Add decompiler IR, basic operations, all-types file (#57)

* framework for basic op * started IR framework * check in type info file * add some basic operations to the first pass ir conversion * use a single condition system * add more basic op decoding * more ir
open-goal · Sep 30, 2020 · 376c273 · 376c273
1 parent c9b53d5
commit 376c273
Show file tree

Hide file tree

Showing 37 changed files with 10,346 additions and 392 deletions.
diff --git a/common/type_system/deftype.cpp b/common/type_system/deftype.cpp
@@ -78,24 +78,6 @@ int64_t get_int(const goos::Object& obj) {
   throw std::runtime_error(obj.print() + " was supposed to be an integer, but isn't");
 }
 
-TypeSpec parse_typespec(TypeSystem* type_system, const goos::Object& src) {
-  if (src.is_symbol()) {
-    return type_system->make_typespec(symbol_string(src));
-  } else if (src.is_pair()) {
-    TypeSpec ts = type_system->make_typespec(symbol_string(car(&src)));
-    const auto& rest = *cdr(&src);
-
-    for_each_in_list(rest,
-                     [&](const goos::Object& o) { ts.add_arg(parse_typespec(type_system, o)); });
-
-    return ts;
-  } else {
-    throw std::runtime_error("invalid typespec: " + src.print());
-  }
-  assert(false);
-  return {};
-}
-
 void add_field(StructureType* structure, TypeSystem* ts, const goos::Object& def) {
   auto rest = &def;
 
@@ -278,6 +260,24 @@ TypeFlags parse_structure_def(StructureType* type,
 
 }  // namespace
 
+TypeSpec parse_typespec(TypeSystem* type_system, const goos::Object& src) {
+  if (src.is_symbol()) {
+    return type_system->make_typespec(symbol_string(src));
+  } else if (src.is_pair()) {
+    TypeSpec ts = type_system->make_typespec(symbol_string(car(&src)));
+    const auto& rest = *cdr(&src);
+
+    for_each_in_list(rest,
+                     [&](const goos::Object& o) { ts.add_arg(parse_typespec(type_system, o)); });
+
+    return ts;
+  } else {
+    throw std::runtime_error("invalid typespec: " + src.print());
+  }
+  assert(false);
+  return {};
+}
+
 DeftypeResult parse_deftype(const goos::Object& deftype, TypeSystem* ts) {
   auto iter = &deftype;
 

diff --git a/common/type_system/deftype.h b/common/type_system/deftype.h
@@ -22,3 +22,4 @@ struct DeftypeResult {
 };
 
 DeftypeResult parse_deftype(const goos::Object& deftype, TypeSystem* ts);
+TypeSpec parse_typespec(TypeSystem* type_system, const goos::Object& src);
diff --git a/decomp.sh b/decomp.sh
@@ -3,4 +3,4 @@
 # Directory of this script
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
 
-$DIR/build/decompiler/decompiler $DIR/decompiler/config/jak1_ntsc_black_label.jsonc $DIR/iso_data $DIR/decompiler_out
+$DIR/build/decompiler/decompiler $DIR/decompiler/config/jak1_ntsc_black_label.jsonc $DIR/iso_data $DIR/decompiler_out
diff --git a/decompiler/CMakeLists.txt b/decompiler/CMakeLists.txt
@@ -12,14 +12,15 @@ add_executable(decompiler
         util/FileIO.cpp
         config.cpp
         util/LispPrint.cpp
+        util/DecompilerTypeSystem.cpp
         Function/BasicBlocks.cpp
         Disasm/InstructionMatching.cpp
-        TypeSystem/GoalType.cpp
-        TypeSystem/GoalFunction.cpp
-        TypeSystem/GoalSymbol.cpp
-        TypeSystem/TypeInfo.cpp
-        TypeSystem/TypeSpec.cpp Function/CfgVtx.cpp Function/CfgVtx.h)
+        Function/CfgVtx.cpp Function/CfgVtx.h
+        IR/BasicOpBuilder.cpp
+        IR/IR.cpp)
 
 target_link_libraries(decompiler
         minilzo
-        common_util)
+        common_util
+        type_system
+        fmt)
diff --git a/decompiler/Disasm/Instruction.h b/decompiler/Disasm/Instruction.h
@@ -44,6 +44,12 @@ struct InstructionAtom {
   std::string to_string(const LinkedObjectFile& file) const;
 
   bool is_link_or_label() const;
+  bool is_reg() const { return kind == REGISTER; }
+  bool is_imm() const { return kind == IMM; }
+  bool is_label() const { return kind == LABEL; }
+  bool is_sym() const { return kind == IMM_SYM; }
+
+  bool is_reg(Register r) const { return kind == REGISTER && reg == r; }
 
  private:
   int32_t imm;

diff --git a/decompiler/Disasm/InstructionMatching.h b/decompiler/Disasm/InstructionMatching.h
@@ -1,5 +1,10 @@
 #pragma once
 
+/*!
+ * @file InstructionMatching.h
+ * Utilities for checking if an instruction matches some criteria.
+ */
+
 #ifndef JAK_DISASSEMBLER_INSTRUCTIONMATCHING_H
 #define JAK_DISASSEMBLER_INSTRUCTIONMATCHING_H
 

diff --git a/decompiler/Disasm/OpcodeInfo.cpp b/decompiler/Disasm/OpcodeInfo.cpp
@@ -1,3 +1,8 @@
+/*!
+ * @file OpcodeInfo.cpp
+ * Decoding info for each opcode.
+ */
+
 #include "OpcodeInfo.h"
 #include <cassert>
 

diff --git a/decompiler/Function/BasicBlocks.h b/decompiler/Function/BasicBlocks.h
@@ -1,8 +1,5 @@
 #pragma once
 
-#ifndef JAK_DISASSEMBLER_BASICBLOCKS_H
-#define JAK_DISASSEMBLER_BASICBLOCKS_H
-
 #include <vector>
 #include <memory>
 
@@ -21,5 +18,3 @@ struct BasicBlock {
 std::vector<BasicBlock> find_blocks_in_function(const LinkedObjectFile& file,
                                                 int seg,
                                                 const Function& func);
-
-#endif  // JAK_DISASSEMBLER_BASICBLOCKS_H
diff --git a/decompiler/Function/CfgVtx.cpp b/decompiler/Function/CfgVtx.cpp
@@ -1637,7 +1637,6 @@ void ControlFlowGraph::flag_early_exit(const std::vector<BasicBlock>& blocks) {
  * Build and resolve a Control Flow Graph as much as possible.
  */
 std::shared_ptr<ControlFlowGraph> build_cfg(const LinkedObjectFile& file, int seg, Function& func) {
-  printf("build cfg : %s\n", func.guessed_name.to_string().c_str());
   auto cfg = std::make_shared<ControlFlowGraph>();
 
   const auto& blocks = cfg->create_blocks(func.basic_blocks.size());
@@ -1716,25 +1715,16 @@ std::shared_ptr<ControlFlowGraph> build_cfg(const LinkedObjectFile& file, int se
 
   cfg->flag_early_exit(func.basic_blocks);
 
-  //  if(func.guessed_name.to_string() == "(method 9 thread)")
-  //    cfg->find_cond_w_else();
-
-  //  if (func.guessed_name.to_string() != "looping-code") {
-  //    return cfg;
-  //  }
-
   bool changed = true;
   while (changed) {
     changed = false;
     // note - we should prioritize finding short-circuiting expressions.
     //    printf("%s\n", cfg->to_dot().c_str());
     //    printf("%s\n", cfg->to_form()->toStringPretty().c_str());
 
-    changed = changed | cfg->find_cond_w_else();
-    changed = changed | cfg->find_cond_n_else();
+    changed = changed || cfg->find_cond_w_else();
+    changed = changed || cfg->find_cond_n_else();
     changed = changed || cfg->find_while_loop_top_level();
-    //    ////    printf("while loops? %d\n", changed);
-    ////    changed = changed || cfg->find_if_else_top_level();
     changed = changed || cfg->find_seq_top_level();
     changed = changed || cfg->find_short_circuits();
 

diff --git a/decompiler/Function/Function.cpp b/decompiler/Function/Function.cpp
@@ -3,7 +3,7 @@
 #include "Function.h"
 #include "decompiler/Disasm/InstructionMatching.h"
 #include "decompiler/ObjectFile/LinkedObjectFile.h"
-#include "decompiler/TypeSystem/TypeInfo.h"
+#include "decompiler/util/DecompilerTypeSystem.h"
 
 namespace {
 std::vector<Register> gpr_backups = {make_gpr(Reg::GP), make_gpr(Reg::S5), make_gpr(Reg::S4),
@@ -418,7 +418,7 @@ void Function::check_epilogue(const LinkedObjectFile& file) {
  *
  * Updates the guessed_name of the function and updates type_info
  */
-void Function::find_global_function_defs(LinkedObjectFile& file) {
+void Function::find_global_function_defs(LinkedObjectFile& file, DecompilerTypeSystem& dts) {
   int state = 0;
   int label_id = -1;
   Register reg;
@@ -457,7 +457,8 @@ void Function::find_global_function_defs(LinkedObjectFile& file) {
           auto& func = file.get_function_at_label(label_id);
           assert(func.guessed_name.empty());
           func.guessed_name.set_as_global(name);
-          get_type_info().inform_symbol(name, TypeSpec("function"));
+          dts.add_symbol(name, "function");
+          ;
           // todo - inform function.
         }
 
@@ -549,4 +550,42 @@ void Function::find_method_defs(LinkedObjectFile& file) {
       }
     }
   }
+}
+
+void Function::add_basic_op(std::shared_ptr<IR> op, int start_instr, int end_instr) {
+  op->is_basic_op = true;
+  assert(end_instr > start_instr);
+
+  for (int i = start_instr; i < end_instr; i++) {
+    instruction_to_basic_op[i] = basic_ops.size();
+  }
+  basic_op_to_instruction[basic_ops.size()] = start_instr;
+  basic_ops.push_back(op);
+}
+
+bool Function::instr_starts_basic_op(int idx) {
+  auto op = instruction_to_basic_op.find(idx);
+  if (op != instruction_to_basic_op.end()) {
+    auto start_instr = basic_op_to_instruction.at(op->second);
+    return start_instr == idx;
+  }
+  return false;
+}
+
+IR* Function::get_basic_op_at_instr(int idx) {
+  return basic_ops.at(instruction_to_basic_op.at(idx)).get();
+}
+
+int Function::get_basic_op_count() {
+  return basic_ops.size();
+}
+
+int Function::get_failed_basic_op_count() {
+  int count = 0;
+  for (auto& x : basic_ops) {
+    if (dynamic_cast<IR_Failed*>(x.get())) {
+      count++;
+    }
+  }
+  return count;
 }
diff --git a/decompiler/Function/Function.h b/decompiler/Function/Function.h
@@ -8,6 +8,9 @@
 #include "decompiler/Disasm/Instruction.h"
 #include "BasicBlocks.h"
 #include "CfgVtx.h"
+#include "decompiler/IR/IR.h"
+
+class DecompilerTypeSystem;
 
 struct FunctionName {
   enum class FunctionKind {
@@ -60,8 +63,14 @@ class Function {
  public:
   Function(int _start_word, int _end_word);
   void analyze_prologue(const LinkedObjectFile& file);
-  void find_global_function_defs(LinkedObjectFile& file);
+  void find_global_function_defs(LinkedObjectFile& file, DecompilerTypeSystem& dts);
   void find_method_defs(LinkedObjectFile& file);
+  void add_basic_op(std::shared_ptr<IR> op, int start_instr, int end_instr);
+  bool has_basic_ops() { return !basic_ops.empty(); }
+  bool instr_starts_basic_op(int idx);
+  IR* get_basic_op_at_instr(int idx);
+  int get_basic_op_count();
+  int get_failed_basic_op_count();
 
   int segment = -1;
   int start_word = -1;
@@ -115,6 +124,9 @@ class Function {
 
  private:
   void check_epilogue(const LinkedObjectFile& file);
+  std::vector<std::shared_ptr<IR>> basic_ops;
+  std::unordered_map<int, int> instruction_to_basic_op;
+  std::unordered_map<int, int> basic_op_to_instruction;
 };
 
 #endif  // NEXT_FUNCTION_H