MallocChecker.cpp revision e17fdb2d5dbf0ffefd417587003eebbe5baf5984 
1//=== MallocChecker.cpp - A malloc/free checker -------------------*- 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 malloc/free checker, which checks for potential memory
11// leaks, double free, and use-after-free problems.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "InterCheckerAPI.h"
17#include "clang/StaticAnalyzer/Core/Checker.h"
18#include "clang/StaticAnalyzer/Core/CheckerManager.h"
19#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
20#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/ObjCMessage.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
24#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
25#include "clang/Basic/SourceManager.h"
26#include "llvm/ADT/ImmutableMap.h"
27#include "llvm/ADT/SmallString.h"
28#include "llvm/ADT/STLExtras.h"
29#include <climits>
30
31using namespace clang;
32using namespace ento;
33
34namespace {
35
36class RefState {
37  enum Kind { AllocateUnchecked, AllocateFailed, Released, Escaped,
38              Relinquished } K;
39  const Stmt *S;
40
41public:
42  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
43
44  bool isAllocated() const { return K == AllocateUnchecked; }
45  bool isReleased() const { return K == Released; }
46
47  const Stmt *getStmt() const { return S; }
48
49  bool operator==(const RefState &X) const {
50    return K == X.K && S == X.S;
51  }
52
53  static RefState getAllocateUnchecked(const Stmt *s) {
54    return RefState(AllocateUnchecked, s);
55  }
56  static RefState getAllocateFailed() {
57    return RefState(AllocateFailed, 0);
58  }
59  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
60  static RefState getEscaped(const Stmt *s) { return RefState(Escaped, s); }
61  static RefState getRelinquished(const Stmt *s) {
62    return RefState(Relinquished, s);
63  }
64
65  void Profile(llvm::FoldingSetNodeID &ID) const {
66    ID.AddInteger(K);
67    ID.AddPointer(S);
68  }
69};
70
71struct ReallocPair {
72  SymbolRef ReallocatedSym;
73  bool IsFreeOnFailure;
74  ReallocPair(SymbolRef S, bool F) : ReallocatedSym(S), IsFreeOnFailure(F) {}
75  void Profile(llvm::FoldingSetNodeID &ID) const {
76    ID.AddInteger(IsFreeOnFailure);
77    ID.AddPointer(ReallocatedSym);
78  }
79  bool operator==(const ReallocPair &X) const {
80    return ReallocatedSym == X.ReallocatedSym &&
81           IsFreeOnFailure == X.IsFreeOnFailure;
82  }
83};
84
85typedef std::pair<const Stmt*, const MemRegion*> LeakInfo;
86
87class MallocChecker : public Checker<check::DeadSymbols,
88                                     check::EndPath,
89                                     check::PreStmt<ReturnStmt>,
90                                     check::PreStmt<CallExpr>,
91                                     check::PostStmt<CallExpr>,
92                                     check::PostStmt<BlockExpr>,
93                                     check::Location,
94                                     check::Bind,
95                                     eval::Assume,
96                                     check::RegionChanges>
97{
98  mutable OwningPtr<BugType> BT_DoubleFree;
99  mutable OwningPtr<BugType> BT_Leak;
100  mutable OwningPtr<BugType> BT_UseFree;
101  mutable OwningPtr<BugType> BT_BadFree;
102  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
103                         *II_valloc, *II_reallocf, *II_strndup, *II_strdup;
104
105public:
106  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
107                    II_valloc(0), II_reallocf(0), II_strndup(0), II_strdup(0) {}
108
109  /// In pessimistic mode, the checker assumes that it does not know which
110  /// functions might free the memory.
111  struct ChecksFilter {
112    DefaultBool CMallocPessimistic;
113    DefaultBool CMallocOptimistic;
114  };
115
116  ChecksFilter Filter;
117
118  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
119  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
120  void checkPostStmt(const BlockExpr *BE, CheckerContext &C) const;
121  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
122  void checkEndPath(CheckerContext &C) const;
123  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
124  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
125                            bool Assumption) const;
126  void checkLocation(SVal l, bool isLoad, const Stmt *S,
127                     CheckerContext &C) const;
128  void checkBind(SVal location, SVal val, const Stmt*S,
129                 CheckerContext &C) const;
130  ProgramStateRef
131  checkRegionChanges(ProgramStateRef state,
132                     const StoreManager::InvalidatedSymbols *invalidated,
133                     ArrayRef<const MemRegion *> ExplicitRegions,
134                     ArrayRef<const MemRegion *> Regions,
135                     const CallOrObjCMessage *Call) const;
136  bool wantsRegionChangeUpdate(ProgramStateRef state) const {
137    return true;
138  }
139
140  void printState(raw_ostream &Out, ProgramStateRef State,
141                  const char *NL, const char *Sep) const;
142
143private:
144  void initIdentifierInfo(ASTContext &C) const;
145
146  /// Check if this is one of the functions which can allocate/reallocate memory
147  /// pointed to by one of its arguments.
148  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
149  bool isFreeFunction(const FunctionDecl *FD, ASTContext &C) const;
150  bool isAllocationFunction(const FunctionDecl *FD, ASTContext &C) const;
151
152  static ProgramStateRef MallocMemReturnsAttr(CheckerContext &C,
153                                              const CallExpr *CE,
154                                              const OwnershipAttr* Att);
155  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
156                                     const Expr *SizeEx, SVal Init,
157                                     ProgramStateRef state) {
158    return MallocMemAux(C, CE,
159                        state->getSVal(SizeEx, C.getLocationContext()),
160                        Init, state);
161  }
162
163  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
164                                     SVal SizeEx, SVal Init,
165                                     ProgramStateRef state);
166
167  /// Update the RefState to reflect the new memory allocation.
168  static ProgramStateRef MallocUpdateRefState(CheckerContext &C,
169                                              const CallExpr *CE,
170                                              ProgramStateRef state);
171
172  ProgramStateRef FreeMemAttr(CheckerContext &C, const CallExpr *CE,
173                              const OwnershipAttr* Att) const;
174  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
175                                 ProgramStateRef state, unsigned Num,
176                                 bool Hold) const;
177
178  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
179                             bool FreesMemOnFailure) const;
180  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
181
182  ///\brief Check if the memory associated with this symbol was released.
183  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
184
185  bool checkEscape(SymbolRef Sym, const Stmt *S, CheckerContext &C) const;
186  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
187                         const Stmt *S = 0) const;
188
189  /// Check if the function is not known to us. So, for example, we could
190  /// conservatively assume it can free/reallocate it's pointer arguments.
191  bool doesNotFreeMemory(const CallOrObjCMessage *Call,
192                         ProgramStateRef State) const;
193
194  static bool SummarizeValue(raw_ostream &os, SVal V);
195  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
196  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
197
198  /// Find the location of the allocation for Sym on the path leading to the
199  /// exploded node N.
200  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
201                             CheckerContext &C) const;
202
203  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
204
205  /// The bug visitor which allows us to print extra diagnostics along the
206  /// BugReport path. For example, showing the allocation site of the leaked
207  /// region.
208  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
209  protected:
210    enum NotificationMode {
211      Normal,
212      ReallocationFailed
213    };
214
215    // The allocated region symbol tracked by the main analysis.
216    SymbolRef Sym;
217
218    // The mode we are in, i.e. what kind of diagnostics will be emitted.
219    NotificationMode Mode;
220
221    // A symbol from when the primary region should have been reallocated.
222    SymbolRef FailedReallocSymbol;
223
224    bool IsLeak;
225
226  public:
227    MallocBugVisitor(SymbolRef S, bool isLeak = false)
228       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
229
230    virtual ~MallocBugVisitor() {}
231
232    void Profile(llvm::FoldingSetNodeID &ID) const {
233      static int X = 0;
234      ID.AddPointer(&X);
235      ID.AddPointer(Sym);
236    }
237
238    inline bool isAllocated(const RefState *S, const RefState *SPrev,
239                            const Stmt *Stmt) {
240      // Did not track -> allocated. Other state (released) -> allocated.
241      return (Stmt && isa<CallExpr>(Stmt) &&
242              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
243    }
244
245    inline bool isReleased(const RefState *S, const RefState *SPrev,
246                           const Stmt *Stmt) {
247      // Did not track -> released. Other state (allocated) -> released.
248      return (Stmt && isa<CallExpr>(Stmt) &&
249              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
250    }
251
252    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
253                                     const Stmt *Stmt) {
254      // If the expression is not a call, and the state change is
255      // released -> allocated, it must be the realloc return value
256      // check. If we have to handle more cases here, it might be cleaner just
257      // to track this extra bit in the state itself.
258      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
259              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
260    }
261
262    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
263                                   const ExplodedNode *PrevN,
264                                   BugReporterContext &BRC,
265                                   BugReport &BR);
266
267    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
268                                    const ExplodedNode *EndPathNode,
269                                    BugReport &BR) {
270      if (!IsLeak)
271        return 0;
272
273      PathDiagnosticLocation L =
274        PathDiagnosticLocation::createEndOfPath(EndPathNode,
275                                                BRC.getSourceManager());
276      // Do not add the statement itself as a range in case of leak.
277      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
278    }
279
280  private:
281    class StackHintGeneratorForReallocationFailed
282        : public StackHintGeneratorForSymbol {
283    public:
284      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
285        : StackHintGeneratorForSymbol(S, M) {}
286
287      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
288        SmallString<200> buf;
289        llvm::raw_svector_ostream os(buf);
290
291        os << "Reallocation of ";
292        // Printed parameters start at 1, not 0.
293        printOrdinal(++ArgIndex, os);
294        os << " parameter failed";
295
296        return os.str();
297      }
298
299      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
300        return "Reallocation of returned value failed";
301      }
302    };
303  };
304};
305} // end anonymous namespace
306
307typedef llvm::ImmutableMap<SymbolRef, RefState> RegionStateTy;
308typedef llvm::ImmutableMap<SymbolRef, ReallocPair > ReallocMap;
309class RegionState {};
310class ReallocPairs {};
311namespace clang {
312namespace ento {
313  template <>
314  struct ProgramStateTrait<RegionState>
315    : public ProgramStatePartialTrait<RegionStateTy> {
316    static void *GDMIndex() { static int x; return &x; }
317  };
318
319  template <>
320  struct ProgramStateTrait<ReallocPairs>
321    : public ProgramStatePartialTrait<ReallocMap> {
322    static void *GDMIndex() { static int x; return &x; }
323  };
324}
325}
326
327namespace {
328class StopTrackingCallback : public SymbolVisitor {
329  ProgramStateRef state;
330public:
331  StopTrackingCallback(ProgramStateRef st) : state(st) {}
332  ProgramStateRef getState() const { return state; }
333
334  bool VisitSymbol(SymbolRef sym) {
335    state = state->remove<RegionState>(sym);
336    return true;
337  }
338};
339} // end anonymous namespace
340
341void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
342  if (II_malloc)
343    return;
344  II_malloc = &Ctx.Idents.get("malloc");
345  II_free = &Ctx.Idents.get("free");
346  II_realloc = &Ctx.Idents.get("realloc");
347  II_reallocf = &Ctx.Idents.get("reallocf");
348  II_calloc = &Ctx.Idents.get("calloc");
349  II_valloc = &Ctx.Idents.get("valloc");
350  II_strdup = &Ctx.Idents.get("strdup");
351  II_strndup = &Ctx.Idents.get("strndup");
352}
353
354bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
355  if (isFreeFunction(FD, C))
356    return true;
357
358  if (isAllocationFunction(FD, C))
359    return true;
360
361  return false;
362}
363
364bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
365                                         ASTContext &C) const {
366  if (!FD)
367    return false;
368
369  IdentifierInfo *FunI = FD->getIdentifier();
370  if (!FunI)
371    return false;
372
373  initIdentifierInfo(C);
374
375  if (FunI == II_malloc || FunI == II_realloc ||
376      FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
377      FunI == II_strdup || FunI == II_strndup)
378    return true;
379
380  if (Filter.CMallocOptimistic && FD->hasAttrs())
381    for (specific_attr_iterator<OwnershipAttr>
382           i = FD->specific_attr_begin<OwnershipAttr>(),
383           e = FD->specific_attr_end<OwnershipAttr>();
384           i != e; ++i)
385      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
386        return true;
387  return false;
388}
389
390bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
391  if (!FD)
392    return false;
393
394  IdentifierInfo *FunI = FD->getIdentifier();
395  if (!FunI)
396    return false;
397
398  initIdentifierInfo(C);
399
400  if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
401    return true;
402
403  if (Filter.CMallocOptimistic && FD->hasAttrs())
404    for (specific_attr_iterator<OwnershipAttr>
405           i = FD->specific_attr_begin<OwnershipAttr>(),
406           e = FD->specific_attr_end<OwnershipAttr>();
407           i != e; ++i)
408      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
409          (*i)->getOwnKind() == OwnershipAttr::Holds)
410        return true;
411  return false;
412}
413
414void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
415  const FunctionDecl *FD = C.getCalleeDecl(CE);
416  if (!FD)
417    return;
418
419  initIdentifierInfo(C.getASTContext());
420  IdentifierInfo *FunI = FD->getIdentifier();
421  if (!FunI)
422    return;
423
424  ProgramStateRef State = C.getState();
425  if (FunI == II_malloc || FunI == II_valloc) {
426    if (CE->getNumArgs() < 1)
427      return;
428    State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
429  } else if (FunI == II_realloc) {
430    State = ReallocMem(C, CE, false);
431  } else if (FunI == II_reallocf) {
432    State = ReallocMem(C, CE, true);
433  } else if (FunI == II_calloc) {
434    State = CallocMem(C, CE);
435  } else if (FunI == II_free) {
436    State = FreeMemAux(C, CE, C.getState(), 0, false);
437  } else if (FunI == II_strdup) {
438    State = MallocUpdateRefState(C, CE, State);
439  } else if (FunI == II_strndup) {
440    State = MallocUpdateRefState(C, CE, State);
441  } else if (Filter.CMallocOptimistic) {
442    // Check all the attributes, if there are any.
443    // There can be multiple of these attributes.
444    if (FD->hasAttrs())
445      for (specific_attr_iterator<OwnershipAttr>
446          i = FD->specific_attr_begin<OwnershipAttr>(),
447          e = FD->specific_attr_end<OwnershipAttr>();
448          i != e; ++i) {
449        switch ((*i)->getOwnKind()) {
450        case OwnershipAttr::Returns:
451          State = MallocMemReturnsAttr(C, CE, *i);
452          break;
453        case OwnershipAttr::Takes:
454        case OwnershipAttr::Holds:
455          State = FreeMemAttr(C, CE, *i);
456          break;
457        }
458      }
459  }
460  C.addTransition(State);
461}
462
463ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
464                                                    const CallExpr *CE,
465                                                    const OwnershipAttr* Att) {
466  if (Att->getModule() != "malloc")
467    return 0;
468
469  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
470  if (I != E) {
471    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
472  }
473  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
474}
475
476ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
477                                           const CallExpr *CE,
478                                           SVal Size, SVal Init,
479                                           ProgramStateRef state) {
480
481  // Bind the return value to the symbolic value from the heap region.
482  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
483  // side effects other than what we model here.
484  unsigned Count = C.getCurrentBlockCount();
485  SValBuilder &svalBuilder = C.getSValBuilder();
486  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
487  DefinedSVal RetVal =
488    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
489  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
490
491  // We expect the malloc functions to return a pointer.
492  if (!isa<Loc>(RetVal))
493    return 0;
494
495  // Fill the region with the initialization value.
496  state = state->bindDefault(RetVal, Init);
497
498  // Set the region's extent equal to the Size parameter.
499  const SymbolicRegion *R =
500      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
501  if (!R)
502    return 0;
503  if (isa<DefinedOrUnknownSVal>(Size)) {
504    SValBuilder &svalBuilder = C.getSValBuilder();
505    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
506    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
507    DefinedOrUnknownSVal extentMatchesSize =
508        svalBuilder.evalEQ(state, Extent, DefinedSize);
509
510    state = state->assume(extentMatchesSize, true);
511    assert(state);
512  }
513
514  return MallocUpdateRefState(C, CE, state);
515}
516
517ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
518                                                    const CallExpr *CE,
519                                                    ProgramStateRef state) {
520  // Get the return value.
521  SVal retVal = state->getSVal(CE, C.getLocationContext());
522
523  // We expect the malloc functions to return a pointer.
524  if (!isa<Loc>(retVal))
525    return 0;
526
527  SymbolRef Sym = retVal.getAsLocSymbol();
528  assert(Sym);
529
530  // Set the symbol's state to Allocated.
531  return state->set<RegionState>(Sym, RefState::getAllocateUnchecked(CE));
532
533}
534
535ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
536                                           const CallExpr *CE,
537                                           const OwnershipAttr* Att) const {
538  if (Att->getModule() != "malloc")
539    return 0;
540
541  ProgramStateRef State = C.getState();
542
543  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
544       I != E; ++I) {
545    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
546                               Att->getOwnKind() == OwnershipAttr::Holds);
547    if (StateI)
548      State = StateI;
549  }
550  return State;
551}
552
553ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
554                                          const CallExpr *CE,
555                                          ProgramStateRef state,
556                                          unsigned Num,
557                                          bool Hold) const {
558  if (CE->getNumArgs() < (Num + 1))
559    return 0;
560
561  const Expr *ArgExpr = CE->getArg(Num);
562  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
563  if (!isa<DefinedOrUnknownSVal>(ArgVal))
564    return 0;
565  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
566
567  // Check for null dereferences.
568  if (!isa<Loc>(location))
569    return 0;
570
571  // The explicit NULL case, no operation is performed.
572  ProgramStateRef notNullState, nullState;
573  llvm::tie(notNullState, nullState) = state->assume(location);
574  if (nullState && !notNullState)
575    return 0;
576
577  // Unknown values could easily be okay
578  // Undefined values are handled elsewhere
579  if (ArgVal.isUnknownOrUndef())
580    return 0;
581
582  const MemRegion *R = ArgVal.getAsRegion();
583
584  // Nonlocs can't be freed, of course.
585  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
586  if (!R) {
587    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
588    return 0;
589  }
590
591  R = R->StripCasts();
592
593  // Blocks might show up as heap data, but should not be free()d
594  if (isa<BlockDataRegion>(R)) {
595    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
596    return 0;
597  }
598
599  const MemSpaceRegion *MS = R->getMemorySpace();
600
601  // Parameters, locals, statics, and globals shouldn't be freed.
602  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
603    // FIXME: at the time this code was written, malloc() regions were
604    // represented by conjured symbols, which are all in UnknownSpaceRegion.
605    // This means that there isn't actually anything from HeapSpaceRegion
606    // that should be freed, even though we allow it here.
607    // Of course, free() can work on memory allocated outside the current
608    // function, so UnknownSpaceRegion is always a possibility.
609    // False negatives are better than false positives.
610
611    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
612    return 0;
613  }
614
615  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
616  // Various cases could lead to non-symbol values here.
617  // For now, ignore them.
618  if (!SR)
619    return 0;
620
621  SymbolRef Sym = SR->getSymbol();
622  const RefState *RS = state->get<RegionState>(Sym);
623
624  // If the symbol has not been tracked, return. This is possible when free() is
625  // called on a pointer that does not get its pointee directly from malloc().
626  // Full support of this requires inter-procedural analysis.
627  if (!RS)
628    return 0;
629
630  // Check double free.
631  if (RS->isReleased()) {
632    if (ExplodedNode *N = C.generateSink()) {
633      if (!BT_DoubleFree)
634        BT_DoubleFree.reset(
635          new BugType("Double free", "Memory Error"));
636      BugReport *R = new BugReport(*BT_DoubleFree,
637                        "Attempt to free released memory", N);
638      R->addRange(ArgExpr->getSourceRange());
639      R->markInteresting(Sym);
640      R->addVisitor(new MallocBugVisitor(Sym));
641      C.EmitReport(R);
642    }
643    return 0;
644  }
645
646  // Normal free.
647  if (Hold)
648    return state->set<RegionState>(Sym, RefState::getRelinquished(CE));
649  return state->set<RegionState>(Sym, RefState::getReleased(CE));
650}
651
652bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
653  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
654    os << "an integer (" << IntVal->getValue() << ")";
655  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
656    os << "a constant address (" << ConstAddr->getValue() << ")";
657  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
658    os << "the address of the label '" << Label->getLabel()->getName() << "'";
659  else
660    return false;
661
662  return true;
663}
664
665bool MallocChecker::SummarizeRegion(raw_ostream &os,
666                                    const MemRegion *MR) {
667  switch (MR->getKind()) {
668  case MemRegion::FunctionTextRegionKind: {
669    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
670    if (FD)
671      os << "the address of the function '" << *FD << '\'';
672    else
673      os << "the address of a function";
674    return true;
675  }
676  case MemRegion::BlockTextRegionKind:
677    os << "block text";
678    return true;
679  case MemRegion::BlockDataRegionKind:
680    // FIXME: where the block came from?
681    os << "a block";
682    return true;
683  default: {
684    const MemSpaceRegion *MS = MR->getMemorySpace();
685
686    if (isa<StackLocalsSpaceRegion>(MS)) {
687      const VarRegion *VR = dyn_cast<VarRegion>(MR);
688      const VarDecl *VD;
689      if (VR)
690        VD = VR->getDecl();
691      else
692        VD = NULL;
693
694      if (VD)
695        os << "the address of the local variable '" << VD->getName() << "'";
696      else
697        os << "the address of a local stack variable";
698      return true;
699    }
700
701    if (isa<StackArgumentsSpaceRegion>(MS)) {
702      const VarRegion *VR = dyn_cast<VarRegion>(MR);
703      const VarDecl *VD;
704      if (VR)
705        VD = VR->getDecl();
706      else
707        VD = NULL;
708
709      if (VD)
710        os << "the address of the parameter '" << VD->getName() << "'";
711      else
712        os << "the address of a parameter";
713      return true;
714    }
715
716    if (isa<GlobalsSpaceRegion>(MS)) {
717      const VarRegion *VR = dyn_cast<VarRegion>(MR);
718      const VarDecl *VD;
719      if (VR)
720        VD = VR->getDecl();
721      else
722        VD = NULL;
723
724      if (VD) {
725        if (VD->isStaticLocal())
726          os << "the address of the static variable '" << VD->getName() << "'";
727        else
728          os << "the address of the global variable '" << VD->getName() << "'";
729      } else
730        os << "the address of a global variable";
731      return true;
732    }
733
734    return false;
735  }
736  }
737}
738
739void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
740                                  SourceRange range) const {
741  if (ExplodedNode *N = C.generateSink()) {
742    if (!BT_BadFree)
743      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
744
745    SmallString<100> buf;
746    llvm::raw_svector_ostream os(buf);
747
748    const MemRegion *MR = ArgVal.getAsRegion();
749    if (MR) {
750      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
751        MR = ER->getSuperRegion();
752
753      // Special case for alloca()
754      if (isa<AllocaRegion>(MR))
755        os << "Argument to free() was allocated by alloca(), not malloc()";
756      else {
757        os << "Argument to free() is ";
758        if (SummarizeRegion(os, MR))
759          os << ", which is not memory allocated by malloc()";
760        else
761          os << "not memory allocated by malloc()";
762      }
763    } else {
764      os << "Argument to free() is ";
765      if (SummarizeValue(os, ArgVal))
766        os << ", which is not memory allocated by malloc()";
767      else
768        os << "not memory allocated by malloc()";
769    }
770
771    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
772    R->markInteresting(MR);
773    R->addRange(range);
774    C.EmitReport(R);
775  }
776}
777
778ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
779                                          const CallExpr *CE,
780                                          bool FreesOnFail) const {
781  if (CE->getNumArgs() < 2)
782    return 0;
783
784  ProgramStateRef state = C.getState();
785  const Expr *arg0Expr = CE->getArg(0);
786  const LocationContext *LCtx = C.getLocationContext();
787  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
788  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
789    return 0;
790  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
791
792  SValBuilder &svalBuilder = C.getSValBuilder();
793
794  DefinedOrUnknownSVal PtrEQ =
795    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
796
797  // Get the size argument. If there is no size arg then give up.
798  const Expr *Arg1 = CE->getArg(1);
799  if (!Arg1)
800    return 0;
801
802  // Get the value of the size argument.
803  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
804  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
805    return 0;
806  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
807
808  // Compare the size argument to 0.
809  DefinedOrUnknownSVal SizeZero =
810    svalBuilder.evalEQ(state, Arg1Val,
811                       svalBuilder.makeIntValWithPtrWidth(0, false));
812
813  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
814  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
815  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
816  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
817  // We only assume exceptional states if they are definitely true; if the
818  // state is under-constrained, assume regular realloc behavior.
819  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
820  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
821
822  // If the ptr is NULL and the size is not 0, the call is equivalent to
823  // malloc(size).
824  if ( PrtIsNull && !SizeIsZero) {
825    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
826                                               UndefinedVal(), StatePtrIsNull);
827    return stateMalloc;
828  }
829
830  if (PrtIsNull && SizeIsZero)
831    return 0;
832
833  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
834  assert(!PrtIsNull);
835  SymbolRef FromPtr = arg0Val.getAsSymbol();
836  SVal RetVal = state->getSVal(CE, LCtx);
837  SymbolRef ToPtr = RetVal.getAsSymbol();
838  if (!FromPtr || !ToPtr)
839    return 0;
840
841  // If the size is 0, free the memory.
842  if (SizeIsZero)
843    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
844      // The semantics of the return value are:
845      // If size was equal to 0, either NULL or a pointer suitable to be passed
846      // to free() is returned.
847      stateFree = stateFree->set<ReallocPairs>(ToPtr,
848                                            ReallocPair(FromPtr, FreesOnFail));
849      C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
850      return stateFree;
851    }
852
853  // Default behavior.
854  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
855    // FIXME: We should copy the content of the original buffer.
856    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
857                                                UnknownVal(), stateFree);
858    if (!stateRealloc)
859      return 0;
860    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
861                                            ReallocPair(FromPtr, FreesOnFail));
862    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
863    return stateRealloc;
864  }
865  return 0;
866}
867
868ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
869  if (CE->getNumArgs() < 2)
870    return 0;
871
872  ProgramStateRef state = C.getState();
873  SValBuilder &svalBuilder = C.getSValBuilder();
874  const LocationContext *LCtx = C.getLocationContext();
875  SVal count = state->getSVal(CE->getArg(0), LCtx);
876  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
877  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
878                                        svalBuilder.getContext().getSizeType());
879  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
880
881  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
882}
883
884LeakInfo
885MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
886                                 CheckerContext &C) const {
887  const LocationContext *LeakContext = N->getLocationContext();
888  // Walk the ExplodedGraph backwards and find the first node that referred to
889  // the tracked symbol.
890  const ExplodedNode *AllocNode = N;
891  const MemRegion *ReferenceRegion = 0;
892
893  while (N) {
894    ProgramStateRef State = N->getState();
895    if (!State->get<RegionState>(Sym))
896      break;
897
898    // Find the most recent expression bound to the symbol in the current
899    // context.
900    if (!ReferenceRegion) {
901      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
902        SVal Val = State->getSVal(MR);
903        if (Val.getAsLocSymbol() == Sym)
904          ReferenceRegion = MR;
905      }
906    }
907
908    // Allocation node, is the last node in the current context in which the
909    // symbol was tracked.
910    if (N->getLocationContext() == LeakContext)
911      AllocNode = N;
912    N = N->pred_empty() ? NULL : *(N->pred_begin());
913  }
914
915  ProgramPoint P = AllocNode->getLocation();
916  const Stmt *AllocationStmt = 0;
917  if (isa<StmtPoint>(P))
918    AllocationStmt = cast<StmtPoint>(P).getStmt();
919
920  return LeakInfo(AllocationStmt, ReferenceRegion);
921}
922
923void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
924                               CheckerContext &C) const {
925  assert(N);
926  if (!BT_Leak) {
927    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
928    // Leaks should not be reported if they are post-dominated by a sink:
929    // (1) Sinks are higher importance bugs.
930    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
931    //     with __noreturn functions such as assert() or exit(). We choose not
932    //     to report leaks on such paths.
933    BT_Leak->setSuppressOnSink(true);
934  }
935
936  // Most bug reports are cached at the location where they occurred.
937  // With leaks, we want to unique them by the location where they were
938  // allocated, and only report a single path.
939  PathDiagnosticLocation LocUsedForUniqueing;
940  const Stmt *AllocStmt = 0;
941  const MemRegion *Region = 0;
942  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
943  if (AllocStmt)
944    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
945                            C.getSourceManager(), N->getLocationContext());
946
947  SmallString<200> buf;
948  llvm::raw_svector_ostream os(buf);
949  os << "Memory is never released; potential leak";
950  if (Region) {
951    os << " of memory pointed to by '";
952    Region->dumpPretty(os);
953    os <<'\'';
954  }
955
956  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
957  R->markInteresting(Sym);
958  R->addVisitor(new MallocBugVisitor(Sym, true));
959  C.EmitReport(R);
960}
961
962void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
963                                     CheckerContext &C) const
964{
965  if (!SymReaper.hasDeadSymbols())
966    return;
967
968  ProgramStateRef state = C.getState();
969  RegionStateTy RS = state->get<RegionState>();
970  RegionStateTy::Factory &F = state->get_context<RegionState>();
971
972  bool generateReport = false;
973  llvm::SmallVector<SymbolRef, 2> Errors;
974  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
975    if (SymReaper.isDead(I->first)) {
976      if (I->second.isAllocated()) {
977        generateReport = true;
978        Errors.push_back(I->first);
979      }
980      // Remove the dead symbol from the map.
981      RS = F.remove(RS, I->first);
982
983    }
984  }
985
986  // Cleanup the Realloc Pairs Map.
987  ReallocMap RP = state->get<ReallocPairs>();
988  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
989    if (SymReaper.isDead(I->first) ||
990        SymReaper.isDead(I->second.ReallocatedSym)) {
991      state = state->remove<ReallocPairs>(I->first);
992    }
993  }
994
995  // Generate leak node.
996  static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
997  ExplodedNode *N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
998
999  if (generateReport) {
1000    for (llvm::SmallVector<SymbolRef, 2>::iterator
1001         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1002      reportLeak(*I, N, C);
1003    }
1004  }
1005  C.addTransition(state->set<RegionState>(RS), N);
1006}
1007
1008void MallocChecker::checkEndPath(CheckerContext &C) const {
1009  ProgramStateRef state = C.getState();
1010  RegionStateTy M = state->get<RegionState>();
1011
1012  // If inside inlined call, skip it.
1013  if (C.getLocationContext()->getParent() != 0)
1014    return;
1015
1016  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1017    RefState RS = I->second;
1018    if (RS.isAllocated()) {
1019      ExplodedNode *N = C.addTransition(state);
1020      if (N)
1021        reportLeak(I->first, N, C);
1022    }
1023  }
1024}
1025
1026bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
1027                                CheckerContext &C) const {
1028  ProgramStateRef state = C.getState();
1029  const RefState *RS = state->get<RegionState>(Sym);
1030  if (!RS)
1031    return false;
1032
1033  if (RS->isAllocated()) {
1034    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
1035    C.addTransition(state);
1036    return true;
1037  }
1038  return false;
1039}
1040
1041void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1042  // We will check for double free in the post visit.
1043  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1044    return;
1045
1046  // Check use after free, when a freed pointer is passed to a call.
1047  ProgramStateRef State = C.getState();
1048  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1049                                    E = CE->arg_end(); I != E; ++I) {
1050    const Expr *A = *I;
1051    if (A->getType().getTypePtr()->isAnyPointerType()) {
1052      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1053      if (!Sym)
1054        continue;
1055      if (checkUseAfterFree(Sym, C, A))
1056        return;
1057    }
1058  }
1059}
1060
1061void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1062  const Expr *E = S->getRetValue();
1063  if (!E)
1064    return;
1065
1066  // Check if we are returning a symbol.
1067  SVal RetVal = C.getState()->getSVal(E, C.getLocationContext());
1068  SymbolRef Sym = RetVal.getAsSymbol();
1069  if (!Sym)
1070    // If we are returning a field of the allocated struct or an array element,
1071    // the callee could still free the memory.
1072    // TODO: This logic should be a part of generic symbol escape callback.
1073    if (const MemRegion *MR = RetVal.getAsRegion())
1074      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1075        if (const SymbolicRegion *BMR =
1076              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1077          Sym = BMR->getSymbol();
1078  if (!Sym)
1079    return;
1080
1081  // Check if we are returning freed memory.
1082  if (checkUseAfterFree(Sym, C, E))
1083    return;
1084
1085  // If this function body is not inlined, check if the symbol is escaping.
1086  if (C.getLocationContext()->getParent() == 0)
1087    checkEscape(Sym, E, C);
1088}
1089
1090// TODO: Blocks should be either inlined or should call invalidate regions
1091// upon invocation. After that's in place, special casing here will not be
1092// needed.
1093void MallocChecker::checkPostStmt(const BlockExpr *BE,
1094                                  CheckerContext &C) const {
1095
1096  // Scan the BlockDecRefExprs for any object the retain count checker
1097  // may be tracking.
1098  if (!BE->getBlockDecl()->hasCaptures())
1099    return;
1100
1101  ProgramStateRef state = C.getState();
1102  const BlockDataRegion *R =
1103    cast<BlockDataRegion>(state->getSVal(BE,
1104                                         C.getLocationContext()).getAsRegion());
1105
1106  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1107                                            E = R->referenced_vars_end();
1108
1109  if (I == E)
1110    return;
1111
1112  SmallVector<const MemRegion*, 10> Regions;
1113  const LocationContext *LC = C.getLocationContext();
1114  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1115
1116  for ( ; I != E; ++I) {
1117    const VarRegion *VR = *I;
1118    if (VR->getSuperRegion() == R) {
1119      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1120    }
1121    Regions.push_back(VR);
1122  }
1123
1124  state =
1125    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1126                                    Regions.data() + Regions.size()).getState();
1127  C.addTransition(state);
1128}
1129
1130bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1131  assert(Sym);
1132  const RefState *RS = C.getState()->get<RegionState>(Sym);
1133  return (RS && RS->isReleased());
1134}
1135
1136bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1137                                      const Stmt *S) const {
1138  if (isReleased(Sym, C)) {
1139    if (ExplodedNode *N = C.generateSink()) {
1140      if (!BT_UseFree)
1141        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1142
1143      BugReport *R = new BugReport(*BT_UseFree,
1144                                   "Use of memory after it is freed",N);
1145      if (S)
1146        R->addRange(S->getSourceRange());
1147      R->markInteresting(Sym);
1148      R->addVisitor(new MallocBugVisitor(Sym));
1149      C.EmitReport(R);
1150      return true;
1151    }
1152  }
1153  return false;
1154}
1155
1156// Check if the location is a freed symbolic region.
1157void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1158                                  CheckerContext &C) const {
1159  SymbolRef Sym = l.getLocSymbolInBase();
1160  if (Sym)
1161    checkUseAfterFree(Sym, C, S);
1162}
1163
1164//===----------------------------------------------------------------------===//
1165// Check various ways a symbol can be invalidated.
1166// TODO: This logic (the next 3 functions) is copied/similar to the
1167// RetainRelease checker. We might want to factor this out.
1168//===----------------------------------------------------------------------===//
1169
1170// Stop tracking symbols when a value escapes as a result of checkBind.
1171// A value escapes in three possible cases:
1172// (1) we are binding to something that is not a memory region.
1173// (2) we are binding to a memregion that does not have stack storage
1174// (3) we are binding to a memregion with stack storage that the store
1175//     does not understand.
1176void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1177                              CheckerContext &C) const {
1178  // Are we storing to something that causes the value to "escape"?
1179  bool escapes = true;
1180  ProgramStateRef state = C.getState();
1181
1182  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1183    escapes = !regionLoc->getRegion()->hasStackStorage();
1184
1185    if (!escapes) {
1186      // To test (3), generate a new state with the binding added.  If it is
1187      // the same state, then it escapes (since the store cannot represent
1188      // the binding).
1189      // Do this only if we know that the store is not supposed to generate the
1190      // same state.
1191      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1192      if (StoredVal != val)
1193        escapes = (state == (state->bindLoc(*regionLoc, val)));
1194    }
1195    if (!escapes) {
1196      // Case 4: We do not currently model what happens when a symbol is
1197      // assigned to a struct field, so be conservative here and let the symbol
1198      // go. TODO: This could definitely be improved upon.
1199      escapes = !isa<VarRegion>(regionLoc->getRegion());
1200    }
1201  }
1202
1203  // If our store can represent the binding and we aren't storing to something
1204  // that doesn't have local storage then just return and have the simulation
1205  // state continue as is.
1206  if (!escapes)
1207      return;
1208
1209  // Otherwise, find all symbols referenced by 'val' that we are tracking
1210  // and stop tracking them.
1211  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1212  C.addTransition(state);
1213}
1214
1215// If a symbolic region is assumed to NULL (or another constant), stop tracking
1216// it - assuming that allocation failed on this path.
1217ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1218                                              SVal Cond,
1219                                              bool Assumption) const {
1220  RegionStateTy RS = state->get<RegionState>();
1221  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1222    // If the symbol is assumed to NULL or another constant, this will
1223    // return an APSInt*.
1224    if (state->getSymVal(I.getKey()))
1225      state = state->remove<RegionState>(I.getKey());
1226  }
1227
1228  // Realloc returns 0 when reallocation fails, which means that we should
1229  // restore the state of the pointer being reallocated.
1230  ReallocMap RP = state->get<ReallocPairs>();
1231  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1232    // If the symbol is assumed to NULL or another constant, this will
1233    // return an APSInt*.
1234    if (state->getSymVal(I.getKey())) {
1235      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1236      const RefState *RS = state->get<RegionState>(ReallocSym);
1237      if (RS) {
1238        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1239          state = state->set<RegionState>(ReallocSym,
1240                             RefState::getAllocateUnchecked(RS->getStmt()));
1241      }
1242      state = state->remove<ReallocPairs>(I.getKey());
1243    }
1244  }
1245
1246  return state;
1247}
1248
1249// Check if the function is known to us. So, for example, we could
1250// conservatively assume it can free/reallocate it's pointer arguments.
1251// (We assume that the pointers cannot escape through calls to system
1252// functions not handled by this checker.)
1253bool MallocChecker::doesNotFreeMemory(const CallOrObjCMessage *Call,
1254                                      ProgramStateRef State) const {
1255  if (!Call)
1256    return false;
1257
1258  // For now, assume that any C++ call can free memory.
1259  // TODO: If we want to be more optimistic here, we'll need to make sure that
1260  // regions escape to C++ containers. They seem to do that even now, but for
1261  // mysterious reasons.
1262  if (Call->isCXXCall())
1263    return false;
1264
1265  const Decl *D = Call->getDecl();
1266  if (!D)
1267    return false;
1268
1269  ASTContext &ASTC = State->getStateManager().getContext();
1270
1271  // If it's one of the allocation functions we can reason about, we model
1272  // its behavior explicitly.
1273  if (isa<FunctionDecl>(D) && isMemFunction(cast<FunctionDecl>(D), ASTC)) {
1274    return true;
1275  }
1276
1277  // If it's not a system call, assume it frees memory.
1278  SourceManager &SM = ASTC.getSourceManager();
1279  if (!SM.isInSystemHeader(D->getLocation()))
1280    return false;
1281
1282  // Process C/ObjC functions.
1283  if (const FunctionDecl *FD  = dyn_cast<FunctionDecl>(D)) {
1284    // White list the system functions whose arguments escape.
1285    const IdentifierInfo *II = FD->getIdentifier();
1286    if (!II)
1287      return true;
1288    StringRef FName = II->getName();
1289
1290    // White list thread local storage.
1291    if (FName.equals("pthread_setspecific"))
1292      return false;
1293
1294    // White list the 'XXXNoCopy' ObjC functions.
1295    if (FName.endswith("NoCopy")) {
1296      // Look for the deallocator argument. We know that the memory ownership
1297      // is not transferred only if the deallocator argument is
1298      // 'kCFAllocatorNull'.
1299      for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1300        const Expr *ArgE = Call->getArg(i)->IgnoreParenCasts();
1301        if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1302          StringRef DeallocatorName = DE->getFoundDecl()->getName();
1303          if (DeallocatorName == "kCFAllocatorNull")
1304            return true;
1305        }
1306      }
1307      return false;
1308    }
1309
1310    // PR12101
1311    // Many CoreFoundation and CoreGraphics might allow a tracked object
1312    // to escape.
1313    if (Call->isCFCGAllowingEscape(FName))
1314      return false;
1315
1316    // Associating streams with malloced buffers. The pointer can escape if
1317    // 'closefn' is specified (and if that function does free memory).
1318    // Currently, we do not inspect the 'closefn' function (PR12101).
1319    if (FName == "funopen")
1320      if (Call->getNumArgs() >= 4 && !Call->getArgSVal(4).isConstant(0))
1321        return false;
1322
1323    // Do not warn on pointers passed to 'setbuf' when used with std streams,
1324    // these leaks might be intentional when setting the buffer for stdio.
1325    // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1326    if (FName == "setbuf" || FName =="setbuffer" ||
1327        FName == "setlinebuf" || FName == "setvbuf") {
1328      if (Call->getNumArgs() >= 1)
1329        if (const DeclRefExpr *Arg =
1330              dyn_cast<DeclRefExpr>(Call->getArg(0)->IgnoreParenCasts()))
1331          if (const VarDecl *D = dyn_cast<VarDecl>(Arg->getDecl()))
1332              if (D->getCanonicalDecl()->getName().find("std")
1333                                                   != StringRef::npos)
1334                return false;
1335    }
1336
1337    // A bunch of other functions, which take ownership of a pointer (See retain
1338    // release checker). Not all the parameters here are invalidated, but the
1339    // Malloc checker cannot differentiate between them. The right way of doing
1340    // this would be to implement a pointer escapes callback.
1341    if (FName == "CVPixelBufferCreateWithBytes" ||
1342        FName == "CGBitmapContextCreateWithData" ||
1343        FName == "CVPixelBufferCreateWithPlanarBytes" ||
1344        FName == "OSAtomicEnqueue") {
1345      return false;
1346    }
1347
1348    // Whitelist NSXXInsertXX, for example NSMapInsertIfAbsent, since they can
1349    // be deallocated by NSMapRemove.
1350    if (FName.startswith("NS") && (FName.find("Insert") != StringRef::npos))
1351      return false;
1352
1353    // If the call has a callback as an argument, assume the memory
1354    // can be freed.
1355    if (Call->hasNonZeroCallbackArg())
1356      return false;
1357
1358    // Otherwise, assume that the function does not free memory.
1359    // Most system calls, do not free the memory.
1360    return true;
1361
1362  // Process ObjC functions.
1363  } else if (const ObjCMethodDecl * ObjCD = dyn_cast<ObjCMethodDecl>(D)) {
1364    Selector S = ObjCD->getSelector();
1365
1366    // White list the ObjC functions which do free memory.
1367    // - Anything containing 'freeWhenDone' param set to 1.
1368    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1369    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1370      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1371        if (Call->getArgSVal(i).isConstant(1))
1372          return false;
1373        else
1374          return true;
1375      }
1376    }
1377
1378    // If the first selector ends with NoCopy, assume that the ownership is
1379    // transferred as well.
1380    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1381    if (S.getNameForSlot(0).endswith("NoCopy")) {
1382      return false;
1383    }
1384
1385    // If the call has a callback as an argument, assume the memory
1386    // can be freed.
1387    if (Call->hasNonZeroCallbackArg())
1388      return false;
1389
1390    // Otherwise, assume that the function does not free memory.
1391    // Most system calls, do not free the memory.
1392    return true;
1393  }
1394
1395  // Otherwise, assume that the function can free memory.
1396  return false;
1397
1398}
1399
1400// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1401// escapes, when we are tracking p), do not track the symbol as we cannot reason
1402// about it anymore.
1403ProgramStateRef
1404MallocChecker::checkRegionChanges(ProgramStateRef State,
1405                            const StoreManager::InvalidatedSymbols *invalidated,
1406                                    ArrayRef<const MemRegion *> ExplicitRegions,
1407                                    ArrayRef<const MemRegion *> Regions,
1408                                    const CallOrObjCMessage *Call) const {
1409  if (!invalidated || invalidated->empty())
1410    return State;
1411  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1412
1413  // If it's a call which might free or reallocate memory, we assume that all
1414  // regions (explicit and implicit) escaped.
1415
1416  // Otherwise, whitelist explicit pointers; we still can track them.
1417  if (!Call || doesNotFreeMemory(Call, State)) {
1418    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1419        E = ExplicitRegions.end(); I != E; ++I) {
1420      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1421        WhitelistedSymbols.insert(R->getSymbol());
1422    }
1423  }
1424
1425  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1426       E = invalidated->end(); I!=E; ++I) {
1427    SymbolRef sym = *I;
1428    if (WhitelistedSymbols.count(sym))
1429      continue;
1430    // The symbol escaped.
1431    if (const RefState *RS = State->get<RegionState>(sym))
1432      State = State->set<RegionState>(sym, RefState::getEscaped(RS->getStmt()));
1433  }
1434  return State;
1435}
1436
1437static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1438                                         ProgramStateRef prevState) {
1439  ReallocMap currMap = currState->get<ReallocPairs>();
1440  ReallocMap prevMap = prevState->get<ReallocPairs>();
1441
1442  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1443       I != E; ++I) {
1444    SymbolRef sym = I.getKey();
1445    if (!currMap.lookup(sym))
1446      return sym;
1447  }
1448
1449  return NULL;
1450}
1451
1452PathDiagnosticPiece *
1453MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1454                                           const ExplodedNode *PrevN,
1455                                           BugReporterContext &BRC,
1456                                           BugReport &BR) {
1457  ProgramStateRef state = N->getState();
1458  ProgramStateRef statePrev = PrevN->getState();
1459
1460  const RefState *RS = state->get<RegionState>(Sym);
1461  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1462  if (!RS && !RSPrev)
1463    return 0;
1464
1465  const Stmt *S = 0;
1466  const char *Msg = 0;
1467  StackHintGeneratorForSymbol *StackHint = 0;
1468
1469  // Retrieve the associated statement.
1470  ProgramPoint ProgLoc = N->getLocation();
1471  if (isa<StmtPoint>(ProgLoc))
1472    S = cast<StmtPoint>(ProgLoc).getStmt();
1473  // If an assumption was made on a branch, it should be caught
1474  // here by looking at the state transition.
1475  if (isa<BlockEdge>(ProgLoc)) {
1476    const CFGBlock *srcBlk = cast<BlockEdge>(ProgLoc).getSrc();
1477    S = srcBlk->getTerminator();
1478  }
1479  if (!S)
1480    return 0;
1481
1482  // Find out if this is an interesting point and what is the kind.
1483  if (Mode == Normal) {
1484    if (isAllocated(RS, RSPrev, S)) {
1485      Msg = "Memory is allocated";
1486      StackHint = new StackHintGeneratorForSymbol(Sym,
1487                                                  "Returned allocated memory");
1488    } else if (isReleased(RS, RSPrev, S)) {
1489      Msg = "Memory is released";
1490      StackHint = new StackHintGeneratorForSymbol(Sym,
1491                                                  "Returned released memory");
1492    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1493      Mode = ReallocationFailed;
1494      Msg = "Reallocation failed";
1495      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1496                                                       "Reallocation failed");
1497
1498      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1499        // Is it possible to fail two reallocs WITHOUT testing in between?
1500        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1501          "We only support one failed realloc at a time.");
1502        BR.markInteresting(sym);
1503        FailedReallocSymbol = sym;
1504      }
1505    }
1506
1507  // We are in a special mode if a reallocation failed later in the path.
1508  } else if (Mode == ReallocationFailed) {
1509    assert(FailedReallocSymbol && "No symbol to look for.");
1510
1511    // Is this is the first appearance of the reallocated symbol?
1512    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1513      // If we ever hit this assert, that means BugReporter has decided to skip
1514      // node pairs or visit them out of order.
1515      assert(state->get<RegionState>(FailedReallocSymbol) &&
1516        "Missed the reallocation point");
1517
1518      // We're at the reallocation point.
1519      Msg = "Attempt to reallocate memory";
1520      StackHint = new StackHintGeneratorForSymbol(Sym,
1521                                                 "Returned reallocated memory");
1522      FailedReallocSymbol = NULL;
1523      Mode = Normal;
1524    }
1525  }
1526
1527  if (!Msg)
1528    return 0;
1529  assert(StackHint);
1530
1531  // Generate the extra diagnostic.
1532  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1533                             N->getLocationContext());
1534  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1535}
1536
1537void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1538                               const char *NL, const char *Sep) const {
1539
1540  RegionStateTy RS = State->get<RegionState>();
1541
1542  if (!RS.isEmpty())
1543    Out << "Has Malloc data" << NL;
1544}
1545
1546#define REGISTER_CHECKER(name) \
1547void ento::register##name(CheckerManager &mgr) {\
1548  registerCStringCheckerBasic(mgr); \
1549  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1550}
1551
1552REGISTER_CHECKER(MallocPessimistic)
1553REGISTER_CHECKER(MallocOptimistic)
1554