Propagation of values between basic blocks (and the stack): initial implementation (probably buggy, but simple cases work).

[#80895676]

libevmjit/BasicBlock.cpp

@@ -30,20 +30,20 @@ BasicBlock::BasicBlock(std::string _name, llvm::Function* _mainFunc) :
 {}
 
 
-void BasicBlock::Stack::push(llvm::Value* _value)
+void BasicBlock::LocalStack::push(llvm::Value* _value)
 {
 	assert(_value->getType() == Type::i256);
 	m_backend.push_back(_value);
 }
 
-llvm::Value* BasicBlock::Stack::pop()
+llvm::Value* BasicBlock::LocalStack::pop()
 {
 	auto top = get(0);
 	m_backend.pop_back();
 	return top;
 }
 
-llvm::Value* BasicBlock::Stack::get(size_t _index)
+llvm::Value* BasicBlock::LocalStack::get(size_t _index)
 {	
 	if (_index >= m_backend.size())
 	{
@@ -55,18 +55,19 @@ llvm::Value* BasicBlock::Stack::get(size_t _index)
 			m_backend[i] = m_llvmBB->empty() ?
 				llvm::PHINode::Create(Type::i256, 0, {}, m_llvmBB) :
 				llvm::PHINode::Create(Type::i256, 0, {}, m_llvmBB->getFirstNonPHI());
+			m_numRequiredStackItems += 1;
 		}
 	}
 
 	return *(m_backend.rbegin() + _index);
 }
 
-void BasicBlock::Stack::dup(size_t _index)
+void BasicBlock::LocalStack::dup(size_t _index)
 {
 	m_backend.push_back(get(_index));
 }
 
-void BasicBlock::Stack::swap(size_t _index)
+void BasicBlock::LocalStack::swap(size_t _index)
 {
 	assert(_index != 0);
 	get(_index); // Create PHI nodes

libevmjit/BasicBlock.h

@@ -15,7 +15,7 @@ using ProgramCounter = uint64_t; // TODO: Rename
 class BasicBlock
 {
 public:
-	class Stack
+	class LocalStack
 	{
 	public:
 		/// Pushes value on stack
@@ -37,17 +37,23 @@ public:
 		/// Size of the stack
 		size_t size() const { return m_backend.size(); }
 
+		size_t initialSize() const { return m_numRequiredStackItems; }
+
 	private:
-		Stack(llvm::BasicBlock* _llvmBB) : m_llvmBB(_llvmBB) {}
-		Stack(const Stack&) = delete;
-		void operator=(const Stack&) = delete;
+		// LocalStack(llvm::BasicBlock* _llvmBB) : m_llvmBB(_llvmBB) {}
+		LocalStack(llvm::BasicBlock* _llvmBB) : m_llvmBB(_llvmBB), m_numRequiredStackItems(0) {}
+		LocalStack(LocalStack const&) = delete;
+		void operator=(LocalStack const&) = delete;
 		friend BasicBlock;
 
 	private:
 		std::vector<llvm::Value*> m_backend;
 
-		/// LLVM Basic Block where phi nodes are inserted
-		llvm::BasicBlock* const m_llvmBB;
+		/** Basic block into which phi nodes are inserted */
+		llvm::BasicBlock* m_llvmBB;
+
+		/** Number of items required on the EVM stack at the beginning of the block */
+		size_t m_numRequiredStackItems;
 	};
 
 	/// Basic block name prefix. The rest is beging instruction index.
@@ -62,7 +68,7 @@ public:
 	operator llvm::BasicBlock*() { return m_llvmBB; }
 	llvm::BasicBlock* llvm() { return m_llvmBB; }
 
-	Stack& getStack() { return m_stack; }
+	LocalStack& getStack() { return m_stack; }
 
 	ProgramCounter begin() { return m_beginInstIdx; }
 	ProgramCounter end() { return m_endInstIdx; }
@@ -74,7 +80,7 @@ private:
 
 	/// Basic black state vector (stack) - current/end values and their positions on stack
 	/// @internal Must be AFTER m_llvmBB
-	Stack m_stack;
+	LocalStack m_stack;
 };
 
 }

libevmjit/Compiler.cpp

@@ -12,6 +12,7 @@
 
 #include "Type.h"
 #include "Memory.h"
+#include "Stack.h"
 #include "Ext.h"
 #include "GasMeter.h"
 #include "Utils.h"
@@ -172,6 +173,7 @@ std::unique_ptr<llvm::Module> Compiler::compile(bytesConstRef bytecode)
 	GasMeter gasMeter(m_builder);
 	Memory memory(m_builder, gasMeter);
 	Ext ext(m_builder);
+	Stack stack(m_builder);
 
 	m_builder.CreateBr(basicBlocks.begin()->second);
 
@@ -214,7 +216,7 @@ std::unique_ptr<llvm::Module> Compiler::compile(bytesConstRef bytecode)
 		m_builder.CreateBr(m_badJumpBlock->llvm());
 	}
 
-	linkBasicBlocks();
+	linkBasicBlocks(stack);
 
 	return module;
 }
@@ -868,61 +870,111 @@ void Compiler::compileBasicBlock(BasicBlock& basicBlock, bytesConstRef bytecode,
 }
 
 
-void Compiler::linkBasicBlocks()
+void Compiler::linkBasicBlocks(Stack& stack)
 {
-	/// Helper function that finds basic block given LLVM basic block pointer
-	auto findBasicBlock = [this](llvm::BasicBlock* _llbb) -> BasicBlock*
+	struct BBInfo
 	{
-		// TODO: Fix for finding jumpTableBlock
-		if (_llbb == this->m_jumpTableBlock->llvm())
-			return this->m_jumpTableBlock.get();
-
-		for (auto&& bb : this->basicBlocks)
-			if (_llbb == bb.second.llvm())
-				return &bb.second;
-		return nullptr;
-
-		// Name is used to get basic block index (index of first instruction)
-		// TODO: If basicBlocs are still a map - multikey map can be used
-		//auto&& idxStr = _llbb->getName().substr(sizeof(BasicBlock::NamePrefix) - 2);
-		//auto idx = std::stoul(idxStr);
-		//return basicBlocks.find(idx)->second;
+		BasicBlock& bblock;
+		std::vector<BBInfo*> predecessors;
+		size_t inputItems;
+		size_t outputItems;
+
+		BBInfo(BasicBlock& _bblock)
+			: bblock(_bblock),
+			  predecessors(),
+			  inputItems(_bblock.getStack().initialSize()),
+			  outputItems(_bblock.getStack().size())
+		{}
 	};
 
-	auto completePhiNodes = [findBasicBlock](llvm::BasicBlock* _llbb) -> void
+	std::map<llvm::BasicBlock*, BBInfo> cfg;
+
+	// Create nodes in cfg
+	for (auto& pair : this->basicBlocks)
+	{
+		auto& bb = pair.second;
+		cfg.emplace(std::piecewise_construct,
+		            std::forward_as_tuple(bb.llvm()),
+		            std::forward_as_tuple(bb));
+	}
+
+	auto& entryBlock = m_mainFunc->getEntryBlock();
+
+	// Create edges in cfg
+	for (auto& pair : cfg)
+	{
+		auto bbPtr = pair.first;
+		auto& bbInfo = pair.second;
+
+		for (auto predIt = llvm::pred_begin(bbPtr); predIt != llvm::pred_end(bbPtr); ++predIt)
+		{
+			if (*predIt != &entryBlock)
+				bbInfo.predecessors.push_back(&cfg.find(*predIt)->second);
+		}
+	}
+
+	// Iteratively compute inputs and outputs of each block, until reaching fixpoint.
+	bool valuesChanged = true;
+	while (valuesChanged)
 	{
-		size_t valueIdx = 0;
-		auto firstNonPhi = _llbb->getFirstNonPHI();
-		for (auto instIt = _llbb->begin(); &*instIt != firstNonPhi; ++instIt, ++valueIdx)
+		valuesChanged = false;
+		for (auto& pair : cfg)
 		{
-			auto phi = llvm::cast<llvm::PHINode>(instIt);
-			for (auto predIt = llvm::pred_begin(_llbb); predIt != llvm::pred_end(_llbb); ++predIt)
+			auto& bbInfo = pair.second;
+
+			for (auto predInfo : bbInfo.predecessors)
 			{
-				// TODO: In case entry block is reached - report error
-				auto predBB = findBasicBlock(*predIt);
-				if (!predBB)
+				if (predInfo->outputItems < bbInfo.inputItems)
+				{
+					bbInfo.inputItems = predInfo->outputItems;
+					valuesChanged = true;
+				}
+				else if (predInfo->outputItems > bbInfo.inputItems)
 				{
-					std::cerr << "Stack too small in " << _llbb->getName().str() << std::endl;
-					std::exit(1);
+					predInfo->outputItems = bbInfo.inputItems;
+					valuesChanged = true;
 				}
-				auto value = predBB->getStack().get(valueIdx);
-				phi->addIncoming(value, predBB->llvm());
 			}
 		}
-	};
+	}
 
-	
-	// TODO: It is crappy visiting of basic blocks.
-	llvm::SmallPtrSet<llvm::BasicBlock*, 32> visitSet;
-	for (auto&& bb : basicBlocks) // TODO: External loop is to visit unreable blocks that can also have phi nodes
+	// Propagate values between blocks.
+	for (auto& pair : cfg)
 	{
-		for (auto it = llvm::po_ext_begin(bb.second.llvm(), visitSet),
-			end = llvm::po_ext_end(bb.second.llvm(), visitSet); it != end; ++it)
+		auto  llbb = pair.first;
+		auto& bbInfo = pair.second;
+		auto& bblock = bbInfo.bblock;
+
+		// Complete phi nodes for the top bbInfo.inputItems placeholder values
+		auto instrIter = llbb->begin();
+		for (size_t valueIdx = 0; valueIdx < bbInfo.inputItems; ++instrIter, ++valueIdx)
 		{
-			// TODO: Use logger
-			//std::cerr << it->getName().str() << std::endl;
-			completePhiNodes(*it);
+			auto phi = llvm::cast<llvm::PHINode>(instrIter);
+			for (auto predIt : bbInfo.predecessors)
+			{
+				assert(valueIdx < predIt->bblock.getStack().size());
+				auto value = predIt->bblock.getStack().get(valueIdx);
+				phi->addIncoming(value, predIt->bblock.llvm());
+			}
+		}
+
+		// Turn the remaining phi nodes into stack.pop's.
+		m_builder.SetInsertPoint(llbb, llvm::BasicBlock::iterator(llbb->getFirstNonPHI()));
+		for (; llvm::isa<llvm::PHINode>(*instrIter); )
+		{
+			auto phi = llvm::cast<llvm::PHINode>(instrIter);
+			auto value = stack.popWord();
+			phi->replaceAllUsesWith(value);
+			++ instrIter;
+			phi->eraseFromParent();
 		}
+
+		// Emit stack push's at the end of the block, just before the terminator;
+		m_builder.SetInsertPoint(llbb, -- llbb->end());
+		auto localStackSize = bblock.getStack().size();
+		assert(localStackSize >= bbInfo.outputItems);
+		for (size_t i = 0; i < localStackSize - bbInfo.outputItems; ++i)
+			stack.pushWord(bblock.getStack().get(localStackSize - 1 - i));
 	}
 
 	// Remove jump table block if not predecessors

libevmjit/Compiler.h

@@ -33,7 +33,7 @@ private:
 
 	void compileBasicBlock(BasicBlock& basicBlock, bytesConstRef bytecode, class Memory& memory, class Ext& ext, class GasMeter& gasMeter, llvm::BasicBlock* nextBasicBlock);
 
-	void linkBasicBlocks();
+	void linkBasicBlocks(class Stack& stack);
 
 
 	llvm::IRBuilder<> m_builder;

libevmjit/Stack.cpp

@@ -0,0 +1,69 @@
+#include "Stack.h"
+#include "Runtime.h"
+
+#include <llvm/IR/Function.h>
+#include <llvm/IR/TypeBuilder.h>
+
+namespace dev
+{
+namespace eth
+{
+namespace jit
+{
+
+Stack::Stack(llvm::IRBuilder<>& _builder)
+	: CompilerHelper(_builder)
+{
+	auto i256Ty = m_builder.getIntNTy(256);
+	auto i256PtrTy = i256Ty->getPointerTo();
+
+	m_arg = m_builder.CreateAlloca(i256Ty, nullptr, "stack.retVal");
+
+	using namespace llvm;
+	using Linkage = GlobalValue::LinkageTypes;
+
+	auto module = getModule();
+	m_push = Function::Create(FunctionType::get(m_builder.getVoidTy(), i256PtrTy, false), Linkage::ExternalLinkage, "stack_push", module);
+	m_pop = Function::Create(FunctionType::get(m_builder.getVoidTy(), i256PtrTy, false), Linkage::ExternalLinkage, "stack_pop", module);
+}
+
+Stack::~Stack()
+{}
+
+llvm::Instruction* Stack::popWord()
+{
+	m_builder.CreateCall(m_pop, m_arg);
+	return m_builder.CreateLoad(m_arg);
+}
+
+void Stack::pushWord(llvm::Value* _word)
+{
+	m_builder.CreateStore(_word, m_arg);
+	m_builder.CreateCall(m_push, m_arg);
+}
+
+}
+}
+}
+
+extern "C"
+{
+
+using namespace dev::eth::jit;
+
+EXPORT void stack_pop(i256* _ret)
+{
+	auto& stack = Runtime::getStack();
+	assert(stack.size() > 0);
+	*_ret = stack.back();
+	stack.pop_back();
+}
+
+EXPORT void stack_push(i256* _word)
+{
+	auto& stack = Runtime::getStack();
+	stack.push_back(*_word);
+}
+
+} // extern "C"
+

libevmjit/Stack.h

@@ -0,0 +1,35 @@
+#pragma once
+
+#include "CompilerHelper.h"
+
+#include <llvm/IR/Module.h>
+
+namespace dev
+{
+namespace eth
+{
+namespace jit
+{
+
+class Stack : public CompilerHelper
+{
+public:
+	Stack(llvm::IRBuilder<>& builder);
+	virtual ~Stack();
+
+	void pushWord(llvm::Value* _word);
+	llvm::Instruction* popWord();
+
+private:
+	llvm::Function* m_push;
+	llvm::Function* m_pop;
+
+	llvm::Value* m_arg;
+};
+
+
+}
+}
+}
+
+