disassemble: initial disassembler

.gitignore

@@ -1,6 +1,8 @@
 main
 scu-dsp-asm
 scu-dsp-asm.*
+scu-dsp-dis
+scu-dsp-dis.*
 .~*
 *.o
 *.gch

Makefile

@@ -3,28 +3,37 @@ CXXFLAGS = -Og -g -Wall -Wextra -Werror -Wfatal-errors -Wno-c99-designator -std=
 LDFLAGS =
 
 TARGET ?=
-CXX = $(TARGET)clang++
+CXX = $(TARGET)g++
 
-SRC = main.cpp
-SRC += lexer.cpp
-SRC += ast_printer.cpp
-SRC += ast_resolver.cpp
-SRC += ast_emitter.cpp
-SRC += parser.cpp
-SRC += stmt_string.cpp
-OBJ = $(patsubst %.cpp,%.o,$(SRC))
-DEP = $(patsubst %.cpp,%.d,$(SRC))
+ASM_SRC = main.cpp
+ASM_SRC += lexer.cpp
+ASM_SRC += ast_printer.cpp
+ASM_SRC += ast_resolver.cpp
+ASM_SRC += ast_emitter.cpp
+ASM_SRC += parser.cpp
+ASM_SRC += stmt_string.cpp
+ASM_OBJ = $(patsubst %.cpp,%.o,$(ASM_SRC))
+ASM_DEP = $(patsubst %.cpp,%.d,$(ASM_SRC))
+ASM_MAIN ?= scu-dsp-asm
 
-MAIN ?= scu-dsp-asm
+DIS_SRC = disassemble.cpp
+DIS_SRC += ast_printer.cpp
+DIS_SRC += stmt_string.cpp
+DIS_OBJ = $(patsubst %.cpp,%.o,$(DIS_SRC))
+DIS_DEP = $(patsubst %.cpp,%.d,$(DIS_SRC))
+DIS_MAIN ?= scu-dsp-dis
 
-all: $(MAIN)
+all: $(ASM_MAIN) $(DIS_MAIN)
 
--include $(DEP)
+-include $(sort $(ASM_DEP) $(DIS_DEP))
 
 %.o: %.cpp
 	$(CXX) $(CXXFLAGS) -MMD -MF $(basename $<).d -c $< -o $@
 
-$(MAIN): $(OBJ)
+$(SRC_MAIN): $(SRC_OBJ)
+	$(CXX) $(STATIC) $(LDFLAGS) $^ -o $@
+
+$(DIS_MAIN): $(DIS_OBJ)
 	$(CXX) $(STATIC) $(LDFLAGS) $^ -o $@
 
 clean:

control_word.hpp

@@ -7,6 +7,7 @@
 
 #include "stmt_base.hpp"
 #include "stmt_ins.hpp"
+#include "ins.hpp"
 
 namespace dsp {
 
@@ -32,7 +33,7 @@ using op_t = std::variant<andl_t,
                           mov_imm_d1_t,
                           mov_ram_d1_t>;
 
-struct control_word_t : stmt_accept_t<control_word_t>
+struct control_word_t : stmt_accept_t<control_word_t>, ins_t<control_word_t>
 {
   control_word_t(std::vector<op_t> ops)
     : ops(ops)
@@ -43,29 +44,26 @@ struct control_word_t : stmt_accept_t<control_word_t>
   {
     std::vector<op_t> ops;
     if (andl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<andl_t>, code});
-    else if (orl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<orl_t>, code});
-    else if (xorl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<xorl_t>, code});
-    else if (add_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<add_t>, code});
-    else if (sub_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sub_t>, code});
-    else if (ad2_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<ad2_t>, code});
-    else if (sr_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sr_t>, code});
-    else if (rr_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rr_t>, code});
-    else if (sl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sl_t>, code});
-    else if (rl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rl_t>, code});
-    else if (rl8_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rl8_t>, code});
-    else if (mov_ram_x_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_x_t>, code});
-    else if (mov_mul_p_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_mul_p_t>, code});
-    else if (mov_ram_p_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_p_t>, code});
-    else if (mov_ram_y_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_y_t>, code});
-    else if (clr_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<clr_a_t>, code});
-    else if (mov_alu_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_alu_a_t>, code});
-    else if (mov_ram_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_a_t>, code});
-    else if (mov_imm_d1_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_imm_d1_t>, code});
-    else if (mov_ram_d1_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_d1_t>, code});
-    else {
-      std::cerr << "invalid control word code: " << (code) << std::endl;
-      throw std::runtime_error("invalid control word code");
-    }
+    if (orl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<orl_t>, code});
+    if (xorl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<xorl_t>, code});
+    if (add_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<add_t>, code});
+    if (sub_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sub_t>, code});
+    if (ad2_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<ad2_t>, code});
+    if (sr_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sr_t>, code});
+    if (rr_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rr_t>, code});
+    if (sl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<sl_t>, code});
+    if (rl_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rl_t>, code});
+    if (rl8_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<rl8_t>, code});
+    if (mov_ram_x_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_x_t>, code});
+    if (mov_mul_p_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_mul_p_t>, code});
+    if (mov_ram_p_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_p_t>, code});
+    if (mov_ram_y_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_y_t>, code});
+    if (clr_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<clr_a_t>, code});
+    if (mov_alu_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_alu_a_t>, code});
+    if (mov_ram_a_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_a_t>, code});
+    if (mov_imm_d1_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_imm_d1_t>, code});
+    if (mov_ram_d1_t::pred(code)) ops.emplace_back(op_t{std::in_place_type<mov_ram_d1_t>, code});
+
     return ops;
   }
   
@@ -74,9 +72,9 @@ struct control_word_t : stmt_accept_t<control_word_t>
 
   const std::vector<op_t> ops;
 
-  uint32_t mask() const { return 0b11 << 30; }
-  uint32_t code() const { return 0b00 << 30; }
-  uint32_t bits() const { return 0; }
+  static constexpr uint32_t mask() { return 0b11 << 30; }
+  static constexpr uint32_t code() { return 0b00 << 30; }
+  static constexpr uint32_t bits() { return 0; }
 };
 
 }

disassemble.cpp

@@ -0,0 +1,89 @@
+#include <cstdint>
+#include <string_view>
+#include <fstream>
+
+#include "stmt.hpp"
+#include "stmt_ins.hpp"
+#include "control_word.hpp"
+#include "ast_printer.hpp"
+
+static uint32_t uint32_be(std::string_view v)
+{
+  return static_cast<uint8_t>(v[0]) << 24
+       | static_cast<uint8_t>(v[1]) << 16
+       | static_cast<uint8_t>(v[2]) << 8
+       | static_cast<uint8_t>(v[3]) << 0;
+}
+
+static dsp::stmt_t * decompile(uint32_t code)
+{
+  using namespace dsp::op;
+  using namespace dsp::load;
+  using namespace dsp::dma;
+  using namespace dsp::jump;
+  using namespace dsp::loop;
+  using namespace dsp::end;
+  using namespace dsp::nop;
+
+  if      (control_word_t::pred(code)) return new control_word_t(code);
+  else if (mvi_t::pred(code))          return new mvi_t(code);
+  else if (mvi_cond_t::pred(code))     return new mvi_cond_t(code);
+  else if (d0_dst_imm_t::pred(code))   return new d0_dst_imm_t(code);
+  else if (src_d0_imm_t::pred(code))   return new src_d0_imm_t(code);
+  else if (d0_dst_ram_t::pred(code))   return new d0_dst_ram_t(code);
+  else if (src_d0_ram_t::pred(code))   return new src_d0_ram_t(code);
+  else if (jmp_t::pred(code))          return new jmp_t(code);
+  else if (jmp_cond_t::pred(code))     return new jmp_cond_t(code);
+  else if (btm_t::pred(code))          return new btm_t(code);
+  else if (lps_t::pred(code))          return new lps_t(code);
+  else if (end_t::pred(code))          return new end_t(code);
+  else if (endi_t::pred(code))         return new endi_t(code);
+  else
+    throw std::runtime_error("undefined code");
+}
+
+static void run(std::string source)
+{
+  std::string_view buf(source);
+  dsp::ast::printer_t printer(std::cout);
+
+  for (decltype(buf)::size_type i = 0; i < (buf.length() / 4); i++) {
+    uint32_t code = uint32_be(buf.substr(i * 4, 4));
+
+    dsp::stmt_t * stmt = decompile(code);
+    stmt->accept(&printer);
+    std::cout << std::endl;
+  }
+}
+
+static int run_file(char const * const input_filename)
+{
+  std::ifstream is {input_filename, std::ios::binary | std::ios::ate};
+  if (!is.is_open()) {
+    std::cerr << "failed to open " << input_filename << std::endl;
+    return -1;
+  }
+  const std::streampos size = is.tellg();
+  std::string buf(size, '\0');
+  is.seekg(0);
+  if (!is.read(&buf[0], size)) {
+    std::cerr << "read failed" << std::endl;
+    return -1;
+  }
+
+  run(buf);
+
+  return 0;
+}
+
+int main(const int argc, char const * const argv[])
+{
+  const std::string c_source("-s");
+
+  switch (argc) {
+  case 2: return run_file(argv[1]);
+  default:
+    std::cerr << "Usage: " << argv[0] << " input-filename" << std::endl;
+    return -1;
+  }
+}

ins.hpp

@@ -9,14 +9,10 @@ namespace dsp {
 template <typename T>
 struct ins_t
 {
-  /*
-  virtual uint32_t bits() const;
-
   uint32_t render() const
   {
-    return T::code() | bits();
+    return T::code() | T::bits();
   }
-  */
 
   static bool pred(uint32_t ins)
   {

main.cpp

@@ -1,8 +1,6 @@
 #include <iostream>
 #include <fstream>
 #include <string>
-#include <optional>
-#include <bitset>
 #include <sstream>
 #include <array>
 #include <iomanip>
@@ -72,7 +70,6 @@ static void run(std::ostream& os, std::string source, bool hex_output, bool verb
   while (auto stmt_o = pass2.statement()) {
     uint32_t output = (*stmt_o)->accept(&emitter);
     if (output != 0xffff'ffff) {
-      //std::cout << std::bitset<32>(output) << std::endl;
       if (hex_output) write_hex(os, output);
       else            write_raw(os, output);
     }

visitable.hpp

@@ -37,7 +37,7 @@ struct mov_ram_d1_t;
 struct control_word_t;
 }
 
-namespace load{
+namespace load {
 struct mvi_t;
 struct mvi_cond_t;
 }