Extend the IL for selecting TLS models (PR9788)

This allows the user/front-end to specify a model that is better
than what LLVM would choose by default. For example, a variable
might be declared as

  @x = thread_local(initialexec) global i32 42

if it will not be used in a shared library that is dlopen'ed.

If the specified model isn't supported by the target, or if LLVM can
make a better choice, a different model may be used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@159077 91177308-0d34-0410-b5e6-96231b3b80d8

docs/BitCodeFormat.html

@@ -864,7 +864,7 @@ library name referenced.</p>
 <h4><a name="MODULE_CODE_GLOBALVAR">MODULE_CODE_GLOBALVAR Record</a></h4>
 
 <div>
-<p><tt>[GLOBALVAR, pointer type, isconst, initid, linkage, alignment, section, visibility, threadlocal]</tt></p>
+<p><tt>[GLOBALVAR, pointer type, isconst, initid, linkage, alignment, section, visibility, threadlocal, unnamed_addr]</tt></p>
 
 <p>The <tt>GLOBALVAR</tt> record (code 7) marks the declaration or
 definition of a global variable. The operand fields are:</p>
@@ -915,8 +915,16 @@ encoding of the visibility of this variable:
   </ul>
 </li>
 
-<li><i>threadlocal</i>: If present and non-zero, indicates that the variable
+<li><i>threadlocal</i>: If present, an encoding of the thread local storage
-is <tt>thread_local</tt></li>
+mode of the variable:
+  <ul>
+    <li><tt>not thread local</tt>: code 0</li>
+    <li><tt>thread local; default TLS model</tt>: code 1</li>
+    <li><tt>localdynamic</tt>: code 2</li>
+    <li><tt>initialexec</tt>: code 3</li>
+    <li><tt>localexec</tt>: code 4</li>
+  </ul>
+</li>
 
 <li><i>unnamed_addr</i>: If present and non-zero, indicates that the variable
 has <tt>unnamed_addr</tt></li>

docs/LangRef.html

@@ -838,9 +838,32 @@ define i32 @main() {   <i>; i32()* </i>&nbsp;
 <p>Global variables define regions of memory allocated at compilation time
    instead of run-time.  Global variables may optionally be initialized, may
    have an explicit section to be placed in, and may have an optional explicit
-   alignment specified.  A variable may be defined as "thread_local", which
+   alignment specified.</p>
+
+<p>A variable may be defined as <tt>thread_local</tt>, which
    means that it will not be shared by threads (each thread will have a
-   separated copy of the variable).  A variable may be defined as a global
+   separated copy of the variable).  Not all targets support thread-local
+   variables.  Optionally, a TLS model may be specified:</p>
+
+<dl>
+  <dt><b><tt>localdynamic</tt></b>:</dt>
+  <dd>For variables that are only used within the current shared library.</dd>
+
+  <dt><b><tt>initialexec</tt></b>:</dt>
+  <dd>For variables in modules that will not be loaded dynamically.</dd>
+
+  <dt><b><tt>localexec</tt></b>:</dt>
+  <dd>For variables defined in the executable and only used within it.</dd>
+</dl>
+
+<p>The models correspond to the ELF TLS models; see
+   <a href="http://people.redhat.com/drepper/tls.pdf">ELF
+   Handling For Thread-Local Storage</a> for more information on under which
+   circumstances the different models may be used.  The target may choose a
+   different TLS model if the specified model is not supported, or if a better
+   choice of model can be made.</p>
+
+<p>A variable may be defined as a global
    "constant," which indicates that the contents of the variable
    will <b>never</b> be modified (enabling better optimization, allowing the
    global data to be placed in the read-only section of an executable, etc).
@@ -893,6 +916,13 @@ define i32 @main() {   <i>; i32()* </i>&nbsp;
 @G = addrspace(5) constant float 1.0, section "foo", align 4
 </pre>
 
+<p>The following example defines a thread-local global with
+   the <tt>initialexec</tt> TLS model:</p>
+
+<pre class="doc_code">
+@G = thread_local(initialexec) global i32 0, align 4
+</pre>
+
 </div>
 
 

include/llvm/GlobalVariable.h

@@ -41,24 +41,47 @@ class GlobalVariable : public GlobalValue, public ilist_node<GlobalVariable> {
   void setParent(Module *parent);
 
   bool isConstantGlobal : 1;           // Is this a global constant?
-  bool isThreadLocalSymbol : 1;        // Is this symbol "Thread Local"?
+  unsigned threadLocalMode : 3;        // Is this symbol "Thread Local",
+                                       // if so, what is the desired model?
 
 public:
   // allocate space for exactly one operand
   void *operator new(size_t s) {
     return User::operator new(s, 1);
   }
-  /// GlobalVariable ctor - If a parent module is specified, the global is
+
-  /// automatically inserted into the end of the specified modules global list.
+  enum ThreadLocalMode {
+    NotThreadLocal = 0,
+    GeneralDynamicTLSModel,
+    LocalDynamicTLSModel,
+    InitialExecTLSModel,
+    LocalExecTLSModel
+  };
+
+  // TODO: Remove these once Clang is updated.
   GlobalVariable(Type *Ty, bool isConstant, LinkageTypes Linkage,
                  Constant *Initializer = 0, const Twine &Name = "",
                  bool ThreadLocal = false, unsigned AddressSpace = 0);
+  GlobalVariable(Module &M, Type *Ty, bool isConstant,
+                 LinkageTypes Linkage, Constant *Initializer,
+                 const Twine &Name = "",
+                 GlobalVariable *InsertBefore = 0, bool ThreadLocal = false,
+                 unsigned AddressSpace = 0);
+
+  /// GlobalVariable ctor - If a parent module is specified, the global is
+  /// automatically inserted into the end of the specified modules global list.
+  // TODO: Put default param values back when ctors above are removed.
+  GlobalVariable(Type *Ty, bool isConstant, LinkageTypes Linkage,
+                 Constant *Initializer, const Twine &Name,
+                 ThreadLocalMode, unsigned AddressSpace = 0);
   /// GlobalVariable ctor - This creates a global and inserts it before the
   /// specified other global.
+  // TODO: Put default param values back when ctors above are removed.
   GlobalVariable(Module &M, Type *Ty, bool isConstant,
                  LinkageTypes Linkage, Constant *Initializer,
                  const Twine &Name,
-                 GlobalVariable *InsertBefore = 0, bool ThreadLocal = false,
+                 GlobalVariable *InsertBefore,
+                 ThreadLocalMode,
                  unsigned AddressSpace = 0);
 
   ~GlobalVariable() {
@@ -135,8 +158,14 @@ public:
   void setConstant(bool Val) { isConstantGlobal = Val; }
 
   /// If the value is "Thread Local", its value isn't shared by the threads.
-  bool isThreadLocal() const { return isThreadLocalSymbol; }
+  bool isThreadLocal() const { return threadLocalMode != NotThreadLocal; }
-  void setThreadLocal(bool Val) { isThreadLocalSymbol = Val; }
+  void setThreadLocal(bool Val) {
+    threadLocalMode = Val ? GeneralDynamicTLSModel : NotThreadLocal;
+  }
+  void setThreadLocalMode(ThreadLocalMode Val) { threadLocalMode = Val; }
+  ThreadLocalMode getThreadLocalMode() const {
+    return static_cast<ThreadLocalMode>(threadLocalMode);
+  }
 
   /// copyAttributesFrom - copy all additional attributes (those not needed to
   /// create a GlobalVariable) from the GlobalVariable Src to this one.

lib/AsmParser/LLLexer.cpp

@@ -474,6 +474,9 @@ lltok::Kind LLLexer::LexIdentifier() {
   KEYWORD(extern_weak);
   KEYWORD(external);
   KEYWORD(thread_local);
+  KEYWORD(localdynamic);
+  KEYWORD(initialexec);
+  KEYWORD(localexec);
   KEYWORD(zeroinitializer);
   KEYWORD(undef);
   KEYWORD(null);

lib/AsmParser/LLParser.cpp

@@ -645,12 +645,13 @@ bool LLParser::ParseGlobal(const std::string &Name, LocTy NameLoc,
                            unsigned Linkage, bool HasLinkage,
                            unsigned Visibility) {
   unsigned AddrSpace;
-  bool ThreadLocal, IsConstant, UnnamedAddr;
+  bool IsConstant, UnnamedAddr;
+  GlobalVariable::ThreadLocalMode TLM;
   LocTy UnnamedAddrLoc;
   LocTy TyLoc;
 
   Type *Ty = 0;
-  if (ParseOptionalToken(lltok::kw_thread_local, ThreadLocal) ||
+  if (ParseOptionalThreadLocal(TLM) ||
       ParseOptionalAddrSpace(AddrSpace) ||
       ParseOptionalToken(lltok::kw_unnamed_addr, UnnamedAddr,
                          &UnnamedAddrLoc) ||
@@ -691,7 +692,8 @@ bool LLParser::ParseGlobal(const std::string &Name, LocTy NameLoc,
 
   if (GV == 0) {
     GV = new GlobalVariable(*M, Ty, false, GlobalValue::ExternalLinkage, 0,
-                            Name, 0, false, AddrSpace);
+                            Name, 0, GlobalVariable::NotThreadLocal,
+                            AddrSpace);
   } else {
     if (GV->getType()->getElementType() != Ty)
       return Error(TyLoc,
@@ -710,7 +712,7 @@ bool LLParser::ParseGlobal(const std::string &Name, LocTy NameLoc,
   GV->setConstant(IsConstant);
   GV->setLinkage((GlobalValue::LinkageTypes)Linkage);
   GV->setVisibility((GlobalValue::VisibilityTypes)Visibility);
-  GV->setThreadLocal(ThreadLocal);
+  GV->setThreadLocalMode(TLM);
   GV->setUnnamedAddr(UnnamedAddr);
 
   // Parse attributes on the global.
@@ -858,6 +860,46 @@ bool LLParser::ParseUInt32(unsigned &Val) {
   return false;
 }
 
+/// ParseTLSModel
+///   := 'localdynamic'
+///   := 'initialexec'
+///   := 'localexec'
+bool LLParser::ParseTLSModel(GlobalVariable::ThreadLocalMode &TLM) {
+  switch (Lex.getKind()) {
+    default:
+      return TokError("expected localdynamic, initialexec or localexec");
+    case lltok::kw_localdynamic:
+      TLM = GlobalVariable::LocalDynamicTLSModel;
+      break;
+    case lltok::kw_initialexec:
+      TLM = GlobalVariable::InitialExecTLSModel;
+      break;
+    case lltok::kw_localexec:
+      TLM = GlobalVariable::LocalExecTLSModel;
+      break;
+  }
+
+  Lex.Lex();
+  return false;
+}
+
+/// ParseOptionalThreadLocal
+///   := /*empty*/
+///   := 'thread_local'
+///   := 'thread_local' '(' tlsmodel ')'
+bool LLParser::ParseOptionalThreadLocal(GlobalVariable::ThreadLocalMode &TLM) {
+  TLM = GlobalVariable::NotThreadLocal;
+  if (!EatIfPresent(lltok::kw_thread_local))
+    return false;
+
+  TLM = GlobalVariable::GeneralDynamicTLSModel;
+  if (Lex.getKind() == lltok::lparen) {
+    Lex.Lex();
+    return ParseTLSModel(TLM) ||
+      ParseToken(lltok::rparen, "expected ')' after thread local model");
+  }
+  return false;
+}
 
 /// ParseOptionalAddrSpace
 ///   := /*empty*/

lib/AsmParser/LLParser.h

@@ -171,6 +171,9 @@ namespace llvm {
       Loc = Lex.getLoc();
       return ParseUInt32(Val);
     }
+
+    bool ParseTLSModel(GlobalVariable::ThreadLocalMode &TLM);
+    bool ParseOptionalThreadLocal(GlobalVariable::ThreadLocalMode &TLM);
     bool ParseOptionalAddrSpace(unsigned &AddrSpace);
     bool ParseOptionalAttrs(Attributes &Attrs, unsigned AttrKind);
     bool ParseOptionalLinkage(unsigned &Linkage, bool &HasLinkage);

lib/AsmParser/LLToken.h

@@ -44,6 +44,7 @@ namespace lltok {
     kw_unnamed_addr,
     kw_extern_weak,
     kw_external, kw_thread_local,
+    kw_localdynamic, kw_initialexec, kw_localexec,
     kw_zeroinitializer,
     kw_undef, kw_null,
     kw_to,

lib/Bitcode/Reader/BitcodeReader.cpp

@@ -102,6 +102,17 @@ static GlobalValue::VisibilityTypes GetDecodedVisibility(unsigned Val) {
   }
 }
 
+static GlobalVariable::ThreadLocalMode GetDecodedThreadLocalMode(unsigned Val) {
+  switch (Val) {
+    case 0: return GlobalVariable::NotThreadLocal;
+    default: // Map unknown non-zero value to general dynamic.
+    case 1: return GlobalVariable::GeneralDynamicTLSModel;
+    case 2: return GlobalVariable::LocalDynamicTLSModel;
+    case 3: return GlobalVariable::InitialExecTLSModel;
+    case 4: return GlobalVariable::LocalExecTLSModel;
+  }
+}
+
 static int GetDecodedCastOpcode(unsigned Val) {
   switch (Val) {
   default: return -1;
@@ -1544,9 +1555,10 @@ bool BitcodeReader::ParseModule(bool Resume) {
       GlobalValue::VisibilityTypes Visibility = GlobalValue::DefaultVisibility;
       if (Record.size() > 6)
         Visibility = GetDecodedVisibility(Record[6]);
-      bool isThreadLocal = false;
+
+      GlobalVariable::ThreadLocalMode TLM = GlobalVariable::NotThreadLocal;
       if (Record.size() > 7)
-        isThreadLocal = Record[7];
+        TLM = GetDecodedThreadLocalMode(Record[7]);
 
       bool UnnamedAddr = false;
       if (Record.size() > 8)
@@ -1554,12 +1566,11 @@ bool BitcodeReader::ParseModule(bool Resume) {
 
       GlobalVariable *NewGV =
         new GlobalVariable(*TheModule, Ty, isConstant, Linkage, 0, "", 0,
-                           isThreadLocal, AddressSpace);
+                           TLM, AddressSpace);
       NewGV->setAlignment(Alignment);
       if (!Section.empty())
         NewGV->setSection(Section);
       NewGV->setVisibility(Visibility);
-      NewGV->setThreadLocal(isThreadLocal);
       NewGV->setUnnamedAddr(UnnamedAddr);
 
       ValueList.push_back(NewGV);

lib/Bitcode/Writer/BitcodeWriter.cpp

@@ -379,6 +379,17 @@ static unsigned getEncodedVisibility(const GlobalValue *GV) {
   llvm_unreachable("Invalid visibility");
 }
 
+static unsigned getEncodedThreadLocalMode(const GlobalVariable *GV) {
+  switch (GV->getThreadLocalMode()) {
+    case GlobalVariable::NotThreadLocal:         return 0;
+    case GlobalVariable::GeneralDynamicTLSModel: return 1;
+    case GlobalVariable::LocalDynamicTLSModel:   return 2;
+    case GlobalVariable::InitialExecTLSModel:    return 3;
+    case GlobalVariable::LocalExecTLSModel:      return 4;
+  }
+  llvm_unreachable("Invalid TLS model");
+}
+
 // Emit top-level description of module, including target triple, inline asm,
 // descriptors for global variables, and function prototype info.
 static void WriteModuleInfo(const Module *M, const ValueEnumerator &VE,
@@ -487,7 +498,7 @@ static void WriteModuleInfo(const Module *M, const ValueEnumerator &VE,
         GV->getVisibility() != GlobalValue::DefaultVisibility ||
         GV->hasUnnamedAddr()) {
       Vals.push_back(getEncodedVisibility(GV));
-      Vals.push_back(GV->isThreadLocal());
+      Vals.push_back(getEncodedThreadLocalMode(GV));
       Vals.push_back(GV->hasUnnamedAddr());
     } else {
       AbbrevToUse = SimpleGVarAbbrev;

lib/Linker/LinkModules.cpp

@@ -684,7 +684,7 @@ bool ModuleLinker::linkAppendingVarProto(GlobalVariable *DstGV,
   GlobalVariable *NG =
     new GlobalVariable(*DstGV->getParent(), NewType, SrcGV->isConstant(),
                        DstGV->getLinkage(), /*init*/0, /*name*/"", DstGV,
-                       DstGV->isThreadLocal(),
+                       DstGV->getThreadLocalMode(),
                        DstGV->getType()->getAddressSpace());
   
   // Propagate alignment, visibility and section info.
@@ -759,7 +759,7 @@ bool ModuleLinker::linkGlobalProto(GlobalVariable *SGV) {
     new GlobalVariable(*DstM, TypeMap.get(SGV->getType()->getElementType()),
                        SGV->isConstant(), SGV->getLinkage(), /*init*/0,
                        SGV->getName(), /*insertbefore*/0,
-                       SGV->isThreadLocal(),
+                       SGV->getThreadLocalMode(),
                        SGV->getType()->getAddressSpace());
   // Propagate alignment, visibility and section info.
   copyGVAttributes(NewDGV, SGV);

lib/Target/CppBackend/CPPBackend.cpp

@@ -130,6 +130,7 @@ namespace {
   private:
     void printLinkageType(GlobalValue::LinkageTypes LT);
     void printVisibilityType(GlobalValue::VisibilityTypes VisTypes);
+    void printThreadLocalMode(GlobalVariable::ThreadLocalMode TLM);
     void printCallingConv(CallingConv::ID cc);
     void printEscapedString(const std::string& str);
     void printCFP(const ConstantFP* CFP);
@@ -325,6 +326,26 @@ void CppWriter::printVisibilityType(GlobalValue::VisibilityTypes VisType) {
   }
 }
 
+void CppWriter::printThreadLocalMode(GlobalVariable::ThreadLocalMode TLM) {
+  switch (TLM) {
+    case GlobalVariable::NotThreadLocal:
+      Out << "GlobalVariable::NotThreadLocal";
+      break;
+    case GlobalVariable::GeneralDynamicTLSModel:
+      Out << "GlobalVariable::GeneralDynamicTLSModel";
+      break;
+    case GlobalVariable::LocalDynamicTLSModel:
+      Out << "GlobalVariable::LocalDynamicTLSModel";
+      break;
+    case GlobalVariable::InitialExecTLSModel:
+      Out << "GlobalVariable::InitialExecTLSModel";
+      break;
+    case GlobalVariable::LocalExecTLSModel:
+      Out << "GlobalVariable::LocalExecTLSModel";
+      break;
+  }
+}
+
 // printEscapedString - Print each character of the specified string, escaping
 // it if it is not printable or if it is an escape char.
 void CppWriter::printEscapedString(const std::string &Str) {
@@ -996,7 +1017,9 @@ void CppWriter::printVariableHead(const GlobalVariable *GV) {
   }
   if (GV->isThreadLocal()) {
     printCppName(GV);
-    Out << "->setThreadLocal(true);";
+    Out << "->setThreadLocalMode(";
+    printThreadLocalMode(GV->getThreadLocalMode());
+    Out << ");";
     nl(Out);
   }
   if (is_inline) {

lib/Target/TargetMachine.cpp

@@ -77,6 +77,24 @@ CodeModel::Model TargetMachine::getCodeModel() const {
   return CodeGenInfo->getCodeModel();
 }
 
+/// Get the IR-specified TLS model for Var.
+static TLSModel::Model getSelectedTLSModel(const GlobalVariable *Var) {
+  switch (Var->getThreadLocalMode()) {
+  case GlobalVariable::NotThreadLocal:
+    llvm_unreachable("getSelectedTLSModel for non-TLS variable");
+    break;
+  case GlobalVariable::GeneralDynamicTLSModel:
+    return TLSModel::GeneralDynamic;
+  case GlobalVariable::LocalDynamicTLSModel:
+    return TLSModel::LocalDynamic;
+  case GlobalVariable::InitialExecTLSModel:
+    return TLSModel::InitialExec;
+  case GlobalVariable::LocalExecTLSModel:
+    return TLSModel::LocalExec;
+  }
+  llvm_unreachable("invalid TLS model");
+}
+
 TLSModel::Model TargetMachine::getTLSModel(const GlobalValue *GV) const {
   // If GV is an alias then use the aliasee for determining
   // thread-localness.
@@ -86,22 +104,31 @@ TLSModel::Model TargetMachine::getTLSModel(const GlobalValue *GV) const {
 
   bool isLocal = Var->hasLocalLinkage();
   bool isDeclaration = Var->isDeclaration();
+  bool isPIC = getRelocationModel() == Reloc::PIC_;
+  bool isPIE = Options.PositionIndependentExecutable;
   // FIXME: what should we do for protected and internal visibility?
   // For variables, is internal different from hidden?
   bool isHidden = Var->hasHiddenVisibility();
 
-  if (getRelocationModel() == Reloc::PIC_ &&
+  TLSModel::Model Model;
-      !Options.PositionIndependentExecutable) {
+  if (isPIC && !isPIE) {
     if (isLocal || isHidden)
-      return TLSModel::LocalDynamic;
+      Model = TLSModel::LocalDynamic;
     else
-      return TLSModel::GeneralDynamic;
+      Model = TLSModel::GeneralDynamic;
   } else {
     if (!isDeclaration || isHidden)
-      return TLSModel::LocalExec;
+      Model = TLSModel::LocalExec;
     else
-      return TLSModel::InitialExec;
+      Model = TLSModel::InitialExec;
   }
+
+  // If the user specified a more specific model, use that.
+  TLSModel::Model SelectedModel = getSelectedTLSModel(Var);
+  if (SelectedModel > Model)
+    return SelectedModel;
+
+  return Model;
 }
 
 /// getOptLevel - Returns the optimization level: None, Less,
@@ -135,4 +162,3 @@ void TargetMachine::setFunctionSections(bool V) {
 void TargetMachine::setDataSections(bool V) {
   DataSections = V;
 }
-

lib/Transforms/IPO/GlobalOpt.cpp

@@ -517,7 +517,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const TargetData &TD) {
       GlobalVariable *NGV = new GlobalVariable(STy->getElementType(i), false,
                                                GlobalVariable::InternalLinkage,
                                                In, GV->getName()+"."+Twine(i),
-                                               GV->isThreadLocal(),
+                                               GV->getThreadLocalMode(),
                                               GV->getType()->getAddressSpace());
       Globals.insert(GV, NGV);
       NewGlobals.push_back(NGV);
@@ -550,7 +550,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const TargetData &TD) {
       GlobalVariable *NGV = new GlobalVariable(STy->getElementType(), false,
                                                GlobalVariable::InternalLinkage,
                                                In, GV->getName()+"."+Twine(i),
-                                               GV->isThreadLocal(),
+                                               GV->getThreadLocalMode(),
                                               GV->getType()->getAddressSpace());
       Globals.insert(GV, NGV);
       NewGlobals.push_back(NGV);
@@ -866,7 +866,7 @@ static GlobalVariable *OptimizeGlobalAddressOfMalloc(GlobalVariable *GV,
                                              UndefValue::get(GlobalType),
                                              GV->getName()+".body",
                                              GV,
-                                             GV->isThreadLocal());
+                                             GV->getThreadLocalMode());
 
   // If there are bitcast users of the malloc (which is typical, usually we have
   // a malloc + bitcast) then replace them with uses of the new global.  Update
@@ -899,7 +899,7 @@ static GlobalVariable *OptimizeGlobalAddressOfMalloc(GlobalVariable *GV,
     new GlobalVariable(Type::getInt1Ty(GV->getContext()), false,
                        GlobalValue::InternalLinkage,
                        ConstantInt::getFalse(GV->getContext()),
-                       GV->getName()+".init", GV->isThreadLocal());
+                       GV->getName()+".init", GV->getThreadLocalMode());
   bool InitBoolUsed = false;
 
   // Loop over all uses of GV, processing them in turn.
@@ -1321,7 +1321,7 @@ static GlobalVariable *PerformHeapAllocSRoA(GlobalVariable *GV, CallInst *CI,
                          PFieldTy, false, GlobalValue::InternalLinkage,
                          Constant::getNullValue(PFieldTy),
                          GV->getName() + ".f" + Twine(FieldNo), GV,
-                         GV->isThreadLocal());
+                         GV->getThreadLocalMode());
     FieldGlobals.push_back(NGV);
 
     unsigned TypeSize = TD->getTypeAllocSize(FieldTy);
@@ -1647,7 +1647,7 @@ static bool TryToShrinkGlobalToBoolean(GlobalVariable *GV, Constant *OtherVal) {
                                              GlobalValue::InternalLinkage,
                                         ConstantInt::getFalse(GV->getContext()),
                                              GV->getName()+".b",
-                                             GV->isThreadLocal());
+                                             GV->getThreadLocalMode());
   GV->getParent()->getGlobalList().insert(GV, NewGV);
 
   Constant *InitVal = GV->getInitializer();
@@ -2054,7 +2054,7 @@ static GlobalVariable *InstallGlobalCtors(GlobalVariable *GCL,
   // Create the new global and insert it next to the existing list.
   GlobalVariable *NGV = new GlobalVariable(CA->getType(), GCL->isConstant(),
                                            GCL->getLinkage(), CA, "",
-                                           GCL->isThreadLocal());
+                                           GCL->getThreadLocalMode());
   GCL->getParent()->getGlobalList().insert(GCL, NGV);
   NGV->takeName(GCL);
 

lib/Transforms/Instrumentation/AddressSanitizer.cpp

@@ -511,7 +511,7 @@ bool AddressSanitizer::insertGlobalRedzones(Module &M) {
     // Create a new global variable with enough space for a redzone.
     GlobalVariable *NewGlobal = new GlobalVariable(
         M, NewTy, G->isConstant(), G->getLinkage(),
-        NewInitializer, "", G, G->isThreadLocal());
+        NewInitializer, "", G, G->getThreadLocalMode());
     NewGlobal->copyAttributesFrom(G);
     NewGlobal->setAlignment(RedzoneSize);
 

lib/Transforms/Instrumentation/GCOVProfiling.cpp

@@ -448,7 +448,7 @@ bool GCOVProfiler::emitProfileArcs() {
         new GlobalVariable(*M, CounterTy, false,
                            GlobalValue::InternalLinkage,
                            Constant::getNullValue(CounterTy),
-                           "__llvm_gcov_ctr", 0, false, 0);
+                           "__llvm_gcov_ctr");
       CountersBySP.push_back(std::make_pair(Counters, (MDNode*)SP));
       
       UniqueVector<BasicBlock *> ComplexEdgePreds;

lib/Transforms/Utils/CloneModule.cpp

@@ -53,7 +53,7 @@ Module *llvm::CloneModule(const Module *M, ValueToValueMapTy &VMap) {
                                             I->isConstant(), I->getLinkage(),
                                             (Constant*) 0, I->getName(),
                                             (GlobalVariable*) 0,
-                                            I->isThreadLocal(),
+                                            I->getThreadLocalMode(),
                                             I->getType()->getAddressSpace());
     GV->copyAttributesFrom(I);
     VMap[I] = GV;

lib/VMCore/AsmWriter.cpp

@@ -1376,6 +1376,26 @@ static void PrintVisibility(GlobalValue::VisibilityTypes Vis,
   }
 }
 
+static void PrintThreadLocalModel(GlobalVariable::ThreadLocalMode TLM,
+                                  formatted_raw_ostream &Out) {
+  switch (TLM) {
+    case GlobalVariable::NotThreadLocal:
+      break;
+    case GlobalVariable::GeneralDynamicTLSModel:
+      Out << "thread_local ";
+      break;
+    case GlobalVariable::LocalDynamicTLSModel:
+      Out << "thread_local(localdynamic) ";
+      break;
+    case GlobalVariable::InitialExecTLSModel:
+      Out << "thread_local(initialexec) ";
+      break;
+    case GlobalVariable::LocalExecTLSModel:
+      Out << "thread_local(localexec) ";
+      break;
+  }
+}
+
 void AssemblyWriter::printGlobal(const GlobalVariable *GV) {
   if (GV->isMaterializable())
     Out << "; Materializable\n";
@@ -1388,8 +1408,8 @@ void AssemblyWriter::printGlobal(const GlobalVariable *GV) {
 
   PrintLinkage(GV->getLinkage(), Out);
   PrintVisibility(GV->getVisibility(), Out);
+  PrintThreadLocalModel(GV->getThreadLocalMode(), Out);
 
-  if (GV->isThreadLocal()) Out << "thread_local ";
   if (unsigned AddressSpace = GV->getType()->getAddressSpace())
     Out << "addrspace(" << AddressSpace << ") ";
   if (GV->hasUnnamedAddr()) Out << "unnamed_addr ";

lib/VMCore/Core.cpp

@@ -1210,7 +1210,7 @@ LLVMValueRef LLVMAddGlobalInAddressSpace(LLVMModuleRef M, LLVMTypeRef Ty,
                                          unsigned AddressSpace) {
   return wrap(new GlobalVariable(*unwrap(M), unwrap(Ty), false,
                                  GlobalValue::ExternalLinkage, 0, Name, 0,
-                                 false, AddressSpace));
+                                 GlobalVariable::NotThreadLocal, AddressSpace));
 }
 
 LLVMValueRef LLVMGetNamedGlobal(LLVMModuleRef M, const char *Name) {

lib/VMCore/Globals.cpp

@@ -80,6 +80,7 @@ bool GlobalValue::isDeclaration() const {
 // GlobalVariable Implementation
 //===----------------------------------------------------------------------===//
 
+// TODO: Remove once clang is updated.
 GlobalVariable::GlobalVariable(Type *Ty, bool constant, LinkageTypes Link,
                                Constant *InitVal, const Twine &Name,
                                bool ThreadLocal, unsigned AddressSpace)
@@ -87,7 +88,51 @@ GlobalVariable::GlobalVariable(Type *Ty, bool constant, LinkageTypes Link,
                 Value::GlobalVariableVal,
                 OperandTraits<GlobalVariable>::op_begin(this),
                 InitVal != 0, Link, Name),
-    isConstantGlobal(constant), isThreadLocalSymbol(ThreadLocal) {
+    isConstantGlobal(constant),
+    threadLocalMode(ThreadLocal ? GeneralDynamicTLSModel : NotThreadLocal) {
+  if (InitVal) {
+    assert(InitVal->getType() == Ty &&
+           "Initializer should be the same type as the GlobalVariable!");
+    Op<0>() = InitVal;
+  }
+
+  LeakDetector::addGarbageObject(this);
+}
+
+// TODO: Remove once clang is updated.
+GlobalVariable::GlobalVariable(Module &M, Type *Ty, bool constant,
+                               LinkageTypes Link, Constant *InitVal,
+                               const Twine &Name,
+                               GlobalVariable *Before, bool ThreadLocal,
+                               unsigned AddressSpace)
+  : GlobalValue(PointerType::get(Ty, AddressSpace),
+                Value::GlobalVariableVal,
+                OperandTraits<GlobalVariable>::op_begin(this),
+                InitVal != 0, Link, Name),
+    isConstantGlobal(constant),
+    threadLocalMode(ThreadLocal ? GeneralDynamicTLSModel : NotThreadLocal) {
+  if (InitVal) {
+    assert(InitVal->getType() == Ty &&
+           "Initializer should be the same type as the GlobalVariable!");
+    Op<0>() = InitVal;
+  }
+
+  LeakDetector::addGarbageObject(this);
+
+  if (Before)
+    Before->getParent()->getGlobalList().insert(Before, this);
+  else
+    M.getGlobalList().push_back(this);
+}
+
+GlobalVariable::GlobalVariable(Type *Ty, bool constant, LinkageTypes Link,
+                               Constant *InitVal, const Twine &Name,
+                               ThreadLocalMode TLMode, unsigned AddressSpace)
+  : GlobalValue(PointerType::get(Ty, AddressSpace),
+                Value::GlobalVariableVal,
+                OperandTraits<GlobalVariable>::op_begin(this),
+                InitVal != 0, Link, Name),
+    isConstantGlobal(constant), threadLocalMode(TLMode) {
   if (InitVal) {
     assert(InitVal->getType() == Ty &&
            "Initializer should be the same type as the GlobalVariable!");
@@ -100,13 +145,13 @@ GlobalVariable::GlobalVariable(Type *Ty, bool constant, LinkageTypes Link,
 GlobalVariable::GlobalVariable(Module &M, Type *Ty, bool constant,
                                LinkageTypes Link, Constant *InitVal,
                                const Twine &Name,
-                               GlobalVariable *Before, bool ThreadLocal,
+                               GlobalVariable *Before, ThreadLocalMode TLMode,
                                unsigned AddressSpace)
   : GlobalValue(PointerType::get(Ty, AddressSpace), 
                 Value::GlobalVariableVal,
                 OperandTraits<GlobalVariable>::op_begin(this),
                 InitVal != 0, Link, Name),
-    isConstantGlobal(constant), isThreadLocalSymbol(ThreadLocal) {
+    isConstantGlobal(constant), threadLocalMode(TLMode) {
   if (InitVal) {
     assert(InitVal->getType() == Ty &&
            "Initializer should be the same type as the GlobalVariable!");

lib/VMCore/IRBuilder.cpp

@@ -28,7 +28,7 @@ Value *IRBuilderBase::CreateGlobalString(StringRef Str, const Twine &Name) {
   Module &M = *BB->getParent()->getParent();
   GlobalVariable *GV = new GlobalVariable(M, StrConstant->getType(),
                                           true, GlobalValue::PrivateLinkage,
-                                          StrConstant, "", 0, false);
+                                          StrConstant);
   GV->setName(Name);
   GV->setUnnamedAddr(true);
   return GV;

test/Assembler/tls-models.ll

@@ -0,0 +1,11 @@
+; RUN: llvm-as < %s | llvm-dis | llvm-as | llvm-dis | FileCheck %s
+
+; CHECK: @a = thread_local global i32 0
+; CHECK: @b = thread_local(localdynamic) global i32 0
+; CHECK: @c = thread_local(initialexec) global i32 0
+; CHECK: @d = thread_local(localexec) global i32 0
+
+@a = thread_local global i32 0
+@b = thread_local(localdynamic) global i32 0
+@c = thread_local(initialexec) global i32 0
+@d = thread_local(localexec) global i32 0

test/CodeGen/ARM/tls-models.ll

@@ -0,0 +1,117 @@
+; RUN: llc -march=arm -mtriple=arm-linux-gnueabi < %s | FileCheck -check-prefix=CHECK-NONPIC %s
+; RUN: llc -march=arm -mtriple=arm-linux-gnueabi -relocation-model=pic < %s | FileCheck -check-prefix=CHECK-PIC %s
+
+
+@external_gd = external thread_local global i32
+@internal_gd = internal thread_local global i32 42
+
+@external_ld = external thread_local(localdynamic) global i32
+@internal_ld = internal thread_local(localdynamic) global i32 42
+
+@external_ie = external thread_local(initialexec) global i32
+@internal_ie = internal thread_local(initialexec) global i32 42
+
+@external_le = external thread_local(localexec) global i32
+@internal_le = internal thread_local(localexec) global i32 42
+
+; ----- no model specified -----
+
+define i32* @f1() {
+entry:
+  ret i32* @external_gd
+
+  ; Non-PIC code can use initial-exec, PIC code has to use general dynamic.
+  ; CHECK-NONPIC:   f1:
+  ; CHECK-NONPIC:   external_gd(gottpoff)
+  ; CHECK-PIC:      f1:
+  ; CHECK-PIC:      external_gd(tlsgd)
+}
+
+define i32* @f2() {
+entry:
+  ret i32* @internal_gd
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic,
+  ; but that is not implemented, so falls back to general dynamic.
+  ; CHECK-NONPIC:   f2:
+  ; CHECK-NONPIC:   internal_gd(tpoff)
+  ; CHECK-PIC:      f2:
+  ; CHECK-PIC:      internal_gd(tlsgd)
+}
+
+
+; ----- localdynamic specified -----
+
+define i32* @f3() {
+entry:
+  ret i32* @external_ld
+
+  ; Non-PIC code can use initial exec, PIC should use local dynamic,
+  ; but that is not implemented, so falls back to general dynamic.
+  ; CHECK-NONPIC:   f3:
+  ; CHECK-NONPIC:   external_ld(gottpoff)
+  ; CHECK-PIC:      f3:
+  ; CHECK-PIC:      external_ld(tlsgd)
+}
+
+define i32* @f4() {
+entry:
+  ret i32* @internal_ld
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic,
+  ; but that is not implemented, so it falls back to general dynamic.
+  ; CHECK-NONPIC:   f4:
+  ; CHECK-NONPIC:   internal_ld(tpoff)
+  ; CHECK-PIC:      f4:
+  ; CHECK-PIC:      internal_ld(tlsgd)
+}
+
+
+; ----- initialexec specified -----
+
+define i32* @f5() {
+entry:
+  ret i32* @external_ie
+
+  ; Non-PIC and PIC code will use initial exec as specified.
+  ; CHECK-NONPIC:   f5:
+  ; CHECK-NONPIC:   external_ie(gottpoff)
+  ; CHECK-PIC:      f5:
+  ; CHECK-PIC:      external_ie(gottpoff)
+}
+
+define i32* @f6() {
+entry:
+  ret i32* @internal_ie
+
+  ; Non-PIC code can use local exec, PIC code use initial exec as specified.
+  ; CHECK-NONPIC:   f6:
+  ; CHECK-NONPIC:   internal_ie(tpoff)
+  ; CHECK-PIC:      f6:
+  ; CHECK-PIC:      internal_ie(gottpoff)
+}
+
+
+; ----- localexec specified -----
+
+define i32* @f7() {
+entry:
+  ret i32* @external_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; CHECK-NONPIC:   f7:
+  ; CHECK-NONPIC:   external_le(tpoff)
+  ; CHECK-PIC:      f7:
+  ; CHECK-PIC:      external_le(tpoff)
+}
+
+define i32* @f8() {
+entry:
+  ret i32* @internal_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; CHECK-NONPIC:   f8:
+  ; CHECK-NONPIC:   internal_le(tpoff)
+  ; CHECK-PIC:      f8:
+  ; CHECK-PIC:      internal_le(tpoff)
+}

test/CodeGen/Mips/tls-models.ll

@@ -0,0 +1,113 @@
+; RUN: llc -march=mipsel < %s | FileCheck -check-prefix=CHECK-PIC %s
+; RUN: llc -march=mipsel -relocation-model=static < %s | FileCheck -check-prefix=CHECK-NONPIC %s
+
+@external_gd = external thread_local global i32
+@internal_gd = internal thread_local global i32 42
+
+@external_ld = external thread_local(localdynamic) global i32
+@internal_ld = internal thread_local(localdynamic) global i32 42
+
+@external_ie = external thread_local(initialexec) global i32
+@internal_ie = internal thread_local(initialexec) global i32 42
+
+@external_le = external thread_local(localexec) global i32
+@internal_le = internal thread_local(localexec) global i32 42
+
+; ----- no model specified -----
+
+define i32* @f1() {
+entry:
+  ret i32* @external_gd
+
+  ; Non-PIC code can use initial-exec, PIC code has to use general dynamic.
+  ; CHECK-NONPIC:   f1:
+  ; CHECK-NONPIC:   %gottprel
+  ; CHECK-PIC:      f1:
+  ; CHECK-PIC:      %tlsgd
+}
+
+define i32* @f2() {
+entry:
+  ret i32* @internal_gd
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic.
+  ; CHECK-NONPIC:   f2:
+  ; CHECK-NONPIC:   %tprel_hi
+  ; CHECK-PIC:      f2:
+  ; CHECK-PIC:      %tlsldm
+}
+
+
+; ----- localdynamic specified -----
+
+define i32* @f3() {
+entry:
+  ret i32* @external_ld
+
+  ; Non-PIC code can use initial exec, PIC should use local dynamic.
+  ; CHECK-NONPIC:   f3:
+  ; CHECK-NONPIC:   %gottprel
+  ; CHECK-PIC:      f3:
+  ; CHECK-PIC:      %tlsldm
+}
+
+define i32* @f4() {
+entry:
+  ret i32* @internal_ld
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic.
+  ; CHECK-NONPIC:   f4:
+  ; CHECK-NONPIC:   %tprel_hi
+  ; CHECK-PIC:      f4:
+  ; CHECK-PIC:      %tlsldm
+}
+
+
+; ----- initialexec specified -----
+
+define i32* @f5() {
+entry:
+  ret i32* @external_ie
+
+  ; Non-PIC and PIC code will use initial exec as specified.
+  ; CHECK-NONPIC:   f5:
+  ; CHECK-NONPIC:   %gottprel
+  ; CHECK-PIC:      f5:
+  ; CHECK-PIC:      %gottprel
+}
+
+define i32* @f6() {
+entry:
+  ret i32* @internal_ie
+
+  ; Non-PIC code can use local exec, PIC code use initial exec as specified.
+  ; CHECK-NONPIC:   f6:
+  ; CHECK-NONPIC:   %tprel_hi
+  ; CHECK-PIC:      f6:
+  ; CHECK-PIC:      %gottprel
+}
+
+
+; ----- localexec specified -----
+
+define i32* @f7() {
+entry:
+  ret i32* @external_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; CHECK-NONPIC:   f7:
+  ; CHECK-NONPIC:   %tprel_hi
+  ; CHECK-PIC:      f7:
+  ; CHECK-PIC:      %tprel_hi
+}
+
+define i32* @f8() {
+entry:
+  ret i32* @internal_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; CHECK-NONPIC:   f8:
+  ; CHECK-NONPIC:   %tprel_hi
+  ; CHECK-PIC:      f8:
+  ; CHECK-PIC:      %tprel_hi
+}

test/CodeGen/X86/tls-models.ll

@@ -0,0 +1,166 @@
+; RUN: llc < %s -march=x86-64 -mtriple=x86_64-linux-gnu | FileCheck -check-prefix=X64 %s
+; RUN: llc < %s -march=x86-64 -mtriple=x86_64-linux-gnu -relocation-model=pic | FileCheck -check-prefix=X64_PIC %s
+; RUN: llc < %s -march=x86 -mtriple=i386-linux-gnu | FileCheck -check-prefix=X32 %s
+; RUN: llc < %s -march=x86 -mtriple=i386-linux-gnu -relocation-model=pic | FileCheck -check-prefix=X32_PIC %s
+
+; Darwin always uses the same model.
+; RUN: llc < %s -march=x86-64 -mtriple=x86_64-apple-darwin | FileCheck -check-prefix=DARWIN %s
+
+@external_gd = external thread_local global i32
+@internal_gd = internal thread_local global i32 42
+
+@external_ld = external thread_local(localdynamic) global i32
+@internal_ld = internal thread_local(localdynamic) global i32 42
+
+@external_ie = external thread_local(initialexec) global i32
+@internal_ie = internal thread_local(initialexec) global i32 42
+
+@external_le = external thread_local(localexec) global i32
+@internal_le = internal thread_local(localexec) global i32 42
+
+; ----- no model specified -----
+
+define i32* @f1() {
+entry:
+  ret i32* @external_gd
+
+  ; Non-PIC code can use initial-exec, PIC code has to use general dynamic.
+  ; X64:     f1:
+  ; X64:     external_gd@GOTTPOFF
+  ; X32:     f1:
+  ; X32:     external_gd@INDNTPOFF
+  ; X64_PIC: f1:
+  ; X64_PIC: external_gd@TLSGD
+  ; X32_PIC: f1:
+  ; X32_PIC: external_gd@TLSGD
+  ; DARWIN:  f1:
+  ; DARWIN:  _external_gd@TLVP
+}
+
+define i32* @f2() {
+entry:
+  ret i32* @internal_gd
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic.
+  ; X64:     f2:
+  ; X64:     internal_gd@TPOFF
+  ; X32:     f2:
+  ; X32:     internal_gd@NTPOFF
+  ; X64_PIC: f2:
+  ; X64_PIC: internal_gd@TLSLD
+  ; X32_PIC: f2:
+  ; X32_PIC: internal_gd@TLSLDM
+  ; DARWIN:  f2:
+  ; DARWIN:  _internal_gd@TLVP
+}
+
+
+; ----- localdynamic specified -----
+
+define i32* @f3() {
+entry:
+  ret i32* @external_ld
+
+  ; Non-PIC code can use initial exec, PIC code use local dynamic as specified.
+  ; X64:     f3:
+  ; X64:     external_ld@GOTTPOFF
+  ; X32:     f3:
+  ; X32:     external_ld@INDNTPOFF
+  ; X64_PIC: f3:
+  ; X64_PIC: external_ld@TLSLD
+  ; X32_PIC: f3:
+  ; X32_PIC: external_ld@TLSLDM
+  ; DARWIN:  f3:
+  ; DARWIN:  _external_ld@TLVP
+}
+
+define i32* @f4() {
+entry:
+  ret i32* @internal_ld
+
+  ; Non-PIC code can use local exec, PIC code can use local dynamic.
+  ; X64:     f4:
+  ; X64:     internal_ld@TPOFF
+  ; X32:     f4:
+  ; X32:     internal_ld@NTPOFF
+  ; X64_PIC: f4:
+  ; X64_PIC: internal_ld@TLSLD
+  ; X32_PIC: f4:
+  ; X32_PIC: internal_ld@TLSLDM
+  ; DARWIN:  f4:
+  ; DARWIN:  _internal_ld@TLVP
+}
+
+
+; ----- initialexec specified -----
+
+define i32* @f5() {
+entry:
+  ret i32* @external_ie
+
+  ; Non-PIC and PIC code will use initial exec as specified.
+  ; X64:     f5:
+  ; X64:     external_ie@GOTTPOFF
+  ; X32:     f5:
+  ; X32:     external_ie@INDNTPOFF
+  ; X64_PIC: f5:
+  ; X64_PIC: external_ie@GOTTPOFF
+  ; X32_PIC: f5:
+  ; X32_PIC: external_ie@GOTNTPOFF
+  ; DARWIN:  f5:
+  ; DARWIN:  _external_ie@TLVP
+}
+
+define i32* @f6() {
+entry:
+  ret i32* @internal_ie
+
+  ; Non-PIC code can use local exec, PIC code use initial exec as specified.
+  ; X64:     f6:
+  ; X64:     internal_ie@TPOFF
+  ; X32:     f6:
+  ; X32:     internal_ie@NTPOFF
+  ; X64_PIC: f6:
+  ; X64_PIC: internal_ie@GOTTPOFF
+  ; X32_PIC: f6:
+  ; X32_PIC: internal_ie@GOTNTPOFF
+  ; DARWIN:  f6:
+  ; DARWIN:  _internal_ie@TLVP
+}
+
+
+; ----- localexec specified -----
+
+define i32* @f7() {
+entry:
+  ret i32* @external_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; X64:     f7:
+  ; X64:     external_le@TPOFF
+  ; X32:     f7:
+  ; X32:     external_le@NTPOFF
+  ; X64_PIC: f7:
+  ; X64_PIC: external_le@TPOFF
+  ; X32_PIC: f7:
+  ; X32_PIC: external_le@NTPOFF
+  ; DARWIN:  f7:
+  ; DARWIN:  _external_le@TLVP
+}
+
+define i32* @f8() {
+entry:
+  ret i32* @internal_le
+
+  ; Non-PIC and PIC code will use local exec as specified.
+  ; X64:     f8:
+  ; X64:     internal_le@TPOFF
+  ; X32:     f8:
+  ; X32:     internal_le@NTPOFF
+  ; X64_PIC: f8:
+  ; X64_PIC: internal_le@TPOFF
+  ; X32_PIC: f8:
+  ; X32_PIC: internal_le@NTPOFF
+  ; DARWIN:  f8:
+  ; DARWIN:  _internal_le@TLVP
+}