Merge pull request #9142 from ethereum/wasm-support-for-i32-in-generated-code

Add support for types in Wasm AST and code generation

libyul/backends/wasm/BinaryTransform.cpp

@@ -22,9 +22,11 @@
 
 #include <libyul/Exceptions.h>
 #include <libsolutil/CommonData.h>
+#include <libsolutil/Visitor.h>
 
 #include <boost/range/adaptor/reversed.hpp>
 #include <boost/range/adaptor/map.hpp>
+#include <boost/range/adaptor/transformed.hpp>
 
 using namespace std;
 using namespace solidity;
@@ -82,6 +84,16 @@ bytes toBytes(ValueType _vt)
 	return toBytes(uint8_t(_vt));
 }
 
+ValueType toValueType(wasm::Type _type)
+{
+	if (_type == wasm::Type::i32)
+		return ValueType::I32;
+	else if (_type == wasm::Type::i64)
+		return ValueType::I64;
+	else
+		yulAssert(false, "Invalid wasm variable type");
+}
+
 enum class Export: uint8_t
 {
 	Function = 0x0,
@@ -131,6 +143,16 @@ bytes toBytes(Opcode _o)
 	return toBytes(uint8_t(_o));
 }
 
+Opcode constOpcodeFor(ValueType _type)
+{
+	if (_type == ValueType::I32)
+		return Opcode::I32Const;
+	else if (_type == ValueType::I64)
+		return Opcode::I64Const;
+	else
+		yulAssert(false, "Values of this type cannot be used with const opcode");
+}
+
 static map<string, uint8_t> const builtins = {
 	{"i32.load", 0x28},
 	{"i64.load", 0x29},
@@ -249,6 +271,34 @@ bytes makeSection(Section _section, bytes _data)
 	return toBytes(_section) + prefixSize(move(_data));
 }
 
+/// This is a kind of run-length-encoding of local types.
+vector<pair<size_t, ValueType>> groupLocalVariables(vector<VariableDeclaration> _localVariables)
+{
+	vector<pair<size_t, ValueType>> localEntries;
+
+	size_t entrySize = 0;
+	ValueType entryType = ValueType::I32; // Any type would work here
+	for (VariableDeclaration const& localVariable: _localVariables)
+	{
+		ValueType variableType = toValueType(localVariable.type);
+
+		if (variableType != entryType)
+		{
+			if (entrySize > 0)
+				localEntries.emplace_back(entrySize, entryType);
+
+			entryType = variableType;
+			entrySize = 0;
+		}
+
+		++entrySize;
+	}
+	if (entrySize > 0)
+		localEntries.emplace_back(entrySize, entryType);
+
+	return localEntries;
+}
+
 }
 
 bytes BinaryTransform::run(Module const& _module)
@@ -297,7 +347,10 @@ bytes BinaryTransform::run(Module const& _module)
 
 bytes BinaryTransform::operator()(Literal const& _literal)
 {
-	return toBytes(Opcode::I64Const) + lebEncodeSigned(_literal.value);
+	return std::visit(GenericVisitor{
+		[&](uint32_t _value) -> bytes { return toBytes(Opcode::I32Const) + lebEncodeSigned(static_cast<int32_t>(_value)); },
+		[&](uint64_t _value) -> bytes { return toBytes(Opcode::I64Const) + lebEncodeSigned(static_cast<int64_t>(_value)); },
+	}, _literal.value);
 }
 
 bytes BinaryTransform::operator()(StringLiteral const&)
@@ -443,21 +496,18 @@ bytes BinaryTransform::operator()(FunctionDefinition const& _function)
 {
 	bytes ret;
 
-	// This is a kind of run-length-encoding of local types. Has to be adapted once
+	vector<pair<size_t, ValueType>> localEntries = groupLocalVariables(_function.locals);
-	// we have locals of different types.
+	ret += lebEncode(localEntries.size());
-	if (_function.locals.size() == 0)
+	for (pair<size_t, ValueType> const& entry: localEntries)
-		ret += lebEncode(0); // number of locals groups
-	else
 	{
-		ret += lebEncode(1); // number of locals groups
+		ret += lebEncode(entry.first);
-		ret += lebEncode(_function.locals.size());
+		ret += toBytes(entry.second);
-		ret += toBytes(ValueType::I64);
 	}
 
 	m_locals.clear();
 	size_t varIdx = 0;
-	for (size_t i = 0; i < _function.parameterNames.size(); ++i)
+	for (size_t i = 0; i < _function.parameters.size(); ++i)
-		m_locals[_function.parameterNames[i]] = varIdx++;
+		m_locals[_function.parameters[i].name] = varIdx++;
 	for (size_t i = 0; i < _function.locals.size(); ++i)
 		m_locals[_function.locals[i].variableName] = varIdx++;
 
@@ -475,38 +525,39 @@ BinaryTransform::Type BinaryTransform::typeOf(FunctionImport const& _import)
 {
 	return {
 		encodeTypes(_import.paramTypes),
-		encodeTypes(_import.returnType ? vector<string>(1, *_import.returnType) : vector<string>())
+		encodeTypes(_import.returnType ? vector<wasm::Type>(1, *_import.returnType) : vector<wasm::Type>())
 	};
 }
 
 BinaryTransform::Type BinaryTransform::typeOf(FunctionDefinition const& _funDef)
 {
-
 	return {
-		encodeTypes(vector<string>(_funDef.parameterNames.size(), "i64")),
+		encodeTypes(_funDef.parameters),
-		encodeTypes(vector<string>(_funDef.returns ? 1 : 0, "i64"))
+		encodeTypes(_funDef.returnType ? vector<wasm::Type>(1, *_funDef.returnType) : vector<wasm::Type>())
 	};
 }
 
-uint8_t BinaryTransform::encodeType(string const& _typeName)
+uint8_t BinaryTransform::encodeType(wasm::Type _type)
 {
-	if (_typeName == "i32")
+	return uint8_t(toValueType(_type));
-		return uint8_t(ValueType::I32);
-	else if (_typeName == "i64")
-		return uint8_t(ValueType::I64);
-	else
-		yulAssert(false, "");
-	return 0;
 }
 
-vector<uint8_t> BinaryTransform::encodeTypes(vector<string> const& _typeNames)
+vector<uint8_t> BinaryTransform::encodeTypes(vector<wasm::Type> const& _types)
 {
 	vector<uint8_t> result;
-	for (auto const& t: _typeNames)
+	for (wasm::Type t: _types)
 		result.emplace_back(encodeType(t));
 	return result;
 }
 
+vector<uint8_t> BinaryTransform::encodeTypes(wasm::TypedNameList const& _typedNameList)
+{
+	vector<uint8_t> result;
+	for (TypedName const& typedName: _typedNameList)
+		result.emplace_back(encodeType(typedName.type));
+	return result;
+}
+
 map<BinaryTransform::Type, vector<string>> BinaryTransform::typeToFunctionMap(
 	vector<wasm::FunctionImport> const& _imports,
 	vector<wasm::FunctionDefinition> const& _functions
@@ -612,13 +663,16 @@ bytes BinaryTransform::memorySection()
 bytes BinaryTransform::globalSection(vector<wasm::GlobalVariableDeclaration> const& _globals)
 {
 	bytes result = lebEncode(_globals.size());
-	for (size_t i = 0; i < _globals.size(); ++i)
+	for (wasm::GlobalVariableDeclaration const& global: _globals)
+	{
+		ValueType globalType = toValueType(global.type);
 		result +=
-			toBytes(ValueType::I64) +
+			toBytes(globalType) +
 			lebEncode(static_cast<uint8_t>(Mutability::Var)) +
-			toBytes(Opcode::I64Const) +
+			toBytes(constOpcodeFor(globalType)) +
 			lebEncodeSigned(0) +
 			toBytes(Opcode::End);
+	}
 
 	return makeSection(Section::GLOBAL, move(result));
 }

libyul/backends/wasm/BinaryTransform.h

@@ -71,8 +71,9 @@ private:
 	static Type typeOf(wasm::FunctionImport const& _import);
 	static Type typeOf(wasm::FunctionDefinition const& _funDef);
 
-	static uint8_t encodeType(std::string const& _typeName);
+	static uint8_t encodeType(wasm::Type _type);
-	static std::vector<uint8_t> encodeTypes(std::vector<std::string> const& _typeNames);
+	static std::vector<uint8_t> encodeTypes(std::vector<wasm::Type> const& _types);
+	static std::vector<uint8_t> encodeTypes(wasm::TypedNameList const& _typedNameList);
 
 	static std::map<Type, std::vector<std::string>> typeToFunctionMap(
 		std::vector<wasm::FunctionImport> const& _imports,

libyul/backends/wasm/TextTransform.cpp

@@ -20,7 +20,10 @@
 
 #include <libyul/backends/wasm/TextTransform.h>
 
+#include <libyul/Exceptions.h>
+
 #include <libsolutil/StringUtils.h>
+#include <libsolutil/Visitor.h>
 
 #include <boost/algorithm/string/join.hpp>
 #include <boost/algorithm/string/replace.hpp>
@@ -44,9 +47,9 @@ string TextTransform::run(wasm::Module const& _module)
 	{
 		ret += "    (import \"" + imp.module + "\" \"" + imp.externalName + "\" (func $" + imp.internalName;
 		if (!imp.paramTypes.empty())
-			ret += " (param" + joinHumanReadablePrefixed(imp.paramTypes, " ", " ") + ")";
+			ret += " (param" + joinHumanReadablePrefixed(imp.paramTypes | boost::adaptors::transformed(encodeType), " ", " ") + ")";
 		if (imp.returnType)
-			ret += " (result " + *imp.returnType + ")";
+			ret += " (result " + encodeType(*imp.returnType) + ")";
 		ret += "))\n";
 	}
 
@@ -56,7 +59,7 @@ string TextTransform::run(wasm::Module const& _module)
 	ret += "    (export \"main\" (func $main))\n";
 
 	for (auto const& g: _module.globals)
-		ret += "    (global $" + g.variableName + " (mut i64) (i64.const 0))\n";
+		ret += "    (global $" + g.variableName + " (mut " + encodeType(g.type) + ") (" + encodeType(g.type) + ".const 0))\n";
 	ret += "\n";
 	for (auto const& f: _module.functions)
 		ret += transform(f) + "\n";
@@ -65,7 +68,10 @@ string TextTransform::run(wasm::Module const& _module)
 
 string TextTransform::operator()(wasm::Literal const& _literal)
 {
-	return "(i64.const " + to_string(_literal.value) + ")";
+	return std::visit(GenericVisitor{
+		[&](uint32_t _value) -> string { return "(i32.const " + to_string(_value) + ")"; },
+		[&](uint64_t _value) -> string { return "(i64.const " + to_string(_value) + ")"; },
+	}, _literal.value);
 }
 
 string TextTransform::operator()(wasm::StringLiteral const& _literal)
@@ -162,12 +168,12 @@ string TextTransform::indented(string const& _in)
 string TextTransform::transform(wasm::FunctionDefinition const& _function)
 {
 	string ret = "(func $" + _function.name + "\n";
-	for (auto const& param: _function.parameterNames)
+	for (auto const& param: _function.parameters)
-		ret += "    (param $" + param + " i64)\n";
+		ret += "    (param $" + param.name + " " + encodeType(param.type) + ")\n";
-	if (_function.returns)
+	if (_function.returnType.has_value())
-		ret += "    (result i64)\n";
+		ret += "    (result " + encodeType(_function.returnType.value()) + ")\n";
 	for (auto const& local: _function.locals)
-		ret += "    (local $" + local.variableName + " i64)\n";
+		ret += "    (local $" + local.variableName + " " + encodeType(local.type) + ")\n";
 	ret += indented(joinTransformed(_function.body, '\n'));
 	if (ret.back() != '\n')
 		ret += '\n';
@@ -193,3 +199,13 @@ string TextTransform::joinTransformed(vector<wasm::Expression> const& _expressio
 	}
 	return ret;
 }
+
+string TextTransform::encodeType(wasm::Type _type)
+{
+	if (_type == wasm::Type::i32)
+		return "i32";
+	else if (_type == wasm::Type::i64)
+		return "i64";
+	else
+		yulAssert(false, "Invalid wasm type");
+}

libyul/backends/wasm/TextTransform.h

@@ -63,6 +63,8 @@ private:
 		std::vector<wasm::Expression> const& _expressions,
 		char _separator = ' '
 	);
+
+	static std::string encodeType(wasm::Type _type);
 };
 
 }

libyul/backends/wasm/WasmAST.h

@@ -30,6 +30,15 @@
 namespace solidity::yul::wasm
 {
 
+enum class Type
+{
+	i32,
+	i64,
+};
+
+struct TypedName { std::string name; Type type; };
+using TypedNameList = std::vector<TypedName>;
+
 struct Literal;
 struct StringLiteral;
 struct LocalVariable;
@@ -50,7 +59,7 @@ using Expression = std::variant<
 	Block, If, Loop, Branch, BranchIf, Return
 >;
 
-struct Literal { uint64_t value; };
+struct Literal { std::variant<uint32_t, uint64_t> value; };
 struct StringLiteral { std::string value; };
 struct LocalVariable { std::string name; };
 struct GlobalVariable { std::string name; };
@@ -70,21 +79,21 @@ struct Branch { Label label; };
 struct Return {};
 struct BranchIf { Label label; std::unique_ptr<Expression> condition; };
 
-struct VariableDeclaration { std::string variableName; };
+struct VariableDeclaration { std::string variableName; Type type; };
-struct GlobalVariableDeclaration { std::string variableName; };
+struct GlobalVariableDeclaration { std::string variableName; Type type; };
 struct FunctionImport {
 	std::string module;
 	std::string externalName;
 	std::string internalName;
-	std::vector<std::string> paramTypes;
+	std::vector<Type> paramTypes;
-	std::optional<std::string> returnType;
+	std::optional<Type> returnType;
 };
 
 struct FunctionDefinition
 {
 	std::string name;
-	std::vector<std::string> parameterNames;
+	std::vector<TypedName> parameters;
-	bool returns;
+	std::optional<Type> returnType;
 	std::vector<VariableDeclaration> locals;
 	std::vector<Expression> body;
 };

libyul/backends/wasm/WasmCodeTransform.cpp

@@ -88,7 +88,7 @@ wasm::Expression WasmCodeTransform::operator()(VariableDeclaration const& _varDe
 	for (auto const& var: _varDecl.variables)
 	{
 		variableNames.emplace_back(var.name.str());
-		m_localVariables.emplace_back(wasm::VariableDeclaration{variableNames.back()});
+		m_localVariables.emplace_back(wasm::VariableDeclaration{variableNames.back(), wasm::Type::i64});
 	}
 
 	if (_varDecl.value)
@@ -127,10 +127,10 @@ wasm::Expression WasmCodeTransform::operator()(FunctionCall const& _call)
 					builtin->name.str().substr(4),
 					builtin->name.str(),
 					{},
-					builtin->returns.empty() ? nullopt : make_optional<string>(builtin->returns.front().str())
+					builtin->returns.empty() ? nullopt : make_optional<wasm::Type>(translatedType(builtin->returns.front()))
 				};
 				for (auto const& param: builtin->parameters)
-					imp.paramTypes.emplace_back(param.str());
+					imp.paramTypes.emplace_back(translatedType(param));
 				m_functionsToImport[builtin->name] = std::move(imp);
 			}
 			typeConversionNeeded = true;
@@ -184,7 +184,7 @@ wasm::Expression WasmCodeTransform::operator()(Literal const& _literal)
 {
 	u256 value = valueOfLiteral(_literal);
 	yulAssert(value <= numeric_limits<uint64_t>::max(), "Literal too large: " + value.str());
-	return wasm::Literal{uint64_t(value)};
+	return wasm::Literal{static_cast<uint64_t>(value)};
 }
 
 wasm::Expression WasmCodeTransform::operator()(If const& _if)
@@ -193,7 +193,7 @@ wasm::Expression WasmCodeTransform::operator()(If const& _if)
 
 	vector<wasm::Expression> args;
 	args.emplace_back(visitReturnByValue(*_if.condition));
-	args.emplace_back(wasm::Literal{0});
+	args.emplace_back(wasm::Literal{static_cast<uint64_t>(0)});
 	return wasm::If{
 		make_unique<wasm::Expression>(wasm::BuiltinCall{"i64.ne", std::move(args)}),
 		visit(_if.body.statements),
@@ -205,7 +205,7 @@ wasm::Expression WasmCodeTransform::operator()(Switch const& _switch)
 {
 	wasm::Block block;
 	string condition = m_nameDispenser.newName("condition"_yulstring).str();
-	m_localVariables.emplace_back(wasm::VariableDeclaration{condition});
+	m_localVariables.emplace_back(wasm::VariableDeclaration{condition, wasm::Type::i64});
 	block.statements.emplace_back(wasm::LocalAssignment{condition, visit(*_switch.expression)});
 
 	vector<wasm::Expression>* currentBlock = &block.statements;
@@ -325,10 +325,11 @@ wasm::FunctionDefinition WasmCodeTransform::translateFunction(yul::FunctionDefin
 	wasm::FunctionDefinition fun;
 	fun.name = _fun.name.str();
 	for (auto const& param: _fun.parameters)
-		fun.parameterNames.emplace_back(param.name.str());
+		fun.parameters.push_back({param.name.str(), wasm::Type::i64});
 	for (auto const& retParam: _fun.returnVariables)
-		fun.locals.emplace_back(wasm::VariableDeclaration{retParam.name.str()});
+		fun.locals.emplace_back(wasm::VariableDeclaration{retParam.name.str(), wasm::Type::i64});
-	fun.returns = !_fun.returnVariables.empty();
+	if (!_fun.returnVariables.empty())
+		fun.returnType = wasm::Type::i64;
 
 	yulAssert(m_localVariables.empty(), "");
 	yulAssert(m_functionBodyLabel.empty(), "");
@@ -361,14 +362,14 @@ wasm::Expression WasmCodeTransform::injectTypeConversionIfNeeded(wasm::FunctionC
 {
 	wasm::FunctionImport const& import = m_functionsToImport.at(YulString{_call.functionName});
 	for (size_t i = 0; i < _call.arguments.size(); ++i)
-		if (import.paramTypes.at(i) == "i32")
+		if (import.paramTypes.at(i) == wasm::Type::i32)
 			_call.arguments[i] = wasm::BuiltinCall{"i32.wrap_i64", make_vector<wasm::Expression>(std::move(_call.arguments[i]))};
 		else
-			yulAssert(import.paramTypes.at(i) == "i64", "Unknown type " + import.paramTypes.at(i));
+			yulAssert(import.paramTypes.at(i) == wasm::Type::i64, "Invalid Wasm type");
 
-	if (import.returnType && *import.returnType != "i64")
+	if (import.returnType && *import.returnType != wasm::Type::i64)
 	{
-		yulAssert(*import.returnType == "i32", "Invalid type " + *import.returnType);
+		yulAssert(*import.returnType == wasm::Type::i32, "Invalid Wasm type");
 		return wasm::BuiltinCall{"i64.extend_i32_u", make_vector<wasm::Expression>(std::move(_call))};
 	}
 	return {std::move(_call)};
@@ -400,6 +401,17 @@ void WasmCodeTransform::allocateGlobals(size_t _amount)
 {
 	while (m_globalVariables.size() < _amount)
 		m_globalVariables.emplace_back(wasm::GlobalVariableDeclaration{
-			m_nameDispenser.newName("global_"_yulstring).str()
+			m_nameDispenser.newName("global_"_yulstring).str(),
+			wasm::Type::i64
 		});
 }
+
+wasm::Type WasmCodeTransform::translatedType(yul::Type _yulType)
+{
+	if (_yulType == "i32"_yulstring)
+		return wasm::Type::i32;
+	else if (_yulType == "i64"_yulstring)
+		return wasm::Type::i64;
+	else
+		yulAssert(false, "This Yul type does not have a corresponding type in Wasm.");
+}

libyul/backends/wasm/WasmCodeTransform.h

@@ -89,6 +89,8 @@ private:
 	/// Makes sure that there are at least @a _amount global variables.
 	void allocateGlobals(size_t _amount);
 
+	static wasm::Type translatedType(yul::Type _yulType);
+
 	Dialect const& m_dialect;
 	NameDispenser m_nameDispenser;
 

test/cmdlineTests/wasm_to_wasm_function_returning_multiple_values/args

@@ -0,0 +1 @@
+--yul --yul-dialect ewasm --machine ewasm

test/cmdlineTests/wasm_to_wasm_function_returning_multiple_values/err

@@ -0,0 +1 @@
+Warning: Yul is still experimental. Please use the output with care.

test/cmdlineTests/wasm_to_wasm_function_returning_multiple_values/input.yul

@@ -0,0 +1,17 @@
+object "object" {
+    code {
+        function main()
+        {
+            let m:i64, n:i32, p:i32, q:i64 := multireturn(1:i32, 2:i64, 3:i64, 4:i32)
+        }
+
+        function multireturn(a:i32, b:i64, c:i64, d:i32) -> x:i64, y:i32, z:i32, w:i64
+        {
+            x := b
+            w := c
+
+            y := a
+            z := d
+        }
+    }
+}

test/cmdlineTests/wasm_to_wasm_function_returning_multiple_values/output

@@ -0,0 +1,73 @@
+
+======= wasm_to_wasm_function_returning_multiple_values/input.yul (Ewasm) =======
+
+Pretty printed source:
+object "object" {
+    code {
+        function main()
+        {
+            let m, n:i32, p:i32, q := multireturn(1:i32, 2, 3, 4:i32)
+        }
+        function multireturn(a:i32, b, c, d:i32) -> x, y:i32, z:i32, w
+        {
+            x := b
+            w := c
+            y := a
+            z := d
+        }
+    }
+}
+
+
+Binary representation:
+0061736d01000000010c0260000060047e7e7e7e017e020100030302000105030100010610037e0142000b7e0142000b7e0142000b071102066d656d6f72790200046d61696e00000a4a022201047e024002404201420242034204100121002300210123012102230221030b0b0b2501047e0240200121042002210720002105200321060b20052400200624012007240220040b
+
+Text representation:
+(module
+    (memory $memory (export "memory") 1)
+    (export "main" (func $main))
+    (global $global_ (mut i64) (i64.const 0))
+    (global $global__1 (mut i64) (i64.const 0))
+    (global $global__2 (mut i64) (i64.const 0))
+
+(func $main
+    (local $m i64)
+    (local $n i64)
+    (local $p i64)
+    (local $q i64)
+    (block $label_
+        (block
+            (local.set $m (call $multireturn (i64.const 1) (i64.const 2) (i64.const 3) (i64.const 4)))
+            (local.set $n (global.get $global_))
+            (local.set $p (global.get $global__1))
+            (local.set $q (global.get $global__2))
+
+        )
+
+    )
+)
+
+(func $multireturn
+    (param $a i64)
+    (param $b i64)
+    (param $c i64)
+    (param $d i64)
+    (result i64)
+    (local $x i64)
+    (local $y i64)
+    (local $z i64)
+    (local $w i64)
+    (block $label__3
+        (local.set $x (local.get $b))
+        (local.set $w (local.get $c))
+        (local.set $y (local.get $a))
+        (local.set $z (local.get $d))
+
+    )
+    (global.set $global_ (local.get $y))
+    (global.set $global__1 (local.get $z))
+    (global.set $global__2 (local.get $w))
+    (local.get $x)
+)
+
+)