Also optimize memory.

libyul/CMakeLists.txt

@@ -96,6 +96,8 @@ add_library(yul
 	optimiser/InlinableExpressionFunctionFinder.h
 	optimiser/KnowledgeBase.cpp
 	optimiser/KnowledgeBase.h
+	optimiser/LoadResolver.cpp
+	optimiser/LoadResolver.h
 	optimiser/MainFunction.cpp
 	optimiser/MainFunction.h
 	optimiser/Metrics.cpp
@@ -111,8 +113,6 @@ add_library(yul
 	optimiser/RedundantAssignEliminator.cpp
 	optimiser/RedundantAssignEliminator.h
 	optimiser/Rematerialiser.cpp
-	optimiser/SLoadResolver.cpp
-	optimiser/SLoadResolver.h
 	optimiser/Rematerialiser.h
 	optimiser/SSAReverser.cpp
 	optimiser/SSAReverser.h

libyul/Dialect.h

@@ -60,6 +60,9 @@ struct BuiltinFunction
 	/// If false, storage of the current contract before and after the function is the same
 	/// under every circumstance. If the function does not return, this can be false.
 	bool invalidatesStorage = true;
+	/// If false, memory before and after the function is the same under every circumstance.
+	/// If the function does not return, this can be false.
+	bool invalidatesMemory = true;
 	/// If true, can only accept literals as arguments and they cannot be moved to variables.
 	bool literalArguments = false;
 };

libyul/backends/evm/EVMDialect.cpp

@@ -55,6 +55,7 @@ pair<YulString, BuiltinFunctionForEVM> createEVMFunction(
 	f.sideEffectFreeIfNoMSize = eth::SemanticInformation::sideEffectFreeIfNoMSize(_instruction);
 	f.isMSize = _instruction == dev::eth::Instruction::MSIZE;
 	f.invalidatesStorage = eth::SemanticInformation::invalidatesStorage(_instruction);
+	f.invalidatesMemory = eth::SemanticInformation::invalidatesMemory(_instruction);
 	f.literalArguments = false;
 	f.instruction = _instruction;
 	f.generateCode = [_instruction](
@@ -78,6 +79,7 @@ pair<YulString, BuiltinFunctionForEVM> createFunction(
 	bool _sideEffectFree,
 	bool _sideEffectFreeIfNoMSize,
 	bool _invalidatesStorage,
+	bool _invalidatesMemory,
 	bool _literalArguments,
 	std::function<void(FunctionCall const&, AbstractAssembly&, BuiltinContext&, std::function<void()>)> _generateCode
 )
@@ -93,6 +95,7 @@ pair<YulString, BuiltinFunctionForEVM> createFunction(
 	f.sideEffectFreeIfNoMSize = _sideEffectFreeIfNoMSize;
 	f.isMSize = false;
 	f.invalidatesStorage = _invalidatesStorage;
+	f.invalidatesMemory = _invalidatesMemory;
 	f.instruction = {};
 	f.generateCode = std::move(_generateCode);
 	return {name, f};
@@ -113,7 +116,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 
 	if (_objectAccess)
 	{
-		builtins.emplace(createFunction("datasize", 1, 1, true, true, true, false, true, [](
+		builtins.emplace(createFunction("datasize", 1, 1, true, true, true, false, false, true, [](
 			FunctionCall const& _call,
 			AbstractAssembly& _assembly,
 			BuiltinContext& _context,
@@ -134,7 +137,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 				_assembly.appendDataSize(_context.subIDs.at(dataName));
 			}
 		}));
-		builtins.emplace(createFunction("dataoffset", 1, 1, true, true, true, false, true, [](
+		builtins.emplace(createFunction("dataoffset", 1, 1, true, true, true, false, false, true, [](
 			FunctionCall const& _call,
 			AbstractAssembly& _assembly,
 			BuiltinContext& _context,
@@ -155,7 +158,7 @@ map<YulString, BuiltinFunctionForEVM> createBuiltins(langutil::EVMVersion _evmVe
 				_assembly.appendDataOffset(_context.subIDs.at(dataName));
 			}
 		}));
-		builtins.emplace(createFunction("datacopy", 3, 0, false, false, false, false, false, [](
+		builtins.emplace(createFunction("datacopy", 3, 0, false, false, false, false, true, false, [](
 			FunctionCall const&,
 			AbstractAssembly& _assembly,
 			BuiltinContext&,

libyul/backends/wasm/WasmDialect.cpp

@@ -84,5 +84,6 @@ void WasmDialect::addFunction(string _name, size_t _params, size_t _returns)
 	f.sideEffectFreeIfNoMSize = false;
 	f.isMSize = false;
 	f.invalidatesStorage = true;
+	f.invalidatesMemory = true;
 	f.literalArguments = false;
 }

libyul/optimiser/DataFlowAnalyzer.cpp

@@ -40,7 +40,7 @@ using namespace yul;
 
 void DataFlowAnalyzer::operator()(ExpressionStatement& _statement)
 {
-	if (boost::optional<pair<YulString, YulString>> vars = isSimpleSStore(_statement))
+	if (auto vars = isSimpleStore(dev::eth::Instruction::SSTORE, _statement))
 	{
 		ASTModifier::operator()(_statement);
 		m_storage.set(vars->first, vars->second);
@@ -54,9 +54,22 @@ void DataFlowAnalyzer::operator()(ExpressionStatement& _statement)
 		for (YulString const& key: keysToErase)
 			m_storage.eraseKey(key);
 	}
+	else if (auto vars = isSimpleStore(dev::eth::Instruction::MSTORE, _statement))
+	{
+		ASTModifier::operator()(_statement);
+		set<YulString> keysToErase;
+		for (auto const& item: m_memory.values)
+			if (!m_knowledgeBase.knownToBeDifferentByAtLeast32(vars->first, item.first))
+				keysToErase.insert(item.first);
+		// TODO is it fine to do that here?
+		// can we also move the storage above?
+		m_memory.set(vars->first, vars->second);
+		for (YulString const& key: keysToErase)
+			m_memory.eraseKey(key);
+	}
 	else
 	{
-		clearStorageKnowledgeIfInvalidated(_statement.expression);
+		clearKnowledgeIfInvalidated(_statement.expression);
 		ASTModifier::operator()(_statement);
 	}
 }
@@ -67,7 +80,7 @@ void DataFlowAnalyzer::operator()(Assignment& _assignment)
 	for (auto const& var: _assignment.variableNames)
 		names.emplace(var.name);
 	assertThrow(_assignment.value, OptimizerException, "");
-	clearStorageKnowledgeIfInvalidated(*_assignment.value);
+	clearKnowledgeIfInvalidated(*_assignment.value);
 	visit(*_assignment.value);
 	handleAssignment(names, _assignment.value.get());
 }
@@ -81,7 +94,7 @@ void DataFlowAnalyzer::operator()(VariableDeclaration& _varDecl)
 
 	if (_varDecl.value)
 	{
-		clearStorageKnowledgeIfInvalidated(*_varDecl.value);
+		clearKnowledgeIfInvalidated(*_varDecl.value);
 		visit(*_varDecl.value);
 	}
 
@@ -90,12 +103,13 @@ void DataFlowAnalyzer::operator()(VariableDeclaration& _varDecl)
 
 void DataFlowAnalyzer::operator()(If& _if)
 {
-	clearStorageKnowledgeIfInvalidated(*_if.condition);
+	clearKnowledgeIfInvalidated(*_if.condition);
 	InvertibleMap<YulString, YulString> storage = m_storage;
+	InvertibleMap<YulString, YulString> memory = m_memory;
 
 	ASTModifier::operator()(_if);
 
-	joinStorageKnowledge(storage);
+	joinKnowledge(storage, memory);
 
 	Assignments assignments;
 	assignments(_if.body);
@@ -104,24 +118,25 @@ void DataFlowAnalyzer::operator()(If& _if)
 
 void DataFlowAnalyzer::operator()(Switch& _switch)
 {
-	clearStorageKnowledgeIfInvalidated(*_switch.expression);
+	clearKnowledgeIfInvalidated(*_switch.expression);
 	visit(*_switch.expression);
 	set<YulString> assignedVariables;
 	for (auto& _case: _switch.cases)
 	{
 		InvertibleMap<YulString, YulString> storage = m_storage;
+		InvertibleMap<YulString, YulString> memory = m_memory;
 		(*this)(_case.body);
-		joinStorageKnowledge(storage);
+		joinKnowledge(storage, memory);
 
 		Assignments assignments;
 		assignments(_case.body);
 		assignedVariables += assignments.names();
 		// This is a little too destructive, we could retain the old values.
 		clearValues(assignments.names());
-		clearStorageKnowledgeIfInvalidated(_case.body);
+		clearKnowledgeIfInvalidated(_case.body);
 	}
 	for (auto& _case: _switch.cases)
-		clearStorageKnowledgeIfInvalidated(_case.body);
+		clearKnowledgeIfInvalidated(_case.body);
 	clearValues(assignedVariables);
 }
 
@@ -132,9 +147,11 @@ void DataFlowAnalyzer::operator()(FunctionDefinition& _fun)
 	map<YulString, Expression const*> value;
 	InvertibleRelation<YulString> references;
 	InvertibleMap<YulString, YulString> storage;
+	InvertibleMap<YulString, YulString> memory;
 	m_value.swap(value);
 	swap(m_references, references);
 	swap(m_storage, storage);
+	swap(m_memory, memory);
 	pushScope(true);
 
 	for (auto const& parameter: _fun.parameters)
@@ -150,6 +167,7 @@ void DataFlowAnalyzer::operator()(FunctionDefinition& _fun)
 	m_value.swap(value);
 	swap(m_references, references);
 	swap(m_storage, storage);
+	swap(m_memory, memory);
 }
 
 void DataFlowAnalyzer::operator()(ForLoop& _for)
@@ -167,19 +185,19 @@ void DataFlowAnalyzer::operator()(ForLoop& _for)
 	clearValues(assignments.names());
 
 	// break/continue are tricky for storage and thus we almost always clear here.
-	clearStorageKnowledgeIfInvalidated(*_for.condition);
-	clearStorageKnowledgeIfInvalidated(_for.post);
-	clearStorageKnowledgeIfInvalidated(_for.body);
+	clearKnowledgeIfInvalidated(*_for.condition);
+	clearKnowledgeIfInvalidated(_for.post);
+	clearKnowledgeIfInvalidated(_for.body);
 
 	visit(*_for.condition);
 	(*this)(_for.body);
 	clearValues(assignmentsSinceCont.names());
-	clearStorageKnowledgeIfInvalidated(_for.body);
+	clearKnowledgeIfInvalidated(_for.body);
 	(*this)(_for.post);
 	clearValues(assignments.names());
-	clearStorageKnowledgeIfInvalidated(*_for.condition);
-	clearStorageKnowledgeIfInvalidated(_for.post);
-	clearStorageKnowledgeIfInvalidated(_for.body);
+	clearKnowledgeIfInvalidated(*_for.condition);
+	clearKnowledgeIfInvalidated(_for.post);
+	clearKnowledgeIfInvalidated(_for.body);
 }
 
 void DataFlowAnalyzer::operator()(Block& _block)
@@ -219,6 +237,10 @@ void DataFlowAnalyzer::handleAssignment(set<YulString> const& _variables, Expres
 		m_storage.eraseKey(name);
 		// assignment to slot contents denoted by "name"
 		m_storage.eraseValue(name);
+		// assignment to slot denoted by "name"
+		m_memory.eraseKey(name);
+		// assignment to slot contents denoted by "name"
+		m_memory.eraseValue(name);
 	}
 }
 
@@ -257,6 +279,10 @@ void DataFlowAnalyzer::clearValues(set<YulString> _variables)
 		m_storage.eraseKey(name);
 		// clear slot contents denoted by "name"
 		m_storage.eraseValue(name);
+		// assignment to slot denoted by "name"
+		m_memory.eraseKey(name);
+		// assignment to slot contents denoted by "name"
+		m_memory.eraseValue(name);
 	}
 
 	// Also clear variables that reference variables to be cleared.
@@ -271,29 +297,51 @@ void DataFlowAnalyzer::clearValues(set<YulString> _variables)
 		m_references.eraseKey(name);
 }
 
-void DataFlowAnalyzer::clearStorageKnowledgeIfInvalidated(Block const& _block)
+void DataFlowAnalyzer::clearKnowledgeIfInvalidated(Block const& _block)
 {
-	if (SideEffectsCollector(m_dialect, _block).invalidatesStorage())
+	SideEffectsCollector sideEffects(m_dialect, _block);
+	if (sideEffects.invalidatesStorage())
 		m_storage.clear();
+	if (sideEffects.invalidatesMemory())
+		m_memory.clear();
 }
 
-void DataFlowAnalyzer::clearStorageKnowledgeIfInvalidated(Expression const& _expr)
+void DataFlowAnalyzer::clearKnowledgeIfInvalidated(Expression const& _expr)
 {
-	if (SideEffectsCollector(m_dialect, _expr).invalidatesStorage())
+	SideEffectsCollector sideEffects(m_dialect, _expr);
+	if (sideEffects.invalidatesStorage())
 		m_storage.clear();
+	if (sideEffects.invalidatesMemory())
+		m_memory.clear();
 }
 
-void DataFlowAnalyzer::joinStorageKnowledge(InvertibleMap<YulString, YulString> const& _other)
+void DataFlowAnalyzer::joinKnowledge(
+	InvertibleMap<YulString, YulString> const& _olderStorage,
+	InvertibleMap<YulString, YulString> const& _olderMemory
+)
 {
+	joinKnowledgeHelper(m_storage, _olderStorage);
+	joinKnowledgeHelper(m_memory, _olderMemory);
+}
+
+void DataFlowAnalyzer::joinKnowledgeHelper(
+	InvertibleMap<YulString, YulString>& _this,
+	InvertibleMap<YulString, YulString> const& _older
+)
+{
+	// We clear if the key does not exist in the older map or if the value is different.
+	// This also works for memory because _older is an "older version"
+	// of m_memory and thus any overlapping write would have cleared the keys
+	// that are not known to be different inside m_memory already.
 	set<YulString> keysToErase;
-	for (auto const& item: m_storage.values)
+	for (auto const& item: _this.values)
 	{
-		auto it = _other.values.find(item.first);
-		if (it == _other.values.end() || it->second != item.second)
+		auto it = _older.values.find(item.first);
+		if (it == _older.values.end() || it->second != item.second)
 			keysToErase.insert(item.first);
 	}
 	for (auto const& key: keysToErase)
-		m_storage.eraseKey(key);
+		_this.eraseKey(key);
 }
 
 bool DataFlowAnalyzer::inScope(YulString _variableName) const
@@ -308,16 +356,22 @@ bool DataFlowAnalyzer::inScope(YulString _variableName) const
 	return false;
 }
 
-boost::optional<pair<YulString, YulString>> DataFlowAnalyzer::isSimpleSStore(
+boost::optional<pair<YulString, YulString>> DataFlowAnalyzer::isSimpleStore(
+	dev::eth::Instruction _store,
 	ExpressionStatement const& _statement
 ) const
 {
+	yulAssert(
+		_store == dev::eth::Instruction::MSTORE ||
+		_store == dev::eth::Instruction::SSTORE,
+		""
+	);
 	if (_statement.expression.type() == typeid(FunctionCall))
 	{
 		FunctionCall const& funCall = boost::get<FunctionCall>(_statement.expression);
 		if (EVMDialect const* dialect = dynamic_cast<EVMDialect const*>(&m_dialect))
 			if (auto const* builtin = dialect->builtin(funCall.functionName.name))
-				if (builtin->instruction == dev::eth::Instruction::SSTORE)
+				if (builtin->instruction == _store)
 					if (
 						funCall.arguments.at(0).type() == typeid(Identifier) &&
 						funCall.arguments.at(1).type() == typeid(Identifier)

libyul/optimiser/DataFlowAnalyzer.h

@@ -27,6 +27,9 @@
 #include <libyul/YulString.h>
 #include <libyul/AsmData.h>
 
+// TODO avoid
+#include <libevmasm/Instruction.h>
+
 #include <libdevcore/InvertibleMap.h>
 
 #include <map>
@@ -93,18 +96,32 @@ protected:
 	/// for example at points where control flow is merged.
 	void clearValues(std::set<YulString> _names);
 
-	/// Clears knowledge about storage if storage may be modified inside the block.
-	void clearStorageKnowledgeIfInvalidated(Block const& _block);
+	/// Clears knowledge about storage or memory if they may be modified inside the block.
+	void clearKnowledgeIfInvalidated(Block const& _block);
 
-	/// Clears knowledge about storage if storage may be modified inside the expression.
-	void clearStorageKnowledgeIfInvalidated(Expression const& _expression);
+	/// Clears knowledge about storage or memory if they may be modified inside the expression.
+	void clearKnowledgeIfInvalidated(Expression const& _expression);
 
-	void joinStorageKnowledge(InvertibleMap<YulString, YulString> const& _other);
+	/// Joins knowledge about storage and memory with an older point in the control-flow.
+	/// This only works if the current state is a direct successor of the older point,
+	/// i.e. `_otherStorage` and `_otherMemory` cannot have additional changes.
+	void joinKnowledge(
+		InvertibleMap<YulString, YulString> const& _olderStorage,
+		InvertibleMap<YulString, YulString> const& _olderMemory
+	);
+
+	static void joinKnowledgeHelper(
+		InvertibleMap<YulString, YulString>& _thisData,
+		InvertibleMap<YulString, YulString> const& _olderData
+	);
 
 	/// Returns true iff the variable is in scope.
 	bool inScope(YulString _variableName) const;
 
-	boost::optional<std::pair<YulString, YulString>> isSimpleSStore(ExpressionStatement const& _statement) const;
+	boost::optional<std::pair<YulString, YulString>> isSimpleStore(
+		dev::eth::Instruction _store,
+		ExpressionStatement const& _statement
+	) const;
 
 	Dialect const& m_dialect;
 
@@ -115,6 +132,7 @@ protected:
 	InvertibleRelation<YulString> m_references;
 
 	InvertibleMap<YulString, YulString> m_storage;
+	InvertibleMap<YulString, YulString> m_memory;
 
 	KnowledgeBase m_knowledgeBase;
 

libyul/optimiser/LoadResolver.cpp

@@ -19,30 +19,48 @@
  * currently stored in storage, if known.
  */
 
-#include <libyul/optimiser/SLoadResolver.h>
+#include <libyul/optimiser/LoadResolver.h>
 
 #include <libyul/backends/evm/EVMDialect.h>
+#include <libyul/optimiser/Semantics.h>
 #include <libyul/AsmData.h>
 
 using namespace std;
 using namespace dev;
 using namespace yul;
 
-void SLoadResolver::visit(Expression& _e)
+void LoadResolver::run(Dialect const& _dialect, Block& _ast)
+{
+	bool containsMSize = SideEffectsCollector(_dialect, _ast).containsMSize();
+	LoadResolver{_dialect, !containsMSize}(_ast);
+}
+
+void LoadResolver::visit(Expression& _e)
 {
 	if (_e.type() == typeid(FunctionCall))
 	{
 		FunctionCall const& funCall = boost::get<FunctionCall>(_e);
 		if (auto const* builtin = dynamic_cast<EVMDialect const&>(m_dialect).builtin(funCall.functionName.name))
-			if (builtin->instruction == dev::eth::Instruction::SLOAD)
-				if (funCall.arguments.at(0).type() == typeid(Identifier))
+			if (!builtin->parameters.empty() && funCall.arguments.at(0).type() == typeid(Identifier))
+			{
+				YulString key = boost::get<Identifier>(funCall.arguments.at(0)).name;
+				if (
+					builtin->instruction == dev::eth::Instruction::SLOAD &&
+					m_storage.values.count(key)
+				)
 				{
-					YulString key = boost::get<Identifier>(funCall.arguments.at(0)).name;
-					if (m_storage.values.count(key))
-					{
-						_e = Identifier{locationOf(_e), m_storage.values[key]};
-						return;
-					}
+					_e = Identifier{locationOf(_e), m_storage.values[key]};
+					return;
 				}
+				else if (
+					m_optimizeMLoad &&
+					builtin->instruction == dev::eth::Instruction::MLOAD &&
+					m_memory.values.count(key)
+				)
+				{
+					_e = Identifier{locationOf(_e), m_memory.values[key]};
+					return;
+				}
+			}
 	}
 }

libyul/optimiser/LoadResolver.h

@@ -29,21 +29,29 @@ namespace yul
 struct EVMDialect;
 
 /**
- * Optimisation stage that replaces expressions of type ``sload(x)`` by the value
- * currently stored in storage, if known.
+ * Optimisation stage that replaces expressions of type ``sload(x)`` and ``mload(x)`` by the value
+ * currently stored in storage resp. memory, if known.
  *
  * Works best if the code is in SSA form.
  *
  * Prerequisite: Disambiguator, ForLoopInitRewriter.
  */
-class SLoadResolver: public DataFlowAnalyzer
+class LoadResolver: public DataFlowAnalyzer
 {
 public:
-	SLoadResolver(Dialect const& _dialect): DataFlowAnalyzer(_dialect) {}
+	static void run(Dialect const& _dialect, Block& _ast);
+
+private:
+	LoadResolver(Dialect const& _dialect, bool _optimizeMLoad):
+		DataFlowAnalyzer(_dialect),
+		m_optimizeMLoad(_optimizeMLoad)
+	{}
 
 protected:
 	using ASTModifier::visit;
 	void visit(Expression& _e) override;
+
+	bool m_optimizeMLoad = false;
 };
 
 }

libyul/optimiser/Semantics.cpp

@@ -65,6 +65,8 @@ void SideEffectsCollector::operator()(FunctionalInstruction const& _instr)
 		m_containsMSize = true;
 	if (eth::SemanticInformation::invalidatesStorage(_instr.instruction))
 		m_invalidatesStorage = true;
+	if (eth::SemanticInformation::invalidatesMemory(_instr.instruction))
+		m_invalidatesMemory = true;
 }
 
 void SideEffectsCollector::operator()(FunctionCall const& _functionCall)
@@ -83,6 +85,8 @@ void SideEffectsCollector::operator()(FunctionCall const& _functionCall)
 			m_containsMSize = true;
 		if (f->invalidatesStorage)
 			m_invalidatesStorage = true;
+		if (f->invalidatesMemory)
+			m_invalidatesMemory = true;
 	}
 	else
 	{
@@ -90,6 +94,7 @@ void SideEffectsCollector::operator()(FunctionCall const& _functionCall)
 		m_sideEffectFree = false;
 		m_sideEffectFreeIfNoMSize = false;
 		m_invalidatesStorage = true;
+		m_invalidatesMemory = true;
 	}
 }
 

libyul/optimiser/Semantics.h

@@ -55,6 +55,7 @@ public:
 	bool sideEffectFreeIfNoMSize() const { return m_sideEffectFreeIfNoMSize; }
 	bool containsMSize() const { return m_containsMSize; }
 	bool invalidatesStorage() const { return m_invalidatesStorage; }
+	bool invalidatesMemory() const { return m_invalidatesMemory; }
 
 private:
 	Dialect const& m_dialect;
@@ -73,6 +74,7 @@ private:
 	/// If false, storage is guaranteed to be unchanged by the coded under all
 	/// circumstances.
 	bool m_invalidatesStorage = false;
+	bool m_invalidatesMemory = false;
 };
 
 /**
@@ -92,7 +94,6 @@ public:
 	void visit(Statement const&) override;
 	using ASTWalker::visit;
 
-
 	std::set<YulString> const& referencedVariables() const { return m_variableReferences; }
 
 private: