Add LiteralKind to Dialect.literalArguments

Co-authored-by: Daniel Kirchner <daniel@ekpyron.org>

libyul/AsmAnalysis.cpp

@@ -272,14 +272,14 @@ vector<YulString> AsmAnalyzer::operator()(FunctionCall const& _funCall)
 	auto watcher = m_errorReporter.errorWatcher();
 	vector<YulString> const* parameterTypes = nullptr;
 	vector<YulString> const* returnTypes = nullptr;
-	vector<bool> const* needsLiteralArguments = nullptr;
+	vector<optional<LiteralKind>> const* literalArguments = nullptr;
 
 	if (BuiltinFunction const* f = m_dialect.builtin(_funCall.functionName.name))
 	{
 		parameterTypes = &f->parameters;
 		returnTypes = &f->returns;
-		if (f->literalArguments)
+		if (!f->literalArguments.empty())
-			needsLiteralArguments = &f->literalArguments.value();
+			literalArguments = &f->literalArguments;
 
 		validateInstructions(_funCall);
 	}
@@ -318,7 +318,7 @@ vector<YulString> AsmAnalyzer::operator()(FunctionCall const& _funCall)
 	for (size_t i = _funCall.arguments.size(); i > 0; i--)
 	{
 		Expression const& arg = _funCall.arguments[i - 1];
-		bool isLiteralArgument = needsLiteralArguments && (*needsLiteralArguments)[i - 1];
+		bool isLiteralArgument = literalArguments && (*literalArguments)[i - 1].has_value();
 		bool isStringLiteral = holds_alternative<Literal>(arg) && get<Literal>(arg).kind == LiteralKind::String;
 
 		if (isLiteralArgument && isStringLiteral)

libyul/Dialect.h

@@ -48,8 +48,13 @@ struct BuiltinFunction
 	ControlFlowSideEffects controlFlowSideEffects;
 	/// If true, this is the msize instruction.
 	bool isMSize = false;
-	/// If set, same length as the arguments, if true at index i, the i'th argument has to be a literal which means it can't be moved to variables.
+	/// Must be empty or the same length as the arguments.
-	std::optional<std::vector<bool>> literalArguments;
+	/// If set at index i, the i'th argument has to be a literal which means it can't be moved to variables.
+	std::vector<std::optional<LiteralKind>> literalArguments{};
+	std::optional<LiteralKind> literalArgument(size_t i) const
+	{
+		return literalArguments.empty() ? std::nullopt : literalArguments.at(i);
+	}
 };
 
 struct Dialect: boost::noncopyable

libyul/backends/evm/EVMDialect.cpp

@@ -70,7 +70,7 @@ pair<YulString, BuiltinFunctionForEVM> createEVMFunction(
 	f.controlFlowSideEffects.terminates = evmasm::SemanticInformation::terminatesControlFlow(_instruction);
 	f.controlFlowSideEffects.reverts = evmasm::SemanticInformation::reverts(_instruction);
 	f.isMSize = _instruction == evmasm::Instruction::MSIZE;
-	f.literalArguments.reset();
+	f.literalArguments.clear();
 	f.instruction = _instruction;
 	f.generateCode = [_instruction](
 		FunctionCall const& _call,
@@ -90,7 +90,7 @@ pair<YulString, BuiltinFunctionForEVM> createFunction(
 	size_t _params,
 	size_t _returns,
 	SideEffects _sideEffects,
-	vector<bool> _literalArguments,
+	vector<optional<LiteralKind>> _literalArguments,
 	std::function<void(FunctionCall const&, AbstractAssembly&, BuiltinContext&, std::function<void(Expression const&)>)> _generateCode
 )
 {
@@ -102,10 +102,7 @@ pair<YulString, BuiltinFunctionForEVM> createFunction(
 	f.parameters.resize(_params);
 	f.returns.resize(_returns);
 	f.sideEffects = std::move(_sideEffects);
-	if (!_literalArguments.empty())
+	f.literalArguments = std::move(_literalArguments);
-		f.literalArguments = std::move(_literalArguments);
-	else
-		f.literalArguments.reset();
 	f.isMSize = false;
 	f.instruction = {};
 	f.generateCode = std::move(_generateCode);
@@ -135,7 +132,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 
 	if (_objectAccess)
 	{
-		builtins.emplace(createFunction("linkersymbol", 1, 1, SideEffects{}, {true}, [](
+		builtins.emplace(createFunction("linkersymbol", 1, 1, SideEffects{}, {LiteralKind::String}, [](
 			FunctionCall const& _call,
 			AbstractAssembly& _assembly,
 			BuiltinContext&,
@@ -145,7 +142,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 			Expression const& arg = _call.arguments.front();
 			_assembly.appendLinkerSymbol(std::get<Literal>(arg).value.str());
 		}));
-		builtins.emplace(createFunction("datasize", 1, 1, SideEffects{}, {true}, [](
+		builtins.emplace(createFunction("datasize", 1, 1, SideEffects{}, {LiteralKind::String}, [](
 			FunctionCall const& _call,
 			AbstractAssembly& _assembly,
 			BuiltinContext& _context,
@@ -167,7 +164,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 				_assembly.appendDataSize(subIdPath);
 			}
 		}));
-		builtins.emplace(createFunction("dataoffset", 1, 1, SideEffects{}, {true}, [](
+		builtins.emplace(createFunction("dataoffset", 1, 1, SideEffects{}, {LiteralKind::String}, [](
 			FunctionCall const& _call,
 			AbstractAssembly& _assembly,
 			BuiltinContext& _context,
@@ -210,7 +207,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 			2,
 			0,
 			SideEffects{false, false, false, false, true},
-			{true, false},
+			{LiteralKind::String, std::nullopt},
 			[](
 				FunctionCall const& _call,
 				AbstractAssembly& _assembly,
@@ -230,7 +227,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 			1,
 			1,
 			SideEffects{},
-			{true},
+			{LiteralKind::String},
 			[](
 				FunctionCall const& _call,
 				AbstractAssembly& _assembly,

libyul/backends/evm/NoOutputAssembly.cpp

@@ -170,7 +170,7 @@ NoOutputEVMDialect::NoOutputEVMDialect(EVMDialect const& _copyFrom):
 			for (size_t j = 0; j < _call.arguments.size(); j++)
 			{
 				size_t const i = _call.arguments.size() - j - 1;
-				if (!(fun.second.literalArguments && (*fun.second.literalArguments)[i]))
+				if (!fun.second.literalArgument(i))
 				{
 					_visitExpression(_call.arguments[i]);
 					visited++;

libyul/backends/wasm/WasmCodeTransform.cpp

@@ -140,22 +140,20 @@ wasm::Expression WasmCodeTransform::operator()(FunctionCall const& _call)
 				m_functionsToImport[builtin->name] = std::move(imp);
 			}
 		}
-		else if (builtin->literalArguments && contains(builtin->literalArguments.value(), true))
-		{
-			vector<wasm::Expression> literals;
-			for (size_t i = 0; i < _call.arguments.size(); i++)
-				if (builtin->literalArguments.value()[i])
-					literals.emplace_back(wasm::StringLiteral{std::get<Literal>(_call.arguments[i]).value.str()});
-				else
-					literals.emplace_back(visitReturnByValue(_call.arguments[i]));
-
-			return wasm::BuiltinCall{_call.functionName.name.str(), std::move(literals)};
-		}
 		else
-			return wasm::BuiltinCall{
+		{
-				_call.functionName.name.str(),
+			vector<wasm::Expression> arguments;
-				visit(_call.arguments)
+			for (size_t i = 0; i < _call.arguments.size(); i++)
-			};
+				if (builtin->literalArgument(i))
+				{
+					yulAssert(builtin->literalArgument(i) == LiteralKind::String, "");
+					arguments.emplace_back(wasm::StringLiteral{std::get<Literal>(_call.arguments[i]).value.str()});
+				}
+				else
+					arguments.emplace_back(visitReturnByValue(_call.arguments[i]));
+
+			return wasm::BuiltinCall{_call.functionName.name.str(), std::move(arguments)};
+		}
 	}
 
 	// If this function returns multiple values, then the first one will

libyul/backends/wasm/WasmDialect.cpp

@@ -21,6 +21,7 @@
 
 #include <libyul/backends/wasm/WasmDialect.h>
 
+#include <libyul/AsmData.h>
 #include <libyul/Exceptions.h>
 
 using namespace std;
@@ -119,8 +120,8 @@ WasmDialect::WasmDialect()
 	m_functions["unreachable"_yulstring].controlFlowSideEffects.terminates = true;
 	m_functions["unreachable"_yulstring].controlFlowSideEffects.reverts = true;
 
-	addFunction("datasize", {i64}, {i64}, true, {true});
+	addFunction("datasize", {i64}, {i64}, true, {LiteralKind::String});
-	addFunction("dataoffset", {i64}, {i64}, true, {true});
+	addFunction("dataoffset", {i64}, {i64}, true, {LiteralKind::String});
 
 	addEthereumExternals();
 }
@@ -221,7 +222,7 @@ void WasmDialect::addEthereumExternals()
 		f.controlFlowSideEffects = ext.controlFlowSideEffects;
 		f.isMSize = false;
 		f.sideEffects.invalidatesStorage = (ext.name == "storageStore");
-		f.literalArguments.reset();
+		f.literalArguments.clear();
 	}
 }
 
@@ -230,7 +231,7 @@ void WasmDialect::addFunction(
 	vector<YulString> _params,
 	vector<YulString> _returns,
 	bool _movable,
-	std::vector<bool> _literalArguments
+	vector<optional<LiteralKind>> _literalArguments
 )
 {
 	YulString name{move(_name)};
@@ -241,8 +242,5 @@ void WasmDialect::addFunction(
 	f.returns = std::move(_returns);
 	f.sideEffects = _movable ? SideEffects{} : SideEffects::worst();
 	f.isMSize = false;
-	if (!_literalArguments.empty())
+	f.literalArguments = std::move(_literalArguments);
-		f.literalArguments = std::move(_literalArguments);
-	else
-		f.literalArguments.reset();
 }

libyul/backends/wasm/WasmDialect.h

@@ -62,7 +62,7 @@ private:
 		std::vector<YulString> _params,
 		std::vector<YulString> _returns,
 		bool _movable = true,
-		std::vector<bool> _literalArguments = std::vector<bool>{}
+		std::vector<std::optional<LiteralKind>> _literalArguments = std::vector<std::optional<LiteralKind>>{}
 	);
 
 	std::map<YulString, BuiltinFunction> m_functions;

libyul/backends/wasm/WordSizeTransform.cpp

@@ -42,16 +42,16 @@ void WordSizeTransform::operator()(FunctionDefinition& _fd)
 
 void WordSizeTransform::operator()(FunctionCall& _fc)
 {
-	vector<bool> const* literalArguments = nullptr;
+	vector<optional<LiteralKind>> const* literalArguments = nullptr;
 
 	if (BuiltinFunction const* fun = m_inputDialect.builtin(_fc.functionName.name))
-		if (fun->literalArguments)
+		if (!fun->literalArguments.empty())
-			literalArguments = &fun->literalArguments.value();
+			literalArguments = &fun->literalArguments;
 
 	vector<Expression> newArgs;
 
 	for (size_t i = 0; i < _fc.arguments.size(); i++)
-		if (!literalArguments || !(*literalArguments)[i])
+		if (!literalArguments || !(*literalArguments)[i].has_value())
 			newArgs += expandValueToVector(_fc.arguments[i]);
 		else
 		{
@@ -109,7 +109,8 @@ void WordSizeTransform::operator()(Block& _block)
 					if (BuiltinFunction const* f = m_inputDialect.builtin(std::get<FunctionCall>(*varDecl.value).functionName.name))
 						if (f->name == "datasize"_yulstring || f->name == "dataoffset"_yulstring)
 						{
-							yulAssert(f->literalArguments && f->literalArguments.value()[0], "");
+							yulAssert(f->literalArguments.size() == 1, "");
+							yulAssert(f->literalArguments.at(0) == LiteralKind::String, "");
 							yulAssert(varDecl.variables.size() == 1, "");
 							auto newLhs = generateU64IdentifierNames(varDecl.variables[0].name);
 							vector<Statement> ret;
@@ -169,7 +170,8 @@ void WordSizeTransform::operator()(Block& _block)
 					if (BuiltinFunction const* f = m_inputDialect.builtin(std::get<FunctionCall>(*assignment.value).functionName.name))
 						if (f->name == "datasize"_yulstring || f->name == "dataoffset"_yulstring)
 						{
-							yulAssert(f->literalArguments && f->literalArguments.value()[0], "");
+							yulAssert(f->literalArguments.size() == 1, "");
+							yulAssert(f->literalArguments[0] == LiteralKind::String, "");
 							yulAssert(assignment.variableNames.size() == 1, "");
 							auto newLhs = generateU64IdentifierNames(assignment.variableNames[0].name);
 							vector<Statement> ret;

libyul/optimiser/CommonSubexpressionEliminator.cpp

@@ -67,7 +67,7 @@ void CommonSubexpressionEliminator::visit(Expression& _e)
 				// We should not modify function arguments that have to be literals
 				// Note that replacing the function call entirely is fine,
 				// if the function call is movable.
-				if (!builtin->literalArguments || !builtin->literalArguments.value()[i - 1])
+				if (!builtin->literalArgument(i - 1))
 					visit(funCall.arguments[i - 1]);
 
 			descend = false;

libyul/optimiser/ExpressionSplitter.cpp

@@ -46,14 +46,10 @@ void ExpressionSplitter::run(OptimiserStepContext& _context, Block& _ast)
 
 void ExpressionSplitter::operator()(FunctionCall& _funCall)
 {
-	vector<bool> const* literalArgs = nullptr;
+	BuiltinFunction const* builtin = m_dialect.builtin(_funCall.functionName.name);
-
-	if (BuiltinFunction const* builtin = m_dialect.builtin(_funCall.functionName.name))
-		if (builtin->literalArguments)
-			literalArgs = &builtin->literalArguments.value();
 
 	for (size_t i = _funCall.arguments.size(); i > 0; i--)
-		if (!literalArgs || !(*literalArgs)[i - 1])
+		if (!builtin || !builtin->literalArgument(i - 1))
 			outlineExpression(_funCall.arguments[i - 1]);
 }