lab4 publish

include/optimization/Dominators.h

+#ifndef SYSYC_DOMINATORS_HPP
+#define SYSYC_DOMINATORS_HPP
+
+#include "BasicBlock.h"
+#include "PassManager.hpp"
+
+#include <list>
+#include <map>
+#include <set>
+
+class Dominators : public Pass {
+  public:
+    explicit Dominators(Module *m) : Pass(m) {}
+    ~Dominators() = default;
+    void run() override;
+    void create_doms(Function *f);
+    void create_reverse_post_order(Function *f);
+    void create_idom(Function *f);
+    void create_dominance_frontier(Function *f);
+    void create_dom_tree_succ(Function *f);
+
+    // for debug
+    void print_idom(Function *f);
+    void print_dominance_frontier(Function *f);
+
+    /****************api about Dominator****************/
+
+    void add_dom(BasicBlock *bb, BasicBlock *dom_bb) { doms_[bb].insert(dom_bb); }
+    std::set<BasicBlock *> &get_doms(BasicBlock *bb) { return doms_[bb]; }
+    void set_doms(BasicBlock *bb, std::set<BasicBlock *> &doms) {
+        doms_[bb].clear();
+        doms_[bb].insert(doms.begin(), doms.end());
+    }
+
+    BasicBlock *get_idom(BasicBlock *bb) { return idom_[bb]; }
+    void set_idom(BasicBlock *bb, BasicBlock *idom) { idom_[bb] = idom; }
+
+    void add_dominance_frontier(BasicBlock *bb, BasicBlock *dom_frontier_bb) {
+        dom_frontier_[bb].insert(dom_frontier_bb);
+    }
+    std::set<BasicBlock *> &get_dominance_frontier(BasicBlock *bb) { return dom_frontier_[bb]; }
+    void set_dominance_frontier(BasicBlock *bb, std::set<BasicBlock *> &df) {
+        dom_frontier_[bb].clear();
+        dom_frontier_[bb].insert(df.begin(), df.end());
+    }
+
+    // successor blocks of this node in dominance tree
+    std::set<BasicBlock *> get_dom_tree_succ_blocks(BasicBlock *bb) { return dom_tree_succ_blocks_[bb]; }
+    void add_dom_tree_succ_block(BasicBlock *bb, BasicBlock *dom_tree_succ_bb) {
+        dom_tree_succ_blocks_[bb].insert(dom_tree_succ_bb);
+    }
+    /****************api about Dominator****************/
+
+  private:
+    void post_order_visit(BasicBlock *bb, std::set<BasicBlock *> &visited);
+    BasicBlock *intersect(BasicBlock *b1, BasicBlock *b2);
+
+    std::list<BasicBlock *> reverse_post_order_{};
+    std::map<BasicBlock *, int> post_order_id_{}; // the root has highest ID
+
+    std::map<BasicBlock *, std::set<BasicBlock *>> doms_{};         // dominance set
+    std::map<BasicBlock *, BasicBlock *> idom_{};                   // immediate dominance
+    std::map<BasicBlock *, std::set<BasicBlock *>> dom_frontier_{}; // dominance frontier set
+    std::map<BasicBlock *, std::set<BasicBlock *>> dom_tree_succ_blocks_{};
+};
+
+#endif

include/optimization/Mem2Reg.hpp

+#ifndef SYSYC_MEM2REG_HPP
+#define SYSYC_MEM2REG_HPP
+
+#include "BasicBlock.h"
+#include "Dominators.h"
+#include "Function.h"
+#include "IRBuilder.h"
+#include "Instruction.h"
+#include "Module.h"
+#include "PassManager.hpp"
+
+#include <memory>
+
+class Mem2Reg : public Pass {
+  private:
+    Function *func_;
+    std::unique_ptr<Dominators> dominators_;
+
+  public:
+    Mem2Reg(Module *m) : Pass(m) {}
+    ~Mem2Reg(){};
+    void run() override;
+    void generate_phi();
+    void re_name(BasicBlock *bb);
+    void remove_alloca();
+};
+
+#endif
\ No newline at end of file

include/optimization/PassManager.hpp

+#ifndef PASSMANAGER_HPP
+#define PASSMANAGER_HPP
+
+#include "Module.h"
+
+#include <memory>
+#include <vector>
+
+class Pass {
+  public:
+    Pass(Module *m) : m_(m) {}
+    virtual ~Pass() = default;
+    virtual void run() = 0;
+
+  protected:
+    Module *m_;
+};
+
+class PassManager {
+  public:
+    PassManager(Module *m) : m_(m) {}
+    template <typename PassType, typename... Args>
+    void add_pass(bool print_ir, Args &&...args) {
+        passes_.push_back(
+            std::pair<std::unique_ptr<Pass>, bool>(new PassType(m_, std::forward<Args>(args)...), print_ir));
+    }
+    void run() {
+        for (auto &pass : passes_) {
+            pass.first->run();
+            if (pass.second) {
+                m_->set_print_name();
+                std::cout << m_->print();
+            }
+        }
+    }
+
+  private:
+    std::vector<std::pair<std::unique_ptr<Pass>, bool>> passes_;
+    Module *m_;
+};
+
+#endif
\ No newline at end of file

src/CMakeLists.txt

@@ -2,4 +2,5 @@ add_subdirectory(parser)
 add_subdirectory(common)
 add_subdirectory(io)
 add_subdirectory(lightir)
-add_subdirectory(cminusfc)
\ No newline at end of file
+add_subdirectory(cminusfc)
+add_subdirectory(optimization)
\ No newline at end of file

src/cminusfc/cminusfc.cpp

+#include "Dominators.h"
+#include "Mem2Reg.hpp"
+#include "PassManager.hpp"
 #include "cminusf_builder.hpp"
+#include "logging.hpp"
+
 #include <fstream>
 #include <iostream>
 #include <memory>
@@ -6,20 +11,15 @@
 using namespace std::literals::string_literals;
 
 void print_help(std::string exe_name) {
-    std::cout << "Usage: " << exe_name <<
-        " [ -h | --help ] [ -o <target-file> ] [ -emit-llvm ] <input-file>" << std::endl;
+    std::cout << "Usage: " << exe_name << " [ -h | --help ] [ -o <target-file> ] [ -emit-llvm ] [-mem2reg] <input-file>"
+              << std::endl;
 }
 
-
 int main(int argc, char **argv) {
     std::string target_path;
     std::string input_path;
 
     bool mem2reg = false;
-    bool const_propagation = false;
-    bool activevars = false;
-    bool loop_inv_hoist = false;
-    bool loop_search = false;
     bool emit = false;
 
     for (int i = 1; i < argc; ++i) {
@@ -38,14 +38,6 @@ int main(int argc, char **argv) {
             emit = true;
         } else if (argv[i] == "-mem2reg"s) {
             mem2reg = true;
-        } else if (argv[i] == "-loop-search"s) {
-            loop_search = true;
-        } else if (argv[i] == "-loop-inv-hoist"s) {
-            loop_inv_hoist = true;
-        } else if (argv[i] == "-const-propagation"s) {
-            const_propagation = true;
-        } else if (argv[i] == "-active-vars"s) {
-            activevars = true;
         } else {
             if (input_path.empty()) {
                 input_path = argv[i];
@@ -85,6 +77,14 @@ int main(int argc, char **argv) {
 
     auto m = builder.getModule();
 
+    m->set_print_name();
+    PassManager PM(m.get());
+
+    if (mem2reg) {
+        PM.add_pass<Mem2Reg>(emit);
+    }
+    PM.run();
+
     auto IR = m->print();
 
     std::ofstream output_stream;

src/optimization/CMakeLists.txt

+add_library(
+        OP_lib STATIC
+        Dominators.cpp
+        Mem2Reg.cpp
+)

src/optimization/Dominators.cpp

+#include "Dominators.h"
+
+#include <algorithm>
+#include <string>
+
+void Dominators::run() {
+    for (auto &f1 : m_->get_functions()) {
+        auto f = &f1;
+        if (f->get_basic_blocks().size() == 0)
+            continue;
+        for (auto &bb1 : f->get_basic_blocks()) {
+            auto bb = &bb1;
+            doms_.insert({bb, {}});
+            idom_.insert({bb, {}});
+            dom_frontier_.insert({bb, {}});
+            dom_tree_succ_blocks_.insert({bb, {}});
+        }
+
+        create_reverse_post_order(f);
+        create_idom(f);
+        create_dominance_frontier(f);
+        create_dom_tree_succ(f);
+        // for debug
+        // print_idom(f);
+        // print_dominance_frontier(f);
+    }
+}
+
+void Dominators::create_doms(Function *f) {
+    // init
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        add_dom(bb, bb);
+    }
+    // iterate
+    bool changed = true;
+    std::vector<BasicBlock *> ret(f->get_num_basic_blocks());
+    std::vector<BasicBlock *> pre(f->get_num_basic_blocks());
+    while (changed) {
+        changed = false;
+        for (auto &bb1 : f->get_basic_blocks()) {
+            auto bb = &bb1;
+            auto &bbs = bb->get_pre_basic_blocks();
+            auto &first = get_doms((*bbs.begin()));
+            pre.insert(pre.begin(), first.begin(), first.end());
+            pre.resize(first.size());
+            ret.resize(f->get_num_basic_blocks());
+            for (auto iter = ++bbs.begin(); iter != bbs.end(); ++iter) {
+                auto &now = get_doms((*iter));
+                auto it = std::set_intersection(pre.begin(), pre.end(), now.begin(), now.end(), ret.begin());
+                ret.resize(it - ret.begin());
+                pre.resize(ret.size());
+                pre.insert(pre.begin(), ret.begin(), ret.end());
+            }
+            std::set<BasicBlock *> doms;
+            doms.insert(bb);
+            doms.insert(pre.begin(), pre.end());
+            if (get_doms(bb) != doms) {
+                set_doms(bb, doms);
+                changed = true;
+            }
+        }
+    }
+}
+
+void Dominators::create_reverse_post_order(Function *f) {
+    reverse_post_order_.clear();
+    post_order_id_.clear();
+    std::set<BasicBlock *> visited;
+    post_order_visit(f->get_entry_block(), visited);
+    reverse_post_order_.reverse();
+}
+
+void Dominators::post_order_visit(BasicBlock *bb, std::set<BasicBlock *> &visited) {
+    visited.insert(bb);
+    for (auto b : bb->get_succ_basic_blocks()) {
+        if (visited.find(b) == visited.end())
+            post_order_visit(b, visited);
+    }
+    post_order_id_[bb] = reverse_post_order_.size();
+    reverse_post_order_.push_back(bb);
+}
+
+void Dominators::create_idom(Function *f) {
+    // init
+    for (auto &bb : f->get_basic_blocks())
+        set_idom(&bb, nullptr);
+    auto root = f->get_entry_block();
+    set_idom(root, root);
+
+    // iterate
+    bool changed = true;
+    while (changed) {
+        changed = false;
+        for (auto bb : this->reverse_post_order_) {
+            if (bb == root) {
+                continue;
+            }
+
+            // find one pred which has idom
+            BasicBlock *pred = nullptr;
+            for (auto p : bb->get_pre_basic_blocks()) {
+                if (get_idom(p)) {
+                    pred = p;
+                    break;
+                }
+            }
+            assert(pred);
+
+            BasicBlock *new_idom = pred;
+            for (auto p : bb->get_pre_basic_blocks()) {
+                if (p == pred)
+                    continue;
+                if (get_idom(p)) {
+                    new_idom = intersect(p, new_idom);
+                }
+            }
+            if (get_idom(bb) != new_idom) {
+                set_idom(bb, new_idom);
+                changed = true;
+            }
+        }
+    }
+}
+
+// find closest parent of b1 and b2
+BasicBlock *Dominators::intersect(BasicBlock *b1, BasicBlock *b2) {
+    while (b1 != b2) {
+        while (post_order_id_[b1] < post_order_id_[b2]) {
+            assert(get_idom(b1));
+            b1 = get_idom(b1);
+        }
+        while (post_order_id_[b2] < post_order_id_[b1]) {
+            assert(get_idom(b2));
+            b2 = get_idom(b2);
+        }
+    }
+    return b1;
+}
+
+void Dominators::create_dominance_frontier(Function *f) {
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        if (bb->get_pre_basic_blocks().size() >= 2) {
+            for (auto p : bb->get_pre_basic_blocks()) {
+                auto runner = p;
+                while (runner != get_idom(bb)) {
+                    add_dominance_frontier(runner, bb);
+                    runner = get_idom(runner);
+                }
+            }
+        }
+    }
+}
+
+void Dominators::create_dom_tree_succ(Function *f) {
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        auto idom = get_idom(bb);
+        // e.g, entry bb
+        if (idom != bb) {
+            add_dom_tree_succ_block(idom, bb);
+        }
+    }
+}
+
+void Dominators::print_idom(Function *f) {
+    int counter = 0;
+    std::map<BasicBlock *, std::string> bb_id;
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        if (bb->get_name().empty())
+            bb_id[bb] = "bb" + std::to_string(counter);
+        else
+            bb_id[bb] = bb->get_name();
+        counter++;
+    }
+    printf("Immediate dominance of function %s:\n", f->get_name().c_str());
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        std::string output;
+        output = bb_id[bb] + ": ";
+        if (get_idom(bb)) {
+            output += bb_id[get_idom(bb)];
+        } else {
+            output += "null";
+        }
+        printf("%s\n", output.c_str());
+    }
+}
+
+void Dominators::print_dominance_frontier(Function *f) {
+    int counter = 0;
+    std::map<BasicBlock *, std::string> bb_id;
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        if (bb->get_name().empty())
+            bb_id[bb] = "bb" + std::to_string(counter);
+        else
+            bb_id[bb] = bb->get_name();
+        counter++;
+    }
+    printf("Dominance Frontier of function %s:\n", f->get_name().c_str());
+    for (auto &bb1 : f->get_basic_blocks()) {
+        auto bb = &bb1;
+        std::string output;
+        output = bb_id[bb] + ": ";
+        if (get_dominance_frontier(bb).empty()) {
+            output += "null";
+        } else {
+            bool first = true;
+            for (auto df : get_dominance_frontier(bb)) {
+                if (first) {
+                    first = false;
+                } else {
+                    output += ", ";
+                }
+                output += bb_id[df];
+            }
+        }
+        printf("%s\n", output.c_str());
+    }
+}

src/optimization/Mem2Reg.cpp

+#include "Mem2Reg.hpp"
+
+#include "IRBuilder.h"
+
+#include <memory>
+
+// 判断是否是全局变量地址
+#define IS_GLOBAL_VARIABLE(l_val) dynamic_cast<GlobalVariable *>(l_val)
+// 判断是否是 getelementptr 指令
+#define IS_GEP_INSTR(l_val) dynamic_cast<GetElementPtrInst *>(l_val)
+
+std::map<Value *, std::vector<Value *>> var_val_stack; // 全局变量初值提前存入栈中
+
+void Mem2Reg::run() {
+    // 创建支配树分析 Pass 的实例
+    dominators_ = std::make_unique<Dominators>(m_);
+    // 建立支配树
+    dominators_->run();
+    // 以函数为单元遍历实现 Mem2Reg 算法
+    for (auto &f : m_->get_functions()) {
+        func_ = &f;
+        if (func_->get_basic_blocks().size() >= 1) {
+            // 对应伪代码中 phi 指令插入的阶段
+            generate_phi();
+            // 对应伪代码中重命名阶段
+            re_name(func_->get_entry_block());
+        }
+        // 移除冗余的局部变量的分配空间
+        remove_alloca();
+    }
+}
+
+void Mem2Reg::generate_phi() {
+    // global_live_var_name 是全局名字集合，以 alloca 出的局部变量来统计。
+    // 步骤一：找到活跃在多个 block 的全局名字集合，以及它们所属的 bb 块
+    std::set<Value *> global_live_var_name;
+    std::map<Value *, std::set<BasicBlock *>> live_var_2blocks;
+    for (auto &bb1 : func_->get_basic_blocks()) {
+        auto bb = &bb1;
+        std::set<Value *> var_is_killed;
+        for (auto &instr1 : bb->get_instructions()) {
+            auto instr = &instr1;
+            if (instr->is_store()) {
+                // store i32 a, i32 *b
+                // a is r_val, b is l_val
+                auto r_val = static_cast<StoreInst *>(instr)->get_rval();
+                auto l_val = static_cast<StoreInst *>(instr)->get_lval();
+
+                if (!IS_GLOBAL_VARIABLE(l_val) && !IS_GEP_INSTR(l_val)) {
+                    global_live_var_name.insert(l_val);
+                    live_var_2blocks[l_val].insert(bb);
+                }
+            }
+        }
+    }
+
+    // 步骤二：从支配树获取支配边界信息，并在对应位置插入 phi 指令
+    std::map<std::pair<BasicBlock *, Value *>, bool> bb_has_var_phi; // bb has phi for var
+    for (auto var : global_live_var_name) {
+        std::vector<BasicBlock *> work_list;
+        work_list.assign(live_var_2blocks[var].begin(), live_var_2blocks[var].end());
+        for (int i = 0; i < work_list.size(); i++) {
+            auto bb = work_list[i];
+            for (auto bb_dominance_frontier_bb : dominators_->get_dominance_frontier(bb)) {
+                if (bb_has_var_phi.find({bb_dominance_frontier_bb, var}) == bb_has_var_phi.end()) {
+                    // generate phi for bb_dominance_frontier_bb & add bb_dominance_frontier_bb to work list
+                    auto phi =
+                        PhiInst::create_phi(var->get_type()->get_pointer_element_type(), bb_dominance_frontier_bb);
+                    phi->set_lval(var);
+                    bb_dominance_frontier_bb->add_instr_begin(phi);
+                    work_list.push_back(bb_dominance_frontier_bb);
+                    bb_has_var_phi[{bb_dominance_frontier_bb, var}] = true;
+                }
+            }
+        }
+    }
+}
+
+void Mem2Reg::re_name(BasicBlock *bb) {
+    std::vector<Instruction *> wait_delete;
+
+    // 步骤三：将 phi 指令作为 lval 的最新定值，lval 即是为局部变量 alloca 出的地址空间
+    for (auto &instr1 : bb->get_instructions()) {
+        auto instr = &instr1;
+        if (instr->is_phi()) {
+            auto l_val = static_cast<PhiInst *>(instr)->get_lval();
+            var_val_stack[l_val].push_back(instr);
+        }
+    }
+
+    for (auto &instr1 : bb->get_instructions()) {
+        auto instr = &instr1;
+        // 步骤四：用 lval 最新的定值替代对应的load指令
+        if (instr->is_load()) {
+            auto l_val = static_cast<LoadInst *>(instr)->get_lval();
+
+            if (!IS_GLOBAL_VARIABLE(l_val) && !IS_GEP_INSTR(l_val)) {
+                if (var_val_stack.find(l_val) != var_val_stack.end()) {
+                    // 此处指令替换会维护 UD 链与 DU 链
+                    instr->replace_all_use_with(var_val_stack[l_val].back());
+                    wait_delete.push_back(instr);
+                }
+            }
+        }
+        // 步骤五：将 store 指令的 rval，也即被存入内存的值，作为 lval 的最新定值
+        if (instr->is_store()) {
+            auto l_val = static_cast<StoreInst *>(instr)->get_lval();
+            auto r_val = static_cast<StoreInst *>(instr)->get_rval();
+
+            if (!IS_GLOBAL_VARIABLE(l_val) && !IS_GEP_INSTR(l_val)) {
+                var_val_stack[l_val].push_back(r_val);
+                wait_delete.push_back(instr);
+            }
+        }
+    }
+
+    // 步骤六：为 lval 对应的 phi 指令参数补充完整
+    for (auto succ_bb : bb->get_succ_basic_blocks()) {
+        for (auto &instr1 : succ_bb->get_instructions()) {
+            auto instr = &instr1;
+            if (instr->is_phi()) {
+                auto l_val = static_cast<PhiInst *>(instr)->get_lval();
+                if (var_val_stack.find(l_val) != var_val_stack.end()) {
+                    assert(var_val_stack[l_val].size() != 0);
+                    static_cast<PhiInst *>(instr)->add_phi_pair_operand(var_val_stack[l_val].back(), bb);
+                }
+                // 对于 phi 参数只有一个前驱定值的情况，将会输出 [ undef, bb ] 的参数格式
+            }
+        }
+    }
+
+    // 步骤七：对 bb 在支配树上的所有后继节点，递归执行 re_name 操作
+    for (auto dom_succ_bb : dominators_->get_dom_tree_succ_blocks(bb)) {
+        re_name(dom_succ_bb);
+    }
+
+    // 步骤八：pop出 lval 的最新定值
+    for (auto &instr1 : bb->get_instructions()) {
+        auto instr = &instr1;
+
+        if (instr->is_store()) {
+            auto l_val = static_cast<StoreInst *>(instr)->get_lval();
+            if (!IS_GLOBAL_VARIABLE(l_val) && !IS_GEP_INSTR(l_val)) {
+                var_val_stack[l_val].pop_back();
+            }
+        } else if (instr->is_phi()) {
+            auto l_val = static_cast<PhiInst *>(instr)->get_lval();
+            if (var_val_stack.find(l_val) != var_val_stack.end()) {
+                var_val_stack[l_val].pop_back();
+            }
+        }
+    }
+
+    // 清除冗余的指令
+    for (auto instr : wait_delete) {
+        bb->erase_instr(instr);
+    }
+}
+
+void Mem2Reg::remove_alloca() {
+    for (auto &bb1 : func_->get_basic_blocks()) {
+        auto bb = &bb1;
+        std::vector<Instruction *> wait_delete;
+        for (auto &instr1 : bb->get_instructions()) {
+            auto instr = &instr1;
+            auto is_alloca = dynamic_cast<AllocaInst *>(instr);
+            if (is_alloca) {
+                bool is_int = is_alloca->get_type()->get_pointer_element_type()->is_integer_type();
+                bool is_float = is_alloca->get_type()->get_pointer_element_type()->is_float_type();
+                if (is_int || is_float) {
+                    wait_delete.push_back(instr);
+                }
+            }
+        }
+        for (auto instr : wait_delete) {
+            bb->erase_instr(instr);
+        }
+    }
+}