Use hash of candidates for CSE.

libyul/optimiser/BlockHasher.cpp

@@ -20,7 +20,6 @@
  */
 
 #include <libyul/optimiser/BlockHasher.h>
-#include <libyul/optimiser/SyntacticalEquality.h>
 #include <libyul/AST.h>
 #include <libyul/Utilities.h>
 
@@ -195,3 +194,35 @@ void BlockHasher::operator()(Block const& _block)
 	for (auto& externalReference: subBlockHasher.m_externalReferences)
 		(*this)(Identifier{{}, externalReference});
 }
+
+uint64_t ExpressionHasher::run(Expression const& _e)
+{
+	ExpressionHasher expressionHasher;
+	expressionHasher.visit(_e);
+	return expressionHasher.m_hash;
+}
+
+void ExpressionHasher::operator()(Literal const& _literal)
+{
+	hash64(compileTimeLiteralHash("Literal"));
+	if (_literal.kind == LiteralKind::Number)
+		hash64(std::hash<u256>{}(valueOfNumberLiteral(_literal)));
+	else
+		hash64(_literal.value.hash());
+	hash64(_literal.type.hash());
+	hash8(static_cast<uint8_t>(_literal.kind));
+}
+
+void ExpressionHasher::operator()(Identifier const& _identifier)
+{
+	hash64(compileTimeLiteralHash("Identifier"));
+	hash64(_identifier.name.hash());
+}
+
+void ExpressionHasher::operator()(FunctionCall const& _funCall)
+{
+	hash64(compileTimeLiteralHash("FunctionCall"));
+	hash64(_funCall.functionName.name.hash());
+	hash64(_funCall.arguments.size());
+	ASTWalker::operator()(_funCall);
+}

libyul/optimiser/BlockHasher.h

@@ -27,6 +27,38 @@
 namespace solidity::yul
 {
 
+class HasherBase
+{
+public:
+	static constexpr uint64_t fnvPrime = 1099511628211u;
+	static constexpr uint64_t fnvEmptyHash = 14695981039346656037u;
+
+protected:
+	void hash8(uint8_t _value)
+	{
+		m_hash *= fnvPrime;
+		m_hash ^= _value;
+	}
+	void hash16(uint16_t _value)
+	{
+		hash8(static_cast<uint8_t>(_value & 0xFF));
+		hash8(static_cast<uint8_t>(_value >> 8));
+	}
+	void hash32(uint32_t _value)
+	{
+		hash16(static_cast<uint16_t>(_value & 0xFFFF));
+		hash16(static_cast<uint16_t>(_value >> 16));
+	}
+	void hash64(uint64_t _value)
+	{
+		hash32(static_cast<uint32_t>(_value & 0xFFFFFFFF));
+		hash32(static_cast<uint32_t>(_value >> 32));
+	}
+
+
+	uint64_t m_hash = fnvEmptyHash;
+};
+
 /**
  * Optimiser component that calculates hash values for blocks.
  * Syntactically equal blocks will have identical hashes and
@@ -41,7 +73,7 @@ namespace solidity::yul
  *
  * Prerequisite: Disambiguator, ForLoopInitRewriter
  */
-class BlockHasher: public ASTWalker
+class BlockHasher: public ASTWalker, public HasherBase
 {
 public:
 
@@ -64,36 +96,12 @@ public:
 
 	static std::map<Block const*, uint64_t> run(Block const& _block);
 
-	static constexpr uint64_t fnvPrime = 1099511628211u;
-	static constexpr uint64_t fnvEmptyHash = 14695981039346656037u;
 
 private:
 	BlockHasher(std::map<Block const*, uint64_t>& _blockHashes): m_blockHashes(_blockHashes) {}
 
-	void hash8(uint8_t _value)
-	{
-		m_hash *= fnvPrime;
-		m_hash ^= _value;
-	}
-	void hash16(uint16_t _value)
-	{
-		hash8(static_cast<uint8_t>(_value & 0xFF));
-		hash8(static_cast<uint8_t>(_value >> 8));
-	}
-	void hash32(uint32_t _value)
-	{
-		hash16(static_cast<uint16_t>(_value & 0xFFFF));
-		hash16(static_cast<uint16_t>(_value >> 16));
-	}
-	void hash64(uint64_t _value)
-	{
-		hash32(static_cast<uint32_t>(_value & 0xFFFFFFFF));
-		hash32(static_cast<uint32_t>(_value >> 32));
-	}
-
 	std::map<Block const*, uint64_t>& m_blockHashes;
 
-	uint64_t m_hash = fnvEmptyHash;
 	struct VariableReference
 	{
 		size_t id = 0;
@@ -106,4 +114,24 @@ private:
 };
 
 
+/**
+ * Computes hashes of expressions that are likely different for syntactically different expressions.
+ * In contrast to the BlockHasher, hashes of identifiers are likely different if the identifiers
+ * have a different name and the same if the name matches.
+ * This means this hasher should only be used on disambiguated sources.
+ */
+class ExpressionHasher: public ASTWalker, public HasherBase
+{
+public:
+	/// Computes a hash of an expression that (in contrast to the behaviour of the class)
+	/// distinguishes (up to hash collisions) variables with different names.
+	static uint64_t run(Expression const& _e);
+
+	using ASTWalker::operator();
+
+	void operator()(Literal const&) override;
+	void operator()(Identifier const&) override;
+	void operator()(FunctionCall const& _funCall) override;
+};
+
 }

libyul/optimiser/CommonSubexpressionEliminator.cpp

@@ -22,6 +22,7 @@
 #include <libyul/optimiser/CommonSubexpressionEliminator.h>
 
 #include <libyul/optimiser/SyntacticalEquality.h>
+#include <libyul/optimiser/BlockHasher.h>
 #include <libyul/optimiser/CallGraphGenerator.h>
 #include <libyul/optimiser/Semantics.h>
 #include <libyul/SideEffects.h>
@@ -55,6 +56,7 @@ CommonSubexpressionEliminator::CommonSubexpressionEliminator(
 void CommonSubexpressionEliminator::operator()(FunctionDefinition& _fun)
 {
 	ScopedSaveAndRestore returnVariables(m_returnVariables, {});
+	ScopedSaveAndRestore replacementCandidates(m_replacementCandidates, {});
 
 	for (auto const& v: _fun.returnVariables)
 		m_returnVariables.insert(v.name);
@@ -103,25 +105,44 @@ void CommonSubexpressionEliminator::visit(Expression& _e)
 					_e = Identifier{debugDataOf(_e), value->name};
 		}
 	}
-	else
-	{
-		// TODO this search is rather inefficient.
-		for (auto const& [variable, value]: allValues())
-		{
-			assertThrow(value.value, OptimizerException, "");
-			// Prevent using the default value of return variables
-			// instead of literal zeros.
-			if (
-				m_returnVariables.count(variable) &&
-				holds_alternative<Literal>(*value.value) &&
-				valueOfLiteral(get<Literal>(*value.value)) == 0
-			)
-				continue;
-			if (SyntacticallyEqual{}(_e, *value.value) && inScope(variable))
+	else if (
+		auto it = m_replacementCandidates.find(&_e);
+		it != m_replacementCandidates.end()
+	)
+		for (auto const& variable: it->second)
+			if (AssignedValue const* value = variableValue(variable))
 			{
-				_e = Identifier{debugDataOf(_e), variable};
-				break;
+				assertThrow(value->value, OptimizerException, "");
+				// Prevent using the default value of return variables
+				// instead of literal zeros.
+				if (
+					m_returnVariables.count(variable) &&
+					holds_alternative<Literal>(*value->value) &&
+					valueOfLiteral(get<Literal>(*value->value)) == 0
+				)
+					continue;
+				// We check for syntactic equality again because the value might have changed.
+				if (inScope(variable) && SyntacticallyEqual{}(_e, *value->value))
+				{
+					_e = Identifier{debugDataOf(_e), variable};
+					break;
+				}
 			}
-		}
-	}
+}
+
+void CommonSubexpressionEliminator::assignValue(YulString _variable, Expression const* _value)
+{
+	if (_value)
+		m_replacementCandidates[_value].insert(_variable);
+	DataFlowAnalyzer::assignValue(_variable, _value);
+}
+
+uint64_t CommonSubexpressionEliminator::ExpressionHash::operator()(Expression const* _e) const
+{
+	return ExpressionHasher::run(*_e);
+}
+
+bool CommonSubexpressionEliminator::ExpressionEqual::operator()(Expression const* _a, Expression const* _b) const
+{
+	return SyntacticallyEqual{}(*_a, *_b);
 }

libyul/optimiser/CommonSubexpressionEliminator.h

@@ -58,8 +58,21 @@ protected:
 	using ASTModifier::visit;
 	void visit(Expression& _e) override;
 
+	void assignValue(YulString _variable, Expression const* _value) override;
 private:
+	struct ExpressionHash {
+		uint64_t operator()(Expression const*) const;
+	};
+	struct ExpressionEqual {
+		bool operator()(Expression const*, Expression const*) const;
+	};
 	std::set<YulString> m_returnVariables;
+	std::unordered_map<
+		Expression const*,
+		std::set<YulString>,
+		ExpressionHash,
+		ExpressionEqual
+	> m_replacementCandidates;
 };
 
 }

libyul/optimiser/DataFlowAnalyzer.h

@@ -123,7 +123,7 @@ protected:
 	/// for example at points where control flow is merged.
 	void clearValues(std::set<YulString> _names);
 
-	void assignValue(YulString _variable, Expression const* _value);
+	virtual void assignValue(YulString _variable, Expression const* _value);
 
 	/// Clears knowledge about storage or memory if they may be modified inside the block.
 	void clearKnowledgeIfInvalidated(Block const& _block);