IR: Add COMDATs to the IR

This new IR facility allows us to represent the object-file semantic of
a COMDAT group.

COMDATs allow us to tie together sections and make the inclusion of one
dependent on another. This is required to implement features like MS
ABI VFTables and optimizing away certain kinds of initialization in C++.

This functionality is only representable in COFF and ELF, Mach-O has no
similar mechanism.

Differential Revision: http://reviews.llvm.org/D4178

llvm-svn: 211920

Author: majnemer

llvm/docs/LangRef.rst

@@ -562,6 +562,8 @@ is zero. The address space qualifier must precede any other attributes.
 
 LLVM allows an explicit section to be specified for globals. If the
 target supports it, it will emit globals to the section specified.
+Additionally, the global can placed in a comdat if the target has the necessary
+support.
 
 By default, global initializers are optimized by assuming that global
 variables defined within the module are not modified from their
@@ -627,8 +629,9 @@ an optional ``unnamed_addr`` attribute, a return type, an optional
 :ref:`parameter attribute <paramattrs>` for the return type, a function
 name, a (possibly empty) argument list (each with optional :ref:`parameter
 attributes <paramattrs>`), optional :ref:`function attributes <fnattrs>`,
-an optional section, an optional alignment, an optional :ref:`garbage
-collector name <gc>`, an optional :ref:`prefix <prefixdata>`, an opening
+an optional section, an optional alignment,
+an optional :ref:`comdat <langref_comdats>`,
+an optional :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`, an opening
 curly brace, a list of basic blocks, and a closing curly brace.
 
 LLVM function declarations consist of the "``declare``" keyword, an
@@ -658,6 +661,7 @@ predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`.
 
 LLVM allows an explicit section to be specified for functions. If the
 target supports it, it will emit functions to the section specified.
+Additionally, the function can placed in a COMDAT.
 
 An explicit alignment may be specified for a function. If not present,
 or if the alignment is set to zero, the alignment of the function is set
@@ -673,8 +677,8 @@ Syntax::
     define [linkage] [visibility] [DLLStorageClass]
            [cconv] [ret attrs]
            <ResultType> @<FunctionName> ([argument list])
-           [unnamed_addr] [fn Attrs] [section "name"] [align N]
-           [gc] [prefix Constant] { ... }
+           [unnamed_addr] [fn Attrs] [section "name"] [comdat $<ComdatName>]
+           [align N] [gc] [prefix Constant] { ... }
 
 .. _langref_aliases:
 
@@ -716,6 +720,89 @@ some can only be checked when producing an object file:
 * No global value in the expression can be a declaration, since that
   would require a relocation, which is not possible.
 
+.. _langref_comdats:
+
+Comdats
+-------
+
+Comdat IR provides access to COFF and ELF object file COMDAT functionality.
+
+Comdats have a name which represents the COMDAT key.  All global objects which
+specify this key will only end up in the final object file if the linker chooses
+that key over some other key.  Aliases are placed in the same COMDAT that their
+aliasee computes to, if any.
+
+Comdats have a selection kind to provide input on how the linker should
+choose between keys in two different object files.
+
+Syntax::
+
+    $<Name> = comdat SelectionKind
+
+The selection kind must be one of the following:
+
+``any``
+    The linker may choose any COMDAT key, the choice is arbitrary.
+``exactmatch``
+    The linker may choose any COMDAT key but the sections must contain the
+    same data.
+``largest``
+    The linker will choose the section containing the largest COMDAT key.
+``noduplicates``
+    The linker requires that only section with this COMDAT key exist.
+``samesize``
+    The linker may choose any COMDAT key but the sections must contain the
+    same amount of data.
+
+Note that the Mach-O platform doesn't support COMDATs and ELF only supports
+``any`` as a selection kind.
+
+Here is an example of a COMDAT group where a function will only be selected if
+the COMDAT key's section is the largest:
+
+.. code-block:: llvm
+
+   $foo = comdat largest
+   @foo = global i32 2, comdat $foo
+
+   define void @bar() comdat $foo {
+     ret void
+   }
+
+In a COFF object file, this will create a COMDAT section with selection kind
+``IMAGE_COMDAT_SELECT_LARGEST`` containing the contents of the ``@foo`` symbol
+and another COMDAT section with selection kind
+``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT
+section and contains the contents of the ``@baz`` symbol.
+
+There are some restrictions on the properties of the global object.
+It, or an alias to it, must have the same name as the COMDAT group when
+targeting COFF.
+The contents and size of this object may be used during link-time to determine
+which COMDAT groups get selected depending on the selection kind.
+Because the name of the object must match the name of the COMDAT group, the
+linkage of the global object must not be local; local symbols can get renamed
+if a collision occurs in the symbol table.
+
+The combined use of COMDATS and section attributes may yield surprising results.
+For example:
+
+.. code-block:: llvm
+
+   $foo = comdat any
+   $bar = comdat any
+   @g1 = global i32 42, section "sec", comdat $foo
+   @g2 = global i32 42, section "sec", comdat $bar
+
+From the object file perspective, this requires the creation of two sections
+with the same name.  This is necessary because both globals belong to different
+COMDAT groups and COMDATs, at the object file level, are represented by
+sections.
+
+Note that certain IR constructs like global variables and functions may create
+COMDATs in the object file in addition to any which are specified using COMDAT
+IR.  This arises, for example, when a global variable has linkonce_odr linkage.
+
 .. _namedmetadatastructure:
 
 Named Metadata

llvm/include/llvm/ADT/UniqueVector.h

@@ -22,13 +22,18 @@ namespace llvm {
 /// class should have an implementation of operator== and of operator<.
 /// Entries can be fetched using operator[] with the entry ID.
 template<class T> class UniqueVector {
+public:
+  typedef typename std::vector<T> VectorType;
+  typedef typename VectorType::iterator iterator;
+  typedef typename VectorType::const_iterator const_iterator;
+
 private:
   // Map - Used to handle the correspondence of entry to ID.
   std::map<T, unsigned> Map;
 
   // Vector - ID ordered vector of entries. Entries can be indexed by ID - 1.
   //
-  std::vector<T> Vector;
+  VectorType Vector;
 
 public:
   /// insert - Append entry to the vector if it doesn't already exist.  Returns
@@ -68,6 +73,18 @@ template<class T> class UniqueVector {
     return Vector[ID - 1];
   }
 
+  /// \brief Return an iterator to the start of the vector.
+  iterator begin() { return Vector.begin(); }
+
+  /// \brief Return an iterator to the start of the vector.
+  const_iterator begin() const { return Vector.begin(); }
+
+  /// \brief Return an iterator to the end of the vector.
+  iterator end() { return Vector.end(); }
+
+  /// \brief Return an iterator to the end of the vector.
+  const_iterator end() const { return Vector.end(); }
+
   /// size - Returns the number of entries in the vector.
   ///
   size_t size() const { return Vector.size(); }

llvm/include/llvm/Bitcode/LLVMBitCodes.h

@@ -71,7 +71,8 @@ namespace bitc {
     // MODULE_CODE_PURGEVALS: [numvals]
     MODULE_CODE_PURGEVALS   = 10,
 
-    MODULE_CODE_GCNAME      = 11   // GCNAME: [strchr x N]
+    MODULE_CODE_GCNAME      = 11,  // GCNAME: [strchr x N]
+    MODULE_CODE_COMDAT      = 12,  // COMDAT: [selection_kind, name]
   };
 
   /// PARAMATTR blocks have code for defining a parameter attribute set.
@@ -376,6 +377,14 @@ namespace bitc {
     ATTR_KIND_JUMP_TABLE = 40
   };
 
+  enum ComdatSelectionKindCodes {
+    COMDAT_SELECTION_KIND_ANY = 1,
+    COMDAT_SELECTION_KIND_EXACT_MATCH = 2,
+    COMDAT_SELECTION_KIND_LARGEST = 3,
+    COMDAT_SELECTION_KIND_NO_DUPLICATES = 4,
+    COMDAT_SELECTION_KIND_SAME_SIZE = 5,
+  };
+
 } // End bitc namespace
 } // End llvm namespace
 

llvm/include/llvm/IR/Comdat.h

@@ -0,0 +1,66 @@
+//===-- llvm/IR/Comdat.h - Comdat definitions -------------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+/// @file
+/// This file contains the declaration of the Comdat class, which represents a
+/// single COMDAT in LLVM.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_IR_COMDAT_H
+#define LLVM_IR_COMDAT_H
+
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/Compiler.h"
+
+namespace llvm {
+
+class raw_ostream;
+template <typename ValueTy> class StringMapEntry;
+
+// This is a Name X SelectionKind pair. The reason for having this be an
+// independent object instead of just adding the name and the SelectionKind
+// to a GlobalObject is that it is invalid to have two Comdats with the same
+// name but different SelectionKind. This structure makes that unrepresentable.
+class Comdat {
+public:
+  enum SelectionKind {
+    Any,          ///< The linker may choose any COMDAT.
+    ExactMatch,   ///< The data referenced by the COMDAT must be the same.
+    Largest,      ///< The linker will choose the largest COMDAT.
+    NoDuplicates, ///< No other Module may specify this COMDAT.
+    SameSize,     ///< The data referenced by the COMDAT must be the same size.
+  };
+
+  Comdat(Comdat &&C);
+  SelectionKind getSelectionKind() const { return SK; }
+  void setSelectionKind(SelectionKind Val) { SK = Val; }
+  StringRef getName() const;
+  void print(raw_ostream &OS) const;
+  void dump() const;
+
+private:
+  friend class Module;
+  Comdat();
+  Comdat(SelectionKind SK, StringMapEntry<Comdat> *Name);
+  Comdat(const Comdat &) LLVM_DELETED_FUNCTION;
+
+  // Points to the map in Module.
+  StringMapEntry<Comdat> *Name;
+  SelectionKind SK;
+};
+
+inline raw_ostream &operator<<(raw_ostream &OS, const Comdat &C) {
+  C.print(OS);
+  return OS;
+}
+
+} // end llvm namespace
+
+#endif

llvm/include/llvm/IR/GlobalAlias.h

@@ -87,6 +87,21 @@ class GlobalAlias : public GlobalValue, public ilist_node<GlobalAlias> {
     return getOperand(0);
   }
 
+  const GlobalObject *getBaseObject() const {
+    return const_cast<GlobalAlias *>(this)->getBaseObject();
+  }
+  GlobalObject *getBaseObject() {
+    return dyn_cast<GlobalObject>(getAliasee()->stripInBoundsOffsets());
+  }
+
+  const GlobalObject *getBaseObject(const DataLayout &DL, APInt &Offset) const {
+    return const_cast<GlobalAlias *>(this)->getBaseObject(DL, Offset);
+  }
+  GlobalObject *getBaseObject(const DataLayout &DL, APInt &Offset) {
+    return dyn_cast<GlobalObject>(
+        getAliasee()->stripAndAccumulateInBoundsConstantOffsets(DL, Offset));
+  }
+
   static bool isValidLinkage(LinkageTypes L) {
     return isExternalLinkage(L) || isLocalLinkage(L) ||
       isWeakLinkage(L) || isLinkOnceLinkage(L);

llvm/include/llvm/IR/GlobalObject.h

@@ -20,7 +20,7 @@
 #include "llvm/IR/GlobalValue.h"
 
 namespace llvm {
-
+class Comdat;
 class Module;
 
 class GlobalObject : public GlobalValue {
@@ -29,11 +29,12 @@ class GlobalObject : public GlobalValue {
 protected:
   GlobalObject(Type *Ty, ValueTy VTy, Use *Ops, unsigned NumOps,
                LinkageTypes Linkage, const Twine &Name)
-      : GlobalValue(Ty, VTy, Ops, NumOps, Linkage, Name) {
+      : GlobalValue(Ty, VTy, Ops, NumOps, Linkage, Name), ObjComdat(nullptr) {
     setGlobalValueSubClassData(0);
   }
 
   std::string Section;     // Section to emit this into, empty means default
+  Comdat *ObjComdat;
 public:
   unsigned getAlignment() const {
     return (1u << getGlobalValueSubClassData()) >> 1;
@@ -44,6 +45,11 @@ class GlobalObject : public GlobalValue {
   const char *getSection() const { return Section.c_str(); }
   void setSection(StringRef S);
 
+  bool hasComdat() const { return getComdat() != nullptr; }
+  const Comdat *getComdat() const { return ObjComdat; }
+  Comdat *getComdat() { return ObjComdat; }
+  void setComdat(Comdat *C) { ObjComdat = C; }
+
   void copyAttributesFrom(const GlobalValue *Src) override;
 
   // Methods for support type inquiry through isa, cast, and dyn_cast:

llvm/include/llvm/IR/GlobalValue.h

@@ -23,6 +23,7 @@
 
 namespace llvm {
 
+class Comdat;
 class PointerType;
 class Module;
 
@@ -110,6 +111,12 @@ class GlobalValue : public Constant {
   bool hasUnnamedAddr() const { return UnnamedAddr; }
   void setUnnamedAddr(bool Val) { UnnamedAddr = Val; }
 
+  bool hasComdat() const { return getComdat() != nullptr; }
+  Comdat *getComdat();
+  const Comdat *getComdat() const {
+    return const_cast<GlobalValue *>(this)->getComdat();
+  }
+
   VisibilityTypes getVisibility() const { return VisibilityTypes(Visibility); }
   bool hasDefaultVisibility() const { return Visibility == DefaultVisibility; }
   bool hasHiddenVisibility() const { return Visibility == HiddenVisibility; }

llvm/include/llvm/IR/Module.h

@@ -16,6 +16,7 @@
 #define LLVM_IR_MODULE_H
 
 #include "llvm/ADT/iterator_range.h"
+#include "llvm/IR/Comdat.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/GlobalAlias.h"
@@ -123,6 +124,8 @@ class Module {
   typedef iplist<GlobalAlias> AliasListType;
   /// The type for the list of named metadata.
   typedef ilist<NamedMDNode> NamedMDListType;
+  /// The type of the comdat "symbol" table.
+  typedef StringMap<Comdat> ComdatSymTabType;
 
   /// The Global Variable iterator.
   typedef GlobalListType::iterator                      global_iterator;
@@ -197,6 +200,7 @@ class Module {
   NamedMDListType NamedMDList;    ///< The named metadata in the module
   std::string GlobalScopeAsm;     ///< Inline Asm at global scope.
   ValueSymbolTable *ValSymTab;    ///< Symbol table for values
+  ComdatSymTabType ComdatSymTab;  ///< Symbol table for COMDATs
   std::unique_ptr<GVMaterializer>
   Materializer;                   ///< Used to materialize GlobalValues
   std::string ModuleID;           ///< Human readable identifier for the module
@@ -403,6 +407,14 @@ class Module {
   /// Remove the given NamedMDNode from this module and delete it.
   void eraseNamedMetadata(NamedMDNode *NMD);
 
+/// @}
+/// @name Comdat Accessors
+/// @{
+
+  /// Return the Comdat in the module with the specified name. It is created
+  /// if it didn't already exist.
+  Comdat *getOrInsertComdat(StringRef Name);
+
 /// @}
 /// @name Module Flags Accessors
 /// @{
@@ -504,6 +516,10 @@ class Module {
   const ValueSymbolTable &getValueSymbolTable() const { return *ValSymTab; }
   /// Get the Module's symbol table of global variable and function identifiers.
   ValueSymbolTable       &getValueSymbolTable()       { return *ValSymTab; }
+  /// Get the Module's symbol table for COMDATs (constant).
+  const ComdatSymTabType &getComdatSymbolTable() const { return ComdatSymTab; }
+  /// Get the Module's symbol table for COMDATs.
+  ComdatSymTabType &getComdatSymbolTable() { return ComdatSymTab; }
 
 /// @}
 /// @name Global Variable Iteration

llvm/include/llvm/Linker/Linker.h

@@ -15,6 +15,8 @@
 
 namespace llvm {
 
+class Comdat;
+class GlobalValue;
 class Module;
 class StringRef;
 class StructType;