the order matters for data flow analysis!

include/codegen/codegen.hpp

@@ -129,15 +129,23 @@ class CodeGen {
                          string reg2,
                          int imm,
                          string tinstr,
+                         int tid = 0,
                          int bits = 12,
-                         string treg = "$t0",
                          bool u = false) {
+        /* this function will tranfser
+         * `addi.d $a0, $fp, imm` to `add.d $a0, $fp, tmp_ireg` if imm
+         * overfloats for `addi.d`. During the time we move `imm` to `tmp_ireg`,
+         * another tmp ireg will be used, they can be same, but we must specify
+         * it.
+         */
+        auto treg = tmpregname(tid, false);
+        assert(treg != reg2 && "it's possible to write tid before reg2's use");
         auto [l, h] = immRange(bits, u);
         if (l <= imm and imm <= h)
             output.push_back(instr_ir + " " + reg1 + ", " + reg2 + ", " +
                              to_string(imm));
         else {
-            assert(value2reg(ConstantInt::get(imm, m), 0, treg) == treg);
+            assert(value2reg(ConstantInt::get(imm, m), tid, treg) == treg);
             output.push_back(tinstr + " " + reg1 + ", " + reg2 + ", " + treg);
         }
     }

include/optimization/GVN.h

@@ -450,6 +450,7 @@ class GVN : public Pass {
     std::unique_ptr<FuncInfo> func_info_;
     std::unique_ptr<GVNExpression::ConstFolder> folder_;
     std::unique_ptr<DeadCode> dce_;
+    std::vector<BasicBlock *> BB_DFS_Order;
 
     // self add member
     // std::uint64_t start_number_;
@@ -463,6 +464,7 @@ class GVN : public Pass {
     //
     // self add function
     //
+    void get_dfs_order(Function *func);
     void add_map_();
     std::uint64_t new_number() { return next_value_number_++; }
     void deal_with_entry(BasicBlock *Entry);

src/codegen/codegen.cpp

@@ -215,8 +215,7 @@ CodeGen::value2reg(Value *v, int i, string recommend) {
     } else if (dynamic_cast<AllocaInst *>(v)) {
         // auto alloc_instr = dynamic_cast<AllocaInst *>(v);
         // give the stack address
-        makeSureInRange(
-            "addi.d", reg_name, FP, -off.at(v), "add.d", 12, reg_name);
+        makeSureInRange("addi.d", reg_name, FP, -off.at(v), "add.d", i);
         /* output.push_back("addi.d " + reg_name + ", $fp, -" +
          *                  to_string(off.at(v))); */
     } else if (dynamic_cast<Argument *>(v)) {
@@ -235,8 +234,7 @@ CodeGen::value2reg(Value *v, int i, string recommend) {
                             FP,
                             func_arg_off.at(cur_func).at(id),
                             instr_ir + "x" + suff,
-                            12,
-                            reg_name);
+                            i);
             /* output.push_back(instr_ir + suff + " " + reg_name + ", $fp, " +
              *                  to_string(func_arg_off.at(cur_func).at(id))); */
         }
@@ -248,8 +246,7 @@ CodeGen::value2reg(Value *v, int i, string recommend) {
                         FP,
                         -off.at(v),
                         instr_ir + "x" + suff,
-                        12,
-                        reg_name);
+                        i);
         /* output.push_back(instr_ir + suff + " " + reg_name +
          *                  ", $fp, -" + to_string(off.at(v)));  */
     }
@@ -350,10 +347,10 @@ CodeGen::IR2assem(SiToFpInst *instr) {
 }
 
 string
-CodeGen::bool2branch(Instruction *instr) {
-    assert(instr->get_type() == m->get_int1_type());
-    auto icmp_instr = dynamic_cast<CmpInst *>(instr);
-    auto fcmp_instr = dynamic_cast<FCmpInst *>(instr);
+CodeGen::bool2branch(Instruction *cond) {
+    assert(cond->get_type() == m->get_int1_type());
+    auto icmp_instr = dynamic_cast<CmpInst *>(cond);
+    auto fcmp_instr = dynamic_cast<FCmpInst *>(cond);
     assert(icmp_instr or fcmp_instr);
 
     string instr_ir;
@@ -365,9 +362,9 @@ CodeGen::bool2branch(Instruction *instr) {
                 break;
             case CmpInst::NE: {
                 instr_ir = "bne";
-                if (instr->get_operand(1) == CONST_0 and
-                    dynamic_cast<Instruction *>(instr->get_operand(0)) and
-                    dynamic_cast<Instruction *>(instr->get_operand(0))
+                if (cond->get_operand(1) == CONST_0 and
+                    dynamic_cast<Instruction *>(cond->get_operand(0)) and
+                    dynamic_cast<Instruction *>(cond->get_operand(0))
                         ->is_zext()) {
                     // something like:
                     // %op0 = icmp slt i32 1, 2 # deepest
@@ -375,7 +372,7 @@ CodeGen::bool2branch(Instruction *instr) {
                     // %op2 = icmp ne i32 %op1, 0
                     // br i1 %op2, label %label3, label %label5
                     auto deepest = static_cast<Instruction *>(
-                        static_cast<Instruction *>(instr->get_operand(0))
+                        static_cast<Instruction *>(cond->get_operand(0))
                             ->get_operand(0));
                     return bool2branch(deepest);
                 }
@@ -395,11 +392,11 @@ CodeGen::bool2branch(Instruction *instr) {
                 reverse = true;
                 break;
         }
-        auto reg1 = value2reg(instr->get_operand(0), 0);
-        auto reg2 = value2reg(instr->get_operand(1), 1);
+        auto reg1 = value2reg(cond->get_operand(0), 0);
+        auto reg2 = value2reg(cond->get_operand(1), 1);
         return instr_ir + " " +
                (reverse ? (reg2 + ", " + reg1) : (reg1 + ", " + reg2)) + ",";
-    } else {
+    } else if (fcmp_instr) {
         switch (fcmp_instr->get_cmp_op()) {
             case FCmpInst::EQ:
                 instr_ir = "fcmp.ceq.s $fcc0";
@@ -422,11 +419,12 @@ CodeGen::bool2branch(Instruction *instr) {
                 instr_ir = "fcmp.cle.s $fcc0";
                 break;
         }
-        auto reg1 = value2reg(instr->get_operand(0), 0);
-        auto reg2 = value2reg(instr->get_operand(1), 1);
+        auto reg1 = value2reg(cond->get_operand(0), 0);
+        auto reg2 = value2reg(cond->get_operand(1), 1);
         output.push_back(instr_ir + ", " + reg1 + ", " + reg2);
         return (reverse ? "bceqz $fcc0," : "bcnez $fcc0,");
-    }
+    } else
+        assert(false);
 }
 
 void
@@ -435,10 +433,20 @@ CodeGen::IR2assem(BranchInst *instr) {
         auto TBB = static_cast<BasicBlock *>(instr->get_operand(1));
         auto FBB = static_cast<BasicBlock *>(instr->get_operand(2));
         // value2reg(instr->get_operand(0));
-        string instr_ir =
-            bool2branch(static_cast<Instruction *>(instr->get_operand(0)));
-        output.push_back(instr_ir + " " + label_in_assem(TBB));
-        output.push_back("b " + label_in_assem(FBB));
+        auto cond = instr->get_operand(0);
+        if (dynamic_cast<ConstantInt *>(cond)) {
+            auto const_bool = static_cast<ConstantInt *>(cond);
+            assert(const_bool->get_type()->is_integer_type() and
+                   static_cast<IntegerType *>(const_bool->get_type())
+                           ->get_num_bits() == 1);
+            bool cond_value = const_bool->get_value();
+            output.push_back("b " + label_in_assem(cond_value ? TBB : FBB));
+        } else {
+            string instr_ir =
+                bool2branch(static_cast<Instruction *>(instr->get_operand(0)));
+            output.push_back(instr_ir + " " + label_in_assem(TBB));
+            output.push_back("b " + label_in_assem(FBB));
+        }
     } else {
         auto bb = static_cast<BasicBlock *>(instr->get_operand(0));
         output.push_back("b " + label_in_assem(bb));
@@ -513,16 +521,15 @@ CodeGen::pass_arguments(CallInst *instr) {
         vis[i] = true;
     }
 
-    order.clear();
-    for (int arg_id = 1; arg_id < instr->get_num_operand(); arg_id++)
-        order.push_back(arg_id); // initialize
-
     // 3. pass arguments' value
-    assert(order.size() == func->get_num_of_args());
-    map<string, bool> wroten;
+    int passed = 0;
+    // assert(order.size() >= func->get_num_of_args());
+    map<string, bool> changed;
     // bool assigned[8] = {false};
     string t0_contained;
     for (auto arg_id : order) {
+        if (arg_id > func->get_num_of_args())
+            continue;
         auto arg_value = instr->get_operand(arg_id);
         auto arg_is_float = arg_value->get_type()->is_float_type();
         auto arg_reg_aid = regname(arg_id, arg_is_float);
@@ -531,12 +538,12 @@ CodeGen::pass_arguments(CallInst *instr) {
         // arg_reg_aid(only right when arg_id <= 8)
         v_reg = value2reg(arg_value, 1);
         if (backup[arg_id]) { // a_id still relied by some argument due to cycle
-            assert(not wroten[arg_reg_aid]);
+            assert(not changed[arg_reg_aid]);
             gencopy(t_reg, arg_reg_aid, arg_is_float);
             t0_contained = arg_reg_aid;
         }
         // in case that the src register has been wroten
-        if (wroten[v_reg]) {
+        if (changed[v_reg]) {
             assert(t0_contained == v_reg);
             v_reg = arg_value->get_type()->is_float_type() ? "$ft0" : "$t0";
         }
@@ -555,8 +562,12 @@ CodeGen::pass_arguments(CallInst *instr) {
              * $sp, " + to_string(func_arg_off.at(func).at(arg_id)));
              */
         }
-        wroten[arg_reg_aid] = true;
+        if (arg_reg_aid != v_reg) {
+            changed[arg_reg_aid] = true;
+        }
+        passed++;
     }
+    assert(passed == func->get_num_of_args());
 }
 
 void

src/optimization/GVN.cpp

@@ -11,6 +11,7 @@
 #include <algorithm>
 #include <cstddef>
 #include <cstdint>
+#include <deque>
 #include <fstream>
 #include <memory>
 #include <sstream>
@@ -386,8 +387,10 @@ GVN::detectEquivalences() {
 #ifdef __DEBUG__
         std::cout << ++times << "th iteration" << std::endl;
 #endif
-        for (auto &_bb : func_->get_basic_blocks()) {
-            auto bb = &_bb;
+        /* for (auto &_bb : func_->get_basic_blocks()) {
+         *     std::cout << _bb.get_name() << std::endl;
+         *     auto bb = &_bb; */
+        for (auto &bb : BB_DFS_Order) {
             if (bb == Entry)
                 continue;
             // get PIN of bb from predecessor(s)
@@ -922,6 +925,7 @@ GVN::run() {
     for (auto &f : m_->get_functions()) {
         if (f.get_basic_blocks().empty())
             continue;
+        get_dfs_order(&f);
         func_ = &f;
         initPerFunction();
         LOG_INFO << "Processing " << f.get_name();
@@ -953,6 +957,26 @@ GVN::run() {
         gvn_json << "]";
 }
 
+void
+GVN::get_dfs_order(Function *func) {
+    std::map<BasicBlock *, bool> vis;
+    std::deque<BasicBlock *> Q;
+
+    BB_DFS_Order.clear();
+    Q.push_back(func->get_entry_block());
+
+    while (not Q.empty()) {
+        auto bb = Q.front();
+        Q.pop_front();
+        if (vis[bb])
+            continue;
+        vis[bb] = true;
+        BB_DFS_Order.push_back(bb);
+        for (auto succ : bb->get_succ_basic_blocks())
+            Q.push_front(succ);
+    }
+}
+
 void
 GVN::dump_tmp(Function &f) {
     std::string gvn_json;