BugReporter.cpp revision 0cf3d471546251b12bdceff360f66c079c40526c 
1// BugReporter.cpp - Generate PathDiagnostics for Bugs ------------*- C++ -*--//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10//  This file defines BugReporter, a utility class for generating
11//  PathDiagnostics.
12//
13//===----------------------------------------------------------------------===//
14
15#include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h"
16#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
17#include "clang/StaticAnalyzer/Core/PathSensitive/ExprEngine.h"
18#include "clang/AST/ASTContext.h"
19#include "clang/Analysis/CFG.h"
20#include "clang/AST/DeclObjC.h"
21#include "clang/AST/Expr.h"
22#include "clang/AST/ParentMap.h"
23#include "clang/AST/StmtObjC.h"
24#include "clang/Basic/Diagnostic.h"
25#include "clang/Basic/SourceManager.h"
26#include "clang/Analysis/ProgramPoint.h"
27#include "clang/StaticAnalyzer/Core/BugReporter/PathDiagnostic.h"
28#include "llvm/Support/raw_ostream.h"
29#include "llvm/ADT/DenseMap.h"
30#include "llvm/ADT/SmallString.h"
31#include "llvm/ADT/STLExtras.h"
32#include "llvm/ADT/OwningPtr.h"
33#include <queue>
34
35using namespace clang;
36using namespace ento;
37
38BugReporterVisitor::~BugReporterVisitor() {}
39
40void BugReporterContext::anchor() {}
41
42//===----------------------------------------------------------------------===//
43// Helper routines for walking the ExplodedGraph and fetching statements.
44//===----------------------------------------------------------------------===//
45
46static inline const Stmt *GetStmt(const ProgramPoint &P) {
47  if (const StmtPoint* SP = dyn_cast<StmtPoint>(&P))
48    return SP->getStmt();
49  else if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P))
50    return BE->getSrc()->getTerminator();
51
52  return 0;
53}
54
55static inline const ExplodedNode*
56GetPredecessorNode(const ExplodedNode *N) {
57  return N->pred_empty() ? NULL : *(N->pred_begin());
58}
59
60static inline const ExplodedNode*
61GetSuccessorNode(const ExplodedNode *N) {
62  return N->succ_empty() ? NULL : *(N->succ_begin());
63}
64
65static const Stmt *GetPreviousStmt(const ExplodedNode *N) {
66  for (N = GetPredecessorNode(N); N; N = GetPredecessorNode(N))
67    if (const Stmt *S = GetStmt(N->getLocation()))
68      return S;
69
70  return 0;
71}
72
73static const Stmt *GetNextStmt(const ExplodedNode *N) {
74  for (N = GetSuccessorNode(N); N; N = GetSuccessorNode(N))
75    if (const Stmt *S = GetStmt(N->getLocation())) {
76      // Check if the statement is '?' or '&&'/'||'.  These are "merges",
77      // not actual statement points.
78      switch (S->getStmtClass()) {
79        case Stmt::ChooseExprClass:
80        case Stmt::BinaryConditionalOperatorClass: continue;
81        case Stmt::ConditionalOperatorClass: continue;
82        case Stmt::BinaryOperatorClass: {
83          BinaryOperatorKind Op = cast<BinaryOperator>(S)->getOpcode();
84          if (Op == BO_LAnd || Op == BO_LOr)
85            continue;
86          break;
87        }
88        default:
89          break;
90      }
91      return S;
92    }
93
94  return 0;
95}
96
97static inline const Stmt*
98GetCurrentOrPreviousStmt(const ExplodedNode *N) {
99  if (const Stmt *S = GetStmt(N->getLocation()))
100    return S;
101
102  return GetPreviousStmt(N);
103}
104
105static inline const Stmt*
106GetCurrentOrNextStmt(const ExplodedNode *N) {
107  if (const Stmt *S = GetStmt(N->getLocation()))
108    return S;
109
110  return GetNextStmt(N);
111}
112
113//===----------------------------------------------------------------------===//
114// PathDiagnosticBuilder and its associated routines and helper objects.
115//===----------------------------------------------------------------------===//
116
117typedef llvm::DenseMap<const ExplodedNode*,
118const ExplodedNode*> NodeBackMap;
119
120namespace {
121class NodeMapClosure : public BugReport::NodeResolver {
122  NodeBackMap& M;
123public:
124  NodeMapClosure(NodeBackMap *m) : M(*m) {}
125  ~NodeMapClosure() {}
126
127  const ExplodedNode *getOriginalNode(const ExplodedNode *N) {
128    NodeBackMap::iterator I = M.find(N);
129    return I == M.end() ? 0 : I->second;
130  }
131};
132
133class PathDiagnosticBuilder : public BugReporterContext {
134  BugReport *R;
135  PathDiagnosticConsumer *PDC;
136  OwningPtr<ParentMap> PM;
137  NodeMapClosure NMC;
138public:
139  PathDiagnosticBuilder(GRBugReporter &br,
140                        BugReport *r, NodeBackMap *Backmap,
141                        PathDiagnosticConsumer *pdc)
142    : BugReporterContext(br),
143      R(r), PDC(pdc), NMC(Backmap) {}
144
145  PathDiagnosticLocation ExecutionContinues(const ExplodedNode *N);
146
147  PathDiagnosticLocation ExecutionContinues(llvm::raw_string_ostream &os,
148                                            const ExplodedNode *N);
149
150  BugReport *getBugReport() { return R; }
151
152  Decl const &getCodeDecl() { return R->getErrorNode()->getCodeDecl(); }
153
154  const LocationContext* getLocationContext() {
155    return R->getErrorNode()->getLocationContext();
156  }
157
158  ParentMap& getParentMap() { return R->getErrorNode()->getParentMap(); }
159
160  const Stmt *getParent(const Stmt *S) {
161    return getParentMap().getParent(S);
162  }
163
164  virtual NodeMapClosure& getNodeResolver() { return NMC; }
165
166  PathDiagnosticLocation getEnclosingStmtLocation(const Stmt *S);
167
168  PathDiagnosticConsumer::PathGenerationScheme getGenerationScheme() const {
169    return PDC ? PDC->getGenerationScheme() : PathDiagnosticConsumer::Extensive;
170  }
171
172  bool supportsLogicalOpControlFlow() const {
173    return PDC ? PDC->supportsLogicalOpControlFlow() : true;
174  }
175};
176} // end anonymous namespace
177
178PathDiagnosticLocation
179PathDiagnosticBuilder::ExecutionContinues(const ExplodedNode *N) {
180  if (const Stmt *S = GetNextStmt(N))
181    return PathDiagnosticLocation(S, getSourceManager(), getLocationContext());
182
183  return PathDiagnosticLocation::createDeclEnd(N->getLocationContext(),
184                                               getSourceManager());
185}
186
187PathDiagnosticLocation
188PathDiagnosticBuilder::ExecutionContinues(llvm::raw_string_ostream &os,
189                                          const ExplodedNode *N) {
190
191  // Slow, but probably doesn't matter.
192  if (os.str().empty())
193    os << ' ';
194
195  const PathDiagnosticLocation &Loc = ExecutionContinues(N);
196
197  if (Loc.asStmt())
198    os << "Execution continues on line "
199       << getSourceManager().getExpansionLineNumber(Loc.asLocation())
200       << '.';
201  else {
202    os << "Execution jumps to the end of the ";
203    const Decl *D = N->getLocationContext()->getDecl();
204    if (isa<ObjCMethodDecl>(D))
205      os << "method";
206    else if (isa<FunctionDecl>(D))
207      os << "function";
208    else {
209      assert(isa<BlockDecl>(D));
210      os << "anonymous block";
211    }
212    os << '.';
213  }
214
215  return Loc;
216}
217
218static bool IsNested(const Stmt *S, ParentMap &PM) {
219  if (isa<Expr>(S) && PM.isConsumedExpr(cast<Expr>(S)))
220    return true;
221
222  const Stmt *Parent = PM.getParentIgnoreParens(S);
223
224  if (Parent)
225    switch (Parent->getStmtClass()) {
226      case Stmt::ForStmtClass:
227      case Stmt::DoStmtClass:
228      case Stmt::WhileStmtClass:
229        return true;
230      default:
231        break;
232    }
233
234  return false;
235}
236
237PathDiagnosticLocation
238PathDiagnosticBuilder::getEnclosingStmtLocation(const Stmt *S) {
239  assert(S && "Null Stmt *passed to getEnclosingStmtLocation");
240  ParentMap &P = getParentMap();
241  SourceManager &SMgr = getSourceManager();
242  const LocationContext *LC = getLocationContext();
243
244  while (IsNested(S, P)) {
245    const Stmt *Parent = P.getParentIgnoreParens(S);
246
247    if (!Parent)
248      break;
249
250    switch (Parent->getStmtClass()) {
251      case Stmt::BinaryOperatorClass: {
252        const BinaryOperator *B = cast<BinaryOperator>(Parent);
253        if (B->isLogicalOp())
254          return PathDiagnosticLocation(S, SMgr, LC);
255        break;
256      }
257      case Stmt::CompoundStmtClass:
258      case Stmt::StmtExprClass:
259        return PathDiagnosticLocation(S, SMgr, LC);
260      case Stmt::ChooseExprClass:
261        // Similar to '?' if we are referring to condition, just have the edge
262        // point to the entire choose expression.
263        if (cast<ChooseExpr>(Parent)->getCond() == S)
264          return PathDiagnosticLocation(Parent, SMgr, LC);
265        else
266          return PathDiagnosticLocation(S, SMgr, LC);
267      case Stmt::BinaryConditionalOperatorClass:
268      case Stmt::ConditionalOperatorClass:
269        // For '?', if we are referring to condition, just have the edge point
270        // to the entire '?' expression.
271        if (cast<AbstractConditionalOperator>(Parent)->getCond() == S)
272          return PathDiagnosticLocation(Parent, SMgr, LC);
273        else
274          return PathDiagnosticLocation(S, SMgr, LC);
275      case Stmt::DoStmtClass:
276          return PathDiagnosticLocation(S, SMgr, LC);
277      case Stmt::ForStmtClass:
278        if (cast<ForStmt>(Parent)->getBody() == S)
279          return PathDiagnosticLocation(S, SMgr, LC);
280        break;
281      case Stmt::IfStmtClass:
282        if (cast<IfStmt>(Parent)->getCond() != S)
283          return PathDiagnosticLocation(S, SMgr, LC);
284        break;
285      case Stmt::ObjCForCollectionStmtClass:
286        if (cast<ObjCForCollectionStmt>(Parent)->getBody() == S)
287          return PathDiagnosticLocation(S, SMgr, LC);
288        break;
289      case Stmt::WhileStmtClass:
290        if (cast<WhileStmt>(Parent)->getCond() != S)
291          return PathDiagnosticLocation(S, SMgr, LC);
292        break;
293      default:
294        break;
295    }
296
297    S = Parent;
298  }
299
300  assert(S && "Cannot have null Stmt for PathDiagnosticLocation");
301
302  // Special case: DeclStmts can appear in for statement declarations, in which
303  //  case the ForStmt is the context.
304  if (isa<DeclStmt>(S)) {
305    if (const Stmt *Parent = P.getParent(S)) {
306      switch (Parent->getStmtClass()) {
307        case Stmt::ForStmtClass:
308        case Stmt::ObjCForCollectionStmtClass:
309          return PathDiagnosticLocation(Parent, SMgr, LC);
310        default:
311          break;
312      }
313    }
314  }
315  else if (isa<BinaryOperator>(S)) {
316    // Special case: the binary operator represents the initialization
317    // code in a for statement (this can happen when the variable being
318    // initialized is an old variable.
319    if (const ForStmt *FS =
320          dyn_cast_or_null<ForStmt>(P.getParentIgnoreParens(S))) {
321      if (FS->getInit() == S)
322        return PathDiagnosticLocation(FS, SMgr, LC);
323    }
324  }
325
326  return PathDiagnosticLocation(S, SMgr, LC);
327}
328
329//===----------------------------------------------------------------------===//
330// ScanNotableSymbols: closure-like callback for scanning Store bindings.
331//===----------------------------------------------------------------------===//
332
333static const VarDecl* GetMostRecentVarDeclBinding(const ExplodedNode *N,
334                                                  ProgramStateManager& VMgr,
335                                                  SVal X) {
336
337  for ( ; N ; N = N->pred_empty() ? 0 : *N->pred_begin()) {
338
339    ProgramPoint P = N->getLocation();
340
341    if (!isa<PostStmt>(P))
342      continue;
343
344    const DeclRefExpr *DR = dyn_cast<DeclRefExpr>(cast<PostStmt>(P).getStmt());
345
346    if (!DR)
347      continue;
348
349    SVal Y = N->getState()->getSVal(DR, N->getLocationContext());
350
351    if (X != Y)
352      continue;
353
354    const VarDecl *VD = dyn_cast<VarDecl>(DR->getDecl());
355
356    if (!VD)
357      continue;
358
359    return VD;
360  }
361
362  return 0;
363}
364
365namespace {
366class NotableSymbolHandler
367: public StoreManager::BindingsHandler {
368
369  SymbolRef Sym;
370  ProgramStateRef PrevSt;
371  const Stmt *S;
372  ProgramStateManager& VMgr;
373  const ExplodedNode *Pred;
374  PathDiagnostic& PD;
375  BugReporter& BR;
376
377public:
378
379  NotableSymbolHandler(SymbolRef sym,
380                       ProgramStateRef prevst,
381                       const Stmt *s,
382                       ProgramStateManager& vmgr,
383                       const ExplodedNode *pred,
384                       PathDiagnostic& pd,
385                       BugReporter& br)
386  : Sym(sym),
387    PrevSt(prevst),
388    S(s),
389    VMgr(vmgr),
390    Pred(pred),
391    PD(pd),
392    BR(br) {}
393
394  bool HandleBinding(StoreManager& SMgr, Store store, const MemRegion* R,
395                     SVal V) {
396
397    SymbolRef ScanSym = V.getAsSymbol();
398
399    if (ScanSym != Sym)
400      return true;
401
402    // Check if the previous state has this binding.
403    SVal X = PrevSt->getSVal(loc::MemRegionVal(R));
404
405    if (X == V) // Same binding?
406      return true;
407
408    // Different binding.  Only handle assignments for now.  We don't pull
409    // this check out of the loop because we will eventually handle other
410    // cases.
411
412    VarDecl *VD = 0;
413
414    if (const BinaryOperator* B = dyn_cast<BinaryOperator>(S)) {
415      if (!B->isAssignmentOp())
416        return true;
417
418      // What variable did we assign to?
419      DeclRefExpr *DR = dyn_cast<DeclRefExpr>(B->getLHS()->IgnoreParenCasts());
420
421      if (!DR)
422        return true;
423
424      VD = dyn_cast<VarDecl>(DR->getDecl());
425    }
426    else if (const DeclStmt *DS = dyn_cast<DeclStmt>(S)) {
427      // FIXME: Eventually CFGs won't have DeclStmts.  Right now we
428      //  assume that each DeclStmt has a single Decl.  This invariant
429      //  holds by construction in the CFG.
430      VD = dyn_cast<VarDecl>(*DS->decl_begin());
431    }
432
433    if (!VD)
434      return true;
435
436    // What is the most recently referenced variable with this binding?
437    const VarDecl *MostRecent = GetMostRecentVarDeclBinding(Pred, VMgr, V);
438
439    if (!MostRecent)
440      return true;
441
442    // Create the diagnostic.
443    if (Loc::isLocType(VD->getType())) {
444      SmallString<64> buf;
445      llvm::raw_svector_ostream os(buf);
446      os << '\'' << *VD << "' now aliases '" << *MostRecent << '\'';
447      PathDiagnosticLocation L =
448        PathDiagnosticLocation::createBegin(S, BR.getSourceManager(),
449                                                   Pred->getLocationContext());
450      PD.push_front(new PathDiagnosticEventPiece(L, os.str()));
451    }
452
453    return true;
454  }
455};
456}
457
458static void HandleNotableSymbol(const ExplodedNode *N,
459                                const Stmt *S,
460                                SymbolRef Sym, BugReporter& BR,
461                                PathDiagnostic& PD) {
462
463  const ExplodedNode *Pred = N->pred_empty() ? 0 : *N->pred_begin();
464  ProgramStateRef PrevSt = Pred ? Pred->getState() : 0;
465
466  if (!PrevSt)
467    return;
468
469  // Look at the region bindings of the current state that map to the
470  // specified symbol.  Are any of them not in the previous state?
471  ProgramStateManager& VMgr = cast<GRBugReporter>(BR).getStateManager();
472  NotableSymbolHandler H(Sym, PrevSt, S, VMgr, Pred, PD, BR);
473  cast<GRBugReporter>(BR).getStateManager().iterBindings(N->getState(), H);
474}
475
476namespace {
477class ScanNotableSymbols
478: public StoreManager::BindingsHandler {
479
480  llvm::SmallSet<SymbolRef, 10> AlreadyProcessed;
481  const ExplodedNode *N;
482  const Stmt *S;
483  GRBugReporter& BR;
484  PathDiagnostic& PD;
485
486public:
487  ScanNotableSymbols(const ExplodedNode *n, const Stmt *s,
488                     GRBugReporter& br, PathDiagnostic& pd)
489  : N(n), S(s), BR(br), PD(pd) {}
490
491  bool HandleBinding(StoreManager& SMgr, Store store,
492                     const MemRegion* R, SVal V) {
493
494    SymbolRef ScanSym = V.getAsSymbol();
495
496    if (!ScanSym)
497      return true;
498
499    if (!BR.isNotable(ScanSym))
500      return true;
501
502    if (AlreadyProcessed.count(ScanSym))
503      return true;
504
505    AlreadyProcessed.insert(ScanSym);
506
507    HandleNotableSymbol(N, S, ScanSym, BR, PD);
508    return true;
509  }
510};
511} // end anonymous namespace
512
513//===----------------------------------------------------------------------===//
514// "Minimal" path diagnostic generation algorithm.
515//===----------------------------------------------------------------------===//
516
517static void CompactPathDiagnostic(PathDiagnostic &PD, const SourceManager& SM);
518
519static void GenerateMinimalPathDiagnostic(PathDiagnostic& PD,
520                                          PathDiagnosticBuilder &PDB,
521                                          const ExplodedNode *N) {
522
523  SourceManager& SMgr = PDB.getSourceManager();
524  const LocationContext *LC = PDB.getLocationContext();
525  const ExplodedNode *NextNode = N->pred_empty()
526                                        ? NULL : *(N->pred_begin());
527  while (NextNode) {
528    N = NextNode;
529    NextNode = GetPredecessorNode(N);
530
531    ProgramPoint P = N->getLocation();
532
533    if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P)) {
534      const CFGBlock *Src = BE->getSrc();
535      const CFGBlock *Dst = BE->getDst();
536      const Stmt *T = Src->getTerminator();
537
538      if (!T)
539        continue;
540
541      PathDiagnosticLocation Start =
542        PathDiagnosticLocation::createBegin(T, SMgr,
543                                                N->getLocationContext());
544
545      switch (T->getStmtClass()) {
546        default:
547          break;
548
549        case Stmt::GotoStmtClass:
550        case Stmt::IndirectGotoStmtClass: {
551          const Stmt *S = GetNextStmt(N);
552
553          if (!S)
554            continue;
555
556          std::string sbuf;
557          llvm::raw_string_ostream os(sbuf);
558          const PathDiagnosticLocation &End = PDB.getEnclosingStmtLocation(S);
559
560          os << "Control jumps to line "
561          << End.asLocation().getExpansionLineNumber();
562          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
563                                                           os.str()));
564          break;
565        }
566
567        case Stmt::SwitchStmtClass: {
568          // Figure out what case arm we took.
569          std::string sbuf;
570          llvm::raw_string_ostream os(sbuf);
571
572          if (const Stmt *S = Dst->getLabel()) {
573            PathDiagnosticLocation End(S, SMgr, LC);
574
575            switch (S->getStmtClass()) {
576              default:
577                os << "No cases match in the switch statement. "
578                "Control jumps to line "
579                << End.asLocation().getExpansionLineNumber();
580                break;
581              case Stmt::DefaultStmtClass:
582                os << "Control jumps to the 'default' case at line "
583                << End.asLocation().getExpansionLineNumber();
584                break;
585
586              case Stmt::CaseStmtClass: {
587                os << "Control jumps to 'case ";
588                const CaseStmt *Case = cast<CaseStmt>(S);
589                const Expr *LHS = Case->getLHS()->IgnoreParenCasts();
590
591                // Determine if it is an enum.
592                bool GetRawInt = true;
593
594                if (const DeclRefExpr *DR = dyn_cast<DeclRefExpr>(LHS)) {
595                  // FIXME: Maybe this should be an assertion.  Are there cases
596                  // were it is not an EnumConstantDecl?
597                  const EnumConstantDecl *D =
598                    dyn_cast<EnumConstantDecl>(DR->getDecl());
599
600                  if (D) {
601                    GetRawInt = false;
602                    os << *D;
603                  }
604                }
605
606                if (GetRawInt)
607                  os << LHS->EvaluateKnownConstInt(PDB.getASTContext());
608
609                os << ":'  at line "
610                << End.asLocation().getExpansionLineNumber();
611                break;
612              }
613            }
614            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
615                                                             os.str()));
616          }
617          else {
618            os << "'Default' branch taken. ";
619            const PathDiagnosticLocation &End = PDB.ExecutionContinues(os, N);
620            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
621                                                             os.str()));
622          }
623
624          break;
625        }
626
627        case Stmt::BreakStmtClass:
628        case Stmt::ContinueStmtClass: {
629          std::string sbuf;
630          llvm::raw_string_ostream os(sbuf);
631          PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
632          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
633                                                           os.str()));
634          break;
635        }
636
637          // Determine control-flow for ternary '?'.
638        case Stmt::BinaryConditionalOperatorClass:
639        case Stmt::ConditionalOperatorClass: {
640          std::string sbuf;
641          llvm::raw_string_ostream os(sbuf);
642          os << "'?' condition is ";
643
644          if (*(Src->succ_begin()+1) == Dst)
645            os << "false";
646          else
647            os << "true";
648
649          PathDiagnosticLocation End = PDB.ExecutionContinues(N);
650
651          if (const Stmt *S = End.asStmt())
652            End = PDB.getEnclosingStmtLocation(S);
653
654          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
655                                                           os.str()));
656          break;
657        }
658
659          // Determine control-flow for short-circuited '&&' and '||'.
660        case Stmt::BinaryOperatorClass: {
661          if (!PDB.supportsLogicalOpControlFlow())
662            break;
663
664          const BinaryOperator *B = cast<BinaryOperator>(T);
665          std::string sbuf;
666          llvm::raw_string_ostream os(sbuf);
667          os << "Left side of '";
668
669          if (B->getOpcode() == BO_LAnd) {
670            os << "&&" << "' is ";
671
672            if (*(Src->succ_begin()+1) == Dst) {
673              os << "false";
674              PathDiagnosticLocation End(B->getLHS(), SMgr, LC);
675              PathDiagnosticLocation Start =
676                PathDiagnosticLocation::createOperatorLoc(B, SMgr);
677              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
678                                                               os.str()));
679            }
680            else {
681              os << "true";
682              PathDiagnosticLocation Start(B->getLHS(), SMgr, LC);
683              PathDiagnosticLocation End = PDB.ExecutionContinues(N);
684              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
685                                                               os.str()));
686            }
687          }
688          else {
689            assert(B->getOpcode() == BO_LOr);
690            os << "||" << "' is ";
691
692            if (*(Src->succ_begin()+1) == Dst) {
693              os << "false";
694              PathDiagnosticLocation Start(B->getLHS(), SMgr, LC);
695              PathDiagnosticLocation End = PDB.ExecutionContinues(N);
696              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
697                                                               os.str()));
698            }
699            else {
700              os << "true";
701              PathDiagnosticLocation End(B->getLHS(), SMgr, LC);
702              PathDiagnosticLocation Start =
703                PathDiagnosticLocation::createOperatorLoc(B, SMgr);
704              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
705                                                               os.str()));
706            }
707          }
708
709          break;
710        }
711
712        case Stmt::DoStmtClass:  {
713          if (*(Src->succ_begin()) == Dst) {
714            std::string sbuf;
715            llvm::raw_string_ostream os(sbuf);
716
717            os << "Loop condition is true. ";
718            PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
719
720            if (const Stmt *S = End.asStmt())
721              End = PDB.getEnclosingStmtLocation(S);
722
723            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
724                                                             os.str()));
725          }
726          else {
727            PathDiagnosticLocation End = PDB.ExecutionContinues(N);
728
729            if (const Stmt *S = End.asStmt())
730              End = PDB.getEnclosingStmtLocation(S);
731
732            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
733                              "Loop condition is false.  Exiting loop"));
734          }
735
736          break;
737        }
738
739        case Stmt::WhileStmtClass:
740        case Stmt::ForStmtClass: {
741          if (*(Src->succ_begin()+1) == Dst) {
742            std::string sbuf;
743            llvm::raw_string_ostream os(sbuf);
744
745            os << "Loop condition is false. ";
746            PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
747            if (const Stmt *S = End.asStmt())
748              End = PDB.getEnclosingStmtLocation(S);
749
750            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
751                                                             os.str()));
752          }
753          else {
754            PathDiagnosticLocation End = PDB.ExecutionContinues(N);
755            if (const Stmt *S = End.asStmt())
756              End = PDB.getEnclosingStmtLocation(S);
757
758            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
759                            "Loop condition is true.  Entering loop body"));
760          }
761
762          break;
763        }
764
765        case Stmt::IfStmtClass: {
766          PathDiagnosticLocation End = PDB.ExecutionContinues(N);
767
768          if (const Stmt *S = End.asStmt())
769            End = PDB.getEnclosingStmtLocation(S);
770
771          if (*(Src->succ_begin()+1) == Dst)
772            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
773                                                        "Taking false branch"));
774          else
775            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
776                                                         "Taking true branch"));
777
778          break;
779        }
780      }
781    }
782
783    if (NextNode) {
784      // Add diagnostic pieces from custom visitors.
785      BugReport *R = PDB.getBugReport();
786      for (BugReport::visitor_iterator I = R->visitor_begin(),
787           E = R->visitor_end(); I!=E; ++I) {
788        if (PathDiagnosticPiece *p = (*I)->VisitNode(N, NextNode, PDB, *R))
789          PD.push_front(p);
790      }
791    }
792
793    if (const PostStmt *PS = dyn_cast<PostStmt>(&P)) {
794      // Scan the region bindings, and see if a "notable" symbol has a new
795      // lval binding.
796      ScanNotableSymbols SNS(N, PS->getStmt(), PDB.getBugReporter(), PD);
797      PDB.getStateManager().iterBindings(N->getState(), SNS);
798    }
799  }
800
801  // After constructing the full PathDiagnostic, do a pass over it to compact
802  // PathDiagnosticPieces that occur within a macro.
803  CompactPathDiagnostic(PD, PDB.getSourceManager());
804}
805
806//===----------------------------------------------------------------------===//
807// "Extensive" PathDiagnostic generation.
808//===----------------------------------------------------------------------===//
809
810static bool IsControlFlowExpr(const Stmt *S) {
811  const Expr *E = dyn_cast<Expr>(S);
812
813  if (!E)
814    return false;
815
816  E = E->IgnoreParenCasts();
817
818  if (isa<AbstractConditionalOperator>(E))
819    return true;
820
821  if (const BinaryOperator *B = dyn_cast<BinaryOperator>(E))
822    if (B->isLogicalOp())
823      return true;
824
825  return false;
826}
827
828namespace {
829class ContextLocation : public PathDiagnosticLocation {
830  bool IsDead;
831public:
832  ContextLocation(const PathDiagnosticLocation &L, bool isdead = false)
833    : PathDiagnosticLocation(L), IsDead(isdead) {}
834
835  void markDead() { IsDead = true; }
836  bool isDead() const { return IsDead; }
837};
838
839class EdgeBuilder {
840  std::vector<ContextLocation> CLocs;
841  typedef std::vector<ContextLocation>::iterator iterator;
842  PathDiagnostic &PD;
843  PathDiagnosticBuilder &PDB;
844  PathDiagnosticLocation PrevLoc;
845
846  bool IsConsumedExpr(const PathDiagnosticLocation &L);
847
848  bool containsLocation(const PathDiagnosticLocation &Container,
849                        const PathDiagnosticLocation &Containee);
850
851  PathDiagnosticLocation getContextLocation(const PathDiagnosticLocation &L);
852
853  PathDiagnosticLocation cleanUpLocation(PathDiagnosticLocation L,
854                                         bool firstCharOnly = false) {
855    if (const Stmt *S = L.asStmt()) {
856      const Stmt *Original = S;
857      while (1) {
858        // Adjust the location for some expressions that are best referenced
859        // by one of their subexpressions.
860        switch (S->getStmtClass()) {
861          default:
862            break;
863          case Stmt::ParenExprClass:
864          case Stmt::GenericSelectionExprClass:
865            S = cast<Expr>(S)->IgnoreParens();
866            firstCharOnly = true;
867            continue;
868          case Stmt::BinaryConditionalOperatorClass:
869          case Stmt::ConditionalOperatorClass:
870            S = cast<AbstractConditionalOperator>(S)->getCond();
871            firstCharOnly = true;
872            continue;
873          case Stmt::ChooseExprClass:
874            S = cast<ChooseExpr>(S)->getCond();
875            firstCharOnly = true;
876            continue;
877          case Stmt::BinaryOperatorClass:
878            S = cast<BinaryOperator>(S)->getLHS();
879            firstCharOnly = true;
880            continue;
881        }
882
883        break;
884      }
885
886      if (S != Original)
887        L = PathDiagnosticLocation(S, L.getManager(), PDB.getLocationContext());
888    }
889
890    if (firstCharOnly)
891      L  = PathDiagnosticLocation::createSingleLocation(L);
892
893    return L;
894  }
895
896  void popLocation() {
897    if (!CLocs.back().isDead() && CLocs.back().asLocation().isFileID()) {
898      // For contexts, we only one the first character as the range.
899      rawAddEdge(cleanUpLocation(CLocs.back(), true));
900    }
901    CLocs.pop_back();
902  }
903
904public:
905  EdgeBuilder(PathDiagnostic &pd, PathDiagnosticBuilder &pdb)
906    : PD(pd), PDB(pdb) {
907
908      // If the PathDiagnostic already has pieces, add the enclosing statement
909      // of the first piece as a context as well.
910      if (!PD.empty()) {
911        PrevLoc = PD.begin()->getLocation();
912
913        if (const Stmt *S = PrevLoc.asStmt())
914          addExtendedContext(PDB.getEnclosingStmtLocation(S).asStmt());
915      }
916  }
917
918  ~EdgeBuilder() {
919    while (!CLocs.empty()) popLocation();
920
921    // Finally, add an initial edge from the start location of the first
922    // statement (if it doesn't already exist).
923    PathDiagnosticLocation L = PathDiagnosticLocation::createDeclBegin(
924                                                       PDB.getLocationContext(),
925                                                       PDB.getSourceManager());
926    if (L.isValid())
927      rawAddEdge(L);
928  }
929
930  void addEdge(PathDiagnosticLocation NewLoc, bool alwaysAdd = false);
931
932  void rawAddEdge(PathDiagnosticLocation NewLoc);
933
934  void addContext(const Stmt *S);
935  void addExtendedContext(const Stmt *S);
936};
937} // end anonymous namespace
938
939
940PathDiagnosticLocation
941EdgeBuilder::getContextLocation(const PathDiagnosticLocation &L) {
942  if (const Stmt *S = L.asStmt()) {
943    if (IsControlFlowExpr(S))
944      return L;
945
946    return PDB.getEnclosingStmtLocation(S);
947  }
948
949  return L;
950}
951
952bool EdgeBuilder::containsLocation(const PathDiagnosticLocation &Container,
953                                   const PathDiagnosticLocation &Containee) {
954
955  if (Container == Containee)
956    return true;
957
958  if (Container.asDecl())
959    return true;
960
961  if (const Stmt *S = Containee.asStmt())
962    if (const Stmt *ContainerS = Container.asStmt()) {
963      while (S) {
964        if (S == ContainerS)
965          return true;
966        S = PDB.getParent(S);
967      }
968      return false;
969    }
970
971  // Less accurate: compare using source ranges.
972  SourceRange ContainerR = Container.asRange();
973  SourceRange ContaineeR = Containee.asRange();
974
975  SourceManager &SM = PDB.getSourceManager();
976  SourceLocation ContainerRBeg = SM.getExpansionLoc(ContainerR.getBegin());
977  SourceLocation ContainerREnd = SM.getExpansionLoc(ContainerR.getEnd());
978  SourceLocation ContaineeRBeg = SM.getExpansionLoc(ContaineeR.getBegin());
979  SourceLocation ContaineeREnd = SM.getExpansionLoc(ContaineeR.getEnd());
980
981  unsigned ContainerBegLine = SM.getExpansionLineNumber(ContainerRBeg);
982  unsigned ContainerEndLine = SM.getExpansionLineNumber(ContainerREnd);
983  unsigned ContaineeBegLine = SM.getExpansionLineNumber(ContaineeRBeg);
984  unsigned ContaineeEndLine = SM.getExpansionLineNumber(ContaineeREnd);
985
986  assert(ContainerBegLine <= ContainerEndLine);
987  assert(ContaineeBegLine <= ContaineeEndLine);
988
989  return (ContainerBegLine <= ContaineeBegLine &&
990          ContainerEndLine >= ContaineeEndLine &&
991          (ContainerBegLine != ContaineeBegLine ||
992           SM.getExpansionColumnNumber(ContainerRBeg) <=
993           SM.getExpansionColumnNumber(ContaineeRBeg)) &&
994          (ContainerEndLine != ContaineeEndLine ||
995           SM.getExpansionColumnNumber(ContainerREnd) >=
996           SM.getExpansionColumnNumber(ContainerREnd)));
997}
998
999void EdgeBuilder::rawAddEdge(PathDiagnosticLocation NewLoc) {
1000  if (!PrevLoc.isValid()) {
1001    PrevLoc = NewLoc;
1002    return;
1003  }
1004
1005  const PathDiagnosticLocation &NewLocClean = cleanUpLocation(NewLoc);
1006  const PathDiagnosticLocation &PrevLocClean = cleanUpLocation(PrevLoc);
1007
1008  if (NewLocClean.asLocation() == PrevLocClean.asLocation())
1009    return;
1010
1011  // FIXME: Ignore intra-macro edges for now.
1012  if (NewLocClean.asLocation().getExpansionLoc() ==
1013      PrevLocClean.asLocation().getExpansionLoc())
1014    return;
1015
1016  PD.push_front(new PathDiagnosticControlFlowPiece(NewLocClean, PrevLocClean));
1017  PrevLoc = NewLoc;
1018}
1019
1020void EdgeBuilder::addEdge(PathDiagnosticLocation NewLoc, bool alwaysAdd) {
1021
1022  if (!alwaysAdd && NewLoc.asLocation().isMacroID())
1023    return;
1024
1025  const PathDiagnosticLocation &CLoc = getContextLocation(NewLoc);
1026
1027  while (!CLocs.empty()) {
1028    ContextLocation &TopContextLoc = CLocs.back();
1029
1030    // Is the top location context the same as the one for the new location?
1031    if (TopContextLoc == CLoc) {
1032      if (alwaysAdd) {
1033        if (IsConsumedExpr(TopContextLoc) &&
1034            !IsControlFlowExpr(TopContextLoc.asStmt()))
1035            TopContextLoc.markDead();
1036
1037        rawAddEdge(NewLoc);
1038      }
1039
1040      return;
1041    }
1042
1043    if (containsLocation(TopContextLoc, CLoc)) {
1044      if (alwaysAdd) {
1045        rawAddEdge(NewLoc);
1046
1047        if (IsConsumedExpr(CLoc) && !IsControlFlowExpr(CLoc.asStmt())) {
1048          CLocs.push_back(ContextLocation(CLoc, true));
1049          return;
1050        }
1051      }
1052
1053      CLocs.push_back(CLoc);
1054      return;
1055    }
1056
1057    // Context does not contain the location.  Flush it.
1058    popLocation();
1059  }
1060
1061  // If we reach here, there is no enclosing context.  Just add the edge.
1062  rawAddEdge(NewLoc);
1063}
1064
1065bool EdgeBuilder::IsConsumedExpr(const PathDiagnosticLocation &L) {
1066  if (const Expr *X = dyn_cast_or_null<Expr>(L.asStmt()))
1067    return PDB.getParentMap().isConsumedExpr(X) && !IsControlFlowExpr(X);
1068
1069  return false;
1070}
1071
1072void EdgeBuilder::addExtendedContext(const Stmt *S) {
1073  if (!S)
1074    return;
1075
1076  const Stmt *Parent = PDB.getParent(S);
1077  while (Parent) {
1078    if (isa<CompoundStmt>(Parent))
1079      Parent = PDB.getParent(Parent);
1080    else
1081      break;
1082  }
1083
1084  if (Parent) {
1085    switch (Parent->getStmtClass()) {
1086      case Stmt::DoStmtClass:
1087      case Stmt::ObjCAtSynchronizedStmtClass:
1088        addContext(Parent);
1089      default:
1090        break;
1091    }
1092  }
1093
1094  addContext(S);
1095}
1096
1097void EdgeBuilder::addContext(const Stmt *S) {
1098  if (!S)
1099    return;
1100
1101  PathDiagnosticLocation L(S, PDB.getSourceManager(), PDB.getLocationContext());
1102
1103  while (!CLocs.empty()) {
1104    const PathDiagnosticLocation &TopContextLoc = CLocs.back();
1105
1106    // Is the top location context the same as the one for the new location?
1107    if (TopContextLoc == L)
1108      return;
1109
1110    if (containsLocation(TopContextLoc, L)) {
1111      CLocs.push_back(L);
1112      return;
1113    }
1114
1115    // Context does not contain the location.  Flush it.
1116    popLocation();
1117  }
1118
1119  CLocs.push_back(L);
1120}
1121
1122static void GenerateExtensivePathDiagnostic(PathDiagnostic& PD,
1123                                            PathDiagnosticBuilder &PDB,
1124                                            const ExplodedNode *N) {
1125  EdgeBuilder EB(PD, PDB);
1126  const SourceManager& SM = PDB.getSourceManager();
1127
1128  const ExplodedNode *NextNode = N->pred_empty() ? NULL : *(N->pred_begin());
1129  while (NextNode) {
1130    N = NextNode;
1131    NextNode = GetPredecessorNode(N);
1132    ProgramPoint P = N->getLocation();
1133
1134    do {
1135      // Block edges.
1136      if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P)) {
1137        const CFGBlock &Blk = *BE->getSrc();
1138        const Stmt *Term = Blk.getTerminator();
1139
1140        // Are we jumping to the head of a loop?  Add a special diagnostic.
1141        if (const Stmt *Loop = BE->getDst()->getLoopTarget()) {
1142          PathDiagnosticLocation L(Loop, SM, PDB.getLocationContext());
1143          const CompoundStmt *CS = NULL;
1144
1145          if (!Term) {
1146            if (const ForStmt *FS = dyn_cast<ForStmt>(Loop))
1147              CS = dyn_cast<CompoundStmt>(FS->getBody());
1148            else if (const WhileStmt *WS = dyn_cast<WhileStmt>(Loop))
1149              CS = dyn_cast<CompoundStmt>(WS->getBody());
1150          }
1151
1152          PathDiagnosticEventPiece *p =
1153            new PathDiagnosticEventPiece(L,
1154                                        "Looping back to the head of the loop");
1155
1156          EB.addEdge(p->getLocation(), true);
1157          PD.push_front(p);
1158
1159          if (CS) {
1160            PathDiagnosticLocation BL =
1161              PathDiagnosticLocation::createEndBrace(CS, SM);
1162            EB.addEdge(BL);
1163          }
1164        }
1165
1166        if (Term)
1167          EB.addContext(Term);
1168
1169        break;
1170      }
1171
1172      if (const BlockEntrance *BE = dyn_cast<BlockEntrance>(&P)) {
1173        if (const CFGStmt *S = BE->getFirstElement().getAs<CFGStmt>()) {
1174          const Stmt *stmt = S->getStmt();
1175          if (IsControlFlowExpr(stmt)) {
1176            // Add the proper context for '&&', '||', and '?'.
1177            EB.addContext(stmt);
1178          }
1179          else
1180            EB.addExtendedContext(PDB.getEnclosingStmtLocation(stmt).asStmt());
1181        }
1182
1183        break;
1184      }
1185    } while (0);
1186
1187    if (!NextNode)
1188      continue;
1189
1190    // Add pieces from custom visitors.
1191    BugReport *R = PDB.getBugReport();
1192    for (BugReport::visitor_iterator I = R->visitor_begin(),
1193                                     E = R->visitor_end(); I!=E; ++I) {
1194      if (PathDiagnosticPiece *p = (*I)->VisitNode(N, NextNode, PDB, *R)) {
1195        const PathDiagnosticLocation &Loc = p->getLocation();
1196        EB.addEdge(Loc, true);
1197        PD.push_front(p);
1198        if (const Stmt *S = Loc.asStmt())
1199          EB.addExtendedContext(PDB.getEnclosingStmtLocation(S).asStmt());
1200      }
1201    }
1202  }
1203}
1204
1205//===----------------------------------------------------------------------===//
1206// Methods for BugType and subclasses.
1207//===----------------------------------------------------------------------===//
1208BugType::~BugType() { }
1209
1210void BugType::FlushReports(BugReporter &BR) {}
1211
1212void BuiltinBug::anchor() {}
1213
1214//===----------------------------------------------------------------------===//
1215// Methods for BugReport and subclasses.
1216//===----------------------------------------------------------------------===//
1217
1218void BugReport::NodeResolver::anchor() {}
1219
1220void BugReport::addVisitor(BugReporterVisitor* visitor) {
1221  if (!visitor)
1222    return;
1223
1224  llvm::FoldingSetNodeID ID;
1225  visitor->Profile(ID);
1226  void *InsertPos;
1227
1228  if (CallbacksSet.FindNodeOrInsertPos(ID, InsertPos)) {
1229    delete visitor;
1230    return;
1231  }
1232
1233  CallbacksSet.InsertNode(visitor, InsertPos);
1234  Callbacks = F.add(visitor, Callbacks);
1235}
1236
1237BugReport::~BugReport() {
1238  for (visitor_iterator I = visitor_begin(), E = visitor_end(); I != E; ++I) {
1239    delete *I;
1240  }
1241}
1242
1243void BugReport::Profile(llvm::FoldingSetNodeID& hash) const {
1244  hash.AddPointer(&BT);
1245  hash.AddString(Description);
1246  if (Location.isValid()) {
1247    Location.Profile(hash);
1248  } else {
1249    assert(ErrorNode);
1250    hash.AddPointer(GetCurrentOrPreviousStmt(ErrorNode));
1251  }
1252
1253  for (SmallVectorImpl<SourceRange>::const_iterator I =
1254      Ranges.begin(), E = Ranges.end(); I != E; ++I) {
1255    const SourceRange range = *I;
1256    if (!range.isValid())
1257      continue;
1258    hash.AddInteger(range.getBegin().getRawEncoding());
1259    hash.AddInteger(range.getEnd().getRawEncoding());
1260  }
1261}
1262
1263const Stmt *BugReport::getStmt() const {
1264  if (!ErrorNode)
1265    return 0;
1266
1267  ProgramPoint ProgP = ErrorNode->getLocation();
1268  const Stmt *S = NULL;
1269
1270  if (BlockEntrance *BE = dyn_cast<BlockEntrance>(&ProgP)) {
1271    CFGBlock &Exit = ProgP.getLocationContext()->getCFG()->getExit();
1272    if (BE->getBlock() == &Exit)
1273      S = GetPreviousStmt(ErrorNode);
1274  }
1275  if (!S)
1276    S = GetStmt(ProgP);
1277
1278  return S;
1279}
1280
1281std::pair<BugReport::ranges_iterator, BugReport::ranges_iterator>
1282BugReport::getRanges() {
1283    // If no custom ranges, add the range of the statement corresponding to
1284    // the error node.
1285    if (Ranges.empty()) {
1286      if (const Expr *E = dyn_cast_or_null<Expr>(getStmt()))
1287        addRange(E->getSourceRange());
1288      else
1289        return std::make_pair(ranges_iterator(), ranges_iterator());
1290    }
1291
1292    // User-specified absence of range info.
1293    if (Ranges.size() == 1 && !Ranges.begin()->isValid())
1294      return std::make_pair(ranges_iterator(), ranges_iterator());
1295
1296    return std::make_pair(Ranges.begin(), Ranges.end());
1297}
1298
1299PathDiagnosticLocation BugReport::getLocation(const SourceManager &SM) const {
1300  if (ErrorNode) {
1301    assert(!Location.isValid() &&
1302     "Either Location or ErrorNode should be specified but not both.");
1303
1304    if (const Stmt *S = GetCurrentOrPreviousStmt(ErrorNode)) {
1305      const LocationContext *LC = ErrorNode->getLocationContext();
1306
1307      // For member expressions, return the location of the '.' or '->'.
1308      if (const MemberExpr *ME = dyn_cast<MemberExpr>(S))
1309        return PathDiagnosticLocation::createMemberLoc(ME, SM);
1310      // For binary operators, return the location of the operator.
1311      if (const BinaryOperator *B = dyn_cast<BinaryOperator>(S))
1312        return PathDiagnosticLocation::createOperatorLoc(B, SM);
1313
1314      return PathDiagnosticLocation::createBegin(S, SM, LC);
1315    }
1316  } else {
1317    assert(Location.isValid());
1318    return Location;
1319  }
1320
1321  return PathDiagnosticLocation();
1322}
1323
1324//===----------------------------------------------------------------------===//
1325// Methods for BugReporter and subclasses.
1326//===----------------------------------------------------------------------===//
1327
1328BugReportEquivClass::~BugReportEquivClass() {
1329  for (iterator I=begin(), E=end(); I!=E; ++I) delete *I;
1330}
1331
1332GRBugReporter::~GRBugReporter() { }
1333BugReporterData::~BugReporterData() {}
1334
1335ExplodedGraph &GRBugReporter::getGraph() { return Eng.getGraph(); }
1336
1337ProgramStateManager&
1338GRBugReporter::getStateManager() { return Eng.getStateManager(); }
1339
1340BugReporter::~BugReporter() {
1341  FlushReports();
1342
1343  // Free the bug reports we are tracking.
1344  typedef std::vector<BugReportEquivClass *> ContTy;
1345  for (ContTy::iterator I = EQClassesVector.begin(), E = EQClassesVector.end();
1346       I != E; ++I) {
1347    delete *I;
1348  }
1349}
1350
1351void BugReporter::FlushReports() {
1352  if (BugTypes.isEmpty())
1353    return;
1354
1355  // First flush the warnings for each BugType.  This may end up creating new
1356  // warnings and new BugTypes.
1357  // FIXME: Only NSErrorChecker needs BugType's FlushReports.
1358  // Turn NSErrorChecker into a proper checker and remove this.
1359  SmallVector<const BugType*, 16> bugTypes;
1360  for (BugTypesTy::iterator I=BugTypes.begin(), E=BugTypes.end(); I!=E; ++I)
1361    bugTypes.push_back(*I);
1362  for (SmallVector<const BugType*, 16>::iterator
1363         I = bugTypes.begin(), E = bugTypes.end(); I != E; ++I)
1364    const_cast<BugType*>(*I)->FlushReports(*this);
1365
1366  typedef llvm::FoldingSet<BugReportEquivClass> SetTy;
1367  for (SetTy::iterator EI=EQClasses.begin(), EE=EQClasses.end(); EI!=EE;++EI){
1368    BugReportEquivClass& EQ = *EI;
1369    FlushReport(EQ);
1370  }
1371
1372  // BugReporter owns and deletes only BugTypes created implicitly through
1373  // EmitBasicReport.
1374  // FIXME: There are leaks from checkers that assume that the BugTypes they
1375  // create will be destroyed by the BugReporter.
1376  for (llvm::StringMap<BugType*>::iterator
1377         I = StrBugTypes.begin(), E = StrBugTypes.end(); I != E; ++I)
1378    delete I->second;
1379
1380  // Remove all references to the BugType objects.
1381  BugTypes = F.getEmptySet();
1382}
1383
1384//===----------------------------------------------------------------------===//
1385// PathDiagnostics generation.
1386//===----------------------------------------------------------------------===//
1387
1388static std::pair<std::pair<ExplodedGraph*, NodeBackMap*>,
1389                 std::pair<ExplodedNode*, unsigned> >
1390MakeReportGraph(const ExplodedGraph* G,
1391                SmallVectorImpl<const ExplodedNode*> &nodes) {
1392
1393  // Create the trimmed graph.  It will contain the shortest paths from the
1394  // error nodes to the root.  In the new graph we should only have one
1395  // error node unless there are two or more error nodes with the same minimum
1396  // path length.
1397  ExplodedGraph* GTrim;
1398  InterExplodedGraphMap* NMap;
1399
1400  llvm::DenseMap<const void*, const void*> InverseMap;
1401  llvm::tie(GTrim, NMap) = G->Trim(nodes.data(), nodes.data() + nodes.size(),
1402                                   &InverseMap);
1403
1404  // Create owning pointers for GTrim and NMap just to ensure that they are
1405  // released when this function exists.
1406  OwningPtr<ExplodedGraph> AutoReleaseGTrim(GTrim);
1407  OwningPtr<InterExplodedGraphMap> AutoReleaseNMap(NMap);
1408
1409  // Find the (first) error node in the trimmed graph.  We just need to consult
1410  // the node map (NMap) which maps from nodes in the original graph to nodes
1411  // in the new graph.
1412
1413  std::queue<const ExplodedNode*> WS;
1414  typedef llvm::DenseMap<const ExplodedNode*, unsigned> IndexMapTy;
1415  IndexMapTy IndexMap;
1416
1417  for (unsigned nodeIndex = 0 ; nodeIndex < nodes.size(); ++nodeIndex) {
1418    const ExplodedNode *originalNode = nodes[nodeIndex];
1419    if (const ExplodedNode *N = NMap->getMappedNode(originalNode)) {
1420      WS.push(N);
1421      IndexMap[originalNode] = nodeIndex;
1422    }
1423  }
1424
1425  assert(!WS.empty() && "No error node found in the trimmed graph.");
1426
1427  // Create a new (third!) graph with a single path.  This is the graph
1428  // that will be returned to the caller.
1429  ExplodedGraph *GNew = new ExplodedGraph();
1430
1431  // Sometimes the trimmed graph can contain a cycle.  Perform a reverse BFS
1432  // to the root node, and then construct a new graph that contains only
1433  // a single path.
1434  llvm::DenseMap<const void*,unsigned> Visited;
1435
1436  unsigned cnt = 0;
1437  const ExplodedNode *Root = 0;
1438
1439  while (!WS.empty()) {
1440    const ExplodedNode *Node = WS.front();
1441    WS.pop();
1442
1443    if (Visited.find(Node) != Visited.end())
1444      continue;
1445
1446    Visited[Node] = cnt++;
1447
1448    if (Node->pred_empty()) {
1449      Root = Node;
1450      break;
1451    }
1452
1453    for (ExplodedNode::const_pred_iterator I=Node->pred_begin(),
1454         E=Node->pred_end(); I!=E; ++I)
1455      WS.push(*I);
1456  }
1457
1458  assert(Root);
1459
1460  // Now walk from the root down the BFS path, always taking the successor
1461  // with the lowest number.
1462  ExplodedNode *Last = 0, *First = 0;
1463  NodeBackMap *BM = new NodeBackMap();
1464  unsigned NodeIndex = 0;
1465
1466  for ( const ExplodedNode *N = Root ;;) {
1467    // Lookup the number associated with the current node.
1468    llvm::DenseMap<const void*,unsigned>::iterator I = Visited.find(N);
1469    assert(I != Visited.end());
1470
1471    // Create the equivalent node in the new graph with the same state
1472    // and location.
1473    ExplodedNode *NewN = GNew->getNode(N->getLocation(), N->getState());
1474
1475    // Store the mapping to the original node.
1476    llvm::DenseMap<const void*, const void*>::iterator IMitr=InverseMap.find(N);
1477    assert(IMitr != InverseMap.end() && "No mapping to original node.");
1478    (*BM)[NewN] = (const ExplodedNode*) IMitr->second;
1479
1480    // Link up the new node with the previous node.
1481    if (Last)
1482      NewN->addPredecessor(Last, *GNew);
1483
1484    Last = NewN;
1485
1486    // Are we at the final node?
1487    IndexMapTy::iterator IMI =
1488      IndexMap.find((const ExplodedNode*)(IMitr->second));
1489    if (IMI != IndexMap.end()) {
1490      First = NewN;
1491      NodeIndex = IMI->second;
1492      break;
1493    }
1494
1495    // Find the next successor node.  We choose the node that is marked
1496    // with the lowest DFS number.
1497    ExplodedNode::const_succ_iterator SI = N->succ_begin();
1498    ExplodedNode::const_succ_iterator SE = N->succ_end();
1499    N = 0;
1500
1501    for (unsigned MinVal = 0; SI != SE; ++SI) {
1502
1503      I = Visited.find(*SI);
1504
1505      if (I == Visited.end())
1506        continue;
1507
1508      if (!N || I->second < MinVal) {
1509        N = *SI;
1510        MinVal = I->second;
1511      }
1512    }
1513
1514    assert(N);
1515  }
1516
1517  assert(First);
1518
1519  return std::make_pair(std::make_pair(GNew, BM),
1520                        std::make_pair(First, NodeIndex));
1521}
1522
1523/// CompactPathDiagnostic - This function postprocesses a PathDiagnostic object
1524///  and collapses PathDiagosticPieces that are expanded by macros.
1525static void CompactPathDiagnostic(PathDiagnostic &PD, const SourceManager& SM) {
1526  typedef std::vector<std::pair<PathDiagnosticMacroPiece*, SourceLocation> >
1527          MacroStackTy;
1528
1529  typedef std::vector<PathDiagnosticPiece*>
1530          PiecesTy;
1531
1532  MacroStackTy MacroStack;
1533  PiecesTy Pieces;
1534
1535  for (PathDiagnostic::iterator I = PD.begin(), E = PD.end(); I!=E; ++I) {
1536    // Get the location of the PathDiagnosticPiece.
1537    const FullSourceLoc Loc = I->getLocation().asLocation();
1538
1539    // Determine the instantiation location, which is the location we group
1540    // related PathDiagnosticPieces.
1541    SourceLocation InstantiationLoc = Loc.isMacroID() ?
1542                                      SM.getExpansionLoc(Loc) :
1543                                      SourceLocation();
1544
1545    if (Loc.isFileID()) {
1546      MacroStack.clear();
1547      Pieces.push_back(&*I);
1548      continue;
1549    }
1550
1551    assert(Loc.isMacroID());
1552
1553    // Is the PathDiagnosticPiece within the same macro group?
1554    if (!MacroStack.empty() && InstantiationLoc == MacroStack.back().second) {
1555      MacroStack.back().first->push_back(&*I);
1556      continue;
1557    }
1558
1559    // We aren't in the same group.  Are we descending into a new macro
1560    // or are part of an old one?
1561    PathDiagnosticMacroPiece *MacroGroup = 0;
1562
1563    SourceLocation ParentInstantiationLoc = InstantiationLoc.isMacroID() ?
1564                                          SM.getExpansionLoc(Loc) :
1565                                          SourceLocation();
1566
1567    // Walk the entire macro stack.
1568    while (!MacroStack.empty()) {
1569      if (InstantiationLoc == MacroStack.back().second) {
1570        MacroGroup = MacroStack.back().first;
1571        break;
1572      }
1573
1574      if (ParentInstantiationLoc == MacroStack.back().second) {
1575        MacroGroup = MacroStack.back().first;
1576        break;
1577      }
1578
1579      MacroStack.pop_back();
1580    }
1581
1582    if (!MacroGroup || ParentInstantiationLoc == MacroStack.back().second) {
1583      // Create a new macro group and add it to the stack.
1584      PathDiagnosticMacroPiece *NewGroup =
1585        new PathDiagnosticMacroPiece(
1586          PathDiagnosticLocation::createSingleLocation(I->getLocation()));
1587
1588      if (MacroGroup)
1589        MacroGroup->push_back(NewGroup);
1590      else {
1591        assert(InstantiationLoc.isFileID());
1592        Pieces.push_back(NewGroup);
1593      }
1594
1595      MacroGroup = NewGroup;
1596      MacroStack.push_back(std::make_pair(MacroGroup, InstantiationLoc));
1597    }
1598
1599    // Finally, add the PathDiagnosticPiece to the group.
1600    MacroGroup->push_back(&*I);
1601  }
1602
1603  // Now take the pieces and construct a new PathDiagnostic.
1604  PD.resetPath(false);
1605
1606  for (PiecesTy::iterator I=Pieces.begin(), E=Pieces.end(); I!=E; ++I) {
1607    if (PathDiagnosticMacroPiece *MP=dyn_cast<PathDiagnosticMacroPiece>(*I))
1608      if (!MP->containsEvent()) {
1609        delete MP;
1610        continue;
1611      }
1612
1613    PD.push_back(*I);
1614  }
1615}
1616
1617void GRBugReporter::GeneratePathDiagnostic(PathDiagnostic& PD,
1618                        SmallVectorImpl<BugReport *> &bugReports) {
1619
1620  assert(!bugReports.empty());
1621  SmallVector<const ExplodedNode *, 10> errorNodes;
1622  for (SmallVectorImpl<BugReport*>::iterator I = bugReports.begin(),
1623    E = bugReports.end(); I != E; ++I) {
1624      errorNodes.push_back((*I)->getErrorNode());
1625  }
1626
1627  // Construct a new graph that contains only a single path from the error
1628  // node to a root.
1629  const std::pair<std::pair<ExplodedGraph*, NodeBackMap*>,
1630  std::pair<ExplodedNode*, unsigned> >&
1631    GPair = MakeReportGraph(&getGraph(), errorNodes);
1632
1633  // Find the BugReport with the original location.
1634  assert(GPair.second.second < bugReports.size());
1635  BugReport *R = bugReports[GPair.second.second];
1636  assert(R && "No original report found for sliced graph.");
1637
1638  OwningPtr<ExplodedGraph> ReportGraph(GPair.first.first);
1639  OwningPtr<NodeBackMap> BackMap(GPair.first.second);
1640  const ExplodedNode *N = GPair.second.first;
1641
1642  // Start building the path diagnostic...
1643  PathDiagnosticBuilder PDB(*this, R, BackMap.get(),
1644                            getPathDiagnosticConsumer());
1645
1646  // Register additional node visitors.
1647  R->addVisitor(new NilReceiverBRVisitor());
1648  R->addVisitor(new ConditionBRVisitor());
1649
1650  // If inlining is turning out, emit diagnostics for CallEnter and
1651  // CallExit at the top level.
1652  bool showTopLevel = Eng.getAnalysisManager().shouldInlineCall();
1653  R->addVisitor(new CallEnterExitBRVisitor(showTopLevel));
1654
1655  // Generate the very last diagnostic piece - the piece is visible before
1656  // the trace is expanded.
1657  PathDiagnosticPiece *LastPiece = 0;
1658  for (BugReport::visitor_iterator I = R->visitor_begin(),
1659                                   E = R->visitor_end(); I!=E; ++I) {
1660    if (PathDiagnosticPiece *Piece = (*I)->getEndPath(PDB, N, *R)) {
1661      assert (!LastPiece &&
1662              "There can only be one final piece in a diagnostic.");
1663      LastPiece = Piece;
1664    }
1665  }
1666  if (!LastPiece)
1667    LastPiece = BugReporterVisitor::getDefaultEndPath(PDB, N, *R);
1668  if (LastPiece)
1669    PD.push_back(LastPiece);
1670  else
1671    return;
1672
1673  switch (PDB.getGenerationScheme()) {
1674    case PathDiagnosticConsumer::Extensive:
1675      GenerateExtensivePathDiagnostic(PD, PDB, N);
1676      break;
1677    case PathDiagnosticConsumer::Minimal:
1678      GenerateMinimalPathDiagnostic(PD, PDB, N);
1679      break;
1680  }
1681}
1682
1683void BugReporter::Register(BugType *BT) {
1684  BugTypes = F.add(BugTypes, BT);
1685}
1686
1687void BugReporter::EmitReport(BugReport* R) {
1688  // Compute the bug report's hash to determine its equivalence class.
1689  llvm::FoldingSetNodeID ID;
1690  R->Profile(ID);
1691
1692  // Lookup the equivance class.  If there isn't one, create it.
1693  BugType& BT = R->getBugType();
1694  Register(&BT);
1695  void *InsertPos;
1696  BugReportEquivClass* EQ = EQClasses.FindNodeOrInsertPos(ID, InsertPos);
1697
1698  if (!EQ) {
1699    EQ = new BugReportEquivClass(R);
1700    EQClasses.InsertNode(EQ, InsertPos);
1701    EQClassesVector.push_back(EQ);
1702  }
1703  else
1704    EQ->AddReport(R);
1705}
1706
1707
1708//===----------------------------------------------------------------------===//
1709// Emitting reports in equivalence classes.
1710//===----------------------------------------------------------------------===//
1711
1712namespace {
1713struct FRIEC_WLItem {
1714  const ExplodedNode *N;
1715  ExplodedNode::const_succ_iterator I, E;
1716
1717  FRIEC_WLItem(const ExplodedNode *n)
1718  : N(n), I(N->succ_begin()), E(N->succ_end()) {}
1719};
1720}
1721
1722static BugReport *
1723FindReportInEquivalenceClass(BugReportEquivClass& EQ,
1724                             SmallVectorImpl<BugReport*> &bugReports) {
1725
1726  BugReportEquivClass::iterator I = EQ.begin(), E = EQ.end();
1727  assert(I != E);
1728  BugReport *R = *I;
1729  BugType& BT = R->getBugType();
1730
1731  // If we don't need to suppress any of the nodes because they are
1732  // post-dominated by a sink, simply add all the nodes in the equivalence class
1733  // to 'Nodes'.  Any of the reports will serve as a "representative" report.
1734  if (!BT.isSuppressOnSink()) {
1735    for (BugReportEquivClass::iterator I=EQ.begin(), E=EQ.end(); I!=E; ++I) {
1736      const ExplodedNode *N = I->getErrorNode();
1737      if (N) {
1738        R = *I;
1739        bugReports.push_back(R);
1740      }
1741    }
1742    return R;
1743  }
1744
1745  // For bug reports that should be suppressed when all paths are post-dominated
1746  // by a sink node, iterate through the reports in the equivalence class
1747  // until we find one that isn't post-dominated (if one exists).  We use a
1748  // DFS traversal of the ExplodedGraph to find a non-sink node.  We could write
1749  // this as a recursive function, but we don't want to risk blowing out the
1750  // stack for very long paths.
1751  BugReport *exampleReport = 0;
1752
1753  for (; I != E; ++I) {
1754    R = *I;
1755    const ExplodedNode *errorNode = R->getErrorNode();
1756
1757    if (!errorNode)
1758      continue;
1759    if (errorNode->isSink()) {
1760      llvm_unreachable(
1761           "BugType::isSuppressSink() should not be 'true' for sink end nodes");
1762    }
1763    // No successors?  By definition this nodes isn't post-dominated by a sink.
1764    if (errorNode->succ_empty()) {
1765      bugReports.push_back(R);
1766      if (!exampleReport)
1767        exampleReport = R;
1768      continue;
1769    }
1770
1771    // At this point we know that 'N' is not a sink and it has at least one
1772    // successor.  Use a DFS worklist to find a non-sink end-of-path node.
1773    typedef FRIEC_WLItem WLItem;
1774    typedef SmallVector<WLItem, 10> DFSWorkList;
1775    llvm::DenseMap<const ExplodedNode *, unsigned> Visited;
1776
1777    DFSWorkList WL;
1778    WL.push_back(errorNode);
1779    Visited[errorNode] = 1;
1780
1781    while (!WL.empty()) {
1782      WLItem &WI = WL.back();
1783      assert(!WI.N->succ_empty());
1784
1785      for (; WI.I != WI.E; ++WI.I) {
1786        const ExplodedNode *Succ = *WI.I;
1787        // End-of-path node?
1788        if (Succ->succ_empty()) {
1789          // If we found an end-of-path node that is not a sink.
1790          if (!Succ->isSink()) {
1791            bugReports.push_back(R);
1792            if (!exampleReport)
1793              exampleReport = R;
1794            WL.clear();
1795            break;
1796          }
1797          // Found a sink?  Continue on to the next successor.
1798          continue;
1799        }
1800        // Mark the successor as visited.  If it hasn't been explored,
1801        // enqueue it to the DFS worklist.
1802        unsigned &mark = Visited[Succ];
1803        if (!mark) {
1804          mark = 1;
1805          WL.push_back(Succ);
1806          break;
1807        }
1808      }
1809
1810      // The worklist may have been cleared at this point.  First
1811      // check if it is empty before checking the last item.
1812      if (!WL.empty() && &WL.back() == &WI)
1813        WL.pop_back();
1814    }
1815  }
1816
1817  // ExampleReport will be NULL if all the nodes in the equivalence class
1818  // were post-dominated by sinks.
1819  return exampleReport;
1820}
1821
1822//===----------------------------------------------------------------------===//
1823// DiagnosticCache.  This is a hack to cache analyzer diagnostics.  It
1824// uses global state, which eventually should go elsewhere.
1825//===----------------------------------------------------------------------===//
1826namespace {
1827class DiagCacheItem : public llvm::FoldingSetNode {
1828  llvm::FoldingSetNodeID ID;
1829public:
1830  DiagCacheItem(BugReport *R, PathDiagnostic *PD) {
1831    R->Profile(ID);
1832    PD->Profile(ID);
1833  }
1834
1835  void Profile(llvm::FoldingSetNodeID &id) {
1836    id = ID;
1837  }
1838
1839  llvm::FoldingSetNodeID &getID() { return ID; }
1840};
1841}
1842
1843static bool IsCachedDiagnostic(BugReport *R, PathDiagnostic *PD) {
1844  // FIXME: Eventually this diagnostic cache should reside in something
1845  // like AnalysisManager instead of being a static variable.  This is
1846  // really unsafe in the long term.
1847  typedef llvm::FoldingSet<DiagCacheItem> DiagnosticCache;
1848  static DiagnosticCache DC;
1849
1850  void *InsertPos;
1851  DiagCacheItem *Item = new DiagCacheItem(R, PD);
1852
1853  if (DC.FindNodeOrInsertPos(Item->getID(), InsertPos)) {
1854    delete Item;
1855    return true;
1856  }
1857
1858  DC.InsertNode(Item, InsertPos);
1859  return false;
1860}
1861
1862void BugReporter::FlushReport(BugReportEquivClass& EQ) {
1863  SmallVector<BugReport*, 10> bugReports;
1864  BugReport *exampleReport = FindReportInEquivalenceClass(EQ, bugReports);
1865  if (!exampleReport)
1866    return;
1867
1868  PathDiagnosticConsumer* PD = getPathDiagnosticConsumer();
1869
1870  // FIXME: Make sure we use the 'R' for the path that was actually used.
1871  // Probably doesn't make a difference in practice.
1872  BugType& BT = exampleReport->getBugType();
1873
1874  OwningPtr<PathDiagnostic>
1875    D(new PathDiagnostic(exampleReport->getBugType().getName(),
1876                         !PD || PD->useVerboseDescription()
1877                         ? exampleReport->getDescription()
1878                         : exampleReport->getShortDescription(),
1879                         BT.getCategory()));
1880
1881  if (!bugReports.empty())
1882    GeneratePathDiagnostic(*D.get(), bugReports);
1883
1884  if (IsCachedDiagnostic(exampleReport, D.get()))
1885    return;
1886
1887  // Get the meta data.
1888  const BugReport::ExtraTextList &Meta =
1889                                  exampleReport->getExtraText();
1890  for (BugReport::ExtraTextList::const_iterator i = Meta.begin(),
1891                                                e = Meta.end(); i != e; ++i) {
1892    D->addMeta(*i);
1893  }
1894
1895  // Emit a summary diagnostic to the regular Diagnostics engine.
1896  BugReport::ranges_iterator Beg, End;
1897  llvm::tie(Beg, End) = exampleReport->getRanges();
1898  DiagnosticsEngine &Diag = getDiagnostic();
1899
1900  // Search the description for '%', as that will be interpretted as a
1901  // format character by FormatDiagnostics.
1902  StringRef desc = exampleReport->getShortDescription();
1903  unsigned ErrorDiag;
1904  {
1905    SmallString<512> TmpStr;
1906    llvm::raw_svector_ostream Out(TmpStr);
1907    for (StringRef::iterator I=desc.begin(), E=desc.end(); I!=E; ++I)
1908      if (*I == '%')
1909        Out << "%%";
1910      else
1911        Out << *I;
1912
1913    Out.flush();
1914    ErrorDiag = Diag.getCustomDiagID(DiagnosticsEngine::Warning, TmpStr);
1915  }
1916
1917  {
1918    DiagnosticBuilder diagBuilder = Diag.Report(
1919      exampleReport->getLocation(getSourceManager()).asLocation(), ErrorDiag);
1920    for (BugReport::ranges_iterator I = Beg; I != End; ++I)
1921      diagBuilder << *I;
1922  }
1923
1924  // Emit a full diagnostic for the path if we have a PathDiagnosticConsumer.
1925  if (!PD)
1926    return;
1927
1928  if (D->empty()) {
1929    PathDiagnosticPiece *piece = new PathDiagnosticEventPiece(
1930                                 exampleReport->getLocation(getSourceManager()),
1931                                 exampleReport->getDescription());
1932
1933    for ( ; Beg != End; ++Beg) piece->addRange(*Beg);
1934    D->push_back(piece);
1935  }
1936
1937  PD->HandlePathDiagnostic(D.take());
1938}
1939
1940void BugReporter::EmitBasicReport(StringRef name, StringRef str,
1941                                  PathDiagnosticLocation Loc,
1942                                  SourceRange* RBeg, unsigned NumRanges) {
1943  EmitBasicReport(name, "", str, Loc, RBeg, NumRanges);
1944}
1945
1946void BugReporter::EmitBasicReport(StringRef name,
1947                                  StringRef category,
1948                                  StringRef str, PathDiagnosticLocation Loc,
1949                                  SourceRange* RBeg, unsigned NumRanges) {
1950
1951  // 'BT' is owned by BugReporter.
1952  BugType *BT = getBugTypeForName(name, category);
1953  BugReport *R = new BugReport(*BT, str, Loc);
1954  for ( ; NumRanges > 0 ; --NumRanges, ++RBeg) R->addRange(*RBeg);
1955  EmitReport(R);
1956}
1957
1958BugType *BugReporter::getBugTypeForName(StringRef name,
1959                                        StringRef category) {
1960  SmallString<136> fullDesc;
1961  llvm::raw_svector_ostream(fullDesc) << name << ":" << category;
1962  llvm::StringMapEntry<BugType *> &
1963      entry = StrBugTypes.GetOrCreateValue(fullDesc);
1964  BugType *BT = entry.getValue();
1965  if (!BT) {
1966    BT = new BugType(name, category);
1967    entry.setValue(BT);
1968  }
1969  return BT;
1970}
1971