[CUDA] Allow parsing of host and device code simultaneously.

 * adds -aux-triple option to specify target triple
 * propagates aux target info to AST context and Preprocessor
 * pulls in target specific preprocessor macros.
 * pulls in target-specific builtins from aux target.
 * sets appropriate host or device attribute on builtins.

Differential Revision: http://reviews.llvm.org/D12917

llvm-svn: 248299

clang/include/clang/AST/ASTContext.h

@@ -437,6 +437,7 @@ class ASTContext : public RefCountedBase<ASTContext> {
   friend class CXXRecordDecl;
 
   const TargetInfo *Target;
+  const TargetInfo *AuxTarget;
   clang::PrintingPolicy PrintingPolicy;
 
 public:
@@ -523,7 +524,8 @@ class ASTContext : public RefCountedBase<ASTContext> {
   }
 
   const TargetInfo &getTargetInfo() const { return *Target; }
-
+  const TargetInfo *getAuxTargetInfo() const { return AuxTarget; }
+
   /// getIntTypeForBitwidth -
   /// sets integer QualTy according to specified details:
   /// bitwidth, signed/unsigned.
@@ -2415,9 +2417,10 @@ class ASTContext : public RefCountedBase<ASTContext> {
   /// This routine may only be invoked once for a given ASTContext object.
   /// It is normally invoked after ASTContext construction.
   ///
-  /// \param Target The target 
-  void InitBuiltinTypes(const TargetInfo &Target);
-
+  /// \param Target The target
+  void InitBuiltinTypes(const TargetInfo &Target,
+                        const TargetInfo *AuxTarget = nullptr);
+
 private:
   void InitBuiltinType(CanQualType &R, BuiltinType::Kind K);
 

clang/include/clang/Basic/Builtins.h

@@ -56,15 +56,23 @@ struct Info {
 
 /// \brief Holds information about both target-independent and
 /// target-specific builtins, allowing easy queries by clients.
+///
+/// Builtins from an optional auxiliary target are stored in
+/// AuxTSRecords. Their IDs are shifted up by NumTSRecords and need to
+/// be translated back with getAuxBuiltinID() before use.
 class Context {
   const Info *TSRecords;
+  const Info *AuxTSRecords;
   unsigned NumTSRecords;
+  unsigned NumAuxTSRecords;
+
 public:
   Context();
 
   /// \brief Perform target-specific initialization
-  void initializeTarget(const TargetInfo &Target);
-
+  /// \param AuxTarget Target info to incorporate builtins from. May be nullptr.
+  void InitializeTarget(const TargetInfo &Target, const TargetInfo *AuxTarget);
+
   /// \brief Mark the identifiers for all the builtins with their
   /// appropriate builtin ID # and mark any non-portable builtin identifiers as
   /// such.
@@ -176,6 +184,15 @@ class Context {
     return getRecord(ID).Features;
   }
 
+  /// \brief Return true if builtin ID belongs to AuxTarget.
+  bool isAuxBuiltinID(unsigned ID) const {
+    return ID >= (Builtin::FirstTSBuiltin + NumTSRecords);
+  }
+
+  /// Return real buitin ID (i.e. ID it would have furing compilation
+  /// for AuxTarget).
+  unsigned getAuxBuiltinID(unsigned ID) const { return ID - NumTSRecords; }
+
 private:
   const Info &getRecord(unsigned ID) const;
 

clang/include/clang/Driver/CC1Options.td

@@ -325,6 +325,8 @@ def cc1as : Flag<["-"], "cc1as">;
 def ast_merge : Separate<["-"], "ast-merge">,
   MetaVarName<"<ast file>">,
   HelpText<"Merge the given AST file into the translation unit being compiled.">;
+def aux_triple : Separate<["-"], "aux-triple">,
+  HelpText<"Auxiliary target triple.">;
 def code_completion_at : Separate<["-"], "code-completion-at">,
   MetaVarName<"<file>:<line>:<column>">,
   HelpText<"Dump code-completion information at a location">;

clang/include/clang/Frontend/CompilerInstance.h

@@ -78,6 +78,9 @@ class CompilerInstance : public ModuleLoader {
   /// The target being compiled for.
   IntrusiveRefCntPtr<TargetInfo> Target;
 
+  /// Auxiliary Target info.
+  IntrusiveRefCntPtr<TargetInfo> AuxTarget;
+
   /// The virtual file system.
   IntrusiveRefCntPtr<vfs::FileSystem> VirtualFileSystem;
 
@@ -351,6 +354,15 @@ class CompilerInstance : public ModuleLoader {
   /// Replace the current Target.
   void setTarget(TargetInfo *Value);
 
+  /// }
+  /// @name AuxTarget Info
+  /// {
+
+  TargetInfo *getAuxTarget() const { return AuxTarget.get(); }
+
+  /// Replace the current AuxTarget.
+  void setAuxTarget(TargetInfo *Value);
+
   /// }
   /// @name Virtual File System
   /// {

clang/include/clang/Frontend/FrontendOptions.h

@@ -256,7 +256,10 @@ class FrontendOptions {
   /// \brief File name of the file that will provide record layouts
   /// (in the format produced by -fdump-record-layouts).
   std::string OverrideRecordLayoutsFile;
-
+
+  /// \brief Auxiliary triple for CUDA compilation.
+  std::string AuxTriple;
+
 public:
   FrontendOptions() :
     DisableFree(false), RelocatablePCH(false), ShowHelp(false),

clang/include/clang/Lex/Preprocessor.h

@@ -98,6 +98,7 @@ class Preprocessor : public RefCountedBase<Preprocessor> {
   DiagnosticsEngine        *Diags;
   LangOptions       &LangOpts;
   const TargetInfo  *Target;
+  const TargetInfo  *AuxTarget;
   FileManager       &FileMgr;
   SourceManager     &SourceMgr;
   std::unique_ptr<ScratchBuffer> ScratchBuf;
@@ -656,7 +657,10 @@ class Preprocessor : public RefCountedBase<Preprocessor> {
   ///
   /// \param Target is owned by the caller and must remain valid for the
   /// lifetime of the preprocessor.
-  void Initialize(const TargetInfo &Target);
+  /// \param AuxTarget is owned by the caller and must remain valid for
+  /// the lifetime of the preprocessor.
+  void Initialize(const TargetInfo &Target,
+                  const TargetInfo *AuxTarget = nullptr);
 
   /// \brief Initialize the preprocessor to parse a model file
   ///
@@ -678,6 +682,7 @@ class Preprocessor : public RefCountedBase<Preprocessor> {
 
   const LangOptions &getLangOpts() const { return LangOpts; }
   const TargetInfo &getTargetInfo() const { return *Target; }
+  const TargetInfo *getAuxTargetInfo() const { return AuxTarget; }
   FileManager &getFileManager() const { return FileMgr; }
   SourceManager &getSourceManager() const { return SourceMgr; }
   HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; }

clang/lib/AST/ASTContext.cpp

@@ -743,10 +743,10 @@ ASTContext::ASTContext(LangOptions &LOpts, SourceManager &SM,
       FirstLocalImport(), LastLocalImport(), ExternCContext(nullptr),
       SourceMgr(SM), LangOpts(LOpts),
       SanitizerBL(new SanitizerBlacklist(LangOpts.SanitizerBlacklistFiles, SM)),
-      AddrSpaceMap(nullptr), Target(nullptr), PrintingPolicy(LOpts),
-      Idents(idents), Selectors(sels), BuiltinInfo(builtins),
-      DeclarationNames(*this), ExternalSource(nullptr), Listener(nullptr),
-      Comments(SM), CommentsLoaded(false),
+      AddrSpaceMap(nullptr), Target(nullptr), AuxTarget(nullptr),
+      PrintingPolicy(LOpts), Idents(idents), Selectors(sels),
+      BuiltinInfo(builtins), DeclarationNames(*this), ExternalSource(nullptr),
+      Listener(nullptr), Comments(SM), CommentsLoaded(false),
       CommentCommandTraits(BumpAlloc, LOpts.CommentOpts), LastSDM(nullptr, 0) {
   TUDecl = TranslationUnitDecl::Create(*this);
 }
@@ -956,13 +956,15 @@ void ASTContext::InitBuiltinType(CanQualType &R, BuiltinType::Kind K) {
   Types.push_back(Ty);
 }
 
-void ASTContext::InitBuiltinTypes(const TargetInfo &Target) {
+void ASTContext::InitBuiltinTypes(const TargetInfo &Target,
+                                  const TargetInfo *AuxTarget) {
   assert((!this->Target || this->Target == &Target) &&
          "Incorrect target reinitialization");
   assert(VoidTy.isNull() && "Context reinitialized?");
 
   this->Target = &Target;
-
+  this->AuxTarget = AuxTarget;
+
   ABI.reset(createCXXABI(Target));
   AddrSpaceMap = getAddressSpaceMap(Target, LangOpts);
   AddrSpaceMapMangling = isAddrSpaceMapManglingEnabled(Target, LangOpts);

clang/lib/Basic/Builtins.cpp

@@ -32,19 +32,27 @@ static const Builtin::Info BuiltinInfo[] = {
 const Builtin::Info &Builtin::Context::getRecord(unsigned ID) const {
   if (ID < Builtin::FirstTSBuiltin)
     return BuiltinInfo[ID];
-  assert(ID - Builtin::FirstTSBuiltin < NumTSRecords && "Invalid builtin ID!");
+  assert(ID - Builtin::FirstTSBuiltin < (NumTSRecords + NumAuxTSRecords) &&
+         "Invalid builtin ID!");
+  if (isAuxBuiltinID(ID))
+    return AuxTSRecords[getAuxBuiltinID(ID) - Builtin::FirstTSBuiltin];
   return TSRecords[ID - Builtin::FirstTSBuiltin];
 }
 
 Builtin::Context::Context() {
   // Get the target specific builtins from the target.
   TSRecords = nullptr;
+  AuxTSRecords = nullptr;
   NumTSRecords = 0;
+  NumAuxTSRecords = 0;
 }
 
-void Builtin::Context::initializeTarget(const TargetInfo &Target) {
+void Builtin::Context::InitializeTarget(const TargetInfo &Target,
+                                        const TargetInfo *AuxTarget) {
   assert(NumTSRecords == 0 && "Already initialized target?");
   Target.getTargetBuiltins(TSRecords, NumTSRecords);
+  if (AuxTarget)
+    AuxTarget->getTargetBuiltins(AuxTSRecords, NumAuxTSRecords);
 }
 
 bool Builtin::Context::builtinIsSupported(const Builtin::Info &BuiltinInfo,
@@ -76,7 +84,12 @@ void Builtin::Context::initializeBuiltins(IdentifierTable &Table,
   // Step #2: Register target-specific builtins.
   for (unsigned i = 0, e = NumTSRecords; i != e; ++i)
     if (builtinIsSupported(TSRecords[i], LangOpts))
-      Table.get(TSRecords[i].Name).setBuiltinID(i+Builtin::FirstTSBuiltin);
+      Table.get(TSRecords[i].Name).setBuiltinID(i + Builtin::FirstTSBuiltin);
+
+  // Step #3: Register target-specific builtins for AuxTarget.
+  for (unsigned i = 0, e = NumAuxTSRecords; i != e; ++i)
+    Table.get(AuxTSRecords[i].Name)
+        .setBuiltinID(i + Builtin::FirstTSBuiltin + NumTSRecords);
 }
 
 void Builtin::Context::forgetBuiltin(unsigned ID, IdentifierTable &Table) {

clang/lib/CodeGen/CGBuiltin.cpp

@@ -1868,40 +1868,54 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
   return GetUndefRValue(E->getType());
 }
 
-Value *CodeGenFunction::EmitTargetBuiltinExpr(unsigned BuiltinID,
-                                              const CallExpr *E) {
-  switch (getTarget().getTriple().getArch()) {
+static Value *EmitTargetArchBuiltinExpr(CodeGenFunction *CGF,
+                                        unsigned BuiltinID, const CallExpr *E,
+                                        llvm::Triple::ArchType Arch) {
+  switch (Arch) {
   case llvm::Triple::arm:
   case llvm::Triple::armeb:
   case llvm::Triple::thumb:
   case llvm::Triple::thumbeb:
-    return EmitARMBuiltinExpr(BuiltinID, E);
+    return CGF->EmitARMBuiltinExpr(BuiltinID, E);
   case llvm::Triple::aarch64:
   case llvm::Triple::aarch64_be:
-    return EmitAArch64BuiltinExpr(BuiltinID, E);
+    return CGF->EmitAArch64BuiltinExpr(BuiltinID, E);
   case llvm::Triple::x86:
   case llvm::Triple::x86_64:
-    return EmitX86BuiltinExpr(BuiltinID, E);
+    return CGF->EmitX86BuiltinExpr(BuiltinID, E);
   case llvm::Triple::ppc:
   case llvm::Triple::ppc64:
   case llvm::Triple::ppc64le:
-    return EmitPPCBuiltinExpr(BuiltinID, E);
+    return CGF->EmitPPCBuiltinExpr(BuiltinID, E);
   case llvm::Triple::r600:
   case llvm::Triple::amdgcn:
-    return EmitAMDGPUBuiltinExpr(BuiltinID, E);
+    return CGF->EmitAMDGPUBuiltinExpr(BuiltinID, E);
   case llvm::Triple::systemz:
-    return EmitSystemZBuiltinExpr(BuiltinID, E);
+    return CGF->EmitSystemZBuiltinExpr(BuiltinID, E);
   case llvm::Triple::nvptx:
   case llvm::Triple::nvptx64:
-    return EmitNVPTXBuiltinExpr(BuiltinID, E);
+    return CGF->EmitNVPTXBuiltinExpr(BuiltinID, E);
   case llvm::Triple::wasm32:
   case llvm::Triple::wasm64:
-    return EmitWebAssemblyBuiltinExpr(BuiltinID, E);
+    return CGF->EmitWebAssemblyBuiltinExpr(BuiltinID, E);
   default:
     return nullptr;
   }
 }
 
+Value *CodeGenFunction::EmitTargetBuiltinExpr(unsigned BuiltinID,
+                                              const CallExpr *E) {
+  if (getContext().BuiltinInfo.isAuxBuiltinID(BuiltinID)) {
+    assert(getContext().getAuxTargetInfo() && "Missing aux target info");
+    return EmitTargetArchBuiltinExpr(
+        this, getContext().BuiltinInfo.getAuxBuiltinID(BuiltinID), E,
+        getContext().getAuxTargetInfo()->getTriple().getArch());
+  }
+
+  return EmitTargetArchBuiltinExpr(this, BuiltinID, E,
+                                   getTarget().getTriple().getArch());
+}
+
 static llvm::VectorType *GetNeonType(CodeGenFunction *CGF,
                                      NeonTypeFlags TypeFlags,
                                      bool V1Ty=false) {

clang/lib/Frontend/CompilerInstance.cpp

@@ -78,9 +78,8 @@ void CompilerInstance::setDiagnostics(DiagnosticsEngine *Value) {
   Diagnostics = Value;
 }
 
-void CompilerInstance::setTarget(TargetInfo *Value) {
-  Target = Value;
-}
+void CompilerInstance::setTarget(TargetInfo *Value) { Target = Value; }
+void CompilerInstance::setAuxTarget(TargetInfo *Value) { AuxTarget = Value; }
 
 void CompilerInstance::setFileManager(FileManager *Value) {
   FileMgr = Value;
@@ -312,7 +311,7 @@ void CompilerInstance::createPreprocessor(TranslationUnitKind TUKind) {
   PP = new Preprocessor(&getPreprocessorOpts(), getDiagnostics(), getLangOpts(),
                         getSourceManager(), *HeaderInfo, *this, PTHMgr,
                         /*OwnsHeaderSearch=*/true, TUKind);
-  PP->Initialize(getTarget());
+  PP->Initialize(getTarget(), getAuxTarget());
 
   // Note that this is different then passing PTHMgr to Preprocessor's ctor.
   // That argument is used as the IdentifierInfoLookup argument to
@@ -396,7 +395,7 @@ void CompilerInstance::createASTContext() {
   auto *Context = new ASTContext(getLangOpts(), PP.getSourceManager(),
                                  PP.getIdentifierTable(), PP.getSelectorTable(),
                                  PP.getBuiltinInfo());
-  Context->InitBuiltinTypes(getTarget());
+  Context->InitBuiltinTypes(getTarget(), getAuxTarget());
   setASTContext(Context);
 }
 
@@ -800,6 +799,13 @@ bool CompilerInstance::ExecuteAction(FrontendAction &Act) {
   if (!hasTarget())
     return false;
 
+  // Create TargetInfo for the other side of CUDA compilation.
+  if (getLangOpts().CUDA && !getFrontendOpts().AuxTriple.empty()) {
+    std::shared_ptr<TargetOptions> TO(new TargetOptions);
+    TO->Triple = getFrontendOpts().AuxTriple;
+    setAuxTarget(TargetInfo::CreateTargetInfo(getDiagnostics(), TO));
+  }
+
   // Inform the target of the language options.
   //
   // FIXME: We shouldn't need to do this, the target should be immutable once

clang/lib/Frontend/CompilerInvocation.cpp

@@ -975,6 +975,9 @@ static InputKind ParseFrontendArgs(FrontendOptions &Opts, ArgList &Args,
 
   Opts.OverrideRecordLayoutsFile
     = Args.getLastArgValue(OPT_foverride_record_layout_EQ);
+  Opts.AuxTriple =
+      llvm::Triple::normalize(Args.getLastArgValue(OPT_aux_triple));
+
   if (const Arg *A = Args.getLastArg(OPT_arcmt_check,
                                      OPT_arcmt_modify,
                                      OPT_arcmt_migrate)) {

clang/lib/Frontend/InitPreprocessor.cpp

@@ -918,6 +918,10 @@ void clang::InitializePreprocessor(
 
   // Install things like __POWERPC__, __GNUC__, etc into the macro table.
   if (InitOpts.UsePredefines) {
+    if (LangOpts.CUDA && PP.getAuxTargetInfo())
+      InitializePredefinedMacros(*PP.getAuxTargetInfo(), LangOpts, FEOpts,
+                                 Builder);
+
     InitializePredefinedMacros(PP.getTargetInfo(), LangOpts, FEOpts, Builder);
 
     // Install definitions to make Objective-C++ ARC work well with various