Merge pull request #16265 from argotorg/optimize_ssacfg_building

Optimize SSA CFG datastructure + construction

Author: clonker

libsolutil/DisjointSet.cpp

@@ -24,7 +24,8 @@
 
 using namespace solidity::util;
 
-ContiguousDisjointSet::ContiguousDisjointSet(size_t const _numNodes):
+template<typename ValueType>
+ContiguousDisjointSet<ValueType>::ContiguousDisjointSet(size_t const _numNodes):
 	m_parents(_numNodes),
 	m_neighbors(_numNodes),
 	m_sizes(_numNodes, static_cast<value_type>(1)),
@@ -35,9 +36,11 @@ ContiguousDisjointSet::ContiguousDisjointSet(size_t const _numNodes):
 	std::iota(m_neighbors.begin(), m_neighbors.end(), 0);
 }
 
-size_t ContiguousDisjointSet::numSets() const { return m_numSets; }
+template<typename ValueType>
+size_t ContiguousDisjointSet<ValueType>::numSets() const { return m_numSets; }
 
-ContiguousDisjointSet::value_type ContiguousDisjointSet::find(value_type const _element) const
+template<typename ValueType>
+typename ContiguousDisjointSet<ValueType>::value_type ContiguousDisjointSet<ValueType>::find(value_type const _element) const
 {
 	solAssert(_element < m_parents.size());
 	// path halving
@@ -50,7 +53,8 @@ ContiguousDisjointSet::value_type ContiguousDisjointSet::find(value_type const _
 	return rootElement;
 }
 
-void ContiguousDisjointSet::merge(value_type const _x, value_type const _y, bool const _mergeBySize)
+template<typename ValueType>
+void ContiguousDisjointSet<ValueType>::merge(value_type const _x, value_type const _y, bool const _mergeBySize)
 {
 	auto xRoot = find(_x);
 	auto yRoot = find(_y);
@@ -69,17 +73,20 @@ void ContiguousDisjointSet::merge(value_type const _x, value_type const _y, bool
 	--m_numSets;
 }
 
-bool ContiguousDisjointSet::sameSubset(value_type const _x, value_type const _y) const
+template<typename ValueType>
+bool ContiguousDisjointSet<ValueType>::sameSubset(value_type const _x, value_type const _y) const
 {
 	return find(_x) == find(_y);
 }
 
-ContiguousDisjointSet::size_type ContiguousDisjointSet::sizeOfSubset(value_type const _x) const
+template<typename ValueType>
+typename  ContiguousDisjointSet<ValueType>::size_type ContiguousDisjointSet<ValueType>::sizeOfSubset(value_type const _x) const
 {
 	return m_sizes[find(_x)];
 }
 
-std::set<ContiguousDisjointSet::value_type> ContiguousDisjointSet::subset(value_type const _x) const
+template<typename ValueType>
+std::set<typename ContiguousDisjointSet<ValueType>::value_type> ContiguousDisjointSet<ValueType>::subset(value_type const _x) const
 {
 	solAssert(_x < m_parents.size());
 	std::set<value_type> result{_x};
@@ -92,7 +99,8 @@ std::set<ContiguousDisjointSet::value_type> ContiguousDisjointSet::subset(value_
 	return result;
 }
 
-std::vector<std::set<ContiguousDisjointSet::value_type>> ContiguousDisjointSet::subsets() const
+template<typename ValueType>
+std::vector<std::set<typename ContiguousDisjointSet<ValueType>::value_type>> ContiguousDisjointSet<ValueType>::subsets() const
 {
 	std::vector<std::set<value_type>> result;
 	std::vector<std::uint8_t> visited(m_parents.size(), false);
@@ -107,3 +115,5 @@ std::vector<std::set<ContiguousDisjointSet::value_type>> ContiguousDisjointSet::
 	}
 	return result;
 }
+
+template class solidity::util::ContiguousDisjointSet<std::uint32_t>;

libsolutil/DisjointSet.h

@@ -31,11 +31,12 @@ namespace solidity::util
 /// [1] https://en.wikipedia.org/wiki/Disjoint-set_data_structure
 /// [2] Tarjan, Robert E., and Jan Van Leeuwen. "Worst-case analysis of set union algorithms."
 ///     Journal of the ACM (JACM) 31.2 (1984): 245-281.
+template<typename ValueType>
 class ContiguousDisjointSet
 {
 public:
 	using size_type = size_t;
-	using value_type = size_t;
+	using value_type = ValueType;
 
 	/// Constructs a new disjoint set datastructure with `_numNodes` elements and each element in its own individual set
 	explicit ContiguousDisjointSet(size_t _numNodes);

libyul/YulStack.cpp

@@ -405,8 +405,7 @@ Json YulStack::cfgJson() const
 			keepLiteralAssignments
 		);
 		std::unique_ptr<ssa::ControlFlowLiveness> liveness = std::make_unique<ssa::ControlFlowLiveness>(*controlFlow);
-		ssa::SSACFGJsonExporter exporter(*controlFlow, liveness.get());
-		return exporter.run();
+		return ssa::json::exportControlFlow(*controlFlow, liveness.get());
 	};
 
 	std::function<Json(std::vector<std::shared_ptr<ObjectNode>>)> exportCFGFromSubObjects;

libyul/backends/evm/ssa/ControlFlow.cpp

@@ -24,8 +24,7 @@ using namespace solidity::yul::ssa;
 
 ControlFlowLiveness::ControlFlowLiveness(ControlFlow const& _controlFlow):
 	controlFlow(_controlFlow),
-	mainLiveness(std::make_unique<LivenessAnalysis>(*_controlFlow.mainGraph)),
-	functionLiveness(_controlFlow.functionGraphs | ranges::views::transform([](auto const& _cfg) { return std::make_unique<LivenessAnalysis>(*_cfg); }) | ranges::to<std::vector>)
+	cfgLiveness(_controlFlow.functionGraphs | ranges::views::transform([](auto const& _cfg) { return std::make_unique<LivenessAnalysis>(*_cfg); }) | ranges::to<std::vector>)
 { }
 
 std::string ControlFlowLiveness::toDot() const

libyul/backends/evm/ssa/ControlFlow.h

@@ -24,6 +24,8 @@
 #include <libyul/AST.h>
 #include <libyul/Scope.h>
 
+#include <range/v3/algorithm/find_if.hpp>
+
 namespace solidity::yul::ssa
 {
 
@@ -33,44 +35,52 @@ struct ControlFlowLiveness{
 	explicit ControlFlowLiveness(ControlFlow const& _controlFlow);
 
 	std::reference_wrapper<ControlFlow const> controlFlow;
-	std::unique_ptr<LivenessAnalysis> mainLiveness;
-	std::vector<std::unique_ptr<LivenessAnalysis>> functionLiveness;
+	std::vector<std::unique_ptr<LivenessAnalysis>> cfgLiveness;
 
 	std::string toDot() const;
 };
 
 struct ControlFlow
 {
-	std::unique_ptr<SSACFG> mainGraph{std::make_unique<SSACFG>()};
-	std::vector<std::unique_ptr<SSACFG>> functionGraphs{};
-	std::vector<std::tuple<Scope::Function const*, SSACFG const*>> functionGraphMapping{};
+	using FunctionGraphID = std::uint32_t;
 
-	SSACFG const* functionGraph(Scope::Function const* _function)
+	static FunctionGraphID constexpr mainGraphID() noexcept { return 0; }
+
+	SSACFG const* mainGraph() const { return functionGraph(mainGraphID()); }
+
+	SSACFG const* functionGraph(Scope::Function const* _function) const
 	{
-		auto it = std::find_if(functionGraphMapping.begin(), functionGraphMapping.end(), [_function](auto const& tup) { return _function == std::get<0>(tup); });
+		auto it = ranges::find_if(functionGraphMapping, [_function](auto const& tup) { return _function == std::get<0>(tup); });
 		if (it != functionGraphMapping.end())
 			return std::get<1>(*it);
 		return nullptr;
 	}
 
+	SSACFG const* functionGraph(FunctionGraphID const _id) const
+	{
+		return functionGraphs.at(_id).get();
+	}
+
 	std::string toDot(ControlFlowLiveness const* _liveness=nullptr) const
 	{
 		if (_liveness)
 			yulAssert(&_liveness->controlFlow.get() == this);
 		std::ostringstream output;
 		output << "digraph SSACFG {\nnodesep=0.7;\ngraph[fontname=\"DejaVu Sans\"]\nnode[shape=box,fontname=\"DejaVu Sans\"];\n\n";
-		output << mainGraph->toDot(false, std::nullopt, _liveness ? _liveness->mainLiveness.get() : nullptr);
 
 		for (size_t index=0; index < functionGraphs.size(); ++index)
 			output << functionGraphs[index]->toDot(
 				false,
-				index+1,
-				_liveness ? _liveness->functionLiveness[index].get() : nullptr
+				index,
+				_liveness ? _liveness->cfgLiveness[index].get() : nullptr
 			);
 
 		output << "}\n";
 		return output.str();
 	}
+
+	std::vector<std::unique_ptr<SSACFG>> functionGraphs{};
+	std::vector<std::tuple<Scope::Function const*, SSACFG const*>> functionGraphMapping{};
 };
 
 }

libyul/backends/evm/ssa/LivenessAnalysis.cpp

@@ -31,18 +31,11 @@ using namespace solidity::yul::ssa;
 
 namespace
 {
-constexpr auto excludingLiteralsFilter(SSACFG const& _cfg)
+constexpr auto excludingLiteralsFilter()
 {
-	return [&_cfg](LivenessAnalysis::LivenessData::Value const& _valueId) -> bool
+	return [](LivenessAnalysis::LivenessData::Value const& _valueId) -> bool
 	{
-		return !std::holds_alternative<SSACFG::LiteralValue>(_cfg.valueInfo(_valueId));
-	};
-}
-constexpr auto unreachableFilter(SSACFG const& _cfg)
-{
-	return [&_cfg](LivenessAnalysis::LivenessData::Value const& _valueId) -> bool
-	{
-		return std::holds_alternative<SSACFG::UnreachableValue>(_cfg.valueInfo(_valueId));
+		return !_valueId.isLiteral();
 	};
 }
 }
@@ -149,18 +142,18 @@ LivenessAnalysis::LivenessData LivenessAnalysis::blockExitValues(SSACFG::BlockId
 		[](SSACFG::BasicBlock::MainExit const&) {},
 		[&](SSACFG::BasicBlock::FunctionReturn const& _functionReturn)
 		{
-			for (auto const& valueId: _functionReturn.returnValues | ranges::views::filter(excludingLiteralsFilter(m_cfg)))
+			for (auto const& valueId: _functionReturn.returnValues | ranges::views::filter(excludingLiteralsFilter()))
 				result.insert(valueId);
 		},
 		[&](SSACFG::BasicBlock::JumpTable const& _jt)
 		{
-			if (excludingLiteralsFilter(m_cfg)(_jt.value))
+			if (excludingLiteralsFilter()(_jt.value))
 				result.insert(_jt.value);
 		},
 		[](SSACFG::BasicBlock::Jump const&) {},
 		[&](SSACFG::BasicBlock::ConditionalJump const& _conditionalJump)
 		{
-			if (excludingLiteralsFilter(m_cfg)(_conditionalJump.condition))
+			if (excludingLiteralsFilter()(_conditionalJump.condition))
 				result.insert(_conditionalJump.condition);
 		},
 		[](SSACFG::BasicBlock::Terminated const&) {}};
@@ -218,12 +211,11 @@ void LivenessAnalysis::runDagDfs()
 			{
 				for (auto const& phi: m_cfg.block(_successor).phis)
 				{
-					auto const& info = m_cfg.valueInfo(phi);
-					yulAssert(std::holds_alternative<SSACFG::PhiValue>(info), "value info of phi wasn't PhiValue");
-					auto const argIndex = m_cfg.phiArgumentIndex(blockId, _successor);
-					yulAssert(argIndex < std::get<SSACFG::PhiValue>(info).arguments.size());
-					auto const arg = std::get<SSACFG::PhiValue>(info).arguments.at(argIndex);
-					if (!std::holds_alternative<SSACFG::LiteralValue>(m_cfg.valueInfo(arg)))
+					auto const& info = m_cfg.phiInfo(phi);
+					auto const& argIndex = m_cfg.phiArgumentIndex(blockId, _successor);
+					yulAssert(argIndex < info.arguments.size());
+					auto const& arg = info.arguments.at(argIndex);
+					if (!arg.isLiteral())
 						live.insert(arg);
 				}
 			});
@@ -242,11 +234,11 @@ void LivenessAnalysis::runDagDfs()
 			});
 
 		if (std::holds_alternative<SSACFG::BasicBlock::FunctionReturn>(block.exit))
-			for (auto const& returnValue: std::get<SSACFG::BasicBlock::FunctionReturn>(block.exit).returnValues | ranges::views::filter(excludingLiteralsFilter(m_cfg)))
+			for (auto const& returnValue: std::get<SSACFG::BasicBlock::FunctionReturn>(block.exit).returnValues | ranges::views::filter(excludingLiteralsFilter()))
 				live.insert(returnValue);
 
 		// clean out unreachables
-		live.eraseIf([&](auto const& _entry) { return unreachableFilter(m_cfg)(_entry.first); });
+		live.eraseIf([&](auto const& _entry) { return _entry.first.isUnreachable(); });
 
 		// LiveOut(B) <- live
 		m_liveOuts[blockId.value] = live;
@@ -259,9 +251,9 @@ void LivenessAnalysis::runDagDfs()
 			for (auto const& op: block.operations | ranges::views::reverse)
 			{
 				// remove variables defined at p from live
-				live.eraseAll(op.outputs | ranges::views::filter(excludingLiteralsFilter(m_cfg)) | ranges::to<std::vector>);
+				live.eraseAll(op.outputs | ranges::views::filter(excludingLiteralsFilter()) | ranges::to<std::vector>);
 				// add uses at p to live
-				live.insertAll(op.inputs | ranges::views::filter(excludingLiteralsFilter(m_cfg)) | ranges::to<std::vector>);
+				live.insertAll(op.inputs | ranges::views::filter(excludingLiteralsFilter()) | ranges::to<std::vector>);
 			}
 		}
 
@@ -272,7 +264,7 @@ void LivenessAnalysis::runDagDfs()
 	}
 }
 
-void LivenessAnalysis::runLoopTreeDfs(size_t const _loopHeader)
+void LivenessAnalysis::runLoopTreeDfs(SSACFG::BlockId::ValueType const _loopHeader)
 {
 	// SSA Book, Algorithm 9.3
 	if (m_loopNestingForest.loopNodes().contains(_loopHeader))
@@ -286,7 +278,7 @@ void LivenessAnalysis::runLoopTreeDfs(size_t const _loopHeader)
 		// must be live out of header if live in of children
 		m_liveOuts[_loopHeader].maxUnion(liveLoop);
 		// for each blockId \in children(loopHeader)
-		for (size_t blockIdValue = 0; blockIdValue < m_cfg.numBlocks(); ++blockIdValue)
+		for (SSACFG::BlockId::ValueType blockIdValue = 0u; blockIdValue < m_cfg.numBlocks(); ++blockIdValue)
 			if (m_loopNestingForest.loopParents()[blockIdValue] == _loopHeader)
 			{
 				// propagate loop liveness information down to the loop header's children
@@ -313,9 +305,9 @@ void LivenessAnalysis::fillOperationsLiveOut()
 			for (auto const& op: operations | ranges::views::reverse)
 			{
 				*rit = live;
-				for (auto const& output: op.outputs | ranges::views::filter(excludingLiteralsFilter(m_cfg)))
+				for (auto const& output: op.outputs | ranges::views::filter(excludingLiteralsFilter()))
 					live.erase(output);
-				for (auto const& input: op.inputs | ranges::views::filter(excludingLiteralsFilter(m_cfg)))
+				for (auto const& input: op.inputs | ranges::views::filter(excludingLiteralsFilter()))
 					live.insert(input);
 				++rit;
 			}

libyul/backends/evm/ssa/LivenessAnalysis.h

@@ -42,8 +42,7 @@ class LivenessAnalysis
 
 		LivenessData() = default;
 		template<std::input_iterator Iter, std::sentinel_for<Iter> Sentinel>
-		LivenessData(Iter begin, Sentinel end): m_liveCounts(begin, end) {
-		}
+		LivenessData(Iter begin, Sentinel end): m_liveCounts(begin, end) {}
 		explicit LivenessData(LiveCounts&& _liveCounts): m_liveCounts(std::move(_liveCounts)) {}
 
 		bool contains(Value const& _valueId) const;
@@ -113,7 +112,7 @@ class LivenessAnalysis
 
 private:
 	void runDagDfs();
-	void runLoopTreeDfs(std::size_t _loopHeader);
+	void runLoopTreeDfs(SSACFG::BlockId::ValueType _loopHeader);
 	void fillOperationsLiveOut();
 	LivenessData blockExitValues(SSACFG::BlockId const& _blockId) const;