BugReporter.cpp revision 00bd44d5677783527d7517c1ffe45e4d75a0f56f 
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/STLExtras.h"
31#include "llvm/ADT/OwningPtr.h"
32#include <queue>
33
34using namespace clang;
35using namespace ento;
36
37BugReporterVisitor::~BugReporterVisitor() {}
38
39void BugReporterContext::anchor() {}
40
41//===----------------------------------------------------------------------===//
42// Helper routines for walking the ExplodedGraph and fetching statements.
43//===----------------------------------------------------------------------===//
44
45static inline const Stmt *GetStmt(const ProgramPoint &P) {
46  if (const StmtPoint* SP = dyn_cast<StmtPoint>(&P))
47    return SP->getStmt();
48  else if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P))
49    return BE->getSrc()->getTerminator();
50
51  return 0;
52}
53
54static inline const ExplodedNode*
55GetPredecessorNode(const ExplodedNode *N) {
56  return N->pred_empty() ? NULL : *(N->pred_begin());
57}
58
59static inline const ExplodedNode*
60GetSuccessorNode(const ExplodedNode *N) {
61  return N->succ_empty() ? NULL : *(N->succ_begin());
62}
63
64static const Stmt *GetPreviousStmt(const ExplodedNode *N) {
65  for (N = GetPredecessorNode(N); N; N = GetPredecessorNode(N))
66    if (const Stmt *S = GetStmt(N->getLocation()))
67      return S;
68
69  return 0;
70}
71
72static const Stmt *GetNextStmt(const ExplodedNode *N) {
73  for (N = GetSuccessorNode(N); N; N = GetSuccessorNode(N))
74    if (const Stmt *S = GetStmt(N->getLocation())) {
75      // Check if the statement is '?' or '&&'/'||'.  These are "merges",
76      // not actual statement points.
77      switch (S->getStmtClass()) {
78        case Stmt::ChooseExprClass:
79        case Stmt::BinaryConditionalOperatorClass: continue;
80        case Stmt::ConditionalOperatorClass: continue;
81        case Stmt::BinaryOperatorClass: {
82          BinaryOperatorKind Op = cast<BinaryOperator>(S)->getOpcode();
83          if (Op == BO_LAnd || Op == BO_LOr)
84            continue;
85          break;
86        }
87        default:
88          break;
89      }
90      return S;
91    }
92
93  return 0;
94}
95
96static inline const Stmt*
97GetCurrentOrPreviousStmt(const ExplodedNode *N) {
98  if (const Stmt *S = GetStmt(N->getLocation()))
99    return S;
100
101  return GetPreviousStmt(N);
102}
103
104static inline const Stmt*
105GetCurrentOrNextStmt(const ExplodedNode *N) {
106  if (const Stmt *S = GetStmt(N->getLocation()))
107    return S;
108
109  return GetNextStmt(N);
110}
111
112//===----------------------------------------------------------------------===//
113// PathDiagnosticBuilder and its associated routines and helper objects.
114//===----------------------------------------------------------------------===//
115
116typedef llvm::DenseMap<const ExplodedNode*,
117const ExplodedNode*> NodeBackMap;
118
119namespace {
120class NodeMapClosure : public BugReport::NodeResolver {
121  NodeBackMap& M;
122public:
123  NodeMapClosure(NodeBackMap *m) : M(*m) {}
124  ~NodeMapClosure() {}
125
126  const ExplodedNode *getOriginalNode(const ExplodedNode *N) {
127    NodeBackMap::iterator I = M.find(N);
128    return I == M.end() ? 0 : I->second;
129  }
130};
131
132class PathDiagnosticBuilder : public BugReporterContext {
133  BugReport *R;
134  PathDiagnosticConsumer *PDC;
135  llvm::OwningPtr<ParentMap> PM;
136  NodeMapClosure NMC;
137public:
138  PathDiagnosticBuilder(GRBugReporter &br,
139                        BugReport *r, NodeBackMap *Backmap,
140                        PathDiagnosticConsumer *pdc)
141    : BugReporterContext(br),
142      R(r), PDC(pdc), NMC(Backmap) {}
143
144  PathDiagnosticLocation ExecutionContinues(const ExplodedNode *N);
145
146  PathDiagnosticLocation ExecutionContinues(llvm::raw_string_ostream &os,
147                                            const ExplodedNode *N);
148
149  BugReport *getBugReport() { return R; }
150
151  Decl const &getCodeDecl() { return R->getErrorNode()->getCodeDecl(); }
152
153  const LocationContext* getLocationContext() {
154    return R->getErrorNode()->getLocationContext();
155  }
156
157  ParentMap& getParentMap() { return R->getErrorNode()->getParentMap(); }
158
159  const Stmt *getParent(const Stmt *S) {
160    return getParentMap().getParent(S);
161  }
162
163  virtual NodeMapClosure& getNodeResolver() { return NMC; }
164
165  PathDiagnosticLocation getEnclosingStmtLocation(const Stmt *S);
166
167  PathDiagnosticConsumer::PathGenerationScheme getGenerationScheme() const {
168    return PDC ? PDC->getGenerationScheme() : PathDiagnosticConsumer::Extensive;
169  }
170
171  bool supportsLogicalOpControlFlow() const {
172    return PDC ? PDC->supportsLogicalOpControlFlow() : true;
173  }
174};
175} // end anonymous namespace
176
177PathDiagnosticLocation
178PathDiagnosticBuilder::ExecutionContinues(const ExplodedNode *N) {
179  if (const Stmt *S = GetNextStmt(N))
180    return PathDiagnosticLocation(S, getSourceManager(), getLocationContext());
181
182  return PathDiagnosticLocation::createDeclEnd(N->getLocationContext(),
183                                               getSourceManager());
184}
185
186PathDiagnosticLocation
187PathDiagnosticBuilder::ExecutionContinues(llvm::raw_string_ostream &os,
188                                          const ExplodedNode *N) {
189
190  // Slow, but probably doesn't matter.
191  if (os.str().empty())
192    os << ' ';
193
194  const PathDiagnosticLocation &Loc = ExecutionContinues(N);
195
196  if (Loc.asStmt())
197    os << "Execution continues on line "
198       << getSourceManager().getExpansionLineNumber(Loc.asLocation())
199       << '.';
200  else {
201    os << "Execution jumps to the end of the ";
202    const Decl *D = N->getLocationContext()->getDecl();
203    if (isa<ObjCMethodDecl>(D))
204      os << "method";
205    else if (isa<FunctionDecl>(D))
206      os << "function";
207    else {
208      assert(isa<BlockDecl>(D));
209      os << "anonymous block";
210    }
211    os << '.';
212  }
213
214  return Loc;
215}
216
217static bool IsNested(const Stmt *S, ParentMap &PM) {
218  if (isa<Expr>(S) && PM.isConsumedExpr(cast<Expr>(S)))
219    return true;
220
221  const Stmt *Parent = PM.getParentIgnoreParens(S);
222
223  if (Parent)
224    switch (Parent->getStmtClass()) {
225      case Stmt::ForStmtClass:
226      case Stmt::DoStmtClass:
227      case Stmt::WhileStmtClass:
228        return true;
229      default:
230        break;
231    }
232
233  return false;
234}
235
236PathDiagnosticLocation
237PathDiagnosticBuilder::getEnclosingStmtLocation(const Stmt *S) {
238  assert(S && "Null Stmt *passed to getEnclosingStmtLocation");
239  ParentMap &P = getParentMap();
240  SourceManager &SMgr = getSourceManager();
241  const LocationContext *LC = getLocationContext();
242
243  while (IsNested(S, P)) {
244    const Stmt *Parent = P.getParentIgnoreParens(S);
245
246    if (!Parent)
247      break;
248
249    switch (Parent->getStmtClass()) {
250      case Stmt::BinaryOperatorClass: {
251        const BinaryOperator *B = cast<BinaryOperator>(Parent);
252        if (B->isLogicalOp())
253          return PathDiagnosticLocation(S, SMgr, LC);
254        break;
255      }
256      case Stmt::CompoundStmtClass:
257      case Stmt::StmtExprClass:
258        return PathDiagnosticLocation(S, SMgr, LC);
259      case Stmt::ChooseExprClass:
260        // Similar to '?' if we are referring to condition, just have the edge
261        // point to the entire choose expression.
262        if (cast<ChooseExpr>(Parent)->getCond() == S)
263          return PathDiagnosticLocation(Parent, SMgr, LC);
264        else
265          return PathDiagnosticLocation(S, SMgr, LC);
266      case Stmt::BinaryConditionalOperatorClass:
267      case Stmt::ConditionalOperatorClass:
268        // For '?', if we are referring to condition, just have the edge point
269        // to the entire '?' expression.
270        if (cast<AbstractConditionalOperator>(Parent)->getCond() == S)
271          return PathDiagnosticLocation(Parent, SMgr, LC);
272        else
273          return PathDiagnosticLocation(S, SMgr, LC);
274      case Stmt::DoStmtClass:
275          return PathDiagnosticLocation(S, SMgr, LC);
276      case Stmt::ForStmtClass:
277        if (cast<ForStmt>(Parent)->getBody() == S)
278          return PathDiagnosticLocation(S, SMgr, LC);
279        break;
280      case Stmt::IfStmtClass:
281        if (cast<IfStmt>(Parent)->getCond() != S)
282          return PathDiagnosticLocation(S, SMgr, LC);
283        break;
284      case Stmt::ObjCForCollectionStmtClass:
285        if (cast<ObjCForCollectionStmt>(Parent)->getBody() == S)
286          return PathDiagnosticLocation(S, SMgr, LC);
287        break;
288      case Stmt::WhileStmtClass:
289        if (cast<WhileStmt>(Parent)->getCond() != S)
290          return PathDiagnosticLocation(S, SMgr, LC);
291        break;
292      default:
293        break;
294    }
295
296    S = Parent;
297  }
298
299  assert(S && "Cannot have null Stmt for PathDiagnosticLocation");
300
301  // Special case: DeclStmts can appear in for statement declarations, in which
302  //  case the ForStmt is the context.
303  if (isa<DeclStmt>(S)) {
304    if (const Stmt *Parent = P.getParent(S)) {
305      switch (Parent->getStmtClass()) {
306        case Stmt::ForStmtClass:
307        case Stmt::ObjCForCollectionStmtClass:
308          return PathDiagnosticLocation(Parent, SMgr, LC);
309        default:
310          break;
311      }
312    }
313  }
314  else if (isa<BinaryOperator>(S)) {
315    // Special case: the binary operator represents the initialization
316    // code in a for statement (this can happen when the variable being
317    // initialized is an old variable.
318    if (const ForStmt *FS =
319          dyn_cast_or_null<ForStmt>(P.getParentIgnoreParens(S))) {
320      if (FS->getInit() == S)
321        return PathDiagnosticLocation(FS, SMgr, LC);
322    }
323  }
324
325  return PathDiagnosticLocation(S, SMgr, LC);
326}
327
328//===----------------------------------------------------------------------===//
329// ScanNotableSymbols: closure-like callback for scanning Store bindings.
330//===----------------------------------------------------------------------===//
331
332static const VarDecl* GetMostRecentVarDeclBinding(const ExplodedNode *N,
333                                                  ProgramStateManager& VMgr,
334                                                  SVal X) {
335
336  for ( ; N ; N = N->pred_empty() ? 0 : *N->pred_begin()) {
337
338    ProgramPoint P = N->getLocation();
339
340    if (!isa<PostStmt>(P))
341      continue;
342
343    const DeclRefExpr *DR = dyn_cast<DeclRefExpr>(cast<PostStmt>(P).getStmt());
344
345    if (!DR)
346      continue;
347
348    SVal Y = N->getState()->getSVal(DR, N->getLocationContext());
349
350    if (X != Y)
351      continue;
352
353    const VarDecl *VD = dyn_cast<VarDecl>(DR->getDecl());
354
355    if (!VD)
356      continue;
357
358    return VD;
359  }
360
361  return 0;
362}
363
364namespace {
365class NotableSymbolHandler
366: public StoreManager::BindingsHandler {
367
368  SymbolRef Sym;
369  ProgramStateRef PrevSt;
370  const Stmt *S;
371  ProgramStateManager& VMgr;
372  const ExplodedNode *Pred;
373  PathDiagnostic& PD;
374  BugReporter& BR;
375
376public:
377
378  NotableSymbolHandler(SymbolRef sym,
379                       ProgramStateRef prevst,
380                       const Stmt *s,
381                       ProgramStateManager& vmgr,
382                       const ExplodedNode *pred,
383                       PathDiagnostic& pd,
384                       BugReporter& br)
385  : Sym(sym),
386    PrevSt(prevst),
387    S(s),
388    VMgr(vmgr),
389    Pred(pred),
390    PD(pd),
391    BR(br) {}
392
393  bool HandleBinding(StoreManager& SMgr, Store store, const MemRegion* R,
394                     SVal V) {
395
396    SymbolRef ScanSym = V.getAsSymbol();
397
398    if (ScanSym != Sym)
399      return true;
400
401    // Check if the previous state has this binding.
402    SVal X = PrevSt->getSVal(loc::MemRegionVal(R));
403
404    if (X == V) // Same binding?
405      return true;
406
407    // Different binding.  Only handle assignments for now.  We don't pull
408    // this check out of the loop because we will eventually handle other
409    // cases.
410
411    VarDecl *VD = 0;
412
413    if (const BinaryOperator* B = dyn_cast<BinaryOperator>(S)) {
414      if (!B->isAssignmentOp())
415        return true;
416
417      // What variable did we assign to?
418      DeclRefExpr *DR = dyn_cast<DeclRefExpr>(B->getLHS()->IgnoreParenCasts());
419
420      if (!DR)
421        return true;
422
423      VD = dyn_cast<VarDecl>(DR->getDecl());
424    }
425    else if (const DeclStmt *DS = dyn_cast<DeclStmt>(S)) {
426      // FIXME: Eventually CFGs won't have DeclStmts.  Right now we
427      //  assume that each DeclStmt has a single Decl.  This invariant
428      //  holds by construction in the CFG.
429      VD = dyn_cast<VarDecl>(*DS->decl_begin());
430    }
431
432    if (!VD)
433      return true;
434
435    // What is the most recently referenced variable with this binding?
436    const VarDecl *MostRecent = GetMostRecentVarDeclBinding(Pred, VMgr, V);
437
438    if (!MostRecent)
439      return true;
440
441    // Create the diagnostic.
442    if (Loc::isLocType(VD->getType())) {
443      llvm::SmallString<64> buf;
444      llvm::raw_svector_ostream os(buf);
445      os << '\'' << *VD << "' now aliases '" << *MostRecent << '\'';
446      PathDiagnosticLocation L =
447        PathDiagnosticLocation::createBegin(S, BR.getSourceManager(),
448                                                   Pred->getLocationContext());
449      PD.push_front(new PathDiagnosticEventPiece(L, os.str()));
450    }
451
452    return true;
453  }
454};
455}
456
457static void HandleNotableSymbol(const ExplodedNode *N,
458                                const Stmt *S,
459                                SymbolRef Sym, BugReporter& BR,
460                                PathDiagnostic& PD) {
461
462  const ExplodedNode *Pred = N->pred_empty() ? 0 : *N->pred_begin();
463  ProgramStateRef PrevSt = Pred ? Pred->getState() : 0;
464
465  if (!PrevSt)
466    return;
467
468  // Look at the region bindings of the current state that map to the
469  // specified symbol.  Are any of them not in the previous state?
470  ProgramStateManager& VMgr = cast<GRBugReporter>(BR).getStateManager();
471  NotableSymbolHandler H(Sym, PrevSt, S, VMgr, Pred, PD, BR);
472  cast<GRBugReporter>(BR).getStateManager().iterBindings(N->getState(), H);
473}
474
475namespace {
476class ScanNotableSymbols
477: public StoreManager::BindingsHandler {
478
479  llvm::SmallSet<SymbolRef, 10> AlreadyProcessed;
480  const ExplodedNode *N;
481  const Stmt *S;
482  GRBugReporter& BR;
483  PathDiagnostic& PD;
484
485public:
486  ScanNotableSymbols(const ExplodedNode *n, const Stmt *s,
487                     GRBugReporter& br, PathDiagnostic& pd)
488  : N(n), S(s), BR(br), PD(pd) {}
489
490  bool HandleBinding(StoreManager& SMgr, Store store,
491                     const MemRegion* R, SVal V) {
492
493    SymbolRef ScanSym = V.getAsSymbol();
494
495    if (!ScanSym)
496      return true;
497
498    if (!BR.isNotable(ScanSym))
499      return true;
500
501    if (AlreadyProcessed.count(ScanSym))
502      return true;
503
504    AlreadyProcessed.insert(ScanSym);
505
506    HandleNotableSymbol(N, S, ScanSym, BR, PD);
507    return true;
508  }
509};
510} // end anonymous namespace
511
512//===----------------------------------------------------------------------===//
513// "Minimal" path diagnostic generation algorithm.
514//===----------------------------------------------------------------------===//
515
516static void CompactPathDiagnostic(PathDiagnostic &PD, const SourceManager& SM);
517
518static void GenerateMinimalPathDiagnostic(PathDiagnostic& PD,
519                                          PathDiagnosticBuilder &PDB,
520                                          const ExplodedNode *N) {
521
522  SourceManager& SMgr = PDB.getSourceManager();
523  const LocationContext *LC = PDB.getLocationContext();
524  const ExplodedNode *NextNode = N->pred_empty()
525                                        ? NULL : *(N->pred_begin());
526  while (NextNode) {
527    N = NextNode;
528    NextNode = GetPredecessorNode(N);
529
530    ProgramPoint P = N->getLocation();
531
532    if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P)) {
533      const CFGBlock *Src = BE->getSrc();
534      const CFGBlock *Dst = BE->getDst();
535      const Stmt *T = Src->getTerminator();
536
537      if (!T)
538        continue;
539
540      PathDiagnosticLocation Start =
541        PathDiagnosticLocation::createBegin(T, SMgr,
542                                                N->getLocationContext());
543
544      switch (T->getStmtClass()) {
545        default:
546          break;
547
548        case Stmt::GotoStmtClass:
549        case Stmt::IndirectGotoStmtClass: {
550          const Stmt *S = GetNextStmt(N);
551
552          if (!S)
553            continue;
554
555          std::string sbuf;
556          llvm::raw_string_ostream os(sbuf);
557          const PathDiagnosticLocation &End = PDB.getEnclosingStmtLocation(S);
558
559          os << "Control jumps to line "
560          << End.asLocation().getExpansionLineNumber();
561          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
562                                                           os.str()));
563          break;
564        }
565
566        case Stmt::SwitchStmtClass: {
567          // Figure out what case arm we took.
568          std::string sbuf;
569          llvm::raw_string_ostream os(sbuf);
570
571          if (const Stmt *S = Dst->getLabel()) {
572            PathDiagnosticLocation End(S, SMgr, LC);
573
574            switch (S->getStmtClass()) {
575              default:
576                os << "No cases match in the switch statement. "
577                "Control jumps to line "
578                << End.asLocation().getExpansionLineNumber();
579                break;
580              case Stmt::DefaultStmtClass:
581                os << "Control jumps to the 'default' case at line "
582                << End.asLocation().getExpansionLineNumber();
583                break;
584
585              case Stmt::CaseStmtClass: {
586                os << "Control jumps to 'case ";
587                const CaseStmt *Case = cast<CaseStmt>(S);
588                const Expr *LHS = Case->getLHS()->IgnoreParenCasts();
589
590                // Determine if it is an enum.
591                bool GetRawInt = true;
592
593                if (const DeclRefExpr *DR = dyn_cast<DeclRefExpr>(LHS)) {
594                  // FIXME: Maybe this should be an assertion.  Are there cases
595                  // were it is not an EnumConstantDecl?
596                  const EnumConstantDecl *D =
597                    dyn_cast<EnumConstantDecl>(DR->getDecl());
598
599                  if (D) {
600                    GetRawInt = false;
601                    os << *D;
602                  }
603                }
604
605                if (GetRawInt)
606                  os << LHS->EvaluateKnownConstInt(PDB.getASTContext());
607
608                os << ":'  at line "
609                << End.asLocation().getExpansionLineNumber();
610                break;
611              }
612            }
613            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
614                                                             os.str()));
615          }
616          else {
617            os << "'Default' branch taken. ";
618            const PathDiagnosticLocation &End = PDB.ExecutionContinues(os, N);
619            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
620                                                             os.str()));
621          }
622
623          break;
624        }
625
626        case Stmt::BreakStmtClass:
627        case Stmt::ContinueStmtClass: {
628          std::string sbuf;
629          llvm::raw_string_ostream os(sbuf);
630          PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
631          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
632                                                           os.str()));
633          break;
634        }
635
636          // Determine control-flow for ternary '?'.
637        case Stmt::BinaryConditionalOperatorClass:
638        case Stmt::ConditionalOperatorClass: {
639          std::string sbuf;
640          llvm::raw_string_ostream os(sbuf);
641          os << "'?' condition is ";
642
643          if (*(Src->succ_begin()+1) == Dst)
644            os << "false";
645          else
646            os << "true";
647
648          PathDiagnosticLocation End = PDB.ExecutionContinues(N);
649
650          if (const Stmt *S = End.asStmt())
651            End = PDB.getEnclosingStmtLocation(S);
652
653          PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
654                                                           os.str()));
655          break;
656        }
657
658          // Determine control-flow for short-circuited '&&' and '||'.
659        case Stmt::BinaryOperatorClass: {
660          if (!PDB.supportsLogicalOpControlFlow())
661            break;
662
663          const BinaryOperator *B = cast<BinaryOperator>(T);
664          std::string sbuf;
665          llvm::raw_string_ostream os(sbuf);
666          os << "Left side of '";
667
668          if (B->getOpcode() == BO_LAnd) {
669            os << "&&" << "' is ";
670
671            if (*(Src->succ_begin()+1) == Dst) {
672              os << "false";
673              PathDiagnosticLocation End(B->getLHS(), SMgr, LC);
674              PathDiagnosticLocation Start =
675                PathDiagnosticLocation::createOperatorLoc(B, SMgr);
676              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
677                                                               os.str()));
678            }
679            else {
680              os << "true";
681              PathDiagnosticLocation Start(B->getLHS(), SMgr, LC);
682              PathDiagnosticLocation End = PDB.ExecutionContinues(N);
683              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
684                                                               os.str()));
685            }
686          }
687          else {
688            assert(B->getOpcode() == BO_LOr);
689            os << "||" << "' is ";
690
691            if (*(Src->succ_begin()+1) == Dst) {
692              os << "false";
693              PathDiagnosticLocation Start(B->getLHS(), SMgr, LC);
694              PathDiagnosticLocation End = PDB.ExecutionContinues(N);
695              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
696                                                               os.str()));
697            }
698            else {
699              os << "true";
700              PathDiagnosticLocation End(B->getLHS(), SMgr, LC);
701              PathDiagnosticLocation Start =
702                PathDiagnosticLocation::createOperatorLoc(B, SMgr);
703              PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
704                                                               os.str()));
705            }
706          }
707
708          break;
709        }
710
711        case Stmt::DoStmtClass:  {
712          if (*(Src->succ_begin()) == Dst) {
713            std::string sbuf;
714            llvm::raw_string_ostream os(sbuf);
715
716            os << "Loop condition is true. ";
717            PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
718
719            if (const Stmt *S = End.asStmt())
720              End = PDB.getEnclosingStmtLocation(S);
721
722            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
723                                                             os.str()));
724          }
725          else {
726            PathDiagnosticLocation End = PDB.ExecutionContinues(N);
727
728            if (const Stmt *S = End.asStmt())
729              End = PDB.getEnclosingStmtLocation(S);
730
731            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
732                              "Loop condition is false.  Exiting loop"));
733          }
734
735          break;
736        }
737
738        case Stmt::WhileStmtClass:
739        case Stmt::ForStmtClass: {
740          if (*(Src->succ_begin()+1) == Dst) {
741            std::string sbuf;
742            llvm::raw_string_ostream os(sbuf);
743
744            os << "Loop condition is false. ";
745            PathDiagnosticLocation End = PDB.ExecutionContinues(os, N);
746            if (const Stmt *S = End.asStmt())
747              End = PDB.getEnclosingStmtLocation(S);
748
749            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
750                                                             os.str()));
751          }
752          else {
753            PathDiagnosticLocation End = PDB.ExecutionContinues(N);
754            if (const Stmt *S = End.asStmt())
755              End = PDB.getEnclosingStmtLocation(S);
756
757            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
758                            "Loop condition is true.  Entering loop body"));
759          }
760
761          break;
762        }
763
764        case Stmt::IfStmtClass: {
765          PathDiagnosticLocation End = PDB.ExecutionContinues(N);
766
767          if (const Stmt *S = End.asStmt())
768            End = PDB.getEnclosingStmtLocation(S);
769
770          if (*(Src->succ_begin()+1) == Dst)
771            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
772                                                        "Taking false branch"));
773          else
774            PD.push_front(new PathDiagnosticControlFlowPiece(Start, End,
775                                                         "Taking true branch"));
776
777          break;
778        }
779      }
780    }
781
782    if (NextNode) {
783      // Add diagnostic pieces from custom visitors.
784      BugReport *R = PDB.getBugReport();
785      for (BugReport::visitor_iterator I = R->visitor_begin(),
786           E = R->visitor_end(); I!=E; ++I) {
787        if (PathDiagnosticPiece *p = (*I)->VisitNode(N, NextNode, PDB, *R))
788          PD.push_front(p);
789      }
790    }
791
792    if (const PostStmt *PS = dyn_cast<PostStmt>(&P)) {
793      // Scan the region bindings, and see if a "notable" symbol has a new
794      // lval binding.
795      ScanNotableSymbols SNS(N, PS->getStmt(), PDB.getBugReporter(), PD);
796      PDB.getStateManager().iterBindings(N->getState(), SNS);
797    }
798  }
799
800  // After constructing the full PathDiagnostic, do a pass over it to compact
801  // PathDiagnosticPieces that occur within a macro.
802  CompactPathDiagnostic(PD, PDB.getSourceManager());
803}
804
805//===----------------------------------------------------------------------===//
806// "Extensive" PathDiagnostic generation.
807//===----------------------------------------------------------------------===//
808
809static bool IsControlFlowExpr(const Stmt *S) {
810  const Expr *E = dyn_cast<Expr>(S);
811
812  if (!E)
813    return false;
814
815  E = E->IgnoreParenCasts();
816
817  if (isa<AbstractConditionalOperator>(E))
818    return true;
819
820  if (const BinaryOperator *B = dyn_cast<BinaryOperator>(E))
821    if (B->isLogicalOp())
822      return true;
823
824  return false;
825}
826
827namespace {
828class ContextLocation : public PathDiagnosticLocation {
829  bool IsDead;
830public:
831  ContextLocation(const PathDiagnosticLocation &L, bool isdead = false)
832    : PathDiagnosticLocation(L), IsDead(isdead) {}
833
834  void markDead() { IsDead = true; }
835  bool isDead() const { return IsDead; }
836};
837
838class EdgeBuilder {
839  std::vector<ContextLocation> CLocs;
840  typedef std::vector<ContextLocation>::iterator iterator;
841  PathDiagnostic &PD;
842  PathDiagnosticBuilder &PDB;
843  PathDiagnosticLocation PrevLoc;
844
845  bool IsConsumedExpr(const PathDiagnosticLocation &L);
846
847  bool containsLocation(const PathDiagnosticLocation &Container,
848                        const PathDiagnosticLocation &Containee);
849
850  PathDiagnosticLocation getContextLocation(const PathDiagnosticLocation &L);
851
852  PathDiagnosticLocation cleanUpLocation(PathDiagnosticLocation L,
853                                         bool firstCharOnly = false) {
854    if (const Stmt *S = L.asStmt()) {
855      const Stmt *Original = S;
856      while (1) {
857        // Adjust the location for some expressions that are best referenced
858        // by one of their subexpressions.
859        switch (S->getStmtClass()) {
860          default:
861            break;
862          case Stmt::ParenExprClass:
863          case Stmt::GenericSelectionExprClass:
864            S = cast<Expr>(S)->IgnoreParens();
865            firstCharOnly = true;
866            continue;
867          case Stmt::BinaryConditionalOperatorClass:
868          case Stmt::ConditionalOperatorClass:
869            S = cast<AbstractConditionalOperator>(S)->getCond();
870            firstCharOnly = true;
871            continue;
872          case Stmt::ChooseExprClass:
873            S = cast<ChooseExpr>(S)->getCond();
874            firstCharOnly = true;
875            continue;
876          case Stmt::BinaryOperatorClass:
877            S = cast<BinaryOperator>(S)->getLHS();
878            firstCharOnly = true;
879            continue;
880        }
881
882        break;
883      }
884
885      if (S != Original)
886        L = PathDiagnosticLocation(S, L.getManager(), PDB.getLocationContext());
887    }
888
889    if (firstCharOnly)
890      L  = PathDiagnosticLocation::createSingleLocation(L);
891
892    return L;
893  }
894
895  void popLocation() {
896    if (!CLocs.back().isDead() && CLocs.back().asLocation().isFileID()) {
897      // For contexts, we only one the first character as the range.
898      rawAddEdge(cleanUpLocation(CLocs.back(), true));
899    }
900    CLocs.pop_back();
901  }
902
903public:
904  EdgeBuilder(PathDiagnostic &pd, PathDiagnosticBuilder &pdb)
905    : PD(pd), PDB(pdb) {
906
907      // If the PathDiagnostic already has pieces, add the enclosing statement
908      // of the first piece as a context as well.
909      if (!PD.empty()) {
910        PrevLoc = PD.begin()->getLocation();
911
912        if (const Stmt *S = PrevLoc.asStmt())
913          addExtendedContext(PDB.getEnclosingStmtLocation(S).asStmt());
914      }
915  }
916
917  ~EdgeBuilder() {
918    while (!CLocs.empty()) popLocation();
919
920    // Finally, add an initial edge from the start location of the first
921    // statement (if it doesn't already exist).
922    PathDiagnosticLocation L = PathDiagnosticLocation::createDeclBegin(
923                                                       PDB.getLocationContext(),
924                                                       PDB.getSourceManager());
925    if (L.isValid())
926      rawAddEdge(L);
927  }
928
929  void addEdge(PathDiagnosticLocation NewLoc, bool alwaysAdd = false);
930
931  void rawAddEdge(PathDiagnosticLocation NewLoc);
932
933  void addContext(const Stmt *S);
934  void addExtendedContext(const Stmt *S);
935};
936} // end anonymous namespace
937
938
939PathDiagnosticLocation
940EdgeBuilder::getContextLocation(const PathDiagnosticLocation &L) {
941  if (const Stmt *S = L.asStmt()) {
942    if (IsControlFlowExpr(S))
943      return L;
944
945    return PDB.getEnclosingStmtLocation(S);
946  }
947
948  return L;
949}
950
951bool EdgeBuilder::containsLocation(const PathDiagnosticLocation &Container,
952                                   const PathDiagnosticLocation &Containee) {
953
954  if (Container == Containee)
955    return true;
956
957  if (Container.asDecl())
958    return true;
959
960  if (const Stmt *S = Containee.asStmt())
961    if (const Stmt *ContainerS = Container.asStmt()) {
962      while (S) {
963        if (S == ContainerS)
964          return true;
965        S = PDB.getParent(S);
966      }
967      return false;
968    }
969
970  // Less accurate: compare using source ranges.
971  SourceRange ContainerR = Container.asRange();
972  SourceRange ContaineeR = Containee.asRange();
973
974  SourceManager &SM = PDB.getSourceManager();
975  SourceLocation ContainerRBeg = SM.getExpansionLoc(ContainerR.getBegin());
976  SourceLocation ContainerREnd = SM.getExpansionLoc(ContainerR.getEnd());
977  SourceLocation ContaineeRBeg = SM.getExpansionLoc(ContaineeR.getBegin());
978  SourceLocation ContaineeREnd = SM.getExpansionLoc(ContaineeR.getEnd());
979
980  unsigned ContainerBegLine = SM.getExpansionLineNumber(ContainerRBeg);
981  unsigned ContainerEndLine = SM.getExpansionLineNumber(ContainerREnd);
982  unsigned ContaineeBegLine = SM.getExpansionLineNumber(ContaineeRBeg);
983  unsigned ContaineeEndLine = SM.getExpansionLineNumber(ContaineeREnd);
984
985  assert(ContainerBegLine <= ContainerEndLine);
986  assert(ContaineeBegLine <= ContaineeEndLine);
987
988  return (ContainerBegLine <= ContaineeBegLine &&
989          ContainerEndLine >= ContaineeEndLine &&
990          (ContainerBegLine != ContaineeBegLine ||
991           SM.getExpansionColumnNumber(ContainerRBeg) <=
992           SM.getExpansionColumnNumber(ContaineeRBeg)) &&
993          (ContainerEndLine != ContaineeEndLine ||
994           SM.getExpansionColumnNumber(ContainerREnd) >=
995           SM.getExpansionColumnNumber(ContainerREnd)));
996}
997
998void EdgeBuilder::rawAddEdge(PathDiagnosticLocation NewLoc) {
999  if (!PrevLoc.isValid()) {
1000    PrevLoc = NewLoc;
1001    return;
1002  }
1003
1004  const PathDiagnosticLocation &NewLocClean = cleanUpLocation(NewLoc);
1005  const PathDiagnosticLocation &PrevLocClean = cleanUpLocation(PrevLoc);
1006
1007  if (NewLocClean.asLocation() == PrevLocClean.asLocation())
1008    return;
1009
1010  // FIXME: Ignore intra-macro edges for now.
1011  if (NewLocClean.asLocation().getExpansionLoc() ==
1012      PrevLocClean.asLocation().getExpansionLoc())
1013    return;
1014
1015  PD.push_front(new PathDiagnosticControlFlowPiece(NewLocClean, PrevLocClean));
1016  PrevLoc = NewLoc;
1017}
1018
1019void EdgeBuilder::addEdge(PathDiagnosticLocation NewLoc, bool alwaysAdd) {
1020
1021  if (!alwaysAdd && NewLoc.asLocation().isMacroID())
1022    return;
1023
1024  const PathDiagnosticLocation &CLoc = getContextLocation(NewLoc);
1025
1026  while (!CLocs.empty()) {
1027    ContextLocation &TopContextLoc = CLocs.back();
1028
1029    // Is the top location context the same as the one for the new location?
1030    if (TopContextLoc == CLoc) {
1031      if (alwaysAdd) {
1032        if (IsConsumedExpr(TopContextLoc) &&
1033            !IsControlFlowExpr(TopContextLoc.asStmt()))
1034            TopContextLoc.markDead();
1035
1036        rawAddEdge(NewLoc);
1037      }
1038
1039      return;
1040    }
1041
1042    if (containsLocation(TopContextLoc, CLoc)) {
1043      if (alwaysAdd) {
1044        rawAddEdge(NewLoc);
1045
1046        if (IsConsumedExpr(CLoc) && !IsControlFlowExpr(CLoc.asStmt())) {
1047          CLocs.push_back(ContextLocation(CLoc, true));
1048          return;
1049        }
1050      }
1051
1052      CLocs.push_back(CLoc);
1053      return;
1054    }
1055
1056    // Context does not contain the location.  Flush it.
1057    popLocation();
1058  }
1059
1060  // If we reach here, there is no enclosing context.  Just add the edge.
1061  rawAddEdge(NewLoc);
1062}
1063
1064bool EdgeBuilder::IsConsumedExpr(const PathDiagnosticLocation &L) {
1065  if (const Expr *X = dyn_cast_or_null<Expr>(L.asStmt()))
1066    return PDB.getParentMap().isConsumedExpr(X) && !IsControlFlowExpr(X);
1067
1068  return false;
1069}
1070
1071void EdgeBuilder::addExtendedContext(const Stmt *S) {
1072  if (!S)
1073    return;
1074
1075  const Stmt *Parent = PDB.getParent(S);
1076  while (Parent) {
1077    if (isa<CompoundStmt>(Parent))
1078      Parent = PDB.getParent(Parent);
1079    else
1080      break;
1081  }
1082
1083  if (Parent) {
1084    switch (Parent->getStmtClass()) {
1085      case Stmt::DoStmtClass:
1086      case Stmt::ObjCAtSynchronizedStmtClass:
1087        addContext(Parent);
1088      default:
1089        break;
1090    }
1091  }
1092
1093  addContext(S);
1094}
1095
1096void EdgeBuilder::addContext(const Stmt *S) {
1097  if (!S)
1098    return;
1099
1100  PathDiagnosticLocation L(S, PDB.getSourceManager(), PDB.getLocationContext());
1101
1102  while (!CLocs.empty()) {
1103    const PathDiagnosticLocation &TopContextLoc = CLocs.back();
1104
1105    // Is the top location context the same as the one for the new location?
1106    if (TopContextLoc == L)
1107      return;
1108
1109    if (containsLocation(TopContextLoc, L)) {
1110      CLocs.push_back(L);
1111      return;
1112    }
1113
1114    // Context does not contain the location.  Flush it.
1115    popLocation();
1116  }
1117
1118  CLocs.push_back(L);
1119}
1120
1121static void GenerateExtensivePathDiagnostic(PathDiagnostic& PD,
1122                                            PathDiagnosticBuilder &PDB,
1123                                            const ExplodedNode *N) {
1124  EdgeBuilder EB(PD, PDB);
1125  const SourceManager& SM = PDB.getSourceManager();
1126
1127  const ExplodedNode *NextNode = N->pred_empty() ? NULL : *(N->pred_begin());
1128  while (NextNode) {
1129    N = NextNode;
1130    NextNode = GetPredecessorNode(N);
1131    ProgramPoint P = N->getLocation();
1132
1133    do {
1134      // Block edges.
1135      if (const BlockEdge *BE = dyn_cast<BlockEdge>(&P)) {
1136        const CFGBlock &Blk = *BE->getSrc();
1137        const Stmt *Term = Blk.getTerminator();
1138
1139        // Are we jumping to the head of a loop?  Add a special diagnostic.
1140        if (const Stmt *Loop = BE->getDst()->getLoopTarget()) {
1141          PathDiagnosticLocation L(Loop, SM, PDB.getLocationContext());
1142          const CompoundStmt *CS = NULL;
1143
1144          if (!Term) {
1145            if (const ForStmt *FS = dyn_cast<ForStmt>(Loop))
1146              CS = dyn_cast<CompoundStmt>(FS->getBody());
1147            else if (const WhileStmt *WS = dyn_cast<WhileStmt>(Loop))
1148              CS = dyn_cast<CompoundStmt>(WS->getBody());
1149          }
1150
1151          PathDiagnosticEventPiece *p =
1152            new PathDiagnosticEventPiece(L,
1153                                        "Looping back to the head of the loop");
1154
1155          EB.addEdge(p->getLocation(), true);
1156          PD.push_front(p);
1157
1158          if (CS) {
1159            PathDiagnosticLocation BL =
1160              PathDiagnosticLocation::createEndBrace(CS, SM);
1161            EB.addEdge(BL);
1162          }
1163        }
1164
1165        if (Term)
1166          EB.addContext(Term);
1167
1168        break;
1169      }
1170
1171      if (const BlockEntrance *BE = dyn_cast<BlockEntrance>(&P)) {
1172        if (const CFGStmt *S = BE->getFirstElement().getAs<CFGStmt>()) {
1173          const Stmt *stmt = S->getStmt();
1174          if (IsControlFlowExpr(stmt)) {
1175            // Add the proper context for '&&', '||', and '?'.
1176            EB.addContext(stmt);
1177          }
1178          else
1179            EB.addExtendedContext(PDB.getEnclosingStmtLocation(stmt).asStmt());
1180        }
1181
1182        break;
1183      }
1184    } while (0);
1185
1186    if (!NextNode)
1187      continue;
1188
1189    // Add pieces from custom visitors.
1190    BugReport *R = PDB.getBugReport();
1191    for (BugReport::visitor_iterator I = R->visitor_begin(),
1192                                     E = R->visitor_end(); I!=E; ++I) {
1193      if (PathDiagnosticPiece *p = (*I)->VisitNode(N, NextNode, PDB, *R)) {
1194        const PathDiagnosticLocation &Loc = p->getLocation();
1195        EB.addEdge(Loc, true);
1196        PD.push_front(p);
1197        if (const Stmt *S = Loc.asStmt())
1198          EB.addExtendedContext(PDB.getEnclosingStmtLocation(S).asStmt());
1199      }
1200    }
1201  }
1202}
1203
1204//===----------------------------------------------------------------------===//
1205// Methods for BugType and subclasses.
1206//===----------------------------------------------------------------------===//
1207BugType::~BugType() { }
1208
1209void BugType::FlushReports(BugReporter &BR) {}
1210
1211void BuiltinBug::anchor() {}
1212
1213//===----------------------------------------------------------------------===//
1214// Methods for BugReport and subclasses.
1215//===----------------------------------------------------------------------===//
1216
1217void BugReport::NodeResolver::anchor() {}
1218
1219void BugReport::addVisitor(BugReporterVisitor* visitor) {
1220  if (!visitor)
1221    return;
1222
1223  llvm::FoldingSetNodeID ID;
1224  visitor->Profile(ID);
1225  void *InsertPos;
1226
1227  if (CallbacksSet.FindNodeOrInsertPos(ID, InsertPos)) {
1228    delete visitor;
1229    return;
1230  }
1231
1232  CallbacksSet.InsertNode(visitor, InsertPos);
1233  Callbacks = F.add(visitor, Callbacks);
1234}
1235
1236BugReport::~BugReport() {
1237  for (visitor_iterator I = visitor_begin(), E = visitor_end(); I != E; ++I) {
1238    delete *I;
1239  }
1240}
1241
1242void BugReport::Profile(llvm::FoldingSetNodeID& hash) const {
1243  hash.AddPointer(&BT);
1244  hash.AddString(Description);
1245  if (Location.isValid()) {
1246    Location.Profile(hash);
1247  } else {
1248    assert(ErrorNode);
1249    hash.AddPointer(GetCurrentOrPreviousStmt(ErrorNode));
1250  }
1251
1252  for (SmallVectorImpl<SourceRange>::const_iterator I =
1253      Ranges.begin(), E = Ranges.end(); I != E; ++I) {
1254    const SourceRange range = *I;
1255    if (!range.isValid())
1256      continue;
1257    hash.AddInteger(range.getBegin().getRawEncoding());
1258    hash.AddInteger(range.getEnd().getRawEncoding());
1259  }
1260}
1261
1262const Stmt *BugReport::getStmt() const {
1263  if (!ErrorNode)
1264    return 0;
1265
1266  ProgramPoint ProgP = ErrorNode->getLocation();
1267  const Stmt *S = NULL;
1268
1269  if (BlockEntrance *BE = dyn_cast<BlockEntrance>(&ProgP)) {
1270    CFGBlock &Exit = ProgP.getLocationContext()->getCFG()->getExit();
1271    if (BE->getBlock() == &Exit)
1272      S = GetPreviousStmt(ErrorNode);
1273  }
1274  if (!S)
1275    S = GetStmt(ProgP);
1276
1277  return S;
1278}
1279
1280std::pair<BugReport::ranges_iterator, BugReport::ranges_iterator>
1281BugReport::getRanges() {
1282    // If no custom ranges, add the range of the statement corresponding to
1283    // the error node.
1284    if (Ranges.empty()) {
1285      if (const Expr *E = dyn_cast_or_null<Expr>(getStmt()))
1286        addRange(E->getSourceRange());
1287      else
1288        return std::make_pair(ranges_iterator(), ranges_iterator());
1289    }
1290
1291    // User-specified absence of range info.
1292    if (Ranges.size() == 1 && !Ranges.begin()->isValid())
1293      return std::make_pair(ranges_iterator(), ranges_iterator());
1294
1295    return std::make_pair(Ranges.begin(), Ranges.end());
1296}
1297
1298PathDiagnosticLocation BugReport::getLocation(const SourceManager &SM) const {
1299  if (ErrorNode) {
1300    assert(!Location.isValid() &&
1301     "Either Location or ErrorNode should be specified but not both.");
1302
1303    if (const Stmt *S = GetCurrentOrPreviousStmt(ErrorNode)) {
1304      const LocationContext *LC = ErrorNode->getLocationContext();
1305
1306      // For member expressions, return the location of the '.' or '->'.
1307      if (const MemberExpr *ME = dyn_cast<MemberExpr>(S))
1308        return PathDiagnosticLocation::createMemberLoc(ME, SM);
1309      // For binary operators, return the location of the operator.
1310      if (const BinaryOperator *B = dyn_cast<BinaryOperator>(S))
1311        return PathDiagnosticLocation::createOperatorLoc(B, SM);
1312
1313      return PathDiagnosticLocation::createBegin(S, SM, LC);
1314    }
1315  } else {
1316    assert(Location.isValid());
1317    return Location;
1318  }
1319
1320  return PathDiagnosticLocation();
1321}
1322
1323//===----------------------------------------------------------------------===//
1324// Methods for BugReporter and subclasses.
1325//===----------------------------------------------------------------------===//
1326
1327BugReportEquivClass::~BugReportEquivClass() {
1328  for (iterator I=begin(), E=end(); I!=E; ++I) delete *I;
1329}
1330
1331GRBugReporter::~GRBugReporter() { }
1332BugReporterData::~BugReporterData() {}
1333
1334ExplodedGraph &GRBugReporter::getGraph() { return Eng.getGraph(); }
1335
1336ProgramStateManager&
1337GRBugReporter::getStateManager() { return Eng.getStateManager(); }
1338
1339BugReporter::~BugReporter() {
1340  FlushReports();
1341
1342  // Free the bug reports we are tracking.
1343  typedef std::vector<BugReportEquivClass *> ContTy;
1344  for (ContTy::iterator I = EQClassesVector.begin(), E = EQClassesVector.end();
1345       I != E; ++I) {
1346    delete *I;
1347  }
1348}
1349
1350void BugReporter::FlushReports() {
1351  if (BugTypes.isEmpty())
1352    return;
1353
1354  // First flush the warnings for each BugType.  This may end up creating new
1355  // warnings and new BugTypes.
1356  // FIXME: Only NSErrorChecker needs BugType's FlushReports.
1357  // Turn NSErrorChecker into a proper checker and remove this.
1358  SmallVector<const BugType*, 16> bugTypes;
1359  for (BugTypesTy::iterator I=BugTypes.begin(), E=BugTypes.end(); I!=E; ++I)
1360    bugTypes.push_back(*I);
1361  for (SmallVector<const BugType*, 16>::iterator
1362         I = bugTypes.begin(), E = bugTypes.end(); I != E; ++I)
1363    const_cast<BugType*>(*I)->FlushReports(*this);
1364
1365  typedef llvm::FoldingSet<BugReportEquivClass> SetTy;
1366  for (SetTy::iterator EI=EQClasses.begin(), EE=EQClasses.end(); EI!=EE;++EI){
1367    BugReportEquivClass& EQ = *EI;
1368    FlushReport(EQ);
1369  }
1370
1371  // BugReporter owns and deletes only BugTypes created implicitly through
1372  // EmitBasicReport.
1373  // FIXME: There are leaks from checkers that assume that the BugTypes they
1374  // create will be destroyed by the BugReporter.
1375  for (llvm::StringMap<BugType*>::iterator
1376         I = StrBugTypes.begin(), E = StrBugTypes.end(); I != E; ++I)
1377    delete I->second;
1378
1379  // Remove all references to the BugType objects.
1380  BugTypes = F.getEmptySet();
1381}
1382
1383//===----------------------------------------------------------------------===//
1384// PathDiagnostics generation.
1385//===----------------------------------------------------------------------===//
1386
1387static std::pair<std::pair<ExplodedGraph*, NodeBackMap*>,
1388                 std::pair<ExplodedNode*, unsigned> >
1389MakeReportGraph(const ExplodedGraph* G,
1390                SmallVectorImpl<const ExplodedNode*> &nodes) {
1391
1392  // Create the trimmed graph.  It will contain the shortest paths from the
1393  // error nodes to the root.  In the new graph we should only have one
1394  // error node unless there are two or more error nodes with the same minimum
1395  // path length.
1396  ExplodedGraph* GTrim;
1397  InterExplodedGraphMap* NMap;
1398
1399  llvm::DenseMap<const void*, const void*> InverseMap;
1400  llvm::tie(GTrim, NMap) = G->Trim(nodes.data(), nodes.data() + nodes.size(),
1401                                   &InverseMap);
1402
1403  // Create owning pointers for GTrim and NMap just to ensure that they are
1404  // released when this function exists.
1405  llvm::OwningPtr<ExplodedGraph> AutoReleaseGTrim(GTrim);
1406  llvm::OwningPtr<InterExplodedGraphMap> AutoReleaseNMap(NMap);
1407
1408  // Find the (first) error node in the trimmed graph.  We just need to consult
1409  // the node map (NMap) which maps from nodes in the original graph to nodes
1410  // in the new graph.
1411
1412  std::queue<const ExplodedNode*> WS;
1413  typedef llvm::DenseMap<const ExplodedNode*, unsigned> IndexMapTy;
1414  IndexMapTy IndexMap;
1415
1416  for (unsigned nodeIndex = 0 ; nodeIndex < nodes.size(); ++nodeIndex) {
1417    const ExplodedNode *originalNode = nodes[nodeIndex];
1418    if (const ExplodedNode *N = NMap->getMappedNode(originalNode)) {
1419      WS.push(N);
1420      IndexMap[originalNode] = nodeIndex;
1421    }
1422  }
1423
1424  assert(!WS.empty() && "No error node found in the trimmed graph.");
1425
1426  // Create a new (third!) graph with a single path.  This is the graph
1427  // that will be returned to the caller.
1428  ExplodedGraph *GNew = new ExplodedGraph();
1429
1430  // Sometimes the trimmed graph can contain a cycle.  Perform a reverse BFS
1431  // to the root node, and then construct a new graph that contains only
1432  // a single path.
1433  llvm::DenseMap<const void*,unsigned> Visited;
1434
1435  unsigned cnt = 0;
1436  const ExplodedNode *Root = 0;
1437
1438  while (!WS.empty()) {
1439    const ExplodedNode *Node = WS.front();
1440    WS.pop();
1441
1442    if (Visited.find(Node) != Visited.end())
1443      continue;
1444
1445    Visited[Node] = cnt++;
1446
1447    if (Node->pred_empty()) {
1448      Root = Node;
1449      break;
1450    }
1451
1452    for (ExplodedNode::const_pred_iterator I=Node->pred_begin(),
1453         E=Node->pred_end(); I!=E; ++I)
1454      WS.push(*I);
1455  }
1456
1457  assert(Root);
1458
1459  // Now walk from the root down the BFS path, always taking the successor
1460  // with the lowest number.
1461  ExplodedNode *Last = 0, *First = 0;
1462  NodeBackMap *BM = new NodeBackMap();
1463  unsigned NodeIndex = 0;
1464
1465  for ( const ExplodedNode *N = Root ;;) {
1466    // Lookup the number associated with the current node.
1467    llvm::DenseMap<const void*,unsigned>::iterator I = Visited.find(N);
1468    assert(I != Visited.end());
1469
1470    // Create the equivalent node in the new graph with the same state
1471    // and location.
1472    ExplodedNode *NewN = GNew->getNode(N->getLocation(), N->getState());
1473
1474    // Store the mapping to the original node.
1475    llvm::DenseMap<const void*, const void*>::iterator IMitr=InverseMap.find(N);
1476    assert(IMitr != InverseMap.end() && "No mapping to original node.");
1477    (*BM)[NewN] = (const ExplodedNode*) IMitr->second;
1478
1479    // Link up the new node with the previous node.
1480    if (Last)
1481      NewN->addPredecessor(Last, *GNew);
1482
1483    Last = NewN;
1484
1485    // Are we at the final node?
1486    IndexMapTy::iterator IMI =
1487      IndexMap.find((const ExplodedNode*)(IMitr->second));
1488    if (IMI != IndexMap.end()) {
1489      First = NewN;
1490      NodeIndex = IMI->second;
1491      break;
1492    }
1493
1494    // Find the next successor node.  We choose the node that is marked
1495    // with the lowest DFS number.
1496    ExplodedNode::const_succ_iterator SI = N->succ_begin();
1497    ExplodedNode::const_succ_iterator SE = N->succ_end();
1498    N = 0;
1499
1500    for (unsigned MinVal = 0; SI != SE; ++SI) {
1501
1502      I = Visited.find(*SI);
1503
1504      if (I == Visited.end())
1505        continue;
1506
1507      if (!N || I->second < MinVal) {
1508        N = *SI;
1509        MinVal = I->second;
1510      }
1511    }
1512
1513    assert(N);
1514  }
1515
1516  assert(First);
1517
1518  return std::make_pair(std::make_pair(GNew, BM),
1519                        std::make_pair(First, NodeIndex));
1520}
1521
1522/// CompactPathDiagnostic - This function postprocesses a PathDiagnostic object
1523///  and collapses PathDiagosticPieces that are expanded by macros.
1524static void CompactPathDiagnostic(PathDiagnostic &PD, const SourceManager& SM) {
1525  typedef std::vector<std::pair<PathDiagnosticMacroPiece*, SourceLocation> >
1526          MacroStackTy;
1527
1528  typedef std::vector<PathDiagnosticPiece*>
1529          PiecesTy;
1530
1531  MacroStackTy MacroStack;
1532  PiecesTy Pieces;
1533
1534  for (PathDiagnostic::iterator I = PD.begin(), E = PD.end(); I!=E; ++I) {
1535    // Get the location of the PathDiagnosticPiece.
1536    const FullSourceLoc Loc = I->getLocation().asLocation();
1537
1538    // Determine the instantiation location, which is the location we group
1539    // related PathDiagnosticPieces.
1540    SourceLocation InstantiationLoc = Loc.isMacroID() ?
1541                                      SM.getExpansionLoc(Loc) :
1542                                      SourceLocation();
1543
1544    if (Loc.isFileID()) {
1545      MacroStack.clear();
1546      Pieces.push_back(&*I);
1547      continue;
1548    }
1549
1550    assert(Loc.isMacroID());
1551
1552    // Is the PathDiagnosticPiece within the same macro group?
1553    if (!MacroStack.empty() && InstantiationLoc == MacroStack.back().second) {
1554      MacroStack.back().first->push_back(&*I);
1555      continue;
1556    }
1557
1558    // We aren't in the same group.  Are we descending into a new macro
1559    // or are part of an old one?
1560    PathDiagnosticMacroPiece *MacroGroup = 0;
1561
1562    SourceLocation ParentInstantiationLoc = InstantiationLoc.isMacroID() ?
1563                                          SM.getExpansionLoc(Loc) :
1564                                          SourceLocation();
1565
1566    // Walk the entire macro stack.
1567    while (!MacroStack.empty()) {
1568      if (InstantiationLoc == MacroStack.back().second) {
1569        MacroGroup = MacroStack.back().first;
1570        break;
1571      }
1572
1573      if (ParentInstantiationLoc == MacroStack.back().second) {
1574        MacroGroup = MacroStack.back().first;
1575        break;
1576      }
1577
1578      MacroStack.pop_back();
1579    }
1580
1581    if (!MacroGroup || ParentInstantiationLoc == MacroStack.back().second) {
1582      // Create a new macro group and add it to the stack.
1583      PathDiagnosticMacroPiece *NewGroup =
1584        new PathDiagnosticMacroPiece(
1585          PathDiagnosticLocation::createSingleLocation(I->getLocation()));
1586
1587      if (MacroGroup)
1588        MacroGroup->push_back(NewGroup);
1589      else {
1590        assert(InstantiationLoc.isFileID());
1591        Pieces.push_back(NewGroup);
1592      }
1593
1594      MacroGroup = NewGroup;
1595      MacroStack.push_back(std::make_pair(MacroGroup, InstantiationLoc));
1596    }
1597
1598    // Finally, add the PathDiagnosticPiece to the group.
1599    MacroGroup->push_back(&*I);
1600  }
1601
1602  // Now take the pieces and construct a new PathDiagnostic.
1603  PD.resetPath(false);
1604
1605  for (PiecesTy::iterator I=Pieces.begin(), E=Pieces.end(); I!=E; ++I) {
1606    if (PathDiagnosticMacroPiece *MP=dyn_cast<PathDiagnosticMacroPiece>(*I))
1607      if (!MP->containsEvent()) {
1608        delete MP;
1609        continue;
1610      }
1611
1612    PD.push_back(*I);
1613  }
1614}
1615
1616void GRBugReporter::GeneratePathDiagnostic(PathDiagnostic& PD,
1617                        SmallVectorImpl<BugReport *> &bugReports) {
1618
1619  assert(!bugReports.empty());
1620  SmallVector<const ExplodedNode *, 10> errorNodes;
1621  for (SmallVectorImpl<BugReport*>::iterator I = bugReports.begin(),
1622    E = bugReports.end(); I != E; ++I) {
1623      errorNodes.push_back((*I)->getErrorNode());
1624  }
1625
1626  // Construct a new graph that contains only a single path from the error
1627  // node to a root.
1628  const std::pair<std::pair<ExplodedGraph*, NodeBackMap*>,
1629  std::pair<ExplodedNode*, unsigned> >&
1630    GPair = MakeReportGraph(&getGraph(), errorNodes);
1631
1632  // Find the BugReport with the original location.
1633  assert(GPair.second.second < bugReports.size());
1634  BugReport *R = bugReports[GPair.second.second];
1635  assert(R && "No original report found for sliced graph.");
1636
1637  llvm::OwningPtr<ExplodedGraph> ReportGraph(GPair.first.first);
1638  llvm::OwningPtr<NodeBackMap> BackMap(GPair.first.second);
1639  const ExplodedNode *N = GPair.second.first;
1640
1641  // Start building the path diagnostic...
1642  PathDiagnosticBuilder PDB(*this, R, BackMap.get(),
1643                            getPathDiagnosticConsumer());
1644
1645  // Register additional node visitors.
1646  R->addVisitor(new NilReceiverBRVisitor());
1647  R->addVisitor(new ConditionBRVisitor());
1648
1649  // Generate the very last diagnostic piece - the piece is visible before
1650  // the trace is expanded.
1651  PathDiagnosticPiece *LastPiece = 0;
1652  for (BugReport::visitor_iterator I = R->visitor_begin(),
1653                                   E = R->visitor_end(); I!=E; ++I) {
1654    if (PathDiagnosticPiece *Piece = (*I)->getEndPath(PDB, N, *R)) {
1655      assert (!LastPiece &&
1656              "There can only be one final piece in a diagnostic.");
1657      LastPiece = Piece;
1658    }
1659  }
1660  if (!LastPiece)
1661    LastPiece = BugReporterVisitor::getDefaultEndPath(PDB, N, *R);
1662  if (LastPiece)
1663    PD.push_back(LastPiece);
1664  else
1665    return;
1666
1667  switch (PDB.getGenerationScheme()) {
1668    case PathDiagnosticConsumer::Extensive:
1669      GenerateExtensivePathDiagnostic(PD, PDB, N);
1670      break;
1671    case PathDiagnosticConsumer::Minimal:
1672      GenerateMinimalPathDiagnostic(PD, PDB, N);
1673      break;
1674  }
1675}
1676
1677void BugReporter::Register(BugType *BT) {
1678  BugTypes = F.add(BugTypes, BT);
1679}
1680
1681void BugReporter::EmitReport(BugReport* R) {
1682  // Compute the bug report's hash to determine its equivalence class.
1683  llvm::FoldingSetNodeID ID;
1684  R->Profile(ID);
1685
1686  // Lookup the equivance class.  If there isn't one, create it.
1687  BugType& BT = R->getBugType();
1688  Register(&BT);
1689  void *InsertPos;
1690  BugReportEquivClass* EQ = EQClasses.FindNodeOrInsertPos(ID, InsertPos);
1691
1692  if (!EQ) {
1693    EQ = new BugReportEquivClass(R);
1694    EQClasses.InsertNode(EQ, InsertPos);
1695    EQClassesVector.push_back(EQ);
1696  }
1697  else
1698    EQ->AddReport(R);
1699}
1700
1701
1702//===----------------------------------------------------------------------===//
1703// Emitting reports in equivalence classes.
1704//===----------------------------------------------------------------------===//
1705
1706namespace {
1707struct FRIEC_WLItem {
1708  const ExplodedNode *N;
1709  ExplodedNode::const_succ_iterator I, E;
1710
1711  FRIEC_WLItem(const ExplodedNode *n)
1712  : N(n), I(N->succ_begin()), E(N->succ_end()) {}
1713};
1714}
1715
1716static BugReport *
1717FindReportInEquivalenceClass(BugReportEquivClass& EQ,
1718                             SmallVectorImpl<BugReport*> &bugReports) {
1719
1720  BugReportEquivClass::iterator I = EQ.begin(), E = EQ.end();
1721  assert(I != E);
1722  BugReport *R = *I;
1723  BugType& BT = R->getBugType();
1724
1725  // If we don't need to suppress any of the nodes because they are
1726  // post-dominated by a sink, simply add all the nodes in the equivalence class
1727  // to 'Nodes'.  Any of the reports will serve as a "representative" report.
1728  if (!BT.isSuppressOnSink()) {
1729    for (BugReportEquivClass::iterator I=EQ.begin(), E=EQ.end(); I!=E; ++I) {
1730      const ExplodedNode *N = I->getErrorNode();
1731      if (N) {
1732        R = *I;
1733        bugReports.push_back(R);
1734      }
1735    }
1736    return R;
1737  }
1738
1739  // For bug reports that should be suppressed when all paths are post-dominated
1740  // by a sink node, iterate through the reports in the equivalence class
1741  // until we find one that isn't post-dominated (if one exists).  We use a
1742  // DFS traversal of the ExplodedGraph to find a non-sink node.  We could write
1743  // this as a recursive function, but we don't want to risk blowing out the
1744  // stack for very long paths.
1745  BugReport *exampleReport = 0;
1746
1747  for (; I != E; ++I) {
1748    R = *I;
1749    const ExplodedNode *errorNode = R->getErrorNode();
1750
1751    if (!errorNode)
1752      continue;
1753    if (errorNode->isSink()) {
1754      llvm_unreachable(
1755           "BugType::isSuppressSink() should not be 'true' for sink end nodes");
1756    }
1757    // No successors?  By definition this nodes isn't post-dominated by a sink.
1758    if (errorNode->succ_empty()) {
1759      bugReports.push_back(R);
1760      if (!exampleReport)
1761        exampleReport = R;
1762      continue;
1763    }
1764
1765    // At this point we know that 'N' is not a sink and it has at least one
1766    // successor.  Use a DFS worklist to find a non-sink end-of-path node.
1767    typedef FRIEC_WLItem WLItem;
1768    typedef SmallVector<WLItem, 10> DFSWorkList;
1769    llvm::DenseMap<const ExplodedNode *, unsigned> Visited;
1770
1771    DFSWorkList WL;
1772    WL.push_back(errorNode);
1773    Visited[errorNode] = 1;
1774
1775    while (!WL.empty()) {
1776      WLItem &WI = WL.back();
1777      assert(!WI.N->succ_empty());
1778
1779      for (; WI.I != WI.E; ++WI.I) {
1780        const ExplodedNode *Succ = *WI.I;
1781        // End-of-path node?
1782        if (Succ->succ_empty()) {
1783          // If we found an end-of-path node that is not a sink.
1784          if (!Succ->isSink()) {
1785            bugReports.push_back(R);
1786            if (!exampleReport)
1787              exampleReport = R;
1788            WL.clear();
1789            break;
1790          }
1791          // Found a sink?  Continue on to the next successor.
1792          continue;
1793        }
1794        // Mark the successor as visited.  If it hasn't been explored,
1795        // enqueue it to the DFS worklist.
1796        unsigned &mark = Visited[Succ];
1797        if (!mark) {
1798          mark = 1;
1799          WL.push_back(Succ);
1800          break;
1801        }
1802      }
1803
1804      // The worklist may have been cleared at this point.  First
1805      // check if it is empty before checking the last item.
1806      if (!WL.empty() && &WL.back() == &WI)
1807        WL.pop_back();
1808    }
1809  }
1810
1811  // ExampleReport will be NULL if all the nodes in the equivalence class
1812  // were post-dominated by sinks.
1813  return exampleReport;
1814}
1815
1816//===----------------------------------------------------------------------===//
1817// DiagnosticCache.  This is a hack to cache analyzer diagnostics.  It
1818// uses global state, which eventually should go elsewhere.
1819//===----------------------------------------------------------------------===//
1820namespace {
1821class DiagCacheItem : public llvm::FoldingSetNode {
1822  llvm::FoldingSetNodeID ID;
1823public:
1824  DiagCacheItem(BugReport *R, PathDiagnostic *PD) {
1825    R->Profile(ID);
1826    PD->Profile(ID);
1827  }
1828
1829  void Profile(llvm::FoldingSetNodeID &id) {
1830    id = ID;
1831  }
1832
1833  llvm::FoldingSetNodeID &getID() { return ID; }
1834};
1835}
1836
1837static bool IsCachedDiagnostic(BugReport *R, PathDiagnostic *PD) {
1838  // FIXME: Eventually this diagnostic cache should reside in something
1839  // like AnalysisManager instead of being a static variable.  This is
1840  // really unsafe in the long term.
1841  typedef llvm::FoldingSet<DiagCacheItem> DiagnosticCache;
1842  static DiagnosticCache DC;
1843
1844  void *InsertPos;
1845  DiagCacheItem *Item = new DiagCacheItem(R, PD);
1846
1847  if (DC.FindNodeOrInsertPos(Item->getID(), InsertPos)) {
1848    delete Item;
1849    return true;
1850  }
1851
1852  DC.InsertNode(Item, InsertPos);
1853  return false;
1854}
1855
1856void BugReporter::FlushReport(BugReportEquivClass& EQ) {
1857  SmallVector<BugReport*, 10> bugReports;
1858  BugReport *exampleReport = FindReportInEquivalenceClass(EQ, bugReports);
1859  if (!exampleReport)
1860    return;
1861
1862  PathDiagnosticConsumer* PD = getPathDiagnosticConsumer();
1863
1864  // FIXME: Make sure we use the 'R' for the path that was actually used.
1865  // Probably doesn't make a difference in practice.
1866  BugType& BT = exampleReport->getBugType();
1867
1868  llvm::OwningPtr<PathDiagnostic>
1869    D(new PathDiagnostic(exampleReport->getBugType().getName(),
1870                         !PD || PD->useVerboseDescription()
1871                         ? exampleReport->getDescription()
1872                         : exampleReport->getShortDescription(),
1873                         BT.getCategory()));
1874
1875  if (!bugReports.empty())
1876    GeneratePathDiagnostic(*D.get(), bugReports);
1877
1878  if (IsCachedDiagnostic(exampleReport, D.get()))
1879    return;
1880
1881  // Get the meta data.
1882  const BugReport::ExtraTextList &Meta =
1883                                  exampleReport->getExtraText();
1884  for (BugReport::ExtraTextList::const_iterator i = Meta.begin(),
1885                                                e = Meta.end(); i != e; ++i) {
1886    D->addMeta(*i);
1887  }
1888
1889  // Emit a summary diagnostic to the regular Diagnostics engine.
1890  BugReport::ranges_iterator Beg, End;
1891  llvm::tie(Beg, End) = exampleReport->getRanges();
1892  DiagnosticsEngine &Diag = getDiagnostic();
1893
1894  // Search the description for '%', as that will be interpretted as a
1895  // format character by FormatDiagnostics.
1896  StringRef desc = exampleReport->getShortDescription();
1897  unsigned ErrorDiag;
1898  {
1899    llvm::SmallString<512> TmpStr;
1900    llvm::raw_svector_ostream Out(TmpStr);
1901    for (StringRef::iterator I=desc.begin(), E=desc.end(); I!=E; ++I)
1902      if (*I == '%')
1903        Out << "%%";
1904      else
1905        Out << *I;
1906
1907    Out.flush();
1908    ErrorDiag = Diag.getCustomDiagID(DiagnosticsEngine::Warning, TmpStr);
1909  }
1910
1911  {
1912    DiagnosticBuilder diagBuilder = Diag.Report(
1913      exampleReport->getLocation(getSourceManager()).asLocation(), ErrorDiag);
1914    for (BugReport::ranges_iterator I = Beg; I != End; ++I)
1915      diagBuilder << *I;
1916  }
1917
1918  // Emit a full diagnostic for the path if we have a PathDiagnosticConsumer.
1919  if (!PD)
1920    return;
1921
1922  if (D->empty()) {
1923    PathDiagnosticPiece *piece = new PathDiagnosticEventPiece(
1924                                 exampleReport->getLocation(getSourceManager()),
1925                                 exampleReport->getDescription());
1926
1927    for ( ; Beg != End; ++Beg) piece->addRange(*Beg);
1928    D->push_back(piece);
1929  }
1930
1931  PD->HandlePathDiagnostic(D.take());
1932}
1933
1934void BugReporter::EmitBasicReport(StringRef name, StringRef str,
1935                                  PathDiagnosticLocation Loc,
1936                                  SourceRange* RBeg, unsigned NumRanges) {
1937  EmitBasicReport(name, "", str, Loc, RBeg, NumRanges);
1938}
1939
1940void BugReporter::EmitBasicReport(StringRef name,
1941                                  StringRef category,
1942                                  StringRef str, PathDiagnosticLocation Loc,
1943                                  SourceRange* RBeg, unsigned NumRanges) {
1944
1945  // 'BT' is owned by BugReporter.
1946  BugType *BT = getBugTypeForName(name, category);
1947  BugReport *R = new BugReport(*BT, str, Loc);
1948  for ( ; NumRanges > 0 ; --NumRanges, ++RBeg) R->addRange(*RBeg);
1949  EmitReport(R);
1950}
1951
1952BugType *BugReporter::getBugTypeForName(StringRef name,
1953                                        StringRef category) {
1954  llvm::SmallString<136> fullDesc;
1955  llvm::raw_svector_ostream(fullDesc) << name << ":" << category;
1956  llvm::StringMapEntry<BugType *> &
1957      entry = StrBugTypes.GetOrCreateValue(fullDesc);
1958  BugType *BT = entry.getValue();
1959  if (!BT) {
1960    BT = new BugType(name, category);
1961    entry.setValue(BT);
1962  }
1963  return BT;
1964}
1965