Link basic blocks with phi functions

[#80113672]

evmcc/BasicBlock.cpp

@@ -6,8 +6,9 @@
 namespace evmcc
 {
 
-BasicBlock::BasicBlock(ProgramCounter _instIdx, llvm::Function* _mainFunc)
+BasicBlock::BasicBlock(ProgramCounter _beginInstIdx, llvm::Function* _mainFunc):
-	: m_llvmBB(llvm::BasicBlock::Create(_mainFunc->getContext(), {"Instr.", std::to_string(_instIdx)}, _mainFunc))
+	m_beginInstIdx(_beginInstIdx),
+	m_llvmBB(llvm::BasicBlock::Create(_mainFunc->getContext(), {"Instr.", std::to_string(_beginInstIdx)}, _mainFunc))
 {}
 
 }

evmcc/BasicBlock.h

@@ -1,4 +1,6 @@
 
+#include <vector>
+
 #include <llvm/IR/BasicBlock.h>
 
 namespace evmcc
@@ -9,14 +11,24 @@ using ProgramCounter = uint64_t;
 class BasicBlock
 {
 public:
-	explicit BasicBlock(ProgramCounter _instIdx, llvm::Function* _mainFunc);
+	using State = std::vector<llvm::Value*>;
+
+	explicit BasicBlock(ProgramCounter _beginInstIdx, llvm::Function* _mainFunc);
+
 	BasicBlock(const BasicBlock&) = delete;
 	void operator=(const BasicBlock&) = delete;
 
 	operator llvm::BasicBlock*() { return m_llvmBB; }
+	llvm::BasicBlock* llvm() { return m_llvmBB; }
+
+	State& getState() { return m_state; }
 
 private:
+	ProgramCounter m_beginInstIdx;
 	llvm::BasicBlock* m_llvmBB;
+
+	/// Basic black state vector - current/end values and their positions
+	State m_state;
 };
 
 }

evmcc/Compiler.cpp

@@ -2,6 +2,7 @@
 #include "Compiler.h"
 
 #include <llvm/IR/IRBuilder.h>
+#include <llvm/IR/CFG.h>
 
 #include <libevmface/Instruction.h>
 
@@ -183,8 +184,8 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 	auto memory = Memory(builder, module.get());
 	auto ext = Ext(builder, module.get());
 
-	llvm::BasicBlock* currentBlock = entryBlock;
+	BasicBlock* currentBlock = &basicBlocks.find(0)->second; // Any value, just to create branch for %entry to %Instr.0
-	BBStack stack(extStack); // Stack for current block
+	BBStack stack(builder, extStack); // Stack for current block
 
 	for (auto pc = bytecode.cbegin(); pc != bytecode.cend(); ++pc)
 	{
@@ -192,6 +193,7 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 
 		ProgramCounter currentPC = pc - bytecode.cbegin();
 
+		// Change basic block
 		auto blockIter = basicBlocks.find(currentPC);
 		if (blockIter != basicBlocks.end())
 		{
@@ -201,8 +203,8 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 				builder.CreateBr(nextBlock);
 			// Insert the next block into the main function.
 			builder.SetInsertPoint(nextBlock);
-			currentBlock = nextBlock;
+			currentBlock = &nextBlock;
-			assert(stack.empty()); // Stack should be empty
+			stack.setBasicBlock(*currentBlock);
 		}
 
 		assert(currentBlock != nullptr);
@@ -569,7 +571,6 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 			// The target address is computed at compile time,
 			// just pop it without looking...
 			stack.pop();
-			stack.reset();
 
 			auto targetBlock = jumpTargets[currentPC];
 			builder.CreateBr(targetBlock);
@@ -589,7 +590,6 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 			auto top = stack.pop();
 			auto zero = ConstantInt::get(Types.word256, 0);
 			auto cond = builder.CreateICmpNE(top, zero, "nonzero");
-			stack.reset();
 			auto targetBlock = jumpTargets[currentPC];
 			auto& followBlock = basicBlocks.find(currentPC + 1)->second;
 			builder.CreateCondBr(cond, targetBlock, followBlock);
@@ -756,7 +756,6 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 			ret = builder.CreateOr(ret, size);
 
 			builder.CreateRet(ret);
-			stack.clear();
 			currentBlock = nullptr;
 			break;
 		}
@@ -770,7 +769,6 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 		case Instruction::STOP:
 		{
 			builder.CreateRet(builder.getInt64(0));
-			stack.clear();
 			currentBlock = nullptr;
 			break;
 		}
@@ -790,6 +788,39 @@ std::unique_ptr<llvm::Module> Compiler::compile(const dev::bytes& bytecode)
 	builder.SetInsertPoint(finalBlock);
 	builder.CreateRet(builder.getInt64(0));
 
+	auto findBB = [this](llvm::BasicBlock* _llvmBB)
+	{
+		for (auto&& bb : basicBlocks)
+		{
+			if (bb.second.llvm() == _llvmBB)
+				return &bb.second;
+		}
+		return (BasicBlock*)nullptr;
+	};
+
+	// Link basic blocks
+	for (auto&& p : basicBlocks)
+	{
+		BasicBlock& bb = p.second;
+		llvm::BasicBlock* llvmBB = bb.llvm();
+
+		size_t i = 0;
+		for (auto& inst : *llvmBB)
+		{
+			if (auto phi = llvm::dyn_cast<llvm::PHINode>(&inst))
+			{
+				for (auto preIt = llvm::pred_begin(llvmBB); preIt != llvm::pred_end(llvmBB); ++preIt)
+				{
+					llvm::BasicBlock* preBB = *preIt;
+					auto pbb = findBB(preBB);
+					assert(i < pbb->getState().size()); // TODO: Report error
+					phi->addIncoming(*(pbb->getState().rbegin() + i), preBB);
+				}
+				++i;
+			}
+		}
+	}
+
 	return module;
 }
 

evmcc/Stack.cpp

@@ -9,6 +9,7 @@
 
 #include <llvm/IR/Function.h>
 
+#include "BasicBlock.h"
 #include "Runtime.h"
 
 #ifdef _MSC_VER
@@ -20,52 +21,55 @@
 namespace evmcc
 {
 
-BBStack::BBStack(Stack& _extStack)
+BBStack::BBStack(llvm::IRBuilder<>& _builder, Stack& _extStack):
-	: m_extStack(_extStack)
+	m_extStack(_extStack),
+	m_builder(_builder)
 {}
 
 void BBStack::push(llvm::Value* _value)
 {
-	m_state.push_back(_value);
+	m_block->getState().push_back(_value);
 }
 
 llvm::Value* BBStack::pop()
 {
-	if (m_state.empty())
+	auto&& state = m_block->getState();
-		return m_extStack.pop();
+	if (state.empty())
+	{
+		// Create PHI node
+		auto first = m_block->llvm()->getFirstNonPHI();
+		auto llvmBB = m_block->llvm();
+		if (llvmBB->getInstList().empty())
+			return llvm::PHINode::Create(m_builder.getIntNTy(256), 0, {}, m_block->llvm());
+		return llvm::PHINode::Create(m_builder.getIntNTy(256), 0, {}, llvmBB->getFirstNonPHI());
+	}
 
-	auto top = m_state.back();
+	auto top = state.back();
-	m_state.pop_back();
+	state.pop_back();
 	return top;
 }
 
-void BBStack::reset()
+void BBStack::setBasicBlock(BasicBlock& _newBlock)
 {
-	for (auto&& value : m_state)
+	// Current block keeps end state
-		m_extStack.push(value);
+	// Just update pointer to current block
-	m_state.clear();
+	// New block should have empty state
-}
+	assert(_newBlock.getState().empty()); 
-
+	m_block = &_newBlock;
-void BBStack::clear()
-{
-	m_state.clear();
-}
-
-bool BBStack::empty() const
-{
-	return m_state.empty();
 }
 
 void BBStack::dup(size_t _index)
 {
-	auto value = *(m_state.rbegin() + _index);
+	auto&& state = m_block->getState();
-	m_state.push_back(value);
+	auto value = *(state.rbegin() + _index);
+	state.push_back(value);
 }
 
 void BBStack::swap(size_t _index)
 {
 	assert(_index != 0);
-	std::swap(*m_state.rbegin(), *(m_state.rbegin() + _index));
+	auto&& state = m_block->getState();
+	std::swap(*state.rbegin(), *(state.rbegin() + _index));
 }
 
 Stack::Stack(llvm::IRBuilder<>& _builder, llvm::Module* _module)

evmcc/Stack.h

@@ -7,6 +7,7 @@
 
 namespace evmcc
 {
+class BasicBlock;
 
 class Stack
 {
@@ -36,7 +37,7 @@ private:
 class BBStack
 {
 public:
-	BBStack(Stack& _extStack);
+	BBStack(llvm::IRBuilder<>& _builder, Stack& _extStack);
 
 	void push(llvm::Value* _value);
 	llvm::Value* pop();
@@ -44,23 +45,14 @@ public:
 	void swap(size_t _index);
 
 	/**
-	 Resets stack on basic block change.
+	 Changes current basic block with a new one with empty state.
-	 Values left on local stack are pushed on external stack.
-	 Local stack is empty after this operation and compilation of new basic block can be started.
 	 */
-	void reset();
+	void setBasicBlock(BasicBlock& _newBlock);
-
-	/**
-	 Dumps values on stack.
-	 */
-	void clear();
-
-	/// Debug only
-	bool empty() const;
 
 private:
-	std::vector<llvm::Value*> m_state;	///< Basic black state vector - current values and their positions
 	Stack& m_extStack;                  ///< External (global) stack
+	BasicBlock* m_block = nullptr;		///< Current basic block
+	llvm::IRBuilder<>& m_builder;
 };