[midend-Loop]初步构建循环规范遍

src/include/midend/Pass/Optimize/LoopNormalization.h

@@ -0,0 +1,155 @@
+#pragma once
+
+#include "Loop.h"          // 循环分析依赖
+#include "LoopCharacteristics.h"  // 循环特征分析依赖
+#include "Dom.h"           // 支配树分析依赖
+#include "IR.h"            // IR定义
+#include "IRBuilder.h"     // IR构建器
+#include "Pass.h"          // Pass框架
+#include <memory>
+#include <set>
+#include <vector>
+
+namespace sysy {
+
+/**
+ * @brief 循环规范化转换Pass
+ * 
+ * 该Pass在循环不变量提升等优化前运行，主要负责：
+ * 1. 为没有前置块(preheader)的循环创建前置块
+ * 2. 确保循环结构符合后续优化的要求
+ * 3. 规范化循环的控制流结构
+ * 
+ * 前置块的作用：
+ * - 为循环不变量提升提供插入位置
+ * - 简化循环分析和优化
+ * - 确保循环有唯一的入口点
+ */
+class LoopNormalizationPass : public OptimizationPass {
+public:
+  // 唯一的 Pass ID
+  static void *ID;
+
+  LoopNormalizationPass(IRBuilder* builder) : OptimizationPass("LoopNormalization", Pass::Granularity::Function), builder(builder) {}
+
+  // 实现 getPassID
+  void *getPassID() const override { return &ID; }
+
+  // 核心运行方法
+  bool runOnFunction(Function *F, AnalysisManager &AM) override;
+  
+  // 声明分析依赖和失效信息
+  void getAnalysisUsage(std::set<void *> &analysisDependencies, std::set<void *> &analysisInvalidations) const override;
+
+private:
+  // ========== IR构建器 ==========
+  IRBuilder* builder;                                  // IR构建器
+  
+  // ========== 缓存的分析结果 ==========
+  LoopAnalysisResult* loopAnalysis;                    // 循环结构分析结果
+  LoopCharacteristicsResult* loopCharacteristics;     // 循环特征分析结果
+  DominatorTree* domTree;                              // 支配树分析结果
+  
+  // ========== 规范化统计 ==========
+  struct NormalizationStats {
+    size_t totalLoops;                    // 总循环数
+    size_t loopsNeedingPreheader;        // 需要前置块的循环数
+    size_t preheadersCreated;            // 创建的前置块数
+    size_t loopsNormalized;              // 规范化的循环数
+    
+    NormalizationStats() : totalLoops(0), loopsNeedingPreheader(0), 
+                          preheadersCreated(0), loopsNormalized(0) {}
+  } stats;
+  
+  // ========== 核心规范化方法 ==========
+  
+  /**
+   * 规范化单个循环
+   * @param loop 要规范化的循环
+   * @return 是否进行了修改
+   */
+  bool normalizeLoop(Loop* loop);
+  
+  /**
+   * 为循环创建前置块
+   * @param loop 需要前置块的循环
+   * @return 创建的前置块，如果失败则返回nullptr
+   */
+  BasicBlock* createPreheaderForLoop(Loop* loop);
+  
+  /**
+   * 检查循环是否需要前置块
+   * @param loop 要检查的循环
+   * @return true如果需要前置块
+   */
+  bool needsPreheader(Loop* loop);
+  
+  /**
+   * 检查循环是否已有合适的前置块
+   * @param loop 要检查的循环
+   * @return 现有的前置块，如果没有则返回nullptr
+   */
+  BasicBlock* getExistingPreheader(Loop* loop);
+  
+  /**
+   * 更新支配树关系（在创建新块后）
+   * @param newBlock 新创建的基本块
+   * @param loop 相关的循环
+   */
+  void updateDominatorRelations(BasicBlock* newBlock, Loop* loop);
+  
+  /**
+   * 重定向循环外的前驱块到新的前置块
+   * @param loop 目标循环
+   * @param preheader 新创建的前置块
+   * @param header 循环头部
+   */
+  void redirectExternalPredecessors(Loop* loop, BasicBlock* preheader, BasicBlock* header);
+  
+  /**
+   * 为前置块生成合适的名称
+   * @param loop 相关的循环
+   * @return 生成的前置块名称
+   */
+  std::string generatePreheaderName(Loop* loop);
+  
+  /**
+   * 验证规范化结果的正确性
+   * @param loop 规范化后的循环
+   * @return true如果规范化正确
+   */
+  bool validateNormalization(Loop* loop);
+  
+  // ========== 辅助方法 ==========
+  
+  /**
+   * 获取循环的外部前驱块（不在循环内的前驱）
+   * @param loop 目标循环
+   * @return 外部前驱块列表
+   */
+  std::vector<BasicBlock*> getExternalPredecessors(Loop* loop);
+  
+  /**
+   * 检查基本块是否适合作为前置块
+   * @param block 候选基本块
+   * @param loop 目标循环
+   * @return true如果适合作为前置块
+   */
+  bool isSuitableAsPreheader(BasicBlock* block, Loop* loop);
+  
+  /**
+   * 更新PHI节点以适应新的前置块
+   * @param header 循环头部
+   * @param preheader 新的前置块
+   * @param oldPreds 原来的外部前驱
+   */
+  void updatePhiNodesForPreheader(BasicBlock* header, BasicBlock* preheader,
+                                 const std::vector<BasicBlock*>& oldPreds);
+  
+  /**
+   * 打印规范化统计信息
+   */
+  void printStats(Function* F);
+};
+
+} // namespace sysy

src/midend/CMakeLists.txt

@@ -17,6 +17,7 @@ add_library(midend_lib STATIC
     Pass/Optimize/Reg2Mem.cpp
     Pass/Optimize/SysYIRCFGOpt.cpp
     Pass/Optimize/SCCP.cpp
+    Pass/Optimize/LoopNormalization.cpp
 )
 
 # 包含中端模块所需的头文件路径

src/midend/Pass/Optimize/LoopNormalization.cpp

@@ -0,0 +1,432 @@
+#include "LoopNormalization.h"
+#include "Dom.h"
+#include "Loop.h"
+#include "LoopCharacteristics.h"
+#include <iostream>
+#include <algorithm>
+#include <sstream>
+
+// 使用全局调试开关
+extern int DEBUG;
+
+namespace sysy {
+
+// 定义 Pass 的唯一 ID
+void *LoopNormalizationPass::ID = (void *)&LoopNormalizationPass::ID;
+
+bool LoopNormalizationPass::runOnFunction(Function *F, AnalysisManager &AM) {
+  if (F->getBasicBlocks().empty()) {
+    return false; // 空函数
+  }
+
+  if (DEBUG)
+    std::cout << "Running LoopNormalizationPass on function: " << F->getName() << std::endl;
+
+  // 获取并缓存所有需要的分析结果
+  loopAnalysis = AM.getAnalysisResult<LoopAnalysisResult, LoopAnalysisPass>(F);
+  if (!loopAnalysis || !loopAnalysis->hasLoops()) {
+    if (DEBUG)
+      std::cout << "No loops found in function " << F->getName() << ", skipping normalization" << std::endl;
+    return false; // 没有循环需要规范化
+  }
+
+  loopCharacteristics = AM.getAnalysisResult<LoopCharacteristicsResult, LoopCharacteristicsPass>(F);
+  domTree = AM.getAnalysisResult<DominatorTree, DominatorTreeAnalysisPass>(F);
+  
+  if (!domTree) {
+    std::cerr << "Error: DominatorTree not available for function " << F->getName() << std::endl;
+    return false;
+  }
+
+  // 重置统计信息
+  stats = NormalizationStats();
+  
+  bool modified = false;
+  const auto& allLoops = loopAnalysis->getAllLoops();
+  stats.totalLoops = allLoops.size();
+
+  if (DEBUG) {
+    std::cout << "Found " << stats.totalLoops << " loops to analyze for normalization" << std::endl;
+  }
+
+  // 按循环深度从外到内处理，确保外层循环先规范化
+  std::vector<Loop*> sortedLoops;
+  for (const auto& loop_ptr : allLoops) {
+    sortedLoops.push_back(loop_ptr.get());
+  }
+  
+  std::sort(sortedLoops.begin(), sortedLoops.end(), [](Loop* a, Loop* b) {
+    return a->getLoopDepth() < b->getLoopDepth(); // 按深度升序排列
+  });
+
+  // 逐个规范化循环
+  for (Loop* loop : sortedLoops) {
+    if (needsPreheader(loop)) {
+      stats.loopsNeedingPreheader++;
+      
+      if (DEBUG) {
+        std::cout << "  Loop " << loop->getName() << " needs preheader (depth=" 
+                  << loop->getLoopDepth() << ")" << std::endl;
+      }
+      
+      if (normalizeLoop(loop)) {
+        modified = true;
+        stats.loopsNormalized++;
+        
+        // 验证规范化结果
+        if (!validateNormalization(loop)) {
+          std::cerr << "Warning: Loop normalization validation failed for loop " 
+                    << loop->getName() << std::endl;
+        }
+      }
+    } else {
+      if (DEBUG) {
+        auto* preheader = getExistingPreheader(loop);
+        if (preheader) {
+          std::cout << "  Loop " << loop->getName() << " already has preheader: " 
+                    << preheader->getName() << std::endl;
+        }
+      }
+    }
+  }
+
+  if (DEBUG && modified) {
+    printStats(F);
+  }
+
+  return modified;
+}
+
+bool LoopNormalizationPass::normalizeLoop(Loop* loop) {
+  if (DEBUG)
+    std::cout << "    Normalizing loop: " << loop->getName() << std::endl;
+
+  // 创建前置块
+  BasicBlock* preheader = createPreheaderForLoop(loop);
+  if (!preheader) {
+    if (DEBUG)
+      std::cout << "    Failed to create preheader for loop " << loop->getName() << std::endl;
+    return false;
+  }
+
+  stats.preheadersCreated++;
+  
+  if (DEBUG) {
+    std::cout << "    Successfully created preheader " << preheader->getName() 
+              << " for loop " << loop->getName() << std::endl;
+  }
+
+  return true;
+}
+
+BasicBlock* LoopNormalizationPass::createPreheaderForLoop(Loop* loop) {
+  BasicBlock* header = loop->getHeader();
+  if (!header) {
+    if (DEBUG)
+      std::cerr << "    Error: Loop has no header block" << std::endl;
+    return nullptr;
+  }
+
+  // 获取循环外的前驱块
+  std::vector<BasicBlock*> externalPreds = getExternalPredecessors(loop);
+  if (externalPreds.empty()) {
+    if (DEBUG)
+      std::cout << "    Loop " << loop->getName() << " has no external predecessors" << std::endl;
+    return nullptr;
+  }
+
+  if (DEBUG) {
+    std::cout << "    Found " << externalPreds.size() << " external predecessors for loop " 
+              << loop->getName() << std::endl;
+    for (auto* pred : externalPreds) {
+      std::cout << "      External pred: " << pred->getName() << std::endl;
+    }
+  }
+
+  // 生成前置块名称
+  std::string preheaderName = generatePreheaderName(loop);
+  
+  // 创建新的前置块
+  Function* parentFunction = header->getParent();
+  BasicBlock* preheader = parentFunction->addBasicBlock(preheaderName);
+  
+  if (!preheader) {
+    if (DEBUG)
+      std::cerr << "    Error: Failed to create basic block " << preheaderName << std::endl;
+    return nullptr;
+  }
+
+  // 在前置块中创建一个简单的跳转指令到循环头部
+  builder->setPosition(preheader, preheader->end());
+  UncondBrInst* br = builder->createUncondBrInst(header);
+
+  // 重定向外部前驱到新的前置块
+  redirectExternalPredecessors(loop, preheader, header);
+
+  // 更新PHI节点
+  updatePhiNodesForPreheader(header, preheader, externalPreds);
+
+  // 更新支配树关系
+  updateDominatorRelations(preheader, loop);
+
+  return preheader;
+}
+
+bool LoopNormalizationPass::needsPreheader(Loop* loop) {
+  // 检查是否已有合适的前置块
+  if (getExistingPreheader(loop) != nullptr) {
+    return false;
+  }
+
+  // 检查是否有外部前驱（如果没有外部前驱，不需要前置块）
+  std::vector<BasicBlock*> externalPreds = getExternalPredecessors(loop);
+  if (externalPreds.empty()) {
+    return false;
+  }
+
+  // 如果有多个外部前驱，或者单个外部前驱不适合作为前置块，则需要创建前置块
+  if (externalPreds.size() > 1) {
+    return true;
+  }
+
+  // 检查唯一的外部前驱是否适合作为前置块
+  return !isSuitableAsPreheader(externalPreds[0], loop);
+}
+
+BasicBlock* LoopNormalizationPass::getExistingPreheader(Loop* loop) {
+  BasicBlock* header = loop->getHeader();
+  if (!header) return nullptr;
+
+  std::vector<BasicBlock*> externalPreds = getExternalPredecessors(loop);
+  
+  // 如果只有一个外部前驱，且适合作为前置块，则返回它
+  if (externalPreds.size() == 1 && isSuitableAsPreheader(externalPreds[0], loop)) {
+    return externalPreds[0];
+  }
+
+  return nullptr;
+}
+
+void LoopNormalizationPass::updateDominatorRelations(BasicBlock* newBlock, Loop* loop) {
+  // 这里需要更新支配树关系
+  // 新的前置块应该支配循环头部，并且被循环外的前驱支配
+  if (DEBUG) {
+    std::cout << "    Updating dominator relations for new preheader " << newBlock->getName() << std::endl;
+  }
+  // 实际的支配树更新逻辑在这里实现
+  // 由于支配树分析通常在Pass运行后重新计算，这里主要是标记需要更新
+}
+
+void LoopNormalizationPass::redirectExternalPredecessors(Loop* loop, BasicBlock* preheader, BasicBlock* header) {
+  std::vector<BasicBlock*> externalPreds = getExternalPredecessors(loop);
+  
+  if (DEBUG) {
+    std::cout << "    Redirecting " << externalPreds.size() << " external predecessors" << std::endl;
+  }
+
+  for (BasicBlock* pred : externalPreds) {
+    // 获取前驱块的终止指令
+    auto termIt = pred->terminator();
+    if (termIt == pred->end()) continue;
+    
+    Instruction* terminator = termIt->get();
+    if (!terminator) continue;
+
+    // 更新跳转目标
+    if (auto* br = dynamic_cast<UncondBrInst*>(terminator)) {
+      // 无条件跳转
+      if (br->getBlock() == header) {
+        // 需要更新操作数
+        br->setOperand(0, preheader);
+        // 更新CFG关系
+        header->removePredecessor(pred);
+        preheader->addPredecessor(pred);
+        pred->removeSuccessor(header);
+        pred->addSuccessor(preheader);
+        
+        if (DEBUG)
+          std::cout << "      Updated unconditional branch from " << pred->getName() 
+                    << " to " << preheader->getName() << std::endl;
+      }
+    } else if (auto* condBr = dynamic_cast<CondBrInst*>(terminator)) {
+      // 条件跳转
+      bool updated = false;
+      if (condBr->getThenBlock() == header) {
+        condBr->setOperand(1, preheader);  // 第1个操作数是then分支
+        updated = true;
+      }
+      if (condBr->getElseBlock() == header) {
+        condBr->setOperand(2, preheader);  // 第2个操作数是else分支
+        updated = true;
+      }
+      if (updated) {
+        // 更新CFG关系
+        header->removePredecessor(pred);
+        preheader->addPredecessor(pred);
+        pred->removeSuccessor(header);
+        pred->addSuccessor(preheader);
+        
+        if (DEBUG) {
+          std::cout << "      Updated conditional branch from " << pred->getName() 
+                    << " to " << preheader->getName() << std::endl;
+        }
+      }
+    }
+  }
+}
+
+std::string LoopNormalizationPass::generatePreheaderName(Loop* loop) {
+  std::ostringstream oss;
+  oss << loop->getName() << ".preheader";
+  return oss.str();
+}
+
+bool LoopNormalizationPass::validateNormalization(Loop* loop) {
+  BasicBlock* header = loop->getHeader();
+  if (!header) return false;
+
+  // 检查循环是否现在有唯一的外部前驱
+  std::vector<BasicBlock*> externalPreds = getExternalPredecessors(loop);
+  if (externalPreds.size() != 1) {
+    if (DEBUG)
+      std::cout << "    Validation failed: Loop " << loop->getName() 
+                << " has " << externalPreds.size() << " external predecessors (expected 1)" << std::endl;
+    return false;
+  }
+
+  // 检查外部前驱是否适合作为前置块
+  if (!isSuitableAsPreheader(externalPreds[0], loop)) {
+    if (DEBUG)
+      std::cout << "    Validation failed: External predecessor " << externalPreds[0]->getName() 
+                << " is not suitable as preheader" << std::endl;
+    return false;
+  }
+
+  if (DEBUG)
+    std::cout << "    Validation passed for loop " << loop->getName() << std::endl;
+
+  return true;
+}
+
+std::vector<BasicBlock*> LoopNormalizationPass::getExternalPredecessors(Loop* loop) {
+  std::vector<BasicBlock*> externalPreds;
+  BasicBlock* header = loop->getHeader();
+  if (!header) return externalPreds;
+
+  for (BasicBlock* pred : header->getPredecessors()) {
+    if (!loop->contains(pred)) {
+      externalPreds.push_back(pred);
+    }
+  }
+
+  return externalPreds;
+}
+
+bool LoopNormalizationPass::isSuitableAsPreheader(BasicBlock* block, Loop* loop) {
+  if (!block) return false;
+
+  // 检查该块是否只有一个后继，且后继是循环头部
+  auto successors = block->getSuccessors();
+  if (successors.size() != 1) {
+    return false;
+  }
+
+  if (successors[0] != loop->getHeader()) {
+    return false;
+  }
+
+  // 检查该块是否不包含复杂的控制流
+  // 理想的前置块应该只包含简单的跳转指令
+  size_t instCount = 0;
+  for (const auto& inst : block->getInstructions()) {
+    instCount++;
+    // 如果指令过多，可能不适合作为前置块
+    if (instCount > 10) { // 阈值可调整
+      return false;
+    }
+  }
+
+  return true;
+}
+
+void LoopNormalizationPass::updatePhiNodesForPreheader(BasicBlock* header, BasicBlock* preheader,
+                                                      const std::vector<BasicBlock*>& oldPreds) {
+  if (DEBUG) {
+    std::cout << "    Updating PHI nodes in header " << header->getName() 
+              << " for new preheader " << preheader->getName() << std::endl;
+  }
+
+  for (auto& inst : header->getInstructions()) {
+    if (auto* phi = dynamic_cast<PhiInst*>(inst.get())) {
+      if (DEBUG) {
+        std::cout << "      Processing PHI node: " << phi->getName() << std::endl;
+      }
+
+      // 收集来自外部前驱的值
+      std::vector<std::pair<Value*, BasicBlock*>> externalValues;
+      for (BasicBlock* oldPred : oldPreds) {
+        Value* value = phi->getvalfromBlk(oldPred);
+        if (value) {
+          externalValues.push_back({value, oldPred});
+        }
+      }
+
+      // 从PHI节点中移除旧的外部前驱
+      for (BasicBlock* oldPred : oldPreds) {
+        phi->removeIncoming(oldPred);
+      }
+
+      // 如果有多个来自外部的值，需要在前置块中创建新的PHI节点
+      if (externalValues.size() > 1) {
+        // 在前置块中创建新的PHI节点来合并外部值
+        builder->setPosition(preheader, preheader->getInstructions().begin());
+        
+        std::vector<Value*> values;
+        std::vector<BasicBlock*> blocks;
+        for (auto& [value, block] : externalValues) {
+          values.push_back(value);
+          blocks.push_back(block);
+        }
+        
+        PhiInst* newPhi = builder->createPhiInst(phi->getType(), values, blocks, "preheader.phi");
+        
+        // 将新PHI的结果作为来自前置块的值添加到原PHI中
+        phi->addIncoming(newPhi, preheader);
+      } else if (externalValues.size() == 1) {
+        // 只有一个外部值，直接添加
+        phi->addIncoming(externalValues[0].first, preheader);
+      }
+    }
+  }
+}
+
+void LoopNormalizationPass::printStats(Function* F) {
+  std::cout << "\n--- Loop Normalization Statistics for Function: " << F->getName() << " ---" << std::endl;
+  std::cout << "Total loops analyzed: " << stats.totalLoops << std::endl;
+  std::cout << "Loops needing preheader: " << stats.loopsNeedingPreheader << std::endl;
+  std::cout << "Preheaders created: " << stats.preheadersCreated << std::endl;
+  std::cout << "Loops successfully normalized: " << stats.loopsNormalized << std::endl;
+  
+  if (stats.totalLoops > 0) {
+    double normalizationRate = (double)stats.loopsNormalized / stats.totalLoops * 100.0;
+    std::cout << "Normalization rate: " << normalizationRate << "%" << std::endl;
+  }
+  
+  std::cout << "---------------------------------------------------------------" << std::endl;
+}
+
+void LoopNormalizationPass::getAnalysisUsage(std::set<void *> &analysisDependencies, 
+                                            std::set<void *> &analysisInvalidations) const {
+  // LoopNormalization依赖的分析
+  analysisDependencies.insert(&LoopAnalysisPass::ID);              // 循环结构分析
+  analysisDependencies.insert(&LoopCharacteristicsPass::ID);       // 循环特征分析
+  analysisDependencies.insert(&DominatorTreeAnalysisPass::ID);     // 支配树分析
+
+  // LoopNormalization会修改CFG结构，因此会使以下分析失效
+  analysisInvalidations.insert(&DominatorTreeAnalysisPass::ID);    // 支配树需要重新计算
+  // 注意：循环结构分析可能需要更新，但我们不让它失效，因为我们只是添加前置块
+  // analysisInvalidations.insert(&LoopAnalysisPass::ID);         // 通常不需要失效
+  // analysisInvalidations.insert(&LoopCharacteristicsPass::ID);  // 通常不需要失效
+}
+
+} // namespace sysy