Generates mov reg, reg

src/binary_translation/CMakeLists.txt

@@ -3,6 +3,7 @@ set(SRCS
 			CodeGen.cpp
 			ModuleGen.cpp
 			Disassembler.cpp
+			InstructionBlock.cpp
 			
 			Instructions/Instruction.cpp
 			Instructions/DataProcessing.cpp
@@ -11,6 +12,7 @@ set(HEADERS
 			CodeGen.h
 			ModuleGen.h
 			Disassembler.h
+			InstructionBlock.h
 			
 			Instructions/Types.h
 			Instructions/Instruction.h

src/binary_translation/InstructionBlock.cpp

@@ -0,0 +1,49 @@
+#include "InstructionBlock.h"
+#include "ModuleGen.h"
+#include "Instructions/Instruction.h"
+#include <sstream>
+#include <iomanip>
+
+InstructionBlock::InstructionBlock(ModuleGen* module, Instruction* instruction)
+    : module(module),
+      instruction(std::unique_ptr<Instruction>(instruction))
+{
+    std::stringstream ss;
+    ss << std::hex << std::setfill('0') << std::setw(8) << instruction->Address();
+    address_string = ss.str();
+}
+
+InstructionBlock::~InstructionBlock()
+{
+}
+
+void InstructionBlock::GenerateEntryBlock()
+{
+    entry_basic_block = llvm::BasicBlock::Create(llvm::getGlobalContext(), address_string + "_Entry");
+}
+
+void InstructionBlock::GenerateCode()
+{
+    Module()->IrBuilder()->SetInsertPoint(entry_basic_block);
+
+    instruction->GenerateCode(this);
+
+    exit_basic_block = Module()->IrBuilder()->GetInsertBlock();
+}
+
+llvm::Value *InstructionBlock::Read(Register reg)
+{
+    auto ib = module->IrBuilder();
+    return ib->CreateAlignedLoad(module->GetRegisterPtr(reg), 4);
+}
+
+llvm::Value *InstructionBlock::Write(Register reg, llvm::Value *value)
+{
+    auto ib = module->IrBuilder();
+    return ib->CreateAlignedStore(value, module->GetRegisterPtr(reg), 4);
+}
+
+size_t InstructionBlock::Address()
+{
+    return instruction->Address();
+}

src/binary_translation/InstructionBlock.h

@@ -0,0 +1,64 @@
+#include <memory>
+#include <string>
+
+namespace llvm
+{
+    class Value;
+    class BasicBlock;
+}
+
+class ModuleGen;
+class Instruction;
+
+enum class Register;
+
+/*
+ * An instruction blocks
+ * Holds the entry and exit points for an instruction
+ * Responsible to generate the code
+ */
+class InstructionBlock
+{
+public:
+    InstructionBlock(ModuleGen *module, Instruction *instruction);
+    ~InstructionBlock();
+
+    /*
+    * Generates the basic block of the instruction
+    */
+    void GenerateEntryBlock();
+
+    /*
+     * Generates the code for the instruction
+     */
+    void GenerateCode();
+
+    /*
+     * Generates code to read the register
+     */
+    llvm::Value *Read(Register reg);
+    /*
+     * Generates code to write the value
+     * Returns the write instruction = written value
+     */
+    llvm::Value *Write(Register reg, llvm::Value *value);
+
+    size_t Address();
+    ModuleGen *Module() { return module; }
+
+    llvm::BasicBlock *GetEntryBasicBlock() { return entry_basic_block; }
+    llvm::BasicBlock *GetExitBasicBlock() { return exit_basic_block; }
+private:
+    // Textual representation of the address
+    // Used to generate names
+    std::string address_string;
+
+    ModuleGen *module;
+    std::unique_ptr<Instruction> instruction;
+
+    // The block at the entry to instruction
+    llvm::BasicBlock *entry_basic_block;
+
+    // The block at the exit from the instruction
+    llvm::BasicBlock *exit_basic_block;
+};

src/binary_translation/Instructions/DataProcessing.cpp

@@ -1,15 +1,35 @@
 #include "DataProcessing.h"
 #include "Disassembler.h"
+#include "InstructionBlock.h"
 
 static RegisterInstruction<DataProcessing> register_instruction;
 
 bool DataProcessing::Decode()
 {
-    if (ReadFields({ FieldDef<4>(&cond), FieldDef<3>(1), FieldDef<4>(&short_op), FieldDef<1>(&s), FieldDef<4>(&rn),
+    // Mov and shifts must have zeroes at some operands of different data processing instructions
-                     FieldDef<4>(&rd), FieldDef<12>(&imm12) }))
+    if (ReadFields({ FieldDef<4>(&cond), FieldDef<3>(0), FieldDef<4>((u32)ShortOpType::MoveAndShifts), FieldDef<1>(&s), FieldDef<4>(0),
+                     FieldDef<4>(&rd), FieldDef<5>(&imm5), FieldDef<3>(0), FieldDef<4>(&rm) }))
     {
-        form = Form::Immediate;
+        form = Form::Register;
+        if (imm5 != 0) return false; // Shifts
+        if (s != 0) return false; // Set flags
+        if (rd == Register::PC) return false; // Jump
         return true;
     }
+    if (ReadFields({ FieldDef<4>(&cond), FieldDef<3>(1), FieldDef<4>(&short_op), FieldDef<1>(&s), FieldDef<4>(&rn),
+        FieldDef<4>(&rd), FieldDef<12>(&imm12) }))
+    {
+        // TODO: not implemented
+        form = Form::Immediate;
+        return false;
+    }
     return false;
 }
+
+void DataProcessing::GenerateCode(InstructionBlock* instruction_block)
+{
+    // Currently supports only mov reg, reg
+
+    auto value = instruction_block->Read(rm);
+    instruction_block->Write(rd, value);
+}

src/binary_translation/Instructions/DataProcessing.h

@@ -16,7 +16,7 @@ public:
     {
         BitwiseAnd = 0, BitwiseXor, Subtract, RevSubtract, Add, AddWithCarry, SubtractWithCarry, ReverseSubtractWithCarry,
         // Compare, Test, Misc
-        BitwiseOr = 12, Move, BitwiseBitClear, BitwiseNot
+        BitwiseOr = 12, MoveAndShifts, BitwiseBitClear, BitwiseNot
     };
     enum class Form
     {
@@ -25,6 +25,7 @@ public:
 
 public:
     virtual bool Decode() override;
+    void GenerateCode(InstructionBlock* instruction_block) override;
 private:
     Form form;
     Condition cond;
@@ -32,5 +33,7 @@ private:
     bool s;
     Register rn;
     Register rd;
+    Register rm;
     u32 imm12;
+    u32 imm5;
 };

src/binary_translation/Instructions/Instruction.h

@@ -2,6 +2,8 @@
 #include "common/common_types.h"
 #include <initializer_list>
 
+class InstructionBlock;
+
 class Instruction
 {
 protected:
@@ -15,9 +17,17 @@ public:
      * Returns true on success, or false otherwise
      */
     bool Read(u32 instruction, u32 address);
+
+    /*
+     * Generates code for the instruction into the instruction block
+     * Derived classes must override this
+     */
+    virtual void GenerateCode(InstructionBlock *instruction_block) = 0;
+
+    u32 Address() { return address; }
 protected:
     /*
-     * Derived classes should override this, and implement it by calling ReadFields
+     * Derived classes must override this, and implement it by calling ReadFields
      */
     virtual bool Decode() = 0;
     /*
@@ -36,7 +46,6 @@ protected:
      */
     template<size_t BitCount, typename Type>
     static FieldDefObject FieldDef(Type *field);
-
 private:
     /*
      * Function used by FieldDefObject to write to a field

src/binary_translation/ModuleGen.cpp

@@ -3,8 +3,11 @@
 #include "core/loader/loader.h"
 #include "core/mem_map.h"
 #include "Instructions/Instruction.h"
+#include "Instructions/Types.h"
+#include "InstructionBlock.h"
 #include <llvm/IR/Function.h>
 #include <llvm/IR/GlobalVariable.h>
+#include <stack>
 
 using namespace llvm;
 
@@ -21,10 +24,37 @@ ModuleGen::~ModuleGen()
 void ModuleGen::Run()
 {
     GenerateGlobals();
+
+    DecodeInstructions();
+    GenerateInstructionsEntry();
+
     GenerateCanRunFunction();
     GenerateRunFunction();
+    GenerateGetBlockAddressFunction();
 
-    GenerateBlocks();
+    GenerateInstructionsCode();
+    GenerateInstructionsTermination();
+    AddInstructionsToRunFunction();
+
+    GenerateBlockAddressArray();
+}
+
+Value *ModuleGen::GetRegisterPtr(Register reg)
+{
+    Value *global;
+    unsigned index;
+    if (reg <= Register::PC)
+    {
+        global = registers_global;
+        index = static_cast<unsigned>(reg)-static_cast<unsigned>(Register::R0);
+    }
+    else
+    {
+        global = flags_global;
+        index = static_cast<unsigned>(reg)-static_cast<unsigned>(Register::N);
+    }
+    auto base = ir_builder->CreateAlignedLoad(global, 4);
+    return ir_builder->CreateConstInBoundsGEP1_32(base, index);
 }
 
 void ModuleGen::GenerateGlobals()
@@ -35,36 +65,193 @@ void ModuleGen::GenerateGlobals()
     // Flags is stored internally as i1* indexed in multiples of 4
     auto flags_global_initializer = ConstantPointerNull::get(IntegerType::getInt1PtrTy(getGlobalContext()));
     flags_global = new GlobalVariable(*module, flags_global_initializer->getType(), false, GlobalValue::ExternalLinkage, flags_global_initializer, "Flags");
+
+    auto get_block_address_function_type = FunctionType::get(ir_builder->getInt8PtrTy(), ir_builder->getInt32Ty(), false);
+    get_block_address_function = Function::Create(get_block_address_function_type, GlobalValue::PrivateLinkage, "GetBlockAddress", module);
+
+    auto can_run_function_type = FunctionType::get(ir_builder->getInt1Ty(), false);
+    can_run_function = Function::Create(can_run_function_type, GlobalValue::ExternalLinkage, "CanRun", module);
+
+    auto run_function_type = FunctionType::get(ir_builder->getVoidTy(), false);
+    run_function = Function::Create(run_function_type, GlobalValue::ExternalLinkage, "Run", module);
+
+    block_address_array_base = Loader::ROMCodeStart / 4;
+    block_address_array_size = Loader::ROMCodeSize / 4;
+
+    block_address_array_type = ArrayType::get(ir_builder->getInt8PtrTy(), block_address_array_size);
+    block_address_array = new GlobalVariable(*module, block_address_array_type, true, GlobalValue::ExternalLinkage, nullptr, "BlockAddressArray");
+}
+
+void ModuleGen::GenerateBlockAddressArray()
+{
+    auto local_block_address_array_values = std::make_unique<Constant*[]>(block_address_array_size);
+    
+    std::fill(
+        local_block_address_array_values.get(),
+        local_block_address_array_values.get() + block_address_array_size,
+        ConstantPointerNull::get(ir_builder->getInt8PtrTy()));
+
+    for (auto i = 0; i < instruction_blocks.size(); ++i)
+    {
+        auto &block = instruction_blocks[i];
+        auto entry_basic_block = block->GetEntryBasicBlock();
+        auto index = block->Address() / 4 - block_address_array_base;
+        local_block_address_array_values[index] = BlockAddress::get(entry_basic_block->getParent(), entry_basic_block);
+    }
+
+    auto local_block_address_array_values_ref = ArrayRef<Constant*>(local_block_address_array_values.get(), block_address_array_size);
+    auto local_blocks_address_array = ConstantArray::get(block_address_array_type, local_block_address_array_values_ref);
+    block_address_array->setInitializer(local_blocks_address_array);
+}
+
+void ModuleGen::GenerateGetBlockAddressFunction()
+{
+    /*
+    entry_basic_block:
+        auto index = (pc - block_address_array_base) / 4;
+        if(index < block_address_array_size)
+        {
+    index_in_bounds_basic_block:
+            return block_address_array[index];
+        }
+        else
+        {
+    index_out_of_bounds_basic_block:
+            return nullptr;
+        }
+    */
+    auto pc = &*get_block_address_function->arg_begin();
+    auto entry_basic_block = BasicBlock::Create(getGlobalContext(), "Entry", get_block_address_function);
+    auto index_in_bounds_basic_block = BasicBlock::Create(getGlobalContext(), "IndexInBounds", get_block_address_function);
+    auto index_out_of_bounds_basic_block = BasicBlock::Create(getGlobalContext(), "IndexOutOfBounds", get_block_address_function);
+
+    ir_builder->SetInsertPoint(entry_basic_block);
+    auto index = ir_builder->CreateUDiv(pc, ir_builder->getInt32(4));
+    index = ir_builder->CreateSub(index, ir_builder->getInt32(block_address_array_base));
+    auto in_bounds_pred = ir_builder->CreateICmpULT(index, ir_builder->getInt32(block_address_array_size));
+    ir_builder->CreateCondBr(in_bounds_pred, index_in_bounds_basic_block, index_out_of_bounds_basic_block);
+
+    ir_builder->SetInsertPoint(index_in_bounds_basic_block);
+    Value *gep_values[] = { ir_builder->getInt32(0), index };
+    auto block_address = ir_builder->CreateLoad(ir_builder->CreateInBoundsGEP(block_address_array, gep_values));
+    ir_builder->CreateRet(block_address);
+
+    ir_builder->SetInsertPoint(index_out_of_bounds_basic_block);
+    ir_builder->CreateRet(ConstantPointerNull::get(ir_builder->getInt8PtrTy()));
 }
 
 void ModuleGen::GenerateCanRunFunction()
 {
-    auto can_run_function_type = FunctionType::get(ir_builder->getInt1Ty(), false);
+    // return GetBlockAddress(Read(PC)) != nullptr;
-    can_run_function = Function::Create(can_run_function_type, GlobalValue::ExternalLinkage, "CanRun", module);
     auto basic_block = BasicBlock::Create(getGlobalContext(), "Entry", can_run_function);
 
     ir_builder->SetInsertPoint(basic_block);
-    ir_builder->CreateRet(ir_builder->getInt1(false));
+    auto block_address = ir_builder->CreateCall(get_block_address_function, ir_builder->CreateAlignedLoad(GetRegisterPtr(Register::PC), 4));
+    ir_builder->CreateRet(ir_builder->CreateICmpNE(block_address, ConstantPointerNull::get(ir_builder->getInt8PtrTy())));
 }
 
 void ModuleGen::GenerateRunFunction()
 {
-    auto run_function_type = FunctionType::get(ir_builder->getVoidTy(), false);
+    /*
-    run_function = Function::Create(run_function_type, GlobalValue::ExternalLinkage, "Run", module);
+    run_function_entry:
-    auto basic_block = BasicBlock::Create(getGlobalContext(), "Entry", run_function);
+    run_function_re_entry:
+        auto block_address = GetBlockAddress(Read(PC))
+        if(index != nullptr)
+        {
+    block_present_basic_block:
+            goto block_address;
+            return;
+        }
+        else
+        {
+    block_not_present_basic_block:
+            return;
+        }
+    */
+    run_function_entry = BasicBlock::Create(getGlobalContext(), "Entry", run_function);
+    // run_function_re_entry is needed because it isn't possible to jump to the first block of a function
+    run_function_re_entry = BasicBlock::Create(getGlobalContext(), "ReEntry", run_function);
+    auto block_present_basic_block = BasicBlock::Create(getGlobalContext(), "BlockPresent", run_function);
+    auto block_not_present_basic_block = BasicBlock::Create(getGlobalContext(), "BlockNotPresent", run_function);
 
-    ir_builder->SetInsertPoint(basic_block);
+    ir_builder->SetInsertPoint(run_function_entry);
+    ir_builder->CreateBr(run_function_re_entry);
+
+    ir_builder->SetInsertPoint(run_function_re_entry);
+    auto block_address = ir_builder->CreateCall(get_block_address_function, ir_builder->CreateAlignedLoad(GetRegisterPtr(Register::PC), 4));
+    auto block_present_pred = ir_builder->CreateICmpNE(block_address, ConstantPointerNull::get(ir_builder->getInt8PtrTy()));
+    ir_builder->CreateCondBr(block_present_pred, block_present_basic_block, block_not_present_basic_block);
+
+    ir_builder->SetInsertPoint(block_present_basic_block);
+    auto indirect_br = ir_builder->CreateIndirectBr(block_address, instruction_blocks.size());
+    for (auto &block : instruction_blocks)
+    {
+        indirect_br->addDestination(block->GetEntryBasicBlock());
+    }
+
+    ir_builder->SetInsertPoint(block_not_present_basic_block);
     ir_builder->CreateRetVoid();
 }
 
-void ModuleGen::GenerateBlocks()
+void ModuleGen::DecodeInstructions()
 {
     for (auto i = Loader::ROMCodeStart; i <= Loader::ROMCodeStart + Loader::ROMCodeSize - 4; i += 4)
     {
         auto instruction = Disassembler::Disassemble(Memory::Read32(i), i);
-        if (instruction != nullptr)
+        if (instruction == nullptr) continue;
+        auto instruction_block = std::make_unique<InstructionBlock>(this, instruction.release());
+        instruction_blocks_by_pc[i] = instruction_block.get();
+        instruction_blocks.push_back(std::move(instruction_block));
+    }
+}
+
+void ModuleGen::GenerateInstructionsEntry()
+{
+    for (auto &instruction : instruction_blocks)
+    {
+        instruction->GenerateEntryBlock();
+    }
+}
+
+void ModuleGen::GenerateInstructionsCode()
+{
+    for (auto &instruction : instruction_blocks)
+    {
+        instruction->GenerateCode();
+    }
+}
+
+void ModuleGen::GenerateInstructionsTermination()
+{
+    // Return to the switch
+    for (auto &block : instruction_blocks)
+    {
+        ir_builder->SetInsertPoint(block->GetExitBasicBlock());
+        ir_builder->CreateAlignedStore(ir_builder->getInt32(block->Address() + 4), GetRegisterPtr(Register::PC), 4);
+        ir_builder->CreateBr(run_function_re_entry);
+    }
+}
+
+void ModuleGen::AddInstructionsToRunFunction()
+{
+    std::stack<BasicBlock *> basic_blocks_stack;
+
+    for (auto &block : instruction_blocks)
+    {
+        basic_blocks_stack.push(block->GetEntryBasicBlock());
+
+        while (basic_blocks_stack.size())
         {
-            LOG_DEBUG(BinaryTranslator, "Instruction at %08x", i);
+            auto basic_block = basic_blocks_stack.top();
+            basic_blocks_stack.pop();
+            basic_block->insertInto(run_function);
+            auto terminator = basic_block->getTerminator();
+            for (auto i = 0; i < terminator->getNumSuccessors(); ++i)
+            {
+                auto new_basic_block = terminator->getSuccessor(i);
+                if (new_basic_block->getParent()) continue; // Already added to run
+                basic_blocks_stack.push(new_basic_block);
+            }
         }
     }
 }

src/binary_translation/ModuleGen.h

@@ -1,4 +1,10 @@
 #include <llvm/IR/IRBuilder.h>
+#include <unordered_map>
+#include <common/common_types.h>
+
+enum class Register;
+
+class InstructionBlock;
 
 namespace llvm
 {
@@ -12,11 +18,30 @@ public:
     ~ModuleGen();
 
     void Run();
+
+    // Returns the address of a register or a flag
+    llvm::Value *GetRegisterPtr(Register reg);
+
+    llvm::IRBuilder<> *IrBuilder() { return ir_builder.get(); }
+    llvm::Module *Module() { return module; }
+private:
+    // Generates the declarations of all the globals of the module
     void GenerateGlobals();
+    void GenerateBlockAddressArray();
+    void GenerateGetBlockAddressFunction();
     void GenerateCanRunFunction();
     void GenerateRunFunction();
-    void GenerateBlocks();
+    // Creates InstructionBlock for each instruction
-private:
+    void DecodeInstructions();
+    // Generates the entry basic blocks for each instruction
+    void GenerateInstructionsEntry();
+    // Generates the code of each instruction
+    void GenerateInstructionsCode();
+    // Terminates each block
+    void GenerateInstructionsTermination();
+    // Adds all the basic blocks of an instruction to the run function
+    void AddInstructionsToRunFunction();
+
     std::unique_ptr<llvm::IRBuilder<>> ir_builder;
     llvm::Module *module;
 
@@ -31,6 +56,21 @@ private:
      *  Orderered N, Z, C, V
      */
     llvm::GlobalVariable *flags_global;
+
+    size_t block_address_array_base;
+    size_t block_address_array_size;
+    /*
+     * i8 **BlockAddressArray;
+     *  The array at [i/4 - block_address_array_base] contains the block address for the instruction at i
+     *  or nullptr if it is not decoded
+     */
+    llvm::ArrayType *block_address_array_type;
+    llvm::GlobalVariable *block_address_array;
+    /*
+     * i8 *GetBlockAddress(u32 pc)
+     *  Returns the address of the block for the instruction at pc
+     */
+    llvm::Function *get_block_address_function;
     /*
      * bool CanRun()
      *  Returns whether there is a binary translation available for a PC
@@ -41,4 +81,12 @@ private:
      *  Runs binary translated opcodes
      */
     llvm::Function *run_function;
+    llvm::BasicBlock *run_function_entry;
+    llvm::BasicBlock *run_function_re_entry;
+
+    /*
+     * All the instruction blocks
+     */
+    std::vector<std::unique_ptr<InstructionBlock>> instruction_blocks;
+    std::unordered_map<u32, InstructionBlock *> instruction_blocks_by_pc;
 };