Reland "[clang-repl] Implement partial translation units and error recovery."

Original commit message:

[clang-repl] Implement partial translation units and error recovery.

https://reviews.llvm.org/D96033 contained a discussion regarding efficient
modeling of error recovery. @rjmccall has outlined the key ideas:

Conceptually, we can split the translation unit into a sequence of partial
translation units (PTUs). Every declaration will be associated with a unique PTU
that owns it.

The first key insight here is that the owning PTU isn't always the "active"
(most recent) PTU, and it isn't always the PTU that the declaration
"comes from". A new declaration (that isn't a redeclaration or specialization of
anything) does belong to the active PTU. A template specialization, however,
belongs to the most recent PTU of all the declarations in its signature - mostly
that means that it can be pulled into a more recent PTU by its template
arguments.

The second key insight is that processing a PTU might extend an earlier PTU.
Rolling back the later PTU shouldn't throw that extension away. For example, if
the second PTU defines a template, and the third PTU requires that template to
be instantiated at float, that template specialization is still part of the
second PTU. Similarly, if the fifth PTU uses an inline function belonging to the
fourth, that definition still belongs to the fourth. When we go to emit code in
a new PTU, we map each declaration we have to emit back to its owning PTU and
emit it in a new module for just the extensions to that PTU. We keep track of
all the modules we've emitted for a PTU so that we can unload them all if we
decide to roll it back.

Most declarations/definitions will only refer to entities from the same or
earlier PTUs. However, it is possible (primarily by defining a
previously-declared entity, but also through templates or ADL) for an entity
that belongs to one PTU to refer to something from a later PTU. We will have to
keep track of this and prevent unwinding to later PTU when we recognize it.
Fortunately, this should be very rare; and crucially, we don't have to do the
bookkeeping for this if we've only got one PTU, e.g. in normal compilation.
Otherwise, PTUs after the first just need to record enough metadata to be able
to revert any changes they've made to declarations belonging to earlier PTUs,
e.g. to redeclaration chains or template specialization lists.

It should even eventually be possible for PTUs to provide their own slab
allocators which can be thrown away as part of rolling back the PTU. We can
maintain a notion of the active allocator and allocate things like Stmt/Expr
nodes in it, temporarily changing it to the appropriate PTU whenever we go to do
something like instantiate a function template. More care will be required when
allocating declarations and types, though.

We would want the PTU to be efficiently recoverable from a Decl; I'm not sure
how best to do that. An easy option that would cover most declarations would be
to make multiple TranslationUnitDecls and parent the declarations appropriately,
but I don't think that's good enough for things like member function templates,
since an instantiation of that would still be parented by its original class.
Maybe we can work this into the DC chain somehow, like how lexical DCs are.

We add a different kind of translation unit `TU_Incremental` which is a
complete translation unit that we might nonetheless incrementally extend later.
Because it is complete (and we might want to generate code for it), we do
perform template instantiation, but because it might be extended later, we don't
warn if it declares or uses undefined internal-linkage symbols.

This patch teaches clang-repl how to recover from errors by disconnecting the
most recent PTU and update the primary PTU lookup tables. For instance:

```./clang-repl
clang-repl> int i = 12; error;
In file included from <<< inputs >>>:1:
input_line_0:1:13: error: C++ requires a type specifier for all declarations
int i = 12; error;
            ^
error: Parsing failed.
clang-repl> int i = 13; extern "C" int printf(const char*,...);
clang-repl> auto r1 = printf("i=%d\n", i);
i=13
clang-repl> quit
```

Differential revision: https://reviews.llvm.org/D104918

Author: vgvassilev

clang/include/clang/AST/ASTContext.h

@@ -459,6 +459,7 @@ class ASTContext : public RefCountedBase<ASTContext> {
   friend class ASTWriter;
   template <class> friend class serialization::AbstractTypeReader;
   friend class CXXRecordDecl;
+  friend class IncrementalParser;
 
   /// A mapping to contain the template or declaration that
   /// a variable declaration describes or was instantiated from,
@@ -567,7 +568,7 @@ class ASTContext : public RefCountedBase<ASTContext> {
   ImportDecl *FirstLocalImport = nullptr;
   ImportDecl *LastLocalImport = nullptr;
 
-  TranslationUnitDecl *TUDecl;
+  TranslationUnitDecl *TUDecl = nullptr;
   mutable ExternCContextDecl *ExternCContext = nullptr;
   mutable BuiltinTemplateDecl *MakeIntegerSeqDecl = nullptr;
   mutable BuiltinTemplateDecl *TypePackElementDecl = nullptr;
@@ -624,6 +625,7 @@ class ASTContext : public RefCountedBase<ASTContext> {
   IdentifierTable &Idents;
   SelectorTable &Selectors;
   Builtin::Context &BuiltinInfo;
+  const TranslationUnitKind TUKind;
   mutable DeclarationNameTable DeclarationNames;
   IntrusiveRefCntPtr<ExternalASTSource> ExternalSource;
   ASTMutationListener *Listener = nullptr;
@@ -1022,7 +1024,18 @@ class ASTContext : public RefCountedBase<ASTContext> {
   /// Get the initializations to perform when importing a module, if any.
   ArrayRef<Decl*> getModuleInitializers(Module *M);
 
-  TranslationUnitDecl *getTranslationUnitDecl() const { return TUDecl; }
+  TranslationUnitDecl *getTranslationUnitDecl() const {
+    return TUDecl->getMostRecentDecl();
+  }
+  void addTranslationUnitDecl() {
+    assert(!TUDecl || TUKind == TU_Incremental);
+    TranslationUnitDecl *NewTUDecl = TranslationUnitDecl::Create(*this);
+    if (TraversalScope.empty() || TraversalScope.back() == TUDecl)
+      TraversalScope = {NewTUDecl};
+    if (TUDecl)
+      NewTUDecl->setPreviousDecl(TUDecl);
+    TUDecl = NewTUDecl;
+  }
 
   ExternCContextDecl *getExternCContextDecl() const;
   BuiltinTemplateDecl *getMakeIntegerSeqDecl() const;
@@ -1099,7 +1112,8 @@ class ASTContext : public RefCountedBase<ASTContext> {
   llvm::DenseSet<const VarDecl *> CUDADeviceVarODRUsedByHost;
 
   ASTContext(LangOptions &LOpts, SourceManager &SM, IdentifierTable &idents,
-             SelectorTable &sels, Builtin::Context &builtins);
+             SelectorTable &sels, Builtin::Context &builtins,
+             TranslationUnitKind TUKind);
   ASTContext(const ASTContext &) = delete;
   ASTContext &operator=(const ASTContext &) = delete;
   ~ASTContext();

clang/include/clang/AST/Decl.h

@@ -79,7 +79,23 @@ class UnresolvedSetImpl;
 class VarTemplateDecl;
 
 /// The top declaration context.
-class TranslationUnitDecl : public Decl, public DeclContext {
+class TranslationUnitDecl : public Decl,
+                            public DeclContext,
+                            public Redeclarable<TranslationUnitDecl> {
+  using redeclarable_base = Redeclarable<TranslationUnitDecl>;
+
+  TranslationUnitDecl *getNextRedeclarationImpl() override {
+    return getNextRedeclaration();
+  }
+
+  TranslationUnitDecl *getPreviousDeclImpl() override {
+    return getPreviousDecl();
+  }
+
+  TranslationUnitDecl *getMostRecentDeclImpl() override {
+    return getMostRecentDecl();
+  }
+
   ASTContext &Ctx;
 
   /// The (most recently entered) anonymous namespace for this
@@ -91,6 +107,16 @@ class TranslationUnitDecl : public Decl, public DeclContext {
   virtual void anchor();
 
 public:
+  using redecl_range = redeclarable_base::redecl_range;
+  using redecl_iterator = redeclarable_base::redecl_iterator;
+
+  using redeclarable_base::getMostRecentDecl;
+  using redeclarable_base::getPreviousDecl;
+  using redeclarable_base::isFirstDecl;
+  using redeclarable_base::redecls;
+  using redeclarable_base::redecls_begin;
+  using redeclarable_base::redecls_end;
+
   ASTContext &getASTContext() const { return Ctx; }
 
   NamespaceDecl *getAnonymousNamespace() const { return AnonymousNamespace; }

clang/include/clang/AST/Redeclarable.h

@@ -193,6 +193,7 @@ class Redeclarable {
 public:
   friend class ASTDeclReader;
   friend class ASTDeclWriter;
+  friend class IncrementalParser;
 
   Redeclarable(const ASTContext &Ctx)
       : RedeclLink(LatestDeclLink(Ctx)),

clang/include/clang/Basic/LangOptions.h

@@ -697,7 +697,11 @@ enum TranslationUnitKind {
   TU_Prefix,
 
   /// The translation unit is a module.
-  TU_Module
+  TU_Module,
+
+  /// The translation unit is a is a complete translation unit that we might
+  /// incrementally extend later.
+  TU_Incremental
 };
 
 } // namespace clang

clang/include/clang/Interpreter/Interpreter.h

@@ -14,7 +14,7 @@
 #ifndef LLVM_CLANG_INTERPRETER_INTERPRETER_H
 #define LLVM_CLANG_INTERPRETER_INTERPRETER_H
 
-#include "clang/Interpreter/Transaction.h"
+#include "clang/Interpreter/PartialTranslationUnit.h"
 
 #include "llvm/Support/Error.h"
 
@@ -55,14 +55,14 @@ class Interpreter {
   static llvm::Expected<std::unique_ptr<Interpreter>>
   create(std::unique_ptr<CompilerInstance> CI);
   const CompilerInstance *getCompilerInstance() const;
-  llvm::Expected<Transaction &> Parse(llvm::StringRef Code);
-  llvm::Error Execute(Transaction &T);
+  llvm::Expected<PartialTranslationUnit &> Parse(llvm::StringRef Code);
+  llvm::Error Execute(PartialTranslationUnit &T);
   llvm::Error ParseAndExecute(llvm::StringRef Code) {
-    auto ErrOrTransaction = Parse(Code);
-    if (auto Err = ErrOrTransaction.takeError())
-      return Err;
-    if (ErrOrTransaction->TheModule)
-      return Execute(*ErrOrTransaction);
+    auto PTU = Parse(Code);
+    if (!PTU)
+      return PTU.takeError();
+    if (PTU->TheModule)
+      return Execute(*PTU);
     return llvm::Error::success();
   }
 };

clang/include/clang/Interpreter/Transaction.h renamed to clang/include/clang/Interpreter/PartialTranslationUnit.h

@@ -11,29 +11,27 @@
 //
 //===----------------------------------------------------------------------===//
 
-#ifndef LLVM_CLANG_INTERPRETER_TRANSACTION_H
-#define LLVM_CLANG_INTERPRETER_TRANSACTION_H
+#ifndef LLVM_CLANG_INTERPRETER_PARTIALTRANSLATIONUNIT_H
+#define LLVM_CLANG_INTERPRETER_PARTIALTRANSLATIONUNIT_H
 
 #include <memory>
-#include <vector>
 
 namespace llvm {
 class Module;
 }
 
 namespace clang {
 
-class DeclGroupRef;
+class TranslationUnitDecl;
 
 /// The class keeps track of various objects created as part of processing
 /// incremental inputs.
-struct Transaction {
-  /// The decls created for the input.
-  std::vector<clang::DeclGroupRef> Decls;
+struct PartialTranslationUnit {
+  TranslationUnitDecl *TUPart = nullptr;
 
   /// The llvm IR produced for the input.
   std::unique_ptr<llvm::Module> TheModule;
 };
 } // namespace clang
 
-#endif // LLVM_CLANG_INTERPRETER_TRANSACTION_H
+#endif // LLVM_CLANG_INTERPRETER_PARTIALTRANSLATIONUNIT_H

clang/include/clang/Lex/Preprocessor.h

@@ -264,9 +264,11 @@ class Preprocessor {
   /// avoid tearing the Lexer and etc. down).
   bool IncrementalProcessing = false;
 
+public:
   /// The kind of translation unit we are processing.
-  TranslationUnitKind TUKind;
+  const TranslationUnitKind TUKind;
 
+private:
   /// The code-completion handler.
   CodeCompletionHandler *CodeComplete = nullptr;
 

clang/include/clang/Sema/Sema.h

@@ -1377,7 +1377,7 @@ class Sema final {
   /// initializers for tentative definitions in C) once parsing has
   /// completed. Modules and precompiled headers perform different kinds of
   /// checks.
-  TranslationUnitKind TUKind;
+  const TranslationUnitKind TUKind;
 
   llvm::BumpPtrAllocator BumpAlloc;
 

clang/lib/AST/ASTContext.cpp

@@ -966,7 +966,7 @@ static bool isAddrSpaceMapManglingEnabled(const TargetInfo &TI,
 
 ASTContext::ASTContext(LangOptions &LOpts, SourceManager &SM,
                        IdentifierTable &idents, SelectorTable &sels,
-                       Builtin::Context &builtins)
+                       Builtin::Context &builtins, TranslationUnitKind TUKind)
     : ConstantArrayTypes(this_()), FunctionProtoTypes(this_()),
       TemplateSpecializationTypes(this_()),
       DependentTemplateSpecializationTypes(this_()), AutoTypes(this_()),
@@ -978,11 +978,10 @@ ASTContext::ASTContext(LangOptions &LOpts, SourceManager &SM,
                                         LangOpts.XRayAttrListFiles, SM)),
       ProfList(new ProfileList(LangOpts.ProfileListFiles, SM)),
       PrintingPolicy(LOpts), Idents(idents), Selectors(sels),
-      BuiltinInfo(builtins), DeclarationNames(*this), Comments(SM),
-      CommentCommandTraits(BumpAlloc, LOpts.CommentOpts),
+      BuiltinInfo(builtins), TUKind(TUKind), DeclarationNames(*this),
+      Comments(SM), CommentCommandTraits(BumpAlloc, LOpts.CommentOpts),
       CompCategories(this_()), LastSDM(nullptr, 0) {
-  TUDecl = TranslationUnitDecl::Create(*this);
-  TraversalScope = {TUDecl};
+  addTranslationUnitDecl();
 }
 
 ASTContext::~ASTContext() {
@@ -1196,9 +1195,10 @@ ExternCContextDecl *ASTContext::getExternCContextDecl() const {
 BuiltinTemplateDecl *
 ASTContext::buildBuiltinTemplateDecl(BuiltinTemplateKind BTK,
                                      const IdentifierInfo *II) const {
-  auto *BuiltinTemplate = BuiltinTemplateDecl::Create(*this, TUDecl, II, BTK);
+  auto *BuiltinTemplate =
+      BuiltinTemplateDecl::Create(*this, getTranslationUnitDecl(), II, BTK);
   BuiltinTemplate->setImplicit();
-  TUDecl->addDecl(BuiltinTemplate);
+  getTranslationUnitDecl()->addDecl(BuiltinTemplate);
 
   return BuiltinTemplate;
 }
@@ -1485,7 +1485,7 @@ void ASTContext::InitBuiltinTypes(const TargetInfo &Target,
   // MSVC predeclares struct _GUID, and we need it to create MSGuidDecls.
   if (LangOpts.MicrosoftExt || LangOpts.Borland) {
     MSGuidTagDecl = buildImplicitRecord("_GUID");
-    TUDecl->addDecl(MSGuidTagDecl);
+    getTranslationUnitDecl()->addDecl(MSGuidTagDecl);
   }
 }
 
@@ -6622,7 +6622,7 @@ QualType ASTContext::getCFConstantStringType() const {
 QualType ASTContext::getObjCSuperType() const {
   if (ObjCSuperType.isNull()) {
     RecordDecl *ObjCSuperTypeDecl = buildImplicitRecord("objc_super");
-    TUDecl->addDecl(ObjCSuperTypeDecl);
+    getTranslationUnitDecl()->addDecl(ObjCSuperTypeDecl);
     ObjCSuperType = getTagDeclType(ObjCSuperTypeDecl);
   }
   return ObjCSuperType;

clang/lib/AST/Decl.cpp

@@ -102,7 +102,7 @@ bool Decl::isOutOfLine() const {
 
 TranslationUnitDecl::TranslationUnitDecl(ASTContext &ctx)
     : Decl(TranslationUnit, nullptr, SourceLocation()),
-      DeclContext(TranslationUnit), Ctx(ctx) {}
+      DeclContext(TranslationUnit), redeclarable_base(ctx), Ctx(ctx) {}
 
 //===----------------------------------------------------------------------===//
 // NamedDecl Implementation

clang/lib/AST/DeclBase.cpp

@@ -1219,7 +1219,6 @@ bool DeclContext::Encloses(const DeclContext *DC) const {
 
 DeclContext *DeclContext::getPrimaryContext() {
   switch (getDeclKind()) {
-  case Decl::TranslationUnit:
   case Decl::ExternCContext:
   case Decl::LinkageSpec:
   case Decl::Export:
@@ -1231,6 +1230,8 @@ DeclContext *DeclContext::getPrimaryContext() {
     // There is only one DeclContext for these entities.
     return this;
 
+  case Decl::TranslationUnit:
+    return static_cast<TranslationUnitDecl *>(this)->getFirstDecl();
   case Decl::Namespace:
     // The original namespace is our primary context.
     return static_cast<NamespaceDecl *>(this)->getOriginalNamespace();
@@ -1285,21 +1286,25 @@ DeclContext *DeclContext::getPrimaryContext() {
   }
 }
 
-void
-DeclContext::collectAllContexts(SmallVectorImpl<DeclContext *> &Contexts){
-  Contexts.clear();
+template <typename T>
+void collectAllContextsImpl(T *Self, SmallVectorImpl<DeclContext *> &Contexts) {
+  for (T *D = Self->getMostRecentDecl(); D; D = D->getPreviousDecl())
+    Contexts.push_back(D);
 
-  if (getDeclKind() != Decl::Namespace) {
-    Contexts.push_back(this);
-    return;
-  }
+  std::reverse(Contexts.begin(), Contexts.end());
+}
 
-  auto *Self = static_cast<NamespaceDecl *>(this);
-  for (NamespaceDecl *N = Self->getMostRecentDecl(); N;
-       N = N->getPreviousDecl())
-    Contexts.push_back(N);
+void DeclContext::collectAllContexts(SmallVectorImpl<DeclContext *> &Contexts) {
+  Contexts.clear();
 
-  std::reverse(Contexts.begin(), Contexts.end());
+  Decl::Kind Kind = getDeclKind();
+
+  if (Kind == Decl::TranslationUnit)
+    collectAllContextsImpl(static_cast<TranslationUnitDecl *>(this), Contexts);
+  else if (Kind == Decl::Namespace)
+    collectAllContextsImpl(static_cast<NamespaceDecl *>(this), Contexts);
+  else
+    Contexts.push_back(this);
 }
 
 std::pair<Decl *, Decl *>

clang/lib/Frontend/ASTUnit.cpp

@@ -807,7 +807,8 @@ std::unique_ptr<ASTUnit> ASTUnit::LoadFromASTFile(
   if (ToLoad >= LoadASTOnly)
     AST->Ctx = new ASTContext(*AST->LangOpts, AST->getSourceManager(),
                               PP.getIdentifierTable(), PP.getSelectorTable(),
-                              PP.getBuiltinInfo());
+                              PP.getBuiltinInfo(),
+                              AST->getTranslationUnitKind());
 
   DisableValidationForModuleKind disableValid =
       DisableValidationForModuleKind::None;

clang/lib/Frontend/CompilerInstance.cpp

@@ -551,7 +551,7 @@ void CompilerInstance::createASTContext() {
   Preprocessor &PP = getPreprocessor();
   auto *Context = new ASTContext(getLangOpts(), PP.getSourceManager(),
                                  PP.getIdentifierTable(), PP.getSelectorTable(),
-                                 PP.getBuiltinInfo());
+                                 PP.getBuiltinInfo(), PP.TUKind);
   Context->InitBuiltinTypes(getTarget(), getAuxTarget());
   setASTContext(Context);
 }