Merge pull request #292 from correctcomputation/RootCauseImprovements

Root cause of wildness improvements

Author: jackastner

clang-tools-extra/clangd/CConvertDiagnostics.cpp

@@ -44,9 +44,9 @@ bool CConvertDiagnostics::PopulateDiagsFromConstraintsInfo(ConstraintsInfo &Line
     return nRange;
   };
 
-  for (auto &WReason : Line.RealWildPtrsWithReasons) {
-    if (Line.PtrSourceMap.find(WReason.first) != Line.PtrSourceMap.end()) {
-      auto *PsInfo = Line.PtrSourceMap[WReason.first];
+  for (auto &WReason : Line.RootWildAtomsWithReason) {
+    if (Line.AtomSourceMap.find(WReason.first) != Line.AtomSourceMap.end()) {
+      auto *PsInfo = Line.AtomSourceMap[WReason.first];
       std::string FilePath = PsInfo->getFileName();
       // If this is not a file in a project? Then ignore.
       if (!IsValidSourceFile(Line, FilePath))
@@ -61,15 +61,15 @@ bool CConvertDiagnostics::PopulateDiagsFromConstraintsInfo(ConstraintsInfo &Line
       NewDiag.Severity = DiagnosticsEngine::Level::Error;
       NewDiag.code = std::to_string(WReason.first);
       NewDiag.Message = "Pointer is wild because of:" +
-                        WReason.second.WildPtrReason;
+                        WReason.second.getWildPtrReason();
 
       // Create notes for the information about root cause.
-      if (WReason.second.IsValid) {
+      PersistentSourceLoc SL = WReason.second.getLocation();
+      if (SL.valid()) {
         Note DiagNote;
-        DiagNote.AbsFile = WReason.second.SourceFileName;
+        DiagNote.AbsFile = SL.getFileName();
         DiagNote.Range =
-            GetLocRange(WReason.second.LineNo, WReason.second.ColStartS,
-                        WReason.second.ColStartE);
+            GetLocRange(SL.getLineNo(), SL.getColSNo(), SL.getColENo());
         DiagNote.Message = "Go here to know the root cause for this.";
         NewDiag.Notes.push_back(DiagNote);
       }
@@ -78,11 +78,11 @@ bool CConvertDiagnostics::PopulateDiagsFromConstraintsInfo(ConstraintsInfo &Line
   }
 
   // For non-direct wild pointers..update the reason and diag information.
-  for (auto NonWildCk : Line.TotalNonDirectWildPointers) {
+  for (auto NonWildCk : Line.TotalNonDirectWildAtoms) {
     if (ProcessedCKeys.find(NonWildCk) == ProcessedCKeys.end()) {
       ProcessedCKeys.insert(NonWildCk);
-      if (Line.PtrSourceMap.find(NonWildCk) != Line.PtrSourceMap.end()) {
-        auto *PsInfo = Line.PtrSourceMap[NonWildCk];
+      if (Line.AtomSourceMap.find(NonWildCk) != Line.AtomSourceMap.end()) {
+        auto *PsInfo = Line.AtomSourceMap[NonWildCk];
         std::string FilePath = PsInfo->getFileName();
         // If this is not a file in a project? Then ignore.
         if (!IsValidSourceFile(Line, FilePath))
@@ -106,15 +106,15 @@ bool CConvertDiagnostics::PopulateDiagsFromConstraintsInfo(ConstraintsInfo &Line
         for (auto tC : DirectWildPtrs) {
           Note DiagNote;
 
-          if (Line.PtrSourceMap.find(tC) != Line.PtrSourceMap.end()) {
-            PsInfo = Line.PtrSourceMap[tC];
+          if (Line.AtomSourceMap.find(tC) != Line.AtomSourceMap.end()) {
+            PsInfo = Line.AtomSourceMap[tC];
             FilePath = PsInfo->getFileName();
             DiagNote.AbsFile = FilePath;
             DiagNote.Range =
                 GetLocRange(PsInfo->getLineNo(), PsInfo->getColSNo(),
                             PsInfo->getColENo());
             MaxPtrReasons--;
-            DiagNote.Message = Line.RealWildPtrsWithReasons[tC].WildPtrReason;
+            DiagNote.Message = Line.RootWildAtomsWithReason[tC].getWildPtrReason();
             if (MaxPtrReasons <= 1)
               DiagNote.Message += " (others)";
             NewDiag.Notes.push_back(DiagNote);

clang-tools-extra/clangd/ClangdServer.cpp

@@ -211,7 +211,7 @@ void ClangdServer::executeCConvCommand(ExecuteCommandParams Params,
   auto Task = [this, Params, ConvCB]() {
     std::string RplMsg;
     auto &WildPtrsInfo = CConvInter.GetWILDPtrsInfo();
-    auto &PtrSourceMap = WildPtrsInfo.PtrSourceMap;
+    auto &PtrSourceMap = WildPtrsInfo.AtomSourceMap;
     if (PtrSourceMap.find(Params.ccConvertManualFix->ptrID) !=
         PtrSourceMap.end()) {
       std::string PtrFileName =

clang-tools-extra/clangd/tool/ClangdMain.cpp

@@ -331,12 +331,20 @@ static llvm::cl::opt<std::string>
                        llvm::cl::init("TotalConstraintStats.json"),
                        llvm::cl::cat(ConvertCategory));
 
-static llvm::cl::opt<std::string>
+static cl::opt<std::string>
     OptWildPtrInfoJson("wildptrstats-output",
-                       llvm::cl::desc("Path to the file where all the info "
-                                "related to WILD ptr will be dumped as json"),
-                       llvm::cl::init("WildPtrStats.json"),
-                       llvm::cl::cat(ConvertCategory));
+                            cl::desc("Path to the file where all the info "
+                                     "related to WILD ptr grouped by reason"
+                                     " will be dumped as json"),
+                            cl::init("WildPtrStats.json"),
+                            cl::cat(ConvertCategory));
+
+static cl::opt<std::string>
+  OptPerPtrWILDInfoJson("perptrstats-output",
+                        cl::desc("Path to the file where all the info "
+                                 "related to each WILD ptr will be dumped as json"),
+                        cl::init("PerWildPtrStats.json"),
+                        cl::cat(ConvertCategory));
 
 static llvm::cl::opt<bool> 
     OptDiableCCTypeChecker("disccty",
@@ -465,6 +473,7 @@ int main(int argc, char *argv[]) {
   CcOptions.DumpIntermediate = DumpIntermediate;
   CcOptions.ConstraintOutputJson = ConstraintOutputJson.getValue();
   CcOptions.WildPtrInfoJson = OptWildPtrInfoJson.getValue();
+  CcOptions.PerPtrInfoJson = OptPerPtrWILDInfoJson.getValue();
   CcOptions.StatsOutputJson = OptStatsOutputJson.getValue();
   CcOptions.AddCheckedRegions = AddCheckedRegions;
   CcOptions.EnableAllTypes = AllTypes;

clang/include/clang/CConv/CConv.h

@@ -38,6 +38,8 @@ struct CConvertOptions {
 
   std::string WildPtrInfoJson;
 
+  std::string PerPtrInfoJson;
+
   std::vector<std::string> AllocatorFunctions;
 
   bool HandleVARARGS;

clang/include/clang/CConv/CConvInteractiveData.h

@@ -17,13 +17,17 @@
 #include "PersistentSourceLoc.h"
 
 // Source info and reason for each wild pointer.
-struct WildPointerInferenceInfo {
-  std::string SourceFileName = "";
+class WildPointerInferenceInfo {
+public:
+  WildPointerInferenceInfo(std::string Reason, const PersistentSourceLoc PSL) :
+      WildPtrReason(Reason), Location(PSL) {}
+
+  const std::string &getWildPtrReason() const { return WildPtrReason; }
+  const PersistentSourceLoc &getLocation() const { return Location; }
+
+private:
   std::string WildPtrReason = "";
-  bool IsValid = false;
-  unsigned LineNo = 0;
-  unsigned ColStartS = 0;
-  unsigned ColStartE = 0;
+  PersistentSourceLoc Location;
 };
 
 // Constraints information.
@@ -37,16 +41,17 @@ class ConstraintsInfo {
   CVars &GetRCVars(ConstraintKey);
   CVars &GetSrcCVars(ConstraintKey);
   CVars getWildAffectedCKeys(const std::set<ConstraintKey> &DWKeys);
-  void print_stats(llvm::raw_ostream &O);
+  void printStats(llvm::raw_ostream &O);
+  void printRootCauseStats(raw_ostream &O, Constraints &CS);
+  int getNumPtrsAffected(ConstraintKey CK);
 
-  std::map<ConstraintKey, struct WildPointerInferenceInfo>
-      RealWildPtrsWithReasons;
-  CVars AllWildPtrs;
-  CVars InSrcWildPtrs;
-  CVars TotalNonDirectWildPointers;
-  CVars InSrcNonDirectWildPointers;
+  std::map<ConstraintKey, WildPointerInferenceInfo> RootWildAtomsWithReason;
+  CVars AllWildAtoms;
+  CVars InSrcWildAtoms;
+  CVars TotalNonDirectWildAtoms;
+  CVars InSrcNonDirectWildAtoms;
   std::set<std::string> ValidSourceFiles;
-  std::map<ConstraintKey, PersistentSourceLoc *> PtrSourceMap;
+  std::map<ConstraintKey, const PersistentSourceLoc *> AtomSourceMap;
 
 private:
   // Root cause map: This is the map of a Constraint var and a set of
@@ -67,6 +72,20 @@ class ConstraintsInfo {
   // the above constraint.
   // For the above case, this contains: p -> {s}, q -> {r, s}
   std::map<ConstraintKey, CVars> SrcWMap;
+
+  std::map<ConstraintVariable *, CVars> PtrRCMap;
+  std::map<ConstraintKey, std::set<ConstraintVariable *>> PtrSrcWMap;
+
+  // Get score for each of the ConstraintKeys, which are wild.
+  // For the above example, the score of s would be 0.5, similarly
+  // the score of r would be 1
+  float getAtomAffectedScore(const CVars &AllKeys);
+
+  float getPtrAffectedScore(const std::set<ConstraintVariable *> CVs);
+
+  void
+  printConstraintStats(raw_ostream &O, Constraints &CS, ConstraintKey Cause);
+
 };
 
 #endif // _CCONVINTERACTIVEDATA_H

clang/include/clang/CConv/ConstraintResolver.h

@@ -60,12 +60,11 @@ class ConstraintResolver {
 
   CVarSet handleDeref(CVarSet T);
 
-  CVarSet getInvalidCastPVCons(Expr *E);
+  CVarSet getInvalidCastPVCons(CastExpr *E);
 
   // Update a PVConstraint with one additional level of indirection
   PVConstraint *addAtom(PVConstraint *PVC, ConstAtom *NewA, Constraints &CS);
   CVarSet addAtomAll(CVarSet CVS, ConstAtom *PtrTyp, Constraints &CS);
-  CVarSet getWildPVConstraint();
   CVarSet PVConstraintFromType(QualType TypE);
 
   CVarSet getAllSubExprConstraintVars(std::vector<Expr *> &Exprs);

clang/include/clang/CConv/ConstraintVariables.h

@@ -204,7 +204,9 @@ class PointerVariableConstraint : public ConstraintVariable {
       RestrictQualification
   };
 
-  static PointerVariableConstraint *getWildPVConstraint(Constraints &CS);
+  static PointerVariableConstraint *
+  getWildPVConstraint(Constraints &CS, const std::string &Rsn,
+                      PersistentSourceLoc *PSL = nullptr);
   static PointerVariableConstraint *getPtrPVConstraint(Constraints &CS);
   static PointerVariableConstraint *getNonPtrPVConstraint(Constraints &CS);
   static PointerVariableConstraint *getNamedNonPtrPVConstraint(StringRef name, Constraints &CS);

clang/include/clang/CConv/Constraints.h

@@ -321,12 +321,11 @@ class Constraint {
   };
 private:
   const ConstraintKind Kind;
+  PersistentSourceLoc PL;
+
 public:
   std::string REASON = DEFAULT_REASON;
-  std::string FileName = "";
-  unsigned LineNo = 0;
-  unsigned ColStart = 0;
-  unsigned ColEnd = 0;
+
   Constraint(ConstraintKind K) : Kind(K) { }
   Constraint(ConstraintKind K, const std::string &rsn) : Kind(K) {
     REASON = rsn;
@@ -349,6 +348,8 @@ class Constraint {
   virtual void setReason(const std::string &Rsn) {
     REASON = Rsn;
   }
+
+  const PersistentSourceLoc &getLocation() const { return PL; }
 };
 
 // a >= b
@@ -579,7 +580,7 @@ class Constraints {
 
   void editConstraintHook(Constraint *C);
 
-  std::pair<Constraints::ConstraintSet, bool> solve();
+  void solve();
   void dump() const;
   void print(llvm::raw_ostream &) const;
   void dump_json(llvm::raw_ostream &) const;
@@ -591,6 +592,10 @@ class Constraints {
                  PersistentSourceLoc *PL, bool isCheckedConstraint = true);
   Implies *createImplies(Geq *Premise, Geq *Conclusion);
 
+  VarAtom *createFreshGEQ(std::string Name, VarAtom::VarKind VK, ConstAtom *Con,
+                          std::string Rsn = DEFAULT_REASON,
+                          PersistentSourceLoc *PSL = nullptr);
+
   VarAtom *getFreshVar(std::string Name, VarAtom::VarKind VK);
   VarAtom *getOrCreateVar(ConstraintKey V, std::string Name,
                           VarAtom::VarKind VK);
@@ -630,7 +635,7 @@ class Constraints {
   VarSolTy getDefaultSolution();
 
   // Solve constraint set via graph-based dynamic transitive closure
-  bool graph_based_solve(ConstraintSet &Conflicts);
+  bool graph_based_solve();
 
   // These atoms can be singletons, so we'll store them in the
   // Constraints class.

clang/include/clang/CConv/PersistentSourceLoc.h

@@ -38,7 +38,7 @@ class PersistentSourceLoc {
   uint32_t getLineNo() const { return LineNo; }
   uint32_t getColSNo() const { return ColNoS; }
   uint32_t getColENo() const { return ColNoE; }
-  bool valid() { return isValid; }
+  bool valid() const { return isValid; }
 
   bool operator<(const PersistentSourceLoc &o) const {
     if (FileName == o.FileName)

clang/include/clang/CConv/ProgramInfo.h

@@ -114,7 +114,8 @@ class ProgramInfo : public ProgramVariableAdder {
   const CallTypeParamBindingsT &getTypeParamBindings(CallExpr *CE,
                                                      ASTContext *C) const;
 
-  void constrainWildIfMacro(ConstraintVariable *CV, SourceLocation Location);
+  void constrainWildIfMacro(ConstraintVariable *CV, SourceLocation Location,
+                            PersistentSourceLoc *PSL = nullptr);
 
 private:
   // List of constraint variables for declarations, indexed by their location in
@@ -184,6 +185,14 @@ class ProgramInfo : public ProgramVariableAdder {
   // For each pointer type in the declaration of D, add a variable to the
   // constraint system for that pointer type.
   void addVariable(clang::DeclaratorDecl *D, clang::ASTContext *AstContext);
+
+  void insertIntoPtrSourceMap(const PersistentSourceLoc *PSL,
+                              ConstraintVariable *CV);
+
+  void computePtrLevelStats();
+
+  void insertCVAtoms(ConstraintVariable *CV,
+                     std::map<ConstraintKey, ConstraintVariable *> &AtomMap);
 };
 
 #endif

clang/include/clang/CConv/RewriteUtils.h

@@ -226,7 +226,7 @@ class RewriteConsumer : public ASTConsumer {
   // A single header file can be included in multiple translations units. This
   // set ensures that the diagnostics for a header file are not emitted each
   // time a translation unit containing the header is vistied.
-  static std::set<PersistentSourceLoc *> EmittedDiagnostics;
+  static std::set<PersistentSourceLoc> EmittedDiagnostics;
 
   void emitRootCauseDiagnostics(ASTContext &Context);
 };