Codegen for object access.

Changelog.md

@@ -9,6 +9,7 @@ Compiler Features:
  * Code Generator: Use binary search for dispatch function if more efficient. The size/speed tradeoff can be tuned using ``--optimize-runs``.
  * SMTChecker: Support mathematical and cryptographic functions in an uninterpreted way.
  * Type Checker: Add an additional reason to be displayed when type conversion fails.
+ * Yul: Support object access via ``datasize``, ``dataoffset`` and ``datacopy`` in standalone assembly mode.
 
 
 Bugfixes:

libsolidity/analysis/ReferencesResolver.cpp

@@ -27,6 +27,7 @@
 #include <libyul/AsmAnalysis.h>
 #include <libyul/AsmAnalysisInfo.h>
 #include <libyul/AsmData.h>
+#include <libyul/backends/evm/EVMDialect.h>
 #include <liblangutil/ErrorReporter.h>
 #include <liblangutil/Exceptions.h>
 
@@ -321,7 +322,7 @@ bool ReferencesResolver::visit(InlineAssembly const& _inlineAssembly)
 		errorsIgnored,
 		EVMVersion(),
 		errorTypeForLoose,
-		yul::Dialect::looseAssemblyForEVM(),
+		yul::EVMDialect::looseAssemblyForEVM(),
 		resolver
 	).analyze(_inlineAssembly.operations());
 	return false;

libsolidity/analysis/TypeChecker.cpp

@@ -27,6 +27,8 @@
 #include <libyul/AsmAnalysisInfo.h>
 #include <libyul/AsmData.h>
 
+#include <libyul/backends/evm/EVMDialect.h>
+
 #include <liblangutil/ErrorReporter.h>
 
 #include <libdevcore/Algorithms.h>
@@ -658,7 +660,7 @@ bool TypeChecker::visit(InlineAssembly const& _inlineAssembly)
 		m_errorReporter,
 		m_evmVersion,
 		Error::Type::SyntaxError,
-		yul::Dialect::looseAssemblyForEVM(),
+		yul::EVMDialect::looseAssemblyForEVM(),
 		identifierAccess
 	);
 	if (!analyzer.analyze(_inlineAssembly.operations()))

libsolidity/codegen/AsmCodeGen.cpp

@@ -188,8 +188,8 @@ void CodeGenerator::assemble(
 		assemblyAdapter,
 		_analysisInfo,
 		_parsedData,
+		*EVMDialect::strictAssemblyForEVM(),
 		_optimize,
-		Dialect::strictAssemblyForEVM(),
 		false,
 		_identifierAccess,
 		_useNamedLabelsForFunctions

libsolidity/codegen/CompilerContext.cpp

@@ -31,6 +31,7 @@
 #include <libyul/AsmAnalysis.h>
 #include <libyul/AsmAnalysisInfo.h>
 #include <libyul/YulString.h>
+#include <libyul/backends/evm/EVMDialect.h>
 #include <liblangutil/ErrorReporter.h>
 #include <liblangutil/Scanner.h>
 
@@ -361,7 +362,7 @@ void CompilerContext::appendInlineAssembly(
 	ErrorList errors;
 	ErrorReporter errorReporter(errors);
 	auto scanner = make_shared<langutil::Scanner>(langutil::CharStream(_assembly, "--CODEGEN--"));
-	auto parserResult = yul::Parser(errorReporter, yul::Dialect::strictAssemblyForEVM()).parse(scanner, false);
+	auto parserResult = yul::Parser(errorReporter, yul::EVMDialect::strictAssemblyForEVM()).parse(scanner, false);
 #ifdef SOL_OUTPUT_ASM
 	cout << yul::AsmPrinter()(*parserResult) << endl;
 #endif
@@ -373,7 +374,7 @@ void CompilerContext::appendInlineAssembly(
 			errorReporter,
 			m_evmVersion,
 			boost::none,
-			yul::Dialect::strictAssemblyForEVM(),
+			yul::EVMDialect::strictAssemblyForEVM(),
 			identifierAccess.resolve
 		).analyze(*parserResult);
 	if (!parserResult || !errorReporter.errors().empty() || !analyzerResult)

libsolidity/interface/AssemblyStack.cpp

@@ -32,6 +32,7 @@
 #include <libyul/backends/evm/EVMObjectCompiler.h>
 #include <libyul/backends/evm/EVMCodeTransform.h>
 #include <libyul/backends/evm/EVMAssembly.h>
+#include <libyul/backends/evm/EVMDialect.h>
 #include <libyul/ObjectParser.h>
 
 #include <libevmasm/Assembly.h>
@@ -45,14 +46,14 @@ using namespace dev::solidity;
 
 namespace
 {
-yul::Dialect languageToDialect(AssemblyStack::Language _language)
+shared_ptr<yul::Dialect> languageToDialect(AssemblyStack::Language _language)
 {
 	switch (_language)
 	{
 	case AssemblyStack::Language::Assembly:
-		return yul::Dialect::looseAssemblyForEVM();
+		return yul::EVMDialect::looseAssemblyForEVM();
 	case AssemblyStack::Language::StrictAssembly:
-		return yul::Dialect::strictAssemblyForEVMObjects();
+		return yul::EVMDialect::strictAssemblyForEVMObjects();
 	case AssemblyStack::Language::Yul:
 		return yul::Dialect::yul();
 	}
@@ -112,6 +113,22 @@ bool AssemblyStack::analyzeParsed(yul::Object& _object)
 	return success;
 }
 
+void AssemblyStack::compileEVM(yul::AbstractAssembly& _assembly, bool _evm15, bool _optimize) const
+{
+	shared_ptr<yul::EVMDialect> dialect;
+
+	if (m_language == Language::Assembly)
+		dialect = yul::EVMDialect::looseAssemblyForEVM();
+	else if (m_language == AssemblyStack::Language::StrictAssembly)
+		dialect = yul::EVMDialect::strictAssemblyForEVMObjects();
+	else if (m_language == AssemblyStack::Language::Yul)
+		dialect = yul::EVMDialect::yulForEVM();
+	else
+		solAssert(false, "Invalid language.");
+
+	yul::EVMObjectCompiler::compile(*m_parserResult, _assembly, *dialect, _evm15, _optimize);
+}
+
 void AssemblyStack::optimize(yul::Object& _object)
 {
 	solAssert(_object.code, "");
@@ -136,7 +153,7 @@ MachineAssemblyObject AssemblyStack::assemble(Machine _machine, bool _optimize)
 		MachineAssemblyObject object;
 		eth::Assembly assembly;
 		EthAssemblyAdapter adapter(assembly);
-		yul::EVMObjectCompiler::compile(*m_parserResult, adapter, languageToDialect(m_language), false, _optimize);
+		compileEVM(adapter, false, _optimize);
 		object.bytecode = make_shared<eth::LinkerObject>(assembly.assemble());
 		object.assembly = assembly.assemblyString();
 		return object;
@@ -145,7 +162,7 @@ MachineAssemblyObject AssemblyStack::assemble(Machine _machine, bool _optimize)
 	{
 		MachineAssemblyObject object;
 		yul::EVMAssembly assembly(true);
-		yul::EVMObjectCompiler::compile(*m_parserResult, assembly, languageToDialect(m_language), true, _optimize);
+		compileEVM(assembly, true, _optimize);
 		object.bytecode = make_shared<eth::LinkerObject>(assembly.finalize());
 		/// TODO: fill out text representation
 		return object;

libsolidity/interface/AssemblyStack.h

@@ -36,6 +36,10 @@ namespace langutil
 {
 class Scanner;
 }
+namespace yul
+{
+class AbstractAssembly;
+}
 
 namespace dev
 {
@@ -86,6 +90,8 @@ private:
 	bool analyzeParsed();
 	bool analyzeParsed(yul::Object& _object);
 
+	void compileEVM(yul::AbstractAssembly& _assembly, bool _evm15, bool _optimize) const;
+
 	void optimize(yul::Object& _object);
 
 	Language m_language = Language::Assembly;

libsolidity/parsing/Parser.cpp

@@ -24,6 +24,7 @@
 #include <vector>
 #include <libsolidity/parsing/Parser.h>
 #include <libyul/AsmParser.h>
+#include <libyul/backends/evm/EVMDialect.h>
 #include <liblangutil/SourceLocation.h>
 #include <liblangutil/ErrorReporter.h>
 #include <liblangutil/Scanner.h>
@@ -1012,7 +1013,7 @@ ASTPointer<InlineAssembly> Parser::parseInlineAssembly(ASTPointer<ASTString> con
 		m_scanner->next();
 	}
 
-	yul::Parser asmParser(m_errorReporter);
+	yul::Parser asmParser(m_errorReporter, yul::EVMDialect::looseAssemblyForEVM());
 	shared_ptr<yul::Block> block = asmParser.parse(m_scanner, true);
 	nodeFactory.markEndPosition();
 	return nodeFactory.createNode<InlineAssembly>(_docString, block);

libyul/AsmAnalysis.cpp

@@ -101,7 +101,7 @@ bool AsmAnalyzer::operator()(Literal const& _literal)
 	}
 	else if (_literal.kind == LiteralKind::Boolean)
 	{
-		solAssert(m_dialect.flavour == AsmFlavour::Yul, "");
+		solAssert(m_dialect->flavour == AsmFlavour::Yul, "");
 		solAssert(_literal.value == YulString{string("true")} || _literal.value == YulString{string("false")}, "");
 	}
 	m_info.stackHeightInfo[&_literal] = m_stackHeight;
@@ -164,7 +164,7 @@ bool AsmAnalyzer::operator()(Identifier const& _identifier)
 
 bool AsmAnalyzer::operator()(FunctionalInstruction const& _instr)
 {
-	solAssert(m_dialect.flavour != AsmFlavour::Yul, "");
+	solAssert(m_dialect->flavour != AsmFlavour::Yul, "");
 	bool success = true;
 	for (auto const& arg: _instr.arguments | boost::adaptors::reversed)
 		if (!expectExpression(arg))
@@ -182,9 +182,9 @@ bool AsmAnalyzer::operator()(ExpressionStatement const& _statement)
 {
 	int initialStackHeight = m_stackHeight;
 	bool success = boost::apply_visitor(*this, _statement.expression);
-	if (m_stackHeight != initialStackHeight && (m_dialect.flavour != AsmFlavour::Loose || m_errorTypeForLoose))
+	if (m_stackHeight != initialStackHeight && (m_dialect->flavour != AsmFlavour::Loose || m_errorTypeForLoose))
 	{
-		Error::Type errorType = m_dialect.flavour == AsmFlavour::Loose ? *m_errorTypeForLoose : Error::Type::TypeError;
+		Error::Type errorType = m_dialect->flavour == AsmFlavour::Loose ? *m_errorTypeForLoose : Error::Type::TypeError;
 		string msg =
 			"Top-level expressions are not supposed to return values (this expression returns " +
 			to_string(m_stackHeight - initialStackHeight) +
@@ -299,7 +299,7 @@ bool AsmAnalyzer::operator()(FunctionCall const& _funCall)
 	bool success = true;
 	size_t parameters = 0;
 	size_t returns = 0;
-	if (BuiltinFunction const* f = m_dialect.builtins->query(_funCall.functionName.name))
+	if (BuiltinFunction const* f = m_dialect->builtin(_funCall.functionName.name))
 	{
 		// TODO: compare types, too
 		parameters = f->parameters.size();
@@ -569,7 +569,7 @@ Scope& AsmAnalyzer::scope(Block const* _block)
 }
 void AsmAnalyzer::expectValidType(string const& type, SourceLocation const& _location)
 {
-	if (m_dialect.flavour != AsmFlavour::Yul)
+	if (m_dialect->flavour != AsmFlavour::Yul)
 		return;
 
 	if (!builtinTypes.count(type))
@@ -629,7 +629,7 @@ void AsmAnalyzer::warnOnInstructions(solidity::Instruction _instr, SourceLocatio
 
 	if (_instr == solidity::Instruction::JUMP || _instr == solidity::Instruction::JUMPI || _instr == solidity::Instruction::JUMPDEST)
 	{
-		if (m_dialect.flavour != AsmFlavour::Loose)
+		if (m_dialect->flavour != AsmFlavour::Loose)
 			solAssert(m_errorTypeForLoose && *m_errorTypeForLoose != Error::Type::Warning, "");
 
 		m_errorReporter.error(
@@ -644,7 +644,7 @@ void AsmAnalyzer::warnOnInstructions(solidity::Instruction _instr, SourceLocatio
 
 void AsmAnalyzer::checkLooseFeature(SourceLocation const& _location, string const& _description)
 {
-	if (m_dialect.flavour != AsmFlavour::Loose)
+	if (m_dialect->flavour != AsmFlavour::Loose)
 		solAssert(false, _description);
 	else if (m_errorTypeForLoose)
 		m_errorReporter.error(*m_errorTypeForLoose, _location, _description);

libyul/AsmAnalysis.h

@@ -59,7 +59,7 @@ public:
 		langutil::ErrorReporter& _errorReporter,
 		dev::solidity::EVMVersion _evmVersion,
 		boost::optional<langutil::Error::Type> _errorTypeForLoose,
-		Dialect _dialect = Dialect::looseAssemblyForEVM(),
+		std::shared_ptr<Dialect> _dialect,
 		ExternalIdentifierAccess::Resolver const& _resolver = ExternalIdentifierAccess::Resolver()
 	):
 		m_resolver(_resolver),
@@ -115,7 +115,7 @@ private:
 	AsmAnalysisInfo& m_info;
 	langutil::ErrorReporter& m_errorReporter;
 	dev::solidity::EVMVersion m_evmVersion;
-	Dialect m_dialect = Dialect::looseAssemblyForEVM();
+	std::shared_ptr<Dialect> m_dialect;
 	boost::optional<langutil::Error::Type> m_errorTypeForLoose;
 };
 

libyul/AsmParser.cpp

@@ -107,14 +107,14 @@ Statement Parser::parseStatement()
 		return parseForLoop();
 	case Token::Assign:
 	{
-		if (m_dialect.flavour != AsmFlavour::Loose)
+		if (m_dialect->flavour != AsmFlavour::Loose)
 			break;
 		StackAssignment assignment = createWithLocation<StackAssignment>();
 		advance();
 		expectToken(Token::Colon);
 		assignment.variableName.location = location();
 		assignment.variableName.name = YulString(currentLiteral());
-		if (m_dialect.builtins->query(assignment.variableName.name))
+		if (m_dialect->builtin(assignment.variableName.name))
 			fatalParserError("Identifier expected, got builtin symbol.");
 		else if (instructions().count(assignment.variableName.name.str()))
 			fatalParserError("Identifier expected, got instruction name.");
@@ -176,9 +176,9 @@ Statement Parser::parseStatement()
 		if (currentToken() == Token::Assign && peekNextToken() != Token::Colon)
 		{
 			Assignment assignment = createWithLocation<Assignment>(identifier.location);
-			if (m_dialect.builtins->query(identifier.name))
+			if (m_dialect->builtin(identifier.name))
 				fatalParserError("Cannot assign to builtin function \"" + identifier.name.str() + "\".");
-			else if (m_dialect.flavour != AsmFlavour::Yul && instructions().count(identifier.name.str()))
+			else if (m_dialect->flavour != AsmFlavour::Yul && instructions().count(identifier.name.str()))
 				fatalParserError("Cannot use instruction names for identifier names.");
 			advance();
 			assignment.variableNames.emplace_back(identifier);
@@ -189,7 +189,7 @@ Statement Parser::parseStatement()
 		else
 		{
 			// label
-			if (m_dialect.flavour != AsmFlavour::Loose)
+			if (m_dialect->flavour != AsmFlavour::Loose)
 				fatalParserError("Labels are not supported.");
 			Label label = createWithLocation<Label>(identifier.location);
 			label.name = identifier.name;
@@ -197,7 +197,7 @@ Statement Parser::parseStatement()
 		}
 	}
 	default:
-		if (m_dialect.flavour != AsmFlavour::Loose)
+		if (m_dialect->flavour != AsmFlavour::Loose)
 			fatalParserError("Call or assignment expected.");
 		break;
 	}
@@ -273,7 +273,7 @@ Expression Parser::parseExpression()
 				instructionNames().at(instr.instruction) +
 				"\" not allowed in this context."
 			);
-		if (m_dialect.flavour != AsmFlavour::Loose && currentToken() != Token::LParen)
+		if (m_dialect->flavour != AsmFlavour::Loose && currentToken() != Token::LParen)
 			fatalParserError(
 				"Non-functional instructions are not allowed in this context."
 			);
@@ -293,7 +293,7 @@ Expression Parser::parseExpression()
 	else if (operation.type() == typeid(Instruction))
 	{
 		// Instructions not taking arguments are allowed as expressions.
-		solAssert(m_dialect.flavour == AsmFlavour::Loose, "");
+		solAssert(m_dialect->flavour == AsmFlavour::Loose, "");
 		Instruction& instr = boost::get<Instruction>(operation);
 		return FunctionalInstruction{std::move(instr.location), instr.instruction, {}};
 	}
@@ -362,9 +362,9 @@ Parser::ElementaryOperation Parser::parseElementaryOperation()
 		else
 			literal = YulString{currentLiteral()};
 		// first search the set of builtins, then the instructions.
-		if (m_dialect.builtins->query(literal))
+		if (m_dialect->builtin(literal))
 			ret = Identifier{location(), literal};
-		else if (m_dialect.flavour != AsmFlavour::Yul && instructions().count(literal.str()))
+		else if (m_dialect->flavour != AsmFlavour::Yul && instructions().count(literal.str()))
 		{
 			dev::solidity::Instruction const& instr = instructions().at(literal.str());
 			ret = Instruction{location(), instr};
@@ -405,7 +405,7 @@ Parser::ElementaryOperation Parser::parseElementaryOperation()
 			{}
 		};
 		advance();
-		if (m_dialect.flavour == AsmFlavour::Yul)
+		if (m_dialect->flavour == AsmFlavour::Yul)
 		{
 			expectToken(Token::Colon);
 			literal.location.end = endPosition();
@@ -418,7 +418,7 @@ Parser::ElementaryOperation Parser::parseElementaryOperation()
 	}
 	default:
 		fatalParserError(
-			m_dialect.flavour == AsmFlavour::Yul ?
+			m_dialect->flavour == AsmFlavour::Yul ?
 			"Literal or identifier expected." :
 			"Literal, identifier or instruction expected."
 		);
@@ -488,7 +488,7 @@ Expression Parser::parseCall(Parser::ElementaryOperation&& _initialOp)
 	RecursionGuard recursionGuard(*this);
 	if (_initialOp.type() == typeid(Instruction))
 	{
-		solAssert(m_dialect.flavour != AsmFlavour::Yul, "Instructions are invalid in Yul");
+		solAssert(m_dialect->flavour != AsmFlavour::Yul, "Instructions are invalid in Yul");
 		Instruction& instruction = boost::get<Instruction>(_initialOp);
 		FunctionalInstruction ret;
 		ret.instruction = instruction.instruction;
@@ -559,7 +559,7 @@ Expression Parser::parseCall(Parser::ElementaryOperation&& _initialOp)
 	}
 	else
 		fatalParserError(
-			m_dialect.flavour == AsmFlavour::Yul ?
+			m_dialect->flavour == AsmFlavour::Yul ?
 			"Function name expected." :
 			"Assembly instruction or function name required in front of \"(\")"
 		);
@@ -572,7 +572,7 @@ TypedName Parser::parseTypedName()
 	RecursionGuard recursionGuard(*this);
 	TypedName typedName = createWithLocation<TypedName>();
 	typedName.name = expectAsmIdentifier();
-	if (m_dialect.flavour == AsmFlavour::Yul)
+	if (m_dialect->flavour == AsmFlavour::Yul)
 	{
 		expectToken(Token::Colon);
 		typedName.location.end = endPosition();
@@ -584,7 +584,7 @@ TypedName Parser::parseTypedName()
 YulString Parser::expectAsmIdentifier()
 {
 	YulString name = YulString{currentLiteral()};
-	if (m_dialect.flavour == AsmFlavour::Yul)
+	if (m_dialect->flavour == AsmFlavour::Yul)
 	{
 		switch (currentToken())
 		{
@@ -598,7 +598,7 @@ YulString Parser::expectAsmIdentifier()
 			break;
 		}
 	}
-	else if (m_dialect.builtins->query(name))
+	else if (m_dialect->builtin(name))
 		fatalParserError("Cannot use builtin function name \"" + name.str() + "\" as identifier name.");
 	else if (instructions().count(name.str()))
 		fatalParserError("Cannot use instruction names for identifier names.");

libyul/AsmParser.h

@@ -38,7 +38,7 @@ namespace yul
 class Parser: public langutil::ParserBase
 {
 public:
-	explicit Parser(langutil::ErrorReporter& _errorReporter, Dialect _dialect = Dialect::looseAssemblyForEVM()):
+	explicit Parser(langutil::ErrorReporter& _errorReporter, std::shared_ptr<Dialect> _dialect):
 		ParserBase(_errorReporter), m_dialect(std::move(_dialect)) {}
 
 	/// Parses an inline assembly block starting with `{` and ending with `}`.
@@ -86,7 +86,7 @@ protected:
 	static bool isValidNumberLiteral(std::string const& _literal);
 
 private:
-	Dialect m_dialect = Dialect::looseAssemblyForEVM();
+	std::shared_ptr<Dialect> m_dialect;
 };
 
 }

libyul/CMakeLists.txt

@@ -10,6 +10,7 @@ add_library(yul
 	ObjectParser.cpp
 	backends/evm/EVMAssembly.cpp
 	backends/evm/EVMCodeTransform.cpp
+	backends/evm/EVMDialect.cpp
 	backends/evm/EVMObjectCompiler.cpp
 	optimiser/ASTCopier.cpp
 	optimiser/ASTWalker.cpp

libyul/Dialect.cpp

@@ -20,57 +20,10 @@
 
 #include <libyul/Dialect.h>
 
+#include <libyul/Object.h>
+#include <libyul/backends/evm/AbstractAssembly.h>
+
 #include <map>
 
 using namespace yul;
 using namespace std;
-
-namespace
-{
-
-void addFunction(
-	map<YulString, BuiltinFunction>& _repository,
-	string const& _name,
-	size_t _params,
-	size_t _returns,
-	bool _movable
-)
-{
-	_repository[YulString{_name}] = BuiltinFunction{
-		YulString{_name},
-		vector<Type>(_params),
-		vector<Type>(_returns),
-		_movable
-	};
-}
-
-class GenericBuiltins: public Builtins
-{
-public:
-	GenericBuiltins(map<YulString, BuiltinFunction> const& _functions): m_functions(_functions) {}
-	BuiltinFunction const* query(YulString _name) const
-	{
-		auto it = m_functions.find(_name);
-		if (it != end(m_functions))
-			return &it->second;
-		else
-			return nullptr;
-	}
-private:
-	map<YulString, BuiltinFunction> const& m_functions;
-};
-
-}
-
-Dialect Dialect::strictAssemblyForEVMObjects()
-{
-	static map<YulString, BuiltinFunction> functions;
-	if (functions.empty())
-	{
-		addFunction(functions, "datasize", 1, 1, true);
-		addFunction(functions, "dataoffset", 1, 1, true);
-		addFunction(functions, "datacopy", 3, 0, false);
-	}
-	// The EVM instructions will be moved to builtins at some point.
-	return Dialect{AsmFlavour::Strict, std::make_shared<GenericBuiltins>(functions)};
-}

libyul/Dialect.h

@@ -22,7 +22,9 @@
 
 #include <libyul/YulString.h>
 
-#include <memory>
+#include <boost/noncopyable.hpp>
+
+#include <vector>
 
 namespace yul
 {
@@ -45,38 +47,19 @@ struct BuiltinFunction
 	bool movable;
 };
 
-/**
+struct Dialect: boost::noncopyable
- * Class to query for builtin functions and their semantics.
- */
-struct Builtins
 {
-	virtual ~Builtins() = default;
+	AsmFlavour const flavour = AsmFlavour::Loose;
 	/// @returns the builtin function of the given name or a nullptr if it is not a builtin function.
-	virtual BuiltinFunction const* query(YulString /*_name*/) const { return nullptr; }
+	virtual BuiltinFunction const* builtin(YulString /*_name*/) const { return nullptr; }
-};
 
-struct Dialect
+	Dialect(AsmFlavour _flavour): flavour(_flavour) {}
-{
+	virtual ~Dialect() {}
-	AsmFlavour flavour = AsmFlavour::Loose;
-	std::shared_ptr<Builtins> builtins;
 
-	Dialect(AsmFlavour _flavour, std::shared_ptr<Builtins> _builtins):
+	static std::shared_ptr<Dialect> yul()
-		flavour(_flavour), builtins(std::move(_builtins))
-	{}
-	static Dialect looseAssemblyForEVM()
-	{
-		return Dialect{AsmFlavour::Loose, std::make_shared<Builtins>()};
-	}
-	static Dialect strictAssemblyForEVM()
-	{
-		// The EVM instructions will be moved to builtins at some point.
-		return Dialect{AsmFlavour::Strict, std::make_shared<Builtins>()};
-	}
-	static Dialect strictAssemblyForEVMObjects();
-	static Dialect yul()
 	{
 		// Will have to add builtins later.
-		return Dialect{AsmFlavour::Yul, std::make_shared<Builtins>()};
+		return std::make_shared<Dialect>(AsmFlavour::Yul);
 	}
 };
 

libyul/ObjectParser.h

@@ -47,7 +47,7 @@ class ObjectParser: public langutil::ParserBase
 public:
 	explicit ObjectParser(
 		langutil::ErrorReporter& _errorReporter,
-		Dialect _dialect = Dialect::looseAssemblyForEVM()
+		std::shared_ptr<Dialect> _dialect
 	):
 		ParserBase(_errorReporter), m_dialect(std::move(_dialect)) {}
 
@@ -67,7 +67,7 @@ private:
 	YulString parseUniqueName(Object const* _containingObject);
 	void addNamedSubObject(Object& _container, YulString _name, std::shared_ptr<ObjectNode> _subObject);
 
-	Dialect m_dialect;
+	std::shared_ptr<Dialect> m_dialect;
 };
 
 }

libyul/backends/evm/EVMCodeTransform.cpp

@@ -97,7 +97,7 @@ CodeTransform::CodeTransform(
 	AsmAnalysisInfo& _analysisInfo,
 	Block const& _block,
 	bool _allowStackOpt,
-	bool _yul,
+	EVMDialect const& _dialect,
 	bool _evm15,
 	ExternalIdentifierAccess const& _identifierAccess,
 	bool _useNamedLabelsForFunctions,
@@ -106,8 +106,8 @@ CodeTransform::CodeTransform(
 ):
 	m_assembly(_assembly),
 	m_info(_analysisInfo),
+	m_dialect(_dialect),
 	m_allowStackOpt(_allowStackOpt),
-	m_yul(_yul),
 	m_evm15(_evm15),
 	m_useNamedLabelsForFunctions(_useNamedLabelsForFunctions),
 	m_identifierAccess(_identifierAccess),
@@ -267,35 +267,46 @@ void CodeTransform::operator()(FunctionCall const& _call)
 {
 	solAssert(m_scope, "");
 
-	m_assembly.setSourceLocation(_call.location);
+	if (BuiltinFunctionForEVM const* builtin = m_dialect.builtin(_call.functionName.name))
-	EVMAssembly::LabelID returnLabel(-1); // only used for evm 1.0
-	if (!m_evm15)
 	{
-		returnLabel = m_assembly.newLabelId();
+		builtin->generateCode(_call, m_assembly, [&]() {
-		m_assembly.appendLabelReference(returnLabel);
+			for (auto const& arg: _call.arguments | boost::adaptors::reversed)
-		m_stackAdjustment++;
+				visitExpression(arg);
+			m_assembly.setSourceLocation(_call.location);
+		});
 	}
-
-	Scope::Function* function = nullptr;
-	solAssert(m_scope->lookup(_call.functionName.name, Scope::NonconstVisitor(
-		[=](Scope::Variable&) { solAssert(false, "Expected function name."); },
-		[=](Scope::Label&) { solAssert(false, "Expected function name."); },
-		[&](Scope::Function& _function) { function = &_function; }
-	)), "Function name not found.");
-	solAssert(function, "");
-	solAssert(function->arguments.size() == _call.arguments.size(), "");
-	for (auto const& arg: _call.arguments | boost::adaptors::reversed)
-		visitExpression(arg);
-	m_assembly.setSourceLocation(_call.location);
-	if (m_evm15)
-		m_assembly.appendJumpsub(functionEntryID(_call.functionName.name, *function), function->arguments.size(), function->returns.size());
 	else
 	{
-		m_assembly.appendJumpTo(functionEntryID(_call.functionName.name, *function), function->returns.size() - function->arguments.size() - 1);
+		m_assembly.setSourceLocation(_call.location);
-		m_assembly.appendLabel(returnLabel);
+		EVMAssembly::LabelID returnLabel(-1); // only used for evm 1.0
-		m_stackAdjustment--;
+		if (!m_evm15)
+		{
+			returnLabel = m_assembly.newLabelId();
+			m_assembly.appendLabelReference(returnLabel);
+			m_stackAdjustment++;
+		}
+
+		Scope::Function* function = nullptr;
+		solAssert(m_scope->lookup(_call.functionName.name, Scope::NonconstVisitor(
+			[=](Scope::Variable&) { solAssert(false, "Expected function name."); },
+			[=](Scope::Label&) { solAssert(false, "Expected function name."); },
+			[&](Scope::Function& _function) { function = &_function; }
+		)), "Function name not found.");
+		solAssert(function, "");
+		solAssert(function->arguments.size() == _call.arguments.size(), "");
+		for (auto const& arg: _call.arguments | boost::adaptors::reversed)
+			visitExpression(arg);
+		m_assembly.setSourceLocation(_call.location);
+		if (m_evm15)
+			m_assembly.appendJumpsub(functionEntryID(_call.functionName.name, *function), function->arguments.size(), function->returns.size());
+		else
+		{
+			m_assembly.appendJumpTo(functionEntryID(_call.functionName.name, *function), function->returns.size() - function->arguments.size() - 1);
+			m_assembly.appendLabel(returnLabel);
+			m_stackAdjustment--;
+		}
+		checkStackHeight(&_call);
 	}
-	checkStackHeight(&_call);
 }
 
 void CodeTransform::operator()(FunctionalInstruction const& _instruction)

libyul/backends/evm/EVMCodeTransform.h

@@ -20,10 +20,10 @@
 
 #include <libyul/backends/evm/EVMAssembly.h>
 
+#include <libyul/backends/evm/EVMDialect.h>
 #include <libyul/optimiser/ASTWalker.h>
 #include <libyul/AsmDataForward.h>
 #include <libyul/AsmScope.h>
-#include <libyul/Dialect.h>
 
 #include <boost/variant.hpp>
 #include <boost/optional.hpp>
@@ -87,8 +87,8 @@ public:
 		AbstractAssembly& _assembly,
 		AsmAnalysisInfo& _analysisInfo,
 		Block const& _block,
+		EVMDialect const& _dialect,
 		bool _allowStackOpt = false,
-		Dialect const& _dialect,
 		bool _evm15 = false,
 		ExternalIdentifierAccess const& _identifierAccess = ExternalIdentifierAccess(),
 		bool _useNamedLabelsForFunctions = false
@@ -115,7 +115,7 @@ protected:
 		AsmAnalysisInfo& _analysisInfo,
 		Block const& _block,
 		bool _allowStackOpt,
-		Dialect const& _dialect,
+		EVMDialect const& _dialect,
 		bool _evm15,
 		ExternalIdentifierAccess const& _identifierAccess,
 		bool _useNamedLabelsForFunctions,
@@ -179,10 +179,10 @@ private:
 	AbstractAssembly& m_assembly;
 	AsmAnalysisInfo& m_info;
 	Scope* m_scope = nullptr;
+	EVMDialect const& m_dialect;
 	bool const m_allowStackOpt = true;
-	Dialect const& m_dialect;
+	bool const m_evm15 = false;
-	bool m_evm15 = false;
+	bool const m_useNamedLabelsForFunctions = false;
-	bool m_useNamedLabelsForFunctions = false;
 	ExternalIdentifierAccess m_identifierAccess;
 	/// Adjustment between the stack height as determined during the analysis phase
 	/// and the stack height in the assembly. This is caused by an initial stack being present

libyul/backends/evm/EVMDialect.cpp

@@ -0,0 +1,141 @@
+/*
+	This file is part of solidity.
+
+	solidity is free software: you can redistribute it and/or modify
+	it under the terms of the GNU General Public License as published by
+	the Free Software Foundation, either version 3 of the License, or
+	(at your option) any later version.
+
+	solidity is distributed in the hope that it will be useful,
+	but WITHOUT ANY WARRANTY; without even the implied warranty of
+	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+	GNU General Public License for more details.
+
+	You should have received a copy of the GNU General Public License
+	along with solidity.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * Yul dialects for EVM.
+ */
+
+#include <libyul/backends/evm/EVMDialect.h>
+
+#include <libyul/AsmAnalysisInfo.h>
+#include <libyul/AsmData.h>
+#include <libyul/Object.h>
+#include <libyul/backends/evm/AbstractAssembly.h>
+
+#include <liblangutil/Exceptions.h>
+
+#include <libyul/Exceptions.h>
+
+#include <boost/range/adaptor/reversed.hpp>
+
+using namespace std;
+using namespace dev;
+using namespace yul;
+using namespace dev::solidity;
+
+
+EVMDialect::EVMDialect(AsmFlavour _flavour, bool _objectAccess):
+	Dialect(_flavour), m_objectAccess(_objectAccess)
+{
+	// The EVM instructions will be moved to builtins at some point.
+	if (!m_objectAccess)
+		return;
+
+	addFunction("datasize", 1, 1, true, [this](
+		FunctionCall const& _call,
+		AbstractAssembly& _assembly,
+		std::function<void()>
+	) {
+		yulAssert(m_currentObject, "No object available.");
+		yulAssert(_call.arguments.size() == 1, "");
+		Expression const& arg = _call.arguments.front();
+		YulString dataName = boost::get<Literal>(arg).value;
+		if (m_currentObject->name == dataName)
+			_assembly.appendAssemblySize();
+		else
+			_assembly.appendDataSize(m_subIDs.at(dataName));
+	});
+	addFunction("dataoffset", 1, 1, true, [this](
+		FunctionCall const& _call,
+		AbstractAssembly& _assembly,
+		std::function<void()>
+	) {
+		yulAssert(m_currentObject, "No object available.");
+		yulAssert(_call.arguments.size() == 1, "");
+		Expression const& arg = _call.arguments.front();
+		YulString dataName = boost::get<Literal>(arg).value;
+		if (m_currentObject->name == dataName)
+			_assembly.appendConstant(0);
+		else
+			_assembly.appendDataOffset(m_subIDs.at(dataName));
+	});
+	addFunction("datacopy", 3, 0, false, [](
+		FunctionCall const&,
+		AbstractAssembly& _assembly,
+		std::function<void()> _visitArguments
+	) {
+		_visitArguments();
+		_assembly.appendInstruction(solidity::Instruction::CODECOPY);
+	});
+}
+
+BuiltinFunctionForEVM const* EVMDialect::builtin(YulString _name) const
+{
+	auto it = m_functions.find(_name);
+	if (it != m_functions.end())
+		return &it->second;
+	else
+		return nullptr;
+}
+
+shared_ptr<EVMDialect> EVMDialect::looseAssemblyForEVM()
+{
+	return make_shared<EVMDialect>(AsmFlavour::Loose, false);
+}
+
+shared_ptr<EVMDialect> EVMDialect::strictAssemblyForEVM()
+{
+	return make_shared<EVMDialect>(AsmFlavour::Strict, false);
+}
+
+shared_ptr<EVMDialect> EVMDialect::strictAssemblyForEVMObjects()
+{
+	return make_shared<EVMDialect>(AsmFlavour::Strict, true);
+}
+
+shared_ptr<yul::EVMDialect> EVMDialect::yulForEVM()
+{
+	return make_shared<EVMDialect>(AsmFlavour::Yul, false);
+}
+
+void EVMDialect::setSubIDs(map<YulString, AbstractAssembly::SubID> _subIDs)
+{
+	yulAssert(m_objectAccess, "Sub IDs set with dialect that does not support object access.");
+	m_subIDs = std::move(_subIDs);
+}
+
+void EVMDialect::setCurrentObject(Object const* _object)
+{
+	yulAssert(m_objectAccess, "Current object set with dialect that does not support object access.");
+	m_currentObject = _object;
+}
+
+void EVMDialect::addFunction(
+	string _name,
+	size_t _params,
+	size_t _returns,
+	bool _movable,
+	std::function<void(FunctionCall const&, AbstractAssembly&, std::function<void()>)> _generateCode
+)
+{
+	YulString name{std::move(_name)};
+	BuiltinFunctionForEVM& f = m_functions[name];
+	f.name = name;
+	f.parameters.resize(_params);
+	f.returns.resize(_returns);
+	f.movable = _movable;
+	f.generateCode = std::move(_generateCode);
+}

libyul/backends/evm/EVMDialect.h

@@ -0,0 +1,85 @@
+/*
+	This file is part of solidity.
+
+	solidity is free software: you can redistribute it and/or modify
+	it under the terms of the GNU General Public License as published by
+	the Free Software Foundation, either version 3 of the License, or
+	(at your option) any later version.
+
+	solidity is distributed in the hope that it will be useful,
+	but WITHOUT ANY WARRANTY; without even the implied warranty of
+	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+	GNU General Public License for more details.
+
+	You should have received a copy of the GNU General Public License
+	along with solidity.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * Yul dialects for EVM.
+ */
+
+#pragma once
+
+#include <libyul/Dialect.h>
+
+#include <libyul/backends/evm/AbstractAssembly.h>
+
+#include <map>
+
+namespace yul
+{
+
+class YulString;
+using Type = YulString;
+struct FunctionCall;
+struct Object;
+
+struct BuiltinFunctionForEVM: BuiltinFunction
+{
+	/// Function to generate code for the given function call and append it to the abstract
+	/// assembly. The third parameter is called to visit (and generate code for) the arguments
+	/// from right to left.
+	std::function<void(FunctionCall const&, AbstractAssembly&, std::function<void()>)> generateCode;
+};
+
+/**
+ * Yul dialect for EVM as a backend.
+ * The main difference is that the builtin functions take an AbstractAssembly for the
+ * code generation.
+ */
+struct EVMDialect: public Dialect
+{
+	EVMDialect(AsmFlavour _flavour, bool _objectAccess);
+
+	/// @returns the builtin function of the given name or a nullptr if it is not a builtin function.
+	BuiltinFunctionForEVM const* builtin(YulString _name) const override;
+
+	static std::shared_ptr<EVMDialect> looseAssemblyForEVM();
+	static std::shared_ptr<EVMDialect> strictAssemblyForEVM();
+	static std::shared_ptr<EVMDialect> strictAssemblyForEVMObjects();
+	static std::shared_ptr<EVMDialect> yulForEVM();
+
+	bool providesObjectAccess() const { return m_objectAccess; }
+
+	/// Sets the mapping of current sub assembly IDs. Used during code generation.
+	void setSubIDs(std::map<YulString, AbstractAssembly::SubID> _subIDs);
+	/// Sets the current object. Used during code generation.
+	void setCurrentObject(Object const* _object);
+
+private:
+	void addFunction(
+		std::string _name,
+		size_t _params,
+		size_t _returns,
+		bool _movable,
+		std::function<void(FunctionCall const&, AbstractAssembly&, std::function<void()>)> _generateCode
+	);
+
+	bool m_objectAccess;
+	Object const* m_currentObject = nullptr;
+	/// Mapping from named objects to abstract assembly sub IDs.
+	std::map<YulString, AbstractAssembly::SubID> m_subIDs;
+	std::map<YulString, BuiltinFunctionForEVM> m_functions;
+};
+
+}

libyul/backends/evm/EVMObjectCompiler.cpp

@@ -21,13 +21,15 @@
 #include <libyul/backends/evm/EVMObjectCompiler.h>
 
 #include <libyul/backends/evm/EVMCodeTransform.h>
+#include <libyul/backends/evm/EVMDialect.h>
+
 #include <libyul/Object.h>
 #include <libyul/Exceptions.h>
 
 using namespace yul;
 using namespace std;
 
-void EVMObjectCompiler::compile(Object& _object, AbstractAssembly& _assembly, Dialect const& _dialect, bool _evm15, bool _optimize)
+void EVMObjectCompiler::compile(Object& _object, AbstractAssembly& _assembly, EVMDialect& _dialect, bool _evm15, bool _optimize)
 {
 	EVMObjectCompiler compiler(_assembly, _dialect, _evm15);
 	compiler.run(_object, _optimize);
@@ -50,7 +52,13 @@ void EVMObjectCompiler::run(Object& _object, bool _optimize)
 			subIDs[data.name] = m_assembly.appendData(data.data);
 		}
 
+	if (m_dialect.providesObjectAccess())
+	{
+		m_dialect.setSubIDs(std::move(subIDs));
+		m_dialect.setCurrentObject(&_object);
+	}
+
 	yulAssert(_object.analysisInfo, "No analysis info.");
 	yulAssert(_object.code, "No code.");
-	CodeTransform{m_assembly, *_object.analysisInfo, *_object.code, _optimize, m_yul, m_evm15, _optimize}(*_object.code);
+	CodeTransform{m_assembly, *_object.analysisInfo, *_object.code, m_dialect, _optimize, m_evm15}(*_object.code);
 }

libyul/backends/evm/EVMObjectCompiler.h

@@ -23,21 +23,21 @@ namespace yul
 {
 struct Object;
 class AbstractAssembly;
-struct Dialect;
+struct EVMDialect;
 
 class EVMObjectCompiler
 {
 public:
-	static void compile(Object& _object, AbstractAssembly& _assembly, Dialect const& _dialect, bool _evm15, bool _optimize);
+	static void compile(Object& _object, AbstractAssembly& _assembly, EVMDialect& _dialect, bool _evm15, bool _optimize);
 private:
-	EVMObjectCompiler(AbstractAssembly& _assembly, Dialect const& _dialect, bool _evm15):
+	EVMObjectCompiler(AbstractAssembly& _assembly, EVMDialect& _dialect, bool _evm15):
 		m_assembly(_assembly), m_dialect(_dialect), m_evm15(_evm15)
 	{}
 
 	void run(Object& _object, bool _optimize);
 
 	AbstractAssembly& m_assembly;
-	Dialect const& m_dialect;
+	EVMDialect& m_dialect;
 	bool m_evm15 = false;
 };
 

test/libyul/Common.cpp

@@ -29,6 +29,7 @@
 #include <libyul/AsmParser.h>
 #include <libyul/AsmAnalysis.h>
 #include <libyul/AsmPrinter.h>
+#include <libyul/backends/evm/EVMDialect.h>
 
 #include <liblangutil/Scanner.h>
 #include <liblangutil/ErrorReporter.h>
@@ -54,7 +55,7 @@ void yul::test::printErrors(ErrorList const& _errors)
 
 pair<shared_ptr<Block>, shared_ptr<yul::AsmAnalysisInfo>> yul::test::parse(string const& _source, bool _yul)
 {
-	Dialect dialect = _yul ? yul::Dialect::yul() : yul::Dialect::strictAssemblyForEVM();
+	shared_ptr<Dialect> dialect = _yul ? yul::Dialect::yul() : yul::EVMDialect::strictAssemblyForEVM();
 	ErrorList errors;
 	ErrorReporter errorReporter(errors);
 	auto scanner = make_shared<Scanner>(CharStream(_source, ""));

test/libyul/Parser.cpp

@@ -49,7 +49,7 @@ namespace test
 namespace
 {
 
-bool parse(string const& _source, Dialect const& _dialect, ErrorReporter& errorReporter)
+bool parse(string const& _source, std::shared_ptr<Dialect> _dialect, ErrorReporter& errorReporter)
 {
 	try
 	{
@@ -74,7 +74,7 @@ bool parse(string const& _source, Dialect const& _dialect, ErrorReporter& errorR
 	return false;
 }
 
-boost::optional<Error> parseAndReturnFirstError(string const& _source, Dialect const& _dialect, bool _allowWarnings = true)
+boost::optional<Error> parseAndReturnFirstError(string const& _source, shared_ptr<Dialect> _dialect, bool _allowWarnings = true)
 {
 	ErrorList errors;
 	ErrorReporter errorReporter(errors);
@@ -99,12 +99,12 @@ boost::optional<Error> parseAndReturnFirstError(string const& _source, Dialect c
 	return {};
 }
 
-bool successParse(std::string const& _source, Dialect const& _dialect = Dialect::yul(), bool _allowWarnings = true)
+bool successParse(std::string const& _source, shared_ptr<Dialect> _dialect = Dialect::yul(), bool _allowWarnings = true)
 {
 	return !parseAndReturnFirstError(_source, _dialect, _allowWarnings);
 }
 
-Error expectError(std::string const& _source, Dialect const& _dialect = Dialect::yul(), bool _allowWarnings = false)
+Error expectError(std::string const& _source, shared_ptr<Dialect> _dialect = Dialect::yul(), bool _allowWarnings = false)
 {
 
 	auto error = parseAndReturnFirstError(_source, _dialect, _allowWarnings);
@@ -306,16 +306,17 @@ BOOST_AUTO_TEST_CASE(if_statement_invalid)
 
 BOOST_AUTO_TEST_CASE(builtins_parser)
 {
-	struct SimpleBuiltins: public Builtins
+	struct SimpleDialect: public Dialect
 	{
-		BuiltinFunction const* query(YulString _name) const override
+		SimpleDialect(): Dialect(AsmFlavour::Strict) {}
+		BuiltinFunction const* builtin(YulString _name) const override
 		{
 			return _name == YulString{"builtin"} ? &f : nullptr;
 		}
 		BuiltinFunction f;
 	};
 
-	Dialect dialect(AsmFlavour::Strict, make_shared<SimpleBuiltins>());
+	shared_ptr<Dialect> dialect = make_shared<SimpleDialect>();
 	CHECK_ERROR_DIALECT("{ let builtin := 6 }", ParserError, "Cannot use builtin function name \"builtin\" as identifier name.", dialect);
 	CHECK_ERROR_DIALECT("{ function builtin() {} }", ParserError, "Cannot use builtin function name \"builtin\" as identifier name.", dialect);
 	CHECK_ERROR_DIALECT("{ builtin := 6 }", ParserError, "Cannot assign to builtin function \"builtin\".", dialect);
@@ -323,16 +324,17 @@ BOOST_AUTO_TEST_CASE(builtins_parser)
 
 BOOST_AUTO_TEST_CASE(builtins_analysis)
 {
-	struct SimpleBuiltinsAnalysis: public Builtins
+	struct SimpleDialect: public Dialect
 	{
-		yul::BuiltinFunction const* query(YulString _name) const override
+		SimpleDialect(): Dialect(AsmFlavour::Strict) {}
+		BuiltinFunction const* builtin(YulString _name) const override
 		{
-			return _name == YulString("builtin") ? &m_builtin : nullptr;
+			return _name == YulString{"builtin"} ? &f : nullptr;
 		}
-		BuiltinFunction m_builtin{YulString{"builtin"}, vector<Type>(2), vector<Type>(3), false};
+		BuiltinFunction f{YulString{"builtin"}, vector<Type>(2), vector<Type>(3), false};
 	};
 
-	Dialect dialect(AsmFlavour::Strict, make_shared<SimpleBuiltinsAnalysis>());
+	shared_ptr<Dialect> dialect = make_shared<SimpleDialect>();
 	BOOST_CHECK(successParse("{ let a, b, c := builtin(1, 2) }", dialect));
 	CHECK_ERROR_DIALECT("{ let a, b, c := builtin(1) }", TypeError, "Function expects 2 arguments but got 1", dialect);
 	CHECK_ERROR_DIALECT("{ let a, b := builtin(1, 2) }", DeclarationError, "Variable count mismatch: 2 variables and 3 values.", dialect);

test/libyul/YulOptimizerTest.cpp

@@ -41,6 +41,7 @@
 #include <libyul/optimiser/RedundantAssignEliminator.h>
 #include <libyul/optimiser/StructuralSimplifier.h>
 #include <libyul/optimiser/Suite.h>
+#include <libyul/backends/evm/EVMDialect.h>
 #include <libyul/AsmPrinter.h>
 #include <libyul/AsmParser.h>
 #include <libyul/AsmAnalysis.h>
@@ -262,7 +263,7 @@ void YulOptimizerTest::printIndented(ostream& _stream, string const& _output, st
 
 bool YulOptimizerTest::parse(ostream& _stream, string const& _linePrefix, bool const _formatted)
 {
-	yul::Dialect dialect = m_yul ? yul::Dialect::yul() : yul::Dialect::strictAssemblyForEVM();
+	shared_ptr<yul::Dialect> dialect = m_yul ? yul::Dialect::yul() : yul::EVMDialect::strictAssemblyForEVM();
 	ErrorList errors;
 	ErrorReporter errorReporter(errors);
 	shared_ptr<Scanner> scanner = make_shared<Scanner>(CharStream(m_source, ""));

test/libyul/objectCompiler/datacopy.yul

@@ -0,0 +1,48 @@
+object "a" {
+  code {
+    datacopy(0, dataoffset("sub"), datasize("sub"))
+    return(0, datasize("sub"))
+  }
+  object "sub" {
+    code {
+      sstore(0, dataoffset("sub"))
+      mstore(0, datasize("data1"))
+    }
+    data "data1" "Hello, World!"
+  }
+}
+// ----
+// Assembly:
+//     /* "source":26:73   */
+//   dataSize(sub_0)
+//   dataOffset(sub_0)
+//     /* "source":35:36   */
+//   0x00
+//     /* "source":26:73   */
+//   codecopy
+//     /* "source":78:104   */
+//   dataSize(sub_0)
+//     /* "source":85:86   */
+//   0x00
+//     /* "source":78:104   */
+//   return
+// stop
+//
+// sub_0: assembly {
+//         /* "source":143:171   */
+//       0x00
+//         /* "source":150:151   */
+//       0x00
+//         /* "source":143:171   */
+//       sstore
+//         /* "source":178:206   */
+//       0x0d
+//         /* "source":185:186   */
+//       0x00
+//         /* "source":178:206   */
+//       mstore
+//     stop
+//     data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// }
+// Bytecode: 600b600d600039600b6000f3fe6000600055600d600052fe
+// Opcodes: PUSH1 0xB PUSH1 0xD PUSH1 0x0 CODECOPY PUSH1 0xB PUSH1 0x0 RETURN INVALID PUSH1 0x0 PUSH1 0x0 SSTORE PUSH1 0xD PUSH1 0x0 MSTORE INVALID

test/libyul/objectCompiler/dataoffset_code.yul

@@ -0,0 +1,29 @@
+object "a" {
+  code { sstore(0, dataoffset("sub")) }
+  object "sub" {
+    code { sstore(0, 8) }
+    data "data1" "Hello, World!"
+  }
+}
+// ----
+// Assembly:
+//     /* "source":22:50   */
+//   dataOffset(sub_0)
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:50   */
+//   sstore
+// stop
+//
+// sub_0: assembly {
+//         /* "source":91:92   */
+//       0x08
+//         /* "source":88:89   */
+//       0x00
+//         /* "source":81:93   */
+//       sstore
+//     stop
+//     data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// }
+// Bytecode: 6006600055fe6008600055fe
+// Opcodes: PUSH1 0x6 PUSH1 0x0 SSTORE INVALID PUSH1 0x8 PUSH1 0x0 SSTORE INVALID

test/libyul/objectCompiler/dataoffset_data.yul

@@ -0,0 +1,16 @@
+object "a" {
+  code { sstore(0, dataoffset("data1")) }
+  data "data1" "Hello, World!"
+}
+// ----
+// Assembly:
+//     /* "source":22:52   */
+//   data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:52   */
+//   sstore
+// stop
+// data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// Bytecode: 6006600055fe48656c6c6f2c20576f726c6421
+// Opcodes: PUSH1 0x6 PUSH1 0x0 SSTORE INVALID 0x48 PUSH6 0x6C6C6F2C2057 PUSH16 0x726C6421000000000000000000000000

test/libyul/objectCompiler/dataoffset_self.yul

@@ -0,0 +1,16 @@
+object "a" {
+  code { sstore(0, dataoffset("a")) }
+  data "data1" "Hello, World!"
+}
+// ----
+// Assembly:
+//     /* "source":22:48   */
+//   0x00
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:48   */
+//   sstore
+// stop
+// data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// Bytecode: 6000600055fe
+// Opcodes: PUSH1 0x0 PUSH1 0x0 SSTORE INVALID

test/libyul/objectCompiler/datasize_code.yul

@@ -0,0 +1,29 @@
+object "a" {
+  code { sstore(0, datasize("sub")) }
+  object "sub" {
+    code { sstore(0, 8) }
+    data "data1" "Hello, World!"
+  }
+}
+// ----
+// Assembly:
+//     /* "source":22:48   */
+//   dataSize(sub_0)
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:48   */
+//   sstore
+// stop
+//
+// sub_0: assembly {
+//         /* "source":89:90   */
+//       0x08
+//         /* "source":86:87   */
+//       0x00
+//         /* "source":79:91   */
+//       sstore
+//     stop
+//     data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// }
+// Bytecode: 6006600055fe
+// Opcodes: PUSH1 0x6 PUSH1 0x0 SSTORE INVALID

test/libyul/objectCompiler/datasize_data.yul

@@ -0,0 +1,16 @@
+object "a" {
+  code { sstore(0, datasize("data1")) }
+  data "data1" "Hello, World!"
+}
+// ----
+// Assembly:
+//     /* "source":22:50   */
+//   0x0d
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:50   */
+//   sstore
+// stop
+// data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// Bytecode: 600d600055fe
+// Opcodes: PUSH1 0xD PUSH1 0x0 SSTORE INVALID

test/libyul/objectCompiler/datasize_self.yul

@@ -0,0 +1,16 @@
+object "a" {
+  code { sstore(0, datasize("a")) }
+  data "data1" "Hello, World!"
+}
+// ----
+// Assembly:
+//     /* "source":22:46   */
+//   bytecodeSize
+//     /* "source":29:30   */
+//   0x00
+//     /* "source":22:46   */
+//   sstore
+// stop
+// data_acaf3289d7b601cbd114fb36c4d29c85bbfd5e133f14cb355c3fd8d99367964f 48656c6c6f2c20576f726c6421
+// Bytecode: 6006600055fe
+// Opcodes: PUSH1 0x6 PUSH1 0x0 SSTORE INVALID

test/tools/yulopti.cpp

@@ -49,6 +49,8 @@
 #include <libyul/optimiser/StructuralSimplifier.h>
 #include <libyul/optimiser/VarDeclPropagator.h>
 
+#include <libyul/backends/evm/EVMDialect.h>
+
 #include <libdevcore/JSON.h>
 
 #include <boost/program_options.hpp>
@@ -83,7 +85,7 @@ public:
 	{
 		ErrorReporter errorReporter(m_errors);
 		shared_ptr<Scanner> scanner = make_shared<Scanner>(CharStream(_input, ""));
-		m_ast = yul::Parser(errorReporter, yul::Dialect::strictAssemblyForEVM()).parse(scanner, false);
+		m_ast = yul::Parser(errorReporter, yul::EVMDialect::strictAssemblyForEVM()).parse(scanner, false);
 		if (!m_ast || !errorReporter.errors().empty())
 		{
 			cout << "Error parsing source." << endl;
@@ -96,7 +98,7 @@ public:
 			errorReporter,
 			EVMVersion::byzantium(),
 			langutil::Error::Type::SyntaxError,
-			Dialect::strictAssemblyForEVM()
+			EVMDialect::strictAssemblyForEVM()
 		);
 		if (!analyzer.analyze(*m_ast) || !errorReporter.errors().empty())
 		{