang->getDiagnostics(), 
-                                    StoredDiagnostics);
+                                    &StoredDiagnostics, nullptr);
   ProcessWarningOptions(Diag, Inv.getDiagnosticOpts());
 
   // Create the target instance.
@@ -2484,7 +2123,7 @@ void ASTUnit::CodeComplete(
   // point is within the main file, after the end of the precompiled
   // preamble.
   std::unique_ptr<llvm::MemoryBuffer> OverrideMainBuffer;
-  if (!getPreambleFile(this).empty()) {
+  if (Preamble) {
     std::string CompleteFilePath(File);
 
     auto VFS = FileMgr.getVirtualFileSystem();
@@ -2506,14 +2145,8 @@ void ASTUnit::CodeComplete(
   // If the main file has been overridden due to the use of a preamble,
   // make that override happen and introduce the preamble.
   if (OverrideMainBuffer) {
-    PreprocessorOpts.addRemappedFile(OriginalSourceFile,
-                                     OverrideMainBuffer.get());
-    PreprocessorOpts.PrecompiledPreambleBytes.first = Preamble.size();
-    PreprocessorOpts.PrecompiledPreambleBytes.second
-                                                    = PreambleEndsAtStartOfLine;
-    PreprocessorOpts.ImplicitPCHInclude = getPreambleFile(this);
-    PreprocessorOpts.DisablePCHValidation = true;
-
+    assert(Preamble && "No preamble was built, but OverrideMainBuffer is not null");
+    Preamble->AddImplicitPreamble(Clang->getInvocation(), OverrideMainBuffer.get());
     OwnedBuffers.push_back(OverrideMainBuffer.release());
   } else {
     PreprocessorOpts.PrecompiledPreambleBytes.first = 0;
@@ -2760,11 +2393,11 @@ SourceLocation ASTUnit::mapLocationFromPreamble(SourceLocation Loc) {
   if (SourceMgr)
     PreambleID = SourceMgr->getPreambleFileID();
 
-  if (Loc.isInvalid() || Preamble.empty() || PreambleID.isInvalid())
+  if (Loc.isInvalid() || !Preamble || PreambleID.isInvalid())
     return Loc;
 
   unsigned Offs;
-  if (SourceMgr->isInFileID(Loc, PreambleID, &Offs) && Offs < Preamble.size()) {
+  if (SourceMgr->isInFileID(Loc, PreambleID, &Offs) && Offs < Preamble->getBounds().Size) {
     SourceLocation FileLoc
         = SourceMgr->getLocForStartOfFile(SourceMgr->getMainFileID());
     return FileLoc.getLocWithOffset(Offs);
@@ -2781,12 +2414,12 @@ SourceLocation ASTUnit::mapLocationToPreamble(SourceLocation Loc) {
   if (SourceMgr)
     PreambleID = SourceMgr->getPreambleFileID();
 
-  if (Loc.isInvalid() || Preamble.empty() || PreambleID.isInvalid())
+  if (Loc.isInvalid() || !Preamble || PreambleID.isInvalid())
     return Loc;
 
   unsigned Offs;
   if (SourceMgr->isInFileID(Loc, SourceMgr->getMainFileID(), &Offs) &&
-      Offs < Preamble.size()) {
+      Offs < Preamble->getBounds().Size) {
     SourceLocation FileLoc = SourceMgr->getLocForStartOfFile(PreambleID);
     return FileLoc.getLocWithOffset(Offs);
   }
@@ -2932,17 +2565,6 @@ InputKind ASTUnit::getInputKind() const {
   return InputKind(Lang, Fmt, PP);
 }
 
-void ASTUnit::PreambleData::countLines() const {
-  NumLines = 0;
-  if (empty())
-    return;
-
-  NumLines = std::count(Buffer.begin(), Buffer.end(), '\n');
-
-  if (Buffer.back() != '\n')
-    ++NumLines;
-}
-
 #ifndef NDEBUG
 ASTUnit::ConcurrencyState::ConcurrencyState() {
   Mutex = new llvm::sys::MutexImpl(/*recursive=*/true);
diff --git a/lib/Frontend/CMakeLists.txt b/lib/Frontend/CMakeLists.txt
index 18abecd2071c..ba3bd7d28c70 100644
--- a/lib/Frontend/CMakeLists.txt
+++ b/lib/Frontend/CMakeLists.txt
@@ -38,6 +38,7 @@ add_clang_library(clangFrontend
   ModuleDependencyCollector.cpp
   MultiplexConsumer.cpp
   PCHContainerOperations.cpp
+  PrecompiledPreamble.cpp
   PrintPreprocessedOutput.cpp
   SerializedDiagnosticPrinter.cpp
   SerializedDiagnosticReader.cpp
diff --git a/lib/Frontend/CompilerInstance.cpp b/lib/Frontend/CompilerInstance.cpp
index 72a8c3818093..e5da2ae4f22e 100644
--- a/lib/Frontend/CompilerInstance.cpp
+++ b/lib/Frontend/CompilerInstance.cpp
@@ -11,6 +11,7 @@
 #include "clang/AST/ASTConsumer.h"
 #include "clang/AST/ASTContext.h"
 #include "clang/AST/Decl.h"
+#include "clang/Basic/CharInfo.h"
 #include "clang/Basic/Diagnostic.h"
 #include "clang/Basic/FileManager.h"
 #include "clang/Basic/MemoryBufferCache.h"
@@ -1902,17 +1903,23 @@ CompilerInstance::loadModule(SourceLocation ImportLoc,
 void CompilerInstance::loadModuleFromSource(SourceLocation ImportLoc,
                                             StringRef ModuleName,
                                             StringRef Source) {
+  // Avoid creating filenames with special characters.
+  SmallString<128> CleanModuleName(ModuleName);
+  for (auto &C : CleanModuleName)
+    if (!isAlphanumeric(C))
+      C = '_';
+
   // FIXME: Using a randomized filename here means that our intermediate .pcm
   // output is nondeterministic (as .pcm files refer to each other by name).
   // Can this affect the output in any way?
   SmallString<128> ModuleFileName;
   if (std::error_code EC = llvm::sys::fs::createTemporaryFile(
-          ModuleName, "pcm", ModuleFileName)) {
+          CleanModuleName, "pcm", ModuleFileName)) {
     getDiagnostics().Report(ImportLoc, diag::err_fe_unable_to_open_output)
         << ModuleFileName << EC.message();
     return;
   }
-  std::string ModuleMapFileName = (ModuleName + ".map").str();
+  std::string ModuleMapFileName = (CleanModuleName + ".map").str();
 
   FrontendInputFile Input(
       ModuleMapFileName,
diff --git a/lib/Frontend/CompilerInvocation.cpp b/lib/Frontend/CompilerInvocation.cpp
index 1667af2d12bb..6254b0013bab 100644
--- a/lib/Frontend/CompilerInvocation.cpp
+++ b/lib/Frontend/CompilerInvocation.cpp
@@ -745,9 +745,22 @@ static bool ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args, InputKind IK,
   Opts.InstrumentForProfiling = Args.hasArg(OPT_pg);
   Opts.CallFEntry = Args.hasArg(OPT_mfentry);
   Opts.EmitOpenCLArgMetadata = Args.hasArg(OPT_cl_kernel_arg_info);
-  // TODO: map this from -gz in the driver and give it a named value
-  if (Args.hasArg(OPT_compress_debug_sections))
-    Opts.setCompressDebugSections(llvm::DebugCompressionType::GNU);
+
+  if (const Arg *A = Args.getLastArg(OPT_compress_debug_sections,
+                                     OPT_compress_debug_sections_EQ)) {
+    if (A->getOption().getID() == OPT_compress_debug_sections) {
+      // TODO: be more clever about the compression type auto-detection
+      Opts.setCompressDebugSections(llvm::DebugCompressionType::GNU);
+    } else {
+      auto DCT = llvm::StringSwitch<llvm::DebugCompressionType>(A->getValue())
+                     .Case("none", llvm::DebugCompressionType::None)
+                     .Case("zlib", llvm::DebugCompressionType::Z)
+                     .Case("zlib-gnu", llvm::DebugCompressionType::GNU)
+                     .Default(llvm::DebugCompressionType::None);
+      Opts.setCompressDebugSections(DCT);
+    }
+  }
+
   Opts.RelaxELFRelocations = Args.hasArg(OPT_mrelax_relocations);
   Opts.DebugCompilationDir = Args.getLastArgValue(OPT_fdebug_compilation_dir);
   for (auto A : Args.filtered(OPT_mlink_bitcode_file, OPT_mlink_cuda_bitcode)) {
@@ -892,14 +905,18 @@ static bool ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args, InputKind IK,
 
   Opts.DiagnosticsWithHotness =
       Args.hasArg(options::OPT_fdiagnostics_show_hotness);
+  bool UsingSampleProfile = !Opts.SampleProfileFile.empty();
+
   if (Opts.DiagnosticsWithHotness &&
-      Opts.getProfileUse() == CodeGenOptions::ProfileNone)
+      Opts.getProfileUse() == CodeGenOptions::ProfileNone &&
+      !UsingSampleProfile) {
     Diags.Report(diag::warn_drv_fdiagnostics_show_hotness_requires_pgo);
+  }
 
   // If the user requested to use a sample profile for PGO, then the
   // backend will need to track source location information so the profile
   // can be incorporated into the IR.
-  if (!Opts.SampleProfileFile.empty())
+  if (UsingSampleProfile)
     NeedLocTracking = true;
 
   // If the user requested a flag that requires source locations available in
@@ -2379,9 +2396,51 @@ static void ParseLangArgs(LangOptions &Opts, ArgList &Args, InputKind IK,
   Opts.AllowEditorPlaceholders = Args.hasArg(OPT_fallow_editor_placeholders);
 }
 
+static bool isStrictlyPreprocessorAction(frontend::ActionKind Action) {
+  switch (Action) {
+  case frontend::ASTDeclList:
+  case frontend::ASTDump:
+  case frontend::ASTPrint:
+  case frontend::ASTView:
+  case frontend::EmitAssembly:
+  case frontend::EmitBC:
+  case frontend::EmitHTML:
+  case frontend::EmitLLVM:
+  case frontend::EmitLLVMOnly:
+  case frontend::EmitCodeGenOnly:
+  case frontend::EmitObj:
+  case frontend::FixIt:
+  case frontend::GenerateModule:
+  case frontend::GenerateModuleInterface:
+  case frontend::GeneratePCH:
+  case frontend::GeneratePTH:
+  case frontend::ParseSyntaxOnly:
+  case frontend::ModuleFileInfo:
+  case frontend::VerifyPCH:
+  case frontend::PluginAction:
+  case frontend::PrintDeclContext:
+  case frontend::RewriteObjC:
+  case frontend::RewriteTest:
+  case frontend::RunAnalysis:
+  case frontend::MigrateSource:
+    return false;
+
+  case frontend::DumpRawTokens:
+  case frontend::DumpTokens:
+  case frontend::InitOnly:
+  case frontend::PrintPreamble:
+  case frontend::PrintPreprocessedInput:
+  case frontend::RewriteMacros:
+  case frontend::RunPreprocessorOnly:
+    return true;
+  }
+  llvm_unreachable("invalid frontend action");
+}
+
 static void ParsePreprocessorArgs(PreprocessorOptions &Opts, ArgList &Args,
                                   FileManager &FileMgr,
-                                  DiagnosticsEngine &Diags) {
+                                  DiagnosticsEngine &Diags,
+                                  frontend::ActionKind Action) {
   using namespace options;
   Opts.ImplicitPCHInclude = Args.getLastArgValue(OPT_include_pch);
   Opts.ImplicitPTHInclude = Args.getLastArgValue(OPT_include_pth);
@@ -2454,6 +2513,12 @@ static void ParsePreprocessorArgs(PreprocessorOptions &Opts, ArgList &Args,
     else
       Opts.ObjCXXARCStandardLibrary = (ObjCXXARCStandardLibraryKind)Library;
   }
+
+  // Always avoid lexing editor placeholders when we're just running the
+  // preprocessor as we never want to emit the
+  // "editor placeholder in source file" error in PP only mode.
+  if (isStrictlyPreprocessorAction(Action))
+    Opts.LexEditorPlaceholders = false;
 }
 
 static void ParsePreprocessorOutputArgs(PreprocessorOutputOptions &Opts,
@@ -2461,45 +2526,10 @@ static void ParsePreprocessorOutputArgs(PreprocessorOutputOptions &Opts,
                                         frontend::ActionKind Action) {
   using namespace options;
 
-  switch (Action) {
-  case frontend::ASTDeclList:
-  case frontend::ASTDump:
-  case frontend::ASTPrint:
-  case frontend::ASTView:
-  case frontend::EmitAssembly:
-  case frontend::EmitBC:
-  case frontend::EmitHTML:
-  case frontend::EmitLLVM:
-  case frontend::EmitLLVMOnly:
-  case frontend::EmitCodeGenOnly:
-  case frontend::EmitObj:
-  case frontend::FixIt:
-  case frontend::GenerateModule:
-  case frontend::GenerateModuleInterface:
-  case frontend::GeneratePCH:
-  case frontend::GeneratePTH:
-  case frontend::ParseSyntaxOnly:
-  case frontend::ModuleFileInfo:
-  case frontend::VerifyPCH:
-  case frontend::PluginAction:
-  case frontend::PrintDeclContext:
-  case frontend::RewriteObjC:
-  case frontend::RewriteTest:
-  case frontend::RunAnalysis:
-  case frontend::MigrateSource:
-    Opts.ShowCPP = 0;
-    break;
-
-  case frontend::DumpRawTokens:
-  case frontend::DumpTokens:
-  case frontend::InitOnly:
-  case frontend::PrintPreamble:
-  case frontend::PrintPreprocessedInput:
-  case frontend::RewriteMacros:
-  case frontend::RunPreprocessorOnly:
+  if (isStrictlyPreprocessorAction(Action))
     Opts.ShowCPP = !Args.hasArg(OPT_dM);
-    break;
-  }
+  else
+    Opts.ShowCPP = 0;
 
   Opts.ShowComments = Args.hasArg(OPT_C);
   Opts.ShowLineMarkers = !Args.hasArg(OPT_P);
@@ -2626,7 +2656,8 @@ bool CompilerInvocation::CreateFromArgs(CompilerInvocation &Res,
   // ParsePreprocessorArgs and remove the FileManager
   // parameters from the function and the "FileManager.h" #include.
   FileManager FileMgr(Res.getFileSystemOpts());
-  ParsePreprocessorArgs(Res.getPreprocessorOpts(), Args, FileMgr, Diags);
+  ParsePreprocessorArgs(Res.getPreprocessorOpts(), Args, FileMgr, Diags,
+                        Res.getFrontendOpts().ProgramAction);
   ParsePreprocessorOutputArgs(Res.getPreprocessorOutputOpts(), Args,
                               Res.getFrontendOpts().ProgramAction);
 
diff --git a/lib/Frontend/FrontendAction.cpp b/lib/Frontend/FrontendAction.cpp
index 7c0b854648bd..f81a06b31869 100644
--- a/lib/Frontend/FrontendAction.cpp
+++ b/lib/Frontend/FrontendAction.cpp
@@ -561,7 +561,6 @@ bool FrontendAction::BeginSourceFile(CompilerInstance &CI,
     CI.setFileManager(&AST->getFileManager());
     CI.createSourceManager(CI.getFileManager());
     CI.getSourceManager().initializeForReplay(AST->getSourceManager());
-    CI.createPreprocessor(getTranslationUnitKind());
 
     // Set up the input file for replay purposes.
     auto Kind = AST->getInputKind();
diff --git a/lib/Frontend/FrontendActions.cpp b/lib/Frontend/FrontendActions.cpp
index c5567a09636b..9621889b27ad 100644
--- a/lib/Frontend/FrontendActions.cpp
+++ b/lib/Frontend/FrontendActions.cpp
@@ -517,7 +517,7 @@ void PrintPreprocessedAction::ExecuteAction() {
     // file.  This is mostly a sanity check in case the file has no 
     // newlines whatsoever.
     if (end - cur > 256) end = cur + 256;
-	  
+
     while (next < end) {
       if (*cur == 0x0D) {  // CR
         if (*next == 0x0A)  // CRLF
diff --git a/lib/Frontend/PrecompiledPreamble.cpp b/lib/Frontend/PrecompiledPreamble.cpp
new file mode 100644
index 000000000000..15b24cbed484
--- /dev/null
+++ b/lib/Frontend/PrecompiledPreamble.cpp
@@ -0,0 +1,563 @@
+//===--- PrecompiledPreamble.cpp - Build precompiled preambles --*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Helper class to build precompiled preamble.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/Frontend/PrecompiledPreamble.h"
+#include "clang/AST/DeclObjC.h"
+#include "clang/Basic/TargetInfo.h"
+#include "clang/Basic/VirtualFileSystem.h"
+#include "clang/Frontend/CompilerInstance.h"
+#include "clang/Frontend/CompilerInvocation.h"
+#include "clang/Frontend/FrontendActions.h"
+#include "clang/Frontend/FrontendOptions.h"
+#include "clang/Lex/Lexer.h"
+#include "clang/Lex/PreprocessorOptions.h"
+#include "clang/Serialization/ASTWriter.h"
+#include "llvm/ADT/StringExtras.h"
+#include "llvm/ADT/StringSet.h"
+#include "llvm/Support/CrashRecoveryContext.h"
+#include "llvm/Support/FileSystem.h"
+#include "llvm/Support/Mutex.h"
+#include "llvm/Support/MutexGuard.h"
+
+using namespace clang;
+
+namespace {
+
+/// Keeps a track of files to be deleted in destructor.
+class TemporaryFiles {
+public:
+  // A static instance to be used by all clients.
+  static TemporaryFiles &getInstance();
+
+private:
+  // Disallow constructing the class directly.
+  TemporaryFiles() = default;
+  // Disallow copy.
+  TemporaryFiles(const TemporaryFiles &) = delete;
+
+public:
+  ~TemporaryFiles();
+
+  /// Adds \p File to a set of tracked files.
+  void addFile(StringRef File);
+
+  /// Remove \p File from disk and from the set of tracked files.
+  void removeFile(StringRef File);
+
+private:
+  llvm::sys::SmartMutex<false> Mutex;
+  llvm::StringSet<> Files;
+};
+
+TemporaryFiles &TemporaryFiles::getInstance() {
+  static TemporaryFiles Instance;
+  return Instance;
+}
+
+TemporaryFiles::~TemporaryFiles() {
+  llvm::MutexGuard Guard(Mutex);
+  for (const auto &File : Files)
+    llvm::sys::fs::remove(File.getKey());
+}
+
+void TemporaryFiles::addFile(StringRef File) {
+  llvm::MutexGuard Guard(Mutex);
+  auto IsInserted = Files.insert(File).second;
+  (void)IsInserted;
+  assert(IsInserted && "File has already been added");
+}
+
+void TemporaryFiles::removeFile(StringRef File) {
+  llvm::MutexGuard Guard(Mutex);
+  auto WasPresent = Files.erase(File);
+  (void)WasPresent;
+  assert(WasPresent && "File was not tracked");
+  llvm::sys::fs::remove(File);
+}
+
+class PreambleMacroCallbacks : public PPCallbacks {
+public:
+  PreambleMacroCallbacks(PreambleCallbacks &Callbacks) : Callbacks(Callbacks) {}
+
+  void MacroDefined(const Token &MacroNameTok,
+                    const MacroDirective *MD) override {
+    Callbacks.HandleMacroDefined(MacroNameTok, MD);
+  }
+
+private:
+  PreambleCallbacks &Callbacks;
+};
+
+class PrecompilePreambleAction : public ASTFrontendAction {
+public:
+  PrecompilePreambleAction(PreambleCallbacks &Callbacks)
+      : Callbacks(Callbacks) {}
+
+  std::unique_ptr<ASTConsumer> CreateASTConsumer(CompilerInstance &CI,
+                                                 StringRef InFile) override;
+
+  bool hasEmittedPreamblePCH() const { return HasEmittedPreamblePCH; }
+
+  void setEmittedPreamblePCH(ASTWriter &Writer) {
+    this->HasEmittedPreamblePCH = true;
+    Callbacks.AfterPCHEmitted(Writer);
+  }
+
+  bool shouldEraseOutputFiles() override { return !hasEmittedPreamblePCH(); }
+  bool hasCodeCompletionSupport() const override { return false; }
+  bool hasASTFileSupport() const override { return false; }
+  TranslationUnitKind getTranslationUnitKind() override { return TU_Prefix; }
+
+private:
+  friend class PrecompilePreambleConsumer;
+
+  bool HasEmittedPreamblePCH = false;
+  PreambleCallbacks &Callbacks;
+};
+
+class PrecompilePreambleConsumer : public PCHGenerator {
+public:
+  PrecompilePreambleConsumer(PrecompilePreambleAction &Action,
+                             const Preprocessor &PP, StringRef isysroot,
+                             std::unique_ptr<raw_ostream> Out)
+      : PCHGenerator(PP, "", isysroot, std::make_shared<PCHBuffer>(),
+                     ArrayRef<std::shared_ptr<ModuleFileExtension>>(),
+                     /*AllowASTWithErrors=*/true),
+        Action(Action), Out(std::move(Out)) {}
+
+  bool HandleTopLevelDecl(DeclGroupRef DG) override {
+    Action.Callbacks.HandleTopLevelDecl(DG);
+    return true;
+  }
+
+  void HandleTranslationUnit(ASTContext &Ctx) override {
+    PCHGenerator::HandleTranslationUnit(Ctx);
+    if (!hasEmittedPCH())
+      return;
+
+    // Write the generated bitstream to "Out".
+    *Out << getPCH();
+    // Make sure it hits disk now.
+    Out->flush();
+    // Free the buffer.
+    llvm::SmallVector<char, 0> Empty;
+    getPCH() = std::move(Empty);
+
+    Action.setEmittedPreamblePCH(getWriter());
+  }
+
+private:
+  PrecompilePreambleAction &Action;
+  std::unique_ptr<raw_ostream> Out;
+};
+
+std::unique_ptr<ASTConsumer>
+PrecompilePreambleAction::CreateASTConsumer(CompilerInstance &CI,
+
+                                            StringRef InFile) {
+  std::string Sysroot;
+  std::string OutputFile;
+  std::unique_ptr<raw_ostream> OS =
+      GeneratePCHAction::ComputeASTConsumerArguments(CI, InFile, Sysroot,
+                                                     OutputFile);
+  if (!OS)
+    return nullptr;
+
+  if (!CI.getFrontendOpts().RelocatablePCH)
+    Sysroot.clear();
+
+  CI.getPreprocessor().addPPCallbacks(
+      llvm::make_unique<PreambleMacroCallbacks>(Callbacks));
+  return llvm::make_unique<PrecompilePreambleConsumer>(
+      *this, CI.getPreprocessor(), Sysroot, std::move(OS));
+}
+
+template <class T> bool moveOnNoError(llvm::ErrorOr<T> Val, T &Output) {
+  if (!Val)
+    return false;
+  Output = std::move(*Val);
+  return true;
+}
+
+} // namespace
+
+PreambleBounds clang::ComputePreambleBounds(const LangOptions &LangOpts,
+                                            llvm::MemoryBuffer *Buffer,
+                                            unsigned MaxLines) {
+  auto Pre = Lexer::ComputePreamble(Buffer->getBuffer(), LangOpts, MaxLines);
+  return PreambleBounds(Pre.first, Pre.second);
+}
+
+llvm::ErrorOr<PrecompiledPreamble> PrecompiledPreamble::Build(
+    const CompilerInvocation &Invocation,
+    const llvm::MemoryBuffer *MainFileBuffer, PreambleBounds Bounds,
+    DiagnosticsEngine &Diagnostics, IntrusiveRefCntPtr<vfs::FileSystem> VFS,
+    std::shared_ptr<PCHContainerOperations> PCHContainerOps,
+    PreambleCallbacks &Callbacks) {
+  assert(VFS && "VFS is null");
+
+  if (!Bounds.Size)
+    return BuildPreambleError::PreambleIsEmpty;
+
+  auto PreambleInvocation = std::make_shared<CompilerInvocation>(Invocation);
+  FrontendOptions &FrontendOpts = PreambleInvocation->getFrontendOpts();
+  PreprocessorOptions &PreprocessorOpts =
+      PreambleInvocation->getPreprocessorOpts();
+
+  // Create a temporary file for the precompiled preamble. In rare
+  // circumstances, this can fail.
+  llvm::ErrorOr<PrecompiledPreamble::TempPCHFile> PreamblePCHFile =
+      PrecompiledPreamble::TempPCHFile::CreateNewPreamblePCHFile();
+  if (!PreamblePCHFile)
+    return BuildPreambleError::CouldntCreateTempFile;
+
+  // Save the preamble text for later; we'll need to compare against it for
+  // subsequent reparses.
+  std::vector<char> PreambleBytes(MainFileBuffer->getBufferStart(),
+                                  MainFileBuffer->getBufferStart() +
+                                      Bounds.Size);
+  bool PreambleEndsAtStartOfLine = Bounds.PreambleEndsAtStartOfLine;
+
+  // Tell the compiler invocation to generate a temporary precompiled header.
+  FrontendOpts.ProgramAction = frontend::GeneratePCH;
+  // FIXME: Generate the precompiled header into memory?
+  FrontendOpts.OutputFile = PreamblePCHFile->getFilePath();
+  PreprocessorOpts.PrecompiledPreambleBytes.first = 0;
+  PreprocessorOpts.PrecompiledPreambleBytes.second = false;
+
+  // Create the compiler instance to use for building the precompiled preamble.
+  std::unique_ptr<CompilerInstance> Clang(
+      new CompilerInstance(std::move(PCHContainerOps)));
+
+  // Recover resources if we crash before exiting this method.
+  llvm::CrashRecoveryContextCleanupRegistrar<CompilerInstance> CICleanup(
+      Clang.get());
+
+  Clang->setInvocation(std::move(PreambleInvocation));
+  Clang->setDiagnostics(&Diagnostics);
+
+  // Create the target instance.
+  Clang->setTarget(TargetInfo::CreateTargetInfo(
+      Clang->getDiagnostics(), Clang->getInvocation().TargetOpts));
+  if (!Clang->hasTarget())
+    return BuildPreambleError::CouldntCreateTargetInfo;
+
+  // Inform the target of the language options.
+  //
+  // FIXME: We shouldn't need to do this, the target should be immutable once
+  // created. This complexity should be lifted elsewhere.
+  Clang->getTarget().adjust(Clang->getLangOpts());
+
+  assert(Clang->getFrontendOpts().Inputs.size() == 1 &&
+         "Invocation must have exactly one source file!");
+  assert(Clang->getFrontendOpts().Inputs[0].getKind().getFormat() ==
+             InputKind::Source &&
+         "FIXME: AST inputs not yet supported here!");
+  assert(Clang->getFrontendOpts().Inputs[0].getKind().getLanguage() !=
+             InputKind::LLVM_IR &&
+         "IR inputs not support here!");
+
+  // Clear out old caches and data.
+  Diagnostics.Reset();
+  ProcessWarningOptions(Diagnostics, Clang->getDiagnosticOpts());
+
+  VFS =
+      createVFSFromCompilerInvocation(Clang->getInvocation(), Diagnostics, VFS);
+  if (!VFS)
+    return BuildPreambleError::CouldntCreateVFSOverlay;
+
+  // Create a file manager object to provide access to and cache the filesystem.
+  Clang->setFileManager(new FileManager(Clang->getFileSystemOpts(), VFS));
+
+  // Create the source manager.
+  Clang->setSourceManager(
+      new SourceManager(Diagnostics, Clang->getFileManager()));
+
+  auto PreambleDepCollector = std::make_shared<DependencyCollector>();
+  Clang->addDependencyCollector(PreambleDepCollector);
+
+  // Remap the main source file to the preamble buffer.
+  StringRef MainFilePath = FrontendOpts.Inputs[0].getFile();
+  auto PreambleInputBuffer = llvm::MemoryBuffer::getMemBufferCopy(
+      MainFileBuffer->getBuffer().slice(0, Bounds.Size), MainFilePath);
+  if (PreprocessorOpts.RetainRemappedFileBuffers) {
+    // MainFileBuffer will be deleted by unique_ptr after leaving the method.
+    PreprocessorOpts.addRemappedFile(MainFilePath, PreambleInputBuffer.get());
+  } else {
+    // In that case, remapped buffer will be deleted by CompilerInstance on
+    // BeginSourceFile, so we call release() to avoid double deletion.
+    PreprocessorOpts.addRemappedFile(MainFilePath,
+                                     PreambleInputBuffer.release());
+  }
+
+  std::unique_ptr<PrecompilePreambleAction> Act;
+  Act.reset(new PrecompilePreambleAction(Callbacks));
+  if (!Act->BeginSourceFile(*Clang.get(), Clang->getFrontendOpts().Inputs[0]))
+    return BuildPreambleError::BeginSourceFileFailed;
+
+  Act->Execute();
+
+  // Run the callbacks.
+  Callbacks.AfterExecute(*Clang);
+
+  Act->EndSourceFile();
+
+  if (!Act->hasEmittedPreamblePCH())
+    return BuildPreambleError::CouldntEmitPCH;
+
+  // Keep track of all of the files that the source manager knows about,
+  // so we can verify whether they have changed or not.
+  llvm::StringMap<PrecompiledPreamble::PreambleFileHash> FilesInPreamble;
+
+  SourceManager &SourceMgr = Clang->getSourceManager();
+  for (auto &Filename : PreambleDepCollector->getDependencies()) {
+    const FileEntry *File = Clang->getFileManager().getFile(Filename);
+    if (!File || File == SourceMgr.getFileEntryForID(SourceMgr.getMainFileID()))
+      continue;
+    if (time_t ModTime = File->getModificationTime()) {
+      FilesInPreamble[File->getName()] =
+          PrecompiledPreamble::PreambleFileHash::createForFile(File->getSize(),
+                                                               ModTime);
+    } else {
+      llvm::MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File);
+      FilesInPreamble[File->getName()] =
+          PrecompiledPreamble::PreambleFileHash::createForMemoryBuffer(Buffer);
+    }
+  }
+
+  return PrecompiledPreamble(
+      std::move(*PreamblePCHFile), std::move(PreambleBytes),
+      PreambleEndsAtStartOfLine, std::move(FilesInPreamble));
+}
+
+PreambleBounds PrecompiledPreamble::getBounds() const {
+  return PreambleBounds(PreambleBytes.size(), PreambleEndsAtStartOfLine);
+}
+
+bool PrecompiledPreamble::CanReuse(const CompilerInvocation &Invocation,
+                                   const llvm::MemoryBuffer *MainFileBuffer,
+                                   PreambleBounds Bounds,
+                                   vfs::FileSystem *VFS) const {
+
+  assert(
+      Bounds.Size <= MainFileBuffer->getBufferSize() &&
+      "Buffer is too large. Bounds were calculated from a different buffer?");
+
+  auto PreambleInvocation = std::make_shared<CompilerInvocation>(Invocation);
+  PreprocessorOptions &PreprocessorOpts =
+      PreambleInvocation->getPreprocessorOpts();
+
+  if (!Bounds.Size)
+    return false;
+
+  // We've previously computed a preamble. Check whether we have the same
+  // preamble now that we did before, and that there's enough space in
+  // the main-file buffer within the precompiled preamble to fit the
+  // new main file.
+  if (PreambleBytes.size() != Bounds.Size ||
+      PreambleEndsAtStartOfLine != Bounds.PreambleEndsAtStartOfLine ||
+      memcmp(PreambleBytes.data(), MainFileBuffer->getBufferStart(),
+             Bounds.Size) != 0)
+    return false;
+  // The preamble has not changed. We may be able to re-use the precompiled
+  // preamble.
+
+  // Check that none of the files used by the preamble have changed.
+  // First, make a record of those files that have been overridden via
+  // remapping or unsaved_files.
+  std::map<llvm::sys::fs::UniqueID, PreambleFileHash> OverriddenFiles;
+  for (const auto &R : PreprocessorOpts.RemappedFiles) {
+    vfs::Status Status;
+    if (!moveOnNoError(VFS->status(R.second), Status)) {
+      // If we can't stat the file we're remapping to, assume that something
+      // horrible happened.
+      return false;
+    }
+
+    OverriddenFiles[Status.getUniqueID()] = PreambleFileHash::createForFile(
+        Status.getSize(), llvm::sys::toTimeT(Status.getLastModificationTime()));
+  }
+
+  for (const auto &RB : PreprocessorOpts.RemappedFileBuffers) {
+    vfs::Status Status;
+    if (!moveOnNoError(VFS->status(RB.first), Status))
+      return false;
+
+    OverriddenFiles[Status.getUniqueID()] =
+        PreambleFileHash::createForMemoryBuffer(RB.second);
+  }
+
+  // Check whether anything has changed.
+  for (const auto &F : FilesInPreamble) {
+    vfs::Status Status;
+    if (!moveOnNoError(VFS->status(F.first()), Status)) {
+      // If we can't stat the file, assume that something horrible happened.
+      return false;
+    }
+
+    std::map<llvm::sys::fs::UniqueID, PreambleFileHash>::iterator Overridden =
+        OverriddenFiles.find(Status.getUniqueID());
+    if (Overridden != OverriddenFiles.end()) {
+      // This file was remapped; check whether the newly-mapped file
+      // matches up with the previous mapping.
+      if (Overridden->second != F.second)
+        return false;
+      continue;
+    }
+
+    // The file was not remapped; check whether it has changed on disk.
+    if (Status.getSize() != uint64_t(F.second.Size) ||
+        llvm::sys::toTimeT(Status.getLastModificationTime()) !=
+            F.second.ModTime)
+      return false;
+  }
+  return true;
+}
+
+void PrecompiledPreamble::AddImplicitPreamble(
+    CompilerInvocation &CI, llvm::MemoryBuffer *MainFileBuffer) const {
+  auto &PreprocessorOpts = CI.getPreprocessorOpts();
+
+  // Configure ImpicitPCHInclude.
+  PreprocessorOpts.PrecompiledPreambleBytes.first = PreambleBytes.size();
+  PreprocessorOpts.PrecompiledPreambleBytes.second = PreambleEndsAtStartOfLine;
+  PreprocessorOpts.ImplicitPCHInclude = PCHFile.getFilePath();
+  PreprocessorOpts.DisablePCHValidation = true;
+
+  // Remap main file to point to MainFileBuffer.
+  auto MainFilePath = CI.getFrontendOpts().Inputs[0].getFile();
+  PreprocessorOpts.addRemappedFile(MainFilePath, MainFileBuffer);
+}
+
+PrecompiledPreamble::PrecompiledPreamble(
+    TempPCHFile PCHFile, std::vector<char> PreambleBytes,
+    bool PreambleEndsAtStartOfLine,
+    llvm::StringMap<PreambleFileHash> FilesInPreamble)
+    : PCHFile(std::move(PCHFile)), FilesInPreamble(FilesInPreamble),
+      PreambleBytes(std::move(PreambleBytes)),
+      PreambleEndsAtStartOfLine(PreambleEndsAtStartOfLine) {}
+
+llvm::ErrorOr<PrecompiledPreamble::TempPCHFile>
+PrecompiledPreamble::TempPCHFile::CreateNewPreamblePCHFile() {
+  // FIXME: This is a hack so that we can override the preamble file during
+  // crash-recovery testing, which is the only case where the preamble files
+  // are not necessarily cleaned up.
+  const char *TmpFile = ::getenv("CINDEXTEST_PREAMBLE_FILE");
+  if (TmpFile)
+    return TempPCHFile::createFromCustomPath(TmpFile);
+  return TempPCHFile::createInSystemTempDir("preamble", "pch");
+}
+
+llvm::ErrorOr<PrecompiledPreamble::TempPCHFile>
+PrecompiledPreamble::TempPCHFile::createInSystemTempDir(const Twine &Prefix,
+                                                        StringRef Suffix) {
+  llvm::SmallString<64> File;
+  auto EC = llvm::sys::fs::createTemporaryFile(Prefix, Suffix, /*ref*/ File);
+  if (EC)
+    return EC;
+  return TempPCHFile(std::move(File).str());
+}
+
+llvm::ErrorOr<PrecompiledPreamble::TempPCHFile>
+PrecompiledPreamble::TempPCHFile::createFromCustomPath(const Twine &Path) {
+  return TempPCHFile(Path.str());
+}
+
+PrecompiledPreamble::TempPCHFile::TempPCHFile(std::string FilePath)
+    : FilePath(std::move(FilePath)) {
+  TemporaryFiles::getInstance().addFile(*this->FilePath);
+}
+
+PrecompiledPreamble::TempPCHFile::TempPCHFile(TempPCHFile &&Other) {
+  FilePath = std::move(Other.FilePath);
+  Other.FilePath = None;
+}
+
+PrecompiledPreamble::TempPCHFile &PrecompiledPreamble::TempPCHFile::
+operator=(TempPCHFile &&Other) {
+  RemoveFileIfPresent();
+
+  FilePath = std::move(Other.FilePath);
+  Other.FilePath = None;
+  return *this;
+}
+
+PrecompiledPreamble::TempPCHFile::~TempPCHFile() { RemoveFileIfPresent(); }
+
+void PrecompiledPreamble::TempPCHFile::RemoveFileIfPresent() {
+  if (FilePath) {
+    TemporaryFiles::getInstance().removeFile(*FilePath);
+    FilePath = None;
+  }
+}
+
+llvm::StringRef PrecompiledPreamble::TempPCHFile::getFilePath() const {
+  assert(FilePath && "TempPCHFile doesn't have a FilePath. Had it been moved?");
+  return *FilePath;
+}
+
+PrecompiledPreamble::PreambleFileHash
+PrecompiledPreamble::PreambleFileHash::createForFile(off_t Size,
+                                                     time_t ModTime) {
+  PreambleFileHash Result;
+  Result.Size = Size;
+  Result.ModTime = ModTime;
+  Result.MD5 = {};
+  return Result;
+}
+
+PrecompiledPreamble::PreambleFileHash
+PrecompiledPreamble::PreambleFileHash::createForMemoryBuffer(
+    const llvm::MemoryBuffer *Buffer) {
+  PreambleFileHash Result;
+  Result.Size = Buffer->getBufferSize();
+  Result.ModTime = 0;
+
+  llvm::MD5 MD5Ctx;
+  MD5Ctx.update(Buffer->getBuffer().data());
+  MD5Ctx.final(Result.MD5);
+
+  return Result;
+}
+
+void PreambleCallbacks::AfterExecute(CompilerInstance &CI) {}
+void PreambleCallbacks::AfterPCHEmitted(ASTWriter &Writer) {}
+void PreambleCallbacks::HandleTopLevelDecl(DeclGroupRef DG) {}
+void PreambleCallbacks::HandleMacroDefined(const Token &MacroNameTok,
+                                           const MacroDirective *MD) {}
+
+std::error_code clang::make_error_code(BuildPreambleError Error) {
+  return std::error_code(static_cast<int>(Error), BuildPreambleErrorCategory());
+}
+
+const char *BuildPreambleErrorCategory::name() const noexcept {
+  return "build-preamble.error";
+}
+
+std::string BuildPreambleErrorCategory::message(int condition) const {
+  switch (static_cast<BuildPreambleError>(condition)) {
+  case BuildPreambleError::PreambleIsEmpty:
+    return "Preamble is empty";
+  case BuildPreambleError::CouldntCreateTempFile:
+    return "Could not create temporary file for PCH";
+  case BuildPreambleError::CouldntCreateTargetInfo:
+    return "CreateTargetInfo() return null";
+  case BuildPreambleError::CouldntCreateVFSOverlay:
+    return "Could not create VFS Overlay";
+  case BuildPreambleError::BeginSourceFileFailed:
+    return "BeginSourceFile() return an error";
+  case BuildPreambleError::CouldntEmitPCH:
+    return "Could not emit PCH";
+  }
+  llvm_unreachable("unexpected BuildPreambleError");
+}
diff --git a/lib/Frontend/PrintPreprocessedOutput.cpp b/lib/Frontend/PrintPreprocessedOutput.cpp
index 832eaf2926f0..5336de1f7468 100644
--- a/lib/Frontend/PrintPreprocessedOutput.cpp
+++ b/lib/Frontend/PrintPreprocessedOutput.cpp
@@ -349,7 +349,7 @@ void PrintPPOutputPPCallbacks::InclusionDirective(SourceLocation HashLoc,
     case tok::pp_include_next:
       startNewLineIfNeeded();
       MoveToLine(HashLoc);
-      OS << "#pragma clang module import " << Imported->getFullModuleName()
+      OS << "#pragma clang module import " << Imported->getFullModuleName(true)
          << " /* clang -E: implicit import for "
          << "#" << PP.getSpelling(IncludeTok) << " "
          << (IsAngled ? '<' : '"') << FileName << (IsAngled ? '>' : '"')
@@ -378,14 +378,14 @@ void PrintPPOutputPPCallbacks::InclusionDirective(SourceLocation HashLoc,
 /// Handle entering the scope of a module during a module compilation.
 void PrintPPOutputPPCallbacks::BeginModule(const Module *M) {
   startNewLineIfNeeded();
-  OS << "#pragma clang module begin " << M->getFullModuleName();
+  OS << "#pragma clang module begin " << M->getFullModuleName(true);
   setEmittedDirectiveOnThisLine();
 }
 
 /// Handle leaving the scope of a module during a module compilation.
 void PrintPPOutputPPCallbacks::EndModule(const Module *M) {
   startNewLineIfNeeded();
-  OS << "#pragma clang module end /*" << M->getFullModuleName() << "*/";
+  OS << "#pragma clang module end /*" << M->getFullModuleName(true) << "*/";
   setEmittedDirectiveOnThisLine();
 }
 
diff --git a/lib/Frontend/Rewrite/FrontendActions.cpp b/lib/Frontend/Rewrite/FrontendActions.cpp
index 45feffbcb5b5..e93f737c47fd 100644
--- a/lib/Frontend/Rewrite/FrontendActions.cpp
+++ b/lib/Frontend/Rewrite/FrontendActions.cpp
@@ -9,6 +9,7 @@
 
 #include "clang/Rewrite/Frontend/FrontendActions.h"
 #include "clang/AST/ASTConsumer.h"
+#include "clang/Basic/CharInfo.h"
 #include "clang/Frontend/CompilerInstance.h"
 #include "clang/Frontend/FrontendActions.h"
 #include "clang/Frontend/FrontendDiagnostic.h"
@@ -224,7 +225,15 @@ public:
     auto OS = Out.lock();
     assert(OS && "loaded module file after finishing rewrite action?");
 
-    (*OS) << "#pragma clang module build " << MF->ModuleName << "\n";
+    (*OS) << "#pragma clang module build ";
+    if (isValidIdentifier(MF->ModuleName))
+      (*OS) << MF->ModuleName;
+    else {
+      (*OS) << '"';
+      OS->write_escaped(MF->ModuleName);
+      (*OS) << '"';
+    }
+    (*OS) << '\n';
 
     // Rewrite the contents of the module in a separate compiler instance.
     CompilerInstance Instance(CI.getPCHContainerOperations(),
@@ -234,9 +243,12 @@ public:
     Instance.createDiagnostics(
         new ForwardingDiagnosticConsumer(CI.getDiagnosticClient()),
         /*ShouldOwnClient=*/true);
+    Instance.getFrontendOpts().DisableFree = false;
     Instance.getFrontendOpts().Inputs.clear();
     Instance.getFrontendOpts().Inputs.emplace_back(
         Filename, InputKind(InputKind::Unknown, InputKind::Precompiled));
+    Instance.getFrontendOpts().ModuleFiles.clear();
+    Instance.getFrontendOpts().ModuleMapFiles.clear();
     // Don't recursively rewrite imports. We handle them all at the top level.
     Instance.getPreprocessorOutputOpts().RewriteImports = false;
 
diff --git a/lib/Frontend/Rewrite/InclusionRewriter.cpp b/lib/Frontend/Rewrite/InclusionRewriter.cpp
index 3564cebba8a8..e0477069b340 100644
--- a/lib/Frontend/Rewrite/InclusionRewriter.cpp
+++ b/lib/Frontend/Rewrite/InclusionRewriter.cpp
@@ -140,7 +140,7 @@ void InclusionRewriter::WriteLineInfo(StringRef Filename, int Line,
 }
 
 void InclusionRewriter::WriteImplicitModuleImport(const Module *Mod) {
-  OS << "#pragma clang module import " << Mod->getFullModuleName()
+  OS << "#pragma clang module import " << Mod->getFullModuleName(true)
      << " /* clang -frewrite-includes: implicit import */" << MainEOL;
 }
 
@@ -471,15 +471,15 @@ void InclusionRewriter::Process(FileID FileId,
             else if (const IncludedFile *Inc = FindIncludeAtLocation(Loc)) {
               const Module *Mod = FindEnteredModule(Loc);
               if (Mod)
-                OS << "#pragma clang module begin " << Mod->getFullModuleName()
-                   << "\n";
+                OS << "#pragma clang module begin "
+                   << Mod->getFullModuleName(true) << "\n";
 
               // Include and recursively process the file.
               Process(Inc->Id, Inc->FileType);
 
               if (Mod)
-                OS << "#pragma clang module end /*" << Mod->getFullModuleName()
-                   << "*/\n";
+                OS << "#pragma clang module end /*"
+                   << Mod->getFullModuleName(true) << "*/\n";
 
               // Add line marker to indicate we're returning from an included
               // file.
diff --git a/lib/Index/IndexDecl.cpp b/lib/Index/IndexDecl.cpp
index 2162c039c48b..d1127722c8ca 100644
--- a/lib/Index/IndexDecl.cpp
+++ b/lib/Index/IndexDecl.cpp
@@ -351,9 +351,11 @@ public:
         IndexCtx.indexTagDecl(D, Relations);
       } else {
         auto *Parent = dyn_cast<NamedDecl>(D->getDeclContext());
+        SmallVector<SymbolRelation, 1> Relations;
+        gatherTemplatePseudoOverrides(D, Relations);
         return IndexCtx.handleReference(D, D->getLocation(), Parent,
                                         D->getLexicalDeclContext(),
-                                        SymbolRoleSet());
+                                        SymbolRoleSet(), Relations);
       }
     }
     return true;
@@ -609,18 +611,16 @@ public:
                                            ClassTemplateSpecializationDecl *D) {
     // FIXME: Notify subsequent callbacks if info comes from implicit
     // instantiation.
-    if (D->isThisDeclarationADefinition()) {
-      llvm::PointerUnion<ClassTemplateDecl *,
-                         ClassTemplatePartialSpecializationDecl *>
-          Template = D->getSpecializedTemplateOrPartial();
-      const Decl *SpecializationOf =
-          Template.is<ClassTemplateDecl *>()
-              ? (Decl *)Template.get<ClassTemplateDecl *>()
-              : Template.get<ClassTemplatePartialSpecializationDecl *>();
-      IndexCtx.indexTagDecl(
-          D, SymbolRelation(SymbolRoleSet(SymbolRole::RelationSpecializationOf),
-                            SpecializationOf));
-    }
+    llvm::PointerUnion<ClassTemplateDecl *,
+                       ClassTemplatePartialSpecializationDecl *>
+        Template = D->getSpecializedTemplateOrPartial();
+    const Decl *SpecializationOf =
+        Template.is<ClassTemplateDecl *>()
+            ? (Decl *)Template.get<ClassTemplateDecl *>()
+            : Template.get<ClassTemplatePartialSpecializationDecl *>();
+    IndexCtx.indexTagDecl(
+        D, SymbolRelation(SymbolRoleSet(SymbolRole::RelationSpecializationOf),
+                          SpecializationOf));
     if (TypeSourceInfo *TSI = D->getTypeAsWritten())
       IndexCtx.indexTypeSourceInfo(TSI, /*Parent=*/nullptr,
                                    D->getLexicalDeclContext());
diff --git a/lib/Lex/Lexer.cpp b/lib/Lex/Lexer.cpp
index 447ff212f06e..012189aa6f9f 100644
--- a/lib/Lex/Lexer.cpp
+++ b/lib/Lex/Lexer.cpp
@@ -19,6 +19,7 @@
 #include "clang/Lex/LexDiagnostic.h"
 #include "clang/Lex/LiteralSupport.h"
 #include "clang/Lex/Preprocessor.h"
+#include "clang/Lex/PreprocessorOptions.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/Support/Compiler.h"
@@ -2750,7 +2751,7 @@ static const char *findPlaceholderEnd(const char *CurPtr,
 
 bool Lexer::lexEditorPlaceholder(Token &Result, const char *CurPtr) {
   assert(CurPtr[-1] == '<' && CurPtr[0] == '#' && "Not a placeholder!");
-  if (!PP || LexingRawMode)
+  if (!PP || !PP->getPreprocessorOpts().LexEditorPlaceholders || LexingRawMode)
     return false;
   const char *End = findPlaceholderEnd(CurPtr + 1, BufferEnd);
   if (!End)
diff --git a/lib/Lex/PPDirectives.cpp b/lib/Lex/PPDirectives.cpp
index 89c2ebd00a68..8c79e50176e1 100644
--- a/lib/Lex/PPDirectives.cpp
+++ b/lib/Lex/PPDirectives.cpp
@@ -538,7 +538,7 @@ void Preprocessor::SkipExcludedConditionalBlock(SourceLocation IfTokenLoc,
           assert(CurPPLexer->LexingRawMode && "We have to be skipping here!");
           CurPPLexer->LexingRawMode = false;
           IdentifierInfo *IfNDefMacro = nullptr;
-          const bool CondValue = EvaluateDirectiveExpression(IfNDefMacro);
+          const bool CondValue = EvaluateDirectiveExpression(IfNDefMacro).Conditional;
           CurPPLexer->LexingRawMode = true;
           if (Callbacks) {
             const SourceLocation CondEnd = CurPPLexer->getSourceLocation();
@@ -635,7 +635,7 @@ void Preprocessor::PTHSkipExcludedConditionalBlock() {
     // Evaluate the condition of the #elif.
     IdentifierInfo *IfNDefMacro = nullptr;
     CurPTHLexer->ParsingPreprocessorDirective = true;
-    bool ShouldEnter = EvaluateDirectiveExpression(IfNDefMacro);
+    bool ShouldEnter = EvaluateDirectiveExpression(IfNDefMacro).Conditional;
     CurPTHLexer->ParsingPreprocessorDirective = false;
 
     // If this condition is true, enter it!
@@ -2654,7 +2654,13 @@ void Preprocessor::HandleIfdefDirective(Token &Result, bool isIfndef,
   }
 
   // Should we include the stuff contained by this directive?
-  if (!MI == isIfndef) {
+  if (PPOpts->SingleFileParseMode && !MI) {
+    // In 'single-file-parse mode' undefined identifiers trigger parsing of all
+    // the directive blocks.
+    CurPPLexer->pushConditionalLevel(DirectiveTok.getLocation(),
+                                     /*wasskip*/false, /*foundnonskip*/false,
+                                     /*foundelse*/false);
+  } else if (!MI == isIfndef) {
     // Yes, remember that we are inside a conditional, then lex the next token.
     CurPPLexer->pushConditionalLevel(DirectiveTok.getLocation(),
                                      /*wasskip*/false, /*foundnonskip*/true,
@@ -2676,7 +2682,8 @@ void Preprocessor::HandleIfDirective(Token &IfToken,
   // Parse and evaluate the conditional expression.
   IdentifierInfo *IfNDefMacro = nullptr;
   const SourceLocation ConditionalBegin = CurPPLexer->getSourceLocation();
-  const bool ConditionalTrue = EvaluateDirectiveExpression(IfNDefMacro);
+  const DirectiveEvalResult DER = EvaluateDirectiveExpression(IfNDefMacro);
+  const bool ConditionalTrue = DER.Conditional;
   const SourceLocation ConditionalEnd = CurPPLexer->getSourceLocation();
 
   // If this condition is equivalent to #ifndef X, and if this is the first
@@ -2695,7 +2702,12 @@ void Preprocessor::HandleIfDirective(Token &IfToken,
                   (ConditionalTrue ? PPCallbacks::CVK_True : PPCallbacks::CVK_False));
 
   // Should we include the stuff contained by this directive?
-  if (ConditionalTrue) {
+  if (PPOpts->SingleFileParseMode && DER.IncludedUndefinedIds) {
+    // In 'single-file-parse mode' undefined identifiers trigger parsing of all
+    // the directive blocks.
+    CurPPLexer->pushConditionalLevel(IfToken.getLocation(), /*wasskip*/false,
+                                     /*foundnonskip*/false, /*foundelse*/false);
+  } else if (ConditionalTrue) {
     // Yes, remember that we are inside a conditional, then lex the next token.
     CurPPLexer->pushConditionalLevel(IfToken.getLocation(), /*wasskip*/false,
                                    /*foundnonskip*/true, /*foundelse*/false);
@@ -2756,6 +2768,14 @@ void Preprocessor::HandleElseDirective(Token &Result) {
   if (Callbacks)
     Callbacks->Else(Result.getLocation(), CI.IfLoc);
 
+  if (PPOpts->SingleFileParseMode && !CI.FoundNonSkip) {
+    // In 'single-file-parse mode' undefined identifiers trigger parsing of all
+    // the directive blocks.
+    CurPPLexer->pushConditionalLevel(CI.IfLoc, /*wasskip*/false,
+                                     /*foundnonskip*/false, /*foundelse*/true);
+    return;
+  }
+
   // Finally, skip the rest of the contents of this block.
   SkipExcludedConditionalBlock(CI.IfLoc, /*Foundnonskip*/true,
                                /*FoundElse*/true, Result.getLocation());
@@ -2791,6 +2811,14 @@ void Preprocessor::HandleElifDirective(Token &ElifToken) {
                     SourceRange(ConditionalBegin, ConditionalEnd),
                     PPCallbacks::CVK_NotEvaluated, CI.IfLoc);
 
+  if (PPOpts->SingleFileParseMode && !CI.FoundNonSkip) {
+    // In 'single-file-parse mode' undefined identifiers trigger parsing of all
+    // the directive blocks.
+    CurPPLexer->pushConditionalLevel(ElifToken.getLocation(), /*wasskip*/false,
+                                     /*foundnonskip*/false, /*foundelse*/false);
+    return;
+  }
+
   // Finally, skip the rest of the contents of this block.
   SkipExcludedConditionalBlock(CI.IfLoc, /*Foundnonskip*/true,
                                /*FoundElse*/CI.FoundElse,
diff --git a/lib/Lex/PPExpressions.cpp b/lib/Lex/PPExpressions.cpp
index 862a4713e4bc..12f5084298df 100644
--- a/lib/Lex/PPExpressions.cpp
+++ b/lib/Lex/PPExpressions.cpp
@@ -73,6 +73,7 @@ public:
 
 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
                                      Token &PeekTok, bool ValueLive,
+                                     bool &IncludedUndefinedIds,
                                      Preprocessor &PP);
 
 /// DefinedTracker - This struct is used while parsing expressions to keep track
@@ -93,6 +94,7 @@ struct DefinedTracker {
   /// TheMacro - When the state is DefinedMacro or NotDefinedMacro, this
   /// indicates the macro that was checked.
   IdentifierInfo *TheMacro;
+  bool IncludedUndefinedIds = false;
 };
 
 /// EvaluateDefined - Process a 'defined(sym)' expression.
@@ -128,6 +130,7 @@ static bool EvaluateDefined(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
   MacroDefinition Macro = PP.getMacroDefinition(II);
   Result.Val = !!Macro;
   Result.Val.setIsUnsigned(false); // Result is signed intmax_t.
+  DT.IncludedUndefinedIds = !Macro;
 
   // If there is a macro, mark it used.
   if (Result.Val != 0 && ValueLive)
@@ -255,6 +258,8 @@ static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
     Result.Val.setIsUnsigned(false);  // "0" is signed intmax_t 0.
     Result.setIdentifier(II);
     Result.setRange(PeekTok.getLocation());
+    DT.IncludedUndefinedIds = (II->getTokenID() != tok::kw_true &&
+                               II->getTokenID() != tok::kw_false);
     PP.LexNonComment(PeekTok);
     return false;
   }
@@ -400,7 +405,8 @@ static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT,
       // Just use DT unmodified as our result.
     } else {
       // Otherwise, we have something like (x+y), and we consumed '(x'.
-      if (EvaluateDirectiveSubExpr(Result, 1, PeekTok, ValueLive, PP))
+      if (EvaluateDirectiveSubExpr(Result, 1, PeekTok, ValueLive,
+                                   DT.IncludedUndefinedIds, PP))
         return true;
 
       if (PeekTok.isNot(tok::r_paren)) {
@@ -532,6 +538,7 @@ static void diagnoseUnexpectedOperator(Preprocessor &PP, PPValue &LHS,
 /// evaluation, such as division by zero warnings.
 static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
                                      Token &PeekTok, bool ValueLive,
+                                     bool &IncludedUndefinedIds,
                                      Preprocessor &PP) {
   unsigned PeekPrec = getPrecedence(PeekTok.getKind());
   // If this token isn't valid, report the error.
@@ -571,6 +578,7 @@ static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
     // Parse the RHS of the operator.
     DefinedTracker DT;
     if (EvaluateValue(RHS, PeekTok, DT, RHSIsLive, PP)) return true;
+    IncludedUndefinedIds = DT.IncludedUndefinedIds;
 
     // Remember the precedence of this operator and get the precedence of the
     // operator immediately to the right of the RHS.
@@ -601,7 +609,8 @@ static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
       RHSPrec = ThisPrec+1;
 
     if (PeekPrec >= RHSPrec) {
-      if (EvaluateDirectiveSubExpr(RHS, RHSPrec, PeekTok, RHSIsLive, PP))
+      if (EvaluateDirectiveSubExpr(RHS, RHSPrec, PeekTok, RHSIsLive,
+                                   IncludedUndefinedIds, PP))
         return true;
       PeekPrec = getPrecedence(PeekTok.getKind());
     }
@@ -769,7 +778,8 @@ static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
       // Parse anything after the : with the same precedence as ?.  We allow
       // things of equal precedence because ?: is right associative.
       if (EvaluateDirectiveSubExpr(AfterColonVal, ThisPrec,
-                                   PeekTok, AfterColonLive, PP))
+                                   PeekTok, AfterColonLive,
+                                   IncludedUndefinedIds, PP))
         return true;
 
       // Now that we have the condition, the LHS and the RHS of the :, evaluate.
@@ -806,7 +816,8 @@ static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec,
 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that
 /// may occur after a #if or #elif directive.  If the expression is equivalent
 /// to "!defined(X)" return X in IfNDefMacro.
-bool Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
+Preprocessor::DirectiveEvalResult
+Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
   SaveAndRestore<bool> PPDir(ParsingIfOrElifDirective, true);
   // Save the current state of 'DisableMacroExpansion' and reset it to false. If
   // 'DisableMacroExpansion' is true, then we must be in a macro argument list
@@ -833,7 +844,7 @@ bool Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
     
     // Restore 'DisableMacroExpansion'.
     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
-    return false;
+    return {false, DT.IncludedUndefinedIds};
   }
 
   // If we are at the end of the expression after just parsing a value, there
@@ -847,20 +858,20 @@ bool Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
 
     // Restore 'DisableMacroExpansion'.
     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
-    return ResVal.Val != 0;
+    return {ResVal.Val != 0, DT.IncludedUndefinedIds};
   }
 
   // Otherwise, we must have a binary operator (e.g. "#if 1 < 2"), so parse the
   // operator and the stuff after it.
   if (EvaluateDirectiveSubExpr(ResVal, getPrecedence(tok::question),
-                               Tok, true, *this)) {
+                               Tok, true, DT.IncludedUndefinedIds, *this)) {
     // Parse error, skip the rest of the macro line.
     if (Tok.isNot(tok::eod))
       DiscardUntilEndOfDirective();
     
     // Restore 'DisableMacroExpansion'.
     DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
-    return false;
+    return {false, DT.IncludedUndefinedIds};
   }
 
   // If we aren't at the tok::eod token, something bad happened, like an extra
@@ -872,5 +883,5 @@ bool Preprocessor::EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) {
 
   // Restore 'DisableMacroExpansion'.
   DisableMacroExpansion = DisableMacroExpansionAtStartOfDirective;
-  return ResVal.Val != 0;
+  return {ResVal.Val != 0, DT.IncludedUndefinedIds};
 }
diff --git a/lib/Lex/Pragma.cpp b/lib/Lex/Pragma.cpp
index c16478dd2be4..bf2363a0a6f4 100644
--- a/lib/Lex/Pragma.cpp
+++ b/lib/Lex/Pragma.cpp
@@ -20,6 +20,7 @@
 #include "clang/Basic/TokenKinds.h"
 #include "clang/Lex/HeaderSearch.h"
 #include "clang/Lex/LexDiagnostic.h"
+#include "clang/Lex/LiteralSupport.h"
 #include "clang/Lex/MacroInfo.h"
 #include "clang/Lex/PPCallbacks.h"
 #include "clang/Lex/Preprocessor.h"
@@ -754,15 +755,52 @@ void Preprocessor::HandlePragmaIncludeAlias(Token &Tok) {
   getHeaderSearchInfo().AddIncludeAlias(OriginalSource, ReplaceFileName);
 }
 
+// Lex a component of a module name: either an identifier or a string literal;
+// for components that can be expressed both ways, the two forms are equivalent.
+static bool LexModuleNameComponent(
+    Preprocessor &PP, Token &Tok,
+    std::pair<IdentifierInfo *, SourceLocation> &ModuleNameComponent,
+    bool First) {
+  PP.LexUnexpandedToken(Tok);
+  if (Tok.is(tok::string_literal) && !Tok.hasUDSuffix()) {
+    StringLiteralParser Literal(Tok, PP);
+    if (Literal.hadError)
+      return true;
+    ModuleNameComponent = std::make_pair(
+        PP.getIdentifierInfo(Literal.GetString()), Tok.getLocation());
+  } else if (!Tok.isAnnotation() && Tok.getIdentifierInfo()) {
+    ModuleNameComponent =
+        std::make_pair(Tok.getIdentifierInfo(), Tok.getLocation());
+  } else {
+    PP.Diag(Tok.getLocation(), diag::err_pp_expected_module_name) << First;
+    return true;
+  }
+  return false;
+}
+
+static bool LexModuleName(
+    Preprocessor &PP, Token &Tok,
+    llvm::SmallVectorImpl<std::pair<IdentifierInfo *, SourceLocation>>
+        &ModuleName) {
+  while (true) {
+    std::pair<IdentifierInfo*, SourceLocation> NameComponent;
+    if (LexModuleNameComponent(PP, Tok, NameComponent, ModuleName.empty()))
+      return true;
+    ModuleName.push_back(NameComponent);
+
+    PP.LexUnexpandedToken(Tok);
+    if (Tok.isNot(tok::period))
+      return false;
+  }
+}
+
 void Preprocessor::HandlePragmaModuleBuild(Token &Tok) {
   SourceLocation Loc = Tok.getLocation();
 
-  LexUnexpandedToken(Tok);
-  if (Tok.isAnnotation() || !Tok.getIdentifierInfo()) {
-    Diag(Tok.getLocation(), diag::err_pp_expected_module_name) << true;
+  std::pair<IdentifierInfo *, SourceLocation> ModuleNameLoc;
+  if (LexModuleNameComponent(*this, Tok, ModuleNameLoc, true))
     return;
-  }
-  IdentifierInfo *ModuleName = Tok.getIdentifierInfo();
+  IdentifierInfo *ModuleName = ModuleNameLoc.first;
 
   LexUnexpandedToken(Tok);
   if (Tok.isNot(tok::eod)) {
@@ -1383,26 +1421,6 @@ public:
   }
 };
 
-static bool LexModuleName(
-    Preprocessor &PP, Token &Tok,
-    llvm::SmallVectorImpl<std::pair<IdentifierInfo *, SourceLocation>>
-        &ModuleName) {
-  while (true) {
-    PP.LexUnexpandedToken(Tok);
-    if (Tok.isAnnotation() || !Tok.getIdentifierInfo()) {
-      PP.Diag(Tok.getLocation(), diag::err_pp_expected_module_name)
-        << ModuleName.empty();
-      return true;
-    }
-
-    ModuleName.emplace_back(Tok.getIdentifierInfo(), Tok.getLocation());
-
-    PP.LexUnexpandedToken(Tok);
-    if (Tok.isNot(tok::period))
-      return false;
-  }
-}
-
 /// Handle the clang \#pragma module import extension. The syntax is:
 /// \code
 ///   #pragma clang module import some.module.name
@@ -1473,7 +1491,7 @@ struct PragmaModuleBeginHandler : public PragmaHandler {
     // be loaded or implicitly loadable.
     // FIXME: We could create the submodule here. We'd need to know whether
     // it's supposed to be explicit, but not much else.
-    Module *M = PP.getHeaderSearchInfo().getModuleMap().findModule(Current);
+    Module *M = PP.getHeaderSearchInfo().lookupModule(Current);
     if (!M) {
       PP.Diag(ModuleName.front().second,
               diag::err_pp_module_begin_no_module_map) << Current;
diff --git a/lib/Parse/ParseDecl.cpp b/lib/Parse/ParseDecl.cpp
index 22696a957a10..d0ce9fc89583 100644
--- a/lib/Parse/ParseDecl.cpp
+++ b/lib/Parse/ParseDecl.cpp
@@ -2629,6 +2629,8 @@ Parser::getDeclSpecContextFromDeclaratorContext(unsigned Context) {
     return DSC_class;
   if (Context == Declarator::FileContext)
     return DSC_top_level;
+  if (Context == Declarator::TemplateParamContext)
+    return DSC_template_param;
   if (Context == Declarator::TemplateTypeArgContext)
     return DSC_template_type_arg;
   if (Context == Declarator::TrailingReturnContext)
@@ -4261,7 +4263,9 @@ void Parser::ParseEnumSpecifier(SourceLocation StartLoc, DeclSpec &DS,
                                    AS, DS.getModulePrivateSpecLoc(), TParams,
                                    Owned, IsDependent, ScopedEnumKWLoc,
                                    IsScopedUsingClassTag, BaseType,
-                                   DSC == DSC_type_specifier, &SkipBody);
+                                   DSC == DSC_type_specifier,
+                                   DSC == DSC_template_param ||
+                                   DSC == DSC_template_type_arg, &SkipBody);
 
   if (SkipBody.ShouldSkip) {
     assert(TUK == Sema::TUK_Definition && "can only skip a definition");
diff --git a/lib/Parse/ParseDeclCXX.cpp b/lib/Parse/ParseDeclCXX.cpp
index 1a4607a84cff..a724fa242268 100644
--- a/lib/Parse/ParseDeclCXX.cpp
+++ b/lib/Parse/ParseDeclCXX.cpp
@@ -1887,7 +1887,8 @@ void Parser::ParseClassSpecifier(tok::TokenKind TagTokKind,
                                        SourceLocation(), false,
                                        clang::TypeResult(),
                                        DSC == DSC_type_specifier,
-                                       &SkipBody);
+                                       DSC == DSC_template_param ||
+                                       DSC == DSC_template_type_arg, &SkipBody);
 
     // If ActOnTag said the type was dependent, try again with the
     // less common call.
diff --git a/lib/Parse/ParseObjc.cpp b/lib/Parse/ParseObjc.cpp
index 77e63efc065e..caa6323d3209 100644
--- a/lib/Parse/ParseObjc.cpp
+++ b/lib/Parse/ParseObjc.cpp
@@ -3627,6 +3627,14 @@ void Parser::ParseLexedObjCMethodDefs(LexedMethod &LM, bool parseMethod) {
   SourceLocation OrigLoc = Tok.getLocation();
 
   assert(!LM.Toks.empty() && "ParseLexedObjCMethodDef - Empty body!");
+  // Store an artificial EOF token to ensure that we don't run off the end of
+  // the method's body when we come to parse it.
+  Token Eof;
+  Eof.startToken();
+  Eof.setKind(tok::eof);
+  Eof.setEofData(MCDecl);
+  Eof.setLocation(OrigLoc);
+  LM.Toks.push_back(Eof);
   // Append the current token at the end of the new token stream so that it
   // doesn't get lost.
   LM.Toks.push_back(Tok);
@@ -3658,7 +3666,7 @@ void Parser::ParseLexedObjCMethodDefs(LexedMethod &LM, bool parseMethod) {
       Actions.ActOnDefaultCtorInitializers(MCDecl);
     ParseFunctionStatementBody(MCDecl, BodyScope);
   }
-  
+
   if (Tok.getLocation() != OrigLoc) {
     // Due to parsing error, we either went over the cached tokens or
     // there are still cached tokens left. If it's the latter case skip the
@@ -3670,4 +3678,6 @@ void Parser::ParseLexedObjCMethodDefs(LexedMethod &LM, bool parseMethod) {
       while (Tok.getLocation() != OrigLoc && Tok.isNot(tok::eof))
         ConsumeAnyToken();
   }
+  // Clean up the remaining EOF token.
+  ConsumeAnyToken();
 }
diff --git a/lib/Parse/ParseTemplate.cpp b/lib/Parse/ParseTemplate.cpp
index 37c80fe5e520..944cd775d52a 100644
--- a/lib/Parse/ParseTemplate.cpp
+++ b/lib/Parse/ParseTemplate.cpp
@@ -674,7 +674,8 @@ Parser::ParseNonTypeTemplateParameter(unsigned Depth, unsigned Position) {
   // FIXME: The type should probably be restricted in some way... Not all
   // declarators (parts of declarators?) are accepted for parameters.
   DeclSpec DS(AttrFactory);
-  ParseDeclarationSpecifiers(DS);
+  ParseDeclarationSpecifiers(DS, ParsedTemplateInfo(), AS_none,
+                             DSC_template_param);
 
   // Parse this as a typename.
   Declarator ParamDecl(DS, Declarator::TemplateParamContext);
diff --git a/lib/Sema/AnalysisBasedWarnings.cpp b/lib/Sema/AnalysisBasedWarnings.cpp
index db688b12cbcf..934e13e72d05 100644
--- a/lib/Sema/AnalysisBasedWarnings.cpp
+++ b/lib/Sema/AnalysisBasedWarnings.cpp
@@ -279,6 +279,150 @@ static void checkRecursiveFunction(Sema &S, const FunctionDecl *FD,
 }
 
 //===----------------------------------------------------------------------===//
+// Check for throw in a non-throwing function.
+//===----------------------------------------------------------------------===//
+enum ThrowState {
+  FoundNoPathForThrow,
+  FoundPathForThrow,
+  FoundPathWithNoThrowOutFunction,
+};
+
+static bool isThrowCaught(const CXXThrowExpr *Throw,
+                          const CXXCatchStmt *Catch) {
+  const Type *ThrowType = nullptr;
+  if (Throw->getSubExpr())
+    ThrowType = Throw->getSubExpr()->getType().getTypePtrOrNull();
+  if (!ThrowType)
+    return false;
+  const Type *CaughtType = Catch->getCaughtType().getTypePtrOrNull();
+  if (!CaughtType)
+    return true;
+  if (ThrowType->isReferenceType())
+    ThrowType = ThrowType->castAs<ReferenceType>()
+                    ->getPointeeType()
+                    ->getUnqualifiedDesugaredType();
+  if (CaughtType->isReferenceType())
+    CaughtType = CaughtType->castAs<ReferenceType>()
+                     ->getPointeeType()
+                     ->getUnqualifiedDesugaredType();
+  if (CaughtType == ThrowType)
+    return true;
+  const CXXRecordDecl *CaughtAsRecordType =
+      CaughtType->getPointeeCXXRecordDecl();
+  const CXXRecordDecl *ThrowTypeAsRecordType = ThrowType->getAsCXXRecordDecl();
+  if (CaughtAsRecordType && ThrowTypeAsRecordType)
+    return ThrowTypeAsRecordType->isDerivedFrom(CaughtAsRecordType);
+  return false;
+}
+
+static bool isThrowCaughtByHandlers(const CXXThrowExpr *CE,
+                                    const CXXTryStmt *TryStmt) {
+  for (unsigned H = 0, E = TryStmt->getNumHandlers(); H < E; ++H) {
+    if (isThrowCaught(CE, TryStmt->getHandler(H)))
+      return true;
+  }
+  return false;
+}
+
+static bool doesThrowEscapePath(CFGBlock Block, SourceLocation &OpLoc) {
+  for (const auto &B : Block) {
+    if (B.getKind() != CFGElement::Statement)
+      continue;
+    const auto *CE = dyn_cast<CXXThrowExpr>(B.getAs<CFGStmt>()->getStmt());
+    if (!CE)
+      continue;
+
+    OpLoc = CE->getThrowLoc();
+    for (const auto &I : Block.succs()) {
+      if (!I.isReachable())
+        continue;
+      if (const auto *Terminator =
+              dyn_cast_or_null<CXXTryStmt>(I->getTerminator()))
+        if (isThrowCaughtByHandlers(CE, Terminator))
+          return false;
+    }
+    return true;
+  }
+  return false;
+}
+
+static bool hasThrowOutNonThrowingFunc(SourceLocation &OpLoc, CFG *BodyCFG) {
+
+  unsigned ExitID = BodyCFG->getExit().getBlockID();
+
+  SmallVector<ThrowState, 16> States(BodyCFG->getNumBlockIDs(),
+                                     FoundNoPathForThrow);
+  States[BodyCFG->getEntry().getBlockID()] = FoundPathWithNoThrowOutFunction;
+
+  SmallVector<CFGBlock *, 16> Stack;
+  Stack.push_back(&BodyCFG->getEntry());
+  while (!Stack.empty()) {
+    CFGBlock *CurBlock = Stack.back();
+    Stack.pop_back();
+
+    unsigned ID = CurBlock->getBlockID();
+    ThrowState CurState = States[ID];
+    if (CurState == FoundPathWithNoThrowOutFunction) {
+      if (ExitID == ID)
+        continue;
+
+      if (doesThrowEscapePath(*CurBlock, OpLoc))
+        CurState = FoundPathForThrow;
+    }
+
+    // Loop over successor blocks and add them to the Stack if their state
+    // changes.
+    for (const auto &I : CurBlock->succs())
+      if (I.isReachable()) {
+        unsigned NextID = I->getBlockID();
+        if (NextID == ExitID && CurState == FoundPathForThrow) {
+          States[NextID] = CurState;
+        } else if (States[NextID] < CurState) {
+          States[NextID] = CurState;
+          Stack.push_back(I);
+        }
+      }
+  }
+  // Return true if the exit node is reachable, and only reachable through
+  // a throw expression.
+  return States[ExitID] == FoundPathForThrow;
+}
+
+static void EmitDiagForCXXThrowInNonThrowingFunc(Sema &S, SourceLocation OpLoc,
+                                                 const FunctionDecl *FD) {
+  if (!S.getSourceManager().isInSystemHeader(OpLoc)) {
+    S.Diag(OpLoc, diag::warn_throw_in_noexcept_func) << FD;
+    if (S.getLangOpts().CPlusPlus11 &&
+        (isa<CXXDestructorDecl>(FD) ||
+         FD->getDeclName().getCXXOverloadedOperator() == OO_Delete ||
+         FD->getDeclName().getCXXOverloadedOperator() == OO_Array_Delete))
+      S.Diag(FD->getLocation(), diag::note_throw_in_dtor);
+    else
+      S.Diag(FD->getLocation(), diag::note_throw_in_function);
+  }
+}
+
+static void checkThrowInNonThrowingFunc(Sema &S, const FunctionDecl *FD,
+                                        AnalysisDeclContext &AC) {
+  CFG *BodyCFG = AC.getCFG();
+  if (!BodyCFG)
+    return;
+  if (BodyCFG->getExit().pred_empty())
+    return;
+  SourceLocation OpLoc;
+  if (hasThrowOutNonThrowingFunc(OpLoc, BodyCFG))
+    EmitDiagForCXXThrowInNonThrowingFunc(S, OpLoc, FD);
+}
+
+static bool isNoexcept(const FunctionDecl *FD) {
+  const auto *FPT = FD->getType()->castAs<FunctionProtoType>();
+  if (FPT->getExceptionSpecType() != EST_None &&
+      FPT->isNothrow(FD->getASTContext()))
+    return true;
+  return false;
+}
+
+//===----------------------------------------------------------------------===//
 // Check for missing return value.
 //===----------------------------------------------------------------------===//
 
@@ -2127,6 +2271,12 @@ AnalysisBasedWarnings::IssueWarnings(sema::AnalysisBasedWarnings::Policy P,
     }
   }
 
+  // Check for throw out of non-throwing function.
+  if (!Diags.isIgnored(diag::warn_throw_in_noexcept_func, D->getLocStart()))
+    if (const FunctionDecl *FD = dyn_cast<FunctionDecl>(D))
+      if (S.getLangOpts().CPlusPlus && isNoexcept(FD))
+        checkThrowInNonThrowingFunc(S, FD, AC);
+
   // If none of the previous checks caused a CFG build, trigger one here
   // for -Wtautological-overlap-compare
   if (!Diags.isIgnored(diag::warn_tautological_overlap_comparison,
diff --git a/lib/Sema/Sema.cpp b/lib/Sema/Sema.cpp
index e7b0914641ff..007a5e483e6c 100644
--- a/lib/Sema/Sema.cpp
+++ b/lib/Sema/Sema.cpp
@@ -740,6 +740,9 @@ void Sema::ActOnEndOfTranslationUnit() {
       // Load pending instantiations from the external source.
       SmallVector<PendingImplicitInstantiation, 4> Pending;
       ExternalSource->ReadPendingInstantiations(Pending);
+      for (auto PII : Pending)
+        if (auto Func = dyn_cast<FunctionDecl>(PII.first))
+          Func->setInstantiationIsPending(true);
       PendingInstantiations.insert(PendingInstantiations.begin(),
                                    Pending.begin(), Pending.end());
     }
diff --git a/lib/Sema/SemaCUDA.cpp b/lib/Sema/SemaCUDA.cpp
index b938ac387c4d..cac5f682275e 100644
--- a/lib/Sema/SemaCUDA.cpp
+++ b/lib/Sema/SemaCUDA.cpp
@@ -629,12 +629,6 @@ static bool IsKnownEmitted(Sema &S, FunctionDecl *FD) {
   // emitted, because (say) the definition could include "inline".
   FunctionDecl *Def = FD->getDefinition();
 
-  // We may currently be parsing the body of FD, in which case
-  // FD->getDefinition() will be null, but we still want to treat FD as though
-  // it's a definition.
-  if (!Def && FD->willHaveBody())
-    Def = FD;
-
   if (Def &&
       !isDiscardableGVALinkage(S.getASTContext().GetGVALinkageForFunction(Def)))
     return true;
diff --git a/lib/Sema/SemaDecl.cpp b/lib/Sema/SemaDecl.cpp
index cba220daf774..e340456bc6da 100644
--- a/lib/Sema/SemaDecl.cpp
+++ b/lib/Sema/SemaDecl.cpp
@@ -7260,11 +7260,11 @@ void Sema::CheckVariableDeclarationType(VarDecl *NewVD) {
         NewVD->setInvalidDecl();
         return;
       }
-      // OpenCL v1.1 s6.5.2 and s6.5.3 no local or constant variables
-      // in functions.
       if (T.getAddressSpace() == LangAS::opencl_constant ||
           T.getAddressSpace() == LangAS::opencl_local) {
         FunctionDecl *FD = getCurFunctionDecl();
+        // OpenCL v1.1 s6.5.2 and s6.5.3: no local or constant variables
+        // in functions.
         if (FD && !FD->hasAttr<OpenCLKernelAttr>()) {
           if (T.getAddressSpace() == LangAS::opencl_constant)
             Diag(NewVD->getLocation(), diag::err_opencl_function_variable)
@@ -7275,6 +7275,20 @@ void Sema::CheckVariableDeclarationType(VarDecl *NewVD) {
           NewVD->setInvalidDecl();
           return;
         }
+        // OpenCL v2.0 s6.5.2 and s6.5.3: local and constant variables must be
+        // in the outermost scope of a kernel function.
+        if (FD && FD->hasAttr<OpenCLKernelAttr>()) {
+          if (!getCurScope()->isFunctionScope()) {
+            if (T.getAddressSpace() == LangAS::opencl_constant)
+              Diag(NewVD->getLocation(), diag::err_opencl_addrspace_scope)
+                  << "constant";
+            else
+              Diag(NewVD->getLocation(), diag::err_opencl_addrspace_scope)
+                  << "local";
+            NewVD->setInvalidDecl();
+            return;
+          }
+        }
       } else if (T.getAddressSpace() != LangAS::Default) {
         // Do not allow other address spaces on automatic variable.
         Diag(NewVD->getLocation(), diag::err_as_qualified_auto_decl) << 1;
@@ -12218,6 +12232,7 @@ Decl *Sema::ActOnFinishFunctionBody(Decl *dcl, Stmt *Body,
 
   if (FD) {
     FD->setBody(Body);
+    FD->setWillHaveBody(false);
 
     if (getLangOpts().CPlusPlus14) {
       if (!FD->isInvalidDecl() && Body && !FD->isDependentContext() &&
@@ -13075,7 +13090,8 @@ Decl *Sema::ActOnTag(Scope *S, unsigned TagSpec, TagUseKind TUK,
                      SourceLocation ScopedEnumKWLoc,
                      bool ScopedEnumUsesClassTag,
                      TypeResult UnderlyingType,
-                     bool IsTypeSpecifier, SkipBodyInfo *SkipBody) {
+                     bool IsTypeSpecifier, bool IsTemplateParamOrArg,
+                     SkipBodyInfo *SkipBody) {
   // If this is not a definition, it must have a name.
   IdentifierInfo *OrigName = Name;
   assert((Name != nullptr || TUK == TUK_Definition) &&
@@ -13345,11 +13361,11 @@ Decl *Sema::ActOnTag(Scope *S, unsigned TagSpec, TagUseKind TUK,
   // also need to do a redeclaration lookup there, just in case
   // there's a shadow friend decl.
   if (Name && Previous.empty() &&
-      (TUK == TUK_Reference || TUK == TUK_Friend)) {
+      (TUK == TUK_Reference || TUK == TUK_Friend || IsTemplateParamOrArg)) {
     if (Invalid) goto CreateNewDecl;
     assert(SS.isEmpty());
 
-    if (TUK == TUK_Reference) {
+    if (TUK == TUK_Reference || IsTemplateParamOrArg) {
       // C++ [basic.scope.pdecl]p5:
       //   -- for an elaborated-type-specifier of the form
       //
@@ -13782,7 +13798,8 @@ CreateNewDecl:
 
   // C++11 [dcl.type]p3:
   //   A type-specifier-seq shall not define a class or enumeration [...].
-  if (getLangOpts().CPlusPlus && IsTypeSpecifier && TUK == TUK_Definition) {
+  if (getLangOpts().CPlusPlus && (IsTypeSpecifier || IsTemplateParamOrArg) &&
+      TUK == TUK_Definition) {
     Diag(New->getLocation(), diag::err_type_defined_in_type_specifier)
       << Context.getTagDeclType(New);
     Invalid = true;
@@ -16089,7 +16106,10 @@ void Sema::ActOnModuleBegin(SourceLocation DirectiveLoc, Module *Mod) {
   // lexically within the module.
   if (getLangOpts().trackLocalOwningModule()) {
     for (auto *DC = CurContext; DC; DC = DC->getLexicalParent()) {
-      cast<Decl>(DC)->setHidden(true);
+      cast<Decl>(DC)->setModuleOwnershipKind(
+          getLangOpts().ModulesLocalVisibility
+              ? Decl::ModuleOwnershipKind::VisibleWhenImported
+              : Decl::ModuleOwnershipKind::Visible);
       cast<Decl>(DC)->setLocalOwningModule(Mod);
     }
   }
@@ -16129,7 +16149,8 @@ void Sema::ActOnModuleEnd(SourceLocation EomLoc, Module *Mod) {
     for (auto *DC = CurContext; DC; DC = DC->getLexicalParent()) {
       cast<Decl>(DC)->setLocalOwningModule(getCurrentModule());
       if (!getCurrentModule())
-        cast<Decl>(DC)->setHidden(false);
+        cast<Decl>(DC)->setModuleOwnershipKind(
+            Decl::ModuleOwnershipKind::Unowned);
     }
   }
 }
diff --git a/lib/Sema/SemaDeclAttr.cpp b/lib/Sema/SemaDeclAttr.cpp
index b43642f5493b..e8503427536d 100644
--- a/lib/Sema/SemaDeclAttr.cpp
+++ b/lib/Sema/SemaDeclAttr.cpp
@@ -6903,6 +6903,32 @@ static bool ShouldDiagnoseAvailabilityInContext(Sema &S, AvailabilityResult K,
   return true;
 }
 
+static bool
+shouldDiagnoseAvailabilityByDefault(const ASTContext &Context,
+                                    const VersionTuple &DeploymentVersion,
+                                    const VersionTuple &DeclVersion) {
+  const auto &Triple = Context.getTargetInfo().getTriple();
+  VersionTuple ForceAvailabilityFromVersion;
+  switch (Triple.getOS()) {
+  case llvm::Triple::IOS:
+  case llvm::Triple::TvOS:
+    ForceAvailabilityFromVersion = VersionTuple(/*Major=*/11);
+    break;
+  case llvm::Triple::WatchOS:
+    ForceAvailabilityFromVersion = VersionTuple(/*Major=*/4);
+    break;
+  case llvm::Triple::Darwin:
+  case llvm::Triple::MacOSX:
+    ForceAvailabilityFromVersion = VersionTuple(/*Major=*/10, /*Minor=*/13);
+    break;
+  default:
+    // New targets should always warn about availability.
+    return Triple.getVendor() == llvm::Triple::Apple;
+  }
+  return DeploymentVersion >= ForceAvailabilityFromVersion ||
+         DeclVersion >= ForceAvailabilityFromVersion;
+}
+
 static void DoEmitAvailabilityWarning(Sema &S, AvailabilityResult K,
                                       Decl *Ctx, const NamedDecl *D,
                                       StringRef Message, SourceLocation Loc,
@@ -6991,13 +7017,26 @@ static void DoEmitAvailabilityWarning(Sema &S, AvailabilityResult K,
     }
     break;
 
-  case AR_NotYetIntroduced:
-    diag = diag::warn_partial_availability;
-    diag_message = diag::warn_partial_message;
-    diag_fwdclass_message = diag::warn_partial_fwdclass_message;
+  case AR_NotYetIntroduced: {
+    // We would like to emit the diagnostic even if -Wunguarded-availability is
+    // not specified for deployment targets >= to iOS 11 or equivalent or
+    // for declarations that were introduced in iOS 11 (macOS 10.13, ...) or
+    // later.
+    const AvailabilityAttr *AA = getAttrForPlatform(S.getASTContext(), D);
+    VersionTuple Introduced = AA->getIntroduced();
+    bool NewWarning = shouldDiagnoseAvailabilityByDefault(
+        S.Context, S.Context.getTargetInfo().getPlatformMinVersion(),
+        Introduced);
+    diag = NewWarning ? diag::warn_partial_availability_new
+                      : diag::warn_partial_availability;
+    diag_message = NewWarning ? diag::warn_partial_message_new
+                              : diag::warn_partial_message;
+    diag_fwdclass_message = NewWarning ? diag::warn_partial_fwdclass_message_new
+                                       : diag::warn_partial_fwdclass_message;
     property_note_select = /* partial */ 2;
     available_here_select_kind = /* partial */ 3;
     break;
+  }
 
   case AR_Available:
     llvm_unreachable("Warning for availability of available declaration?");
@@ -7317,7 +7356,18 @@ void DiagnoseUnguardedAvailability::DiagnoseDeclAvailability(
     if (!ShouldDiagnoseAvailabilityInContext(SemaRef, Result, Introduced, Ctx))
       return;
 
-    SemaRef.Diag(Range.getBegin(), diag::warn_unguarded_availability)
+    // We would like to emit the diagnostic even if -Wunguarded-availability is
+    // not specified for deployment targets >= to iOS 11 or equivalent or
+    // for declarations that were introduced in iOS 11 (macOS 10.13, ...) or
+    // later.
+    unsigned DiagKind =
+        shouldDiagnoseAvailabilityByDefault(
+            SemaRef.Context,
+            SemaRef.Context.getTargetInfo().getPlatformMinVersion(), Introduced)
+            ? diag::warn_unguarded_availability_new
+            : diag::warn_unguarded_availability;
+
+    SemaRef.Diag(Range.getBegin(), DiagKind)
         << Range << D
         << AvailabilityAttr::getPrettyPlatformName(
                SemaRef.getASTContext().getTargetInfo().getPlatformName())
diff --git a/lib/Sema/SemaDeclCXX.cpp b/lib/Sema/SemaDeclCXX.cpp
index 844299bb87cf..453ece9d9c4d 100644
--- a/lib/Sema/SemaDeclCXX.cpp
+++ b/lib/Sema/SemaDeclCXX.cpp
@@ -13394,7 +13394,8 @@ Decl *Sema::ActOnTemplatedFriendTag(Scope *S, SourceLocation FriendLoc,
                       /*ScopedEnumKWLoc=*/SourceLocation(),
                       /*ScopedEnumUsesClassTag=*/false,
                       /*UnderlyingType=*/TypeResult(),
-                      /*IsTypeSpecifier=*/false);
+                      /*IsTypeSpecifier=*/false,
+                      /*IsTemplateParamOrArg=*/false);
     }
 
     NestedNameSpecifierLoc QualifierLoc = SS.getWithLocInContext(Context);
@@ -13878,6 +13879,9 @@ void Sema::SetDeclDeleted(Decl *Dcl, SourceLocation DelLoc) {
     return;
   }
 
+  // Deleted function does not have a body.
+  Fn->setWillHaveBody(false);
+
   if (const FunctionDecl *Prev = Fn->getPreviousDecl()) {
     // Don't consider the implicit declaration we generate for explicit
     // specializations. FIXME: Do not generate these implicit declarations.
diff --git a/lib/Sema/SemaExpr.cpp b/lib/Sema/SemaExpr.cpp
index 75a6903392ea..f49df6b3216d 100644
--- a/lib/Sema/SemaExpr.cpp
+++ b/lib/Sema/SemaExpr.cpp
@@ -13732,6 +13732,7 @@ void Sema::MarkFunctionReferenced(SourceLocation Loc, FunctionDecl *Func,
         // call to such a function.
         InstantiateFunctionDefinition(PointOfInstantiation, Func);
       else {
+        Func->setInstantiationIsPending(true);
         PendingInstantiations.push_back(std::make_pair(Func,
                                                        PointOfInstantiation));
         // Notify the consumer that a function was implicitly instantiated.
diff --git a/lib/Sema/SemaExprCXX.cpp b/lib/Sema/SemaExprCXX.cpp
index a9ff21bc41ab..710ead30790f 100644
--- a/lib/Sema/SemaExprCXX.cpp
+++ b/lib/Sema/SemaExprCXX.cpp
@@ -2630,7 +2630,7 @@ void Sema::DeclareGlobalAllocationFunction(DeclarationName Name,
           // Make the function visible to name lookup, even if we found it in
           // an unimported module. It either is an implicitly-declared global
           // allocation function, or is suppressing that function.
-          Func->setHidden(false);
+          Func->setVisibleDespiteOwningModule();
           return;
         }
       }
@@ -2662,7 +2662,7 @@ void Sema::DeclareGlobalAllocationFunction(DeclarationName Name,
         FnType, /*TInfo=*/nullptr, SC_None, false, true);
     Alloc->setImplicit();
     // Global allocation functions should always be visible.
-    Alloc->setHidden(false);
+    Alloc->setVisibleDespiteOwningModule();
 
     // Implicit sized deallocation functions always have default visibility.
     Alloc->addAttr(
diff --git a/lib/Sema/SemaLookup.cpp b/lib/Sema/SemaLookup.cpp
index 1fb25f4e0e7c..2e7fb875a276 100644
--- a/lib/Sema/SemaLookup.cpp
+++ b/lib/Sema/SemaLookup.cpp
@@ -862,6 +862,16 @@ static bool LookupDirect(Sema &S, LookupResult &R, const DeclContext *DC) {
   if (!Record->isCompleteDefinition())
     return Found;
 
+  // For conversion operators, 'operator auto' should only match
+  // 'operator auto'.  Since 'auto' is not a type, it shouldn't be considered
+  // as a candidate for template substitution.
+  auto *ContainedDeducedType =
+      R.getLookupName().getCXXNameType()->getContainedDeducedType();
+  if (R.getLookupName().getNameKind() ==
+          DeclarationName::CXXConversionFunctionName &&
+      ContainedDeducedType && ContainedDeducedType->isUndeducedType())
+    return Found;
+
   for (CXXRecordDecl::conversion_iterator U = Record->conversion_begin(),
          UEnd = Record->conversion_end(); U != UEnd; ++U) {
     FunctionTemplateDecl *ConvTemplate = dyn_cast<FunctionTemplateDecl>(*U);
@@ -1331,7 +1341,7 @@ void Sema::makeMergedDefinitionVisible(NamedDecl *ND) {
     Context.mergeDefinitionIntoModule(ND, M);
   else
     // We're not building a module; just make the definition visible.
-    ND->setHidden(false);
+    ND->setVisibleDespiteOwningModule();
 
   // If ND is a template declaration, make the template parameters
   // visible too. They're not (necessarily) within a mergeable DeclContext.
@@ -1518,7 +1528,7 @@ bool LookupResult::isVisibleSlow(Sema &SemaRef, NamedDecl *D) {
           !SemaRef.getLangOpts().ModulesLocalVisibility) {
         // Cache the fact that this declaration is implicitly visible because
         // its parent has a visible definition.
-        D->setHidden(false);
+        D->setVisibleDespiteOwningModule();
       }
       return true;
     }
diff --git a/lib/Sema/SemaTemplate.cpp b/lib/Sema/SemaTemplate.cpp
index 1eea151a4ec8..a8923ce9e27d 100644
--- a/lib/Sema/SemaTemplate.cpp
+++ b/lib/Sema/SemaTemplate.cpp
@@ -8612,7 +8612,8 @@ Sema::ActOnExplicitInstantiation(Scope *S,
                         /*ModulePrivateLoc=*/SourceLocation(),
                         MultiTemplateParamsArg(), Owned, IsDependent,
                         SourceLocation(), false, TypeResult(),
-                        /*IsTypeSpecifier*/false);
+                        /*IsTypeSpecifier*/false,
+                        /*IsTemplateParamOrArg*/false);
   assert(!IsDependent && "explicit instantiation of dependent name not yet handled");
 
   if (!TagD)
diff --git a/lib/Sema/SemaTemplateInstantiate.cpp b/lib/Sema/SemaTemplateInstantiate.cpp
index fe92dd8ac653..f4f0c804aee1 100644
--- a/lib/Sema/SemaTemplateInstantiate.cpp
+++ b/lib/Sema/SemaTemplateInstantiate.cpp
@@ -2045,7 +2045,7 @@ Sema::InstantiateClass(SourceLocation PointOfInstantiation,
 
   // The instantiation is visible here, even if it was first declared in an
   // unimported module.
-  Instantiation->setHidden(false);
+  Instantiation->setVisibleDespiteOwningModule();
 
   // FIXME: This loses the as-written tag kind for an explicit instantiation.
   Instantiation->setTagKind(Pattern->getTagKind());
@@ -2247,7 +2247,7 @@ bool Sema::InstantiateEnum(SourceLocation PointOfInstantiation,
 
   // The instantiation is visible here, even if it was first declared in an
   // unimported module.
-  Instantiation->setHidden(false);
+  Instantiation->setVisibleDespiteOwningModule();
 
   // Enter the scope of this instantiation. We don't use
   // PushDeclContext because we don't have a scope.
diff --git a/lib/Sema/SemaTemplateInstantiateDecl.cpp b/lib/Sema/SemaTemplateInstantiateDecl.cpp
index 148ce24293a0..abe912fb548b 100644
--- a/lib/Sema/SemaTemplateInstantiateDecl.cpp
+++ b/lib/Sema/SemaTemplateInstantiateDecl.cpp
@@ -1782,6 +1782,9 @@ Decl *TemplateDeclInstantiator::VisitFunctionDecl(FunctionDecl *D,
       Previous.clear();
   }
 
+  if (isFriend)
+    Function->setObjectOfFriendDecl();
+
   SemaRef.CheckFunctionDeclaration(/*Scope*/ nullptr, Function, Previous,
                                    isExplicitSpecialization);
 
@@ -3782,6 +3785,7 @@ void Sema::InstantiateFunctionDefinition(SourceLocation PointOfInstantiation,
       // Try again at the end of the translation unit (at which point a
       // definition will be required).
       assert(!Recursive);
+      Function->setInstantiationIsPending(true);
       PendingInstantiations.push_back(
         std::make_pair(Function, PointOfInstantiation));
     } else if (TSK == TSK_ImplicitInstantiation) {
@@ -3801,6 +3805,7 @@ void Sema::InstantiateFunctionDefinition(SourceLocation PointOfInstantiation,
   // Postpone late parsed template instantiations.
   if (PatternDecl->isLateTemplateParsed() &&
       !LateTemplateParser) {
+    Function->setInstantiationIsPending(true);
     PendingInstantiations.push_back(
       std::make_pair(Function, PointOfInstantiation));
     return;
@@ -3863,7 +3868,7 @@ void Sema::InstantiateFunctionDefinition(SourceLocation PointOfInstantiation,
 
   // The instantiation is visible here, even if it was first declared in an
   // unimported module.
-  Function->setHidden(false);
+  Function->setVisibleDespiteOwningModule();
 
   // Copy the inner loc start from the pattern.
   Function->setInnerLocStart(PatternDecl->getInnerLocStart());
@@ -4269,7 +4274,7 @@ void Sema::InstantiateVariableDefinition(SourceLocation PointOfInstantiation,
 
       // The instantiation is visible here, even if it was first declared in an
       // unimported module.
-      Var->setHidden(false);
+      Var->setVisibleDespiteOwningModule();
 
       // If we're performing recursive template instantiation, create our own
       // queue of pending implicit instantiations that we will instantiate
@@ -4285,9 +4290,6 @@ void Sema::InstantiateVariableDefinition(SourceLocation PointOfInstantiation,
       InstantiateVariableInitializer(Var, PatternDecl, TemplateArgs);
       PreviousContext.pop();
 
-      // FIXME: Need to inform the ASTConsumer that we instantiated the
-      // initializer?
-
       // This variable may have local implicit instantiations that need to be
       // instantiated within this scope.
       LocalInstantiations.perform();
@@ -4397,7 +4399,6 @@ void Sema::InstantiateVariableDefinition(SourceLocation PointOfInstantiation,
   if (Def->isStaticDataMember() && !Def->isOutOfLine()) {
     // We're instantiating an inline static data member whose definition was
     // provided inside the class.
-    // FIXME: Update record?
     InstantiateVariableInitializer(Var, Def, TemplateArgs);
   } else if (!VarSpec) {
     Var = cast_or_null<VarDecl>(SubstDecl(Def, Var->getDeclContext(),
@@ -5146,6 +5147,8 @@ void Sema::PerformPendingInstantiations(bool LocalOnly) {
                                 TSK_ExplicitInstantiationDefinition;
       InstantiateFunctionDefinition(/*FIXME:*/Inst.second, Function, true,
                                     DefinitionRequired, true);
+      if (Function->isDefined())
+        Function->setInstantiationIsPending(false);
       continue;
     }
 
diff --git a/lib/Serialization/ASTReader.cpp b/lib/Serialization/ASTReader.cpp
index eeb0132c1690..ef2841849ff6 100644
--- a/lib/Serialization/ASTReader.cpp
+++ b/lib/Serialization/ASTReader.cpp
@@ -3580,8 +3580,8 @@ static void moveMethodToBackOfGlobalList(Sema &S, ObjCMethodDecl *Method) {
 void ASTReader::makeNamesVisible(const HiddenNames &Names, Module *Owner) {
   assert(Owner->NameVisibility != Module::Hidden && "nothing to make visible?");
   for (Decl *D : Names) {
-    bool wasHidden = D->Hidden;
-    D->Hidden = false;
+    bool wasHidden = D->isHidden();
+    D->setVisibleDespiteOwningModule();
 
     if (wasHidden && SemaObj) {
       if (ObjCMethodDecl *Method = dyn_cast<ObjCMethodDecl>(D)) {
@@ -3648,7 +3648,7 @@ void ASTReader::mergeDefinitionVisibility(NamedDecl *Def,
   if (Def->isHidden()) {
     // If MergedDef is visible or becomes visible, make the definition visible.
     if (!MergedDef->isHidden())
-      Def->Hidden = false;
+      Def->setVisibleDespiteOwningModule();
     else if (getContext().getLangOpts().ModulesLocalVisibility) {
       getContext().mergeDefinitionIntoModule(
           Def, MergedDef->getImportedOwningModule(),
@@ -9324,9 +9324,20 @@ void ASTReader::diagnoseOdrViolations() {
              diag::err_module_odr_violation_different_definitions)
             << FirstRecord << FirstModule.empty() << FirstModule;
 
+        if (FirstDecl) {
+          Diag(FirstDecl->getLocation(), diag::note_first_module_difference)
+              << FirstRecord << FirstDecl->getSourceRange();
+        }
+
         Diag(SecondRecord->getLocation(),
              diag::note_module_odr_violation_different_definitions)
             << SecondModule;
+
+        if (SecondDecl) {
+          Diag(SecondDecl->getLocation(), diag::note_second_module_difference)
+              << SecondDecl->getSourceRange();
+        }
+
         Diagnosed = true;
         break;
       }
diff --git a/lib/Serialization/ASTReaderDecl.cpp b/lib/Serialization/ASTReaderDecl.cpp
index ed103e629216..07ab421cfc51 100644
--- a/lib/Serialization/ASTReaderDecl.cpp
+++ b/lib/Serialization/ASTReaderDecl.cpp
@@ -522,31 +522,29 @@ void ASTDeclReader::VisitDecl(Decl *D) {
   D->setTopLevelDeclInObjCContainer(Record.readInt());
   D->setAccess((AccessSpecifier)Record.readInt());
   D->FromASTFile = true;
-  D->setModulePrivate(Record.readInt());
-  D->Hidden = D->isModulePrivate();
+  bool ModulePrivate = Record.readInt();
 
   // Determine whether this declaration is part of a (sub)module. If so, it
   // may not yet be visible.
   if (unsigned SubmoduleID = readSubmoduleID()) {
     // Store the owning submodule ID in the declaration.
+    D->setModuleOwnershipKind(
+        ModulePrivate ? Decl::ModuleOwnershipKind::ModulePrivate
+                      : Decl::ModuleOwnershipKind::VisibleWhenImported);
     D->setOwningModuleID(SubmoduleID);
 
-    if (D->Hidden) {
-      // Module-private declarations are never visible, so there is no work to do.
+    if (ModulePrivate) {
+      // Module-private declarations are never visible, so there is no work to
+      // do.
     } else if (Reader.getContext().getLangOpts().ModulesLocalVisibility) {
       // If local visibility is being tracked, this declaration will become
-      // hidden and visible as the owning module does. Inform Sema that this
-      // declaration might not be visible.
-      D->Hidden = true;
+      // hidden and visible as the owning module does.
     } else if (Module *Owner = Reader.getSubmodule(SubmoduleID)) {
-      if (Owner->NameVisibility != Module::AllVisible) {
-        // The owning module is not visible. Mark this declaration as hidden.
-        D->Hidden = true;
-
-        // Note that this declaration was hidden because its owning module is 
-        // not yet visible.
+      // Mark the declaration as visible when its owning module becomes visible.
+      if (Owner->NameVisibility == Module::AllVisible)
+        D->setVisibleDespiteOwningModule();
+      else
         Reader.HiddenNamesMap[Owner].push_back(D);
-      }
     }
   }
 }
@@ -3934,10 +3932,21 @@ void ASTDeclReader::UpdateDecl(Decl *D) {
       break;
     }
 
-    case UPD_CXX_INSTANTIATED_STATIC_DATA_MEMBER:
-      cast<VarDecl>(D)->getMemberSpecializationInfo()->setPointOfInstantiation(
+    case UPD_CXX_INSTANTIATED_STATIC_DATA_MEMBER: {
+      VarDecl *VD = cast<VarDecl>(D);
+      VD->getMemberSpecializationInfo()->setPointOfInstantiation(
           ReadSourceLocation());
+      uint64_t Val = Record.readInt();
+      if (Val && !VD->getInit()) {
+        VD->setInit(Record.readExpr());
+        if (Val > 1) { // IsInitKnownICE = 1, IsInitNotICE = 2, IsInitICE = 3
+          EvaluatedStmt *Eval = VD->ensureEvaluatedStmt();
+          Eval->CheckedICE = true;
+          Eval->IsICE = Val == 3;
+        }
+      }
       break;
+    }
 
     case UPD_CXX_INSTANTIATED_DEFAULT_ARGUMENT: {
       auto Param = cast<ParmVarDecl>(D);
@@ -4133,7 +4142,7 @@ void ASTDeclReader::UpdateDecl(Decl *D) {
         Reader.HiddenNamesMap[Owner].push_back(Exported);
       } else {
         // The declaration is now visible.
-        Exported->Hidden = false;
+        Exported->setVisibleDespiteOwningModule();
       }
       break;
     }
diff --git a/lib/Serialization/ASTWriter.cpp b/lib/Serialization/ASTWriter.cpp
index dcacabec1225..1c0db14ced14 100644
--- a/lib/Serialization/ASTWriter.cpp
+++ b/lib/Serialization/ASTWriter.cpp
@@ -5033,9 +5033,18 @@ void ASTWriter::WriteDeclUpdatesBlocks(RecordDataImpl &OffsetsRecord) {
       case UPD_CXX_ADDED_FUNCTION_DEFINITION:
         break;
 
-      case UPD_CXX_INSTANTIATED_STATIC_DATA_MEMBER:
+      case UPD_CXX_INSTANTIATED_STATIC_DATA_MEMBER: {
+        const VarDecl *VD = cast<VarDecl>(D);
         Record.AddSourceLocation(Update.getLoc());
+        if (VD->getInit()) {
+          Record.push_back(!VD->isInitKnownICE() ? 1
+                                                 : (VD->isInitICE() ? 3 : 2));
+          Record.AddStmt(const_cast<Expr*>(VD->getInit()));
+        } else {
+          Record.push_back(0);
+        }
         break;
+      }
 
       case UPD_CXX_INSTANTIATED_DEFAULT_ARGUMENT:
         Record.AddStmt(const_cast<Expr *>(
diff --git a/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp b/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp
index 48d6cd8a527c..097d4198800d 100644
--- a/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp
+++ b/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp
@@ -50,8 +50,10 @@ bool BuiltinFunctionChecker::evalCall(const CallExpr *CE,
     state = state->assume(ArgSVal.castAs<DefinedOrUnknownSVal>(), true);
     // FIXME: do we want to warn here? Not right now. The most reports might
     // come from infeasible paths, thus being false positives.
-    if (!state)
+    if (!state) {
+      C.generateSink(C.getState(), C.getPredecessor());
       return true;
+    }
 
     C.addTransition(state);
     return true;
diff --git a/lib/StaticAnalyzer/Checkers/CStringChecker.cpp b/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
index 32e3ce9270aa..77c24629d71e 100644
--- a/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
+++ b/lib/StaticAnalyzer/Checkers/CStringChecker.cpp
@@ -120,6 +120,7 @@ public:
   void evalStdCopy(CheckerContext &C, const CallExpr *CE) const;
   void evalStdCopyBackward(CheckerContext &C, const CallExpr *CE) const;
   void evalStdCopyCommon(CheckerContext &C, const CallExpr *CE) const;
+  void evalMemset(CheckerContext &C, const CallExpr *CE) const;
 
   // Utility methods
   std::pair<ProgramStateRef , ProgramStateRef >
@@ -1999,6 +2000,54 @@ void CStringChecker::evalStdCopyCommon(CheckerContext &C,
   C.addTransition(State);
 }
 
+void CStringChecker::evalMemset(CheckerContext &C, const CallExpr *CE) const {
+  if (CE->getNumArgs() != 3)
+    return;
+
+  CurrentFunctionDescription = "memory set function";
+
+  const Expr *Mem = CE->getArg(0);
+  const Expr *Size = CE->getArg(2);
+  ProgramStateRef State = C.getState();
+
+  // See if the size argument is zero.
+  const LocationContext *LCtx = C.getLocationContext();
+  SVal SizeVal = State->getSVal(Size, LCtx);
+  QualType SizeTy = Size->getType();
+
+  ProgramStateRef StateZeroSize, StateNonZeroSize;
+  std::tie(StateZeroSize, StateNonZeroSize) =
+    assumeZero(C, State, SizeVal, SizeTy);
+
+  // Get the value of the memory area.
+  SVal MemVal = State->getSVal(Mem, LCtx);
+
+  // If the size is zero, there won't be any actual memory access, so
+  // just bind the return value to the Mem buffer and return.
+  if (StateZeroSize && !StateNonZeroSize) {
+    StateZeroSize = StateZeroSize->BindExpr(CE, LCtx, MemVal);
+    C.addTransition(StateZeroSize);
+    return;
+  }
+
+  // Ensure the memory area is not null.
+  // If it is NULL there will be a NULL pointer dereference.
+  State = checkNonNull(C, StateNonZeroSize, Mem, MemVal);
+  if (!State)
+    return;
+
+  State = CheckBufferAccess(C, State, Size, Mem);
+  if (!State)
+    return;
+  State = InvalidateBuffer(C, State, Mem, C.getSVal(Mem),
+                           /*IsSourceBuffer*/false, Size);
+  if (!State)
+    return;
+
+  State = State->BindExpr(CE, LCtx, MemVal);
+  C.addTransition(State);
+}
+
 static bool isCPPStdLibraryFunction(const FunctionDecl *FD, StringRef Name) {
   IdentifierInfo *II = FD->getIdentifier();
   if (!II)
@@ -2032,6 +2081,8 @@ bool CStringChecker::evalCall(const CallExpr *CE, CheckerContext &C) const {
     evalFunction =  &CStringChecker::evalMemcmp;
   else if (C.isCLibraryFunction(FDecl, "memmove"))
     evalFunction =  &CStringChecker::evalMemmove;
+  else if (C.isCLibraryFunction(FDecl, "memset"))
+    evalFunction =  &CStringChecker::evalMemset;
   else if (C.isCLibraryFunction(FDecl, "strcpy"))
     evalFunction =  &CStringChecker::evalStrcpy;
   else if (C.isCLibraryFunction(FDecl, "strncpy"))
diff --git a/lib/StaticAnalyzer/Checkers/CloneChecker.cpp b/lib/StaticAnalyzer/Checkers/CloneChecker.cpp
index 1885b0e39203..83955c586b68 100644
--- a/lib/StaticAnalyzer/Checkers/CloneChecker.cpp
+++ b/lib/StaticAnalyzer/Checkers/CloneChecker.cpp
@@ -73,12 +73,17 @@ void CloneChecker::checkEndOfTranslationUnit(const TranslationUnitDecl *TU,
   bool ReportNormalClones = Mgr.getAnalyzerOptions().getBooleanOption(
       "ReportNormalClones", true, this);
 
+  StringRef IgnoredFilesPattern = Mgr.getAnalyzerOptions().getOptionAsString(
+      "IgnoredFilesPattern", "", this);
+
   // Let the CloneDetector create a list of clones from all the analyzed
   // statements. We don't filter for matching variable patterns at this point
   // because reportSuspiciousClones() wants to search them for errors.
   std::vector<CloneDetector::CloneGroup> AllCloneGroups;
 
-  Detector.findClones(AllCloneGroups, RecursiveCloneTypeIIConstraint(),
+  Detector.findClones(AllCloneGroups,
+                      FilenamePatternConstraint(IgnoredFilesPattern),
+                      RecursiveCloneTypeIIConstraint(),
                       MinComplexityConstraint(MinComplexity),
                       MinGroupSizeConstraint(2), OnlyLargestCloneConstraint());
 
diff --git a/lib/StaticAnalyzer/Checkers/LocalizationChecker.cpp b/lib/StaticAnalyzer/Checkers/LocalizationChecker.cpp
index af35c2b0e991..6bbaaac05e6b 100644
--- a/lib/StaticAnalyzer/Checkers/LocalizationChecker.cpp
+++ b/lib/StaticAnalyzer/Checkers/LocalizationChecker.cpp
@@ -281,6 +281,9 @@ void NonLocalizedStringChecker::initUIMethods(ASTContext &Ctx) const {
   IdentifierInfo *setLabelNSSegmentedControl[] = {
       &Ctx.Idents.get("setLabel"), &Ctx.Idents.get("forSegment")};
   ADD_METHOD(NSSegmentedControl, setLabelNSSegmentedControl, 2, 0)
+  IdentifierInfo *setToolTipNSSegmentedControl[] = {
+      &Ctx.Idents.get("setToolTip"), &Ctx.Idents.get("forSegment")};
+  ADD_METHOD(NSSegmentedControl, setToolTipNSSegmentedControl, 2, 0)
 
   NEW_RECEIVER(NSButtonCell)
   ADD_UNARY_METHOD(NSButtonCell, setTitle, 0)
@@ -562,6 +565,46 @@ void NonLocalizedStringChecker::initUIMethods(ASTContext &Ctx) const {
   IdentifierInfo *setTitleUISegmentedControl[] = {
       &Ctx.Idents.get("setTitle"), &Ctx.Idents.get("forSegmentAtIndex")};
   ADD_METHOD(UISegmentedControl, setTitleUISegmentedControl, 2, 0)
+
+  NEW_RECEIVER(NSAccessibilityCustomRotorItemResult)
+  IdentifierInfo
+      *initWithItemLoadingTokenNSAccessibilityCustomRotorItemResult[] = {
+          &Ctx.Idents.get("initWithItemLoadingToken"),
+          &Ctx.Idents.get("customLabel")};
+  ADD_METHOD(NSAccessibilityCustomRotorItemResult,
+             initWithItemLoadingTokenNSAccessibilityCustomRotorItemResult, 2, 1)
+  ADD_UNARY_METHOD(NSAccessibilityCustomRotorItemResult, setCustomLabel, 0)
+
+  NEW_RECEIVER(UIContextualAction)
+  IdentifierInfo *contextualActionWithStyleUIContextualAction[] = {
+      &Ctx.Idents.get("contextualActionWithStyle"), &Ctx.Idents.get("title"),
+      &Ctx.Idents.get("handler")};
+  ADD_METHOD(UIContextualAction, contextualActionWithStyleUIContextualAction, 3,
+             1)
+  ADD_UNARY_METHOD(UIContextualAction, setTitle, 0)
+
+  NEW_RECEIVER(NSAccessibilityCustomRotor)
+  IdentifierInfo *initWithLabelNSAccessibilityCustomRotor[] = {
+      &Ctx.Idents.get("initWithLabel"), &Ctx.Idents.get("itemSearchDelegate")};
+  ADD_METHOD(NSAccessibilityCustomRotor,
+             initWithLabelNSAccessibilityCustomRotor, 2, 0)
+  ADD_UNARY_METHOD(NSAccessibilityCustomRotor, setLabel, 0)
+
+  NEW_RECEIVER(NSWindowTab)
+  ADD_UNARY_METHOD(NSWindowTab, setTitle, 0)
+  ADD_UNARY_METHOD(NSWindowTab, setToolTip, 0)
+
+  NEW_RECEIVER(NSAccessibilityCustomAction)
+  IdentifierInfo *initWithNameNSAccessibilityCustomAction[] = {
+      &Ctx.Idents.get("initWithName"), &Ctx.Idents.get("handler")};
+  ADD_METHOD(NSAccessibilityCustomAction,
+             initWithNameNSAccessibilityCustomAction, 2, 0)
+  IdentifierInfo *initWithNameTargetNSAccessibilityCustomAction[] = {
+      &Ctx.Idents.get("initWithName"), &Ctx.Idents.get("target"),
+      &Ctx.Idents.get("selector")};
+  ADD_METHOD(NSAccessibilityCustomAction,
+             initWithNameTargetNSAccessibilityCustomAction, 3, 0)
+  ADD_UNARY_METHOD(NSAccessibilityCustomAction, setName, 0)
 }
 
 #define LSF_INSERT(function_name) LSF.insert(&Ctx.Idents.get(function_name));
diff --git a/lib/StaticAnalyzer/Core/AnalyzerOptions.cpp b/lib/StaticAnalyzer/Core/AnalyzerOptions.cpp
index 45ef612ee1d5..11b9f8c4f725 100644
--- a/lib/StaticAnalyzer/Core/AnalyzerOptions.cpp
+++ b/lib/StaticAnalyzer/Core/AnalyzerOptions.cpp
@@ -293,7 +293,7 @@ unsigned AnalyzerOptions::getMaxInlinableSize() {
         DefaultValue = 4;
         break;
       case UMK_Deep:
-        DefaultValue = 50;
+        DefaultValue = 100;
         break;
     }
 
@@ -332,7 +332,7 @@ unsigned AnalyzerOptions::getMaxNodesPerTopLevelFunction() {
         DefaultValue = 75000;
         break;
       case UMK_Deep:
-        DefaultValue = 150000;
+        DefaultValue = 225000;
         break;
     }
     MaxNodesPerTopLevelFunction = getOptionAsInteger("max-nodes", DefaultValue);
diff --git a/lib/StaticAnalyzer/Core/ExprEngineC.cpp b/lib/StaticAnalyzer/Core/ExprEngineC.cpp
index 8f720a2067b1..6f1e8391e67c 100644
--- a/lib/StaticAnalyzer/Core/ExprEngineC.cpp
+++ b/lib/StaticAnalyzer/Core/ExprEngineC.cpp
@@ -980,10 +980,9 @@ void ExprEngine::VisitUnaryOperator(const UnaryOperator* U, ExplodedNode *Pred,
           //    transfer functions as "0 == E".
           SVal Result;
           if (Optional<Loc> LV = V.getAs<Loc>()) {
-            Loc X = svalBuilder.makeNull();
+            Loc X = svalBuilder.makeNullWithType(Ex->getType());
             Result = evalBinOp(state, BO_EQ, *LV, X, U->getType());
-          }
-          else if (Ex->getType()->isFloatingType()) {
+          } else if (Ex->getType()->isFloatingType()) {
             // FIXME: handle floating point types.
             Result = UnknownVal();
           } else {
diff --git a/test/Analysis/analyzer-config.c b/test/Analysis/analyzer-config.c
index c0153a50532a..70521c63fbad 100644
--- a/test/Analysis/analyzer-config.c
+++ b/test/Analysis/analyzer-config.c
@@ -19,8 +19,8 @@ void foo() {
 // CHECK-NEXT: ipa = dynamic-bifurcate
 // CHECK-NEXT: ipa-always-inline-size = 3
 // CHECK-NEXT: leak-diagnostics-reference-allocation = false
-// CHECK-NEXT: max-inlinable-size = 50
-// CHECK-NEXT: max-nodes = 150000
+// CHECK-NEXT: max-inlinable-size = 100
+// CHECK-NEXT: max-nodes = 225000
 // CHECK-NEXT: max-times-inline-large = 32
 // CHECK-NEXT: min-cfg-size-treat-functions-as-large = 14
 // CHECK-NEXT: mode = deep
diff --git a/test/Analysis/analyzer-config.cpp b/test/Analysis/analyzer-config.cpp
index f84be1781160..60c03c1e497b 100644
--- a/test/Analysis/analyzer-config.cpp
+++ b/test/Analysis/analyzer-config.cpp
@@ -30,8 +30,8 @@ public:
 // CHECK-NEXT: ipa = dynamic-bifurcate
 // CHECK-NEXT: ipa-always-inline-size = 3
 // CHECK-NEXT: leak-diagnostics-reference-allocation = false
-// CHECK-NEXT: max-inlinable-size = 50
-// CHECK-NEXT: max-nodes = 150000
+// CHECK-NEXT: max-inlinable-size = 100
+// CHECK-NEXT: max-nodes = 225000
 // CHECK-NEXT: max-times-inline-large = 32
 // CHECK-NEXT: min-cfg-size-treat-functions-as-large = 14
 // CHECK-NEXT: mode = deep
diff --git a/test/Analysis/builtin-assume.c b/test/Analysis/builtin-assume.c
deleted file mode 100644
index 00d651d9e3be..000000000000
--- a/test/Analysis/builtin-assume.c
+++ /dev/null
@@ -1,8 +0,0 @@
-// RUN: %clang_analyze_cc1 -analyzer-checker=core,debug.ExprInspection -verify %s
-
-void clang_analyzer_eval(int);
-
-void f(int i) {
-  __builtin_assume(i < 10);
-  clang_analyzer_eval(i < 15); // expected-warning {{TRUE}}
-}
diff --git a/test/Analysis/builtin-functions.cpp b/test/Analysis/builtin-functions.cpp
index 4e9859754d62..2c1950251145 100644
--- a/test/Analysis/builtin-functions.cpp
+++ b/test/Analysis/builtin-functions.cpp
@@ -1,6 +1,7 @@
 // RUN: %clang_analyze_cc1 -triple x86_64-apple-darwin10 -analyzer-checker=core,debug.ExprInspection %s -std=c++11 -verify
 
 void clang_analyzer_eval(bool);
+void clang_analyzer_warnIfReached();
 
 void testAddressof(int x) {
   clang_analyzer_eval(&x == __builtin_addressof(x)); // expected-warning{{TRUE}}
@@ -50,3 +51,16 @@ void test_assume_aligned_4(char *p) {
   q = (char*) __builtin_assume_aligned(p + 1, 16);
   clang_analyzer_eval(p == q); // expected-warning{{FALSE}}
 }
+
+void f(int i) {
+  __builtin_assume(i < 10);
+  clang_analyzer_eval(i < 15); // expected-warning {{TRUE}}
+}
+
+void g(int i) {
+  if (i > 5) {
+    __builtin_assume(i < 5);
+    clang_analyzer_warnIfReached(); // Assumtion contradicts constraints.
+                                    // We give up the analysis on this path.
+  }
+}
diff --git a/test/Analysis/copypaste/autogenerated_automoc.cpp b/test/Analysis/copypaste/autogenerated_automoc.cpp
new file mode 100644
index 000000000000..55963c4545c9
--- /dev/null
+++ b/test/Analysis/copypaste/autogenerated_automoc.cpp
@@ -0,0 +1,19 @@
+// RUN: %clang_analyze_cc1 -std=c++11 -analyzer-checker=alpha.clone.CloneChecker -analyzer-config alpha.clone.CloneChecker:IgnoredFilesPattern="moc_|.*_automoc.cpp" -verify %s
+
+// Because files that have `_automoc.' in their names are most likely autogenerated,
+// we suppress copy-paste warnings here.
+
+// expected-no-diagnostics
+
+void f1() {
+  int *p1 = new int[1];
+  int *p2 = new int[1];
+  if (p1) {
+    delete [] p1;
+    p1 = nullptr;
+  }
+  if (p2) {
+    delete [] p1; // no-warning
+    p2 = nullptr;
+  }
+}
diff --git a/test/Analysis/copypaste/dbus_autogenerated.cpp b/test/Analysis/copypaste/dbus_autogenerated.cpp
new file mode 100644
index 000000000000..182437565813
--- /dev/null
+++ b/test/Analysis/copypaste/dbus_autogenerated.cpp
@@ -0,0 +1,19 @@
+// RUN: %clang_analyze_cc1 -std=c++11 -analyzer-checker=alpha.clone.CloneChecker -analyzer-config alpha.clone.CloneChecker:IgnoredFilesPattern="moc_|dbus_|.*_automoc" -verify %s
+
+// Because files that have `dbus_' in their names are most likely autogenerated,
+// we suppress copy-paste warnings here.
+
+// expected-no-diagnostics
+
+void f1() {
+  int *p1 = new int[1];
+  int *p2 = new int[1];
+  if (p1) {
+    delete [] p1;
+    p1 = nullptr;
+  }
+  if (p2) {
+    delete [] p1; // no-warning
+    p2 = nullptr;
+  }
+}
diff --git a/test/Analysis/copypaste/moc_autogenerated.cpp b/test/Analysis/copypaste/moc_autogenerated.cpp
new file mode 100644
index 000000000000..626fe2a3dd07
--- /dev/null
+++ b/test/Analysis/copypaste/moc_autogenerated.cpp
@@ -0,0 +1,19 @@
+// RUN: %clang_analyze_cc1 -std=c++11 -analyzer-checker=alpha.clone.CloneChecker -analyzer-config alpha.clone.CloneChecker:IgnoredFilesPattern="moc_|.*_automoc" -verify %s
+
+// Because files that have `moc_' in their names are most likely autogenerated,
+// we suppress copy-paste warnings here.
+
+// expected-no-diagnostics
+
+void f1() {
+  int *p1 = new int[1];
+  int *p2 = new int[1];
+  if (p1) {
+    delete [] p1;
+    p1 = nullptr;
+  }
+  if (p2) {
+    delete [] p1; // no-warning
+    p2 = nullptr;
+  }
+}
diff --git a/test/Analysis/copypaste/not-autogenerated.cpp b/test/Analysis/copypaste/not-autogenerated.cpp
new file mode 100644
index 000000000000..765e7aaf2aab
--- /dev/null
+++ b/test/Analysis/copypaste/not-autogenerated.cpp
@@ -0,0 +1,14 @@
+// RUN: %clang_analyze_cc1 -std=c++11 -analyzer-checker=alpha.clone.CloneChecker -analyzer-config alpha.clone.CloneChecker:IgnoredFilesPattern="moc_|ui_|dbus_|.*_automoc" -verify %s
+
+void f1() {
+  int *p1 = new int[1];
+  int *p2 = new int[1];
+  if (p1) {
+    delete [] p1; // expected-note{{Similar code using 'p1' here}}
+    p1 = nullptr;
+  }
+  if (p2) {
+    delete [] p1; // expected-warning{{Potential copy-paste error; did you really mean to use 'p1' here?}}
+    p2 = nullptr;
+  }
+}
diff --git a/test/Analysis/copypaste/ui_autogenerated.cpp b/test/Analysis/copypaste/ui_autogenerated.cpp
new file mode 100644
index 000000000000..a08c33fe9e2a
--- /dev/null
+++ b/test/Analysis/copypaste/ui_autogenerated.cpp
@@ -0,0 +1,19 @@
+// RUN: %clang_analyze_cc1 -std=c++11 -analyzer-checker=alpha.clone.CloneChecker -analyzer-config alpha.clone.CloneChecker:IgnoredFilesPattern="moc_|ui_|.*_automoc" -verify %s
+
+// Because files that have `ui_' in their names are most likely autogenerated,
+// we suppress copy-paste warnings here.
+
+// expected-no-diagnostics
+
+void f1() {
+  int *p1 = new int[1];
+  int *p2 = new int[1];
+  if (p1) {
+    delete [] p1;
+    p1 = nullptr;
+  }
+  if (p2) {
+    delete [] p1; // no-warning
+    p2 = nullptr;
+  }
+}
diff --git a/test/Analysis/null-deref-ps-region.c b/test/Analysis/null-deref-ps-region.c
index 6ef99ae473ca..c46ca6c52ae3 100644
--- a/test/Analysis/null-deref-ps-region.c
+++ b/test/Analysis/null-deref-ps-region.c
@@ -1,6 +1,11 @@
-// RUN: %clang_analyze_cc1 -analyzer-checker=core,alpha.core -std=gnu99 -analyzer-store=region -verify %s
-// expected-no-diagnostics
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,alpha.core,unix,alpha.unix -std=gnu99 -analyzer-store=region -verify %s
 
+#include "Inputs/system-header-simulator.h"
+
+typedef __typeof(sizeof(int)) size_t;
+void *memset(void *__s, int __c, size_t __n);
+void *malloc(size_t __size);
+void free(void *__ptr);
 
 // The store for 'a[1]' should not be removed mistakenly. SymbolicRegions may
 // also be live roots.
@@ -13,3 +18,55 @@ void f14(int *a) {
     i = *p; // no-warning
   }
 }
+
+void foo() {
+  int *x = malloc(sizeof(int));
+  memset(x, 0, sizeof(int));
+  int n = 1 / *x; // FIXME: no-warning
+  free(x);
+}
+
+void bar() {
+  int *x = malloc(sizeof(int));
+  memset(x, 0, 1);
+  int n = 1 / *x; // no-warning
+  free(x);
+}
+
+void testConcreteNull() {
+  int *x = 0;
+  memset(x, 0, 1); // expected-warning {{Null pointer argument in call to memory set function}}
+}
+
+void testStackArray() {
+  char buf[13];
+  memset(buf, 0, 1); // no-warning
+}
+
+void testHeapSymbol() {
+  char *buf = (char *)malloc(13);
+  memset(buf, 0, 1); // no-warning
+  free(buf);
+}
+
+void testStackArrayOutOfBound() {
+  char buf[1];
+  memset(buf, 0, 1024); // expected-warning {{Memory set function accesses out-of-bound array element}}
+}
+
+void testHeapSymbolOutOfBound() {
+  char *buf = (char *)malloc(1);
+  memset(buf, 0, 1024); // expected-warning {{Memory set function accesses out-of-bound array element}}
+  free(buf);
+}
+
+void testStackArraySameSize() {
+  char buf[1];
+  memset(buf, 0, sizeof(buf)); // no-warning
+}
+
+void testHeapSymbolSameSize() {
+  char *buf = (char *)malloc(1);
+  memset(buf, 0, 1); // no-warning
+  free(buf);
+}
diff --git a/test/CXX/except/except.spec/p11.cpp b/test/CXX/except/except.spec/p11.cpp
index 1d0a647fb4f4..196f84c557ea 100644
--- a/test/CXX/except/except.spec/p11.cpp
+++ b/test/CXX/except/except.spec/p11.cpp
@@ -1,12 +1,11 @@
 // RUN: %clang_cc1 -std=c++11 -fexceptions -fcxx-exceptions -fsyntax-only -verify %s
-// expected-no-diagnostics
 
 // This is the "let the user shoot themselves in the foot" clause.
-void f() noexcept {
-  throw 0; // no-error
+void f() noexcept { // expected-note {{non-throwing function declare here}}
+  throw 0; // expected-warning {{has a non-throwing exception specification but}} 
 }
-void g() throw() {
-  throw 0; // no-error
+void g() throw() { // expected-note {{non-throwing function declare here}}
+  throw 0; // expected-warning {{has a non-throwing exception specification but}} 
 }
 void h() throw(int) {
   throw 0.0; // no-error
diff --git a/test/CodeGen/64bit-swiftcall.c b/test/CodeGen/64bit-swiftcall.c
index 06c314501552..92ba37cd7fe6 100644
--- a/test/CodeGen/64bit-swiftcall.c
+++ b/test/CodeGen/64bit-swiftcall.c
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 -triple x86_64-apple-darwin10 -target-cpu core2 -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-apple-darwin10 -target-cpu core2 -emit-llvm -o - %s | FileCheck %s --check-prefix=X86-64
 // RUN: %clang_cc1 -triple arm64-apple-ios9 -target-cpu cyclone -emit-llvm -o - %s | FileCheck %s
 // RUN: %clang_cc1 -triple arm64-apple-ios9 -target-cpu cyclone -emit-llvm -o - %s | FileCheck %s --check-prefix=ARM64
 
@@ -1014,3 +1015,20 @@ typedef struct {
 TEST(struct_v1f3)
 // ARM64-LABEL: define swiftcc { <2 x float>, float } @return_struct_v1f3()
 // ARM64-LABEL: define swiftcc void @take_struct_v1f3(<2 x float>, float)
+
+typedef struct {
+  int3 vect;
+  unsigned long long val;
+} __attribute__((packed)) padded_alloc_size_vector;
+TEST(padded_alloc_size_vector)
+// X86-64-LABEL: take_padded_alloc_size_vector(<3 x i32>, i64)
+// X86-64-NOT: [4 x i8]
+// x86-64: ret void
+
+typedef union {
+  float f1;
+  float3 fv2;
+} union_hom_fp_partial2;
+TEST(union_hom_fp_partial2)
+// X86-64-LABEL: take_union_hom_fp_partial2(i64, float)
+// ARM64-LABEL: take_union_hom_fp_partial2(i64, float)
diff --git a/test/CodeGen/aarch64-neon-intrinsics.c b/test/CodeGen/aarch64-neon-intrinsics.c
index bcb680c4b518..cbc2e72fcbac 100644
--- a/test/CodeGen/aarch64-neon-intrinsics.c
+++ b/test/CodeGen/aarch64-neon-intrinsics.c
@@ -9037,10 +9037,9 @@ int64x2_t test_vld1q_s64(int64_t const *a) {
 
 // CHECK-LABEL: @test_vld1q_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <8 x i16>*
-// CHECK:   [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* [[TMP1]]
-// CHECK:   [[TMP3:%.*]] = bitcast <8 x i16> [[TMP2]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP3]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <8 x half>*
+// CHECK:   [[TMP2:%.*]] = load <8 x half>, <8 x half>* [[TMP1]]
+// CHECK:   ret <8 x half> [[TMP2]]
 float16x8_t test_vld1q_f16(float16_t const *a) {
   return vld1q_f16(a);
 }
@@ -9152,10 +9151,9 @@ int64x1_t test_vld1_s64(int64_t const *a) {
 
 // CHECK-LABEL: @test_vld1_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <4 x i16>*
-// CHECK:   [[TMP2:%.*]] = load <4 x i16>, <4 x i16>* [[TMP1]]
-// CHECK:   [[TMP3:%.*]] = bitcast <4 x i16> [[TMP2]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP3]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <4 x half>*
+// CHECK:   [[TMP2:%.*]] = load <4 x half>, <4 x half>* [[TMP1]]
+// CHECK:   ret <4 x half> [[TMP2]]
 float16x4_t test_vld1_f16(float16_t const *a) {
   return vld1_f16(a);
 }
@@ -9342,10 +9340,10 @@ int64x2x2_t test_vld2q_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x2_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x i16>*
-// CHECK:   [[VLD2:%.*]] = call { <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld2.v8i16.p0v8i16(<8 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16> } [[VLD2]], { <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x half>*
+// CHECK:   [[VLD2:%.*]] = call { <8 x half>, <8 x half> } @llvm.aarch64.neon.ld2.v8f16.p0v8f16(<8 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half> } [[VLD2]], { <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 16, i1 false)
@@ -9573,10 +9571,10 @@ int64x1x2_t test_vld2_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x2_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x i16>*
-// CHECK:   [[VLD2:%.*]] = call { <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld2.v4i16.p0v4i16(<4 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16> } [[VLD2]], { <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x half>*
+// CHECK:   [[VLD2:%.*]] = call { <4 x half>, <4 x half> } @llvm.aarch64.neon.ld2.v4f16.p0v4f16(<4 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half> } [[VLD2]], { <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 16, i32 8, i1 false)
@@ -9804,10 +9802,10 @@ int64x2x3_t test_vld3q_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x3_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x i16>*
-// CHECK:   [[VLD3:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld3.v8i16.p0v8i16(<8 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16> } [[VLD3]], { <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x half>*
+// CHECK:   [[VLD3:%.*]] = call { <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld3.v8f16.p0v8f16(<8 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half> } [[VLD3]], { <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 48, i32 16, i1 false)
@@ -10035,10 +10033,10 @@ int64x1x3_t test_vld3_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x3_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x i16>*
-// CHECK:   [[VLD3:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld3.v4i16.p0v4i16(<4 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16> } [[VLD3]], { <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x half>*
+// CHECK:   [[VLD3:%.*]] = call { <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld3.v4f16.p0v4f16(<4 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half> } [[VLD3]], { <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 24, i32 8, i1 false)
@@ -10266,10 +10264,10 @@ int64x2x4_t test_vld4q_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x4_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x i16>*
-// CHECK:   [[VLD4:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld4.v8i16.p0v8i16(<8 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } [[VLD4]], { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <8 x half>*
+// CHECK:   [[VLD4:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld4.v8f16.p0v8f16(<8 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half>, <8 x half> } [[VLD4]], { <8 x half>, <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 64, i32 16, i1 false)
@@ -10497,10 +10495,10 @@ int64x1x4_t test_vld4_s64(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x4_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x i16>*
-// CHECK:   [[VLD4:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld4.v4i16.p0v4i16(<4 x i16>* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } [[VLD4]], { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <4 x half>*
+// CHECK:   [[VLD4:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld4.v4f16.p0v4f16(<4 x half>* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half>, <4 x half> } [[VLD4]], { <4 x half>, <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 8, i1 false)
@@ -10666,9 +10664,9 @@ void test_vst1q_s64(int64_t *a, int64x2_t b) {
 // CHECK-LABEL: @test_vst1q_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <8 x i16>*
-// CHECK:   [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   store <8 x i16> [[TMP3]], <8 x i16>* [[TMP2]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <8 x half>*
+// CHECK:   [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   store <8 x half> [[TMP3]], <8 x half>* [[TMP2]]
 // CHECK:   ret void
 void test_vst1q_f16(float16_t *a, float16x8_t b) {
   vst1q_f16(a, b);
@@ -10800,9 +10798,9 @@ void test_vst1_s64(int64_t *a, int64x1_t b) {
 // CHECK-LABEL: @test_vst1_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <4 x i16>*
-// CHECK:   [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   store <4 x i16> [[TMP3]], <4 x i16>* [[TMP2]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <4 x half>*
+// CHECK:   [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   store <4 x half> [[TMP3]], <4 x half>* [[TMP2]]
 // CHECK:   ret void
 void test_vst1_f16(float16_t *a, float16x4_t b) {
   vst1_f16(a, b);
@@ -11056,9 +11054,9 @@ void test_vst2q_s64(int64_t *a, int64x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP6:%.*]] = bitcast <8 x half> [[TMP5]] to <16 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st2.v8i16.p0i8(<8 x i16> [[TMP7]], <8 x i16> [[TMP8]], i8* [[TMP2]])
+// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st2.v8f16.p0i8(<8 x half> [[TMP7]], <8 x half> [[TMP8]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst2q_f16(float16_t *a, float16x8x2_t b) {
   vst2q_f16(a, b);
@@ -11366,9 +11364,9 @@ void test_vst2_s64(int64_t *a, int64x1x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP6:%.*]] = bitcast <4 x half> [[TMP5]] to <8 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st2.v4i16.p0i8(<4 x i16> [[TMP7]], <4 x i16> [[TMP8]], i8* [[TMP2]])
+// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st2.v4f16.p0i8(<4 x half> [[TMP7]], <4 x half> [[TMP8]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst2_f16(float16_t *a, float16x4x2_t b) {
   vst2_f16(a, b);
@@ -11716,10 +11714,10 @@ void test_vst3q_s64(int64_t *a, int64x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP8:%.*]] = bitcast <8 x half> [[TMP7]] to <16 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st3.v8i16.p0i8(<8 x i16> [[TMP9]], <8 x i16> [[TMP10]], <8 x i16> [[TMP11]], i8* [[TMP2]])
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st3.v8f16.p0i8(<8 x half> [[TMP9]], <8 x half> [[TMP10]], <8 x half> [[TMP11]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst3q_f16(float16_t *a, float16x8x3_t b) {
   vst3q_f16(a, b);
@@ -12085,10 +12083,10 @@ void test_vst3_s64(int64_t *a, int64x1x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP8:%.*]] = bitcast <4 x half> [[TMP7]] to <8 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st3.v4i16.p0i8(<4 x i16> [[TMP9]], <4 x i16> [[TMP10]], <4 x i16> [[TMP11]], i8* [[TMP2]])
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st3.v4f16.p0i8(<4 x half> [[TMP9]], <4 x half> [[TMP10]], <4 x half> [[TMP11]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst3_f16(float16_t *a, float16x4x3_t b) {
   vst3_f16(a, b);
@@ -12494,11 +12492,11 @@ void test_vst4q_s64(int64_t *a, int64x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP10:%.*]] = bitcast <8 x half> [[TMP9]] to <16 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st4.v8i16.p0i8(<8 x i16> [[TMP11]], <8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], i8* [[TMP2]])
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st4.v8f16.p0i8(<8 x half> [[TMP11]], <8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst4q_f16(float16_t *a, float16x8x4_t b) {
   vst4q_f16(a, b);
@@ -12922,11 +12920,11 @@ void test_vst4_s64(int64_t *a, int64x1x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP10:%.*]] = bitcast <4 x half> [[TMP9]] to <8 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st4.v4i16.p0i8(<4 x i16> [[TMP11]], <4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], i8* [[TMP2]])
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st4.v4f16.p0i8(<4 x half> [[TMP11]], <4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst4_f16(float16_t *a, float16x4x4_t b) {
   vst4_f16(a, b);
@@ -13208,10 +13206,10 @@ int64x2x2_t test_vld1q_s64_x2(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x2_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld1x2.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16> } [[VLD1XN]], { <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <8 x half>, <8 x half> } @llvm.aarch64.neon.ld1x2.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half> } [[VLD1XN]], { <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 16, i1 false)
@@ -13454,10 +13452,10 @@ int64x1x2_t test_vld1_s64_x2(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x2_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld1x2.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16> } [[VLD1XN]], { <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <4 x half>, <4 x half> } @llvm.aarch64.neon.ld1x2.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half> } [[VLD1XN]], { <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 16, i32 8, i1 false)
@@ -13700,10 +13698,10 @@ int64x2x3_t test_vld1q_s64_x3(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x3_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld1x3.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16> } [[VLD1XN]], { <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld1x3.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half> } [[VLD1XN]], { <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 48, i32 16, i1 false)
@@ -13946,10 +13944,10 @@ int64x1x3_t test_vld1_s64_x3(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x3_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld1x3.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16> } [[VLD1XN]], { <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld1x3.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half> } [[VLD1XN]], { <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 24, i32 8, i1 false)
@@ -14192,10 +14190,10 @@ int64x2x4_t test_vld1q_s64_x4(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x4_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld1x4.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } [[VLD1XN]], { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld1x4.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half>, <8 x half> } [[VLD1XN]], { <8 x half>, <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 64, i32 16, i1 false)
@@ -14438,10 +14436,10 @@ int64x1x4_t test_vld1_s64_x4(int64_t const *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x4_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD1XN:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld1x4.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } [[VLD1XN]], { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD1XN:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld1x4.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half>, <4 x half> } [[VLD1XN]], { <4 x half>, <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 8, i1 false)
@@ -14752,10 +14750,10 @@ void test_vst1q_s64_x2(int64_t *a, int64x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP6:%.*]] = bitcast <8 x half> [[TMP5]] to <16 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x2.v8i16.p0i16(<8 x i16> [[TMP7]], <8 x i16> [[TMP8]], i16* [[TMP9]])
+// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x2.v8f16.p0f16(<8 x half> [[TMP7]], <8 x half> [[TMP8]], half* [[TMP9]])
 // CHECK:   ret void
 void test_vst1q_f16_x2(float16_t *a, float16x8x2_t b) {
   vst1q_f16_x2(a, b);
@@ -15098,10 +15096,10 @@ void test_vst1_s64_x2(int64_t *a, int64x1x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP6:%.*]] = bitcast <4 x half> [[TMP5]] to <8 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x2.v4i16.p0i16(<4 x i16> [[TMP7]], <4 x i16> [[TMP8]], i16* [[TMP9]])
+// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x2.v4f16.p0f16(<4 x half> [[TMP7]], <4 x half> [[TMP8]], half* [[TMP9]])
 // CHECK:   ret void
 void test_vst1_f16_x2(float16_t *a, float16x4x2_t b) {
   vst1_f16_x2(a, b);
@@ -15484,11 +15482,11 @@ void test_vst1q_s64_x3(int64_t *a, int64x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP8:%.*]] = bitcast <8 x half> [[TMP7]] to <16 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x3.v8i16.p0i16(<8 x i16> [[TMP9]], <8 x i16> [[TMP10]], <8 x i16> [[TMP11]], i16* [[TMP12]])
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x3.v8f16.p0f16(<8 x half> [[TMP9]], <8 x half> [[TMP10]], <8 x half> [[TMP11]], half* [[TMP12]])
 // CHECK:   ret void
 void test_vst1q_f16_x3(float16_t *a, float16x8x3_t b) {
   vst1q_f16_x3(a, b);
@@ -15894,11 +15892,11 @@ void test_vst1_s64_x3(int64_t *a, int64x1x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP8:%.*]] = bitcast <4 x half> [[TMP7]] to <8 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x3.v4i16.p0i16(<4 x i16> [[TMP9]], <4 x i16> [[TMP10]], <4 x i16> [[TMP11]], i16* [[TMP12]])
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x3.v4f16.p0f16(<4 x half> [[TMP9]], <4 x half> [[TMP10]], <4 x half> [[TMP11]], half* [[TMP12]])
 // CHECK:   ret void
 void test_vst1_f16_x3(float16_t *a, float16x4x3_t b) {
   vst1_f16_x3(a, b);
@@ -16344,12 +16342,12 @@ void test_vst1q_s64_x4(int64_t *a, int64x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP10:%.*]] = bitcast <8 x half> [[TMP9]] to <16 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x4.v8i16.p0i16(<8 x i16> [[TMP11]], <8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], i16* [[TMP15]])
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x4.v8f16.p0f16(<8 x half> [[TMP11]], <8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], half* [[TMP15]])
 // CHECK:   ret void
 void test_vst1q_f16_x4(float16_t *a, float16x8x4_t b) {
   vst1q_f16_x4(a, b);
@@ -16818,12 +16816,12 @@ void test_vst1_s64_x4(int64_t *a, int64x1x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP10:%.*]] = bitcast <4 x half> [[TMP9]] to <8 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast i8* [[TMP2]] to i16*
-// CHECK:   call void @llvm.aarch64.neon.st1x4.v4i16.p0i16(<4 x i16> [[TMP11]], <4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], i16* [[TMP15]])
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast i8* [[TMP2]] to half*
+// CHECK:   call void @llvm.aarch64.neon.st1x4.v4f16.p0f16(<4 x half> [[TMP11]], <4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], half* [[TMP15]])
 // CHECK:   ret void
 void test_vst1_f16_x4(float16_t *a, float16x4x4_t b) {
   vst1_f16_x4(a, b);
diff --git a/test/CodeGen/aarch64-neon-ldst-one.c b/test/CodeGen/aarch64-neon-ldst-one.c
index 9bd9ab1cb61b..a3c5b140a0d2 100644
--- a/test/CodeGen/aarch64-neon-ldst-one.c
+++ b/test/CodeGen/aarch64-neon-ldst-one.c
@@ -90,12 +90,11 @@ int64x2_t test_vld1q_dup_s64(int64_t  *a) {
 
 // CHECK-LABEL: define <8 x half> @test_vld1q_dup_f16(half* %a) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP2:%.*]] = load i16, i16* [[TMP1]]
-// CHECK:   [[TMP3:%.*]] = insertelement <8 x i16> undef, i16 [[TMP2]], i32 0
-// CHECK:   [[LANE:%.*]] = shufflevector <8 x i16> [[TMP3]], <8 x i16> [[TMP3]], <8 x i32> zeroinitializer
-// CHECK:   [[TMP4:%.*]] = bitcast <8 x i16> [[LANE]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP4]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP2:%.*]] = load half, half* [[TMP1]]
+// CHECK:   [[TMP3:%.*]] = insertelement <8 x half> undef, half [[TMP2]], i32 0
+// CHECK:   [[LANE:%.*]] = shufflevector <8 x half> [[TMP3]], <8 x half> [[TMP3]], <8 x i32> zeroinitializer
+// CHECK:   ret <8 x half> [[LANE]]
 float16x8_t test_vld1q_dup_f16(float16_t  *a) {
   return vld1q_dup_f16(a);
 }
@@ -239,12 +238,11 @@ int64x1_t test_vld1_dup_s64(int64_t  *a) {
 
 // CHECK-LABEL: define <4 x half> @test_vld1_dup_f16(half* %a) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP2:%.*]] = load i16, i16* [[TMP1]]
-// CHECK:   [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[TMP2]], i32 0
-// CHECK:   [[LANE:%.*]] = shufflevector <4 x i16> [[TMP3]], <4 x i16> [[TMP3]], <4 x i32> zeroinitializer
-// CHECK:   [[TMP4:%.*]] = bitcast <4 x i16> [[LANE]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP4]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP2:%.*]] = load half, half* [[TMP1]]
+// CHECK:   [[TMP3:%.*]] = insertelement <4 x half> undef, half [[TMP2]], i32 0
+// CHECK:   [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <4 x i32> zeroinitializer
+// CHECK:   ret <4 x half> [[LANE]]
 float16x4_t test_vld1_dup_f16(float16_t  *a) {
   return vld1_dup_f16(a);
 }
@@ -447,10 +445,10 @@ int64x2x2_t test_vld2q_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x2_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD2:%.*]] = call { <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld2r.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16> } [[VLD2]], { <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD2:%.*]] = call { <8 x half>, <8 x half> } @llvm.aarch64.neon.ld2r.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half> } [[VLD2]], { <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 16, i1 false)
@@ -693,10 +691,10 @@ int64x1x2_t test_vld2_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x2_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD2:%.*]] = call { <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld2r.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16> } [[VLD2]], { <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD2:%.*]] = call { <4 x half>, <4 x half> } @llvm.aarch64.neon.ld2r.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half> } [[VLD2]], { <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 16, i32 8, i1 false)
@@ -947,10 +945,10 @@ int64x2x3_t test_vld3q_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x3_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD3:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld3r.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16> } [[VLD3]], { <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD3:%.*]] = call { <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld3r.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half> } [[VLD3]], { <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 48, i32 16, i1 false)
@@ -1207,10 +1205,10 @@ int64x1x3_t test_vld3_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x3_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD3:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld3r.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16> } [[VLD3]], { <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD3:%.*]] = call { <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld3r.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half> } [[VLD3]], { <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 24, i32 8, i1 false)
@@ -1459,10 +1457,10 @@ int64x2x4_t test_vld4q_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x4_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD4:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld4r.v8i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } [[VLD4]], { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD4:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld4r.v8f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <8 x half>, <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half>, <8 x half> } [[VLD4]], { <8 x half>, <8 x half>, <8 x half>, <8 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x8x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 64, i32 16, i1 false)
@@ -1705,10 +1703,10 @@ int64x1x4_t test_vld4_dup_s64(int64_t  *a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x4_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to i16*
-// CHECK:   [[VLD4:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld4r.v4i16.p0i16(i16* [[TMP2]])
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } [[VLD4]], { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP3]]
+// CHECK:   [[TMP2:%.*]] = bitcast i8* [[TMP1]] to half*
+// CHECK:   [[VLD4:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld4r.v4f16.p0f16(half* [[TMP2]])
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <4 x half>, <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half>, <4 x half> } [[VLD4]], { <4 x half>, <4 x half>, <4 x half>, <4 x half> }* [[TMP3]]
 // CHECK:   [[TMP4:%.*]] = bitcast %struct.float16x4x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 8, i1 false)
@@ -1897,12 +1895,11 @@ int64x2_t test_vld1q_lane_s64(int64_t  *a, int64x2_t b) {
 // CHECK-LABEL: define <8 x half> @test_vld1q_lane_f16(half* %a, <8 x half> %b) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP4:%.*]] = load i16, i16* [[TMP3]]
-// CHECK:   [[VLD1_LANE:%.*]] = insertelement <8 x i16> [[TMP2]], i16 [[TMP4]], i32 7
-// CHECK:   [[TMP5:%.*]] = bitcast <8 x i16> [[VLD1_LANE]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP5]]
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP4:%.*]] = load half, half* [[TMP3]]
+// CHECK:   [[VLD1_LANE:%.*]] = insertelement <8 x half> [[TMP2]], half [[TMP4]], i32 7
+// CHECK:   ret <8 x half> [[VLD1_LANE]]
 float16x8_t test_vld1q_lane_f16(float16_t  *a, float16x8_t b) {
   return vld1q_lane_f16(a, b, 7);
 }
@@ -2054,12 +2051,11 @@ int64x1_t test_vld1_lane_s64(int64_t  *a, int64x1_t b) {
 // CHECK-LABEL: define <4 x half> @test_vld1_lane_f16(half* %a, <4 x half> %b) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP4:%.*]] = load i16, i16* [[TMP3]]
-// CHECK:   [[VLD1_LANE:%.*]] = insertelement <4 x i16> [[TMP2]], i16 [[TMP4]], i32 3
-// CHECK:   [[TMP5:%.*]] = bitcast <4 x i16> [[VLD1_LANE]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP5]]
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP4:%.*]] = load half, half* [[TMP3]]
+// CHECK:   [[VLD1_LANE:%.*]] = insertelement <4 x half> [[TMP2]], half [[TMP4]], i32 3
+// CHECK:   ret <4 x half> [[VLD1_LANE]]
 float16x4_t test_vld1_lane_f16(float16_t  *a, float16x4_t b) {
   return vld1_lane_f16(a, b, 3);
 }
@@ -2495,11 +2491,11 @@ int64x2x2_t test_vld2q_lane_s64(int64_t  *a, int64x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP6:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP7:%.*]] = bitcast <8 x half> [[TMP6]] to <16 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[VLD2_LANE:%.*]] = call { <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld2lane.v8i16.p0i8(<8 x i16> [[TMP8]], <8 x i16> [[TMP9]], i64 7, i8* [[TMP3]])
-// CHECK:   [[TMP10:%.*]] = bitcast i8* [[TMP2]] to { <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16> } [[VLD2_LANE]], { <8 x i16>, <8 x i16> }* [[TMP10]]
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[VLD2_LANE:%.*]] = call { <8 x half>, <8 x half> } @llvm.aarch64.neon.ld2lane.v8f16.p0i8(<8 x half> [[TMP8]], <8 x half> [[TMP9]], i64 7, i8* [[TMP3]])
+// CHECK:   [[TMP10:%.*]] = bitcast i8* [[TMP2]] to { <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half> } [[VLD2_LANE]], { <8 x half>, <8 x half> }* [[TMP10]]
 // CHECK:   [[TMP11:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP12:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP11]], i8* [[TMP12]], i64 32, i32 16, i1 false)
@@ -2927,11 +2923,11 @@ int64x1x2_t test_vld2_lane_s64(int64_t  *a, int64x1x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP6:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP7:%.*]] = bitcast <4 x half> [[TMP6]] to <8 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[VLD2_LANE:%.*]] = call { <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld2lane.v4i16.p0i8(<4 x i16> [[TMP8]], <4 x i16> [[TMP9]], i64 3, i8* [[TMP3]])
-// CHECK:   [[TMP10:%.*]] = bitcast i8* [[TMP2]] to { <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16> } [[VLD2_LANE]], { <4 x i16>, <4 x i16> }* [[TMP10]]
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[VLD2_LANE:%.*]] = call { <4 x half>, <4 x half> } @llvm.aarch64.neon.ld2lane.v4f16.p0i8(<4 x half> [[TMP8]], <4 x half> [[TMP9]], i64 3, i8* [[TMP3]])
+// CHECK:   [[TMP10:%.*]] = bitcast i8* [[TMP2]] to { <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half> } [[VLD2_LANE]], { <4 x half>, <4 x half> }* [[TMP10]]
 // CHECK:   [[TMP11:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP12:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP11]], i8* [[TMP12]], i64 16, i32 8, i1 false)
@@ -3364,12 +3360,12 @@ int64x2x3_t test_vld3q_lane_s64(int64_t  *a, int64x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP8:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP9:%.*]] = bitcast <8 x half> [[TMP8]] to <16 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   [[VLD3_LANE:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld3lane.v8i16.p0i8(<8 x i16> [[TMP10]], <8 x i16> [[TMP11]], <8 x i16> [[TMP12]], i64 7, i8* [[TMP3]])
-// CHECK:   [[TMP13:%.*]] = bitcast i8* [[TMP2]] to { <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16> } [[VLD3_LANE]], { <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP13]]
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   [[VLD3_LANE:%.*]] = call { <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld3lane.v8f16.p0i8(<8 x half> [[TMP10]], <8 x half> [[TMP11]], <8 x half> [[TMP12]], i64 7, i8* [[TMP3]])
+// CHECK:   [[TMP13:%.*]] = bitcast i8* [[TMP2]] to { <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half> } [[VLD3_LANE]], { <8 x half>, <8 x half>, <8 x half> }* [[TMP13]]
 // CHECK:   [[TMP14:%.*]] = bitcast %struct.float16x8x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP15:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP14]], i8* [[TMP15]], i64 48, i32 16, i1 false)
@@ -3889,12 +3885,12 @@ int64x1x3_t test_vld3_lane_s64(int64_t  *a, int64x1x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP8:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP9:%.*]] = bitcast <4 x half> [[TMP8]] to <8 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   [[VLD3_LANE:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld3lane.v4i16.p0i8(<4 x i16> [[TMP10]], <4 x i16> [[TMP11]], <4 x i16> [[TMP12]], i64 3, i8* [[TMP3]])
-// CHECK:   [[TMP13:%.*]] = bitcast i8* [[TMP2]] to { <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16> } [[VLD3_LANE]], { <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP13]]
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   [[VLD3_LANE:%.*]] = call { <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld3lane.v4f16.p0i8(<4 x half> [[TMP10]], <4 x half> [[TMP11]], <4 x half> [[TMP12]], i64 3, i8* [[TMP3]])
+// CHECK:   [[TMP13:%.*]] = bitcast i8* [[TMP2]] to { <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half> } [[VLD3_LANE]], { <4 x half>, <4 x half>, <4 x half> }* [[TMP13]]
 // CHECK:   [[TMP14:%.*]] = bitcast %struct.float16x4x3_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP15:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP14]], i8* [[TMP15]], i64 24, i32 8, i1 false)
@@ -4454,13 +4450,13 @@ int64x2x4_t test_vld4q_lane_s64(int64_t  *a, int64x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP10:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP11:%.*]] = bitcast <8 x half> [[TMP10]] to <16 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x i16>
-// CHECK:   [[VLD4_LANE:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.aarch64.neon.ld4lane.v8i16.p0i8(<8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], <8 x i16> [[TMP15]], i64 7, i8* [[TMP3]])
-// CHECK:   [[TMP16:%.*]] = bitcast i8* [[TMP2]] to { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }*
-// CHECK:   store { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } [[VLD4_LANE]], { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> }* [[TMP16]]
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x half>
+// CHECK:   [[VLD4_LANE:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half> } @llvm.aarch64.neon.ld4lane.v8f16.p0i8(<8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], <8 x half> [[TMP15]], i64 7, i8* [[TMP3]])
+// CHECK:   [[TMP16:%.*]] = bitcast i8* [[TMP2]] to { <8 x half>, <8 x half>, <8 x half>, <8 x half> }*
+// CHECK:   store { <8 x half>, <8 x half>, <8 x half>, <8 x half> } [[VLD4_LANE]], { <8 x half>, <8 x half>, <8 x half>, <8 x half> }* [[TMP16]]
 // CHECK:   [[TMP17:%.*]] = bitcast %struct.float16x8x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP18:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP17]], i8* [[TMP18]], i64 64, i32 16, i1 false)
@@ -5043,13 +5039,13 @@ int64x1x4_t test_vld4_lane_s64(int64_t  *a, int64x1x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP10:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP11:%.*]] = bitcast <4 x half> [[TMP10]] to <8 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x i16>
-// CHECK:   [[VLD4_LANE:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.aarch64.neon.ld4lane.v4i16.p0i8(<4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], <4 x i16> [[TMP15]], i64 3, i8* [[TMP3]])
-// CHECK:   [[TMP16:%.*]] = bitcast i8* [[TMP2]] to { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }*
-// CHECK:   store { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } [[VLD4_LANE]], { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> }* [[TMP16]]
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x half>
+// CHECK:   [[VLD4_LANE:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half> } @llvm.aarch64.neon.ld4lane.v4f16.p0i8(<4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], <4 x half> [[TMP15]], i64 3, i8* [[TMP3]])
+// CHECK:   [[TMP16:%.*]] = bitcast i8* [[TMP2]] to { <4 x half>, <4 x half>, <4 x half>, <4 x half> }*
+// CHECK:   store { <4 x half>, <4 x half>, <4 x half>, <4 x half> } [[VLD4_LANE]], { <4 x half>, <4 x half>, <4 x half>, <4 x half> }* [[TMP16]]
 // CHECK:   [[TMP17:%.*]] = bitcast %struct.float16x4x4_t* [[RETVAL]] to i8*
 // CHECK:   [[TMP18:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP17]], i8* [[TMP18]], i64 32, i32 8, i1 false)
@@ -5361,10 +5357,10 @@ void test_vst1q_lane_s64(int64_t  *a, int64x2_t b) {
 // CHECK-LABEL: define void @test_vst1q_lane_f16(half* %a, <8 x half> %b) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   [[TMP3:%.*]] = extractelement <8 x i16> [[TMP2]], i32 7
-// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   store i16 [[TMP3]], i16* [[TMP4]]
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   [[TMP3:%.*]] = extractelement <8 x half> [[TMP2]], i32 7
+// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   store half [[TMP3]], half* [[TMP4]]
 // CHECK:   ret void
 void test_vst1q_lane_f16(float16_t  *a, float16x8_t b) {
   vst1q_lane_f16(a, b, 7);
@@ -5517,10 +5513,10 @@ void test_vst1_lane_s64(int64_t  *a, int64x1_t b) {
 // CHECK-LABEL: define void @test_vst1_lane_f16(half* %a, <4 x half> %b) #0 {
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   [[TMP3:%.*]] = extractelement <4 x i16> [[TMP2]], i32 3
-// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   store i16 [[TMP3]], i16* [[TMP4]]
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   [[TMP3:%.*]] = extractelement <4 x half> [[TMP2]], i32 3
+// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   store half [[TMP3]], half* [[TMP4]]
 // CHECK:   ret void
 void test_vst1_lane_f16(float16_t  *a, float16x4_t b) {
   vst1_lane_f16(a, b, 3);
@@ -5789,9 +5785,9 @@ void test_vst2q_lane_s64(int64_t  *a, int64x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP6:%.*]] = bitcast <8 x half> [[TMP5]] to <16 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st2lane.v8i16.p0i8(<8 x i16> [[TMP7]], <8 x i16> [[TMP8]], i64 7, i8* [[TMP2]])
+// CHECK:   [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st2lane.v8f16.p0i8(<8 x half> [[TMP7]], <8 x half> [[TMP8]], i64 7, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst2q_lane_f16(float16_t  *a, float16x8x2_t b) {
   vst2q_lane_f16(a, b, 7);
@@ -6124,9 +6120,9 @@ void test_vst2_lane_s64(int64_t  *a, int64x1x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i64 0, i64 1
 // CHECK:   [[TMP5:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP6:%.*]] = bitcast <4 x half> [[TMP5]] to <8 x i8>
-// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st2lane.v4i16.p0i8(<4 x i16> [[TMP7]], <4 x i16> [[TMP8]], i64 3, i8* [[TMP2]])
+// CHECK:   [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st2lane.v4f16.p0i8(<4 x half> [[TMP7]], <4 x half> [[TMP8]], i64 3, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst2_lane_f16(float16_t  *a, float16x4x2_t b) {
   vst2_lane_f16(a, b, 3);
@@ -6499,10 +6495,10 @@ void test_vst3q_lane_s64(int64_t  *a, int64x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP8:%.*]] = bitcast <8 x half> [[TMP7]] to <16 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st3lane.v8i16.p0i8(<8 x i16> [[TMP9]], <8 x i16> [[TMP10]], <8 x i16> [[TMP11]], i64 7, i8* [[TMP2]])
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st3lane.v8f16.p0i8(<8 x half> [[TMP9]], <8 x half> [[TMP10]], <8 x half> [[TMP11]], i64 7, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst3q_lane_f16(float16_t  *a, float16x8x3_t b) {
   vst3q_lane_f16(a, b, 7);
@@ -6898,10 +6894,10 @@ void test_vst3_lane_s64(int64_t  *a, int64x1x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i64 0, i64 2
 // CHECK:   [[TMP7:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP8:%.*]] = bitcast <4 x half> [[TMP7]] to <8 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st3lane.v4i16.p0i8(<4 x i16> [[TMP9]], <4 x i16> [[TMP10]], <4 x i16> [[TMP11]], i64 3, i8* [[TMP2]])
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st3lane.v4f16.p0i8(<4 x half> [[TMP9]], <4 x half> [[TMP10]], <4 x half> [[TMP11]], i64 3, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst3_lane_f16(float16_t  *a, float16x4x3_t b) {
   vst3_lane_f16(a, b, 3);
@@ -7337,11 +7333,11 @@ void test_vst4q_lane_s64(int64_t  *a, int64x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP10:%.*]] = bitcast <8 x half> [[TMP9]] to <16 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st4lane.v8i16.p0i8(<8 x i16> [[TMP11]], <8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], i64 7, i8* [[TMP2]])
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x half>
+// CHECK:   call void @llvm.aarch64.neon.st4lane.v8f16.p0i8(<8 x half> [[TMP11]], <8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], i64 7, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst4q_lane_f16(float16_t  *a, float16x8x4_t b) {
   vst4q_lane_f16(a, b, 7);
@@ -7800,11 +7796,11 @@ void test_vst4_lane_s64(int64_t  *a, int64x1x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i64 0, i64 3
 // CHECK:   [[TMP9:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP10:%.*]] = bitcast <4 x half> [[TMP9]] to <8 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x i16>
-// CHECK:   call void @llvm.aarch64.neon.st4lane.v4i16.p0i8(<4 x i16> [[TMP11]], <4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], i64 3, i8* [[TMP2]])
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x half>
+// CHECK:   call void @llvm.aarch64.neon.st4lane.v4f16.p0i8(<4 x half> [[TMP11]], <4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], i64 3, i8* [[TMP2]])
 // CHECK:   ret void
 void test_vst4_lane_f16(float16_t  *a, float16x4x4_t b) {
   vst4_lane_f16(a, b, 3);
diff --git a/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c b/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
new file mode 100644
index 000000000000..3f61238b64fb
--- /dev/null
+++ b/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
@@ -0,0 +1,1633 @@
+// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -target-feature +fullfp16 -target-feature +v8.2a\
+// RUN: -fallow-half-arguments-and-returns -S -disable-O0-optnone -emit-llvm -o - %s \
+// RUN: | opt -S -mem2reg \
+// RUN: | FileCheck %s
+
+// REQUIRES: aarch64-registered-target
+
+#include <arm_neon.h>
+
+// CHECK-LABEL: test_vabs_f16
+// CHECK:  [[ABS:%.*]] =  call <4 x half> @llvm.fabs.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[ABS]]
+float16x4_t test_vabs_f16(float16x4_t a) {
+  return vabs_f16(a);
+}
+
+// CHECK-LABEL: test_vabsq_f16
+// CHECK:  [[ABS:%.*]] = call <8 x half> @llvm.fabs.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[ABS]]
+float16x8_t test_vabsq_f16(float16x8_t a) {
+  return vabsq_f16(a);
+}
+
+// CHECK-LABEL: test_vceqz_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oeq <4 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vceqz_f16(float16x4_t a) {
+  return vceqz_f16(a);
+}
+
+// CHECK-LABEL: test_vceqzq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oeq <8 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vceqzq_f16(float16x8_t a) {
+  return vceqzq_f16(a);
+}
+
+// CHECK-LABEL: test_vcgez_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oge <4 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcgez_f16(float16x4_t a) {
+  return vcgez_f16(a);
+}
+
+// CHECK-LABEL: test_vcgezq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oge <8 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcgezq_f16(float16x8_t a) {
+  return vcgezq_f16(a);
+}
+
+// CHECK-LABEL: test_vcgtz_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ogt <4 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcgtz_f16(float16x4_t a) {
+  return vcgtz_f16(a);
+}
+
+// CHECK-LABEL: test_vcgtzq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ogt <8 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcgtzq_f16(float16x8_t a) {
+  return vcgtzq_f16(a);
+}
+
+// CHECK-LABEL: test_vclez_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ole <4 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vclez_f16(float16x4_t a) {
+  return vclez_f16(a);
+}
+
+// CHECK-LABEL: test_vclezq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ole <8 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vclezq_f16(float16x8_t a) {
+  return vclezq_f16(a);
+}
+
+// CHECK-LABEL: test_vcltz_f16
+// CHECK:  [[TMP1:%.*]] = fcmp olt <4 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcltz_f16(float16x4_t a) {
+  return vcltz_f16(a);
+}
+
+// CHECK-LABEL: test_vcltzq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp olt <8 x half> %a, zeroinitializer
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcltzq_f16(float16x8_t a) {
+  return vcltzq_f16(a);
+}
+
+// CHECK-LABEL: test_vcvt_f16_s16
+// CHECK:  [[VCVT:%.*]] = sitofp <4 x i16> %a to <4 x half>
+// CHECK:  ret <4 x half> [[VCVT]]
+float16x4_t test_vcvt_f16_s16 (int16x4_t a) {
+  return vcvt_f16_s16(a);
+}
+
+// CHECK-LABEL: test_vcvtq_f16_s16
+// CHECK:  [[VCVT:%.*]] = sitofp <8 x i16> %a to <8 x half>
+// CHECK:  ret <8 x half> [[VCVT]]
+float16x8_t test_vcvtq_f16_s16 (int16x8_t a) {
+  return vcvtq_f16_s16(a);
+}
+
+// CHECK-LABEL: test_vcvt_f16_u16
+// CHECK:  [[VCVT:%.*]] = uitofp <4 x i16> %a to <4 x half>
+// CHECK:  ret <4 x half> [[VCVT]]
+float16x4_t test_vcvt_f16_u16 (uint16x4_t a) {
+  return vcvt_f16_u16(a);
+}
+
+// CHECK-LABEL: test_vcvtq_f16_u16
+// CHECK:  [[VCVT:%.*]] = uitofp <8 x i16> %a to <8 x half>
+// CHECK:  ret <8 x half> [[VCVT]]
+float16x8_t test_vcvtq_f16_u16 (uint16x8_t a) {
+  return vcvtq_f16_u16(a);
+}
+
+// CHECK-LABEL: test_vcvt_s16_f16
+// CHECK:  [[VCVT:%.*]] = fptosi <4 x half> %a to <4 x i16>
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvt_s16_f16 (float16x4_t a) {
+  return vcvt_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtq_s16_f16
+// CHECK:  [[VCVT:%.*]] = fptosi <8 x half> %a to <8 x i16>
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtq_s16_f16 (float16x8_t a) {
+  return vcvtq_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvt_u16_f16
+// CHECK:  [[VCVT:%.*]] = fptoui <4 x half> %a to <4 x i16>
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvt_u16_f16 (float16x4_t a) {
+  return vcvt_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtq_u16_f16
+// CHECK:  [[VCVT:%.*]] = fptoui <8 x half> %a to <8 x i16>
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtq_u16_f16 (float16x8_t a) {
+  return vcvtq_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvta_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtas.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvta_s16_f16 (float16x4_t a) {
+  return vcvta_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtaq_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtas.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtaq_s16_f16 (float16x8_t a) {
+  return vcvtaq_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtm_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtms.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvtm_s16_f16 (float16x4_t a) {
+  return vcvtm_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtmq_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtms.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtmq_s16_f16 (float16x8_t a) {
+  return vcvtmq_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtm_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtmu.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+uint16x4_t test_vcvtm_u16_f16 (float16x4_t a) {
+  return vcvtm_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtmq_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtmu.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+uint16x8_t test_vcvtmq_u16_f16 (float16x8_t a) {
+  return vcvtmq_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtn_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtns.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvtn_s16_f16 (float16x4_t a) {
+  return vcvtn_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtnq_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtns.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtnq_s16_f16 (float16x8_t a) {
+  return vcvtnq_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtn_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtnu.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+uint16x4_t test_vcvtn_u16_f16 (float16x4_t a) {
+  return vcvtn_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtnq_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtnu.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+uint16x8_t test_vcvtnq_u16_f16 (float16x8_t a) {
+  return vcvtnq_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtp_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtps.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+int16x4_t test_vcvtp_s16_f16 (float16x4_t a) {
+  return vcvtp_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtpq_s16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtps.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+int16x8_t test_vcvtpq_s16_f16 (float16x8_t a) {
+  return vcvtpq_s16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtp_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtpu.v4i16.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x i16> [[VCVT]]
+uint16x4_t test_vcvtp_u16_f16 (float16x4_t a) {
+  return vcvtp_u16_f16(a);
+}
+
+// CHECK-LABEL: test_vcvtpq_u16_f16
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtpu.v8i16.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x i16> [[VCVT]]
+uint16x8_t test_vcvtpq_u16_f16 (float16x8_t a) {
+  return vcvtpq_u16_f16(a);
+}
+
+// FIXME: Fix the zero constant when fp16 non-storage-only type becomes available.
+// CHECK-LABEL: test_vneg_f16
+// CHECK:  [[NEG:%.*]] = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %a
+// CHECK:  ret <4 x half> [[NEG]]
+float16x4_t test_vneg_f16(float16x4_t a) {
+  return vneg_f16(a);
+}
+
+// CHECK-LABEL: test_vnegq_f16
+// CHECK:  [[NEG:%.*]] = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %a
+// CHECK:  ret <8 x half> [[NEG]]
+float16x8_t test_vnegq_f16(float16x8_t a) {
+  return vnegq_f16(a);
+}
+
+// CHECK-LABEL: test_vrecpe_f16
+// CHECK:  [[RCP:%.*]] = call <4 x half> @llvm.aarch64.neon.frecpe.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RCP]]
+float16x4_t test_vrecpe_f16(float16x4_t a) {
+  return vrecpe_f16(a);
+}
+
+// CHECK-LABEL: test_vrecpeq_f16
+// CHECK:  [[RCP:%.*]] = call <8 x half> @llvm.aarch64.neon.frecpe.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RCP]]
+float16x8_t test_vrecpeq_f16(float16x8_t a) {
+  return vrecpeq_f16(a);
+}
+
+// CHECK-LABEL: test_vrnd_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.trunc.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrnd_f16(float16x4_t a) {
+  return vrnd_f16(a);
+}
+
+// CHECK-LABEL: test_vrndq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.trunc.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndq_f16(float16x8_t a) {
+  return vrndq_f16(a);
+}
+
+// CHECK-LABEL: test_vrnda_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.round.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrnda_f16(float16x4_t a) {
+  return vrnda_f16(a);
+}
+
+// CHECK-LABEL: test_vrndaq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.round.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndaq_f16(float16x8_t a) {
+  return vrndaq_f16(a);
+}
+
+// CHECK-LABEL: test_vrndi_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.nearbyint.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrndi_f16(float16x4_t a) {
+  return vrndi_f16(a);
+}
+
+// CHECK-LABEL: test_vrndiq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.nearbyint.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndiq_f16(float16x8_t a) {
+  return vrndiq_f16(a);
+}
+
+// CHECK-LABEL: test_vrndm_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.floor.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrndm_f16(float16x4_t a) {
+  return vrndm_f16(a);
+}
+
+// CHECK-LABEL: test_vrndmq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.floor.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndmq_f16(float16x8_t a) {
+  return vrndmq_f16(a);
+}
+
+// CHECK-LABEL: test_vrndn_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.aarch64.neon.frintn.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrndn_f16(float16x4_t a) {
+  return vrndn_f16(a);
+}
+
+// CHECK-LABEL: test_vrndnq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.aarch64.neon.frintn.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndnq_f16(float16x8_t a) {
+  return vrndnq_f16(a);
+}
+
+// CHECK-LABEL: test_vrndp_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.ceil.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrndp_f16(float16x4_t a) {
+  return vrndp_f16(a);
+}
+
+// CHECK-LABEL: test_vrndpq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.ceil.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndpq_f16(float16x8_t a) {
+  return vrndpq_f16(a);
+}
+
+// CHECK-LABEL: test_vrndx_f16
+// CHECK:  [[RND:%.*]] =  call <4 x half> @llvm.rint.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrndx_f16(float16x4_t a) {
+  return vrndx_f16(a);
+}
+
+// CHECK-LABEL: test_vrndxq_f16
+// CHECK:  [[RND:%.*]] =  call <8 x half> @llvm.rint.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrndxq_f16(float16x8_t a) {
+  return vrndxq_f16(a);
+}
+
+// CHECK-LABEL: test_vrsqrte_f16
+// CHECK:  [[RND:%.*]] = call <4 x half> @llvm.aarch64.neon.frsqrte.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[RND]]
+float16x4_t test_vrsqrte_f16(float16x4_t a) {
+  return vrsqrte_f16(a);
+}
+
+// CHECK-LABEL: test_vrsqrteq_f16
+// CHECK:  [[RND:%.*]] = call <8 x half> @llvm.aarch64.neon.frsqrte.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[RND]]
+float16x8_t test_vrsqrteq_f16(float16x8_t a) {
+  return vrsqrteq_f16(a);
+}
+
+// CHECK-LABEL: test_vsqrt_f16
+// CHECK:  [[SQR:%.*]] = call <4 x half> @llvm.sqrt.v4f16(<4 x half> %a)
+// CHECK:  ret <4 x half> [[SQR]]
+float16x4_t test_vsqrt_f16(float16x4_t a) {
+  return vsqrt_f16(a);
+}
+
+// CHECK-LABEL: test_vsqrtq_f16
+// CHECK:  [[SQR:%.*]] = call <8 x half> @llvm.sqrt.v8f16(<8 x half> %a)
+// CHECK:  ret <8 x half> [[SQR]]
+float16x8_t test_vsqrtq_f16(float16x8_t a) {
+  return vsqrtq_f16(a);
+}
+
+// CHECK-LABEL: test_vadd_f16
+// CHECK:  [[ADD:%.*]] = fadd <4 x half> %a, %b
+// CHECK:  ret <4 x half> [[ADD]]
+float16x4_t test_vadd_f16(float16x4_t a, float16x4_t b) {
+  return vadd_f16(a, b);
+}
+
+// CHECK-LABEL: test_vaddq_f16
+// CHECK:  [[ADD:%.*]] = fadd <8 x half> %a, %b
+// CHECK:  ret <8 x half> [[ADD]]
+float16x8_t test_vaddq_f16(float16x8_t a, float16x8_t b) {
+  return vaddq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vabd_f16
+// CHECK:  [[ABD:%.*]] = call <4 x half> @llvm.aarch64.neon.fabd.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[ABD]]
+float16x4_t test_vabd_f16(float16x4_t a, float16x4_t b) {
+  return vabd_f16(a, b);
+}
+
+// CHECK-LABEL: test_vabdq_f16
+// CHECK:  [[ABD:%.*]] = call <8 x half> @llvm.aarch64.neon.fabd.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[ABD]]
+float16x8_t test_vabdq_f16(float16x8_t a, float16x8_t b) {
+  return vabdq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcage_f16
+// CHECK:  [[ABS:%.*]] = call <4 x i16> @llvm.aarch64.neon.facge.v4i16.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x i16> [[ABS]]
+uint16x4_t test_vcage_f16(float16x4_t a, float16x4_t b) {
+  return vcage_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcageq_f16
+// CHECK:  [[ABS:%.*]] = call <8 x i16> @llvm.aarch64.neon.facge.v8i16.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x i16> [[ABS]]
+uint16x8_t test_vcageq_f16(float16x8_t a, float16x8_t b) {
+  return vcageq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcagt_f16
+// CHECK:  [[ABS:%.*]] = call <4 x i16> @llvm.aarch64.neon.facgt.v4i16.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x i16> [[ABS]]
+uint16x4_t test_vcagt_f16(float16x4_t a, float16x4_t b) {
+  return vcagt_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcagtq_f16
+// CHECK:  [[ABS:%.*]] = call <8 x i16> @llvm.aarch64.neon.facgt.v8i16.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x i16> [[ABS]]
+uint16x8_t test_vcagtq_f16(float16x8_t a, float16x8_t b) {
+  return vcagtq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcale_f16
+// CHECK:  [[ABS:%.*]] = call <4 x i16> @llvm.aarch64.neon.facge.v4i16.v4f16(<4 x half> %b, <4 x half> %a)
+// CHECK:  ret <4 x i16> [[ABS]]
+uint16x4_t test_vcale_f16(float16x4_t a, float16x4_t b) {
+  return vcale_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcaleq_f16
+// CHECK:  [[ABS:%.*]] = call <8 x i16> @llvm.aarch64.neon.facge.v8i16.v8f16(<8 x half> %b, <8 x half> %a)
+// CHECK:  ret <8 x i16> [[ABS]]
+uint16x8_t test_vcaleq_f16(float16x8_t a, float16x8_t b) {
+  return vcaleq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcalt_f16
+// CHECK:  [[ABS:%.*]] = call <4 x i16> @llvm.aarch64.neon.facgt.v4i16.v4f16(<4 x half> %b, <4 x half> %a)
+// CHECK:  ret <4 x i16> [[ABS]]
+uint16x4_t test_vcalt_f16(float16x4_t a, float16x4_t b) {
+  return vcalt_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcaltq_f16
+// CHECK:  [[ABS:%.*]] = call <8 x i16> @llvm.aarch64.neon.facgt.v8i16.v8f16(<8 x half> %b, <8 x half> %a)
+// CHECK:  ret <8 x i16> [[ABS]]
+uint16x8_t test_vcaltq_f16(float16x8_t a, float16x8_t b) {
+  return vcaltq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vceq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oeq <4 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vceq_f16(float16x4_t a, float16x4_t b) {
+  return vceq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vceqq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oeq <8 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vceqq_f16(float16x8_t a, float16x8_t b) {
+  return vceqq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcge_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oge <4 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcge_f16(float16x4_t a, float16x4_t b) {
+  return vcge_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcgeq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp oge <8 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcgeq_f16(float16x8_t a, float16x8_t b) {
+  return vcgeq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcgt_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ogt <4 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcgt_f16(float16x4_t a, float16x4_t b) {
+  return vcgt_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcgtq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ogt <8 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcgtq_f16(float16x8_t a, float16x8_t b) {
+  return vcgtq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcle_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ole <4 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vcle_f16(float16x4_t a, float16x4_t b) {
+  return vcle_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcleq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp ole <8 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcleq_f16(float16x8_t a, float16x8_t b) {
+  return vcleq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vclt_f16
+// CHECK:  [[TMP1:%.*]] = fcmp olt <4 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <4 x i1> [[TMP1]] to <4 x i16>
+// CHECK:  ret <4 x i16> [[TMP2]]
+uint16x4_t test_vclt_f16(float16x4_t a, float16x4_t b) {
+  return vclt_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcltq_f16
+// CHECK:  [[TMP1:%.*]] = fcmp olt <8 x half> %a, %b
+// CHECK:  [[TMP2:%.*]] = sext <8 x i1> [[TMP1:%.*]] to <8 x i16>
+// CHECK:  ret <8 x i16> [[TMP2]]
+uint16x8_t test_vcltq_f16(float16x8_t a, float16x8_t b) {
+  return vcltq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vcvt_n_f16_s16
+// CHECK:  [[CVT:%.*]] = call <4 x half> @llvm.aarch64.neon.vcvtfxs2fp.v4f16.v4i16(<4 x i16> %vcvt_n, i32 2)
+// CHECK:  ret <4 x half> [[CVT]]
+float16x4_t test_vcvt_n_f16_s16(int16x4_t a) {
+  return vcvt_n_f16_s16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvtq_n_f16_s16
+// CHECK:  [[CVT:%.*]] = call <8 x half> @llvm.aarch64.neon.vcvtfxs2fp.v8f16.v8i16(<8 x i16> %vcvt_n, i32 2)
+// CHECK:  ret <8 x half> [[CVT]]
+float16x8_t test_vcvtq_n_f16_s16(int16x8_t a) {
+  return vcvtq_n_f16_s16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvt_n_f16_u16
+// CHECK:  [[CVT:%.*]] = call <4 x half> @llvm.aarch64.neon.vcvtfxu2fp.v4f16.v4i16(<4 x i16> %vcvt_n, i32 2)
+// CHECK:  ret <4 x half> [[CVT]]
+float16x4_t test_vcvt_n_f16_u16(uint16x4_t a) {
+  return vcvt_n_f16_u16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvtq_n_f16_u16
+// CHECK:  [[CVT:%.*]] = call <8 x half> @llvm.aarch64.neon.vcvtfxu2fp.v8f16.v8i16(<8 x i16> %vcvt_n, i32 2)
+// CHECK:  ret <8 x half> [[CVT]]
+float16x8_t test_vcvtq_n_f16_u16(uint16x8_t a) {
+  return vcvtq_n_f16_u16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvt_n_s16_f16
+// CHECK:  [[CVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.vcvtfp2fxs.v4i16.v4f16(<4 x half> %vcvt_n, i32 2)
+// CHECK:  ret <4 x i16> [[CVT]]
+int16x4_t test_vcvt_n_s16_f16(float16x4_t a) {
+  return vcvt_n_s16_f16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvtq_n_s16_f16
+// CHECK:  [[CVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.vcvtfp2fxs.v8i16.v8f16(<8 x half> %vcvt_n, i32 2)
+// CHECK:  ret <8 x i16> [[CVT]]
+int16x8_t test_vcvtq_n_s16_f16(float16x8_t a) {
+  return vcvtq_n_s16_f16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvt_n_u16_f16
+// CHECK:  [[CVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.vcvtfp2fxu.v4i16.v4f16(<4 x half> %vcvt_n, i32 2)
+// CHECK:  ret <4 x i16> [[CVT]]
+uint16x4_t test_vcvt_n_u16_f16(float16x4_t a) {
+  return vcvt_n_u16_f16(a, 2);
+}
+
+// CHECK-LABEL: test_vcvtq_n_u16_f16
+// CHECK:  [[CVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.vcvtfp2fxu.v8i16.v8f16(<8 x half> %vcvt_n, i32 2)
+// CHECK:  ret <8 x i16> [[CVT]]
+uint16x8_t test_vcvtq_n_u16_f16(float16x8_t a) {
+  return vcvtq_n_u16_f16(a, 2);
+}
+
+// CHECK-LABEL: test_vdiv_f16
+// CHECK:  [[DIV:%.*]] = fdiv <4 x half> %a, %b
+// CHECK:  ret <4 x half> [[DIV]]
+float16x4_t test_vdiv_f16(float16x4_t a, float16x4_t b) {
+  return vdiv_f16(a, b);
+}
+
+// CHECK-LABEL: test_vdivq_f16
+// CHECK:  [[DIV:%.*]] = fdiv <8 x half> %a, %b
+// CHECK:  ret <8 x half> [[DIV]]
+float16x8_t test_vdivq_f16(float16x8_t a, float16x8_t b) {
+  return vdivq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmax_f16
+// CHECK:  [[MAX:%.*]] = call <4 x half> @llvm.aarch64.neon.fmax.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MAX]]
+float16x4_t test_vmax_f16(float16x4_t a, float16x4_t b) {
+  return vmax_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmaxq_f16
+// CHECK:  [[MAX:%.*]] = call <8 x half> @llvm.aarch64.neon.fmax.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MAX]]
+float16x8_t test_vmaxq_f16(float16x8_t a, float16x8_t b) {
+  return vmaxq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmaxnm_f16
+// CHECK:  [[MAX:%.*]] = call <4 x half> @llvm.aarch64.neon.fmaxnm.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MAX]]
+float16x4_t test_vmaxnm_f16(float16x4_t a, float16x4_t b) {
+  return vmaxnm_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmaxnmq_f16
+// CHECK:  [[MAX:%.*]] = call <8 x half> @llvm.aarch64.neon.fmaxnm.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MAX]]
+float16x8_t test_vmaxnmq_f16(float16x8_t a, float16x8_t b) {
+  return vmaxnmq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmin_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.fmin.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vmin_f16(float16x4_t a, float16x4_t b) {
+  return vmin_f16(a, b);
+}
+
+// CHECK-LABEL: test_vminq_f16
+// CHECK:  [[MIN:%.*]] = call <8 x half> @llvm.aarch64.neon.fmin.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vminq_f16(float16x8_t a, float16x8_t b) {
+  return vminq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vminnm_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.fminnm.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vminnm_f16(float16x4_t a, float16x4_t b) {
+  return vminnm_f16(a, b);
+}
+
+// CHECK-LABEL: test_vminnmq_f16
+// CHECK:  [[MIN:%.*]] = call <8 x half> @llvm.aarch64.neon.fminnm.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vminnmq_f16(float16x8_t a, float16x8_t b) {
+  return vminnmq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmul_f16
+// CHECK:  [[MUL:%.*]] = fmul <4 x half> %a, %b
+// CHECK:  ret <4 x half> [[MUL]]
+float16x4_t test_vmul_f16(float16x4_t a, float16x4_t b) {
+  return vmul_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmulq_f16
+// CHECK:  [[MUL:%.*]] = fmul <8 x half> %a, %b
+// CHECK:  ret <8 x half> [[MUL]]
+float16x8_t test_vmulq_f16(float16x8_t a, float16x8_t b) {
+  return vmulq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmulx_f16
+// CHECK:  [[MUL:%.*]] = call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MUL]]
+float16x4_t test_vmulx_f16(float16x4_t a, float16x4_t b) {
+  return vmulx_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmulxq_f16
+// CHECK:  [[MUL:%.*]] = call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MUL]]
+float16x8_t test_vmulxq_f16(float16x8_t a, float16x8_t b) {
+  return vmulxq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpadd_f16
+// CHECK:  [[ADD:%.*]] = call <4 x half> @llvm.aarch64.neon.addp.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[ADD]]
+float16x4_t test_vpadd_f16(float16x4_t a, float16x4_t b) {
+  return vpadd_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpaddq_f16
+// CHECK:  [[ADD:%.*]] = call <8 x half> @llvm.aarch64.neon.addp.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[ADD]]
+float16x8_t test_vpaddq_f16(float16x8_t a, float16x8_t b) {
+  return vpaddq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpmax_f16
+// CHECK:  [[MAX:%.*]] = call <4 x half> @llvm.aarch64.neon.fmaxp.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MAX]]
+float16x4_t test_vpmax_f16(float16x4_t a, float16x4_t b) {
+  return vpmax_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpmaxq_f16
+// CHECK:  [[MAX:%.*]] = call <8 x half> @llvm.aarch64.neon.fmaxp.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MAX]]
+float16x8_t test_vpmaxq_f16(float16x8_t a, float16x8_t b) {
+  return vpmaxq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpmaxnm_f16
+// CHECK:  [[MAX:%.*]] = call <4 x half> @llvm.aarch64.neon.fmaxnmp.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MAX]]
+float16x4_t test_vpmaxnm_f16(float16x4_t a, float16x4_t b) {
+  return vpmaxnm_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpmaxnmq_f16
+// CHECK:  [[MAX:%.*]] = call <8 x half> @llvm.aarch64.neon.fmaxnmp.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MAX]]
+float16x8_t test_vpmaxnmq_f16(float16x8_t a, float16x8_t b) {
+  return vpmaxnmq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpmin_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.fminp.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vpmin_f16(float16x4_t a, float16x4_t b) {
+  return vpmin_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpminq_f16
+// CHECK:  [[MIN:%.*]] = call <8 x half> @llvm.aarch64.neon.fminp.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vpminq_f16(float16x8_t a, float16x8_t b) {
+  return vpminq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpminnm_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.fminnmp.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vpminnm_f16(float16x4_t a, float16x4_t b) {
+  return vpminnm_f16(a, b);
+}
+
+// CHECK-LABEL: test_vpminnmq_f16
+// CHECK:  [[MIN:%.*]] = call <8 x half> @llvm.aarch64.neon.fminnmp.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vpminnmq_f16(float16x8_t a, float16x8_t b) {
+  return vpminnmq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vrecps_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.frecps.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vrecps_f16(float16x4_t a, float16x4_t b) {
+  return vrecps_f16(a, b);
+}
+
+// CHECK-LABEL: test_vrecpsq_f16
+// CHECK:  [[MIN:%.*]] =  call <8 x half> @llvm.aarch64.neon.frecps.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vrecpsq_f16(float16x8_t a, float16x8_t b) {
+  return vrecpsq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vrsqrts_f16
+// CHECK:  [[MIN:%.*]] = call <4 x half> @llvm.aarch64.neon.frsqrts.v4f16(<4 x half> %a, <4 x half> %b)
+// CHECK:  ret <4 x half> [[MIN]]
+float16x4_t test_vrsqrts_f16(float16x4_t a, float16x4_t b) {
+  return vrsqrts_f16(a, b);
+}
+
+// CHECK-LABEL: test_vrsqrtsq_f16
+// CHECK:  [[MIN:%.*]] =  call <8 x half> @llvm.aarch64.neon.frsqrts.v8f16(<8 x half> %a, <8 x half> %b)
+// CHECK:  ret <8 x half> [[MIN]]
+float16x8_t test_vrsqrtsq_f16(float16x8_t a, float16x8_t b) {
+  return vrsqrtsq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vsub_f16
+// CHECK:  [[ADD:%.*]] = fsub <4 x half> %a, %b 
+// CHECK:  ret <4 x half> [[ADD]]
+float16x4_t test_vsub_f16(float16x4_t a, float16x4_t b) {
+  return vsub_f16(a, b);
+}
+
+// CHECK-LABEL: test_vsubq_f16
+// CHECK:  [[ADD:%.*]] = fsub <8 x half> %a, %b
+// CHECK:  ret <8 x half> [[ADD]]
+float16x8_t test_vsubq_f16(float16x8_t a, float16x8_t b) {
+  return vsubq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vfma_f16
+// CHECK:  [[ADD:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> %b, <4 x half> %c, <4 x half> %a)
+// CHECK:  ret <4 x half> [[ADD]]
+float16x4_t test_vfma_f16(float16x4_t a, float16x4_t b, float16x4_t c) {
+  return vfma_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmaq_f16
+// CHECK:  [[ADD:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %c, <8 x half> %a)
+// CHECK:  ret <8 x half> [[ADD]]
+float16x8_t test_vfmaq_f16(float16x8_t a, float16x8_t b, float16x8_t c) {
+  return vfmaq_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfms_f16
+// CHECK:  [[SUB:%.*]] = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK:  [[ADD:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> [[SUB]], <4 x half> %c, <4 x half> %a)
+// CHECK:  ret <4 x half> [[ADD]]
+float16x4_t test_vfms_f16(float16x4_t a, float16x4_t b, float16x4_t c) {
+  return vfms_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmsq_f16
+// CHECK:  [[SUB:%.*]] = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK:  [[ADD:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> [[SUB]], <8 x half> %c, <8 x half> %a)
+// CHECK:  ret <8 x half> [[ADD]]
+float16x8_t test_vfmsq_f16(float16x8_t a, float16x8_t b, float16x8_t c) {
+  return vfmsq_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfma_lane_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[FMLA:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> [[TMP4]], <4 x half> [[LANE]], <4 x half> [[TMP5]])
+// CHECK: ret <4 x half> [[FMLA]]
+float16x4_t test_vfma_lane_f16(float16x4_t a, float16x4_t b, float16x4_t c) {
+  return vfma_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfmaq_lane_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[TMP4:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[FMLA:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> [[TMP4]], <8 x half> [[LANE]], <8 x half> [[TMP5]])
+// CHECK: ret <8 x half> [[FMLA]]
+float16x8_t test_vfmaq_lane_f16(float16x8_t a, float16x8_t b, float16x4_t c) {
+  return vfmaq_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfma_laneq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <8 x half> [[TMP5]], <8 x half> [[TMP5]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[FMLA:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> [[LANE]], <4 x half> [[TMP4]], <4 x half> [[TMP3]])
+// CHECK: ret <4 x half> [[FMLA]]
+float16x4_t test_vfma_laneq_f16(float16x4_t a, float16x4_t b, float16x8_t c) {
+  return vfma_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vfmaq_laneq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[TMP4:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <8 x half> [[TMP5]], <8 x half> [[TMP5]], <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[FMLA:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> [[LANE]], <8 x half> [[TMP4]], <8 x half> [[TMP3]])
+// CHECK: ret <8 x half> [[FMLA]]
+float16x8_t test_vfmaq_laneq_f16(float16x8_t a, float16x8_t b, float16x8_t c) {
+  return vfmaq_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vfma_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <4 x half> undef, half %c, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %c, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %c, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %c, i32 3
+// CHECK: [[FMA:%.*]]  = call <4 x half> @llvm.fma.v4f16(<4 x half> %b, <4 x half> [[TMP3]], <4 x half> %a)
+// CHECK: ret <4 x half> [[FMA]]
+float16x4_t test_vfma_n_f16(float16x4_t a, float16x4_t b, float16_t c) {
+  return vfma_n_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmaq_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <8 x half> undef, half %c, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %c, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %c, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %c, i32 3
+// CHECK: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %c, i32 4
+// CHECK: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %c, i32 5
+// CHECK: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %c, i32 6
+// CHECK: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %c, i32 7
+// CHECK: [[FMA:%.*]]  = call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> [[TMP7]], <8 x half> %a)
+// CHECK: ret <8 x half> [[FMA]]
+float16x8_t test_vfmaq_n_f16(float16x8_t a, float16x8_t b, float16_t c) {
+  return vfmaq_n_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmah_lane_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[EXTR:%.*]] = extractelement <4 x half> [[TMP1]], i32 3
+// CHECK: [[FMA:%.*]]  = call half @llvm.fma.f16(half %b, half [[EXTR]], half %a)
+// CHECK: ret half [[FMA]]
+float16_t test_vfmah_lane_f16(float16_t a, float16_t b, float16x4_t c) {
+  return vfmah_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfmah_laneq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[EXTR:%.*]] = extractelement <8 x half> [[TMP1]], i32 7
+// CHECK: [[FMA:%.*]]  = call half @llvm.fma.f16(half %b, half [[EXTR]], half %a)
+// CHECK: ret half [[FMA]]
+float16_t test_vfmah_laneq_f16(float16_t a, float16_t b, float16x8_t c) {
+  return vfmah_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vfms_lane_f16
+// CHECK: [[SUB:%.*]]  = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x half> [[SUB]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[FMA:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> [[TMP4]], <4 x half> [[LANE]], <4 x half> [[TMP5]])
+// CHECK: ret <4 x half> [[FMA]]
+float16x4_t test_vfms_lane_f16(float16x4_t a, float16x4_t b, float16x4_t c) {
+  return vfms_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfmsq_lane_f16
+// CHECK: [[SUB:%.*]]  = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x half> [[SUB]] to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[TMP4:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[FMLA:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> [[TMP4]], <8 x half> [[LANE]], <8 x half> [[TMP5]])
+// CHECK: ret <8 x half> [[FMLA]]
+float16x8_t test_vfmsq_lane_f16(float16x8_t a, float16x8_t b, float16x4_t c) {
+  return vfmsq_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfms_laneq_f16
+// CHECK: [[SUB:%.*]]  = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x half> [[SUB]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <8 x half> [[TMP5]], <8 x half> [[TMP5]], <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[FMLA:%.*]] = call <4 x half> @llvm.fma.v4f16(<4 x half> [[LANE]], <4 x half> [[TMP4]], <4 x half> [[TMP3]])
+// CHECK: ret <4 x half> [[FMLA]]
+float16x4_t test_vfms_laneq_f16(float16x4_t a, float16x4_t b, float16x8_t c) {
+  return vfms_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vfmsq_laneq_f16
+// CHECK: [[SUB:%.*]]  = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x half> [[SUB]] to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[TMP4:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x half>
+// CHECK: [[LANE:%.*]] = shufflevector <8 x half> [[TMP5]], <8 x half> [[TMP5]], <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[FMLA:%.*]] = call <8 x half> @llvm.fma.v8f16(<8 x half> [[LANE]], <8 x half> [[TMP4]], <8 x half> [[TMP3]])
+// CHECK: ret <8 x half> [[FMLA]]
+float16x8_t test_vfmsq_laneq_f16(float16x8_t a, float16x8_t b, float16x8_t c) {
+  return vfmsq_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vfms_n_f16
+// CHECK: [[SUB:%.*]]  = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = insertelement <4 x half> undef, half %c, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %c, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %c, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %c, i32 3
+// CHECK: [[FMA:%.*]]  = call <4 x half> @llvm.fma.v4f16(<4 x half> [[SUB]], <4 x half> [[TMP3]], <4 x half> %a)
+// CHECK: ret <4 x half> [[FMA]]
+float16x4_t test_vfms_n_f16(float16x4_t a, float16x4_t b, float16_t c) {
+  return vfms_n_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmsq_n_f16
+// CHECK: [[SUB:%.*]]  = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+// CHECK: [[TMP0:%.*]] = insertelement <8 x half> undef, half %c, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %c, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %c, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %c, i32 3
+// CHECK: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %c, i32 4
+// CHECK: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %c, i32 5
+// CHECK: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %c, i32 6
+// CHECK: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %c, i32 7
+// CHECK: [[FMA:%.*]]  = call <8 x half> @llvm.fma.v8f16(<8 x half> [[SUB]], <8 x half> [[TMP7]], <8 x half> %a)
+// CHECK: ret <8 x half> [[FMA]]
+float16x8_t test_vfmsq_n_f16(float16x8_t a, float16x8_t b, float16_t c) {
+  return vfmsq_n_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vfmsh_lane_f16
+// CHECK: [[TMP0:%.*]] = fpext half %b to float
+// CHECK: [[TMP1:%.*]] = fsub float -0.000000e+00, [[TMP0]]
+// CHECK: [[SUB:%.*]]  = fptrunc float [[TMP1]] to half
+// CHECK: [[TMP2:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x half>
+// CHECK: [[EXTR:%.*]] = extractelement <4 x half> [[TMP3]], i32 3
+// CHECK: [[FMA:%.*]]  = call half @llvm.fma.f16(half [[SUB]], half [[EXTR]], half %a)
+// CHECK: ret half [[FMA]]
+float16_t test_vfmsh_lane_f16(float16_t a, float16_t b, float16x4_t c) {
+  return vfmsh_lane_f16(a, b, c, 3);
+}
+
+// CHECK-LABEL: test_vfmsh_laneq_f16
+// CHECK: [[TMP0:%.*]] = fpext half %b to float
+// CHECK: [[TMP1:%.*]] = fsub float -0.000000e+00, [[TMP0]]
+// CHECK: [[SUB:%.*]]  = fptrunc float [[TMP1]] to half
+// CHECK: [[TMP2:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x half>
+// CHECK: [[EXTR:%.*]] = extractelement <8 x half> [[TMP3]], i32 7
+// CHECK: [[FMA:%.*]]  = call half @llvm.fma.f16(half [[SUB]], half [[EXTR]], half %a)
+// CHECK: ret half [[FMA]]
+float16_t test_vfmsh_laneq_f16(float16_t a, float16_t b, float16x8_t c) {
+  return vfmsh_laneq_f16(a, b, c, 7);
+}
+
+// CHECK-LABEL: test_vmul_lane_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <4 x half> %b, <4 x half> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[MUL:%.*]]  = fmul <4 x half> %a, [[TMP0]]
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmul_lane_f16(float16x4_t a, float16x4_t b) {
+  return vmul_lane_f16(a, b, 3);
+}
+
+// CHECK-LABEL: test_vmulq_lane_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <4 x half> %b, <4 x half> %b, <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]]  = fmul <8 x half> %a, [[TMP0]]
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulq_lane_f16(float16x8_t a, float16x4_t b) {
+  return vmulq_lane_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmul_laneq_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <8 x half> %b, <8 x half> %b, <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]]  = fmul <4 x half> %a, [[TMP0]]
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmul_laneq_f16(float16x4_t a, float16x8_t b) {
+  return vmul_laneq_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmulq_laneq_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <8 x half> %b, <8 x half> %b, <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]]  = fmul <8 x half> %a, [[TMP0]]
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulq_laneq_f16(float16x8_t a, float16x8_t b) {
+  return vmulq_laneq_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmul_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <4 x half> undef, half %b, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %b, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %b, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %b, i32 3
+// CHECK: [[MUL:%.*]]  = fmul <4 x half> %a, [[TMP3]]
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmul_n_f16(float16x4_t a, float16_t b) {
+  return vmul_n_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmulq_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <8 x half> undef, half %b, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %b, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %b, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %b, i32 3
+// CHECK: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %b, i32 4
+// CHECK: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %b, i32 5
+// CHECK: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %b, i32 6
+// CHECK: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %b, i32 7
+// CHECK: [[MUL:%.*]]  = fmul <8 x half> %a, [[TMP7]]
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulq_n_f16(float16x8_t a, float16_t b) {
+  return vmulq_n_f16(a, b);
+}
+
+// FIXME: Fix it when fp16 non-storage-only type becomes available.
+// CHECK-LABEL: test_vmulh_lane_f16
+// CHECK: [[CONV0:%.*]] = fpext half %a to float
+// CHECK: [[CONV1:%.*]] = fpext half %{{.*}} to float
+// CHECK: [[MUL:%.*]]   = fmul float [[CONV0:%.*]], [[CONV0:%.*]]
+// CHECK: [[CONV3:%.*]] = fptrunc float %mul to half
+// CHECK: ret half [[CONV3:%.*]]
+float16_t test_vmulh_lane_f16(float16_t a, float16x4_t b) {
+  return vmulh_lane_f16(a, b, 3);
+}
+
+// CHECK-LABEL: test_vmulh_laneq_f16
+// CHECK: [[CONV0:%.*]] = fpext half %a to float
+// CHECK: [[CONV1:%.*]] = fpext half %{{.*}} to float
+// CHECK: [[MUL:%.*]]   = fmul float [[CONV0:%.*]], [[CONV0:%.*]]
+// CHECK: [[CONV3:%.*]] = fptrunc float %mul to half
+// CHECK: ret half [[CONV3:%.*]]
+float16_t test_vmulh_laneq_f16(float16_t a, float16x8_t b) {
+  return vmulh_laneq_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmulx_lane_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <4 x half> %b, <4 x half> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+// CHECK: [[MUL:%.*]] = call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> [[TMP0]])
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmulx_lane_f16(float16x4_t a, float16x4_t b) {
+  return vmulx_lane_f16(a, b, 3);
+}
+
+// CHECK-LABEL: test_vmulxq_lane_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <4 x half> %b, <4 x half> %b, <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]] = call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> [[TMP0]])
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulxq_lane_f16(float16x8_t a, float16x4_t b) {
+  return vmulxq_lane_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmulx_laneq_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <8 x half> %b, <8 x half> %b, <4 x i32> <i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]]  = call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> [[TMP0]])
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmulx_laneq_f16(float16x4_t a, float16x8_t b) {
+  return vmulx_laneq_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmulxq_laneq_f16
+// CHECK: [[TMP0:%.*]] = shufflevector <8 x half> %b, <8 x half> %b, <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK: [[MUL:%.*]]  = call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> [[TMP0]])
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulxq_laneq_f16(float16x8_t a, float16x8_t b) {
+  return vmulxq_laneq_f16(a, b, 7);
+}
+
+// CHECK-LABEL: test_vmulx_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <4 x half> undef, half %b, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %b, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %b, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %b, i32 3
+// CHECK: [[MUL:%.*]]  = call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> [[TMP3]])
+// CHECK: ret <4 x half> [[MUL]]
+float16x4_t test_vmulx_n_f16(float16x4_t a, float16_t b) {
+  return vmulx_n_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmulxq_n_f16
+// CHECK: [[TMP0:%.*]] = insertelement <8 x half> undef, half %b, i32 0
+// CHECK: [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %b, i32 1
+// CHECK: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %b, i32 2
+// CHECK: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %b, i32 3
+// CHECK: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %b, i32 4
+// CHECK: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %b, i32 5
+// CHECK: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %b, i32 6
+// CHECK: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %b, i32 7
+// CHECK: [[MUL:%.*]]  = call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> [[TMP7]])
+// CHECK: ret <8 x half> [[MUL]]
+float16x8_t test_vmulxq_n_f16(float16x8_t a, float16_t b) {
+  return vmulxq_n_f16(a, b);
+}
+
+/* TODO: Not implemented yet (needs scalar intrinsic from arm_fp16.h)
+// CCHECK-LABEL: test_vmulxh_lane_f16
+// CCHECK: [[CONV0:%.*]] = fpext half %a to float
+// CCHECK: [[CONV1:%.*]] = fpext half %{{.*}} to float
+// CCHECK: [[MUL:%.*]]   = fmul float [[CONV0:%.*]], [[CONV0:%.*]]
+// CCHECK: [[CONV3:%.*]] = fptrunc float %mul to half
+// CCHECK: ret half [[CONV3:%.*]]
+float16_t test_vmulxh_lane_f16(float16_t a, float16x4_t b) {
+  return vmulxh_lane_f16(a, b, 3);
+}
+
+// CCHECK-LABEL: test_vmulxh_laneq_f16
+// CCHECK: [[CONV0:%.*]] = fpext half %a to float
+// CCHECK: [[CONV1:%.*]] = fpext half %{{.*}} to float
+// CCHECK: [[MUL:%.*]]   = fmul float [[CONV0:%.*]], [[CONV0:%.*]]
+// CCHECK: [[CONV3:%.*]] = fptrunc float %mul to half
+// CCHECK: ret half [[CONV3:%.*]]
+float16_t test_vmulxh_laneq_f16(float16_t a, float16x8_t b) {
+  return vmulxh_laneq_f16(a, b, 7);
+}
+*/
+
+// CHECK-LABEL: test_vmaxv_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fmaxv.f16.v4f16(<4 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vmaxv_f16(float16x4_t a) {
+  return vmaxv_f16(a);
+}
+
+// CHECK-LABEL: test_vmaxvq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fmaxv.f16.v8f16(<8 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vmaxvq_f16(float16x8_t a) {
+  return vmaxvq_f16(a);
+}
+
+// CHECK-LABEL: test_vminv_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fminv.f16.v4f16(<4 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vminv_f16(float16x4_t a) {
+  return vminv_f16(a);
+}
+
+// CHECK-LABEL: test_vminvq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fminv.f16.v8f16(<8 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vminvq_f16(float16x8_t a) {
+  return vminvq_f16(a);
+}
+
+// CHECK-LABEL: test_vmaxnmv_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fmaxnmv.f16.v4f16(<4 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vmaxnmv_f16(float16x4_t a) {
+  return vmaxnmv_f16(a);
+}
+
+// CHECK-LABEL: test_vmaxnmvq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fmaxnmv.f16.v8f16(<8 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vmaxnmvq_f16(float16x8_t a) {
+  return vmaxnmvq_f16(a);
+}
+
+// CHECK-LABEL: test_vminnmv_f16
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fminnmv.f16.v4f16(<4 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vminnmv_f16(float16x4_t a) {
+  return vminnmv_f16(a);
+}
+
+// CHECK-LABEL: test_vminnmvq_f16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK: [[MAX:%.*]]  = call half @llvm.aarch64.neon.fminnmv.f16.v8f16(<8 x half> [[TMP1]])
+// CHECK: ret half [[MAX]]
+float16_t test_vminnmvq_f16(float16x8_t a) {
+  return vminnmvq_f16(a);
+}
+
+// CHECK-LABEL: test_vbsl_f16
+// CHECK:  [[TMP0:%.*]] = bitcast <4 x half> %b to <8 x i8>
+// CHECK:  [[TMP1:%.*]] = bitcast <4 x half> %c to <8 x i8>
+// CHECK:  [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK:  [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
+// CHECK:  [[TMP4:%.*]] = and <4 x i16> %a, [[TMP2]]
+// CHECK:  [[TMP5:%.*]] = xor <4 x i16> %a, <i16 -1, i16 -1, i16 -1, i16 -1>
+// CHECK:  [[TMP6:%.*]] = and <4 x i16> [[TMP5]], [[TMP3]]
+// CHECK:  [[TMP7:%.*]] = or <4 x i16> [[TMP4]], [[TMP6]]
+// CHECK:  [[TMP8:%.*]] = bitcast <4 x i16> [[TMP7]] to <4 x half>
+// CHECK:  ret <4 x half> [[TMP8]]
+float16x4_t test_vbsl_f16(uint16x4_t a, float16x4_t b, float16x4_t c) {
+  return vbsl_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vbslq_f16
+// CHECK:  [[TMP0:%.*]] = bitcast <8 x half> %b to <16 x i8>
+// CHECK:  [[TMP1:%.*]] = bitcast <8 x half> %c to <16 x i8>
+// CHECK:  [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK:  [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
+// CHECK:  [[TMP4:%.*]] = and <8 x i16> %a, [[TMP2]]
+// CHECK:  [[TMP5:%.*]] = xor <8 x i16> %a, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1>
+// CHECK:  [[TMP6:%.*]] = and <8 x i16> [[TMP5]], [[TMP3]]
+// CHECK:  [[TMP7:%.*]] = or <8 x i16> [[TMP4]], [[TMP6]]
+// CHECK:  [[TMP8:%.*]] = bitcast <8 x i16> [[TMP7]] to <8 x half>
+// CHECK:  ret <8 x half> [[TMP8]]
+float16x8_t test_vbslq_f16(uint16x8_t a, float16x8_t b, float16x8_t c) {
+  return vbslq_f16(a, b, c);
+}
+
+// CHECK-LABEL: test_vzip_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <4 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
+// CHECK:   store <4 x half> [[VZIP0_I]], <4 x half>* [[TMP1]]
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <4 x half>, <4 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
+// CHECK:   store <4 x half> [[VZIP1_I]], <4 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 16, i32 8, i1 false)
+float16x4x2_t test_vzip_f16(float16x4_t a, float16x4_t b) {
+  return vzip_f16(a, b);
+}
+
+// CHECK-LABEL: test_vzipq_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <8 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
+// CHECK:   store <8 x half> [[VZIP0_I]], <8 x half>* [[TMP1]]
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <8 x half>, <8 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
+// CHECK:   store <8 x half> [[VZIP1_I]], <8 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 32, i32 16, i1 false)
+float16x8x2_t test_vzipq_f16(float16x8_t a, float16x8_t b) {
+  return vzipq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vuzp_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <4 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+// CHECK:   store <4 x half> [[VZIP0_I]], <4 x half>* [[TMP1]]
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <4 x half>, <4 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+// CHECK:   store <4 x half> [[VZIP1_I]], <4 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 16, i32 8, i1 false)
+float16x4x2_t test_vuzp_f16(float16x4_t a, float16x4_t b) {
+  return vuzp_f16(a, b);
+}
+
+// CHECK-LABEL: test_vuzpq_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <8 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+// CHECK:   store <8 x half> [[VZIP0_I]], <8 x half>* [[TMP1]]
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <8 x half>, <8 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+// CHECK:   store <8 x half> [[VZIP1_I]], <8 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 32, i32 16, i1 false)
+float16x8x2_t test_vuzpq_f16(float16x8_t a, float16x8_t b) {
+  return vuzpq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vtrn_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x4x2_t, align 8
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <4 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+// CHECK:   store <4 x half> [[VZIP0_I]], <4 x half>* [[TMP1]]
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <4 x half>, <4 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+// CHECK:   store <4 x half> [[VZIP1_I]], <4 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x4x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x4x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 16, i32 8, i1 false)
+float16x4x2_t test_vtrn_f16(float16x4_t a, float16x4_t b) {
+  return vtrn_f16(a, b);
+}
+
+// CHECK-LABEL: test_vtrnq_f16
+// CHECK:   [[RETVAL:%.*]]  = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[__RET_I:%.*]] = alloca %struct.float16x8x2_t, align 16
+// CHECK:   [[TMP0:%.*]]  = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   [[TMP1:%.*]]  = bitcast i8* [[TMP0]] to <8 x half>*
+// CHECK:   [[VZIP0_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+// CHECK:   store <8 x half> [[VZIP0_I]], <8 x half>* [[TMP1]] 
+// CHECK:   [[TMP2:%.*]] = getelementptr inbounds <8 x half>, <8 x half>* [[TMP1]], i32 1
+// CHECK:   [[VZIP1_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32>  <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+// CHECK:   store <8 x half> [[VZIP1_I]], <8 x half>* [[TMP2]]
+// CHECK:   [[TMP5:%.*]] = bitcast %struct.float16x8x2_t* [[RETVAL]] to i8*
+// CHECK:   [[TMP6:%.*]] = bitcast %struct.float16x8x2_t* [[__RET_I]] to i8*
+// CHECK:   call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP5]], i8* [[TMP6]], i64 32, i32 16, i1 false)
+float16x8x2_t test_vtrnq_f16(float16x8_t a, float16x8_t b) {
+  return vtrnq_f16(a, b);
+}
+
+// CHECK-LABEL: test_vmov_n_f16
+// CHECK:   [[TMP0:%.*]] = insertelement <4 x half> undef, half %a, i32 0
+// CHECK:   [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %a, i32 1
+// CHECK:   [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %a, i32 2
+// CHECK:   [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %a, i32 3
+// CHECK:   ret <4 x half> [[TMP3]]
+float16x4_t test_vmov_n_f16(float16_t a) {
+  return vmov_n_f16(a);
+}
+
+// CHECK-LABEL: test_vmovq_n_f16
+// CHECK:   [[TMP0:%.*]] = insertelement <8 x half> undef, half %a, i32 0
+// CHECK:   [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %a, i32 1
+// CHECK:   [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %a, i32 2
+// CHECK:   [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %a, i32 3
+// CHECK:   [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %a, i32 4
+// CHECK:   [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %a, i32 5
+// CHECK:   [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %a, i32 6
+// CHECK:   [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %a, i32 7
+// CHECK:   ret <8 x half> [[TMP7]]
+float16x8_t test_vmovq_n_f16(float16_t a) {
+  return vmovq_n_f16(a);
+}
+
+// CHECK-LABEL: test_vdup_n_f16
+// CHECK:   [[TMP0:%.*]] = insertelement <4 x half> undef, half %a, i32 0
+// CHECK:   [[TMP1:%.*]] = insertelement <4 x half> [[TMP0]], half %a, i32 1
+// CHECK:   [[TMP2:%.*]] = insertelement <4 x half> [[TMP1]], half %a, i32 2
+// CHECK:   [[TMP3:%.*]] = insertelement <4 x half> [[TMP2]], half %a, i32 3
+// CHECK:   ret <4 x half> [[TMP3]]
+float16x4_t test_vdup_n_f16(float16_t a) {
+  return vdup_n_f16(a);
+}
+
+// CHECK-LABEL: test_vdupq_n_f16
+// CHECK:   [[TMP0:%.*]] = insertelement <8 x half> undef, half %a, i32 0
+// CHECK:   [[TMP1:%.*]] = insertelement <8 x half> [[TMP0]], half %a, i32 1
+// CHECK:   [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half %a, i32 2
+// CHECK:   [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half %a, i32 3
+// CHECK:   [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half %a, i32 4
+// CHECK:   [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half %a, i32 5
+// CHECK:   [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half %a, i32 6
+// CHECK:   [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half %a, i32 7
+// CHECK:   ret <8 x half> [[TMP7]]
+float16x8_t test_vdupq_n_f16(float16_t a) {
+  return vdupq_n_f16(a);
+}
+
+// CHECK-LABEL: test_vdup_lane_f16
+// CHECK:   [[SHFL:%.*]] = shufflevector <4 x half> %a, <4 x half> %a, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+// CHECK:   ret <4 x half> [[SHFL]]
+float16x4_t test_vdup_lane_f16(float16x4_t a) {
+  return vdup_lane_f16(a, 3);
+}
+
+// CHECK-LABEL: test_vdupq_lane_f16
+// CHECK:   [[SHFL:%.*]] = shufflevector <4 x half> %a, <4 x half> %a, <8 x i32> <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7>
+// CHECK:   ret <8 x half> [[SHFL]]
+float16x8_t test_vdupq_lane_f16(float16x4_t a) {
+  return vdupq_lane_f16(a, 7);
+}
+
+// CHECK-LABEL: @test_vext_f16(
+// CHECK:   [[TMP0:%.*]] = bitcast <4 x half> %a to <8 x i8>
+// CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   [[VEXT:%.*]] = shufflevector <4 x half> [[TMP2]], <4 x half> [[TMP3]], <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+// CHECK:   ret <4 x half> [[VEXT]]
+float16x4_t test_vext_f16(float16x4_t a, float16x4_t b) {
+  return vext_f16(a, b, 2);
+}
+
+// CHECK-LABEL: @test_vextq_f16(
+// CHECK:   [[TMP0:%.*]] = bitcast <8 x half> %a to <16 x i8>
+// CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   [[VEXT:%.*]] = shufflevector <8 x half> [[TMP2]], <8 x half> [[TMP3]], <8 x i32> <i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12>
+// CHECK:   ret <8 x half> [[VEXT]]
+float16x8_t test_vextq_f16(float16x8_t a, float16x8_t b) {
+  return vextq_f16(a, b, 5);
+}
+
+// CHECK-LABEL: @test_vrev64_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <4 x half> %a, <4 x half> %a, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+// CHECK:   ret <4 x half> [[SHFL]]
+float16x4_t test_vrev64_f16(float16x4_t a) {
+  return vrev64_f16(a);
+}
+
+// CHECK-LABEL: @test_vrev64q_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <8 x half> %a, <8 x half> %a, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
+// CHECK:   ret <8 x half> [[SHFL]]
+float16x8_t test_vrev64q_f16(float16x8_t a) {
+  return vrev64q_f16(a);
+}
+
+// CHECK-LABEL: @test_vzip1_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
+// CHECK:   ret <4 x half> [[SHFL]]
+float16x4_t test_vzip1_f16(float16x4_t a, float16x4_t b) {
+  return vzip1_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vzip1q_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
+// CHECK:   ret <8 x half> [[SHFL]]
+float16x8_t test_vzip1q_f16(float16x8_t a, float16x8_t b) {
+  return vzip1q_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vzip2_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
+// CHECK:   ret <4 x half> [[SHFL]]
+float16x4_t test_vzip2_f16(float16x4_t a, float16x4_t b) {
+  return vzip2_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vzip2q_f16(
+// CHECK:   [[SHFL:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
+// CHECK:   ret <8 x half> [[SHFL]]
+float16x8_t test_vzip2q_f16(float16x8_t a, float16x8_t b) {
+  return vzip2q_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vuzp1_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+// CHECK:   ret <4 x half> [[SHUFFLE_I]]
+float16x4_t test_vuzp1_f16(float16x4_t a, float16x4_t b) {
+  return vuzp1_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vuzp1q_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+// CHECK:   ret <8 x half> [[SHUFFLE_I]]
+float16x8_t test_vuzp1q_f16(float16x8_t a, float16x8_t b) {
+  return vuzp1q_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vuzp2_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+// CHECK:   ret <4 x half> [[SHUFFLE_I]]
+float16x4_t test_vuzp2_f16(float16x4_t a, float16x4_t b) {
+  return vuzp2_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vuzp2q_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+// CHECK:   ret <8 x half> [[SHUFFLE_I]]
+float16x8_t test_vuzp2q_f16(float16x8_t a, float16x8_t b) {
+  return vuzp2q_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vtrn1_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+// CHECK:   ret <4 x half> [[SHUFFLE_I]]
+float16x4_t test_vtrn1_f16(float16x4_t a, float16x4_t b) {
+  return vtrn1_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vtrn1q_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32>  <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+// CHECK:   ret <8 x half> [[SHUFFLE_I]]
+float16x8_t test_vtrn1q_f16(float16x8_t a, float16x8_t b) {
+  return vtrn1q_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vtrn2_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <4 x half> %a, <4 x half> %b, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+// CHECK:   ret <4 x half> [[SHUFFLE_I]]
+float16x4_t test_vtrn2_f16(float16x4_t a, float16x4_t b) {
+  return vtrn2_f16(a, b);
+}
+
+// CHECK-LABEL: @test_vtrn2q_f16(
+// CHECK:   [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %b, <8 x i32> <i32 1, i32 9, i32 3, i32 11, i32 5, i32 13, i32 7, i32 15>
+// CHECK:   ret <8 x half> [[SHUFFLE_I]]
+float16x8_t test_vtrn2q_f16(float16x8_t a, float16x8_t b) {
+  return vtrn2q_f16(a, b);
+}
+
diff --git a/test/CodeGen/address-space.c b/test/CodeGen/address-space.c
index 35e3dbdcfa73..54e059385772 100644
--- a/test/CodeGen/address-space.c
+++ b/test/CodeGen/address-space.c
@@ -1,6 +1,6 @@
-// RUN: %clang_cc1 -triple x86_64-apple-darwin -emit-llvm < %s | FileCheck -check-prefixes=CHECK,GIZ %s
+// RUN: %clang_cc1 -triple x86_64-apple-darwin -emit-llvm < %s | FileCheck -check-prefixes=CHECK,X86,GIZ %s
 // RUN: %clang_cc1 -triple amdgcn -emit-llvm < %s | FileCheck -check-prefixes=CHECK,PIZ %s
-// RUN: %clang_cc1 -triple amdgcn---amdgiz -emit-llvm < %s | FileCheck -check-prefixes=CHECK,GIZ %s
+// RUN: %clang_cc1 -triple amdgcn---amdgiz -emit-llvm < %s | FileCheck -check-prefixes=CHECK,AMDGIZ,GIZ %s
 
 // CHECK: @foo = common addrspace(1) global
 int foo __attribute__((address_space(1)));
diff --git a/test/CodeGen/arm_neon_intrinsics.c b/test/CodeGen/arm_neon_intrinsics.c
index 62888dd73339..b01c90c03a96 100644
--- a/test/CodeGen/arm_neon_intrinsics.c
+++ b/test/CodeGen/arm_neon_intrinsics.c
@@ -3896,9 +3896,8 @@ int64x2_t test_vld1q_s64(int64_t const * a) {
 
 // CHECK-LABEL: @test_vld1q_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD1:%.*]] = call <8 x i16> @llvm.arm.neon.vld1.v8i16.p0i8(i8* [[TMP0]], i32 2)
-// CHECK:   [[TMP1:%.*]] = bitcast <8 x i16> [[VLD1]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP1]]
+// CHECK:   [[VLD1:%.*]] = call <8 x half> @llvm.arm.neon.vld1.v8f16.p0i8(i8* [[TMP0]], i32 2)
+// CHECK:   ret <8 x half> [[VLD1]]
 float16x8_t test_vld1q_f16(float16_t const * a) {
   return vld1q_f16(a);
 }
@@ -3990,9 +3989,8 @@ int64x1_t test_vld1_s64(int64_t const * a) {
 
 // CHECK-LABEL: @test_vld1_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD1:%.*]] = call <4 x i16> @llvm.arm.neon.vld1.v4i16.p0i8(i8* [[TMP0]], i32 2)
-// CHECK:   [[TMP1:%.*]] = bitcast <4 x i16> [[VLD1]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP1]]
+// CHECK:   [[VLD1:%.*]] = call <4 x half> @llvm.arm.neon.vld1.v4f16.p0i8(i8* [[TMP0]], i32 2)
+// CHECK:   ret <4 x half> [[VLD1]]
 float16x4_t test_vld1_f16(float16_t const * a) {
   return vld1_f16(a);
 }
@@ -4106,12 +4104,11 @@ int64x2_t test_vld1q_dup_s64(int64_t const * a) {
 
 // CHECK-LABEL: @test_vld1q_dup_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP2:%.*]] = load i16, i16* [[TMP1]], align 2
-// CHECK:   [[TMP3:%.*]] = insertelement <8 x i16> undef, i16 [[TMP2]], i32 0
-// CHECK:   [[LANE:%.*]] = shufflevector <8 x i16> [[TMP3]], <8 x i16> [[TMP3]], <8 x i32> zeroinitializer
-// CHECK:   [[TMP4:%.*]] = bitcast <8 x i16> [[LANE]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP4]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP2:%.*]] = load half, half* [[TMP1]], align 2
+// CHECK:   [[TMP3:%.*]] = insertelement <8 x half> undef, half [[TMP2]], i32 0
+// CHECK:   [[LANE:%.*]] = shufflevector <8 x half> [[TMP3]], <8 x half> [[TMP3]], <8 x i32> zeroinitializer
+// CHECK:   ret <8 x half> [[LANE]]
 float16x8_t test_vld1q_dup_f16(float16_t const * a) {
   return vld1q_dup_f16(a);
 }
@@ -4233,12 +4230,11 @@ int64x1_t test_vld1_dup_s64(int64_t const * a) {
 
 // CHECK-LABEL: @test_vld1_dup_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP2:%.*]] = load i16, i16* [[TMP1]], align 2
-// CHECK:   [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[TMP2]], i32 0
-// CHECK:   [[LANE:%.*]] = shufflevector <4 x i16> [[TMP3]], <4 x i16> [[TMP3]], <4 x i32> zeroinitializer
-// CHECK:   [[TMP4:%.*]] = bitcast <4 x i16> [[LANE]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP4]]
+// CHECK:   [[TMP1:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP2:%.*]] = load half, half* [[TMP1]], align 2
+// CHECK:   [[TMP3:%.*]] = insertelement <4 x half> undef, half [[TMP2]], i32 0
+// CHECK:   [[LANE:%.*]] = shufflevector <4 x half> [[TMP3]], <4 x half> [[TMP3]], <4 x i32> zeroinitializer
+// CHECK:   ret <4 x half> [[LANE]]
 float16x4_t test_vld1_dup_f16(float16_t const * a) {
   return vld1_dup_f16(a);
 }
@@ -4365,12 +4361,11 @@ int64x2_t test_vld1q_lane_s64(int64_t const * a, int64x2_t b) {
 // CHECK-LABEL: @test_vld1q_lane_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP4:%.*]] = load i16, i16* [[TMP3]], align 2
-// CHECK:   [[VLD1_LANE:%.*]] = insertelement <8 x i16> [[TMP2]], i16 [[TMP4]], i32 7
-// CHECK:   [[TMP5:%.*]] = bitcast <8 x i16> [[VLD1_LANE]] to <8 x half>
-// CHECK:   ret <8 x half> [[TMP5]]
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP4:%.*]] = load half, half* [[TMP3]], align 2
+// CHECK:   [[VLD1_LANE:%.*]] = insertelement <8 x half> [[TMP2]], half [[TMP4]], i32 7
+// CHECK:   ret <8 x half> [[VLD1_LANE]]
 float16x8_t test_vld1q_lane_f16(float16_t const * a, float16x8_t b) {
   return vld1q_lane_f16(a, b, 7);
 }
@@ -4498,12 +4493,11 @@ int64x1_t test_vld1_lane_s64(int64_t const * a, int64x1_t b) {
 // CHECK-LABEL: @test_vld1_lane_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   [[TMP4:%.*]] = load i16, i16* [[TMP3]], align 2
-// CHECK:   [[VLD1_LANE:%.*]] = insertelement <4 x i16> [[TMP2]], i16 [[TMP4]], i32 3
-// CHECK:   [[TMP5:%.*]] = bitcast <4 x i16> [[VLD1_LANE]] to <4 x half>
-// CHECK:   ret <4 x half> [[TMP5]]
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   [[TMP3:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   [[TMP4:%.*]] = load half, half* [[TMP3]], align 2
+// CHECK:   [[VLD1_LANE:%.*]] = insertelement <4 x half> [[TMP2]], half [[TMP4]], i32 3
+// CHECK:   ret <4 x half> [[VLD1_LANE]]
 float16x4_t test_vld1_lane_f16(float16_t const * a, float16x4_t b) {
   return vld1_lane_f16(a, b, 3);
 }
@@ -4596,7 +4590,7 @@ int32x4x2_t test_vld2q_s32(int32_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x2_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD2Q_V:%.*]] = call { <8 x i16>, <8 x i16>
+// CHECK:   [[VLD2Q_V:%.*]] = call { <8 x half>, <8 x half>
 float16x8x2_t test_vld2q_f16(float16_t const * a) {
   return vld2q_f16(a);
 }
@@ -4701,7 +4695,7 @@ int64x1x2_t test_vld2_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x2_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD2_V:%.*]] = call { <4 x i16>, <4 x i16>
+// CHECK:   [[VLD2_V:%.*]] = call { <4 x half>, <4 x half>
 float16x4x2_t test_vld2_f16(float16_t const * a) {
   return vld2_f16(a);
 }
@@ -4806,7 +4800,7 @@ int64x1x2_t test_vld2_dup_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x2_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x2_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD_DUP:%.*]] = call { <4 x i16>, <4 x i16>
+// CHECK:   [[VLD_DUP:%.*]] = call { <4 x half>, <4 x half>
 float16x4x2_t test_vld2_dup_f16(float16_t const * a) {
   return vld2_dup_f16(a);
 }
@@ -4965,9 +4959,9 @@ int32x4x2_t test_vld2q_lane_s32(int32_t const * a, int32x4x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP7:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP8:%.*]] = bitcast <8 x half> [[TMP7]] to <16 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[VLD2Q_LANE_V:%.*]] = call { <8 x i16>, <8 x i16>
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[VLD2Q_LANE_V:%.*]] = call { <8 x half>, <8 x half>
 float16x8x2_t test_vld2q_lane_f16(float16_t const * a, float16x8x2_t b) {
   return vld2q_lane_f16(a, b, 7);
 }
@@ -5198,9 +5192,9 @@ int32x2x2_t test_vld2_lane_s32(int32_t const * a, int32x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP7:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP8:%.*]] = bitcast <4 x half> [[TMP7]] to <8 x i8>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[VLD2_LANE_V:%.*]] = call { <4 x i16>, <4 x i16>
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[VLD2_LANE_V:%.*]] = call { <4 x half>, <4 x half>
 float16x4x2_t test_vld2_lane_f16(float16_t const * a, float16x4x2_t b) {
   return vld2_lane_f16(a, b, 3);
 }
@@ -5337,7 +5331,7 @@ int32x4x3_t test_vld3q_s32(int32_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x3_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD3Q_V:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>
+// CHECK:   [[VLD3Q_V:%.*]] = call { <8 x half>, <8 x half>, <8 x half>
 float16x8x3_t test_vld3q_f16(float16_t const * a) {
   return vld3q_f16(a);
 }
@@ -5442,7 +5436,7 @@ int64x1x3_t test_vld3_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x3_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD3_V:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[VLD3_V:%.*]] = call { <4 x half>, <4 x half>, <4 x half>
 float16x4x3_t test_vld3_f16(float16_t const * a) {
   return vld3_f16(a);
 }
@@ -5547,7 +5541,7 @@ int64x1x3_t test_vld3_dup_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x3_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x3_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD_DUP:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[VLD_DUP:%.*]] = call { <4 x half>, <4 x half>, <4 x half>
 float16x4x3_t test_vld3_dup_f16(float16_t const * a) {
   return vld3_dup_f16(a);
 }
@@ -5730,10 +5724,10 @@ int32x4x3_t test_vld3q_lane_s32(int32_t const * a, int32x4x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP9:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP10:%.*]] = bitcast <8 x half> [[TMP9]] to <16 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x i16>
-// CHECK:   [[VLD3Q_LANE_V:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x half>
+// CHECK:   [[VLD3Q_LANE_V:%.*]] = call { <8 x half>, <8 x half>, <8 x half>
 float16x8x3_t test_vld3q_lane_f16(float16_t const * a, float16x8x3_t b) {
   return vld3q_lane_f16(a, b, 7);
 }
@@ -6004,10 +5998,10 @@ int32x2x3_t test_vld3_lane_s32(int32_t const * a, int32x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP9:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP10:%.*]] = bitcast <4 x half> [[TMP9]] to <8 x i8>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x i16>
-// CHECK:   [[VLD3_LANE_V:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x half>
+// CHECK:   [[VLD3_LANE_V:%.*]] = call { <4 x half>, <4 x half>, <4 x half>
 float16x4x3_t test_vld3_lane_f16(float16_t const * a, float16x4x3_t b) {
   return vld3_lane_f16(a, b, 3);
 }
@@ -6157,7 +6151,7 @@ int32x4x4_t test_vld4q_s32(int32_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x8x4_t, align 16
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x8x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD4Q_V:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>
+// CHECK:   [[VLD4Q_V:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half>
 float16x8x4_t test_vld4q_f16(float16_t const * a) {
   return vld4q_f16(a);
 }
@@ -6262,7 +6256,7 @@ int64x1x4_t test_vld4_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x4_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD4_V:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[VLD4_V:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half>
 float16x4x4_t test_vld4_f16(float16_t const * a) {
   return vld4_f16(a);
 }
@@ -6367,7 +6361,7 @@ int64x1x4_t test_vld4_dup_s64(int64_t const * a) {
 // CHECK:   [[__RET:%.*]] = alloca %struct.float16x4x4_t, align 8
 // CHECK:   [[TMP0:%.*]] = bitcast %struct.float16x4x4_t* [[__RET]] to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast half* %a to i8*
-// CHECK:   [[VLD_DUP:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[VLD_DUP:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half>
 float16x4x4_t test_vld4_dup_f16(float16_t const * a) {
   return vld4_dup_f16(a);
 }
@@ -6574,11 +6568,11 @@ int32x4x4_t test_vld4q_lane_s32(int32_t const * a, int32x4x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP11:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP12:%.*]] = bitcast <8 x half> [[TMP11]] to <16 x i8>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x i16>
-// CHECK:   [[TMP16:%.*]] = bitcast <16 x i8> [[TMP12]] to <8 x i16>
-// CHECK:   [[VLD4Q_LANE_V:%.*]] = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP6]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP8]] to <8 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP10]] to <8 x half>
+// CHECK:   [[TMP16:%.*]] = bitcast <16 x i8> [[TMP12]] to <8 x half>
+// CHECK:   [[VLD4Q_LANE_V:%.*]] = call { <8 x half>, <8 x half>, <8 x half>, <8 x half>
 float16x8x4_t test_vld4q_lane_f16(float16_t const * a, float16x8x4_t b) {
   return vld4q_lane_f16(a, b, 7);
 }
@@ -6889,11 +6883,11 @@ int32x2x4_t test_vld4_lane_s32(int32_t const * a, int32x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP11:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP12:%.*]] = bitcast <4 x half> [[TMP11]] to <8 x i8>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x i16>
-// CHECK:   [[TMP16:%.*]] = bitcast <8 x i8> [[TMP12]] to <4 x i16>
-// CHECK:   [[VLD4_LANE_V:%.*]] = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP6]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP8]] to <4 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP10]] to <4 x half>
+// CHECK:   [[TMP16:%.*]] = bitcast <8 x i8> [[TMP12]] to <4 x half>
+// CHECK:   [[VLD4_LANE_V:%.*]] = call { <4 x half>, <4 x half>, <4 x half>, <4 x half>
 float16x4x4_t test_vld4_lane_f16(float16_t const * a, float16x4x4_t b) {
   return vld4_lane_f16(a, b, 3);
 }
@@ -15784,8 +15778,8 @@ void test_vst1q_s64(int64_t * a, int64x2_t b) {
 // CHECK-LABEL: @test_vst1q_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst1.p0i8.v8i16(i8* [[TMP0]], <8 x i16> [[TMP2]], i32 2)
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst1.p0i8.v8f16(i8* [[TMP0]], <8 x half> [[TMP2]], i32 2)
 // CHECK:   ret void
 void test_vst1q_f16(float16_t * a, float16x8_t b) {
   vst1q_f16(a, b);
@@ -15895,8 +15889,8 @@ void test_vst1_s64(int64_t * a, int64x1_t b) {
 // CHECK-LABEL: @test_vst1_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst1.p0i8.v4i16(i8* [[TMP0]], <4 x i16> [[TMP2]], i32 2)
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst1.p0i8.v4f16(i8* [[TMP0]], <4 x half> [[TMP2]], i32 2)
 // CHECK:   ret void
 void test_vst1_f16(float16_t * a, float16x4_t b) {
   vst1_f16(a, b);
@@ -16018,10 +16012,10 @@ void test_vst1q_lane_s64(int64_t * a, int64x2_t b) {
 // CHECK-LABEL: @test_vst1q_lane_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <8 x half> %b to <16 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16>
-// CHECK:   [[TMP3:%.*]] = extractelement <8 x i16> [[TMP2]], i32 7
-// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   store i16 [[TMP3]], i16* [[TMP4]], align 2
+// CHECK:   [[TMP2:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x half>
+// CHECK:   [[TMP3:%.*]] = extractelement <8 x half> [[TMP2]], i32 7
+// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   store half [[TMP3]], half* [[TMP4]], align 2
 // CHECK:   ret void
 void test_vst1q_lane_f16(float16_t * a, float16x8_t b) {
   vst1q_lane_f16(a, b, 7);
@@ -16150,10 +16144,10 @@ void test_vst1_lane_s64(int64_t * a, int64x1_t b) {
 // CHECK-LABEL: @test_vst1_lane_f16(
 // CHECK:   [[TMP0:%.*]] = bitcast half* %a to i8*
 // CHECK:   [[TMP1:%.*]] = bitcast <4 x half> %b to <8 x i8>
-// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16>
-// CHECK:   [[TMP3:%.*]] = extractelement <4 x i16> [[TMP2]], i32 3
-// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to i16*
-// CHECK:   store i16 [[TMP3]], i16* [[TMP4]], align 2
+// CHECK:   [[TMP2:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x half>
+// CHECK:   [[TMP3:%.*]] = extractelement <4 x half> [[TMP2]], i32 3
+// CHECK:   [[TMP4:%.*]] = bitcast i8* [[TMP0]] to half*
+// CHECK:   store half [[TMP3]], half* [[TMP4]], align 2
 // CHECK:   ret void
 void test_vst1_lane_f16(float16_t * a, float16x4_t b) {
   vst1_lane_f16(a, b, 3);
@@ -16355,9 +16349,9 @@ void test_vst2q_s32(int32_t * a, int32x4x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP6:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP7:%.*]] = bitcast <8 x half> [[TMP6]] to <16 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst2.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP8]], <8 x i16> [[TMP9]], i32 2)
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst2.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP8]], <8 x half> [[TMP9]], i32 2)
 // CHECK:   ret void
 void test_vst2q_f16(float16_t * a, float16x8x2_t b) {
   vst2q_f16(a, b);
@@ -16652,9 +16646,9 @@ void test_vst2_s64(int64_t * a, int64x1x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP6:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP7:%.*]] = bitcast <4 x half> [[TMP6]] to <8 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst2.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP8]], <4 x i16> [[TMP9]], i32 2)
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst2.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP8]], <4 x half> [[TMP9]], i32 2)
 // CHECK:   ret void
 void test_vst2_f16(float16_t * a, float16x4x2_t b) {
   vst2_f16(a, b);
@@ -16855,9 +16849,9 @@ void test_vst2q_lane_s32(int32_t * a, int32x4x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <8 x half>], [2 x <8 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP6:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX2]], align 16
 // CHECK:   [[TMP7:%.*]] = bitcast <8 x half> [[TMP6]] to <16 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst2lane.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP8]], <8 x i16> [[TMP9]], i32 7, i32 2)
+// CHECK:   [[TMP8:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst2lane.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP8]], <8 x half> [[TMP9]], i32 7, i32 2)
 // CHECK:   ret void
 void test_vst2q_lane_f16(float16_t * a, float16x8x2_t b) {
   vst2q_lane_f16(a, b, 7);
@@ -17079,9 +17073,9 @@ void test_vst2_lane_s32(int32_t * a, int32x2x2_t b) {
 // CHECK:   [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x <4 x half>], [2 x <4 x half>]* [[VAL1]], i32 0, i32 1
 // CHECK:   [[TMP6:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX2]], align 8
 // CHECK:   [[TMP7:%.*]] = bitcast <4 x half> [[TMP6]] to <8 x i8>
-// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst2lane.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP8]], <4 x i16> [[TMP9]], i32 3, i32 2)
+// CHECK:   [[TMP8:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP9:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst2lane.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP8]], <4 x half> [[TMP9]], i32 3, i32 2)
 // CHECK:   ret void
 void test_vst2_lane_f16(float16_t * a, float16x4x2_t b) {
   vst2_lane_f16(a, b, 3);
@@ -17354,10 +17348,10 @@ void test_vst3q_s32(int32_t * a, int32x4x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP8:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP9:%.*]] = bitcast <8 x half> [[TMP8]] to <16 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst3.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP10]], <8 x i16> [[TMP11]], <8 x i16> [[TMP12]], i32 2)
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst3.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP10]], <8 x half> [[TMP11]], <8 x half> [[TMP12]], i32 2)
 // CHECK:   ret void
 void test_vst3q_f16(float16_t * a, float16x8x3_t b) {
   vst3q_f16(a, b);
@@ -17705,10 +17699,10 @@ void test_vst3_s64(int64_t * a, int64x1x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP8:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP9:%.*]] = bitcast <4 x half> [[TMP8]] to <8 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst3.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP10]], <4 x i16> [[TMP11]], <4 x i16> [[TMP12]], i32 2)
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst3.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP10]], <4 x half> [[TMP11]], <4 x half> [[TMP12]], i32 2)
 // CHECK:   ret void
 void test_vst3_f16(float16_t * a, float16x4x3_t b) {
   vst3_f16(a, b);
@@ -17946,10 +17940,10 @@ void test_vst3q_lane_s32(int32_t * a, int32x4x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <8 x half>], [3 x <8 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP8:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX4]], align 16
 // CHECK:   [[TMP9:%.*]] = bitcast <8 x half> [[TMP8]] to <16 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst3lane.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP10]], <8 x i16> [[TMP11]], <8 x i16> [[TMP12]], i32 7, i32 2)
+// CHECK:   [[TMP10:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst3lane.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP10]], <8 x half> [[TMP11]], <8 x half> [[TMP12]], i32 7, i32 2)
 // CHECK:   ret void
 void test_vst3q_lane_f16(float16_t * a, float16x8x3_t b) {
   vst3q_lane_f16(a, b, 7);
@@ -18211,10 +18205,10 @@ void test_vst3_lane_s32(int32_t * a, int32x2x3_t b) {
 // CHECK:   [[ARRAYIDX4:%.*]] = getelementptr inbounds [3 x <4 x half>], [3 x <4 x half>]* [[VAL3]], i32 0, i32 2
 // CHECK:   [[TMP8:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX4]], align 8
 // CHECK:   [[TMP9:%.*]] = bitcast <4 x half> [[TMP8]] to <8 x i8>
-// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst3lane.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP10]], <4 x i16> [[TMP11]], <4 x i16> [[TMP12]], i32 3, i32 2)
+// CHECK:   [[TMP10:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP11:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst3lane.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP10]], <4 x half> [[TMP11]], <4 x half> [[TMP12]], i32 3, i32 2)
 // CHECK:   ret void
 void test_vst3_lane_f16(float16_t * a, float16x4x3_t b) {
   vst3_lane_f16(a, b, 3);
@@ -18530,11 +18524,11 @@ void test_vst4q_s32(int32_t * a, int32x4x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP10:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP11:%.*]] = bitcast <8 x half> [[TMP10]] to <16 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst4.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], <8 x i16> [[TMP15]], i32 2)
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst4.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], <8 x half> [[TMP15]], i32 2)
 // CHECK:   ret void
 void test_vst4q_f16(float16_t * a, float16x8x4_t b) {
   vst4q_f16(a, b);
@@ -18935,11 +18929,11 @@ void test_vst4_s64(int64_t * a, int64x1x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP10:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP11:%.*]] = bitcast <4 x half> [[TMP10]] to <8 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst4.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], <4 x i16> [[TMP15]], i32 2)
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst4.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], <4 x half> [[TMP15]], i32 2)
 // CHECK:   ret void
 void test_vst4_f16(float16_t * a, float16x4x4_t b) {
   vst4_f16(a, b);
@@ -19214,11 +19208,11 @@ void test_vst4q_lane_s32(int32_t * a, int32x4x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <8 x half>], [4 x <8 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP10:%.*]] = load <8 x half>, <8 x half>* [[ARRAYIDX6]], align 16
 // CHECK:   [[TMP11:%.*]] = bitcast <8 x half> [[TMP10]] to <16 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x i16>
-// CHECK:   call void @llvm.arm.neon.vst4lane.p0i8.v8i16(i8* [[TMP3]], <8 x i16> [[TMP12]], <8 x i16> [[TMP13]], <8 x i16> [[TMP14]], <8 x i16> [[TMP15]], i32 7, i32 2)
+// CHECK:   [[TMP12:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <16 x i8> [[TMP7]] to <8 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <16 x i8> [[TMP9]] to <8 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <16 x i8> [[TMP11]] to <8 x half>
+// CHECK:   call void @llvm.arm.neon.vst4lane.p0i8.v8f16(i8* [[TMP3]], <8 x half> [[TMP12]], <8 x half> [[TMP13]], <8 x half> [[TMP14]], <8 x half> [[TMP15]], i32 7, i32 2)
 // CHECK:   ret void
 void test_vst4q_lane_f16(float16_t * a, float16x8x4_t b) {
   vst4q_lane_f16(a, b, 7);
@@ -19520,11 +19514,11 @@ void test_vst4_lane_s32(int32_t * a, int32x2x4_t b) {
 // CHECK:   [[ARRAYIDX6:%.*]] = getelementptr inbounds [4 x <4 x half>], [4 x <4 x half>]* [[VAL5]], i32 0, i32 3
 // CHECK:   [[TMP10:%.*]] = load <4 x half>, <4 x half>* [[ARRAYIDX6]], align 8
 // CHECK:   [[TMP11:%.*]] = bitcast <4 x half> [[TMP10]] to <8 x i8>
-// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
-// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x i16>
-// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x i16>
-// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x i16>
-// CHECK:   call void @llvm.arm.neon.vst4lane.p0i8.v4i16(i8* [[TMP3]], <4 x i16> [[TMP12]], <4 x i16> [[TMP13]], <4 x i16> [[TMP14]], <4 x i16> [[TMP15]], i32 3, i32 2)
+// CHECK:   [[TMP12:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x half>
+// CHECK:   [[TMP13:%.*]] = bitcast <8 x i8> [[TMP7]] to <4 x half>
+// CHECK:   [[TMP14:%.*]] = bitcast <8 x i8> [[TMP9]] to <4 x half>
+// CHECK:   [[TMP15:%.*]] = bitcast <8 x i8> [[TMP11]] to <4 x half>
+// CHECK:   call void @llvm.arm.neon.vst4lane.p0i8.v4f16(i8* [[TMP3]], <4 x half> [[TMP12]], <4 x half> [[TMP13]], <4 x half> [[TMP14]], <4 x half> [[TMP15]], i32 3, i32 2)
 // CHECK:   ret void
 void test_vst4_lane_f16(float16_t * a, float16x4x4_t b) {
   vst4_lane_f16(a, b, 3);
diff --git a/test/CodeGen/default-address-space.c b/test/CodeGen/default-address-space.c
index 07ddf48fac2f..fc5f55ffd6f4 100644
--- a/test/CodeGen/default-address-space.c
+++ b/test/CodeGen/default-address-space.c
@@ -22,9 +22,10 @@ int *B;
 int test1() { return foo; }
 
 // COM-LABEL: define i32 @test2(i32 %i)
-// PIZ: load i32, i32 addrspace(4)*
+// COM: %[[addr:.*]] = getelementptr
+// PIZ: load i32, i32 addrspace(4)* %[[addr]]
 // PIZ-NEXT: ret i32
-// CHECK: load i32, i32*
+// CHECK: load i32, i32* %[[addr]]
 // CHECK-NEXT: ret i32
 int test2(int i) { return ban[i]; }
 
@@ -42,15 +43,17 @@ void test3() {
 }
 
 // PIZ-LABEL: define void @test4(i32 addrspace(4)* %a)
-// PIZ: %[[a_addr:.*]] = alloca i32 addrspace(4)*
-// PIZ: store i32 addrspace(4)* %a, i32 addrspace(4)** %[[a_addr]]
-// PIZ: %[[r0:.*]] = load i32 addrspace(4)*, i32 addrspace(4)** %[[a_addr]]
+// PIZ: %[[alloca:.*]] = alloca i32 addrspace(4)*
+// PIZ: %[[a_addr:.*]] = addrspacecast{{.*}} %[[alloca]] to i32 addrspace(4)* addrspace(4)*
+// PIZ: store i32 addrspace(4)* %a, i32 addrspace(4)* addrspace(4)* %[[a_addr]]
+// PIZ: %[[r0:.*]] = load i32 addrspace(4)*, i32 addrspace(4)* addrspace(4)* %[[a_addr]]
 // PIZ: %[[arrayidx:.*]] = getelementptr inbounds i32, i32 addrspace(4)* %[[r0]]
 // PIZ: store i32 0, i32 addrspace(4)* %[[arrayidx]]
 // CHECK-LABEL: define void @test4(i32* %a)
-// CHECK: %[[a_addr:.*]] = alloca i32*, align 4, addrspace(5)
-// CHECK: store i32* %a, i32* addrspace(5)* %[[a_addr]]
-// CHECK: %[[r0:.*]] = load i32*, i32* addrspace(5)* %[[a_addr]]
+// CHECK: %[[alloca:.*]] = alloca i32*, align 4, addrspace(5)
+// CHECK: %[[a_addr:.*]] = addrspacecast{{.*}} %[[alloca]] to i32**
+// CHECK: store i32* %a, i32** %[[a_addr]]
+// CHECK: %[[r0:.*]] = load i32*, i32** %[[a_addr]]
 // CHECK: %[[arrayidx:.*]] = getelementptr inbounds i32, i32* %[[r0]]
 // CHECK: store i32 0, i32* %[[arrayidx]]
 void test4(int *a) {
diff --git a/test/CodeGen/mcount.c b/test/CodeGen/mcount.c
index 7f915841952e..2839d8ef6af3 100644
--- a/test/CodeGen/mcount.c
+++ b/test/CodeGen/mcount.c
@@ -1,18 +1,18 @@
 // RUN: %clang_cc1 -pg -triple i386-unknown-unknown -emit-llvm -o - %s | FileCheck %s
 // RUN: %clang_cc1 -pg -triple i386-unknown-unknown -emit-llvm -O2 -o - %s | FileCheck %s
-// RUN: %clang_cc1 -pg -triple powerpc-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple powerpc64-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple powerpc64le-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple i386-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple x86_64-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple arm-netbsd-eabi -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple aarch64-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple mips-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple powerpc-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple powerpc64-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple powerpc64le-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple sparc-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
-// RUN: %clang_cc1 -pg -triple sparc64-netbsd -emit-llvm -o - %s | FileCheck -check-prefix=CHECK-PREFIXED %s
+// RUN: %clang_cc1 -pg -triple powerpc-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple powerpc64-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple powerpc64le-unknown-gnu-linux -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple i386-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple x86_64-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple arm-netbsd-eabi -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple aarch64-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple mips-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple powerpc-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple powerpc64-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple powerpc64le-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple sparc-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
+// RUN: %clang_cc1 -pg -triple sparc64-netbsd -emit-llvm -o - %s | FileCheck -check-prefixes=CHECK-PREFIXED,NO-MCOUNT1 %s
 // RUN: %clang_cc1 -emit-llvm -o - %s | FileCheck %s -check-prefix=NO-MCOUNT
 
 int bar(void) {
@@ -23,10 +23,17 @@ int foo(void) {
   return bar();
 }
 
-int main(void) {
+int __attribute__((no_instrument_function)) no_instrument(void) {
   return foo();
 }
 
-// CHECK: attributes #{{[0-9]+}} = { {{.*}}"counting-function"="mcount"{{.*}} }
-// CHECK-PREFIXED: attributes #{{[0-9]+}} = { {{.*}}"counting-function"="_mcount"{{.*}} }
+int main(void) {
+  return no_instrument();
+}
+
+// CHECK: attributes #0 = { {{.*}}"counting-function"="mcount"{{.*}} }
+// CHECK: attributes #1 = { {{.*}} }
+// CHECK-PREFIXED: attributes #0 = { {{.*}}"counting-function"="_mcount"{{.*}} }
+// CHECK-PREFIXED: attributes #1 = { {{.*}} }
 // NO-MCOUNT-NOT: attributes #{{[0-9]}} = { {{.*}}"counting-function"={{.*}} }
+// NO-MCOUNT1-NOT: attributes #1 = { {{.*}}"counting-function"={{.*}} }
diff --git a/test/CodeGen/ms-inline-asm.c b/test/CodeGen/ms-inline-asm.c
index 5182d7f2e81a..d26a660c9b0a 100644
--- a/test/CodeGen/ms-inline-asm.c
+++ b/test/CodeGen/ms-inline-asm.c
@@ -627,6 +627,12 @@ void t43() {
 // CHECK: call void asm sideeffect inteldialect "mov eax, $0", "*m,~{eax},~{dirflag},~{fpsr},~{flags}"(i32* %{{.*}})
 }
 
+void dot_operator(){
+// CHECK-LABEL: define void @dot_operator
+	__asm { mov eax, 3[ebx]A.b}
+// CHECK: call void asm sideeffect inteldialect "mov eax, $$3[ebx].4", "~{eax},~{dirflag},~{fpsr},~{flags}"
+}
+
 void call_clobber() {
   __asm call t41
   // CHECK-LABEL: define void @call_clobber
diff --git a/test/CodeGen/ms-intrinsics-other.c b/test/CodeGen/ms-intrinsics-other.c
new file mode 100644
index 000000000000..d23bc7301801
--- /dev/null
+++ b/test/CodeGen/ms-intrinsics-other.c
@@ -0,0 +1,161 @@
+// RUN: %clang_cc1 -ffreestanding -fms-extensions \
+// RUN:         -triple x86_64--darwin -Oz -emit-llvm %s -o - \
+// RUN:         | FileCheck %s
+// RUN: %clang_cc1 -ffreestanding -fms-extensions \
+// RUN:         -triple x86_64--linux -Oz -emit-llvm %s -o - \
+// RUN:         | FileCheck %s
+
+// LP64 targets use 'long' as 'int' for MS intrinsics (-fms-extensions)
+#ifdef __LP64__
+#define LONG int
+#else
+#define LONG long
+#endif
+
+unsigned char test_BitScanForward(unsigned LONG *Index, unsigned LONG Mask) {
+  return _BitScanForward(Index, Mask);
+}
+// CHECK: define{{.*}}i8 @test_BitScanForward(i32* {{[a-z_ ]*}}%Index, i32 {{[a-z_ ]*}}%Mask){{.*}}{
+// CHECK:   [[ISNOTZERO:%[a-z0-9._]+]] = icmp eq i32 %Mask, 0
+// CHECK:   br i1 [[ISNOTZERO]], label %[[END_LABEL:[a-z0-9._]+]], label %[[ISNOTZERO_LABEL:[a-z0-9._]+]]
+// CHECK:   [[END_LABEL]]:
+// CHECK:   [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK:   [[ISNOTZERO_LABEL]]:
+// CHECK:   [[INDEX:%[0-9]+]] = tail call i32 @llvm.cttz.i32(i32 %Mask, i1 true)
+// CHECK:   store i32 [[INDEX]], i32* %Index, align 4
+// CHECK:   br label %[[END_LABEL]]
+
+unsigned char test_BitScanReverse(unsigned LONG *Index, unsigned LONG Mask) {
+  return _BitScanReverse(Index, Mask);
+}
+// CHECK: define{{.*}}i8 @test_BitScanReverse(i32* {{[a-z_ ]*}}%Index, i32 {{[a-z_ ]*}}%Mask){{.*}}{
+// CHECK:   [[ISNOTZERO:%[0-9]+]] = icmp eq i32 %Mask, 0
+// CHECK:   br i1 [[ISNOTZERO]], label %[[END_LABEL:[a-z0-9._]+]], label %[[ISNOTZERO_LABEL:[a-z0-9._]+]]
+// CHECK:   [[END_LABEL]]:
+// CHECK:   [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK:   [[ISNOTZERO_LABEL]]:
+// CHECK:   [[REVINDEX:%[0-9]+]] = tail call i32 @llvm.ctlz.i32(i32 %Mask, i1 true)
+// CHECK:   [[INDEX:%[0-9]+]] = xor i32 [[REVINDEX]], 31
+// CHECK:   store i32 [[INDEX]], i32* %Index, align 4
+// CHECK:   br label %[[END_LABEL]]
+
+#if defined(__x86_64__)
+unsigned char test_BitScanForward64(unsigned LONG *Index, unsigned __int64 Mask) {
+  return _BitScanForward64(Index, Mask);
+}
+// CHECK: define{{.*}}i8 @test_BitScanForward64(i32* {{[a-z_ ]*}}%Index, i64 {{[a-z_ ]*}}%Mask){{.*}}{
+// CHECK:   [[ISNOTZERO:%[a-z0-9._]+]] = icmp eq i64 %Mask, 0
+// CHECK:   br i1 [[ISNOTZERO]], label %[[END_LABEL:[a-z0-9._]+]], label %[[ISNOTZERO_LABEL:[a-z0-9._]+]]
+// CHECK:   [[END_LABEL]]:
+// CHECK:   [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK:   [[ISNOTZERO_LABEL]]:
+// CHECK:   [[INDEX:%[0-9]+]] = tail call i64 @llvm.cttz.i64(i64 %Mask, i1 true)
+// CHECK:   [[TRUNC_INDEX:%[0-9]+]] = trunc i64 [[INDEX]] to i32
+// CHECK:   store i32 [[TRUNC_INDEX]], i32* %Index, align 4
+// CHECK:   br label %[[END_LABEL]]
+
+unsigned char test_BitScanReverse64(unsigned LONG *Index, unsigned __int64 Mask) {
+  return _BitScanReverse64(Index, Mask);
+}
+// CHECK: define{{.*}}i8 @test_BitScanReverse64(i32* {{[a-z_ ]*}}%Index, i64 {{[a-z_ ]*}}%Mask){{.*}}{
+// CHECK:   [[ISNOTZERO:%[0-9]+]] = icmp eq i64 %Mask, 0
+// CHECK:   br i1 [[ISNOTZERO]], label %[[END_LABEL:[a-z0-9._]+]], label %[[ISNOTZERO_LABEL:[a-z0-9._]+]]
+// CHECK:   [[END_LABEL]]:
+// CHECK:   [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK:   [[ISNOTZERO_LABEL]]:
+// CHECK:   [[REVINDEX:%[0-9]+]] = tail call i64 @llvm.ctlz.i64(i64 %Mask, i1 true)
+// CHECK:   [[TRUNC_REVINDEX:%[0-9]+]] = trunc i64 [[REVINDEX]] to i32
+// CHECK:   [[INDEX:%[0-9]+]] = xor i32 [[TRUNC_REVINDEX]], 63
+// CHECK:   store i32 [[INDEX]], i32* %Index, align 4
+// CHECK:   br label %[[END_LABEL]]
+#endif
+
+LONG test_InterlockedExchange(LONG volatile *value, LONG mask) {
+  return _InterlockedExchange(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedExchange(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw xchg i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedExchangeAdd(LONG volatile *value, LONG mask) {
+  return _InterlockedExchangeAdd(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedExchangeAdd(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw add i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedExchangeSub(LONG volatile *value, LONG mask) {
+  return _InterlockedExchangeSub(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedExchangeSub(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw sub i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedOr(LONG volatile *value, LONG mask) {
+  return _InterlockedOr(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedOr(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw or i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedXor(LONG volatile *value, LONG mask) {
+  return _InterlockedXor(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedXor(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw xor i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedAnd(LONG volatile *value, LONG mask) {
+  return _InterlockedAnd(value, mask);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedAnd(i32*{{[a-z_ ]*}}%value, i32{{[a-z_ ]*}}%mask){{.*}}{
+// CHECK:   [[RESULT:%[0-9]+]] = atomicrmw and i32* %value, i32 %mask seq_cst
+// CHECK:   ret i32 [[RESULT:%[0-9]+]]
+// CHECK: }
+
+LONG test_InterlockedCompareExchange(LONG volatile *Destination, LONG Exchange, LONG Comperand) {
+  return _InterlockedCompareExchange(Destination, Exchange, Comperand);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedCompareExchange(i32*{{[a-z_ ]*}}%Destination, i32{{[a-z_ ]*}}%Exchange, i32{{[a-z_ ]*}}%Comperand){{.*}}{
+// CHECK: [[TMP:%[0-9]+]] = cmpxchg volatile i32* %Destination, i32 %Comperand, i32 %Exchange seq_cst seq_cst
+// CHECK: [[RESULT:%[0-9]+]] = extractvalue { i32, i1 } [[TMP]], 0
+// CHECK: ret i32 [[RESULT]]
+// CHECK: }
+
+LONG test_InterlockedIncrement(LONG volatile *Addend) {
+  return _InterlockedIncrement(Addend);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedIncrement(i32*{{[a-z_ ]*}}%Addend){{.*}}{
+// CHECK: [[TMP:%[0-9]+]] = atomicrmw add i32* %Addend, i32 1 seq_cst
+// CHECK: [[RESULT:%[0-9]+]] = add i32 [[TMP]], 1
+// CHECK: ret i32 [[RESULT]]
+// CHECK: }
+
+LONG test_InterlockedDecrement(LONG volatile *Addend) {
+  return _InterlockedDecrement(Addend);
+}
+// CHECK: define{{.*}}i32 @test_InterlockedDecrement(i32*{{[a-z_ ]*}}%Addend){{.*}}{
+// CHECK: [[TMP:%[0-9]+]] = atomicrmw sub i32* %Addend, i32 1 seq_cst
+// CHECK: [[RESULT:%[0-9]+]] = add i32 [[TMP]], -1
+// CHECK: ret i32 [[RESULT]]
+// CHECK: }
+
+unsigned char test_interlockedbittestandset(volatile LONG *ptr, LONG bit) {
+  return _interlockedbittestandset(ptr, bit);
+}
+// CHECK-LABEL: define{{.*}} i8 @test_interlockedbittestandset
+// CHECK: [[MASKBIT:%[0-9]+]] = shl i32 1, %bit
+// CHECK: [[OLD:%[0-9]+]] = atomicrmw or i32* %ptr, i32 [[MASKBIT]] seq_cst
+// CHECK: [[SHIFT:%[0-9]+]] = lshr i32 [[OLD]], %bit
+// CHECK: [[TRUNC:%[0-9]+]] = trunc i32 [[SHIFT]] to i8
+// CHECK: [[AND:%[0-9]+]] = and i8 [[TRUNC]], 1
+// CHECK: ret i8 [[AND]]
diff --git a/test/CodeGen/ms-intrinsics-rotations.c b/test/CodeGen/ms-intrinsics-rotations.c
index 65d25cbe85eb..9533e6c3c6a2 100644
--- a/test/CodeGen/ms-intrinsics-rotations.c
+++ b/test/CodeGen/ms-intrinsics-rotations.c
@@ -1,181 +1,169 @@
 // RUN: %clang_cc1 -ffreestanding -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 \
 // RUN:         -triple i686--windows -emit-llvm %s -o - \
-// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG

+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
 // RUN: %clang_cc1 -ffreestanding -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 \
 // RUN:         -triple thumbv7--windows -emit-llvm %s -o - \
-// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG

+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
 // RUN: %clang_cc1 -ffreestanding -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 \
 // RUN:         -triple x86_64--windows -emit-llvm %s -o - \
-// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG

+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
 // RUN: %clang_cc1 -ffreestanding -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 \
 // RUN:         -triple i686--linux -emit-llvm %s -o - \
-// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG

+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
 // RUN: %clang_cc1 -ffreestanding -fms-extensions -fms-compatibility -fms-compatibility-version=17.00 \
 // RUN:         -triple x86_64--linux -emit-llvm %s -o - \
-// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-64BIT-LONG

-

-// rotate left

-

-unsigned char test_rotl8(unsigned char value, unsigned char shift) {

-  return _rotl8(value, shift);

-}

-// CHECK: i8 @test_rotl8

-// CHECK:   [[SHIFT:%[0-9]+]] = and i8 %{{[0-9]+}}, 7

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i8 8, [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i8 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i8 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i8 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i8 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i8 [[VALUE]], i8 [[ROTATED]]

-// CHECK:   ret i8 [[RESULT]]

-// CHECK  }

-

-unsigned short test_rotl16(unsigned short value, unsigned char shift) {

-  return _rotl16(value, shift);

-}

-// CHECK: i16 @test_rotl16

-// CHECK:   [[SHIFT:%[0-9]+]] = and i16 %{{[0-9]+}}, 15

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i16 16, [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i16 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i16 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i16 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i16 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i16 [[VALUE]], i16 [[ROTATED]]

-// CHECK:   ret i16 [[RESULT]]

-// CHECK  }

-

-unsigned int test_rotl(unsigned int value, int shift) {

-  return _rotl(value, shift);

-}

-// CHECK: i32 @test_rotl

-// CHECK:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]

-// CHECK:   ret i32 [[RESULT]]

-// CHECK  }

-

-unsigned long test_lrotl(unsigned long value, int shift) {

-  return _lrotl(value, shift);

-}

-// CHECK-32BIT-LONG: i32 @test_lrotl

-// CHECK-32BIT-LONG:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31

-// CHECK-32BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]

-// CHECK-32BIT-LONG:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK-32BIT-LONG:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE]], [[NEGSHIFT]]

-// CHECK-32BIT-LONG:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]

-// CHECK-32BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0

-// CHECK-32BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]

-// CHECK-32BIT-LONG:   ret i32 [[RESULT]]

-// CHECK-32BIT-LONG  }

-

-// CHECK-64BIT-LONG: i64 @test_lrotl

-// CHECK-64BIT-LONG:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63

-// CHECK-64BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]

-// CHECK-64BIT-LONG:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK-64BIT-LONG:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE]], [[NEGSHIFT]]

-// CHECK-64BIT-LONG:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]

-// CHECK-64BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0

-// CHECK-64BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]

-// CHECK-64BIT-LONG:   ret i64 [[RESULT]]

-// CHECK-64BIT-LONG  }

-

-unsigned __int64 test_rotl64(unsigned __int64 value, int shift) {

-  return _rotl64(value, shift);

-}

-// CHECK: i64 @test_rotl64

-// CHECK:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]

-// CHECK:   ret i64 [[RESULT]]

-// CHECK  }

-

-// rotate right

-

-unsigned char test_rotr8(unsigned char value, unsigned char shift) {

-  return _rotr8(value, shift);

-}

-// CHECK: i8 @test_rotr8

-// CHECK:   [[SHIFT:%[0-9]+]] = and i8 %{{[0-9]+}}, 7

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i8 8, [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i8 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i8 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i8 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i8 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i8 [[VALUE]], i8 [[ROTATED]]

-// CHECK:   ret i8 [[RESULT]]

-// CHECK  }

-

-unsigned short test_rotr16(unsigned short value, unsigned char shift) {

-  return _rotr16(value, shift);

-}

-// CHECK: i16 @test_rotr16

-// CHECK:   [[SHIFT:%[0-9]+]] = and i16 %{{[0-9]+}}, 15

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i16 16, [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i16 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i16 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i16 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i16 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i16 [[VALUE]], i16 [[ROTATED]]

-// CHECK:   ret i16 [[RESULT]]

-// CHECK  }

-

-unsigned int test_rotr(unsigned int value, int shift) {

-  return _rotr(value, shift);

-}

-// CHECK: i32 @test_rotr

-// CHECK:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]

-// CHECK:   ret i32 [[RESULT]]

-// CHECK  }

-

-unsigned long test_lrotr(unsigned long value, int shift) {

-  return _lrotr(value, shift);

-}

-// CHECK-32BIT-LONG: i32 @test_lrotr

-// CHECK-32BIT-LONG:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31

-// CHECK-32BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]

-// CHECK-32BIT-LONG:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK-32BIT-LONG:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE]], [[NEGSHIFT]]

-// CHECK-32BIT-LONG:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]

-// CHECK-32BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0

-// CHECK-32BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]

-// CHECK-32BIT-LONG:   ret i32 [[RESULT]]

-// CHECK-32BIT-LONG  }

-

-// CHECK-64BIT-LONG: i64 @test_lrotr

-// CHECK-64BIT-LONG:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63

-// CHECK-64BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]

-// CHECK-64BIT-LONG:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK-64BIT-LONG:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE]], [[NEGSHIFT]]

-// CHECK-64BIT-LONG:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]

-// CHECK-64BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0

-// CHECK-64BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]

-// CHECK-64BIT-LONG:   ret i64 [[RESULT]]

-// CHECK-64BIT-LONG  }

-

-unsigned __int64 test_rotr64(unsigned __int64 value, int shift) {

-  return _rotr64(value, shift);

-}

-// CHECK: i64 @test_rotr64

-// CHECK:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63

-// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]

-// CHECK:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE:%[0-9]+]], [[SHIFT]]

-// CHECK:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE]], [[NEGSHIFT]]

-// CHECK:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]

-// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0

-// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]

-// CHECK:   ret i64 [[RESULT]]

-// CHECK  }

+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+// RUN: %clang_cc1 -ffreestanding -fms-extensions \
+// RUN:         -triple x86_64--darwin -emit-llvm %s -o - \
+// RUN:         | FileCheck %s --check-prefixes CHECK,CHECK-32BIT-LONG
+
+// LP64 targets use 'long' as 'int' for MS intrinsics (-fms-extensions)
+#ifdef __LP64__
+#define LONG int
+#else
+#define LONG long
+#endif
+
+// rotate left
+
+unsigned char test_rotl8(unsigned char value, unsigned char shift) {
+  return _rotl8(value, shift);
+}
+// CHECK: i8 @test_rotl8
+// CHECK:   [[SHIFT:%[0-9]+]] = and i8 %{{[0-9]+}}, 7
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i8 8, [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i8 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i8 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i8 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i8 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i8 [[VALUE]], i8 [[ROTATED]]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK  }
+
+unsigned short test_rotl16(unsigned short value, unsigned char shift) {
+  return _rotl16(value, shift);
+}
+// CHECK: i16 @test_rotl16
+// CHECK:   [[SHIFT:%[0-9]+]] = and i16 %{{[0-9]+}}, 15
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i16 16, [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i16 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i16 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i16 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i16 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i16 [[VALUE]], i16 [[ROTATED]]
+// CHECK:   ret i16 [[RESULT]]
+// CHECK  }
+
+unsigned int test_rotl(unsigned int value, int shift) {
+  return _rotl(value, shift);
+}
+// CHECK: i32 @test_rotl
+// CHECK:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]
+// CHECK:   ret i32 [[RESULT]]
+// CHECK  }
+
+unsigned LONG test_lrotl(unsigned LONG value, int shift) {
+  return _lrotl(value, shift);
+}
+// CHECK-32BIT-LONG: i32 @test_lrotl
+// CHECK-32BIT-LONG:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31
+// CHECK-32BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]
+// CHECK-32BIT-LONG:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK-32BIT-LONG:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE]], [[NEGSHIFT]]
+// CHECK-32BIT-LONG:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]
+// CHECK-32BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0
+// CHECK-32BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]
+// CHECK-32BIT-LONG:   ret i32 [[RESULT]]
+// CHECK-32BIT-LONG  }
+
+unsigned __int64 test_rotl64(unsigned __int64 value, int shift) {
+  return _rotl64(value, shift);
+}
+// CHECK: i64 @test_rotl64
+// CHECK:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]
+// CHECK:   ret i64 [[RESULT]]
+// CHECK  }
+
+// rotate right
+
+unsigned char test_rotr8(unsigned char value, unsigned char shift) {
+  return _rotr8(value, shift);
+}
+// CHECK: i8 @test_rotr8
+// CHECK:   [[SHIFT:%[0-9]+]] = and i8 %{{[0-9]+}}, 7
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i8 8, [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i8 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i8 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i8 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i8 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i8 [[VALUE]], i8 [[ROTATED]]
+// CHECK:   ret i8 [[RESULT]]
+// CHECK  }
+
+unsigned short test_rotr16(unsigned short value, unsigned char shift) {
+  return _rotr16(value, shift);
+}
+// CHECK: i16 @test_rotr16
+// CHECK:   [[SHIFT:%[0-9]+]] = and i16 %{{[0-9]+}}, 15
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i16 16, [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i16 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i16 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i16 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i16 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i16 [[VALUE]], i16 [[ROTATED]]
+// CHECK:   ret i16 [[RESULT]]
+// CHECK  }
+
+unsigned int test_rotr(unsigned int value, int shift) {
+  return _rotr(value, shift);
+}
+// CHECK: i32 @test_rotr
+// CHECK:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]
+// CHECK:   ret i32 [[RESULT]]
+// CHECK  }
+
+unsigned LONG test_lrotr(unsigned LONG value, int shift) {
+  return _lrotr(value, shift);
+}
+// CHECK-32BIT-LONG: i32 @test_lrotr
+// CHECK-32BIT-LONG:   [[SHIFT:%[0-9]+]] = and i32 %{{[0-9]+}}, 31
+// CHECK-32BIT-LONG:   [[NEGSHIFT:%[0-9]+]] = sub i32 32, [[SHIFT]]
+// CHECK-32BIT-LONG:   [[LOW:%[0-9]+]] = lshr i32 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK-32BIT-LONG:   [[HIGH:%[0-9]+]] = shl i32 [[VALUE]], [[NEGSHIFT]]
+// CHECK-32BIT-LONG:   [[ROTATED:%[0-9]+]] = or i32 [[HIGH]], [[LOW]]
+// CHECK-32BIT-LONG:   [[ISZERO:%[0-9]+]] = icmp eq i32 [[SHIFT]], 0
+// CHECK-32BIT-LONG:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i32 [[VALUE]], i32 [[ROTATED]]
+// CHECK-32BIT-LONG:   ret i32 [[RESULT]]
+// CHECK-32BIT-LONG  }
+
+unsigned __int64 test_rotr64(unsigned __int64 value, int shift) {
+  return _rotr64(value, shift);
+}
+// CHECK: i64 @test_rotr64
+// CHECK:   [[SHIFT:%[0-9]+]] = and i64 %{{[0-9]+}}, 63
+// CHECK:   [[NEGSHIFT:%[0-9]+]] = sub i64 64, [[SHIFT]]
+// CHECK:   [[LOW:%[0-9]+]] = lshr i64 [[VALUE:%[0-9]+]], [[SHIFT]]
+// CHECK:   [[HIGH:%[0-9]+]] = shl i64 [[VALUE]], [[NEGSHIFT]]
+// CHECK:   [[ROTATED:%[0-9]+]] = or i64 [[HIGH]], [[LOW]]
+// CHECK:   [[ISZERO:%[0-9]+]] = icmp eq i64 [[SHIFT]], 0
+// CHECK:   [[RESULT:%[0-9]+]] = select i1 [[ISZERO]], i64 [[VALUE]], i64 [[ROTATED]]
+// CHECK:   ret i64 [[RESULT]]
+// CHECK  }
diff --git a/test/CodeGen/no-devirt.cpp b/test/CodeGen/no-devirt.cpp
new file mode 100644
index 000000000000..4333b7cde7c6
--- /dev/null
+++ b/test/CodeGen/no-devirt.cpp
@@ -0,0 +1,59 @@
+// RUN: %clang_cc1 %s -DUSEIT -triple %itanium_abi_triple -emit-llvm -o - |  FileCheck %s
+
+// Test with decls and template defs in pch, and just use in .cpp
+// RUN:  %clang_cc1 %s -DTMPL_DEF_IN_HEADER -triple %itanium_abi_triple -emit-pch -o %t
+// RUN:  %clang_cc1 %s -DTMPL_DEF_IN_HEADER -DUSEIT -triple %itanium_abi_triple -include-pch %t -emit-llvm -o - | FileCheck %s
+
+// Test with A in pch, and B and C in main
+// Test with just decls in pch, and template defs and use in .cpp
+// RUN:  %clang_cc1 %s -triple %itanium_abi_triple -emit-pch -o %t
+// RUN:  %clang_cc1 %s -DUSEIT -triple %itanium_abi_triple -include-pch %t -emit-llvm -o - | FileCheck %s
+
+#ifndef HEADER
+#define HEADER
+template < typename T, int N = 0 > class TmplWithArray {
+public:
+  virtual T& operator [] (int idx);
+  virtual T& func1 (int idx);
+  virtual T& func2 (int idx);
+  T ar[N+1];
+};
+struct Wrapper {
+  TmplWithArray<bool, 10> data;
+  bool indexIt(int a) {
+    if (a > 6) return data[a] ;      // Should not devirtualize
+    if (a > 4) return data.func1(a); // Should devirtualize
+    return data.func2(a);            // Should devirtualize
+  }
+};
+
+#ifdef TMPL_DEF_IN_HEADER
+template <typename T, int N> T& TmplWithArray<T, N >::operator[](int idx) {
+  return ar[idx];
+}
+template <typename T, int N> T& TmplWithArray<T, N >::func1(int idx) {
+  return ar[idx];
+}
+#endif // TMPL_DEF_IN_HEADER
+#endif // HEADER
+
+#ifdef USEIT
+#ifndef TMPL_DEF_IN_HEADER
+template <typename T, int N> T& TmplWithArray<T, N >::operator[](int idx) {
+  return ar[idx];
+}
+template <typename T, int N> T& TmplWithArray<T, N >::func1(int idx) {
+  return ar[idx];
+}
+#endif // !TMPL_DEF_IN_HEADER
+extern Wrapper ew;
+bool stuff(int p)
+{
+  return ew.indexIt(p);
+}
+#endif
+
+// CHECK-NOT: call {{.*}} @_ZN13TmplWithArrayIbLi10EEixEi
+// CHECK-DAG: call {{.*}} @_ZN13TmplWithArrayIbLi10EE5func1Ei
+// CHECK-DAG: call {{.*}} @_ZN13TmplWithArrayIbLi10EE5func2Ei
+
diff --git a/test/CodeGen/pr27892.c b/test/CodeGen/pr27892.c
deleted file mode 100644
index 57722c4671ab..000000000000
--- a/test/CodeGen/pr27892.c
+++ /dev/null
@@ -1,23 +0,0 @@
-// RUN: %clang_cc1 -triple x86_64-linux-gnu -fms-extensions %s -emit-llvm -o - | FileCheck %s
-
-long test1(long *p) {
-  return _InterlockedIncrement(p);
-}
-// CHECK-DAG: define i64 @test1(
-// CHECK:   %[[p_addr:.*]] = alloca i64*, align 8
-// CHECK:   store i64* %p, i64** %[[p_addr]], align 8
-// CHECK:   %[[p_load:.*]] = load i64*, i64** %[[p_addr]], align 8
-// CHECK:   %[[atomic_add:.*]] = atomicrmw add i64* %[[p_load]], i64 1 seq_cst
-// CHECK:   %[[res:.*]] = add i64 %[[atomic_add]], 1
-// CHECK:   ret i64 %[[res]]
-
-long test2(long *p) {
-  return _InterlockedDecrement(p);
-}
-// CHECK-DAG: define i64 @test2(
-// CHECK:   %[[p_addr:.*]] = alloca i64*, align 8
-// CHECK:   store i64* %p, i64** %[[p_addr]], align 8
-// CHECK:   %[[p_load:.*]] = load i64*, i64** %[[p_addr]], align 8
-// CHECK:   %[[atomic_sub:.*]] = atomicrmw sub i64* %[[p_load]], i64 1 seq_cst
-// CHECK:   %[[res:.*]] = sub i64 %[[atomic_sub]], 1
-// CHECK:   ret i64 %[[res]]
diff --git a/test/CodeGen/target-data.c b/test/CodeGen/target-data.c
index 1e8ce6a2fd12..68ee8f02d2ee 100644
--- a/test/CodeGen/target-data.c
+++ b/test/CodeGen/target-data.c
@@ -175,7 +175,7 @@
 
 // RUN: %clang_cc1 -triple msp430-unknown -o - -emit-llvm %s | \
 // RUN: FileCheck %s -check-prefix=MSP430
-// MSP430: target datalayout = "e-m:e-p:16:16-i32:16:32-a:16-n8:16"
+// MSP430: target datalayout = "e-m:e-p:16:16-i32:16-i64:16-f32:16-f64:16-a:8-n8:16-S16"
 
 // RUN: %clang_cc1 -triple tce-unknown -o - -emit-llvm %s | \
 // RUN: FileCheck %s -check-prefix=TCE
diff --git a/test/CodeGen/vectorcall.c b/test/CodeGen/vectorcall.c
index 167f72ca2cfd..fa244fb908e0 100644
--- a/test/CodeGen/vectorcall.c
+++ b/test/CodeGen/vectorcall.c
@@ -100,8 +100,19 @@ void __vectorcall odd_size_hva(struct OddSizeHVA a) {}
 // X32: define x86_vectorcallcc void @"\01odd_size_hva@@32"(%struct.OddSizeHVA inreg %a.coerce)
 // X64: define x86_vectorcallcc void @"\01odd_size_hva@@32"(%struct.OddSizeHVA inreg %a.coerce)
 
-// The Vectorcall ABI only allows passing the first 6 items in registers, so this shouldn't 
+// The Vectorcall ABI only allows passing the first 6 items in registers in x64, so this shouldn't
 // consider 'p7' as a register.  Instead p5 gets put into the register on the second pass.
-struct HFA2 __vectorcall AddParticles(struct HFA2 p1, float p2, struct HFA4 p3, int p4, struct HFA2 p5, float p6, float p7){ return p1;}
-// X32: define x86_vectorcallcc %struct.HFA2 @"\01AddParticles@@80"(%struct.HFA2 inreg %p1.coerce, float %p2, %struct.HFA4* inreg %p3, i32 inreg %p4, %struct.HFA2 inreg %p5.coerce, float %p6, float %p7)
-// X64: define x86_vectorcallcc %struct.HFA2 @"\01AddParticles@@96"(%struct.HFA2 inreg %p1.coerce, float %p2, %struct.HFA4* %p3, i32 %p4, %struct.HFA2 inreg %p5.coerce, float %p6, float %p7)
+// x86 should pass p2, p6 and p7 in registers, then p1 in the second pass.
+struct HFA2 __vectorcall AddParticles(struct HFA2 p1, float p2, struct HFA4 p3, int p4, struct HFA2 p5, float p6, float p7, int p8){ return p1;}
+// X32: define x86_vectorcallcc %struct.HFA2 @"\01AddParticles@@84"(%struct.HFA2 inreg %p1.coerce, float %p2, %struct.HFA4* inreg %p3, i32 inreg %p4, %struct.HFA2* %p5, float %p6, float %p7, i32 %p8)
+// X64: define x86_vectorcallcc %struct.HFA2 @"\01AddParticles@@104"(%struct.HFA2 inreg %p1.coerce, float %p2, %struct.HFA4* %p3, i32 %p4, %struct.HFA2 inreg %p5.coerce, float %p6, float %p7, i32 %p8)
+
+// Vectorcall in both architectures allows passing of an HVA as long as there is room,
+// even if it is not one of the first 6 arguments.  First pass puts p4 into a
+// register on both.  p9 ends up in a register in x86 only.  Second pass puts p1
+// in a register, does NOT put p7 in a register (since theres no room), then puts 
+// p8 in a register.
+void __vectorcall HVAAnywhere(struct HFA2 p1, int p2, int p3, float p4, int p5, int p6, struct HFA4 p7, struct HFA2 p8, float p9){}
+// X32: define x86_vectorcallcc void @"\01HVAAnywhere@@88"(%struct.HFA2 inreg %p1.coerce, i32 inreg %p2, i32 inreg %p3, float %p4, i32 %p5, i32 %p6, %struct.HFA4* %p7, %struct.HFA2 inreg %p8.coerce, float %p9)
+// X64: define x86_vectorcallcc void @"\01HVAAnywhere@@112"(%struct.HFA2 inreg %p1.coerce, i32 %p2, i32 %p3, float %p4, i32 %p5, i32 %p6, %struct.HFA4* %p7, %struct.HFA2 inreg %p8.coerce, float %p9) 
+
diff --git a/test/CodeGen/x86_64-arguments.c b/test/CodeGen/x86_64-arguments.c
index 9f375d780c94..d24ea4dbab3d 100644
--- a/test/CodeGen/x86_64-arguments.c
+++ b/test/CodeGen/x86_64-arguments.c
@@ -460,7 +460,7 @@ void test54() {
   test54_helper(x54, x54, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0i);
 }
 // AVX: @test54_helper(<8 x float> {{%[a-zA-Z0-9]+}}, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double {{%[a-zA-Z0-9]+}}, double {{%[a-zA-Z0-9]+}})
-// AVX: @test54_helper(<8 x float> {{%[a-zA-Z0-9]+}}, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, { double, double }* byval align 8 {{%[a-zA-Z0-9]+}})
+// AVX: @test54_helper(<8 x float> {{%[a-zA-Z0-9]+}}, <8 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, { double, double }* byval align 8 {{%[^)]+}})
 
 typedef float __m512 __attribute__ ((__vector_size__ (64)));
 typedef struct {
@@ -529,7 +529,7 @@ void f63(__m512 *m, __builtin_va_list argList) {
 }
 
 // AVX512: @f64_helper(<16 x float> {{%[a-zA-Z0-9]+}}, <16 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double {{%[a-zA-Z0-9]+}}, double {{%[a-zA-Z0-9]+}})
-// AVX512: @f64_helper(<16 x float> {{%[a-zA-Z0-9]+}}, <16 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, { double, double }* byval align 8 {{%[a-zA-Z0-9]+}})
+// AVX512: @f64_helper(<16 x float> {{%[a-zA-Z0-9]+}}, <16 x float> {{%[a-zA-Z0-9]+}}, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, { double, double }* byval align 8 {{%[^)]+}})
 void f64_helper(__m512, ...);
 __m512 x64;
 void f64() {
diff --git a/test/CodeGenCXX/amdgcn-automatic-variable.cpp b/test/CodeGenCXX/amdgcn-automatic-variable.cpp
index aab720770d12..7df27c28e6d2 100644
--- a/test/CodeGenCXX/amdgcn-automatic-variable.cpp
+++ b/test/CodeGenCXX/amdgcn-automatic-variable.cpp
@@ -3,9 +3,10 @@
 // CHECK-LABEL: define void @_Z5func1Pi(i32* %x)
 void func1(int *x) {
   // CHECK: %[[x_addr:.*]] = alloca i32*{{.*}}addrspace(5)
-  // CHECK: store i32* %x, i32* addrspace(5)* %[[x_addr]]
-  // CHECK: %[[r0:.*]] = load i32*, i32* addrspace(5)* %[[x_addr]]
-  // CHECK: store i32 1, i32* %[[r0]]
+  // CHECK: %[[r0:.*]] = addrspacecast i32* addrspace(5)* %[[x_addr]] to i32**
+  // CHECK: store i32* %x, i32** %[[r0]]
+  // CHECK: %[[r1:.*]] = load i32*, i32** %[[r0]]
+  // CHECK: store i32 1, i32* %[[r1]]
   *x = 1;
 }
 
@@ -70,3 +71,12 @@ void func3() {
   // CHECK: call void @_ZN1AD1Ev(%class.A* %[[r0]])
   A a;
 }
+
+// CHECK-LABEL: define void @_Z5func4i
+void func4(int x) {
+  // CHECK: %[[x_addr:.*]] = alloca i32, align 4, addrspace(5)
+  // CHECK: %[[r0:.*]] = addrspacecast i32 addrspace(5)* %[[x_addr]] to i32*
+  // CHECK: store i32 %x, i32* %[[r0]], align 4
+  // CHECK: call void @_Z5func1Pi(i32* %[[r0]])
+  func1(&x);
+}
diff --git a/test/CodeGenObjC/objc_copyStruct.m b/test/CodeGenObjC/objc_copyStruct.m
new file mode 100644
index 000000000000..7fa0ad54f3ae
--- /dev/null
+++ b/test/CodeGenObjC/objc_copyStruct.m
@@ -0,0 +1,16 @@
+// RUN: %clang -target x86_64-unknown-windows-msvc -fobjc-runtime=ios -Wno-objc-root-class -S -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang -target x86_64-apple-ios -fobjc-runtime=ios -Wno-objc-root-class -S -o - -emit-llvm %s | FileCheck %s
+
+struct S {
+  float f, g;
+};
+
+@interface I
+@property struct S s;
+@end
+
+@implementation I
+@end
+
+// CHECK: declare {{.*}}void @objc_copyStruct(i8*, i8*, i64, i1, i1)
+
diff --git a/test/CodeGenObjC/ubsan-nonnull-and-nullability.m b/test/CodeGenObjC/ubsan-nonnull-and-nullability.m
index b927f55cd402..db6588dac310 100644
--- a/test/CodeGenObjC/ubsan-nonnull-and-nullability.m
+++ b/test/CodeGenObjC/ubsan-nonnull-and-nullability.m
@@ -7,16 +7,26 @@
 // CHECK-LABEL: define nonnull i32* @f1
 __attribute__((returns_nonnull)) int *_Nonnull f1(int *_Nonnull p) {
   // CHECK: entry:
+  // CHECK-NEXT: [[SLOC_PTR:%.*]] = alloca i8*
   // CHECK-NEXT: [[ADDR:%.*]] = alloca i32*
+  // CHECK-NEXT: store i8* null, i8** [[SLOC_PTR]]
   // CHECK-NEXT: store i32* [[P:%.*]], i32** [[ADDR]]
+  // CHECK-NEXT: store {{.*}} [[SLOC_PTR]]
   // CHECK-NEXT: [[ARG:%.*]] = load i32*, i32** [[ADDR]]
+  // CHECK-NEXT: [[SLOC:%.*]] = load {{.*}} [[SLOC_PTR]]
+  // CHECK-NEXT: [[SLOC_NONNULL:%.*]] = icmp ne i8* [[SLOC]], null
+  // CHECK-NEXT: br i1 [[SLOC_NONNULL]], label %nullcheck
+  // 
+  // CHECK: nullcheck:
   // CHECK-NEXT: [[ICMP:%.*]] = icmp ne i32* [[ARG]], null, !nosanitize
   // CHECK-NEXT: br i1 [[ICMP]], label %[[CONT:.+]], label %[[HANDLE:[^,]+]]
   // CHECK: [[HANDLE]]:
-  // CHECK-NEXT:   call void @__ubsan_handle_nonnull_return_abort
+  // CHECK:      call void @__ubsan_handle_nonnull_return
   // CHECK-NEXT:   unreachable, !nosanitize
   // CHECK: [[CONT]]:
-  // CHECK-NEXT:   ret i32*
+  // CHECK-NEXT:   br label %no.nullcheck
+  // CHECK: no.nullcheck:
+  // CHECK-NEXT: ret i32* [[ARG]]
   return p;
 }
 
@@ -29,3 +39,23 @@ void call_f2() {
   // CHECK-NOT: call void @__ubsan_handle_nonnull_arg_abort
   f2((void *)0);
 }
+
+// If the return value isn't meant to be checked, make sure we don't check it.
+// CHECK-LABEL: define i32* @f3
+int *f3(int *p) {
+  // CHECK-NOT: return.sloc
+  // CHECK-NOT: call{{.*}}ubsan
+  return p;
+}
+
+// Check for a valid "return" source location, even when there is no return
+// statement, to avoid accidentally calling the runtime.
+
+// CHECK-LABEL: define nonnull i32* @f4
+__attribute__((returns_nonnull)) int *f4() {
+  // CHECK: store i8* null, i8** [[SLOC_PTR:%.*]]
+  // CHECK: [[SLOC:%.*]] = load {{.*}} [[SLOC_PTR]]
+  // CHECK: [[SLOC_NONNULL:%.*]] = icmp ne i8* [[SLOC]], null
+  // CHECK: br i1 [[SLOC_NONNULL]], label %nullcheck
+  // CHECK: nullcheck:
+}
diff --git a/test/CodeGenObjC/ubsan-nullability.m b/test/CodeGenObjC/ubsan-nullability.m
index 7f53ea6292ee..eeb24b03c868 100644
--- a/test/CodeGenObjC/ubsan-nullability.m
+++ b/test/CodeGenObjC/ubsan-nullability.m
@@ -2,7 +2,7 @@
 // RUN: %clang_cc1 -x objective-c -emit-llvm -triple x86_64-apple-macosx10.10.0 -fsanitize=nullability-arg,nullability-assign,nullability-return -w %s -o - | FileCheck %s
 // RUN: %clang_cc1 -x objective-c++ -emit-llvm -triple x86_64-apple-macosx10.10.0 -fsanitize=nullability-arg,nullability-assign,nullability-return -w %s -o - | FileCheck %s
 
-// CHECK: [[NONNULL_RV_LOC1:@.*]] = private unnamed_addr global {{.*}} i32 109, i32 1 {{.*}} i32 100, i32 6
+// CHECK: [[NONNULL_RV_LOC1:@.*]] = private unnamed_addr global {{.*}} i32 100, i32 6
 // CHECK: [[NONNULL_ARG_LOC:@.*]] = private unnamed_addr global {{.*}} i32 204, i32 15 {{.*}} i32 190, i32 23
 // CHECK: [[NONNULL_ASSIGN1_LOC:@.*]] = private unnamed_addr global {{.*}} i32 305, i32 9
 // CHECK: [[NONNULL_ASSIGN2_LOC:@.*]] = private unnamed_addr global {{.*}} i32 405, i32 10
@@ -10,7 +10,7 @@
 // CHECK: [[NONNULL_INIT1_LOC:@.*]] = private unnamed_addr global {{.*}} i32 604, i32 25
 // CHECK: [[NONNULL_INIT2_LOC1:@.*]] = private unnamed_addr global {{.*}} i32 707, i32 26
 // CHECK: [[NONNULL_INIT2_LOC2:@.*]] = private unnamed_addr global {{.*}} i32 707, i32 29
-// CHECK: [[NONNULL_RV_LOC2:@.*]] = private unnamed_addr global {{.*}} i32 817, i32 1 {{.*}} i32 800, i32 6
+// CHECK: [[NONNULL_RV_LOC2:@.*]] = private unnamed_addr global {{.*}} i32 800, i32 6
 
 #define NULL ((void *)0)
 #define INULL ((int *)NULL)
@@ -19,14 +19,11 @@
 // CHECK-LABEL: define i32* @{{.*}}nonnull_retval1
 #line 100
 int *_Nonnull nonnull_retval1(int *p) {
-  // CHECK: br i1 true, label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
-  // CHECK: [[NULL]]:
   // CHECK: [[ICMP:%.*]] = icmp ne i32* {{.*}}, null, !nosanitize
-  // CHECK-NEXT: br i1 [[ICMP]], {{.*}}, !nosanitize
+  // CHECK: br i1 [[ICMP]], {{.*}}, !nosanitize
   // CHECK: call void @__ubsan_handle_nullability_return{{.*}}[[NONNULL_RV_LOC1]]
   return p;
-  // CHECK: [[NONULL]]:
-  // CHECK-NEXT: ret i32*
+  // CHECK: ret i32*
 }
 
 #line 190
@@ -108,10 +105,13 @@ int *_Nonnull nonnull_retval2(int *_Nonnull arg1,  //< Test this.
   // CHECK-NEXT: [[DO_RV_CHECK_1:%.*]] = and i1 true, [[ARG1CMP]], !nosanitize
   // CHECK: [[ARG2CMP:%.*]] = icmp ne i32* %arg2, null, !nosanitize
   // CHECK-NEXT: [[DO_RV_CHECK_2:%.*]] = and i1 [[DO_RV_CHECK_1]], [[ARG2CMP]]
-  // CHECK: br i1 [[DO_RV_CHECK_2]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
+  // CHECK: [[SLOC_PTR:%.*]] = load i8*, i8** %return.sloc.ptr
+  // CHECK-NEXT: [[SLOC_NONNULL:%.*]] = icmp ne i8* [[SLOC_PTR]], null
+  // CHECK-NEXT: [[DO_RV_CHECK_3:%.*]] = and i1 [[SLOC_NONNULL]], [[DO_RV_CHECK_2]]
+  // CHECK: br i1 [[DO_RV_CHECK_3]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
   // CHECK: [[NULL]]:
   // CHECK-NEXT: [[ICMP:%.*]] = icmp ne i32* {{.*}}, null, !nosanitize
-  // CHECK-NEXT: br i1 [[ICMP]], {{.*}}, !nosanitize
+  // CHECK: br i1 [[ICMP]], {{.*}}, !nosanitize
   // CHECK: call void @__ubsan_handle_nullability_return{{.*}}[[NONNULL_RV_LOC2]]
   return arg1;
   // CHECK: [[NONULL]]:
@@ -129,10 +129,13 @@ int *_Nonnull nonnull_retval2(int *_Nonnull arg1,  //< Test this.
 +(int *_Nonnull) objc_clsmethod: (int *_Nonnull) arg1 {
   // CHECK: [[ARG1CMP:%.*]] = icmp ne i32* %arg1, null, !nosanitize
   // CHECK-NEXT: [[DO_RV_CHECK:%.*]] = and i1 true, [[ARG1CMP]]
-  // CHECK: br i1 [[DO_RV_CHECK]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
+  // CHECK: [[SLOC_PTR:%.*]] = load i8*, i8** %return.sloc.ptr
+  // CHECK-NEXT: [[SLOC_NONNULL:%.*]] = icmp ne i8* [[SLOC_PTR]], null
+  // CHECK-NEXT: [[DO_RV_CHECK_2:%.*]] = and i1 [[SLOC_NONNULL]], [[DO_RV_CHECK]]
+  // CHECK: br i1 [[DO_RV_CHECK_2]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
   // CHECK: [[NULL]]:
   // CHECK-NEXT: [[ICMP:%.*]] = icmp ne i32* {{.*}}, null, !nosanitize
-  // CHECK-NEXT: br i1 [[ICMP]], {{.*}}, !nosanitize
+  // CHECK: br i1 [[ICMP]], {{.*}}, !nosanitize
   // CHECK: call void @__ubsan_handle_nullability_return{{.*}}
   return arg1;
   // CHECK: [[NONULL]]:
@@ -143,10 +146,13 @@ int *_Nonnull nonnull_retval2(int *_Nonnull arg1,  //< Test this.
 -(int *_Nonnull) objc_method: (int *_Nonnull) arg1 {
   // CHECK: [[ARG1CMP:%.*]] = icmp ne i32* %arg1, null, !nosanitize
   // CHECK-NEXT: [[DO_RV_CHECK:%.*]] = and i1 true, [[ARG1CMP]]
-  // CHECK: br i1 [[DO_RV_CHECK]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
+  // CHECK: [[SLOC_PTR:%.*]] = load i8*, i8** %return.sloc.ptr
+  // CHECK-NEXT: [[SLOC_NONNULL:%.*]] = icmp ne i8* [[SLOC_PTR]], null
+  // CHECK-NEXT: [[DO_RV_CHECK_2:%.*]] = and i1 [[SLOC_NONNULL]], [[DO_RV_CHECK]]
+  // CHECK: br i1 [[DO_RV_CHECK_2]], label %[[NULL:.*]], label %[[NONULL:.*]], !nosanitize
   // CHECK: [[NULL]]:
   // CHECK-NEXT: [[ICMP:%.*]] = icmp ne i32* {{.*}}, null, !nosanitize
-  // CHECK-NEXT: br i1 [[ICMP]], {{.*}}, !nosanitize
+  // CHECK: br i1 [[ICMP]], {{.*}}, !nosanitize
   // CHECK: call void @__ubsan_handle_nullability_return{{.*}}
   return arg1;
   // CHECK: [[NONULL]]:
diff --git a/test/CodeGenOpenCL/spir_version.cl b/test/CodeGenOpenCL/spir_version.cl
index 8a191282b3c7..ac5b8e8c7fa5 100644
--- a/test/CodeGenOpenCL/spir_version.cl
+++ b/test/CodeGenOpenCL/spir_version.cl
@@ -10,17 +10,18 @@
 // RUN: %clang_cc1 %s -triple "amdgcn--amdhsa" -emit-llvm -o - -cl-std=CL2.0 | FileCheck %s --check-prefix=CHECK-AMDGCN-CL20
 
 kernel void foo() {}
+kernel void bar() {}
 
-// CHECK-SPIR-CL10: !opencl.spir.version = !{[[SPIR:![0-9]+]]}
-// CHECK-SPIR-CL10: !opencl.ocl.version = !{[[OCL:![0-9]+]]}
-// CHECK-SPIR-CL10: [[SPIR]] = !{i32 1, i32 2}
-// CHECK-SPIR-CL10: [[OCL]] = !{i32 1, i32 0}
-// CHECK-SPIR-CL12: !opencl.spir.version = !{[[VER:![0-9]+]]}
-// CHECK-SPIR-CL12: !opencl.ocl.version = !{[[VER]]}
+// CHECK-SPIR-CL10-DAG: !opencl.spir.version = !{[[SPIR:![0-9]+]]}
+// CHECK-SPIR-CL10-DAG: !opencl.ocl.version = !{[[OCL:![0-9]+]]}
+// CHECK-SPIR-CL10-DAG: [[SPIR]] = !{i32 1, i32 2}
+// CHECK-SPIR-CL10-DAG: [[OCL]] = !{i32 1, i32 0}
+// CHECK-SPIR-CL12-DAG: !opencl.spir.version = !{[[VER:![0-9]+]]}
+// CHECK-SPIR-CL12-DAG: !opencl.ocl.version = !{[[VER]]}
 // CHECK-SPIR-CL12: [[VER]] = !{i32 1, i32 2}
 
-// CHECK-SPIR-CL20: !opencl.spir.version = !{[[VER:![0-9]+]]}
-// CHECK-SPIR-CL20: !opencl.ocl.version = !{[[VER]]}
+// CHECK-SPIR-CL20-DAG: !opencl.spir.version = !{[[VER:![0-9]+]]}
+// CHECK-SPIR-CL20-DAG: !opencl.ocl.version = !{[[VER]]}
 // CHECK-SPIR-CL20: [[VER]] = !{i32 2, i32 0}
 
 // CHECK-AMDGCN-CL10-NOT: !opencl.spir.version
diff --git a/test/Driver/ananas.c b/test/Driver/ananas.c
new file mode 100644
index 000000000000..2a5b35ed6cec
--- /dev/null
+++ b/test/Driver/ananas.c
@@ -0,0 +1,9 @@
+// RUN: %clang -no-canonical-prefixes -target x86_64-unknown-ananas -static %s \
+// RUN:   --sysroot=%S/Inputs/ananas-tree -### 2>&1 \
+// RUN:   | FileCheck --check-prefix=CHECK-STATIC %s
+// CHECK-STATIC: ld{{.*}}" "-Bstatic"
+// CHECK-STATIC: crt0.o
+// CHECK-STATIC: crti.o
+// CHECK-STATIC: crtbegin.o
+// CHECK-STATIC: crtend.o
+// CHECK-STATIC: crtn.o
diff --git a/test/Driver/autocomplete.c b/test/Driver/autocomplete.c
index 94649f243738..f0bdcce5707c 100644
--- a/test/Driver/autocomplete.c
+++ b/test/Driver/autocomplete.c
@@ -2,5 +2,37 @@
 // FSYN: -fsyntax-only
 // RUN: %clang --autocomplete=-s | FileCheck %s -check-prefix=STD
 // STD: -std={{.*}}-stdlib=
-// RUN: %clang --autocomplete=foo | not FileCheck %s -check-prefix=NONE
-// NONE: foo
+// RUN: %clang --autocomplete=foo | FileCheck %s -check-prefix=FOO
+// FOO-NOT: foo
+// RUN: %clang --autocomplete=-stdlib=,l | FileCheck %s -check-prefix=STDLIB
+// STDLIB: libc++ libstdc++
+// RUN: %clang --autocomplete=-stdlib=, | FileCheck %s -check-prefix=STDLIBALL
+// STDLIBALL: libc++ libstdc++ platform
+// RUN: %clang --autocomplete=-meabi,d | FileCheck %s -check-prefix=MEABI
+// MEABI: default
+// RUN: %clang --autocomplete=-meabi, | FileCheck %s -check-prefix=MEABIALL
+// MEABIALL: 4 5 default gnu
+// RUN: %clang --autocomplete=-cl-std=,CL2 | FileCheck %s -check-prefix=CLSTD
+// CLSTD: CL2.0
+// RUN: %clang --autocomplete=-cl-std=, | FileCheck %s -check-prefix=CLSTDALL
+// CLSTDALL: cl CL cl1.1 CL1.1 cl1.2 CL1.2 cl2.0 CL2.0
+// RUN: %clang --autocomplete=-fno-sanitize-coverage=,f | FileCheck %s -check-prefix=FNOSANICOVER
+// FNOSANICOVER: func
+// RUN: %clang --autocomplete=-fno-sanitize-coverage=, | FileCheck %s -check-prefix=FNOSANICOVERALL
+// FNOSANICOVERALL: 8bit-counters bb edge func indirect-calls inline-8bit-counters no-prune trace-bb trace-cmp trace-div trace-gep trace-pc trace-pc-guard
+// RUN: %clang --autocomplete=-ffp-contract=, | FileCheck %s -check-prefix=FFPALL
+// FFPALL: fast off on
+// RUN: %clang --autocomplete=-flto=, | FileCheck %s -check-prefix=FLTOALL
+// FLTOALL: full thin
+// RUN: %clang --autocomplete=-fveclib=, | FileCheck %s -check-prefix=FVECLIBALL
+// FVECLIBALL: Accelerate none SVML
+// RUN: %clang --autocomplete=-fshow-overloads=, | FileCheck %s -check-prefix=FSOVERALL
+// FSOVERALL: all best
+// RUN: %clang --autocomplete=-fvisibility=, | FileCheck %s -check-prefix=FVISIBILITYALL
+// FVISIBILITYALL: default hidden
+// RUN: %clang --autocomplete=-mfloat-abi=, | FileCheck %s -check-prefix=MFLOATABIALL
+// MFLOATABIALL: hard soft softfp
+// RUN: %clang --autocomplete=-mthread-model, | FileCheck %s -check-prefix=MTHREADMODELALL
+// MTHREADMODELALL: posix single
+// RUN: %clang --autocomplete=-mrelocation-model, | FileCheck %s -check-prefix=MRELOCMODELALL
+// MRELOCMODELALL: dynamic-no-pic pic ropi ropi-rwpi rwpi static
diff --git a/test/Driver/compress-noias.c b/test/Driver/compress-noias.c
new file mode 100644
index 000000000000..d20cf366b8fc
--- /dev/null
+++ b/test/Driver/compress-noias.c
@@ -0,0 +1,37 @@
+// REQUIRES: zlib
+// REQUIRES: x86-registered-target
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -Wa,-compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-_COMPRESS_DEBUG_SECTIONS %s
+// CHECK-_COMPRESS_DEBUG_SECTIONS: "-compress-debug-sections"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -Wa,--compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-__COMPRESS_DEBUG_SECTIONS %s
+// CHECK-__COMPRESS_DEBUG_SECTIONS: "--compress-debug-sections"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -Wa,--compress-debug-sections -Wa,--nocompress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-POSNEG %s
+// CHECK-POSNEG: "--compress-debug-sections"
+// CHECK-POSNEG: "--nocompress-debug-sections"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -Wa,-compress-debug-sections -Wa,--compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-MULTIPLE %s
+// CHECK-MULTIPLE: "-compress-debug-sections"
+// CHECK-MULTIPLE: "--compress-debug-sections"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ %s
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ %s
+// CHECK-OPT_GZ: "-compress-debug-sections"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=none -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_NONE %s
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=none -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_NONE %s
+// CHECK-OPT_GZ_EQ_NONE: "-compress-debug-sections=none"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=zlib -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB %s
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=zlib -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB %s
+// CHECK-OPT_GZ_EQ_ZLIB: "-compress-debug-sections=zlib"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=zlib-gnu -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB_GNU %s
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=zlib-gnu -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB_GNU %s
+// CHECK-OPT_GZ_EQ_ZLIB_GNU: "-compress-debug-sections=zlib-gnu"
+
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=invalid -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_INVALID %s
+// RUN: %clang -### -target i686-unknown-linux-gnu -fno-integrated-as -gz=invalid -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_INVALID %s
+// CHECK-OPT_GZ_EQ_INVALID: error: unsupported argument 'invalid' to option 'gz='
+
diff --git a/test/Driver/compress.c b/test/Driver/compress.c
index 6cdc6b72243e..a00ce91225bd 100644
--- a/test/Driver/compress.c
+++ b/test/Driver/compress.c
@@ -1,8 +1,36 @@
-// RUN: %clang -### -c -integrated-as -Wa,-compress-debug-sections %s 2>&1 | FileCheck --check-prefix=COMPRESS_DEBUG %s
-// RUN: %clang -### -c -integrated-as -Wa,--compress-debug-sections %s 2>&1 | FileCheck --check-prefix=COMPRESS_DEBUG %s
 // REQUIRES: zlib
 
-// COMPRESS_DEBUG: "-compress-debug-sections"
+// RUN: %clang -### -fintegrated-as -Wa,-compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-_COMPRESS_DEBUG_SECTIONS %s
+// CHECK-_COMPRESS_DEBUG_SECTIONS: "-compress-debug-sections"
+
+// RUN: %clang -### -fintegrated-as -Wa,--compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-__COMPRESS_DEBUG_SECTIONS %s
+// CHECK-__COMPRESS_DEBUG_SECTIONS: "--compress-debug-sections"
+
+// RUN: %clang -### -fintegrated-as -Wa,--compress-debug-sections -Wa,--nocompress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-POSNEG %s
+// CHECK-POSNEG: "--compress-debug-sections"
+// CHECK-POSNEG: "--nocompress-debug-sections"
+
+// RUN: %clang -### -fintegrated-as -Wa,-compress-debug-sections -Wa,--compress-debug-sections -c %s 2>&1 | FileCheck -check-prefix CHECK-MULTIPLE %s
+// CHECK-MULTIPLE: "-compress-debug-sections"
+// CHECK-MULTIPLE: "--compress-debug-sections"
+
+// RUN: %clang -### -fintegrated-as -gz -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ %s
+// RUN: %clang -### -fintegrated-as -gz -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ %s
+// CHECK-OPT_GZ: "-compress-debug-sections"
+
+// RUN: %clang -### -fintegrated-as -gz=none -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_NONE %s
+// RUN: %clang -### -fintegrated-as -gz=none -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_NONE %s
+// CHECK-OPT_GZ_EQ_NONE: "-compress-debug-sections=none"
+
+// RUN: %clang -### -fintegrated-as -gz=zlib -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB %s
+// RUN: %clang -### -fintegrated-as -gz=zlib -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB %s
+// CHECK-OPT_GZ_EQ_ZLIB: "-compress-debug-sections=zlib"
+
+// RUN: %clang -### -fintegrated-as -gz=zlib-gnu -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB_GNU %s
+// RUN: %clang -### -fintegrated-as -gz=zlib-gnu -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_ZLIB_GNU %s
+// CHECK-OPT_GZ_EQ_ZLIB_GNU: "-compress-debug-sections=zlib-gnu"
+
+// RUN: %clang -### -fintegrated-as -gz=invalid -x assembler -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_INVALID %s
+// RUN: %clang -### -fintegrated-as -gz=invalid -c %s 2>&1 | FileCheck -check-prefix CHECK-OPT_GZ_EQ_INVALID %s
+// CHECK-OPT_GZ_EQ_INVALID: error: unsupported argument 'invalid' to option 'gz='
 
-// RUN: %clang -### -c -integrated-as -Wa,--compress-debug-sections -Wa,--nocompress-debug-sections %s 2>&1 | FileCheck --check-prefix=NOCOMPRESS_DEBUG %s
-// NOCOMPRESS_DEBUG-NOT: "-compress-debug-sections"
diff --git a/test/Driver/fsanitize-object-size.c b/test/Driver/fsanitize-object-size.c
new file mode 100644
index 000000000000..b96221e0fd43
--- /dev/null
+++ b/test/Driver/fsanitize-object-size.c
@@ -0,0 +1,31 @@
+// Check that the object size check is disabled at -O0.
+//
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size %s -O0 -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=null,object-size %s -### 2>&1 | FileCheck %s --check-prefixes=CHECK-NO-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined %s -### 2>&1 | FileCheck %s --check-prefixes=CHECK-NO-OSIZE-NO-WARNING
+
+// Check that the object size check is enabled at other optimization levels.
+//
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined -O1 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -O2 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -O3 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -O4 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -Ofast %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -Os %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -Oz %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=object-size -Og %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+
+// Use of trap mode shouldn't affect the object size check.
+//
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined -fsanitize-trap=undefined -O1 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined-trap -O1 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined-trap -fsanitize-undefined-trap-on-error -O1 %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-HAS-OSIZE
+
+// CHECK-HAS-OSIZE-NOT: warning: the object size sanitizer
+// CHECK-HAS-OSIZE: -fsanitize={{[^ ]*}}object-size
+
+// CHECK-NO-OSIZE: warning: the object size sanitizer has no effect at -O0, but is explicitly enabled: -fsanitize={{[^ ]*}}object-size
+// CHECK-NO-OSIZE-NOT: -fsanitize={{[^ ]*}}object-size
+
+// CHECK-NO-OSIZE-NO-WARNING-NOT: -fsanitize={{[^ ]*}}object-size
diff --git a/test/Driver/fsanitize.c b/test/Driver/fsanitize.c
index f14459df6391..0752ef6df090 100644
--- a/test/Driver/fsanitize.c
+++ b/test/Driver/fsanitize.c
@@ -3,18 +3,18 @@
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined -fsanitize-undefined-trap-on-error %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-TRAP
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined-trap -fsanitize-undefined-trap-on-error %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-TRAP
 // RUN: %clang -target x86_64-linux-gnu -fsanitize-undefined-trap-on-error -fsanitize=undefined-trap %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-TRAP
-// CHECK-UNDEFINED-TRAP: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute|function),?){19}"}}
-// CHECK-UNDEFINED-TRAP: "-fsanitize-trap=alignment,array-bounds,bool,enum,float-cast-overflow,float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,null,object-size,pointer-overflow,return,returns-nonnull-attribute,shift-base,shift-exponent,signed-integer-overflow,unreachable,vla-bound"
-// CHECK-UNDEFINED-TRAP2: "-fsanitize-trap=alignment,array-bounds,bool,enum,float-cast-overflow,float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,null,object-size,pointer-overflow,return,returns-nonnull-attribute,shift-base,shift-exponent,unreachable,vla-bound"
+// CHECK-UNDEFINED-TRAP: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute|function),?){18}"}}
+// CHECK-UNDEFINED-TRAP: "-fsanitize-trap=alignment,array-bounds,bool,enum,float-cast-overflow,float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,null,pointer-overflow,return,returns-nonnull-attribute,shift-base,shift-exponent,signed-integer-overflow,unreachable,vla-bound"
+// CHECK-UNDEFINED-TRAP2: "-fsanitize-trap=alignment,array-bounds,bool,enum,float-cast-overflow,float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,null,pointer-overflow,return,returns-nonnull-attribute,shift-base,shift-exponent,unreachable,vla-bound"
 
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED
-// CHECK-UNDEFINED: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|vptr|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){20}"}}
+// CHECK-UNDEFINED: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|vptr|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){19}"}}
 
 // RUN: %clang -target x86_64-apple-darwin10 -fsanitize=undefined %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-DARWIN
-// CHECK-UNDEFINED-DARWIN: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){18}"}}
+// CHECK-UNDEFINED-DARWIN: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){17}"}}
 
 // RUN: %clang -target i386-unknown-openbsd -fsanitize=undefined %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-OPENBSD
-// CHECK-UNDEFINED-OPENBSD: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){18}"}}
+// CHECK-UNDEFINED-OPENBSD: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){17}"}}
 
 // RUN: %clang -target i386-pc-win32 -fsanitize=undefined %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-WIN --check-prefix=CHECK-UNDEFINED-WIN32
 // RUN: %clang -target i386-pc-win32 -fsanitize=undefined -x c++ %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-UNDEFINED-WIN --check-prefix=CHECK-UNDEFINED-WIN32 --check-prefix=CHECK-UNDEFINED-WIN-CXX
@@ -23,7 +23,7 @@
 // CHECK-UNDEFINED-WIN32: "--dependent-lib={{[^"]*}}ubsan_standalone-i386.lib"
 // CHECK-UNDEFINED-WIN64: "--dependent-lib={{[^"]*}}ubsan_standalone-x86_64.lib"
 // CHECK-UNDEFINED-WIN-CXX: "--dependent-lib={{[^"]*}}ubsan_standalone_cxx{{[^"]*}}.lib"
-// CHECK-UNDEFINED-WIN-SAME: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){18}"}}
+// CHECK-UNDEFINED-WIN-SAME: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){17}"}}
 
 // RUN: %clang -target i386-pc-win32 -fsanitize-coverage=bb %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-COVERAGE-WIN32
 // CHECK-COVERAGE-WIN32: "--dependent-lib={{[^"]*}}ubsan_standalone-i386.lib"
@@ -43,7 +43,7 @@
 // CHECK-FNO-SANITIZE-ALL: "-fsanitize=thread"
 
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=thread,undefined -fno-sanitize=thread -fno-sanitize=float-cast-overflow,vptr,bool,enum %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-PARTIAL-UNDEFINED
-// CHECK-PARTIAL-UNDEFINED: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|object-size|pointer-overflow|array-bounds|returns-nonnull-attribute|nonnull-attribute),?){16}"}}
+// CHECK-PARTIAL-UNDEFINED: "-fsanitize={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|unreachable|return|vla-bound|alignment|null|pointer-overflow|array-bounds|returns-nonnull-attribute|nonnull-attribute),?){15}"}}
 
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=shift -fno-sanitize=shift-base %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-FSANITIZE-SHIFT-PARTIAL
 // CHECK-FSANITIZE-SHIFT-PARTIAL: "-fsanitize=shift-exponent"
@@ -217,11 +217,11 @@
 // RUN: %clang -target x86_64-linux-gnu %s -fsanitize=undefined -fno-sanitize-recover=undefined -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-RECOVER-UBSAN
 // RUN: %clang -target x86_64-linux-gnu %s -fsanitize=undefined -fno-sanitize-recover=all -fsanitize-recover=thread -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-RECOVER-UBSAN
 // RUN: %clang -target x86_64-linux-gnu %s -fsanitize=undefined -fsanitize-recover=all -fno-sanitize-recover=undefined -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-RECOVER-UBSAN
-// CHECK-RECOVER-UBSAN: "-fsanitize-recover={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|vla-bound|alignment|null|vptr|object-size|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){18}"}}
+// CHECK-RECOVER-UBSAN: "-fsanitize-recover={{((signed-integer-overflow|integer-divide-by-zero|float-divide-by-zero|function|shift-base|shift-exponent|vla-bound|alignment|null|vptr|pointer-overflow|float-cast-overflow|array-bounds|enum|bool|returns-nonnull-attribute|nonnull-attribute),?){17}"}}
 // CHECK-NO-RECOVER-UBSAN-NOT: sanitize-recover
 
 // RUN: %clang -target x86_64-linux-gnu %s -fsanitize=undefined -fno-sanitize-recover=all -fsanitize-recover=object-size,shift-base -### 2>&1 | FileCheck %s --check-prefix=CHECK-PARTIAL-RECOVER
-// CHECK-PARTIAL-RECOVER: "-fsanitize-recover={{((object-size|shift-base),?){2}"}}
+// CHECK-PARTIAL-RECOVER: "-fsanitize-recover={{((shift-base),?){1}"}}
 
 // RUN: %clang -target x86_64-linux-gnu %s -fsanitize=address -fsanitize-recover=all -### 2>&1 | FileCheck %s --check-prefix=CHECK-RECOVER-ASAN
 // CHECK-RECOVER-ASAN: "-fsanitize-recover=address"
diff --git a/test/Driver/nozlibcompress.c b/test/Driver/nozlibcompress.c
index 9986c85d79ae..41e1794ad9c3 100644
--- a/test/Driver/nozlibcompress.c
+++ b/test/Driver/nozlibcompress.c
@@ -1,6 +1,7 @@
-// RUN: %clang -c %s -Wa,--compress-debug-sections 2>&1 | FileCheck %s
-// RUN: %clang -c %s -Wa,--compress-debug-sections -Wa,--nocompress-debug-sections 2>&1 | FileCheck --allow-empty --check-prefix=NOWARN %s
 // REQUIRES: nozlib
 
-// CHECK: warning: cannot compress debug sections (zlib not installed)
-// NOWARN-NOT: warning: cannot compress debug sections (zlib not installed)
+// RUN: %clang -### -fintegrated-as -gz -c %s 2>&1 | FileCheck %s -check-prefix CHECK-WARN
+// RUN: %clang -### -fintegrated-as -gz=none -c %s 2>&1 | FileCheck -allow-empty -check-prefix CHECK-NOWARN %s
+
+// CHECK-WARN: warning: cannot compress debug sections (zlib not installed)
+// CHECK-NOWARN-NOT: warning: cannot compress debug sections (zlib not installed)
diff --git a/test/Driver/wasm-toolchain.c b/test/Driver/wasm-toolchain.c
index 3be60df92672..8debc0287007 100644
--- a/test/Driver/wasm-toolchain.c
+++ b/test/Driver/wasm-toolchain.c
@@ -27,18 +27,18 @@
 
 // RUN: %clang -### -no-canonical-prefixes -target wasm32-unknown-unknown --sysroot=/foo %s 2>&1 | FileCheck -check-prefix=LINK %s
 // LINK: clang{{.*}}" "-cc1" {{.*}} "-o" "[[temp:[^"]*]]"
-// LINK: lld{{.*}}" "-flavor" "wasm" "-L/foo/lib32" "crt1.o" "crti.o" "[[temp]]" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
+// LINK: lld{{.*}}" "-flavor" "wasm" "-L/foo/lib" "crt1.o" "crti.o" "[[temp]]" "-allow-undefined-file" "wasm.syms" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
 
 // A basic C link command-line with optimization. WebAssembly is somewhat
 // special in enabling --gc-sections by default.
 
 // RUN: %clang -### -O2 -no-canonical-prefixes -target wasm32-unknown-unknown --sysroot=/foo %s 2>&1 | FileCheck -check-prefix=LINK_OPT %s
 // LINK_OPT: clang{{.*}}" "-cc1" {{.*}} "-o" "[[temp:[^"]*]]"
-// LINK_OPT: lld{{.*}}" "-flavor" "wasm" "--gc-sections" "-L/foo/lib32" "crt1.o" "crti.o" "[[temp]]" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
+// LINK_OPT: lld{{.*}}" "-flavor" "wasm" "--gc-sections" "-L/foo/lib" "crt1.o" "crti.o" "[[temp]]" "-allow-undefined-file" "wasm.syms" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
 
 // Ditto, but ensure that a user --no-gc-sections comes after the
 // default --gc-sections.
 
 // RUN: %clang -### -O2 -no-canonical-prefixes -target wasm32-unknown-unknown --sysroot=/foo -Wl,--no-gc-sections %s 2>&1 | FileCheck -check-prefix=NO_GC_SECTIONS %s
 // NO_GC_SECTIONS: clang{{.*}}" "-cc1" {{.*}} "-o" "[[temp:[^"]*]]"
-// NO_GC_SECTIONS: lld{{.*}}" "-flavor" "wasm" "--gc-sections" "-L/foo/lib32" "crt1.o" "crti.o" "--no-gc-sections" "[[temp]]" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
+// NO_GC_SECTIONS: lld{{.*}}" "-flavor" "wasm" "--gc-sections" "-L/foo/lib" "crt1.o" "crti.o" "--no-gc-sections" "[[temp]]" "-allow-undefined-file" "wasm.syms" "-lc" "-lcompiler_rt" "crtn.o" "-o" "a.out"
diff --git a/test/FixIt/fixit-format-darwin.m b/test/FixIt/fixit-format-darwin.m
index bfc71291a5c3..077cc0cf21bc 100644
--- a/test/FixIt/fixit-format-darwin.m
+++ b/test/FixIt/fixit-format-darwin.m
@@ -57,3 +57,20 @@ void test() {
   Log3("test 7: %s", getNSInteger(), getNSUInteger());
   // CHECK: Log3("test 7: %ld", (long)getNSInteger(), (unsigned long)getNSUInteger());
 }
+
+#define Outer1(...) \
+do { \
+  printf(__VA_ARGS__); \
+} while (0)
+
+#define Outer2(...) \
+do { \
+  Outer1(__VA_ARGS__); Outer1(__VA_ARGS__); \
+} while (0)
+
+void bug33447() {
+  Outer2("test 8: %s", getNSInteger());  
+  // CHECK: Outer2("test 8: %ld", (long)getNSInteger());
+  Outer2("test 9: %s %s", getNSInteger(), getNSInteger());
+  // CHECK: Outer2("test 9: %ld %ld", (long)getNSInteger(), (long)getNSInteger());
+}
diff --git a/test/Frontend/Inputs/optimization-remark-with-hotness-sample.proftext b/test/Frontend/Inputs/optimization-remark-with-hotness-sample.proftext
new file mode 100644
index 000000000000..aeea583de4d4
--- /dev/null
+++ b/test/Frontend/Inputs/optimization-remark-with-hotness-sample.proftext
@@ -0,0 +1,7 @@
+foo:0:0
+ 0: 0
+bar:29:29
+ 6: foo:0
+main:0:0
+ 0: 0 bar:0
+
diff --git a/test/Frontend/optimization-remark-with-hotness.c b/test/Frontend/optimization-remark-with-hotness.c
index 708f5ec8d4bf..30ead64b8eb5 100644
--- a/test/Frontend/optimization-remark-with-hotness.c
+++ b/test/Frontend/optimization-remark-with-hotness.c
@@ -1,11 +1,21 @@
+// Generate instrumentation and sampling profile data.
 // RUN: llvm-profdata merge \
-// RUN:     %S/Inputs/optimization-remark-with-hotness.proftext   \
+// RUN:     %S/Inputs/optimization-remark-with-hotness.proftext \
 // RUN:     -o %t.profdata
+// RUN: llvm-profdata merge -sample \
+// RUN:     %S/Inputs/optimization-remark-with-hotness-sample.proftext \
+// RUN:     -o %t-sample.profdata
+//
 // RUN: %clang_cc1 -triple x86_64-apple-macosx10.9 -main-file-name \
 // RUN:     optimization-remark-with-hotness.c %s -emit-llvm-only \
 // RUN:     -fprofile-instrument-use-path=%t.profdata -Rpass=inline \
 // RUN:     -Rpass-analysis=inline -Rpass-missed=inline \
 // RUN:     -fdiagnostics-show-hotness -verify
+// RUN: %clang_cc1 -triple x86_64-apple-macosx10.9 -main-file-name \
+// RUN:     optimization-remark-with-hotness.c %s -emit-llvm-only \
+// RUN:     -fprofile-sample-use=%t-sample.profdata -Rpass=inline \
+// RUN:     -Rpass-analysis=inline -Rpass-missed=inline \
+// RUN:     -fdiagnostics-show-hotness -verify
 // The clang version of the previous test.
 // RUN: %clang -target x86_64-apple-macosx10.9 %s -c -emit-llvm -o /dev/null \
 // RUN:     -fprofile-instr-use=%t.profdata -Rpass=inline \
diff --git a/test/Frontend/pp-only-no-editor-placeholders.c b/test/Frontend/pp-only-no-editor-placeholders.c
new file mode 100644
index 000000000000..515d9893fd76
--- /dev/null
+++ b/test/Frontend/pp-only-no-editor-placeholders.c
@@ -0,0 +1,4 @@
+// RUN: %clang_cc1 -E -verify -o - %s | FileCheck %s
+// expected-no-diagnostics
+
+<#placeholder#>; // CHECK: <#placeholder#>;
diff --git a/test/Import/indirect-struct-member-access/Inputs/S.c b/test/Import/indirect-struct-member-access/Inputs/S.c
new file mode 100644
index 000000000000..b0876d27df44
--- /dev/null
+++ b/test/Import/indirect-struct-member-access/Inputs/S.c
@@ -0,0 +1,3 @@
+struct S {
+  int a;
+};
diff --git a/test/Import/indirect-struct-member-access/test.c b/test/Import/indirect-struct-member-access/test.c
new file mode 100644
index 000000000000..cbf7e65ec1cc
--- /dev/null
+++ b/test/Import/indirect-struct-member-access/test.c
@@ -0,0 +1,4 @@
+// RUN: clang-import-test -import %S/Inputs/S.c -expression %s
+void expr(struct S *MyS) {
+  MyS->a = 3;
+}
diff --git a/test/Index/Core/index-source.cpp b/test/Index/Core/index-source.cpp
index 10f2d8f77747..6d20fdd48e50 100644
--- a/test/Index/Core/index-source.cpp
+++ b/test/Index/Core/index-source.cpp
@@ -134,6 +134,9 @@ class PseudoOverridesInSpecializations {
 
   template<typename U> struct InnerTemplate { };
   template<typename U> struct InnerTemplate <U*> { };
+
+  template<typename U>
+  class InnerClass { };
 };
 
 template<>
@@ -195,7 +198,21 @@ class PseudoOverridesInSpecializations<double, int> {
 // CHECK-NEXT: RelChild
 // CHECK-NEXT: RelSpecialization | InnerTemplate | c:@ST>2#T#T@PseudoOverridesInSpecializations@ST>1#T@InnerTemplate
   template<typename U> struct InnerTemplate <U*> { };
+
+  template<typename U>
+  class InnerClass;
+// CHECK: [[@LINE-1]]:9 | class(Gen)/C++ | InnerClass | c:@S@PseudoOverridesInSpecializations>#d#I@ST>1#T@InnerClass | <no-cgname> | Ref,RelCont,RelSpecialization | rel: 2
+// CHECK-NEXT: RelCont
+// CHECK-NEXT: RelSpecialization | InnerClass | c:@ST>2#T#T@PseudoOverridesInSpecializations@ST>1#T@InnerClass
+};
+
+template<typename U>
+class PseudoOverridesInSpecializations<double, int>::InnerClass {
 };
+// CHECK: [[@LINE-2]]:54 | class(Gen)/C++ | InnerClass | c:@S@PseudoOverridesInSpecializations>#d#I@ST>1#T@InnerClass | <no-cgname> | Def,RelChild | rel: 1
+// CHECK-NEXT: RelChild
+// CHECK: [[@LINE-4]]:7 | class(Gen)/C++ | PseudoOverridesInSpecializations | c:@ST>2#T#T@PseudoOverridesInSpecializations | <no-cgname> | Ref,RelCont | rel: 1
+// CHECK-NEXT: RelCont
 
 template<typename S>
 class PseudoOverridesInSpecializations<float, S> {
@@ -265,7 +282,9 @@ class SpecializationDecl { };
 
 template<>
 class SpecializationDecl<int>;
-// CHECK: [[@LINE-1]]:7 | class(Gen)/C++ | SpecializationDecl | c:@ST>1#T@SpecializationDecl | <no-cgname> | Ref | rel: 0
+// CHECK: [[@LINE-1]]:7 | class(Gen,TS)/C++ | SpecializationDecl | c:@S@SpecializationDecl>#I | <no-cgname> | Decl,RelSpecialization | rel: 1
+// CHECK-NEXT: RelSpecialization | SpecializationDecl | c:@ST>1#T@SpecializationDecl
+// CHECK: [[@LINE-3]]:7 | class(Gen)/C++ | SpecializationDecl | c:@ST>1#T@SpecializationDecl | <no-cgname> | Ref | rel: 0
 
 template<>
 class SpecializationDecl<int> { };
@@ -275,8 +294,10 @@ class SpecializationDecl<int> { };
 
 template<typename T>
 class PartialSpecilizationClass<Cls, T>;
-// CHECK: [[@LINE-1]]:7 | class(Gen)/C++ | PartialSpecilizationClass | c:@ST>2#T#T@PartialSpecilizationClass | <no-cgname> | Ref | rel: 0
-// CHECK-NEXT: [[@LINE-2]]:33 | class/C++ | Cls | c:@S@Cls | <no-cgname> | Ref | rel: 0
+// CHECK: [[@LINE-1]]:7 | class(Gen,TPS)/C++ | PartialSpecilizationClass | c:@SP>1#T@PartialSpecilizationClass>#$@S@Cls#t0.0 | <no-cgname> | Decl,RelSpecialization | rel: 1
+// CHECK-NEXT: RelSpecialization | PartialSpecilizationClass | c:@ST>2#T#T@PartialSpecilizationClass
+// CHECK: [[@LINE-3]]:7 | class(Gen)/C++ | PartialSpecilizationClass | c:@ST>2#T#T@PartialSpecilizationClass | <no-cgname> | Ref | rel: 0
+// CHECK-NEXT: [[@LINE-4]]:33 | class/C++ | Cls | c:@S@Cls | <no-cgname> | Ref | rel: 0
 
 template<>
 class PartialSpecilizationClass<Cls, Cls> : Cls { };
diff --git a/test/Index/singe-file-parse.m b/test/Index/singe-file-parse.m
deleted file mode 100644
index d13915b30c5f..000000000000
--- a/test/Index/singe-file-parse.m
+++ /dev/null
@@ -1,11 +0,0 @@
-// RUN: c-index-test -single-file-parse %s | FileCheck %s
-
-#include <stdint.h>
-
-// CHECK-NOT: TypedefDecl=intptr_t
-
-// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=MyCls
-@interface MyCls
-// CHECK: [[@LINE+1]]:8: ObjCInstanceMethodDecl=some_meth
--(void)some_meth;
-@end
diff --git a/test/Index/single-file-parse.m b/test/Index/single-file-parse.m
new file mode 100644
index 000000000000..f75b9bd0ee54
--- /dev/null
+++ b/test/Index/single-file-parse.m
@@ -0,0 +1,121 @@
+// RUN: c-index-test -single-file-parse %s | FileCheck %s
+
+#include <stdint.h>
+
+// CHECK-NOT: TypedefDecl=intptr_t
+
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=MyCls
+@interface MyCls
+// CHECK: [[@LINE+1]]:8: ObjCInstanceMethodDecl=some_meth
+-(void)some_meth;
+@end
+
+#if 1
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test1
+@interface Test1 @end
+#else
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test2 @end
+#endif
+
+#if 0
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test3 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test4
+@interface Test4 @end
+#endif
+
+#if SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test5
+@interface Test5 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test6
+@interface Test6 @end
+#endif
+
+#define SOMETHING_DEFINED 1
+#if SOMETHING_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test7
+@interface Test7 @end
+#else
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test8 @end
+#endif
+
+#if defined(SOMETHING_NOT_DEFINED)
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test9
+@interface Test9 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test10
+@interface Test10 @end
+#endif
+
+#if defined(SOMETHING_DEFINED)
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test11
+@interface Test11 @end
+#else
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test12 @end
+#endif
+
+#if SOMETHING_NOT_DEFINED1
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test13
+@interface Test13 @end
+#elif SOMETHING_NOT_DEFINED2
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test14
+@interface Test14 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test15
+@interface Test15 @end
+#endif
+
+#ifdef SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test19
+@interface Test19 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test20
+@interface Test20 @end
+#endif
+
+#ifdef SOMETHING_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test21
+@interface Test21 @end
+#else
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test22 @end
+#endif
+
+#ifndef SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test23
+@interface Test23 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test24
+@interface Test24 @end
+#endif
+
+#ifndef SOMETHING_DEFINED
+// CHECK-NOT: [[@LINE+1]]:12:
+@interface Test25 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test26
+@interface Test26 @end
+#endif
+
+#if 1 < SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test27
+@interface Test27 @end
+#else
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test28
+@interface Test28 @end
+#endif
+
+#if SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test29
+@interface Test29 @end
+#endif
+
+#ifdef SOMETHING_NOT_DEFINED
+// CHECK: [[@LINE+1]]:12: ObjCInterfaceDecl=Test30
+@interface Test30 @end
+#endif
diff --git a/test/Misc/Inputs/module.modulemap b/test/Misc/Inputs/module.modulemap
new file mode 100644
index 000000000000..a8ecb09390a2
--- /dev/null
+++ b/test/Misc/Inputs/module.modulemap
@@ -0,0 +1 @@
+module X {}
diff --git a/test/Misc/ast-dump-decl.c b/test/Misc/ast-dump-decl.c
index 45edea26b4f8..313c808c4ac8 100644
--- a/test/Misc/ast-dump-decl.c
+++ b/test/Misc/ast-dump-decl.c
@@ -1,8 +1,13 @@
 // RUN: %clang_cc1 -triple x86_64-unknown-unknown -ast-dump -ast-dump-filter Test %s | FileCheck -check-prefix CHECK -strict-whitespace %s
 // RUN: %clang_cc1 -triple x86_64-unknown-unknown -ast-dump %s | FileCheck -check-prefix CHECK-TU -strict-whitespace %s
+// RUN: %clang_cc1 -fmodules -fmodules-local-submodule-visibility -fmodule-name=X -triple x86_64-unknown-unknown -fmodule-map-file=%S/Inputs/module.modulemap -ast-dump -ast-dump-filter Test %s -DMODULES | FileCheck -check-prefix CHECK -check-prefix CHECK-MODULES -strict-whitespace %s
 
 int TestLocation;
-// CHECK: VarDecl 0x{{[^ ]*}} <{{.*}}:4:1, col:5> col:5 TestLocation
+// CHECK: VarDecl 0x{{[^ ]*}} <{{.*}}:[[@LINE-1]]:1, col:5> col:5 TestLocation
+
+#ifdef MODULES
+#pragma clang module begin X
+#endif
 
 struct TestIndent {
   int x;
@@ -33,7 +38,7 @@ typedef int TestTypedefDecl;
 // CHECK:      TypedefDecl{{.*}} TestTypedefDecl 'int'
 
 __module_private__ typedef int TestTypedefDeclPrivate;
-// CHECK:      TypedefDecl{{.*}} TestTypedefDeclPrivate 'int' __module_private__
+// CHECK-MODULE:      TypedefDecl{{.*}} TestTypedefDeclPrivate 'int' __module_private__
 
 enum TestEnumDecl {
   testEnumDecl
@@ -53,7 +58,7 @@ enum TestEnumDeclForward;
 // CHECK:      EnumDecl{{.*}} TestEnumDeclForward
 
 __module_private__ enum TestEnumDeclPrivate;
-// CHECK:      EnumDecl{{.*}} TestEnumDeclPrivate __module_private__
+// CHECK-MODULE:      EnumDecl{{.*}} TestEnumDeclPrivate __module_private__
 
 struct TestRecordDecl {
   int i;
@@ -83,7 +88,7 @@ struct TestRecordDeclForward;
 // CHECK:      RecordDecl{{.*}} struct TestRecordDeclForward
 
 __module_private__ struct TestRecordDeclPrivate;
-// CHECK:      RecordDecl{{.*}} struct TestRecordDeclPrivate __module_private__
+// CHECK-MODULE:      RecordDecl{{.*}} struct TestRecordDeclPrivate __module_private__
 
 enum testEnumConstantDecl {
   TestEnumConstantDecl,
@@ -136,7 +141,7 @@ struct testFieldDecl {
 // CHECK:      FieldDecl{{.*}} TestFieldDecl 'int'
 // CHECK:      FieldDecl{{.*}} TestFieldDeclWidth 'int'
 // CHECK-NEXT:   IntegerLiteral
-// CHECK:      FieldDecl{{.*}} TestFieldDeclPrivate 'int' __module_private__
+// CHECK-MODULE:      FieldDecl{{.*}} TestFieldDeclPrivate 'int' __module_private__
 
 int TestVarDecl;
 // CHECK:      VarDecl{{.*}} TestVarDecl 'int'
@@ -148,7 +153,7 @@ __thread int TestVarDeclThread;
 // CHECK:      VarDecl{{.*}} TestVarDeclThread 'int' tls{{$}}
 
 __module_private__ int TestVarDeclPrivate;
-// CHECK:      VarDecl{{.*}} TestVarDeclPrivate 'int' __module_private__
+// CHECK-MODULE:      VarDecl{{.*}} TestVarDeclPrivate 'int' __module_private__
 
 int TestVarDeclInit = 0;
 // CHECK:      VarDecl{{.*}} TestVarDeclInit 'int'
@@ -156,3 +161,8 @@ int TestVarDeclInit = 0;
 
 void testParmVarDecl(int TestParmVarDecl);
 // CHECK: ParmVarDecl{{.*}} TestParmVarDecl 'int'
+
+#ifdef MODULES
+#pragma clang module end
+#endif
+
diff --git a/test/Misc/ast-dump-decl.cpp b/test/Misc/ast-dump-decl.cpp
index e1cdeb0995fa..7370f46c5b32 100644
--- a/test/Misc/ast-dump-decl.cpp
+++ b/test/Misc/ast-dump-decl.cpp
@@ -95,17 +95,12 @@ class TestCXXRecordDeclPack : public T... {
 thread_local int TestThreadLocalInt;
 // CHECK: TestThreadLocalInt {{.*}} tls_dynamic
 
-__module_private__ class TestCXXRecordDeclPrivate;
-// CHECK: CXXRecordDecl{{.*}} class TestCXXRecordDeclPrivate __module_private__
-
 class testCXXMethodDecl {
-  __module_private__ void TestCXXMethodDeclPrivate();
   virtual void TestCXXMethodDeclPure() = 0;
   void TestCXXMethodDeclDelete() = delete;
   void TestCXXMethodDeclThrow() throw();
   void TestCXXMethodDeclThrowType() throw(int);
 };
-// CHECK: CXXMethodDecl{{.*}} TestCXXMethodDeclPrivate 'void (void)' __module_private__
 // CHECK: CXXMethodDecl{{.*}} TestCXXMethodDeclPure 'void (void)' virtual pure
 // CHECK: CXXMethodDecl{{.*}} TestCXXMethodDeclDelete 'void (void)' delete
 // CHECK: CXXMethodDecl{{.*}} TestCXXMethodDeclThrow 'void (void) throw()'
diff --git a/test/Misc/cc1as-compress.s b/test/Misc/cc1as-compress.s
new file mode 100644
index 000000000000..beba227138d8
--- /dev/null
+++ b/test/Misc/cc1as-compress.s
@@ -0,0 +1,8 @@
+// REQUIRES: zlib
+// REQUIRES: x86-registered-target
+
+// RUN: %clang -cc1as -triple i686 --compress-debug-sections -filetype asm %s -o /dev/null 2>&1 | FileCheck -allow-empty %s
+// RUN: %clang -cc1as -triple i686 -compress-debug-sections -filetype asm %s -o /dev/null 2>&1 | FileCheck -allow-empty %s
+
+// CHECK-NOT: error: unknown argument:
+
diff --git a/test/Modules/Inputs/preprocess/file.h b/test/Modules/Inputs/preprocess/file.h
index 808ade5768b1..84cf22a33740 100644
--- a/test/Modules/Inputs/preprocess/file.h
+++ b/test/Modules/Inputs/preprocess/file.h
@@ -1,3 +1,9 @@
+#include "other.h"
+
+#ifndef FILE_H
+#define FILE_H
 struct __FILE;
 #include "fwd.h"
 typedef struct __FILE FILE;
+typedef foo bar;
+#endif
diff --git a/test/Modules/Inputs/preprocess/fwd.h b/test/Modules/Inputs/preprocess/fwd.h
index 4a19c6d0c057..f6de1800c000 100644
--- a/test/Modules/Inputs/preprocess/fwd.h
+++ b/test/Modules/Inputs/preprocess/fwd.h
@@ -1 +1,2 @@
+typedef struct foo foo;
 struct __FILE;
diff --git a/test/Modules/Inputs/preprocess/module.modulemap b/test/Modules/Inputs/preprocess/module.modulemap
index f700db03beac..5be2e5c4ff98 100644
--- a/test/Modules/Inputs/preprocess/module.modulemap
+++ b/test/Modules/Inputs/preprocess/module.modulemap
@@ -1,5 +1,5 @@
 module fwd { header "fwd.h" export * }
-module file { header "file.h" header "file2.h" export * }
+module file { header "file.h" header "file2.h" header "other.h" export * }
 module nested {
   module a { header "a.h" }
   module b { header "b.h" }
diff --git a/test/Modules/Inputs/preprocess/other.h b/test/Modules/Inputs/preprocess/other.h
new file mode 100644
index 000000000000..84c4d1d5eb23
--- /dev/null
+++ b/test/Modules/Inputs/preprocess/other.h
@@ -0,0 +1 @@
+// other.h: empty
diff --git a/test/Modules/const-var-init-update.cpp b/test/Modules/const-var-init-update.cpp
new file mode 100644
index 000000000000..61080eb83917
--- /dev/null
+++ b/test/Modules/const-var-init-update.cpp
@@ -0,0 +1,30 @@
+// RUN: %clang_cc1 -std=c++1z -fmodules %s -verify
+// expected-no-diagnostics
+
+#pragma clang module build std
+module std { module limits {} module other {} }
+#pragma clang module contents
+#pragma clang module begin std.limits
+template<typename T> struct numeric_limits {
+  static constexpr T __max = 5;
+  static constexpr T max() { return __max; }
+};
+#pragma clang module end
+#pragma clang module begin std.other
+inline void f() { numeric_limits<int> nl; }
+#pragma clang module end
+#pragma clang module endbuild
+
+#pragma clang module build module_b
+module module_b {}
+#pragma clang module contents
+#pragma clang module begin module_b
+#pragma clang module import std.limits
+constexpr int a = numeric_limits<int>::max();
+#pragma clang module end
+#pragma clang module endbuild
+
+#pragma clang module import std.limits
+#pragma clang module import module_b
+constexpr int b = a;
+static_assert(b == 5);
diff --git a/test/Modules/interface-visibility.m b/test/Modules/interface-visibility.m
new file mode 100644
index 000000000000..2bb124ce0956
--- /dev/null
+++ b/test/Modules/interface-visibility.m
@@ -0,0 +1,29 @@
+// RUN: %clang_cc1 -fmodules -fobjc-arc -x objective-c-module-map %s -fmodule-name=Foo -verify
+
+module Foo {}
+
+#pragma clang module contents
+#pragma clang module begin Foo
+
+// expected-no-diagnostics
+
+#pragma clang module build Foundation
+module Foundation {}
+#pragma clang module contents
+#pragma clang module begin Foundation
+@interface NSIndexSet
+@end
+#pragma clang module end
+#pragma clang module endbuild
+
+#pragma clang module import Foundation
+
+@interface NSIndexSet (Testing)
+- (int)foo;
+@end
+
+static inline int test(NSIndexSet *obj) {
+  return [obj foo];
+}
+
+#pragma clang module end
diff --git a/test/Modules/preprocess-module.cpp b/test/Modules/preprocess-module.cpp
index 000290fc971b..b0cbac18e780 100644
--- a/test/Modules/preprocess-module.cpp
+++ b/test/Modules/preprocess-module.cpp
@@ -29,15 +29,15 @@
 // RUN: %clang_cc1 -fmodules -fmodule-file=%t/rewrite.pcm %s -I%t -verify -fno-modules-error-recovery -DREWRITE -DINCLUDE -I%S/Inputs/preprocess
 
 // Now try building the module when the header files are missing.
-// RUN: cp %S/Inputs/preprocess/fwd.h %S/Inputs/preprocess/file.h %S/Inputs/preprocess/file2.h %S/Inputs/preprocess/module.modulemap %t
+// RUN: cp %S/Inputs/preprocess/fwd.h %S/Inputs/preprocess/file.h %S/Inputs/preprocess/file2.h %S/Inputs/preprocess/other.h %S/Inputs/preprocess/module.modulemap %t
 // RUN: %clang_cc1 -fmodules -fmodule-name=file -fmodule-file=%t/fwd.pcm -I%t -x c++-module-map %t/module.modulemap -E -frewrite-includes -o %t/copy.ii
-// RUN: rm %t/fwd.h %t/file.h %t/file2.h %t/module.modulemap
+// RUN: rm %t/fwd.h %t/file.h %t/file2.h %t/other.h %t/module.modulemap
 // RUN: %clang_cc1 -fmodules -fmodule-name=file -fmodule-file=%t/fwd.pcm -x c++-module-map-cpp-output %t/copy.ii -emit-module -o %t/copy.pcm
 
 // Check that our module contains correct mapping information for the headers.
-// RUN: cp %S/Inputs/preprocess/fwd.h %S/Inputs/preprocess/file.h %S/Inputs/preprocess/file2.h %S/Inputs/preprocess/module.modulemap %t
+// RUN: cp %S/Inputs/preprocess/fwd.h %S/Inputs/preprocess/file.h %S/Inputs/preprocess/file2.h %S/Inputs/preprocess/other.h %S/Inputs/preprocess/module.modulemap %t
 // RUN: %clang_cc1 -fmodules -fmodule-file=%t/copy.pcm %s -I%t -verify -fno-modules-error-recovery -DCOPY -DINCLUDE
-// RUN: rm %t/fwd.h %t/file.h %t/file2.h %t/module.modulemap
+// RUN: rm %t/fwd.h %t/file.h %t/file2.h %t/other.h %t/module.modulemap
 
 // Check that we can preprocess from a .pcm file and that we get the same result as preprocessing from the original sources.
 // RUN: %clang_cc1 -fmodules -fmodule-name=file -fmodule-file=%t/fwd.pcm -I%S/Inputs/preprocess -x c++-module-map %S/Inputs/preprocess/module.modulemap -emit-module -o %t/file.pcm
@@ -50,6 +50,10 @@
 // RUN: %clang_cc1 -fmodules -fmodule-file=%t/file.rewrite.pcm %s -I%t -verify -fno-modules-error-recovery -DFILE_REWRITE
 // RUN: %clang_cc1 -fmodules -fmodule-file=%t/file.rewrite.pcm %s -I%t -verify -fno-modules-error-recovery -DFILE_REWRITE -DINCLUDE -I%S/Inputs/preprocess
 //
+// Check that we can preprocess this user of the .pcm file.
+// RUN: %clang_cc1 -fmodules -fmodule-file=%t/file.pcm %s -I%t -E -frewrite-imports -o %t/preprocess-module.ii
+// RUN: %clang_cc1 -fmodules %t/preprocess-module.ii -verify -fno-modules-error-recovery -DFILE_REWRITE_FULL
+//
 // Check that language / header search options are ignored when preprocessing from a .pcm file.
 // RUN: %clang_cc1 %t/file.pcm -E -frewrite-includes -o %t/file.rewrite.ii.2
 // RUN: cmp %t/file.rewrite.ii %t/file.rewrite.ii.2
@@ -78,12 +82,15 @@
 // REWRITE: #pragma clang module begin file
 // CHECK: # 1 "{{.*}}file.h" 1
 // NO-REWRITE: #pragma clang module begin file
-// NO-REWRITE: # 1 "{{.*}}file.h"{{$}}
 //
-// CHECK: struct __FILE;
+// REWRITE: #ifndef FILE_H
+// REWRITE: #define FILE_H
+//
 // CHECK: #pragma clang module import fwd /* clang {{-E|-frewrite-includes}}: implicit import
 // CHECK: typedef struct __FILE FILE;
 //
+// REWRITE: #endif
+//
 // REWRITE: #pragma clang module end
 // CHECK: # 2 "<module-includes>" 2
 // NO-REWRITE: #pragma clang module end
@@ -105,11 +112,16 @@
 // REWRITE: #pragma clang module begin file
 // CHECK: # 1 "{{.*}}file.h" 1
 // NO-REWRITE: #pragma clang module begin file
-// NO-REWRITE: # 1 "{{.*}}file.h"{{$}}
 //
-// CHECK: struct __FILE;
-// CHECK: #pragma clang module import fwd /* clang {{-E|-frewrite-includes}}: implicit import
-// CHECK: typedef struct __FILE FILE;
+// REWRITE: #ifndef FILE_H
+// REWRITE: #define FILE_H
+// REWRITE: #if 0
+// REWRITE: #include "fwd.h"
+// REWRITE: #endif
+// REWRITE-NOT: #pragma clang module import fwd
+// REWRITE: #endif
+//
+// NO-REWRITE-NOT: struct __FILE;
 //
 // REWRITE: #pragma clang module end
 // CHECK: # 2 "{{.*}}file2.h" 2
@@ -124,15 +136,17 @@
 // NO-REWRITE: #pragma clang module end
 
 
-__FILE *a; // expected-error {{declaration of '__FILE' must be imported}}
+__FILE *a; // expected-error-re {{{{declaration of '__FILE' must be imported|unknown type name '__FILE'}}}}
 #if FILE_REWRITE
-// expected-note@file.rewrite.ii:1 {{here}}
+// expected-note@file.rewrite.ii:* {{here}}
+#elif FILE_REWRITE_FULL
+// No note diagnostic at all in this case: we've built the 'file' module but not loaded it into this compilation yet.
 #elif REWRITE
-// expected-note@rewrite.ii:1 {{here}}
+// expected-note@rewrite.ii:* {{here}}
 #elif COPY
-// expected-note@copy.ii:1 {{here}}
+// expected-note@copy.ii:* {{here}}
 #else
-// expected-note@no-rewrite.ii:1 {{here}}
+// expected-note@no-rewrite.ii:* {{here}}
 #endif
 
 #ifdef INCLUDE
diff --git a/test/Modules/string_names.cpp b/test/Modules/string_names.cpp
index 43068f13c012..a6503d048d6b 100644
--- a/test/Modules/string_names.cpp
+++ b/test/Modules/string_names.cpp
@@ -1,6 +1,10 @@
 // RUN: rm -rf %t
 // RUN: %clang_cc1 -fmodules-cache-path=%t -fmodules -fimplicit-module-maps -fmodules-decluse -I %S/Inputs/string_names %s -fmodule-name="my/module-a" -verify
 
+// Check that we can preprocess with string module names.
+// RUN: %clang_cc1 -fmodules-cache-path=%t -fmodules -fimplicit-module-maps -I %S/Inputs/string_names %s -fmodule-name="my/module-a" -E -frewrite-imports -o %t/test.ii
+// RUN: %clang_cc1 -fmodules -fimplicit-module-maps -fmodules-decluse -I %S/Inputs/string_names %t/test.ii -fmodule-name="my/module-a"
+
 #include "a.h"
 #include "b.h" // expected-error {{does not depend on a module exporting}}
 #include "c.h"
diff --git a/test/Parser/objc-at-implementation-eof-crash.m b/test/Parser/objc-at-implementation-eof-crash.m
new file mode 100644
index 000000000000..76e56c10703e
--- /dev/null
+++ b/test/Parser/objc-at-implementation-eof-crash.m
@@ -0,0 +1,24 @@
+// RUN: %clang_cc1 -verify -Wno-objc-root-class %s
+
+@interface ClassA
+
+- (void)fileExistsAtPath:(int)x;
+
+@end
+
+@interface ClassB
+
+@end
+
+@implementation ClassB // expected-note {{implementation started here}}
+
+- (void) method:(ClassA *)mgr { // expected-note {{to match this '{'}}
+  [mgr fileExistsAtPath:0
+} // expected-error {{expected ']'}}
+
+@implementation ClassC //              \
+  // expected-error {{missing '@end'}} \
+  // expected-error {{expected '}'}}   \
+  // expected-warning {{cannot find interface declaration for 'ClassC'}}
+
+@end
diff --git a/test/Parser/objc-at-interface-eof-crash.m b/test/Parser/objc-at-interface-eof-crash.m
new file mode 100644
index 000000000000..2c7bfd688f06
--- /dev/null
+++ b/test/Parser/objc-at-interface-eof-crash.m
@@ -0,0 +1,23 @@
+// RUN: %clang_cc1 -verify -Wno-objc-root-class %s
+
+@interface ClassA
+
+- (void)fileExistsAtPath:(int)x;
+
+@end
+
+@interface ClassB
+
+@end
+
+@implementation ClassB // expected-note {{implementation started here}}
+
+- (void) method:(ClassA *)mgr { // expected-note {{to match this '{'}}
+  [mgr fileExistsAtPath:0
+} // expected-error {{expected ']'}}
+
+@interface ClassC //                   \
+  // expected-error {{missing '@end'}} \
+  // expected-error {{expected '}'}}
+
+@end
diff --git a/test/Sema/asm.c b/test/Sema/asm.c
index e49a1663a896..04b7cb19eb83 100644
--- a/test/Sema/asm.c
+++ b/test/Sema/asm.c
@@ -160,6 +160,41 @@ double test15() {
   return ret;
 }
 
+void iOutputConstraint(int x){
+  __asm ("nop" : "=ir" (x) : :); // no-error
+  __asm ("nop" : "=ri" (x) : :); // no-error
+  __asm ("nop" : "=ig" (x) : :); // no-error
+  __asm ("nop" : "=im" (x) : :); // no-error
+  __asm ("nop" : "=imr" (x) : :); // no-error
+  __asm ("nop" : "=i" (x) : :); // expected-error{{invalid output constraint '=i' in asm}}
+  __asm ("nop" : "+i" (x) : :); // expected-error{{invalid output constraint '+i' in asm}}
+  __asm ("nop" : "=ii" (x) : :); // expected-error{{invalid output constraint '=ii' in asm}}
+  __asm ("nop" : "=nr" (x) : :); // no-error
+  __asm ("nop" : "=rn" (x) : :); // no-error
+  __asm ("nop" : "=ng" (x) : :); // no-error
+  __asm ("nop" : "=nm" (x) : :); // no-error
+  __asm ("nop" : "=nmr" (x) : :); // no-error
+  __asm ("nop" : "=n" (x) : :); // expected-error{{invalid output constraint '=n' in asm}}
+  __asm ("nop" : "+n" (x) : :); // expected-error{{invalid output constraint '+n' in asm}}
+  __asm ("nop" : "=nn" (x) : :); // expected-error{{invalid output constraint '=nn' in asm}}
+  __asm ("nop" : "=Fr" (x) : :); // no-error
+  __asm ("nop" : "=rF" (x) : :); // no-error
+  __asm ("nop" : "=Fg" (x) : :); // no-error
+  __asm ("nop" : "=Fm" (x) : :); // no-error
+  __asm ("nop" : "=Fmr" (x) : :); // no-error
+  __asm ("nop" : "=F" (x) : :); // expected-error{{invalid output constraint '=F' in asm}}
+  __asm ("nop" : "+F" (x) : :); // expected-error{{invalid output constraint '+F' in asm}}
+  __asm ("nop" : "=FF" (x) : :); // expected-error{{invalid output constraint '=FF' in asm}}
+  __asm ("nop" : "=Er" (x) : :); // no-error
+  __asm ("nop" : "=rE" (x) : :); // no-error
+  __asm ("nop" : "=Eg" (x) : :); // no-error
+  __asm ("nop" : "=Em" (x) : :); // no-error
+  __asm ("nop" : "=Emr" (x) : :); // no-error
+  __asm ("nop" : "=E" (x) : :); // expected-error{{invalid output constraint '=E' in asm}}
+  __asm ("nop" : "+E" (x) : :); // expected-error{{invalid output constraint '+E' in asm}}
+  __asm ("nop" : "=EE" (x) : :); // expected-error{{invalid output constraint '=EE' in asm}}
+}
+
 // PR19837
 struct foo {
   int a;
diff --git a/test/Sema/overloadable.c b/test/Sema/overloadable.c
index 49d8085651d4..be9b862f2934 100644
--- a/test/Sema/overloadable.c
+++ b/test/Sema/overloadable.c
@@ -106,8 +106,8 @@ void fn_type_conversions() {
   void foo(char *c) __attribute__((overloadable));
   void (*ptr1)(void *) = &foo;
   void (*ptr2)(char *) = &foo;
-  void (*ambiguous)(int *) = &foo; // expected-error{{initializing 'void (*)(int *)' with an expression of incompatible type '<overloaded function type>'}} expected-note@105{{candidate function}} expected-note@106{{candidate function}}
-  void *vp_ambiguous = &foo; // expected-error{{initializing 'void *' with an expression of incompatible type '<overloaded function type>'}} expected-note@105{{candidate function}} expected-note@106{{candidate function}}
+  void (*ambiguous)(int *) = &foo; // expected-error{{initializing 'void (*)(int *)' with an expression of incompatible type '<overloaded function type>'}} expected-note@-4{{candidate function}} expected-note@-3{{candidate function}}
+  void *vp_ambiguous = &foo; // expected-error{{initializing 'void *' with an expression of incompatible type '<overloaded function type>'}} expected-note@-5{{candidate function}} expected-note@-4{{candidate function}}
 
   void (*specific1)(int *) = (void (*)(void *))&foo; // expected-warning{{incompatible function pointer types initializing 'void (*)(int *)' with an expression of type 'void (*)(void *)'}}
   void *specific2 = (void (*)(void *))&foo;
@@ -117,8 +117,8 @@ void fn_type_conversions() {
   void disabled(char *c) __attribute__((overloadable, enable_if(1, "The function name lies.")));
   // To be clear, these should all point to the last overload of 'disabled'
   void (*dptr1)(char *c) = &disabled;
-  void (*dptr2)(void *c) = &disabled; // expected-warning{{incompatible pointer types initializing 'void (*)(void *)' with an expression of type '<overloaded function type>'}} expected-note@115{{candidate function made ineligible by enable_if}} expected-note@116{{candidate function made ineligible by enable_if}} expected-note@117{{candidate function has type mismatch at 1st parameter (expected 'void *' but has 'char *')}}
-  void (*dptr3)(int *c) = &disabled; // expected-warning{{incompatible pointer types initializing 'void (*)(int *)' with an expression of type '<overloaded function type>'}} expected-note@115{{candidate function made ineligible by enable_if}} expected-note@116{{candidate function made ineligible by enable_if}} expected-note@117{{candidate function has type mismatch at 1st parameter (expected 'int *' but has 'char *')}}
+  void (*dptr2)(void *c) = &disabled; // expected-warning{{incompatible pointer types initializing 'void (*)(void *)' with an expression of type '<overloaded function type>'}} expected-note@-5{{candidate function made ineligible by enable_if}} expected-note@-4{{candidate function made ineligible by enable_if}} expected-note@-3{{candidate function has type mismatch at 1st parameter (expected 'void *' but has 'char *')}}
+  void (*dptr3)(int *c) = &disabled; // expected-warning{{incompatible pointer types initializing 'void (*)(int *)' with an expression of type '<overloaded function type>'}} expected-note@-6{{candidate function made ineligible by enable_if}} expected-note@-5{{candidate function made ineligible by enable_if}} expected-note@-4{{candidate function has type mismatch at 1st parameter (expected 'int *' but has 'char *')}}
 
   void *specific_disabled = &disabled;
 }
@@ -131,14 +131,14 @@ void incompatible_pointer_type_conversions() {
   void foo(char *c) __attribute__((overloadable));
   void foo(short *c) __attribute__((overloadable));
   foo(charbuf);
-  foo(ucharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@131{{candidate function}} expected-note@132{{candidate function}}
-  foo(intbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@131{{candidate function}} expected-note@132{{candidate function}}
+  foo(ucharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@-3{{candidate function}} expected-note@-2{{candidate function}}
+  foo(intbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@-4{{candidate function}} expected-note@-3{{candidate function}}
 
   void bar(unsigned char *c) __attribute__((overloadable));
   void bar(signed char *c) __attribute__((overloadable));
-  bar(charbuf); // expected-error{{call to 'bar' is ambiguous}} expected-note@137{{candidate function}} expected-note@138{{candidate function}}
+  bar(charbuf); // expected-error{{call to 'bar' is ambiguous}} expected-note@-2{{candidate function}} expected-note@-1{{candidate function}}
   bar(ucharbuf);
-  bar(intbuf); // expected-error{{call to 'bar' is ambiguous}} expected-note@137{{candidate function}} expected-note@138{{candidate function}}
+  bar(intbuf); // expected-error{{call to 'bar' is ambiguous}} expected-note@-4{{candidate function}} expected-note@-3{{candidate function}}
 }
 
 void dropping_qualifiers_is_incompatible() {
@@ -148,8 +148,8 @@ void dropping_qualifiers_is_incompatible() {
   void foo(char *c) __attribute__((overloadable));
   void foo(const volatile unsigned char *c) __attribute__((overloadable));
 
-  foo(ccharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@148{{candidate function}} expected-note@149{{candidate function}}
-  foo(vcharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@148{{candidate function}} expected-note@149{{candidate function}}
+  foo(ccharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@-3{{candidate function}} expected-note@-2{{candidate function}}
+  foo(vcharbuf); // expected-error{{call to 'foo' is ambiguous}} expected-note@-4{{candidate function}} expected-note@-3{{candidate function}}
 }
 
 // Bug: we used to treat `__typeof__(foo)` as though it was `__typeof__(&foo)`
diff --git a/test/SemaCXX/PR16677.cpp b/test/SemaCXX/PR16677.cpp
index 7140ac79f089..efa4faaacd69 100644
--- a/test/SemaCXX/PR16677.cpp
+++ b/test/SemaCXX/PR16677.cpp
@@ -10,7 +10,6 @@ class Base { };
 template<class T,  // Should be angle bracket instead of comma
 class Derived : public Base<T> { // expected-error{{'Derived' cannot be defined in a type specifier}}
   Class_With_Destructor member;
-}; // expected-error{{a non-type template parameter cannot have type 'class Derived'}}
-   // expected-error@-1{{expected ',' or '>' in template-parameter-list}}
-   // expected-warning@-2{{declaration does not declare anything}}
+}; // expected-error{{expected ',' or '>' in template-parameter-list}}
+   // expected-warning@-1{{declaration does not declare anything}}
 
diff --git a/test/SemaCXX/cxx1y-deduced-return-type.cpp b/test/SemaCXX/cxx1y-deduced-return-type.cpp
index bfe0ab9dcdbc..13ff751acae4 100644
--- a/test/SemaCXX/cxx1y-deduced-return-type.cpp
+++ b/test/SemaCXX/cxx1y-deduced-return-type.cpp
@@ -55,6 +55,25 @@ auto b(bool k) {
   return "goodbye";
 }
 
+// Allow 'operator auto' to call only the explicit operator auto.
+struct BothOps {
+  template <typename T> operator T();
+  template <typename T> operator T *();
+  operator auto() { return 0; }
+  operator auto *() { return this; }
+};
+struct JustTemplateOp {
+  template <typename T> operator T();
+  template <typename T> operator T *();
+};
+
+auto c() {
+  BothOps().operator auto(); // ok
+  BothOps().operator auto *(); // ok
+  JustTemplateOp().operator auto(); // expected-error {{no member named 'operator auto' in 'JustTemplateOp'}}
+  JustTemplateOp().operator auto *(); // expected-error {{no member named 'operator auto *' in 'JustTemplateOp'}}
+}
+
 auto *ptr_1() {
   return 100; // expected-error {{cannot deduce return type 'auto *' from returned value of type 'int'}}
 }
diff --git a/test/SemaCXX/cxx1z-noexcept-function-type.cpp b/test/SemaCXX/cxx1z-noexcept-function-type.cpp
index 40dc3a22e530..524ea8e53c10 100644
--- a/test/SemaCXX/cxx1z-noexcept-function-type.cpp
+++ b/test/SemaCXX/cxx1z-noexcept-function-type.cpp
@@ -1,5 +1,6 @@
 // RUN: %clang_cc1 -std=c++14 -verify -fexceptions -fcxx-exceptions %s
 // RUN: %clang_cc1 -std=c++1z -verify -fexceptions -fcxx-exceptions %s -Wno-dynamic-exception-spec
+// RUN: %clang_cc1 -std=c++14 -verify -fexceptions -fcxx-exceptions -Wno-c++1z-compat-mangling -DNO_COMPAT_MANGLING %s
 
 #if __cplusplus > 201402L
 
@@ -81,7 +82,7 @@ namespace CompatWarning {
   auto f5() -> void (*)() throw();
   auto f6() -> void (&)() throw();
   auto f7() -> void (X::*)() throw();
-#if __cplusplus <= 201402L
+#if __cplusplus <= 201402L && !defined(NO_COMPAT_MANGLING)
   // expected-warning@-8 {{mangled name of 'f1' will change in C++17 due to non-throwing exception specification in function signature}}
   // expected-warning@-8 {{mangled name of 'f2' will change in C++17 due to non-throwing exception specification in function signature}}
   // expected-warning@-8 {{mangled name of 'f3' will change in C++17 due to non-throwing exception specification in function signature}}
diff --git a/test/SemaCXX/friend2.cpp b/test/SemaCXX/friend2.cpp
index 347af0d61b1b..d1d4b628ba2d 100644
--- a/test/SemaCXX/friend2.cpp
+++ b/test/SemaCXX/friend2.cpp
@@ -170,3 +170,40 @@ struct Test {
 template class Test<int>;
 
 }
+
+namespace pr14785 {
+template<typename T>
+struct Somewhat {
+  void internal() const { }
+  friend void operator+(int const &, Somewhat<T> const &) {}  // expected-error{{redefinition of 'operator+'}}
+};
+
+void operator+(int const &, Somewhat<char> const &x) {  // expected-note {{previous definition is here}}
+  x.internal();  // expected-note{{in instantiation of template class 'pr14785::Somewhat<char>' requested here}}
+}
+}
+
+namespace D30375 {
+template <typename K> struct B {
+  template <typename A> bool insert(A &);
+};
+
+template <typename K>
+template <typename A> bool B<K>::insert(A &x) { return x < x; }
+
+template <typename K> class D {
+  B<K> t;
+
+public:
+  K x;
+  bool insert() { return t.insert(x); }
+  template <typename K1> friend bool operator<(const D<K1> &, const D<K1> &);
+};
+
+template <typename K> bool operator<(const D<K> &, const D<K> &);
+
+void func() {
+  D<D<int>> cache;
+  cache.insert();
+}
+}
diff --git a/test/SemaCXX/invalid-template-params.cpp b/test/SemaCXX/invalid-template-params.cpp
new file mode 100644
index 000000000000..0c463fe13d5b
--- /dev/null
+++ b/test/SemaCXX/invalid-template-params.cpp
@@ -0,0 +1,23 @@
+// RUN: %clang_cc1 -fsyntax-only -std=c++11 -verify %s
+
+template<class> class Foo {
+  template<class UBar // expected-error {{expected ';' after class}}
+                      // expected-note@-1 {{'UBar' declared here}}
+  void foo1(); // expected-error {{a non-type template parameter cannot have type 'class UBar'}}
+               // expected-error@-1 {{expected ',' or '>' in template-parameter-list}}
+               // expected-warning@-2 {{declaration does not declare anything}}
+};
+
+Foo<int>::UBar g1; // expected-error {{no type named 'UBar' in 'Foo<int>'}}
+
+class C0 {
+public:
+  template<typename T0, typename T1 = T0 // missing closing angle bracket
+  struct S0 {}; // expected-error {{'S0' cannot be defined in a type specifier}}
+                // expected-error@-1 {{cannot combine with previous 'type-name' declaration specifier}}
+                // expected-error@-2 {{expected ',' or '>' in template-parameter-list}}
+                // expected-warning@-3 {{declaration does not declare anything}}
+  C0() : m(new S0<int>) {} // expected-error {{expected '(' for function-style cast or type construction}}
+                           // expected-error@-1 {{expected expression}}
+  S0<int> *m; // expected-error {{expected member name or ';' after declaration specifiers}}
+};
diff --git a/test/SemaCXX/warn-throw-out-noexcept-func.cpp b/test/SemaCXX/warn-throw-out-noexcept-func.cpp
new file mode 100644
index 000000000000..dfd1ff9065ab
--- /dev/null
+++ b/test/SemaCXX/warn-throw-out-noexcept-func.cpp
@@ -0,0 +1,265 @@
+// RUN: %clang_cc1 %s  -fdelayed-template-parsing -fcxx-exceptions -fsyntax-only -Wexceptions -verify -std=c++11
+struct A_ShouldDiag {
+  ~A_ShouldDiag(); // implicitly noexcept(true)
+};
+A_ShouldDiag::~A_ShouldDiag() { // expected-note {{destructor or deallocator has a (possibly implicit) non-throwing excepton specification}}
+  throw 1; // expected-warning {{has a non-throwing exception specification but can still throw, resulting in unexpected program termination}}
+}
+struct B_ShouldDiag {
+  int i;
+  ~B_ShouldDiag() noexcept(true) {} //no disg, no throw stmt
+};
+struct R_ShouldDiag : A_ShouldDiag {
+  B_ShouldDiag b;
+  ~R_ShouldDiag() {     // expected-note  {{destructor or deallocator has a}}
+    throw 1; // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+
+struct M_ShouldNotDiag {
+  B_ShouldDiag b;
+  ~M_ShouldNotDiag() noexcept(false);
+};
+
+M_ShouldNotDiag::~M_ShouldNotDiag() noexcept(false) {
+  throw 1;
+}
+
+struct N_ShouldDiag {
+  B_ShouldDiag b;
+  ~N_ShouldDiag(); //implicitly noexcept(true)
+};
+
+N_ShouldDiag::~N_ShouldDiag() {  // expected-note  {{destructor or deallocator has a}}
+  throw 1; // expected-warning {{has a non-throwing exception specification but}}
+}
+struct X_ShouldDiag {
+  B_ShouldDiag b;
+  ~X_ShouldDiag() noexcept { // expected-note  {{destructor or deallocator has a}}
+    throw 1;      // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+struct Y_ShouldDiag : A_ShouldDiag {
+  ~Y_ShouldDiag() noexcept(true) { // expected-note  {{destructor or deallocator has a}}
+    throw 1;            // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+struct C_ShouldNotDiag {
+  int i;
+  ~C_ShouldNotDiag() noexcept(false) {}
+};
+struct D_ShouldNotDiag {
+  C_ShouldNotDiag c;
+  ~D_ShouldNotDiag() { //implicitly noexcept(false)
+    throw 1;
+  }
+};
+struct E_ShouldNotDiag  {
+  C_ShouldNotDiag c;
+  ~E_ShouldNotDiag(); //implicitly noexcept(false)
+};
+E_ShouldNotDiag::~E_ShouldNotDiag() //implicitly noexcept(false)
+{
+  throw 1;
+}
+
+template <typename T>
+class A1_ShouldDiag {
+  T b;
+
+public:
+  ~A1_ShouldDiag() {    // expected-note  {{destructor or deallocator has a}}
+    throw 1; // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+template <typename T>
+struct B1_ShouldDiag {
+  T i;
+  ~B1_ShouldDiag() noexcept(true) {}
+};
+template <typename T>
+struct R1_ShouldDiag : A1_ShouldDiag<T> //expected-note {{in instantiation of member function}}
+{
+  B1_ShouldDiag<T> b;
+  ~R1_ShouldDiag() {    // expected-note  {{destructor or deallocator has a}}
+    throw 1; // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+template <typename T>
+struct S1_ShouldDiag : A1_ShouldDiag<T> {
+  B1_ShouldDiag<T> b;
+  ~S1_ShouldDiag() noexcept { // expected-note  {{destructor or deallocator has a}}
+    throw 1;       // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+void operator delete(void *ptr) noexcept { // expected-note  {{destructor or deallocator has a}}
+  throw 1;                                 // expected-warning {{has a non-throwing exception specification but}}
+}
+struct except_fun {
+  static const bool i = false;
+};
+struct noexcept_fun {
+  static const bool i = true;
+};
+template <typename T>
+struct dependent_warn {
+  ~dependent_warn() noexcept(T::i) {
+    throw 1;
+  }
+};
+template <typename T>
+struct dependent_warn_noexcept {
+  ~dependent_warn_noexcept() noexcept(T::i) { // expected-note  {{destructor or deallocator has a}}
+    throw 1;                                  // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+template <typename T>
+struct dependent_warn_both {
+  ~dependent_warn_both() noexcept(T::i) { // expected-note  {{destructor or deallocator has a}}
+    throw 1;                              // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+void foo() noexcept { //expected-note {{non-throwing function declare here}}
+  throw 1;            // expected-warning {{has a non-throwing exception specification but}}
+}
+struct Throws {
+  ~Throws() noexcept(false);
+};
+
+struct ShouldDiagnose {
+  Throws T;
+  ~ShouldDiagnose() noexcept { //expected-note {{destructor or deallocator has a}}
+    throw; // expected-warning {{has a non-throwing exception specification but}}
+  }
+};
+struct ShouldNotDiagnose {
+  Throws T;
+  ~ShouldNotDiagnose() { 
+    throw; 
+  }
+};
+
+void bar_ShouldNotDiag() noexcept {
+  try {
+    throw 1;
+  } catch (...) {
+  }
+}
+void f_ShouldNotDiag() noexcept {
+  try {
+    throw 12;
+  } catch (int) {
+  }
+}
+void g_ShouldNotDiag() noexcept {
+  try {
+    throw 12;
+  } catch (...) {
+  }
+}
+
+void h_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw 12; // expected-warning {{has a non-throwing exception specification but}}
+  } catch (const char *) {
+  }
+}
+
+void i_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw 12;
+  } catch (int) {
+    throw; // expected-warning {{has a non-throwing exception specification but}}
+  }
+}
+void j_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw 12;
+  } catch (int) {
+    throw "haha"; // expected-warning {{has a non-throwing exception specification but}}
+  }
+}
+
+void k_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw 12;
+  } catch (...) {
+    throw; // expected-warning {{has a non-throwing exception specification but}}
+  }
+}
+
+void loo_ShouldDiag(int i) noexcept { //expected-note {{non-throwing function declare here}}
+  if (i)
+    try {
+      throw 12;
+    } catch (int) {
+      throw "haha"; //expected-warning {{has a non-throwing exception specification but}}
+    }
+  i = 10;
+}
+
+void loo1_ShouldNotDiag() noexcept {
+  if (0)
+    throw 12;
+}
+
+void loo2_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  if (1)
+    throw 12; // expected-warning {{has a non-throwing exception specification but}}
+}
+struct S {};
+
+void l_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw S{}; //expected-warning {{has a non-throwing exception specification but}}
+  } catch (S *s) {
+  }
+}
+
+void m_ShouldNotDiag() noexcept {
+  try {
+    const S &s = S{};
+    throw s;
+  } catch (S s) {
+  }
+
+}
+void n_ShouldNotDiag() noexcept {
+  try {
+    S s = S{};
+    throw s;
+  } catch (const S &s) {
+  }
+}
+void o_ShouldDiag() noexcept { //expected-note {{non-throwing function declare here}}
+  try {
+    throw; //expected-warning {{has a non-throwing exception specification but}}
+  } catch (...) {
+  }
+}
+
+#define NOEXCEPT noexcept
+void with_macro() NOEXCEPT { //expected-note {{non-throwing function declare here}}
+  throw 1; // expected-warning {{has a non-throwing exception specification but}}
+}
+
+void with_try_block() try {
+  throw 2;
+} catch (...) {
+}
+
+void with_try_block1() noexcept try { //expected-note {{non-throwing function declare here}}
+  throw 2;  // expected-warning {{has a non-throwing exception specification but}}
+} catch (char *) {
+}
+
+int main() {
+  R1_ShouldDiag<int> o; //expected-note {{in instantiation of member function}}
+  S1_ShouldDiag<int> b; //expected-note {{in instantiation of member function}}
+  dependent_warn<except_fun> f;
+  dependent_warn_noexcept<noexcept_fun> f1; //expected-note {{in instantiation of member function}}
+  dependent_warn_both<except_fun> f2;
+  dependent_warn_both<noexcept_fun> f3; //expected-note {{in instantiation of member function}}
+  ShouldDiagnose obj;
+  ShouldNotDiagnose obj1;
+}
diff --git a/test/SemaObjC/unguarded-availability-new.m b/test/SemaObjC/unguarded-availability-new.m
new file mode 100644
index 000000000000..33baedebc279
--- /dev/null
+++ b/test/SemaObjC/unguarded-availability-new.m
@@ -0,0 +1,160 @@
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -xobjective-c++ -DMAC -triple x86_64-apple-macosx10.13 -fblocks -fsyntax-only -verify %s
+
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -Wunguarded-availability-new -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -Wno-unguarded-availability-new -DNO_WARNING -fblocks -fsyntax-only -verify %s
+
+// unguarded-availability implies unguarded-availability-new:
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -Wunguarded-availability -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.11 -Wunguarded-availability -Wno-unguarded-availability-new -DNO_WARNING -DWARN_PREV -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -Wno-unguarded-availability -DNO_WARNING  -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -Wno-unguarded-availability -Wunguarded-availability-new -fblocks -fsyntax-only -verify %s
+
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -D TEST_FUNC_CURRENT -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.13 -D TEST_FUNC_NEXT -DNO_WARNING -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-ios11 -DNO_WARNING -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DMAC -triple x86_64-apple-macosx10.12 -DWARN_CURRENT -fblocks -fsyntax-only -verify %s
+
+// RUN: %clang_cc1 -DIOS -triple x86_64-apple-ios11 -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DIOS -triple x86_64-apple-ios11 -D TEST_FUNC_CURRENT -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DIOS -triple x86_64-apple-ios11 -D TEST_FUNC_NEXT -DNO_WARNING -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DIOS -triple x86_64-apple-ios10.3 -DWARN_CURRENT -fblocks -fsyntax-only -verify %s
+
+// RUN: %clang_cc1 -DTVOS -triple x86_64-apple-tvos11 -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DTVOS -triple x86_64-apple-tvos11 -D TEST_FUNC_CURRENT -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DTVOS -triple x86_64-apple-tvos11 -D TEST_FUNC_NEXT -DNO_WARNING -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DTVOS -triple x86_64-apple-tvos10 -DWARN_CURRENT -fblocks -fsyntax-only -verify %s
+
+// RUN: %clang_cc1 -DWATCHOS -triple i386-apple-watchos4 -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DWATCHOS -triple i386-apple-watchos4 -D TEST_FUNC_CURRENT -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DWATCHOS -triple i386-apple-watchos4 -D TEST_FUNC_NEXT -DNO_WARNING -fblocks -fsyntax-only -verify %s
+// RUN: %clang_cc1 -DWATCHOS -triple i386-apple-watchos3 -DWARN_CURRENT -fblocks -fsyntax-only -verify %s
+
+#ifdef MAC
+#define PLATFORM macos
+#define NEXT 10.14
+
+#define AVAILABLE_PREV __attribute__((availability(macos, introduced = 10.12)))
+#define AVAILABLE_CURRENT __attribute__((availability(macos, introduced = 10.13)))
+#define AVAILABLE_NEXT __attribute__((availability(macos, introduced = 10.14)))
+#endif
+
+#ifdef IOS
+#define PLATFORM ios
+#define NEXT 12
+
+#define AVAILABLE_PREV __attribute__((availability(ios, introduced = 10)))
+#define AVAILABLE_CURRENT __attribute__((availability(ios, introduced = 11)))
+#define AVAILABLE_NEXT __attribute__((availability(ios, introduced = 12)))
+#endif
+
+#ifdef TVOS
+#define PLATFORM tvos
+#define NEXT 13
+
+#define AVAILABLE_PREV __attribute__((availability(tvos, introduced = 10)))
+#define AVAILABLE_CURRENT __attribute__((availability(tvos, introduced = 11)))
+#define AVAILABLE_NEXT __attribute__((availability(tvos, introduced = 13)))
+#endif
+
+#ifdef WATCHOS
+#define PLATFORM watchos
+#define NEXT 5
+
+#define AVAILABLE_PREV __attribute__((availability(watchos, introduced = 3)))
+#define AVAILABLE_CURRENT __attribute__((availability(watchos, introduced = 4)))
+#define AVAILABLE_NEXT __attribute__((availability(watchos, introduced = 5)))
+#endif
+
+void previouslyAvailable() AVAILABLE_PREV;
+#ifdef WARN_PREV
+  // expected-note@-2 {{'previouslyAvailable' has been explicitly marked partial here}}
+#endif
+void currentlyAvailable() AVAILABLE_CURRENT;
+#ifdef WARN_CURRENT
+  // expected-note@-2 {{'currentlyAvailable' has been explicitly marked partial here}}
+#endif
+void willBeAvailabile() AVAILABLE_NEXT;
+#ifndef NO_WARNING
+  // expected-note@-2 {{'willBeAvailabile' has been explicitly marked partial here}}
+#endif
+
+#ifdef TEST_FUNC_CURRENT
+#define FUNC_AVAILABLE AVAILABLE_CURRENT
+#endif
+#ifdef TEST_FUNC_NEXT
+#define FUNC_AVAILABLE AVAILABLE_NEXT
+#endif
+#ifndef FUNC_AVAILABLE
+#define FUNC_AVAILABLE
+#endif
+
+typedef int AVAILABLE_NEXT new_int;
+#ifndef NO_WARNING
+  // expected-note@-2 {{'new_int' has been explicitly marked partial here}}
+#endif
+FUNC_AVAILABLE new_int x;
+#ifndef NO_WARNING
+#ifdef MAC
+  // expected-warning@-3 {{'new_int' is partial: introduced in macOS 10.14}} expected-note@-3 {{explicitly redeclare 'new_int' to silence this warning}}
+#endif
+#ifdef IOS
+  // expected-warning@-6 {{'new_int' is partial: introduced in iOS 12}} expected-note@-6 {{explicitly redeclare 'new_int' to silence this warning}}
+#endif
+#ifdef TVOS
+  // expected-warning@-9 {{'new_int' is partial: introduced in tvOS 13}} expected-note@-9 {{explicitly redeclare 'new_int' to silence this warning}}
+#endif
+#ifdef WATCHOS
+  // expected-warning@-12 {{'new_int' is partial: introduced in watchOS 5}} expected-note@-12 {{explicitly redeclare 'new_int' to silence this warning}}
+#endif
+#endif
+
+void test() FUNC_AVAILABLE {
+  previouslyAvailable();
+#ifdef WARN_PREV
+#ifdef MAC
+  // expected-warning@-3 {{'previouslyAvailable' is only available on macOS 10.12 or newer}}
+#endif
+  // expected-note@-5 {{enclose 'previouslyAvailable' in an @available check to silence this warning}}
+#endif
+  currentlyAvailable();
+#ifdef WARN_CURRENT
+#ifdef MAC
+  // expected-warning@-3 {{'currentlyAvailable' is only available on macOS 10.13 or newer}}
+#endif
+#ifdef IOS
+  // expected-warning@-6 {{'currentlyAvailable' is only available on iOS 11 or newer}}
+#endif
+#ifdef TVOS
+  // expected-warning@-9 {{'currentlyAvailable' is only available on tvOS 11 or newer}}
+#endif
+#ifdef WATCHOS
+  // expected-warning@-12 {{'currentlyAvailable' is only available on watchOS 4 or newer}}
+#endif
+  // expected-note@-14 {{enclose 'currentlyAvailable' in an @available check to silence this warning}}
+#endif
+  willBeAvailabile();
+#ifndef NO_WARNING
+#ifdef MAC
+  // expected-warning@-3 {{'willBeAvailabile' is only available on macOS 10.14 or newer}}
+#endif
+#ifdef IOS
+  // expected-warning@-6 {{'willBeAvailabile' is only available on iOS 12 or newer}}
+#endif
+#ifdef TVOS
+  // expected-warning@-9 {{'willBeAvailabile' is only available on tvOS 13 or newer}}
+#endif
+#ifdef WATCHOS
+  // expected-warning@-12 {{'willBeAvailabile' is only available on watchOS 5 or newer}}
+#endif
+  // expected-note@-14 {{enclose 'willBeAvailabile' in an @available check to silence this warning}}
+#endif
+  if (@available(PLATFORM NEXT, *))
+    willBeAvailabile(); // OK
+}
+
+#ifdef NO_WARNING
+#ifndef WARN_PREV
+// expected-no-diagnostics
+#endif
+#endif
diff --git a/test/SemaOpenCL/storageclass.cl b/test/SemaOpenCL/storageclass.cl
index e611313b4a70..9a461068f237 100644
--- a/test/SemaOpenCL/storageclass.cl
+++ b/test/SemaOpenCL/storageclass.cl
@@ -13,6 +13,11 @@ void kernel foo(int x) {
   constant int L1 = 0;
   local int L2;
 
+  if (true) {
+    local int L1; // expected-error {{variables in the local address space can only be declared in the outermost scope of a kernel function}}
+    constant int L1 = 42; // expected-error {{variables in the constant address space can only be declared in the outermost scope of a kernel function}}
+  }
+
   auto int L3 = 7;                            // expected-error{{OpenCL version 1.2 does not support the 'auto' storage class specifier}}
   global int L4;                              // expected-error{{function scope variable cannot be declared in global address space}}
   __attribute__((address_space(100))) int L5; // expected-error{{automatic variable qualified with an invalid address space}}
diff --git a/tools/clang-format/clang-format.py b/tools/clang-format/clang-format.py
index ae8a6ebf74e9..2412566346f2 100644
--- a/tools/clang-format/clang-format.py
+++ b/tools/clang-format/clang-format.py
@@ -63,8 +63,19 @@ def main():
   # Determine range to format.
   if vim.eval('exists("l:lines")') == '1':
     lines = vim.eval('l:lines')
+  elif vim.eval('exists("l:formatdiff")') == '1':
+    with open(vim.current.buffer.name, 'r') as f:
+      ondisk = f.read().splitlines();
+    sequence = difflib.SequenceMatcher(None, ondisk, vim.current.buffer)
+    lines = []
+    for op in reversed(sequence.get_opcodes()):
+      if op[0] not in ['equal', 'delete']:
+        lines += ['-lines', '%s:%s' % (op[3] + 1, op[4])]
+    if lines == []:
+      return
   else:
-    lines = '%s:%s' % (vim.current.range.start + 1, vim.current.range.end + 1)
+    lines = ['-lines', '%s:%s' % (vim.current.range.start + 1,
+                                  vim.current.range.end + 1)]
 
   # Determine the cursor position.
   cursor = int(vim.eval('line2byte(line("."))+col(".")')) - 2
@@ -82,7 +93,7 @@ def main():
   # Call formatter.
   command = [binary, '-style', style, '-cursor', str(cursor)]
   if lines != 'all':
-    command.extend(['-lines', lines])
+    command += lines
   if fallback_style:
     command.extend(['-fallback-style', fallback_style])
   if vim.current.buffer.name:
diff --git a/tools/clang-import-test/clang-import-test.cpp b/tools/clang-import-test/clang-import-test.cpp
index 567a4bb4f0a2..db4dc76eded5 100644
--- a/tools/clang-import-test/clang-import-test.cpp
+++ b/tools/clang-import-test/clang-import-test.cpp
@@ -17,7 +17,9 @@
 #include "clang/Basic/TargetInfo.h"
 #include "clang/Basic/TargetOptions.h"
 #include "clang/CodeGen/ModuleBuilder.h"
+#include "clang/Frontend/ASTConsumers.h"
 #include "clang/Frontend/CompilerInstance.h"
+#include "clang/Frontend/MultiplexConsumer.h"
 #include "clang/Frontend/TextDiagnosticBuffer.h"
 #include "clang/Lex/Lexer.h"
 #include "clang/Lex/Preprocessor.h"
@@ -51,6 +53,10 @@ static llvm::cl::list<std::string>
               llvm::cl::desc("Argument to pass to the CompilerInvocation"),
               llvm::cl::CommaSeparated);
 
+static llvm::cl::opt<bool>
+DumpAST("dump-ast", llvm::cl::init(false),
+        llvm::cl::desc("Dump combined AST"));
+
 namespace init_convenience {
 class TestDiagnosticConsumer : public DiagnosticConsumer {
 private:
@@ -233,7 +239,7 @@ std::unique_ptr<CompilerInstance> BuildIndirect(std::unique_ptr<CompilerInstance
 }
 
 llvm::Error ParseSource(const std::string &Path, CompilerInstance &CI,
-                        CodeGenerator &CG) {
+                        ASTConsumer &Consumer) {
   SourceManager &SM = CI.getSourceManager();
   const FileEntry *FE = CI.getFileManager().getFile(Path);
   if (!FE) {
@@ -241,13 +247,14 @@ llvm::Error ParseSource(const std::string &Path, CompilerInstance &CI,
         llvm::Twine("Couldn't open ", Path), std::error_code());
   }
   SM.setMainFileID(SM.createFileID(FE, SourceLocation(), SrcMgr::C_User));
-  ParseAST(CI.getPreprocessor(), &CG, CI.getASTContext());
+  ParseAST(CI.getPreprocessor(), &Consumer, CI.getASTContext());
   return llvm::Error::success();
 }
 
 llvm::Expected<std::unique_ptr<CompilerInstance>>
 Parse(const std::string &Path,
-      llvm::ArrayRef<std::unique_ptr<CompilerInstance>> Imports) {
+      llvm::ArrayRef<std::unique_ptr<CompilerInstance>> Imports,
+      bool ShouldDumpAST) {
   std::vector<const char *> ClangArgv(ClangArgs.size());
   std::transform(ClangArgs.begin(), ClangArgs.end(), ClangArgv.begin(),
                  [](const std::string &s) -> const char * { return s.data(); });
@@ -261,14 +268,20 @@ Parse(const std::string &Path,
   if (Imports.size())
     AddExternalSource(*CI, Imports);
 
+  std::vector<std::unique_ptr<ASTConsumer>> ASTConsumers;
+
   auto LLVMCtx = llvm::make_unique<llvm::LLVMContext>();
-  std::unique_ptr<CodeGenerator> CG =
-      init_convenience::BuildCodeGen(*CI, *LLVMCtx);
-  CG->Initialize(CI->getASTContext());
+  ASTConsumers.push_back(init_convenience::BuildCodeGen(*CI, *LLVMCtx));
+
+  if (ShouldDumpAST)
+    ASTConsumers.push_back(CreateASTDumper("", true, false, false));
 
   CI->getDiagnosticClient().BeginSourceFile(CI->getLangOpts(),
                                             &CI->getPreprocessor());
-  if (llvm::Error PE = ParseSource(Path, *CI, *CG)) {
+  MultiplexConsumer Consumers(std::move(ASTConsumers));
+  Consumers.Initialize(CI->getASTContext());
+
+  if (llvm::Error PE = ParseSource(Path, *CI, Consumers)) {
     return std::move(PE);
   }
   CI->getDiagnosticClient().EndSourceFile();
@@ -288,7 +301,8 @@ int main(int argc, const char **argv) {
   llvm::cl::ParseCommandLineOptions(argc, argv);
   std::vector<std::unique_ptr<CompilerInstance>> ImportCIs;
   for (auto I : Imports) {
-    llvm::Expected<std::unique_ptr<CompilerInstance>> ImportCI = Parse(I, {});
+    llvm::Expected<std::unique_ptr<CompilerInstance>> ImportCI =
+      Parse(I, {}, false);
     if (auto E = ImportCI.takeError()) {
       llvm::errs() << llvm::toString(std::move(E));
       exit(-1);
@@ -310,7 +324,7 @@ int main(int argc, const char **argv) {
     }
   }
   llvm::Expected<std::unique_ptr<CompilerInstance>> ExpressionCI =
-      Parse(Expression, Direct ? ImportCIs : IndirectCIs);
+      Parse(Expression, Direct ? ImportCIs : IndirectCIs, DumpAST);
   if (auto E = ExpressionCI.takeError()) {
     llvm::errs() << llvm::toString(std::move(E));
     exit(-1);
diff --git a/tools/driver/cc1as_main.cpp b/tools/driver/cc1as_main.cpp
index 37c14f4a26e1..2fc2b508ef21 100644
--- a/tools/driver/cc1as_main.cpp
+++ b/tools/driver/cc1as_main.cpp
@@ -202,9 +202,22 @@ bool AssemblerInvocation::CreateFromArgs(AssemblerInvocation &Opts,
   Opts.SaveTemporaryLabels = Args.hasArg(OPT_msave_temp_labels);
   // Any DebugInfoKind implies GenDwarfForAssembly.
   Opts.GenDwarfForAssembly = Args.hasArg(OPT_debug_info_kind_EQ);
-  // TODO: base this on -gz instead
-  if (Args.hasArg(OPT_compress_debug_sections))
-    Opts.CompressDebugSections = llvm::DebugCompressionType::GNU;
+
+  if (const Arg *A = Args.getLastArg(OPT_compress_debug_sections,
+                                     OPT_compress_debug_sections_EQ)) {
+    if (A->getOption().getID() == OPT_compress_debug_sections) {
+      // TODO: be more clever about the compression type auto-detection
+      Opts.CompressDebugSections = llvm::DebugCompressionType::GNU;
+    } else {
+      Opts.CompressDebugSections =
+          llvm::StringSwitch<llvm::DebugCompressionType>(A->getValue())
+              .Case("none", llvm::DebugCompressionType::None)
+              .Case("zlib", llvm::DebugCompressionType::Z)
+              .Case("zlib-gnu", llvm::DebugCompressionType::GNU)
+              .Default(llvm::DebugCompressionType::None);
+    }
+  }
+
   Opts.RelaxELFRelocations = Args.hasArg(OPT_mrelax_relocations);
   Opts.DwarfVersion = getLastArgIntValue(Args, OPT_dwarf_version_EQ, 2, Diags);
   Opts.DwarfDebugFlags = Args.getLastArgValue(OPT_dwarf_debug_flags);
diff --git a/tools/driver/driver.cpp b/tools/driver/driver.cpp
index 4bd3b228d0ff..af25d9598021 100644
--- a/tools/driver/driver.cpp
+++ b/tools/driver/driver.cpp
@@ -53,8 +53,15 @@ using namespace clang::driver;
 using namespace llvm::opt;
 
 std::string GetExecutablePath(const char *Argv0, bool CanonicalPrefixes) {
-  if (!CanonicalPrefixes)
-    return Argv0;
+  if (!CanonicalPrefixes) {
+    SmallString<128> ExecutablePath(Argv0);
+    // Do a PATH lookup if Argv0 isn't a valid path.
+    if (!llvm::sys::fs::exists(ExecutablePath))
+      if (llvm::ErrorOr<std::string> P =
+              llvm::sys::findProgramByName(ExecutablePath))
+        ExecutablePath = *P;
+    return ExecutablePath.str();
+  }
 
   // This just needs to be some symbol in the binary; C++ doesn't
   // allow taking the address of ::main however.
diff --git a/unittests/Format/CMakeLists.txt b/unittests/Format/CMakeLists.txt
index 507d643ba102..5c04ba1143c6 100644
--- a/unittests/Format/CMakeLists.txt
+++ b/unittests/Format/CMakeLists.txt
@@ -14,6 +14,7 @@ add_clang_unittest(FormatTests
   NamespaceEndCommentsFixerTest.cpp
   SortImportsTestJS.cpp
   SortIncludesTest.cpp
+  UsingDeclarationsSorterTest.cpp
   )
 
 target_link_libraries(FormatTests
diff --git a/unittests/Format/FormatTest.cpp b/unittests/Format/FormatTest.cpp
index f4937171038d..64bb28e0be23 100644
--- a/unittests/Format/FormatTest.cpp
+++ b/unittests/Format/FormatTest.cpp
@@ -341,6 +341,18 @@ TEST_F(FormatTest, FormatIfWithoutCompoundStatement) {
   verifyFormat("if (true)\n  f();\ng();");
   verifyFormat("if (a)\n  if (b)\n    if (c)\n      g();\nh();");
   verifyFormat("if (a)\n  if (b) {\n    f();\n  }\ng();");
+  verifyFormat("if constexpr (true)\n"
+               "  f();\ng();");
+  verifyFormat("if constexpr (a)\n"
+               "  if constexpr (b)\n"
+               "    if constexpr (c)\n"
+               "      g();\n"
+               "h();");
+  verifyFormat("if constexpr (a)\n"
+               "  if constexpr (b) {\n"
+               "    f();\n"
+               "  }\n"
+               "g();");
 
   FormatStyle AllowsMergedIf = getLLVMStyle();
   AllowsMergedIf.AlignEscapedNewlines = FormatStyle::ENAS_Left;
@@ -423,9 +435,11 @@ TEST_F(FormatTest, FormatShortBracedStatements) {
   AllowSimpleBracedStatements.AllowShortLoopsOnASingleLine = true;
 
   verifyFormat("if (true) {}", AllowSimpleBracedStatements);
+  verifyFormat("if constexpr (true) {}", AllowSimpleBracedStatements);
   verifyFormat("while (true) {}", AllowSimpleBracedStatements);
   verifyFormat("for (;;) {}", AllowSimpleBracedStatements);
   verifyFormat("if (true) { f(); }", AllowSimpleBracedStatements);
+  verifyFormat("if constexpr (true) { f(); }", AllowSimpleBracedStatements);
   verifyFormat("while (true) { f(); }", AllowSimpleBracedStatements);
   verifyFormat("for (;;) { f(); }", AllowSimpleBracedStatements);
   verifyFormat("if (true) { //\n"
@@ -444,6 +458,14 @@ TEST_F(FormatTest, FormatShortBracedStatements) {
                "}",
                AllowSimpleBracedStatements);
 
+  verifyFormat("struct A2 {\n"
+               "  int X;\n"
+               "};",
+               AllowSimpleBracedStatements);
+  verifyFormat("typedef struct A2 {\n"
+               "  int X;\n"
+               "} A2_t;",
+               AllowSimpleBracedStatements);
   verifyFormat("template <int> struct A2 {\n"
                "  struct B {};\n"
                "};",
@@ -496,6 +518,19 @@ TEST_F(FormatTest, ParseIfElse) {
                "else {\n"
                "  i();\n"
                "}");
+  verifyFormat("if (true)\n"
+               "  if constexpr (true)\n"
+               "    if (true) {\n"
+               "      if constexpr (true)\n"
+               "        f();\n"
+               "    } else {\n"
+               "      g();\n"
+               "    }\n"
+               "  else\n"
+               "    h();\n"
+               "else {\n"
+               "  i();\n"
+               "}");
   verifyFormat("void f() {\n"
                "  if (a) {\n"
                "  } else {\n"
@@ -511,6 +546,12 @@ TEST_F(FormatTest, ElseIf) {
                "  g();\n"
                "else\n"
                "  h();");
+  verifyFormat("if constexpr (a)\n"
+               "  f();\n"
+               "else if constexpr (b)\n"
+               "  g();\n"
+               "else\n"
+               "  h();");
   verifyFormat("if (a) {\n"
                "  f();\n"
                "}\n"
@@ -528,6 +569,11 @@ TEST_F(FormatTest, ElseIf) {
                "    aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {\n"
                "}",
                getLLVMStyleWithColumns(62));
+  verifyFormat("if (a) {\n"
+               "} else if constexpr (\n"
+               "    aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {\n"
+               "}",
+               getLLVMStyleWithColumns(62));
 }
 
 TEST_F(FormatTest, FormatsForLoop) {
@@ -1658,6 +1704,19 @@ TEST_F(FormatTest, DesignatedInitializers) {
                "    .eeeeeeeeeeeeeeeeeeeeeeeeeee = 5};");
 
   verifyGoogleFormat("const struct A a = {.a = 1, .b = 2};");
+
+  verifyFormat("const struct A a = {[0] = 1, [1] = 2};");
+  verifyFormat("const struct A a = {[1] = aaaaaaaaaa,\n"
+               "                    [2] = bbbbbbbbbb,\n"
+               "                    [3] = cccccccccc,\n"
+               "                    [4] = dddddddddd,\n"
+               "                    [5] = eeeeeeeeee};");
+  verifyFormat("const struct Aaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa = {\n"
+               "    [1] = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa,\n"
+               "    [2] = bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb,\n"
+               "    [3] = cccccccccccccccccccccccccccccccccccccc,\n"
+               "    [4] = dddddddddddddddddddddddddddddddddddddd,\n"
+               "    [5] = eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee};");
 }
 
 TEST_F(FormatTest, NestedStaticInitializers) {
@@ -2506,6 +2565,9 @@ TEST_F(FormatTest, LineBreakingInBinaryExpressions) {
   verifyFormat("if ((aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||\n"
                "     bbbbbbbbbbbbbbbbbb) && // aaaaaaaaaaaaaaaa\n"
                "    cccccc) {\n}");
+  verifyFormat("if constexpr ((aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ||\n"
+               "               bbbbbbbbbbbbbbbbbb) && // aaaaaaaaaaa\n"
+               "              cccccc) {\n}");
   verifyFormat("b = a &&\n"
                "    // Comment\n"
                "    b.c && d;");
@@ -5961,6 +6023,8 @@ TEST_F(FormatTest, LayoutCxx11BraceInitializers) {
                "  T member = {arg1, arg2};\n"
                "};");
   verifyFormat("vector<int> foo = {::SomeGlobalFunction()};");
+  verifyFormat("const struct A a = {.a = 1, .b = 2};");
+  verifyFormat("const struct A a = {[0] = 1, [1] = 2};");
   verifyFormat("static_assert(std::is_integral<int>{} + 0, \"\");");
   verifyFormat("int a = std::is_integral<int>{} + 0;");
 
@@ -6126,6 +6190,8 @@ TEST_F(FormatTest, LayoutCxx11BraceInitializers) {
                "    aaaaaaa,\n"
                "    a};");
   verifyFormat("vector<int> foo = { ::SomeGlobalFunction() };", ExtraSpaces);
+  verifyFormat("const struct A a = { .a = 1, .b = 2 };", ExtraSpaces);
+  verifyFormat("const struct A a = { [0] = 1, [1] = 2 };", ExtraSpaces);
 }
 
 TEST_F(FormatTest, FormatsBracedListsInColumnLayout) {
@@ -6443,6 +6509,52 @@ TEST_F(FormatTest, PullInlineFunctionDefinitionsIntoSingleLine) {
                MergeInlineOnly);
 }
 
+TEST_F(FormatTest, PullInlineOnlyFunctionDefinitionsIntoSingleLine) {
+  FormatStyle MergeInlineOnly = getLLVMStyle();
+  MergeInlineOnly.AllowShortFunctionsOnASingleLine =
+      FormatStyle::SFS_InlineOnly;
+  verifyFormat("class C {\n"
+               "  int f() { return 42; }\n"
+               "};",
+               MergeInlineOnly);
+  verifyFormat("int f() {\n"
+               "  return 42;\n"
+               "}",
+               MergeInlineOnly);
+
+  // SFS_InlineOnly does not imply SFS_Empty
+  verifyFormat("class C {\n"
+               "  int f() {}\n"
+               "};",
+               MergeInlineOnly);
+  verifyFormat("int f() {\n"
+               "}",
+               MergeInlineOnly);
+
+  // Also verify behavior when BraceWrapping.AfterFunction = true
+  MergeInlineOnly.BreakBeforeBraces = FormatStyle::BS_Custom;
+  MergeInlineOnly.BraceWrapping.AfterFunction = true;
+  verifyFormat("class C {\n"
+               "  int f() { return 42; }\n"
+               "};",
+               MergeInlineOnly);
+  verifyFormat("int f()\n"
+               "{\n"
+               "  return 42;\n"
+               "}",
+               MergeInlineOnly);
+
+  // SFS_InlineOnly does not imply SFS_Empty
+  verifyFormat("int f()\n"
+               "{\n"
+               "}",
+               MergeInlineOnly);
+  verifyFormat("class C {\n"
+               "  int f() {}\n"
+               "};",
+               MergeInlineOnly);
+}
+
 TEST_F(FormatTest, SplitEmptyFunctionBody) {
   FormatStyle Style = getLLVMStyle();
   Style.AllowShortFunctionsOnASingleLine = FormatStyle::SFS_None;
@@ -6627,6 +6739,10 @@ TEST_F(FormatTest, MergeHandlingInTheFaceOfPreprocessorDirectives) {
                "  if (true) continue;\n"
                "}",
                ShortMergedIf);
+  ShortMergedIf.ColumnLimit = 33;
+  verifyFormat("#define A \\\n"
+               "  if constexpr (true) return 42;",
+               ShortMergedIf);
   ShortMergedIf.ColumnLimit = 29;
   verifyFormat("#define A                   \\\n"
                "  if (aaaaaaaaaa) return 1; \\\n"
@@ -6638,6 +6754,11 @@ TEST_F(FormatTest, MergeHandlingInTheFaceOfPreprocessorDirectives) {
                "    return 1;     \\\n"
                "  return 2;",
                ShortMergedIf);
+  verifyFormat("#define A                \\\n"
+               "  if constexpr (aaaaaaa) \\\n"
+               "    return 1;            \\\n"
+               "  return 2;",
+               ShortMergedIf);
 }
 
 TEST_F(FormatTest, FormatStarDependingOnContext) {
@@ -8847,11 +8968,24 @@ TEST_F(FormatTest, AllmanBraceBreaking) {
                BreakBeforeBraceShortIfs);
   verifyFormat("void f(bool b)\n"
                "{\n"
+               "  if constexpr (b)\n"
+               "  {\n"
+               "    return;\n"
+               "  }\n"
+               "}\n",
+               BreakBeforeBraceShortIfs);
+  verifyFormat("void f(bool b)\n"
+               "{\n"
                "  if (b) return;\n"
                "}\n",
                BreakBeforeBraceShortIfs);
   verifyFormat("void f(bool b)\n"
                "{\n"
+               "  if constexpr (b) return;\n"
+               "}\n",
+               BreakBeforeBraceShortIfs);
+  verifyFormat("void f(bool b)\n"
+               "{\n"
                "  while (b)\n"
                "  {\n"
                "    return;\n"
@@ -9199,6 +9333,7 @@ TEST_F(FormatTest, ParsesConfigurationBools) {
   CHECK_PARSE_BOOL(Cpp11BracedListStyle);
   CHECK_PARSE_BOOL(ReflowComments);
   CHECK_PARSE_BOOL(SortIncludes);
+  CHECK_PARSE_BOOL(SortUsingDeclarations);
   CHECK_PARSE_BOOL(SpacesInParentheses);
   CHECK_PARSE_BOOL(SpacesInSquareBrackets);
   CHECK_PARSE_BOOL(SpacesInAngles);
@@ -10098,6 +10233,11 @@ TEST_F(FormatTest, FormatsLambdas) {
                "      doo_dah();\n"
                "    })) {\n"
                "}");
+  verifyFormat("if constexpr (blah_blah(whatever, whatever, [] {\n"
+               "                doo_dah();\n"
+               "                doo_dah();\n"
+               "              })) {\n"
+               "}");
   verifyFormat("auto lambda = []() {\n"
                "  int a = 2\n"
                "#if A\n"
@@ -10735,6 +10875,13 @@ TEST_F(ReplacementTest, SortIncludesAfterReplacement) {
   EXPECT_EQ(Expected, *Result);
 }
 
+TEST_F(FormatTest, FormatSortsUsingDeclarations) {
+  EXPECT_EQ("using std::cin;\n"
+            "using std::cout;",
+            format("using std::cout;\n"
+                   "using std::cin;", getGoogleStyle()));
+}
+
 TEST_F(FormatTest, UTF8CharacterLiteralCpp03) {
   format::FormatStyle Style = format::getLLVMStyle();
   Style.Standard = FormatStyle::LS_Cpp03;
diff --git a/unittests/Format/UsingDeclarationsSorterTest.cpp b/unittests/Format/UsingDeclarationsSorterTest.cpp
new file mode 100644
index 000000000000..858a62c2d799
--- /dev/null
+++ b/unittests/Format/UsingDeclarationsSorterTest.cpp
@@ -0,0 +1,234 @@
+//===- UsingDeclarationsSorterTest.cpp - Formatting unit tests ------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/Format/Format.h"
+
+#include "llvm/Support/Debug.h"
+#include "gtest/gtest.h"
+
+#define DEBUG_TYPE "using-declarations-sorter-test"
+
+namespace clang {
+namespace format {
+namespace {
+
+class UsingDeclarationsSorterTest : public ::testing::Test {
+protected:
+  std::string sortUsingDeclarations(llvm::StringRef Code,
+                                    const std::vector<tooling::Range> &Ranges,
+                                    const FormatStyle &Style = getLLVMStyle()) {
+    DEBUG(llvm::errs() << "---\n");
+    DEBUG(llvm::errs() << Code << "\n\n");
+    tooling::Replacements Replaces =
+        clang::format::sortUsingDeclarations(Style, Code, Ranges, "<stdin>");
+    auto Result = applyAllReplacements(Code, Replaces);
+    EXPECT_TRUE(static_cast<bool>(Result));
+    DEBUG(llvm::errs() << "\n" << *Result << "\n\n");
+    return *Result;
+  }
+
+  std::string sortUsingDeclarations(llvm::StringRef Code,
+                                    const FormatStyle &Style = getLLVMStyle()) {
+    return sortUsingDeclarations(Code,
+                                 /*Ranges=*/{1, tooling::Range(0, Code.size())},
+                                 Style);
+  }
+};
+
+TEST_F(UsingDeclarationsSorterTest, SwapsTwoConsecutiveUsingDeclarations) {
+  EXPECT_EQ("using a;\n"
+            "using b;",
+            sortUsingDeclarations("using a;\n"
+                                  "using b;"));
+  EXPECT_EQ("using a;\n"
+            "using aa;",
+            sortUsingDeclarations("using aa;\n"
+                                  "using a;"));
+  EXPECT_EQ("using ::a;\n"
+            "using a;",
+            sortUsingDeclarations("using a;\n"
+                                  "using ::a;"));
+
+  EXPECT_EQ("using a::bcd;\n"
+            "using a::cd;",
+            sortUsingDeclarations("using a::cd;\n"
+                                  "using a::bcd;"));
+
+  EXPECT_EQ("using a;\n"
+            "using a::a;",
+            sortUsingDeclarations("using a::a;\n"
+                                  "using a;"));
+
+  EXPECT_EQ("using a::ba::aa;\n"
+            "using a::bb::ccc;",
+            sortUsingDeclarations("using a::bb::ccc;\n"
+                                  "using a::ba::aa;"));
+
+  EXPECT_EQ("using a;\n"
+            "using typename a;",
+            sortUsingDeclarations("using typename a;\n"
+                                  "using a;"));
+
+  EXPECT_EQ("using typename z;\n"
+            "using typenamea;",
+            sortUsingDeclarations("using typenamea;\n"
+                                  "using typename z;"));
+
+  EXPECT_EQ("using a, b;\n"
+            "using aa;",
+            sortUsingDeclarations("using aa;\n"
+                                  "using a, b;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, SortsMultipleTopLevelDeclarations) {
+  EXPECT_EQ("using a;\n"
+            "using b;\n"
+            "using c;\n"
+            "using d;\n"
+            "using e;",
+            sortUsingDeclarations("using d;\n"
+                                  "using b;\n"
+                                  "using e;\n"
+                                  "using a;\n"
+                                  "using c;"));
+
+  EXPECT_EQ("#include <iostream>\n"
+            "using ::std::endl;\n"
+            "using std::cin;\n"
+            "using std::cout;\n"
+            "int main();",
+            sortUsingDeclarations("#include <iostream>\n"
+                                  "using std::cout;\n"
+                                  "using std::cin;\n"
+                                  "using ::std::endl;\n"
+                                  "int main();"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, BreaksOnEmptyLines) {
+  EXPECT_EQ("using b;\n"
+            "using c;\n"
+            "\n"
+            "using a;\n"
+            "using d;",
+            sortUsingDeclarations("using c;\n"
+                                  "using b;\n"
+                                  "\n"
+                                  "using d;\n"
+                                  "using a;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, BreaksOnUsingNamespace) {
+  EXPECT_EQ("using b;\n"
+            "using namespace std;\n"
+            "using a;",
+            sortUsingDeclarations("using b;\n"
+                                  "using namespace std;\n"
+                                  "using a;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, KeepsUsingDeclarationsInPPDirectives) {
+  EXPECT_EQ("#define A \\\n"
+            "using b;\\\n"
+            "using a;",
+            sortUsingDeclarations("#define A \\\n"
+                                  "using b;\\\n"
+                                  "using a;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, KeepsTypeAliases) {
+  auto Code = "struct C { struct B { struct A; }; };\n"
+              "using B = C::B;\n"
+              "using A = B::A;";
+  EXPECT_EQ(Code, sortUsingDeclarations(Code));
+}
+
+TEST_F(UsingDeclarationsSorterTest, MovesTrailingCommentsWithDeclarations) {
+  EXPECT_EQ("using a; // line a1\n"
+            "using b; /* line b1\n"
+            "          * line b2\n"
+            "          * line b3 */\n"
+            "using c; // line c1\n"
+            "         // line c2",
+            sortUsingDeclarations("using c; // line c1\n"
+                                  "         // line c2\n"
+                                  "using b; /* line b1\n"
+                                  "          * line b2\n"
+                                  "          * line b3 */\n"
+                                  "using a; // line a1"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, SortsInStructScope) {
+  EXPECT_EQ("struct pt3 : pt2 {\n"
+            "  using pt2::x;\n"
+            "  using pt2::y;\n"
+            "  float z;\n"
+            "};",
+            sortUsingDeclarations("struct pt3 : pt2 {\n"
+                                  "  using pt2::y;\n"
+                                  "  using pt2::x;\n"
+                                  "  float z;\n"
+                                  "};"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, KeepsOperators) {
+  EXPECT_EQ("using a::operator();\n"
+            "using a::operator-;\n"
+            "using a::operator+;",
+            sortUsingDeclarations("using a::operator();\n"
+                                  "using a::operator-;\n"
+                                  "using a::operator+;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, SortsUsingDeclarationsInsideNamespaces) {
+  EXPECT_EQ("namespace A {\n"
+            "struct B;\n"
+            "struct C;\n"
+            "}\n"
+            "namespace X {\n"
+            "using A::B;\n"
+            "using A::C;\n"
+            "}",
+            sortUsingDeclarations("namespace A {\n"
+                                  "struct B;\n"
+                                  "struct C;\n"
+                                  "}\n"
+                                  "namespace X {\n"
+                                  "using A::C;\n"
+                                  "using A::B;\n"
+                                  "}"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, SupportsClangFormatOff) {
+  EXPECT_EQ("// clang-format off\n"
+            "using b;\n"
+            "using a;\n"
+            "// clang-format on\n"
+            "using c;\n"
+            "using d;",
+            sortUsingDeclarations("// clang-format off\n"
+                                  "using b;\n"
+                                  "using a;\n"
+                                  "// clang-format on\n"
+                                  "using d;\n"
+                                  "using c;"));
+}
+
+TEST_F(UsingDeclarationsSorterTest, SortsPartialRangeOfUsingDeclarations) {
+  EXPECT_EQ("using b;\n"
+            "using a;\n"
+            "using c;",
+            sortUsingDeclarations("using b;\n"
+                                  "using c;\n" // starts at offset 10
+                                  "using a;",
+                                  {tooling::Range(10, 15)}));
+}
+
+} // end namespace
+} // end namespace format
+} // end namespace clang
diff --git a/utils/TableGen/NeonEmitter.cpp b/utils/TableGen/NeonEmitter.cpp
index 49c1edce3220..62fcccbacb55 100644
--- a/utils/TableGen/NeonEmitter.cpp
+++ b/utils/TableGen/NeonEmitter.cpp
@@ -860,6 +860,10 @@ void Type::applyModifier(char Mod) {
     Float = true;
     ElementBitwidth = 64;
     break;
+  case 'H':
+    Float = true;
+    ElementBitwidth = 16;
+    break;
   case 'g':
     if (AppliedQuad)
       Bitwidth /= 2;
@@ -1006,7 +1010,7 @@ std::string Intrinsic::getInstTypeCode(Type T, ClassKind CK) const {
 }
 
 static bool isFloatingPointProtoModifier(char Mod) {
-  return Mod == 'F' || Mod == 'f';
+  return Mod == 'F' || Mod == 'f' || Mod == 'H';
 }
 
 std::string Intrinsic::getBuiltinTypeStr() {
diff --git a/utils/bash-autocomplete.sh b/utils/bash-autocomplete.sh
index de4a435b8849..ba908bcc0bc3 100755
--- a/utils/bash-autocomplete.sh
+++ b/utils/bash-autocomplete.sh
@@ -1,13 +1,38 @@
 # Please add "source /path/to/bash-autocomplete.sh" to your .bashrc to use this.
 _clang()
 {
-  local cur prev words cword flags
+  local cur prev words cword arg
   _init_completion -n : || return
 
-  flags=$( clang --autocomplete="$cur" )
-  if [[ "$flags" == "" || "$cur" == "" ]]; then
+  # bash always separates '=' as a token even if there's no space before/after '='.
+  # On the other hand, '=' is just a regular character for clang options that
+  # contain '='. For example, "-stdlib=" is defined as is, instead of "-stdlib" and "=".
+  # So, we need to partially undo bash tokenization here for integrity.
+  local w1="${COMP_WORDS[$cword - 1]}"
+  local w2="${COMP_WORDS[$cword - 2]}"
+  if [[ "$cur" == -* ]]; then
+    # -foo<tab>
+    arg="$cur"
+  elif [[ "$w1" == -*  && "$cur" == '=' ]]; then
+    # -foo=<tab>
+    arg="$w1=,"
+  elif [[ "$w1" == -* ]]; then
+    # -foo <tab> or -foo bar<tab>
+    arg="$w1,$cur"
+  elif [[ "$w2" == -* && "$w1" == '=' ]]; then
+    # -foo=bar<tab>
+    arg="$w2=,$cur"
+  fi
+
+  local flags=$( clang --autocomplete="$arg" )
+  if [[ "$cur" == '=' ]]; then
+    COMPREPLY=( $( compgen -W "$flags" -- "") )
+  elif [[ "$flags" == "" || "$arg" == "" ]]; then
     _filedir
   else
+    # Bash automatically appends a space after '=' by default.
+    # Disable it so that it works nicely for options in the form of -foo=bar.
+    [[ "${flags: -1}" == '=' ]] && compopt -o nospace
     COMPREPLY=( $( compgen -W "$flags" -- "$cur" ) )
   fi
 }