first workable Mem2Reg

.vscode/launch.json

@@ -9,7 +9,7 @@
             "request": "launch",
             "name": "Debug",
             "program": "${workspaceFolder}/build/sysy",
-            "args": ["../sysytests/functional_2022/21_if_test2.sy", "-S", "-o", "build/21_my.ll", "-emit-llvm"],
+            "args": ["../sysytests/functional_2022/21_if_test2.sy", "-S", "-o", "build/21_my.ll", "-O1", "-emit-llvm"],
             "cwd": "${workspaceFolder}"
         }
     ]

include/llir_type.h

@@ -1,10 +1,10 @@
 #pragma once
 
 #include "common.h"
+#include <cassert>
 #include <memory>
 #include <string>
 #include <vector>
-#include <cassert>
 
 namespace antlrSysY {
 

include/llir_value.h

@@ -9,8 +9,8 @@
 #include <shared_mutex>
 #include <string>
 #include <tuple>
-#include <vector>
 #include <unordered_set>
+#include <vector>
 
 namespace antlrSysY {
 
@@ -25,7 +25,7 @@ DEF_PTR_T(ConstantInt);
 // typedef std::tuple<ValuePtr_t, UserPtr_t, int> UseEdge_t;
 struct Use {
   ValuePtr_t value;
-  User* user;
+  User *user;
   unsigned op_index;
 };
 //                  Define, User, operand-index
@@ -66,7 +66,7 @@ public:
     panic("No applicable for IR gen");
   }
 
-  void u_remove_use(User* user) {
+  void u_remove_use(User *user) {
     // use_list.erase(
     //   std::remove_if(
     //     use_list.begin(), use_list.end(), [user](const UseEdge_t &use) { return std::get<1>(use) == user; }
@@ -78,7 +78,7 @@ public:
         itr = use_list.erase(itr);
       }
       else {
-        ++ itr;
+        ++itr;
       }
     }
   }
@@ -101,7 +101,7 @@ public:
     }
     for (auto use : use_list) {
       auto user                 = use.user;
-      auto index                 = use.op_index;
+      auto index                = use.op_index;
       user->operand_list[index] = value;
       assert(value);
       value->use_list.push_back({value, user, index});
@@ -162,7 +162,7 @@ public:
   std::list<BasicBlockPtr_t> successors;
   std::list<BasicBlockPtr_t> predecessors;
 
-  BasicBlockPtr_t IDOM;               // dominating node
+  BasicBlockPtr_t IDOM;                         // dominating node
   std::unordered_set<BasicBlockPtr_t> idom_set; // immediate dominated nodes
   std::unordered_set<BasicBlockPtr_t> DOM_set;  // dominated nodes
   std::unordered_set<BasicBlockPtr_t> DF_set;

include/visitor.h

@@ -81,7 +81,12 @@ std::shared_ptr<InstGEP> build_InstGEP(
   BasicBlockPtr_t parent_bb
 );
 
-InstPhiPtr_t build_InstPhi(TypePtr_t type, const decltype(Function::bb_list) &incoming_vals, BasicBlockPtr_t parent_bb);
+InstPhiPtr_t build_InstPhi(
+  TypePtr_t type,
+  const decltype(Function::bb_list) &incoming_vals,
+  BasicBlockPtr_t parent_bb,
+  const std::string &name
+);
 
 #pragma endregion
 

scripts/mytester.py

@@ -228,7 +228,7 @@ scheme_ref = {
 
 scheme_my = {
     "name": "my",
-    "sy_ir": f"build/sysy -S {{sy}} -o {{ir}} -emit-llvm",
+    "sy_ir": f"build/sysy -S {{sy}} -o {{ir}} -emit-llvm -O1",
     "ir_asm": "llc --march=x86 --relocation-model=pic {ir} -o {asm}",
     "asm_obj": "as --32 {asm} -o {obj}",
     "obj_bin": f"clang -m32 -Ofast -fPIE {{obj}} {LIB_PATH} -o {{bin}}",

src/main.cpp

@@ -62,11 +62,11 @@ int main(int argc, const char **argv) {
 #pragma region Logger
   el::Configurations defaultConf;
   defaultConf.setToDefault();
-  defaultConf.set(el::Level::Debug, el::ConfigurationType::Format, "%level %loc %msg");
-  defaultConf.set(el::Level::Warning, el::ConfigurationType::Format, "%level %loc %msg");
-  defaultConf.set(el::Level::Error, el::ConfigurationType::Format, "%level %loc %msg");
-  defaultConf.set(el::Level::Info, el::ConfigurationType::Format, "%level %loc %msg");
-  defaultConf.set(el::Level::Verbose, el::ConfigurationType::Format, "%level %loc %msg");
+  defaultConf.set(el::Level::Debug, el::ConfigurationType::Format, "%levshort %loc %msg");
+  defaultConf.set(el::Level::Warning, el::ConfigurationType::Format, "%levshort %loc %msg");
+  defaultConf.set(el::Level::Error, el::ConfigurationType::Format, "%levshort %loc %msg");
+  defaultConf.set(el::Level::Info, el::ConfigurationType::Format, "%levshort %loc %msg");
+  defaultConf.set(el::Level::Verbose, el::ConfigurationType::Format, "%levshort%vlevel %loc %msg");
   el::Loggers::addFlag(el::LoggingFlag::ColoredTerminalOutput);
   el::Loggers::reconfigureLogger("default", defaultConf);
 #pragma endregion

src/pass_build_cfg.cpp

@@ -39,9 +39,9 @@ static void connect(BasicBlockPtr_t pred, BasicBlockPtr_t succ) {
 }
 
 static void _build_cfg(const FunctionPtr_t func) {
-  for (auto itr = std::next(func->bb_list.begin()); itr != func->bb_list.end(); ) {
+  for (auto itr = std::next(func->bb_list.begin()); itr != func->bb_list.end();) {
     auto bb = *itr;
-    if (bb->inst_list.size() == 1 && Value::is<InstBranch>(bb->inst_list.front())){
+    if (bb->inst_list.size() == 1 && Value::is<InstBranch>(bb->inst_list.front())) {
       auto br = Value::as<InstBranch>(bb->inst_list.front());
       if (br->operand_list.size() == 1) {
         VLOG(6) << "remove dummy jump block:" << bb->name;
@@ -53,7 +53,7 @@ static void _build_cfg(const FunctionPtr_t func) {
         continue;
       }
     }
-    ++ itr;
+    ++itr;
   }
   for (auto basicblock : func->bb_list) {
     auto _inst_br = basicblock->inst_list.back();
@@ -71,9 +71,9 @@ static void _build_cfg(const FunctionPtr_t func) {
     }
   }
   // remove unreachable blocks
-  for (auto itr = std::next(func->bb_list.begin()); itr != func->bb_list.end(); ){
+  for (auto itr = std::next(func->bb_list.begin()); itr != func->bb_list.end();) {
     auto bb = *itr;
-    if (bb->predecessors.size() == 0){
+    if (bb->predecessors.size() == 0) {
       for (auto succ : bb->successors) {
         succ->predecessors.remove(bb);
       }
@@ -81,10 +81,9 @@ static void _build_cfg(const FunctionPtr_t func) {
       itr = func->bb_list.erase(itr);
     }
     else
-      ++ itr;
+      ++itr;
   }
-  
-  print_cfg(func->bb_list.front(), true);
+  if (VLOG_IS_ON(6)) print_cfg(func->bb_list.front(), true);
 }
 
 void PassBuildCFG::run(const Module &module) {

src/pass_mem2reg.cpp

@@ -14,20 +14,20 @@ namespace antlrSysY {
 static bool is_alloca_promotable(InstAllocaPtr_t inst) {
   for (const auto use : inst->use_list) {
     auto user = use.user;
-    if (dynamic_cast<InstLoad*>(user)) {
-      const auto li = dynamic_cast<InstLoad*>(user);
+    if (dynamic_cast<InstLoad *>(user)) {
+      const auto li = dynamic_cast<InstLoad *>(user);
       if (li->type != inst->type) {
         return false;
       }
     }
-    else if (dynamic_cast<InstStore*>(user)) {
-      const auto si = dynamic_cast<InstStore*>(user);
+    else if (dynamic_cast<InstStore *>(user)) {
+      const auto si = dynamic_cast<InstStore *>(user);
       if (si->operand_list[1] == inst || si->type != inst->type) {
         return false;
       }
     }
-    else if (dynamic_cast<InstGEP*>(user)) {
-      const auto gep = dynamic_cast<InstGEP*>(user);
+    else if (dynamic_cast<InstGEP *>(user)) {
+      const auto gep = dynamic_cast<InstGEP *>(user);
       for (int i = 1; i < gep->operand_list.size(); ++i) {
         if (!Value::is<ConstantInt>(gep->operand_list[i]) || !Value::as<ConstantInt>(gep->operand_list[i])->value) {
           return false;
@@ -46,20 +46,20 @@ struct AllocaInfo {
   std::vector<BasicBlockPtr_t> use_blocks = {};
   bool only_in_1_block                    = true;
   BasicBlockPtr_t only_block              = nullptr;
-  InstStore* only_store               = nullptr;
+  InstStore *only_store                   = nullptr;
 };
 
 static void analyze_alloca(InstAllocaPtr_t ai, AllocaInfo &info) {
   for (auto use : ai->use_list) {
-    auto user = dynamic_cast<Instruction*>(use.user);
-    if (dynamic_cast<InstStore*>(user)) {
-      const auto si = dynamic_cast<InstStore*>(user);
+    auto user = dynamic_cast<Instruction *>(use.user);
+    if (dynamic_cast<InstStore *>(user)) {
+      const auto si = dynamic_cast<InstStore *>(user);
       info.def_blocks.push_back(si->parent_bb);
       info.only_store = si;
     }
     else {
-      assert(dynamic_cast<InstLoad*>(user));
-      const auto li = dynamic_cast<InstLoad*>(user);
+      assert(dynamic_cast<InstLoad *>(user));
+      const auto li = dynamic_cast<InstLoad *>(user);
       info.use_blocks.push_back(li->parent_bb);
     }
     if (info.only_in_1_block) {
@@ -69,18 +69,20 @@ static void analyze_alloca(InstAllocaPtr_t ai, AllocaInfo &info) {
         info.only_in_1_block = false;
     }
   }
-  std::cout << "Alloca " << ai->name;
-  std::cout << "  Def: [";
-  for (auto blk : info.def_blocks) {
-    std::cout << blk->name << ", ";
+  if (VLOG_IS_ON(6)) {
+    std::cout << "Alloca " << ai->name;
+    std::cout << "  Def: [";
+    for (auto blk : info.def_blocks) {
+      std::cout << blk->name << ", ";
+    }
+    std::cout << "]\n";
+    std::cout << "  Use: [";
+    for (auto blk : info.use_blocks) {
+      std::cout << blk->name << ", ";
+    }
+    std::cout << "]\n";
+    std::cout.flush();
   }
-  std::cout << "]\n";
-  std::cout << "  Use: [";
-  for (auto blk : info.def_blocks) {
-    std::cout << blk->name << ", ";
-  }
-  std::cout << "]\n";
-  std::cout.flush();
 }
 
 // static bool rewrite_single_store(InstAllocaPtr_t ai, AllocaInfo& alloca_info) {
@@ -149,8 +151,10 @@ static void print_dom_info(BasicBlockPtr_t bb) {
   }
   std::cout << "]\n";
   std::cout << "  idomer: ";
-  if (bb->IDOM) std::cout << bb->IDOM->name;
-  else std::cout << "null";
+  if (bb->IDOM)
+    std::cout << bb->IDOM->name;
+  else
+    std::cout << "null";
   std::cout << "\n";
   std::cout << "  dom: [";
   for (auto dom : bb->DOM_set) {
@@ -163,7 +167,7 @@ static void print_dom_info(BasicBlockPtr_t bb) {
   }
   std::cout << "]\n";
   std::cout << "  DF: [";
-    for (auto df : bb->DF_set) {
+  for (auto df : bb->DF_set) {
     std::cout << df->name << ", ";
   }
   std::cout << "]\n";
@@ -185,7 +189,7 @@ static void _mem_2_reg(FunctionPtr_t func) {
   VLOG(4) << "  Gen Dominance Tree & Frontier";
   gen_dominance(func);
   gen_dominance_frontier(func);
-  print_dom_tree(func->bb_list.front());
+  if (VLOG_IS_ON(6)) print_dom_tree(func->bb_list.front());
   // actually, all variable alloca is placed at block head, so we collect them first
   std::vector<InstAllocaPtr_t> alloca_list;
   std::unordered_map<InstAllocaPtr_t, int> alloca_to_id;
@@ -197,12 +201,12 @@ static void _mem_2_reg(FunctionPtr_t func) {
   for (auto bb : func->bb_list) {
     for (auto inst : bb->inst_list) {
       if (Value::is<InstAlloca>(inst)) {
-        assert(inst->parent_bb == bb);
+        // assert(inst->parent_bb == bb);
+        // 貌似没问题，parent_bb是插入alloca的时候的那个block，而alloca实际上是被插入在entry_block的
         assert(bb == func->bb_list.front() && "Alloca should be at front of a func");
         auto ai = Value::as<InstAlloca>(inst);
         // assert(is_alloca_promotable(ai) && "Invalid alloca");
-        if (!Type::isType<IntegerType>(PointerType::pointedType(ai->type)))
-          continue;
+        if (!Type::isType<IntegerType>(PointerType::pointedType(ai->type))) continue;
         alloca_list.push_back(ai);
       }
     }
@@ -248,13 +252,14 @@ static void _mem_2_reg(FunctionPtr_t func) {
       auto bb = worklist.front();
       worklist.pop();
       for (auto frontier : bb->DF_set) {
+        // LOG(DEBUG) << "visiting " << bb->name <<" : " << frontier->name;
         auto frontier_index = bb_to_id.at(frontier);
         if (!visited[frontier_index]) {
           visited[frontier_index] = true;
-          if (livein_blocks.count(frontier)) {
-            auto inst_phi = build_InstPhi(TypeHelper::TYPE_I32, bb->predecessors, bb);
+          // if (livein_blocks.count(frontier)) {
+            auto inst_phi = build_InstPhi(TypeHelper::TYPE_I32, frontier->predecessors, frontier, ai->name);
             phi_to_allocaid.insert({inst_phi, i});
-          }
+          // }
           if (!def_blocks.count(frontier)) {
             worklist.push(frontier);
           }
@@ -266,8 +271,7 @@ static void _mem_2_reg(FunctionPtr_t func) {
   // renaming
   VLOG(4) << "  variable renaming";
   std::vector<ValuePtr_t> _init_values;
-  for (int i = 0; i < alloca_list.size(); ++ i)
-    _init_values.push_back(ConstantInt::make_shared(0));
+  for (int i = 0; i < alloca_list.size(); ++i) _init_values.push_back(ConstantInt::make_shared(0));
   std::vector<RenameInfo> rename_list = {{func->bb_list.front(), nullptr, _init_values}};
   std::vector<bool> visited(bb_to_id.size(), 0);
   while (!rename_list.empty()) {
@@ -292,8 +296,8 @@ static void _mem_2_reg(FunctionPtr_t func) {
     // process instruction
     for (auto itr = rename_info.bb->inst_list.begin(); itr != rename_info.bb->inst_list.end();) {
       auto inst = *itr++;
-      if (Value::is<InstAlloca>(inst)) {
-        assert(alloca_to_id.count(Value::as<InstAlloca>(inst)));
+      // we skip non-integer alloca, they are not in our alloca_list
+      if (Value::is<InstAlloca>(inst) && alloca_to_id.count(Value::as<InstAlloca>(inst))) {
         rename_info.bb->inst_list.remove(inst);
       }
       else if (Value::is<InstLoad>(inst)) {

src/visitor_factory.cpp

@@ -143,9 +143,11 @@ std::shared_ptr<InstGEP> build_InstGEP(
 InstPhiPtr_t build_InstPhi(
   TypePtr_t type,
   const decltype(Function::bb_list) &incoming_vals,
-  BasicBlockPtr_t parent_bb
+  BasicBlockPtr_t parent_bb,
+  const std::string &name
 ) {
-  auto inst = std::make_shared<InstPhi>(type, incoming_vals, parent_bb);
+  auto inst  = std::make_shared<InstPhi>(type, incoming_vals, parent_bb);
+  inst->name = name;
   parent_bb->inst_list.insert(parent_bb->inst_list.begin(), inst);
   return inst;
 }

src/visitor_llir_gen.cpp

@@ -286,7 +286,7 @@ static void _gen_blocks(std::ostream &ostr, const std::list<BasicBlockPtr_t> &bl
               }
               else if (Value::is<FParam>(inst->operand_list[i])) {
                 auto op0 = Value::as<FParam>(inst->operand_list[i]);
-                ostr << op0->to_IR_string() << ", ";
+                ostr << op0->to_IR_string();
               }
               else {
                 LOG(ERROR) << "Unexpected type of op_i: " << inst->operand_list[i]->to_string();
@@ -467,6 +467,7 @@ static void _gen_blocks(std::ostream &ostr, const std::list<BasicBlockPtr_t> &bl
               ostr << ", ";
             }
           }
+          ostr << "    ; alloca: " << inst->name;
           break;
         }
         // These should not be generated in frontend stage