Refine types based on debug metadata

include/rellic/AST/IRToASTVisitor.h

@@ -31,6 +31,7 @@ using IRToValDeclMap = std::unordered_map<llvm::Value *, clang::ValueDecl *>;
 using IRToStmtMap = std::unordered_map<llvm::Value *, clang::Stmt *>;
 using DIToTypedefMap =
     std::unordered_map<llvm::DIDerivedType *, clang::TypedefNameDecl *>;
+using ArgToTempMap = std::unordered_map<llvm::Argument *, clang::VarDecl *>;
 
 class IRToASTVisitor : public llvm::InstVisitor<IRToASTVisitor> {
  private:
@@ -43,6 +44,7 @@ class IRToASTVisitor : public llvm::InstVisitor<IRToASTVisitor> {
   DIToTypedefMap typedef_decls;
   IRToStmtMap stmts;
   DebugInfoCollector &dic;
+  ArgToTempMap temp_decls;
 
   clang::Expr *GetOperandExpr(llvm::Value *val);
   clang::QualType GetQualType(llvm::Type *type, llvm::DIType *ditype);

include/rellic/BC/Util.h

@@ -24,6 +24,7 @@ namespace rellic {
 // Serialize an LLVM object into a string.
 std::string LLVMThingToString(llvm::Value *thing);
 std::string LLVMThingToString(llvm::Type *thing);
+std::string LLVMThingToString(llvm::DIType *thing);
 
 // Try to verify a module.
 bool VerifyModule(llvm::Module *module);

lib/AST/IRToASTVisitor.cpp

@@ -66,6 +66,7 @@ clang::QualType IRToASTVisitor::GetQualType(llvm::Type *type,
           }
           return ast_ctx.getTypedefType(tdef_decl);
         } break;
+        case llvm::dwarf::DW_TAG_inheritance:
         case llvm::dwarf::DW_TAG_member: {
           return GetQualType(type, derived->getBaseType());
         };
@@ -526,8 +527,41 @@ void IRToASTVisitor::VisitArgument(llvm::Argument &arg) {
       ditype = ditype_array[arg.getArgNo() + 1];
     }
   }
+
+  auto argtype{arg.getType()};
+  if (arg.hasByValAttr()) {
+    auto byval{arg.getAttribute(llvm::Attribute::ByVal)};
+    argtype = byval.getValueAsType();
+  }
   // Create a declaration
-  parm = ast.CreateParamDecl(fdecl, GetQualType(arg.getType(), ditype), name);
+  parm = ast.CreateParamDecl(fdecl, GetQualType(argtype, ditype), name);
+}
+
+// This function fixes function types for those functions that have arguments
+// that are passed by value using the `byval` attribute.
+// They need special treatment because those arguments, instead of actually
+// being passed by value, are instead passed "by reference" from a bitcode point
+// of view, with the caveat that the actual semantics are more like "create a
+// copy of the reference before calling, and pass a pointer to that copy
+// instead" (this is done implicitly).
+// Thus, we need to convert a function type like
+//   i32 @do_foo(%struct.foo* byval(%struct.foo) align 4 %f)
+// into
+//   i32 @do_foo(%struct.foo %f)
+static llvm::FunctionType *GetFixedFunctionType(llvm::Function &func) {
+  std::vector<llvm::Type *> new_arg_types{};
+
+  for (auto &arg : func.args()) {
+    if (arg.hasByValAttr()) {
+      auto ptrtype{llvm::cast<llvm::PointerType>(arg.getType())};
+      new_arg_types.push_back(ptrtype->getElementType());
+    } else {
+      new_arg_types.push_back(arg.getType());
+    }
+  }
+
+  return llvm::FunctionType::get(func.getReturnType(), new_arg_types,
+                                 func.isVarArg());
 }
 
 void IRToASTVisitor::VisitFunctionDecl(llvm::Function &func) {
@@ -546,7 +580,7 @@ void IRToASTVisitor::VisitFunctionDecl(llvm::Function &func) {
 
   DLOG(INFO) << "Creating FunctionDecl for " << name;
   auto tudecl{ast_ctx.getTranslationUnitDecl()};
-  auto ftype{func.getFunctionType()};
+  auto ftype{GetFixedFunctionType(func)};
   auto type{GetQualType(ftype, dic.GetIRFuncToDITypeMap()[&func])};
   decl = ast.CreateFunctionDecl(tudecl, type, name);
 

lib/AST/StructFieldRenamer.cpp

@@ -60,6 +60,9 @@ bool StructFieldRenamer::VisitRecordDecl(clang::RecordDecl *decl) {
     // FIXME(frabert): Is a clash between field names actually possible?
     // Can this mechanism actually be left out?
     auto name{di_field->getName().str()};
+    if (di_field->getTag() == llvm::dwarf::DW_TAG_inheritance) {
+      name = di_field->getBaseType()->getName().str() + "_base";
+    }
     if (seen_names.find(name) == seen_names.end()) {
       seen_names.insert(name);
       decl_field->setDeclName(ast.CreateIdentifier(name));

lib/BC/Util.cpp

@@ -55,6 +55,10 @@ std::string LLVMThingToString(llvm::Type *thing) {
   return DoLLVMThingToString(thing);
 }
 
+std::string LLVMThingToString(llvm::DIType *thing) {
+  return DoLLVMThingToString(thing);
+}
+
 // Try to verify a module.
 bool VerifyModule(llvm::Module *module) {
   std::string error;

tests/tools/decomp/byval_struct.c

@@ -14,4 +14,5 @@ int main() {
     struct foo f = {atoi("1"), atoi("2"), atoi("3"), atoi("4")};
     long long x = get_3x(f);
     printf("%lld %lld\n", f.x, x);
+    return 0;
 }