Fix crashing name mangler (#1910)

* Fix broken name mangler

* Remove llvm backend mangeling

src/CppParser/ASTNameMangler.cpp

@@ -0,0 +1,125 @@
+/************************************************************************
+*
+* CppSharp
+* Licensed under the simplified BSD license. All rights reserved.
+*
+************************************************************************/
+
+#include "ASTNameMangler.h"
+
+#include <clang/AST/GlobalDecl.h>
+#include <clang/AST/Mangle.h>
+#include <clang/AST/VTableBuilder.h>
+#include <clang/Basic/TargetInfo.h>
+#include <llvm/IR/Mangler.h>
+
+using namespace clang;
+using namespace CppSharp::CppParser;
+
+namespace {
+    enum ObjCKind {
+        ObjCClass,
+        ObjCMetaclass,
+    };
+
+    StringRef getClassSymbolPrefix(ObjCKind Kind, const ASTContext& Context) {
+        if (Context.getLangOpts().ObjCRuntime.isGNUFamily())
+            return Kind == ObjCMetaclass ? "_OBJC_METACLASS_" : "_OBJC_CLASS_";
+        return Kind == ObjCMetaclass ? "OBJC_METACLASS_$_" : "OBJC_CLASS_$_";
+    }
+
+    void WriteObjCClassName(const ObjCInterfaceDecl* D, raw_ostream& OS) {
+        OS << getClassSymbolPrefix(ObjCClass, D->getASTContext());
+        OS << D->getObjCRuntimeNameAsString();
+    }
+}
+
+ASTNameMangler::ASTNameMangler(ASTContext& Ctx)
+    : DL(Ctx.getTargetInfo().getDataLayoutString())
+    , MC(Ctx.createMangleContext())
+{
+}
+
+std::string ASTNameMangler::GetName(const Decl* D) {
+    std::string Name;
+    {
+        llvm::raw_string_ostream OS(Name);
+        WriteName(D, OS);
+    }
+    return Name;
+}
+
+bool ASTNameMangler::WriteName(const Decl* D, raw_ostream& OS) {
+    // First apply frontend mangling.
+    if (auto* FD = dyn_cast<FunctionDecl>(D)) {
+        if (FD->isDependentContext())
+            return true;
+        if (WriteFuncOrVarName(FD, OS))
+            return true;
+    }
+    else if (auto* VD = dyn_cast<VarDecl>(D)) {
+        if (WriteFuncOrVarName(VD, OS))
+            return true;
+    }
+    else if (auto* MD = dyn_cast<ObjCMethodDecl>(D)) {
+        MC->mangleObjCMethodName(MD, OS, /*includePrefixByte=*/false,
+            /*includeCategoryNamespace=*/true);
+        return false;
+    }
+    else if (auto* ID = dyn_cast<ObjCInterfaceDecl>(D)) {
+        WriteObjCClassName(ID, OS);
+    }
+    else {
+        return true;
+    }
+
+    return false;
+}
+
+std::string ASTNameMangler::GetMangledStructor(const NamedDecl* ND, unsigned StructorType) {
+    std::string FrontendBuf;
+    llvm::raw_string_ostream FOS(FrontendBuf);
+
+    GlobalDecl GD;
+    if (const auto* CD = dyn_cast_or_null<CXXConstructorDecl>(ND))
+        GD = GlobalDecl(CD, static_cast<CXXCtorType>(StructorType));
+    else if (const auto* DD = dyn_cast_or_null<CXXDestructorDecl>(ND))
+        GD = GlobalDecl(DD, static_cast<CXXDtorType>(StructorType));
+    MC->mangleName(GD, FOS);
+
+    return FrontendBuf;
+}
+
+std::string ASTNameMangler::GetMangledThunk(const CXXMethodDecl* MD, const ThunkInfo& T, bool /*ElideOverrideInfo*/) {
+    std::string FrontendBuf;
+    llvm::raw_string_ostream FOS(FrontendBuf);
+
+    // TODO: Enable `ElideOverrideInfo` param if clang is updated to 19
+    MC->mangleThunk(MD, T, /*ElideOverrideInfo,*/ FOS);
+
+    return FrontendBuf;
+}
+
+bool ASTNameMangler::WriteFuncOrVarName(const NamedDecl* D, raw_ostream& OS) const
+{
+    if (!MC->shouldMangleDeclName(D)) {
+        const IdentifierInfo* II = D->getIdentifier();
+        if (!II)
+            return true;
+        OS << II->getName();
+        return false;
+    }
+
+    GlobalDecl GD;
+    if (const auto* CtorD = dyn_cast<CXXConstructorDecl>(D))
+        GD = GlobalDecl(CtorD, Ctor_Base);
+    else if (const auto* DtorD = dyn_cast<CXXDestructorDecl>(D))
+        GD = GlobalDecl(DtorD, Dtor_Base);
+    else if (D->hasAttr<CUDAGlobalAttr>())
+        GD = GlobalDecl(cast<FunctionDecl>(D));
+    else
+        GD = GlobalDecl(D);
+
+    MC->mangleName(GD, OS);
+    return false;
+}

src/CppParser/ASTNameMangler.h

@@ -0,0 +1,50 @@
+/************************************************************************
+*
+* CppSharp
+* Licensed under the simplified BSD license. All rights reserved.
+*
+************************************************************************/
+
+#pragma once
+
+#include <clang/AST/ASTFwd.h>
+#include <llvm/IR/DataLayout.h>
+
+#include <string>
+
+namespace clang
+{
+    class ASTContext;
+    class MangleContext;
+    struct ThunkInfo;
+}
+
+namespace llvm
+{
+    class raw_ostream;
+}
+
+namespace CppSharp::CppParser {
+
+/// <summary>
+/// Helper class for getting the mangled name of a declaration
+/// </summary>
+/// <remarks>Source adapted from https://clang.llvm.org/doxygen/Mangle_8cpp_source.html#l00394</remarks>
+class ASTNameMangler
+{
+public:
+    explicit ASTNameMangler(clang::ASTContext& Ctx);
+
+    std::string GetName(const clang::Decl* D);
+    bool WriteName(const clang::Decl* D, llvm::raw_ostream& OS);
+
+private:
+    std::string GetMangledStructor(const clang::NamedDecl* ND, unsigned StructorType);
+    std::string GetMangledThunk(const clang::CXXMethodDecl* MD, const clang::ThunkInfo& T, bool ElideOverrideInfo);
+    bool WriteFuncOrVarName(const clang::NamedDecl* D, llvm::raw_ostream& OS) const;
+    
+    llvm::DataLayout DL;
+    std::unique_ptr<clang::MangleContext> MC;
+};
+
+}

src/CppParser/Parser.cpp

@@ -58,6 +58,8 @@
 #include <Driver/ToolChains/Linux.h>
 #include <Driver/ToolChains/MSVC.h>
 
+#include "ASTNameMangler.h"
+
 #if defined(__APPLE__) || defined(__linux__)
 #ifndef _GNU_SOURCE
 #define _GNU_SOURCE
@@ -468,69 +470,55 @@ void Parser::Setup(bool Compile)
         PP.getLangOpts());
 
     c->createASTContext();
+    NameMangler.reset(new ASTNameMangler(c->getASTContext()));
 }
 
 //-----------------------------------//
 
-std::string Parser::GetDeclMangledName(const clang::Decl* D)
+std::string Parser::GetDeclMangledName(const clang::Decl* D) const
 {
-    using namespace clang;
+    // Source adapted from https://clang.llvm.org/doxygen/JSONNodeDumper_8cpp_source.html#l00845
 
-    if(!D || !isa<NamedDecl>(D))
-        return "";
-
-    bool CanMangle = isa<FunctionDecl>(D) || isa<VarDecl>(D)
-        || isa<CXXConstructorDecl>(D) || isa<CXXDestructorDecl>(D);
+    using namespace clang;
 
-    if (!CanMangle) return "";
+    auto ND = dyn_cast_or_null<NamedDecl>(D);
+    if (!ND || !ND->getDeclName())
+        return {};
 
-    auto ND = cast<NamedDecl>(D);
-    std::unique_ptr<MangleContext> MC;
+    // If the declaration is dependent or is in a dependent context, then the
+    // mangling is unlikely to be meaningful (and in some cases may cause
+    // "don't know how to mangle this" assertion failures.)
+    if (ND->isTemplated())
+        return {};
 
-    auto& AST = c->getASTContext();
-    auto targetABI = c->getTarget().getCXXABI().getKind();
-    switch(targetABI)
-    {
-    default:
-       MC.reset(ItaniumMangleContext::create(AST, AST.getDiagnostics()));
-       break;
-    case TargetCXXABI::Microsoft:
-       MC.reset(MicrosoftMangleContext::create(AST, AST.getDiagnostics()));
-       break;
-    }
-
-    if (!MC)
-        llvm_unreachable("Unknown mangling ABI");
-
-    std::string Mangled;
-    llvm::raw_string_ostream Out(Mangled);
+    /*bool CanMangle = isa<FunctionDecl>(D) || isa<VarDecl>(D)
+        || isa<CXXConstructorDecl>(D) || isa<CXXDestructorDecl>(D);
 
-    bool IsDependent = false;
-    if (const ValueDecl *VD = dyn_cast<ValueDecl>(ND))
-        IsDependent |= VD->getType()->isDependentType();
+    if (!CanMangle)
+        return {};*/
 
-    if (!IsDependent)
-        IsDependent |= ND->getDeclContext()->isDependentContext();
+    // FIXME: There are likely other contexts in which it makes no sense to ask
+    // for a mangled name.
+    if (isa<RequiresExprBodyDecl>(ND->getDeclContext()))
+        return {};
 
-    if (!MC->shouldMangleDeclName(ND) || IsDependent)
-        return ND->getDeclName().getAsString();
+    // Do not mangle template deduction guides.
+    if (isa<CXXDeductionGuideDecl>(ND))
+        return {};
 
-    GlobalDecl GD;
-    if (const CXXConstructorDecl *CD = dyn_cast<CXXConstructorDecl>(ND))
-        GD = GlobalDecl(CD, Ctor_Base);
-    else if (const CXXDestructorDecl *DD = dyn_cast<CXXDestructorDecl>(ND))
-        GD = GlobalDecl(DD, Dtor_Base);
-    else
-        GD = GlobalDecl(ND);
-    MC->mangleName(GD, Out);
+    // Mangled names are not meaningful for locals, and may not be well-defined
+    // in the case of VLAs.
+    auto* VD = dyn_cast<VarDecl>(ND);
+    if (VD && VD->hasLocalStorage())
+        return {};
 
-    Out.flush();
+    std::string MangledName = NameMangler->GetName(ND);
 
     // Strip away LLVM name marker.
-    if(!Mangled.empty() && Mangled[0] == '\01')
-        Mangled = Mangled.substr(1);
+    if (!MangledName.empty() && MangledName[0] == '\01')
+        MangledName = MangledName.substr(1);
 
-    return Mangled;
+    return MangledName;
 }
 
 //-----------------------------------//
@@ -632,7 +620,7 @@ static clang::SourceLocation GetDeclStartLocation(clang::CompilerInstance* C,
     return GetDeclStartLocation(C, prevDecl);
 }
 
-std::string Parser::GetTypeName(const clang::Type* Type)
+std::string Parser::GetTypeName(const clang::Type* Type) const
 {
     using namespace clang;
 

src/CppParser/Parser.h

@@ -47,6 +47,7 @@ namespace clang {
 #define Debug printf
 
 namespace CppSharp { namespace CppParser {
+    class ASTNameMangler;
 
 class Parser
 {
@@ -133,8 +134,8 @@ private:
     // Clang helpers
     SourceLocationKind GetLocationKind(const clang::SourceLocation& Loc);
     bool IsValidDeclaration(const clang::SourceLocation& Loc);
-    std::string GetDeclMangledName(const clang::Decl* D);
-    std::string GetTypeName(const clang::Type* Type);
+    std::string GetDeclMangledName(const clang::Decl* D) const;
+    std::string GetTypeName(const clang::Type* Type) const;
     bool CanCheckCodeGenInfo(const clang::Type* Ty);
     void CompleteIfSpecializationType(const clang::QualType& QualType);
     Parameter* WalkParameter(const clang::ParmVarDecl* PVD,
@@ -181,6 +182,7 @@ private:
     int index;
     std::unique_ptr<clang::CompilerInstance> c;
     llvm::LLVMContext LLVMCtx;
+    std::unique_ptr<ASTNameMangler> NameMangler;
     std::unique_ptr<llvm::Module> LLVMModule;
     std::unique_ptr<clang::CodeGen::CodeGenModule> CGM;
     std::unique_ptr<clang::CodeGen::CodeGenTypes> codeGenTypes;