MIPS-Assembler-Simulator

CSC3050 Project 1 MIPS Assembler & Simulator

This project implements a MIPS assembler and simulator in a single simulator.cpp file. The assembler and the simulator are designed separately so that each of them is supposed to work independently.

Code: https://github.com/doutv/MIPS-Assembler-Simulator

Ref:

• http://courses.missouristate.edu/kenvollmar/mars/help/syscallhelp.html
• https://github.com/SamuelGong/MIPSAssemblerAndDisassembler

Features

Nested class

Ref:

• https://stackoverflow.com/questions/486099/can-inner-classes-access-private-variables
  • Unlike Java though, there is no correlation between an object of type Outer::Inner and an object of the parent class. You have to make the parent child relationship manually.
    Wrap Scanner and Parser into Assembler using nested class, which enhances code reusability.

class Assembler
{
  public:
    inline static vector<string> data_seg;
    inline static vector<string> text_seg;
    inline static vector<string> output;
    class Scanner
    {
      Assembler &assembler;
    };
    class Parser
    {
      Assembler &assembler;
    };
}

Hash table mapping machine code to function pointer

Ref:

• https://stackoverflow.com/questions/2136998/using-a-stl-map-of-function-pointers
• https://stackoverflow.com/questions/55520876/creating-an-unordered-map-of-stdfunctions-with-any-arguments
• https://stackoverflow.com/questions/23962019/how-to-initialize-stdfunction-with-a-member-function
  • print_add is a non-static member function of foo, which means it must be invoked on an instance of Foo; hence it has an implicit first argument, the this pointer.

Hash table unordered_map<string, function<void(const string &)>> can replace numerous conditional statements (if else if).
In its generating function gen_opcode_to_func , std::bind should be used to locate the function pointer of the current Simulator instance. e.g.

// in void Simulator::gen_opcode_to_func(unordered_map<string, function<void(const string &)>> &m)
m.emplace("000100", bind(&Simulator::instr_beq, this, placeholders::_1));

Thus, the code is much more neater when calling the corresponding function.

// in void Simulator::exec_instr(const string &mc)
if (opcode == R_opcode[0] || opcode == R_opcode[1])
{
    // R instructions
    string funct = mc.substr(mc.size() - 6, 6);
    auto it = opcode_funct_to_func.find(opcode + funct);
    if (it == opcode_funct_to_func.end())
        signal_exception("function not found!");
    (it->second)(mc);
}

Throw exception instead of exit

Ref:

• https://stackoverflow.com/questions/32257840/properly-terminating-program-using-exceptions

exit() will not go back through the call stack up to main cleaning everything up.
I define istream and ostream in main() function, if I call exit() during the program,
the streams cannot destruct properly, thus no output to files.

static void signal_exception(const string &err)
{
    throw invalid_argument(err);
}
// in main()
try
{
    if it should exit
    {
        signal_exception(exit_msg);
    }
}
catch (const exception &e)
{
    cerr << e.what() << endl;
}

Test multiple cases with makefile

Ref:

• https://stackoverflow.com/questions/4927676/implementing-make-check-or-make-test
• https://blog.csdn.net/u011630575/article/details/52151995
  • 2>&1 意思是把 标准错误输出 重定向到 标准输出.
  • /dev/null是一个文件，这个文件比较特殊，所有传给它的东西它都丢弃掉
    With this makefile, I can compile and test multiple cases by a single command make.
    Take sim_test for example, which tests both assembler and simulator, it will echo fail message and exit when the current case fails.

PROM = simulator
TEST_DIR = ./test
ASM_TESTS = 1 2 3 4 5 6 7 8 9 10 11 12 a-plus-b fib memcpy-hello-world
SIM_TESTS = a-plus-b fib memcpy-hello-world

.PHONY: all clean
.ONESHELL:

all: $(PROM) asm_test sim_test
    @echo "All tests passed!"

$(PROM): $(PROM).cpp
    g++ $(PROM).cpp -o $(PROM) -std=c++17

clean:
    rm $(PROM)
    rm $(TEST_DIR)/*.tasmout
    rm $(TEST_DIR)/*.out

asm_test: $(PROM)
    for t in $(ASM_TESTS); do \
        ./$(PROM) $(TEST_DIR)/$$t.asm $(TEST_DIR)/$$t.tasmout 2>&1; \
        diff -q $(TEST_DIR)/$$t.asmout $(TEST_DIR)/$$t.tasmout > /dev/null || \
        echo "Test $$t failed"; \
    done
    echo -e "All assembler tests passed!\n"


sim_test: $(PROM)
    for t in $(SIM_TESTS); do \
        ./$(PROM) $(TEST_DIR)/$$t.asm $(TEST_DIR)/$$t.in $(TEST_DIR)/$$t.out 2>&1; \
        diff -q $(TEST_DIR)/$$t.out $(TEST_DIR)/$$t.simout > /dev/null || \
        echo "Test $$t failed"; \
    done
    echo -e "All simulator tests passed!\n"

Tricks

• Use fixed width integer types such as int32_t instead of int.
• Get the maximum possible number of a data type, e.g. the maximum number for int16_t is numeric_limits<int16_t>::max()
• Conversion between binary string std::string and integer int32_t
  • int to binary string: bitset<str_length>(int32_t_number).to_string()
  • binary string to int: stoi(str,nullptr,base=2)
• Detect overflow by __builtin_add_overflow , __builtin_sub_overflow , __builtin_mul_overflow.
  • https://stackoverflow.com/questions/199333/how-do-i-detect-unsigned-integer-multiply-overflow

  • unsigned long b, c, c_test;
    if (__builtin_umull_overflow(b, c, &c_test))
    {
        // Returned non-zero: there has been an overflow
    }
    else
    {
        // Return zero: there hasn't been an overflow
    }
    

• inline static Declare and initialize static member variables together in class.
  • https://www.zhihu.com/question/375445099
  • 在 C++17 加入了内联静态成员变量，只要在声明成员变量时， 在static前加入 inline，就能顺便定义了它，还可初始化它，这就不用在类外定义了。
• static const Declare and initialize constant member variables in class.
• #define , #ifdef , #endif Use C preprocessor to help debug.
• Conversion between std::string and std::array
  • std::string -> std::array<char,size>
  • copy

  • word_t word;
    string word_str = bitset<word_size>(get_regv(rt)).to_string();
    copy(word_str.begin(), word_str.end(), word.data());
    

  • std::array<char,size> -> std::string
  • std::string constructor

  • word_t word = get_word_from_memory(addr);
    string word_str(&word[0], word_size);