Extended calculated statistics (#546)

* extended statistics

* variable renaming for statistics calculation

* added statistics entry for PHI nodes

* varibale name typo in statistics fixed

* removed separate counter for terminators

* overload of << operator for GeneralStatistics

* removed json stats from IRDB and instead expose the whole GeneralStatistics result

* moved printAsJson to GeneralStatistics from IRDB

* updated print statistics usage

* returning a reference to the statistics object

* minor

Co-authored-by: Fabian Schiebel <fabian.schiebel@iem.fraunhofer.de>

Author: janniclas

include/phasar/DB/ProjectIRDB.h

@@ -22,6 +22,7 @@
 #include "llvm/Passes/PassBuilder.h"
 
 #include "nlohmann/json.hpp"
+#include "phasar/PhasarLLVM/Passes/GeneralStatisticsAnalysis.h"
 #include "phasar/Utils/EnumFlags.h"
 
 namespace llvm {
@@ -63,7 +64,7 @@ class ProjectIRDB {
   std::map<std::size_t, llvm::Instruction *> IDInstructionMapping;
   size_t NumGlobals = 0;
   size_t NumberCallsites = 0;
-  nlohmann::json StatsJson;
+  GeneralStatistics GSPResult;
 
   void buildIDModuleMapping(llvm::Module *M);
 
@@ -160,6 +161,10 @@ class ProjectIRDB {
     return AllocatedTypes;
   };
 
+  [[nodiscard]] const psr::GeneralStatistics &getStatistics() const {
+    return GSPResult;
+  };
+
   [[nodiscard]] std::set<const llvm::StructType *>
   getAllocatedStructTypes() const;
 
@@ -188,8 +193,6 @@ class ProjectIRDB {
 
   [[nodiscard]] static std::size_t getInstructionID(const llvm::Instruction *I);
 
-  void printAsJson(llvm::raw_ostream &OS = llvm::outs()) const;
-
   void print() const;
 
   void emitPreprocessedIR(llvm::raw_ostream &OS = llvm::outs(),

include/phasar/PhasarLLVM/Passes/GeneralStatisticsAnalysis.h

@@ -45,6 +45,9 @@ class GeneralStatistics {
   size_t LoadInstructions = 0;
   size_t MemIntrinsics = 0;
   size_t GlobalPointers = 0;
+  size_t Branches = 0;
+  size_t GetElementPtrs = 0;
+  size_t PhiNodes = 0;
   std::set<const llvm::Type *> AllocatedTypes;
   std::set<const llvm::Instruction *> AllocaInstructions;
   std::set<const llvm::Instruction *> RetResInstructions;
@@ -118,6 +121,10 @@ class GeneralStatistics {
   [[nodiscard]] std::set<const llvm::Instruction *>
   getRetResInstructions() const;
   [[nodiscard]] nlohmann::json getAsJson() const;
+  void printAsJson(llvm::raw_ostream &OS = llvm::outs()) const;
+
+  friend llvm::raw_ostream &operator<<(llvm::raw_ostream &OS,
+                                       const GeneralStatistics &Statistics);
 };
 
 /**

lib/Controller/AnalysisController.cpp

@@ -230,15 +230,7 @@ void AnalysisController::emitRequestedHelperAnalysisResults() {
       PT.printAsJson();
     }
   }
-  // if (EmitterOptions & AnalysisControllerEmitterOptions::EmitCGAsText) {
-  //   if (!ResultDirectory.empty()) {
-  //     if (auto OFS = openFileStream("/psr-cg.txt")) {
-  //       ICF.print(*OFS);
-  //     }
-  //   } else {
-  //     ICF.print();
-  //   }
-  // }
+
   if (EmitterOptions & AnalysisControllerEmitterOptions::EmitCGAsDot) {
     if (!ResultDirectory.empty()) {
       if (auto OFS = openFileStream("/psr-cg.dot")) {
@@ -249,23 +241,14 @@ void AnalysisController::emitRequestedHelperAnalysisResults() {
     }
   }
 
-  // if (EmitterOptions & AnalysisControllerEmitterOptions::EmitCGAsJson) {
-  //   if (!ResultDirectory.empty()) {
-  //     if (auto OFS = openFileStream("/psr-cg.json")) {
-  //       ICF.printAsJson(*OFS);
-  //     }
-  //   } else {
-  //     ICF.printAsJson();
-  //   }
-  // }
-
   if (EmitterOptions & AnalysisControllerEmitterOptions::EmitStatisticsAsJson) {
+    const auto &Stats = IRDB.getStatistics();
     if (!ResultDirectory.empty()) {
       if (auto OFS = openFileStream("/psr-IrStatistics.json")) {
-        IRDB.printAsJson(*OFS);
+        Stats.printAsJson(*OFS);
       }
     } else {
-      IRDB.printAsJson();
+      Stats.printAsJson();
     }
   }
 }

lib/DB/ProjectIRDB.cpp

@@ -128,8 +128,8 @@ void ProjectIRDB::preprocessModule(llvm::Module *M) {
   PHASAR_LOG_LEVEL(INFO, "Preprocess module: " << M->getModuleIdentifier());
   MPM.run(*M, MAM);
   // retrieve data from the GeneralStatisticsAnalysis registered earlier
-  auto GSPResult = MAM.getResult<GeneralStatisticsAnalysis>(*M);
-  StatsJson = GSPResult.getAsJson();
+  GSPResult = MAM.getResult<GeneralStatisticsAnalysis>(*M);
+
   NumberCallsites = GSPResult.getFunctioncalls();
   auto Allocas = GSPResult.getAllocaInstructions();
   AllocaInstructions.insert(Allocas.begin(), Allocas.end());
@@ -279,10 +279,6 @@ void ProjectIRDB::print() const {
   }
 }
 
-void ProjectIRDB::printAsJson(llvm::raw_ostream &OS) const {
-  OS << StatsJson.dump(4) << '\n';
-}
-
 void ProjectIRDB::emitPreprocessedIR(llvm::raw_ostream &OS,
                                      bool ShortenIR) const {
   for (const auto &[File, Module] : Modules) {

lib/PhasarLLVM/Passes/GeneralStatisticsAnalysis.cpp

@@ -58,14 +58,15 @@ GeneralStatisticsAnalysis::run(llvm::Module &M,
           // do not add allocas from llvm internal functions
           Stats.AllocaInstructions.insert(&I);
           ++Stats.AllocationSites;
-        } // check bitcast instructions for possible types
-        else {
-          for (auto *User : I.users()) {
-            if (const llvm::BitCastInst *Cast =
-                    llvm::dyn_cast<llvm::BitCastInst>(User)) {
-              // types.insert(cast->getDestTy());
-            }
-          }
+        }
+        if (llvm::isa<llvm::PHINode>(I)) {
+          ++Stats.PhiNodes;
+        }
+        if (llvm::isa<llvm::BranchInst>(I)) {
+          ++Stats.Branches;
+        }
+        if (llvm::isa<llvm::GetElementPtrInst>(I)) {
+          ++Stats.GetElementPtrs;
         }
         // check for return or resume instructions
         if (llvm::isa<llvm::ReturnInst>(I) || llvm::isa<llvm::ResumeInst>(I)) {
@@ -125,7 +126,7 @@ GeneralStatisticsAnalysis::run(llvm::Module &M,
     }
   }
   // check for global pointers
-  for (auto &Global : M.globals()) {
+  for (auto const &Global : M.globals()) {
     if (Global.getType()->isPointerTy()) {
       ++Stats.GlobalPointers;
     }
@@ -217,6 +218,10 @@ GeneralStatistics::getRetResInstructions() const {
   return RetResInstructions;
 }
 
+void GeneralStatistics::printAsJson(llvm::raw_ostream &OS) const {
+  OS << getAsJson().dump(4) << '\n';
+}
+
 nlohmann::json GeneralStatistics::getAsJson() const {
   nlohmann::json J;
   J["ModuleName"] = GeneralStatistics::ModuleName;
@@ -225,7 +230,28 @@ nlohmann::json GeneralStatistics::getAsJson() const {
   J["AllocaInstructions"] = AllocaInstructions.size();
   J["CallSites"] = CallSites;
   J["GlobalVariables"] = Globals;
+  J["Branches"] = Branches;
+  J["GetElementPtrs"] = GetElementPtrs;
+  J["BasicBlocks"] = BasicBlocks;
+  J["PhiNodes"] = PhiNodes;
   return J;
 }
 
+llvm::raw_ostream &operator<<(llvm::raw_ostream &OS,
+                              const GeneralStatistics &Statistics) {
+  return OS << "General LLVM IR Statistics"
+            << "\n"
+            << "Module " << Statistics.ModuleName << ":\n"
+            << "LLVM IR instructions:\t" << Statistics.Instructions << "\n"
+            << "Functions:\t" << Statistics.Functions << "\n"
+            << "Global Variables:\t" << Statistics.Globals << "\n"
+            << "Alloca Instructions:\t" << Statistics.AllocaInstructions.size()
+            << "\n"
+            << "Call Sites:\t" << Statistics.CallSites << "\n"
+            << "Branches:\t" << Statistics.Branches << "\n"
+            << "GetElementPtrs:\t" << Statistics.GetElementPtrs << "\n"
+            << "Phi Nodes:\t" << Statistics.PhiNodes << "\n"
+            << "Basic Blocks:\t" << Statistics.BasicBlocks << "\n";
+}
+
 } // namespace psr

tools/phasar-llvm/phasar-llvm.cpp

@@ -336,17 +336,11 @@ int main(int Argc, const char **Argv) {
   // setup IRDB as source code manager
   ProjectIRDB IRDB(std::vector(ModuleOpt.begin(), ModuleOpt.end()));
   if (StatisticsOpt) {
-    llvm::outs() << "Module " << IRDB.getWPAModule()->getName() << ":\n";
-    llvm::outs() << "> LLVM IR instructions:\t" << IRDB.getNumInstructions()
-                 << "\n";
-    llvm::outs() << "> Functions:\t\t" << IRDB.getWPAModule()->size() << "\n";
-    llvm::outs() << "> Global variables:\t"
-                 << IRDB.getWPAModule()->global_size() << "\n";
-    llvm::outs() << "> Alloca instructions:\t"
-                 << IRDB.getAllocaInstructions().size() << "\n";
+    llvm::outs() << IRDB.getStatistics();
+    // the way we construct memory locations in IRDB is not included in
+    // the GeneralStatistics class right now, thus we print it here separately.
     llvm::outs() << "> Memory Locations:\t"
                  << IRDB.getAllMemoryLocations().size() << "\n";
-    llvm::outs() << "> Call Sites:\t\t" << IRDB.getNumCallsites() << "\n";
   }
 
   // setup the emitter options to display the computed analysis results