MallocChecker.cpp revision 48d798ce32447607144db70a484cdb99c1180663 
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  // Get the return value.
481  SVal retVal = state->getSVal(CE, C.getLocationContext());
482
483  // We expect the malloc functions to return a pointer.
484  if (!isa<Loc>(retVal))
485    return 0;
486
487  // Fill the region with the initialization value.
488  state = state->bindDefault(retVal, Init);
489
490  // Set the region's extent equal to the Size parameter.
491  const SymbolicRegion *R =
492      dyn_cast_or_null<SymbolicRegion>(retVal.getAsRegion());
493  if (!R)
494    return 0;
495  if (isa<DefinedOrUnknownSVal>(Size)) {
496    SValBuilder &svalBuilder = C.getSValBuilder();
497    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
498    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
499    DefinedOrUnknownSVal extentMatchesSize =
500        svalBuilder.evalEQ(state, Extent, DefinedSize);
501
502    state = state->assume(extentMatchesSize, true);
503    assert(state);
504  }
505
506  return MallocUpdateRefState(C, CE, state);
507}
508
509ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
510                                                    const CallExpr *CE,
511                                                    ProgramStateRef state) {
512  // Get the return value.
513  SVal retVal = state->getSVal(CE, C.getLocationContext());
514
515  // We expect the malloc functions to return a pointer.
516  if (!isa<Loc>(retVal))
517    return 0;
518
519  SymbolRef Sym = retVal.getAsLocSymbol();
520  assert(Sym);
521
522  // Set the symbol's state to Allocated.
523  return state->set<RegionState>(Sym, RefState::getAllocateUnchecked(CE));
524
525}
526
527ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
528                                           const CallExpr *CE,
529                                           const OwnershipAttr* Att) const {
530  if (Att->getModule() != "malloc")
531    return 0;
532
533  ProgramStateRef State = C.getState();
534
535  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
536       I != E; ++I) {
537    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
538                               Att->getOwnKind() == OwnershipAttr::Holds);
539    if (StateI)
540      State = StateI;
541  }
542  return State;
543}
544
545ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
546                                          const CallExpr *CE,
547                                          ProgramStateRef state,
548                                          unsigned Num,
549                                          bool Hold) const {
550  if (CE->getNumArgs() < (Num + 1))
551    return 0;
552
553  const Expr *ArgExpr = CE->getArg(Num);
554  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
555  if (!isa<DefinedOrUnknownSVal>(ArgVal))
556    return 0;
557  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
558
559  // Check for null dereferences.
560  if (!isa<Loc>(location))
561    return 0;
562
563  // The explicit NULL case, no operation is performed.
564  ProgramStateRef notNullState, nullState;
565  llvm::tie(notNullState, nullState) = state->assume(location);
566  if (nullState && !notNullState)
567    return 0;
568
569  // Unknown values could easily be okay
570  // Undefined values are handled elsewhere
571  if (ArgVal.isUnknownOrUndef())
572    return 0;
573
574  const MemRegion *R = ArgVal.getAsRegion();
575
576  // Nonlocs can't be freed, of course.
577  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
578  if (!R) {
579    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
580    return 0;
581  }
582
583  R = R->StripCasts();
584
585  // Blocks might show up as heap data, but should not be free()d
586  if (isa<BlockDataRegion>(R)) {
587    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
588    return 0;
589  }
590
591  const MemSpaceRegion *MS = R->getMemorySpace();
592
593  // Parameters, locals, statics, and globals shouldn't be freed.
594  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
595    // FIXME: at the time this code was written, malloc() regions were
596    // represented by conjured symbols, which are all in UnknownSpaceRegion.
597    // This means that there isn't actually anything from HeapSpaceRegion
598    // that should be freed, even though we allow it here.
599    // Of course, free() can work on memory allocated outside the current
600    // function, so UnknownSpaceRegion is always a possibility.
601    // False negatives are better than false positives.
602
603    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
604    return 0;
605  }
606
607  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
608  // Various cases could lead to non-symbol values here.
609  // For now, ignore them.
610  if (!SR)
611    return 0;
612
613  SymbolRef Sym = SR->getSymbol();
614  const RefState *RS = state->get<RegionState>(Sym);
615
616  // If the symbol has not been tracked, return. This is possible when free() is
617  // called on a pointer that does not get its pointee directly from malloc().
618  // Full support of this requires inter-procedural analysis.
619  if (!RS)
620    return 0;
621
622  // Check double free.
623  if (RS->isReleased()) {
624    if (ExplodedNode *N = C.generateSink()) {
625      if (!BT_DoubleFree)
626        BT_DoubleFree.reset(
627          new BugType("Double free", "Memory Error"));
628      BugReport *R = new BugReport(*BT_DoubleFree,
629                        "Attempt to free released memory", N);
630      R->addRange(ArgExpr->getSourceRange());
631      R->markInteresting(Sym);
632      R->addVisitor(new MallocBugVisitor(Sym));
633      C.EmitReport(R);
634    }
635    return 0;
636  }
637
638  // Normal free.
639  if (Hold)
640    return state->set<RegionState>(Sym, RefState::getRelinquished(CE));
641  return state->set<RegionState>(Sym, RefState::getReleased(CE));
642}
643
644bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
645  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
646    os << "an integer (" << IntVal->getValue() << ")";
647  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
648    os << "a constant address (" << ConstAddr->getValue() << ")";
649  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
650    os << "the address of the label '" << Label->getLabel()->getName() << "'";
651  else
652    return false;
653
654  return true;
655}
656
657bool MallocChecker::SummarizeRegion(raw_ostream &os,
658                                    const MemRegion *MR) {
659  switch (MR->getKind()) {
660  case MemRegion::FunctionTextRegionKind: {
661    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
662    if (FD)
663      os << "the address of the function '" << *FD << '\'';
664    else
665      os << "the address of a function";
666    return true;
667  }
668  case MemRegion::BlockTextRegionKind:
669    os << "block text";
670    return true;
671  case MemRegion::BlockDataRegionKind:
672    // FIXME: where the block came from?
673    os << "a block";
674    return true;
675  default: {
676    const MemSpaceRegion *MS = MR->getMemorySpace();
677
678    if (isa<StackLocalsSpaceRegion>(MS)) {
679      const VarRegion *VR = dyn_cast<VarRegion>(MR);
680      const VarDecl *VD;
681      if (VR)
682        VD = VR->getDecl();
683      else
684        VD = NULL;
685
686      if (VD)
687        os << "the address of the local variable '" << VD->getName() << "'";
688      else
689        os << "the address of a local stack variable";
690      return true;
691    }
692
693    if (isa<StackArgumentsSpaceRegion>(MS)) {
694      const VarRegion *VR = dyn_cast<VarRegion>(MR);
695      const VarDecl *VD;
696      if (VR)
697        VD = VR->getDecl();
698      else
699        VD = NULL;
700
701      if (VD)
702        os << "the address of the parameter '" << VD->getName() << "'";
703      else
704        os << "the address of a parameter";
705      return true;
706    }
707
708    if (isa<GlobalsSpaceRegion>(MS)) {
709      const VarRegion *VR = dyn_cast<VarRegion>(MR);
710      const VarDecl *VD;
711      if (VR)
712        VD = VR->getDecl();
713      else
714        VD = NULL;
715
716      if (VD) {
717        if (VD->isStaticLocal())
718          os << "the address of the static variable '" << VD->getName() << "'";
719        else
720          os << "the address of the global variable '" << VD->getName() << "'";
721      } else
722        os << "the address of a global variable";
723      return true;
724    }
725
726    return false;
727  }
728  }
729}
730
731void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
732                                  SourceRange range) const {
733  if (ExplodedNode *N = C.generateSink()) {
734    if (!BT_BadFree)
735      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
736
737    SmallString<100> buf;
738    llvm::raw_svector_ostream os(buf);
739
740    const MemRegion *MR = ArgVal.getAsRegion();
741    if (MR) {
742      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
743        MR = ER->getSuperRegion();
744
745      // Special case for alloca()
746      if (isa<AllocaRegion>(MR))
747        os << "Argument to free() was allocated by alloca(), not malloc()";
748      else {
749        os << "Argument to free() is ";
750        if (SummarizeRegion(os, MR))
751          os << ", which is not memory allocated by malloc()";
752        else
753          os << "not memory allocated by malloc()";
754      }
755    } else {
756      os << "Argument to free() is ";
757      if (SummarizeValue(os, ArgVal))
758        os << ", which is not memory allocated by malloc()";
759      else
760        os << "not memory allocated by malloc()";
761    }
762
763    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
764    R->markInteresting(MR);
765    R->addRange(range);
766    C.EmitReport(R);
767  }
768}
769
770ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
771                                          const CallExpr *CE,
772                                          bool FreesOnFail) const {
773  if (CE->getNumArgs() < 2)
774    return 0;
775
776  ProgramStateRef state = C.getState();
777  const Expr *arg0Expr = CE->getArg(0);
778  const LocationContext *LCtx = C.getLocationContext();
779  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
780  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
781    return 0;
782  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
783
784  SValBuilder &svalBuilder = C.getSValBuilder();
785
786  DefinedOrUnknownSVal PtrEQ =
787    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
788
789  // Get the size argument. If there is no size arg then give up.
790  const Expr *Arg1 = CE->getArg(1);
791  if (!Arg1)
792    return 0;
793
794  // Get the value of the size argument.
795  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
796  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
797    return 0;
798  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
799
800  // Compare the size argument to 0.
801  DefinedOrUnknownSVal SizeZero =
802    svalBuilder.evalEQ(state, Arg1Val,
803                       svalBuilder.makeIntValWithPtrWidth(0, false));
804
805  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
806  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
807  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
808  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
809  // We only assume exceptional states if they are definitely true; if the
810  // state is under-constrained, assume regular realloc behavior.
811  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
812  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
813
814  // If the ptr is NULL and the size is not 0, the call is equivalent to
815  // malloc(size).
816  if ( PrtIsNull && !SizeIsZero) {
817    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
818                                               UndefinedVal(), StatePtrIsNull);
819    return stateMalloc;
820  }
821
822  if (PrtIsNull && SizeIsZero)
823    return 0;
824
825  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
826  assert(!PrtIsNull);
827  SymbolRef FromPtr = arg0Val.getAsSymbol();
828  SVal RetVal = state->getSVal(CE, LCtx);
829  SymbolRef ToPtr = RetVal.getAsSymbol();
830  if (!FromPtr || !ToPtr)
831    return 0;
832
833  // If the size is 0, free the memory.
834  if (SizeIsZero)
835    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
836      // The semantics of the return value are:
837      // If size was equal to 0, either NULL or a pointer suitable to be passed
838      // to free() is returned.
839      stateFree = stateFree->set<ReallocPairs>(ToPtr,
840                                            ReallocPair(FromPtr, FreesOnFail));
841      C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
842      return stateFree;
843    }
844
845  // Default behavior.
846  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
847    // FIXME: We should copy the content of the original buffer.
848    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
849                                                UnknownVal(), stateFree);
850    if (!stateRealloc)
851      return 0;
852    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
853                                            ReallocPair(FromPtr, FreesOnFail));
854    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
855    return stateRealloc;
856  }
857  return 0;
858}
859
860ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
861  if (CE->getNumArgs() < 2)
862    return 0;
863
864  ProgramStateRef state = C.getState();
865  SValBuilder &svalBuilder = C.getSValBuilder();
866  const LocationContext *LCtx = C.getLocationContext();
867  SVal count = state->getSVal(CE->getArg(0), LCtx);
868  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
869  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
870                                        svalBuilder.getContext().getSizeType());
871  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
872
873  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
874}
875
876LeakInfo
877MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
878                                 CheckerContext &C) const {
879  const LocationContext *LeakContext = N->getLocationContext();
880  // Walk the ExplodedGraph backwards and find the first node that referred to
881  // the tracked symbol.
882  const ExplodedNode *AllocNode = N;
883  const MemRegion *ReferenceRegion = 0;
884
885  while (N) {
886    ProgramStateRef State = N->getState();
887    if (!State->get<RegionState>(Sym))
888      break;
889
890    // Find the most recent expression bound to the symbol in the current
891    // context.
892    if (!ReferenceRegion) {
893      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
894        SVal Val = State->getSVal(MR);
895        if (Val.getAsLocSymbol() == Sym)
896          ReferenceRegion = MR;
897      }
898    }
899
900    // Allocation node, is the last node in the current context in which the
901    // symbol was tracked.
902    if (N->getLocationContext() == LeakContext)
903      AllocNode = N;
904    N = N->pred_empty() ? NULL : *(N->pred_begin());
905  }
906
907  ProgramPoint P = AllocNode->getLocation();
908  const Stmt *AllocationStmt = 0;
909  if (isa<StmtPoint>(P))
910    AllocationStmt = cast<StmtPoint>(P).getStmt();
911
912  return LeakInfo(AllocationStmt, ReferenceRegion);
913}
914
915void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
916                               CheckerContext &C) const {
917  assert(N);
918  if (!BT_Leak) {
919    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
920    // Leaks should not be reported if they are post-dominated by a sink:
921    // (1) Sinks are higher importance bugs.
922    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
923    //     with __noreturn functions such as assert() or exit(). We choose not
924    //     to report leaks on such paths.
925    BT_Leak->setSuppressOnSink(true);
926  }
927
928  // Most bug reports are cached at the location where they occurred.
929  // With leaks, we want to unique them by the location where they were
930  // allocated, and only report a single path.
931  PathDiagnosticLocation LocUsedForUniqueing;
932  const Stmt *AllocStmt = 0;
933  const MemRegion *Region = 0;
934  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
935  if (AllocStmt)
936    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
937                            C.getSourceManager(), N->getLocationContext());
938
939  SmallString<200> buf;
940  llvm::raw_svector_ostream os(buf);
941  os << "Memory is never released; potential leak";
942  if (Region) {
943    os << " of memory pointed to by '";
944    Region->dumpPretty(os);
945    os <<'\'';
946  }
947
948  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
949  R->markInteresting(Sym);
950  R->addVisitor(new MallocBugVisitor(Sym, true));
951  C.EmitReport(R);
952}
953
954void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
955                                     CheckerContext &C) const
956{
957  if (!SymReaper.hasDeadSymbols())
958    return;
959
960  ProgramStateRef state = C.getState();
961  RegionStateTy RS = state->get<RegionState>();
962  RegionStateTy::Factory &F = state->get_context<RegionState>();
963
964  bool generateReport = false;
965  llvm::SmallVector<SymbolRef, 2> Errors;
966  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
967    if (SymReaper.isDead(I->first)) {
968      if (I->second.isAllocated()) {
969        generateReport = true;
970        Errors.push_back(I->first);
971      }
972      // Remove the dead symbol from the map.
973      RS = F.remove(RS, I->first);
974
975    }
976  }
977
978  // Cleanup the Realloc Pairs Map.
979  ReallocMap RP = state->get<ReallocPairs>();
980  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
981    if (SymReaper.isDead(I->first) ||
982        SymReaper.isDead(I->second.ReallocatedSym)) {
983      state = state->remove<ReallocPairs>(I->first);
984    }
985  }
986
987  // Generate leak node.
988  static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
989  ExplodedNode *N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
990
991  if (generateReport) {
992    for (llvm::SmallVector<SymbolRef, 2>::iterator
993         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
994      reportLeak(*I, N, C);
995    }
996  }
997  C.addTransition(state->set<RegionState>(RS), N);
998}
999
1000void MallocChecker::checkEndPath(CheckerContext &C) const {
1001  ProgramStateRef state = C.getState();
1002  RegionStateTy M = state->get<RegionState>();
1003
1004  // If inside inlined call, skip it.
1005  if (C.getLocationContext()->getParent() != 0)
1006    return;
1007
1008  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1009    RefState RS = I->second;
1010    if (RS.isAllocated()) {
1011      ExplodedNode *N = C.addTransition(state);
1012      if (N)
1013        reportLeak(I->first, N, C);
1014    }
1015  }
1016}
1017
1018bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
1019                                CheckerContext &C) const {
1020  ProgramStateRef state = C.getState();
1021  const RefState *RS = state->get<RegionState>(Sym);
1022  if (!RS)
1023    return false;
1024
1025  if (RS->isAllocated()) {
1026    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
1027    C.addTransition(state);
1028    return true;
1029  }
1030  return false;
1031}
1032
1033void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1034  // We will check for double free in the post visit.
1035  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1036    return;
1037
1038  // Check use after free, when a freed pointer is passed to a call.
1039  ProgramStateRef State = C.getState();
1040  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1041                                    E = CE->arg_end(); I != E; ++I) {
1042    const Expr *A = *I;
1043    if (A->getType().getTypePtr()->isAnyPointerType()) {
1044      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1045      if (!Sym)
1046        continue;
1047      if (checkUseAfterFree(Sym, C, A))
1048        return;
1049    }
1050  }
1051}
1052
1053void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1054  const Expr *E = S->getRetValue();
1055  if (!E)
1056    return;
1057
1058  // Check if we are returning a symbol.
1059  SVal RetVal = C.getState()->getSVal(E, C.getLocationContext());
1060  SymbolRef Sym = RetVal.getAsSymbol();
1061  if (!Sym)
1062    // If we are returning a field of the allocated struct or an array element,
1063    // the callee could still free the memory.
1064    // TODO: This logic should be a part of generic symbol escape callback.
1065    if (const MemRegion *MR = RetVal.getAsRegion())
1066      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1067        if (const SymbolicRegion *BMR =
1068              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1069          Sym = BMR->getSymbol();
1070  if (!Sym)
1071    return;
1072
1073  // Check if we are returning freed memory.
1074  if (checkUseAfterFree(Sym, C, E))
1075    return;
1076
1077  // If this function body is not inlined, check if the symbol is escaping.
1078  if (C.getLocationContext()->getParent() == 0)
1079    checkEscape(Sym, E, C);
1080}
1081
1082// TODO: Blocks should be either inlined or should call invalidate regions
1083// upon invocation. After that's in place, special casing here will not be
1084// needed.
1085void MallocChecker::checkPostStmt(const BlockExpr *BE,
1086                                  CheckerContext &C) const {
1087
1088  // Scan the BlockDecRefExprs for any object the retain count checker
1089  // may be tracking.
1090  if (!BE->getBlockDecl()->hasCaptures())
1091    return;
1092
1093  ProgramStateRef state = C.getState();
1094  const BlockDataRegion *R =
1095    cast<BlockDataRegion>(state->getSVal(BE,
1096                                         C.getLocationContext()).getAsRegion());
1097
1098  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1099                                            E = R->referenced_vars_end();
1100
1101  if (I == E)
1102    return;
1103
1104  SmallVector<const MemRegion*, 10> Regions;
1105  const LocationContext *LC = C.getLocationContext();
1106  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1107
1108  for ( ; I != E; ++I) {
1109    const VarRegion *VR = *I;
1110    if (VR->getSuperRegion() == R) {
1111      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1112    }
1113    Regions.push_back(VR);
1114  }
1115
1116  state =
1117    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1118                                    Regions.data() + Regions.size()).getState();
1119  C.addTransition(state);
1120}
1121
1122bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1123  assert(Sym);
1124  const RefState *RS = C.getState()->get<RegionState>(Sym);
1125  return (RS && RS->isReleased());
1126}
1127
1128bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1129                                      const Stmt *S) const {
1130  if (isReleased(Sym, C)) {
1131    if (ExplodedNode *N = C.generateSink()) {
1132      if (!BT_UseFree)
1133        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1134
1135      BugReport *R = new BugReport(*BT_UseFree,
1136                                   "Use of memory after it is freed",N);
1137      if (S)
1138        R->addRange(S->getSourceRange());
1139      R->markInteresting(Sym);
1140      R->addVisitor(new MallocBugVisitor(Sym));
1141      C.EmitReport(R);
1142      return true;
1143    }
1144  }
1145  return false;
1146}
1147
1148// Check if the location is a freed symbolic region.
1149void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1150                                  CheckerContext &C) const {
1151  SymbolRef Sym = l.getLocSymbolInBase();
1152  if (Sym)
1153    checkUseAfterFree(Sym, C, S);
1154}
1155
1156//===----------------------------------------------------------------------===//
1157// Check various ways a symbol can be invalidated.
1158// TODO: This logic (the next 3 functions) is copied/similar to the
1159// RetainRelease checker. We might want to factor this out.
1160//===----------------------------------------------------------------------===//
1161
1162// Stop tracking symbols when a value escapes as a result of checkBind.
1163// A value escapes in three possible cases:
1164// (1) we are binding to something that is not a memory region.
1165// (2) we are binding to a memregion that does not have stack storage
1166// (3) we are binding to a memregion with stack storage that the store
1167//     does not understand.
1168void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1169                              CheckerContext &C) const {
1170  // Are we storing to something that causes the value to "escape"?
1171  bool escapes = true;
1172  ProgramStateRef state = C.getState();
1173
1174  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1175    escapes = !regionLoc->getRegion()->hasStackStorage();
1176
1177    if (!escapes) {
1178      // To test (3), generate a new state with the binding added.  If it is
1179      // the same state, then it escapes (since the store cannot represent
1180      // the binding).
1181      // Do this only if we know that the store is not supposed to generate the
1182      // same state.
1183      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1184      if (StoredVal != val)
1185        escapes = (state == (state->bindLoc(*regionLoc, val)));
1186    }
1187    if (!escapes) {
1188      // Case 4: We do not currently model what happens when a symbol is
1189      // assigned to a struct field, so be conservative here and let the symbol
1190      // go. TODO: This could definitely be improved upon.
1191      escapes = !isa<VarRegion>(regionLoc->getRegion());
1192    }
1193  }
1194
1195  // If our store can represent the binding and we aren't storing to something
1196  // that doesn't have local storage then just return and have the simulation
1197  // state continue as is.
1198  if (!escapes)
1199      return;
1200
1201  // Otherwise, find all symbols referenced by 'val' that we are tracking
1202  // and stop tracking them.
1203  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1204  C.addTransition(state);
1205}
1206
1207// If a symbolic region is assumed to NULL (or another constant), stop tracking
1208// it - assuming that allocation failed on this path.
1209ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1210                                              SVal Cond,
1211                                              bool Assumption) const {
1212  RegionStateTy RS = state->get<RegionState>();
1213  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1214    // If the symbol is assumed to NULL or another constant, this will
1215    // return an APSInt*.
1216    if (state->getSymVal(I.getKey()))
1217      state = state->remove<RegionState>(I.getKey());
1218  }
1219
1220  // Realloc returns 0 when reallocation fails, which means that we should
1221  // restore the state of the pointer being reallocated.
1222  ReallocMap RP = state->get<ReallocPairs>();
1223  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1224    // If the symbol is assumed to NULL or another constant, this will
1225    // return an APSInt*.
1226    if (state->getSymVal(I.getKey())) {
1227      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1228      const RefState *RS = state->get<RegionState>(ReallocSym);
1229      if (RS) {
1230        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1231          state = state->set<RegionState>(ReallocSym,
1232                             RefState::getAllocateUnchecked(RS->getStmt()));
1233      }
1234      state = state->remove<ReallocPairs>(I.getKey());
1235    }
1236  }
1237
1238  return state;
1239}
1240
1241// Check if the function is known to us. So, for example, we could
1242// conservatively assume it can free/reallocate it's pointer arguments.
1243// (We assume that the pointers cannot escape through calls to system
1244// functions not handled by this checker.)
1245bool MallocChecker::doesNotFreeMemory(const CallOrObjCMessage *Call,
1246                                      ProgramStateRef State) const {
1247  if (!Call)
1248    return false;
1249
1250  // For now, assume that any C++ call can free memory.
1251  // TODO: If we want to be more optimistic here, we'll need to make sure that
1252  // regions escape to C++ containers. They seem to do that even now, but for
1253  // mysterious reasons.
1254  if (Call->isCXXCall())
1255    return false;
1256
1257  const Decl *D = Call->getDecl();
1258  if (!D)
1259    return false;
1260
1261  ASTContext &ASTC = State->getStateManager().getContext();
1262
1263  // If it's one of the allocation functions we can reason about, we model
1264  // its behavior explicitly.
1265  if (isa<FunctionDecl>(D) && isMemFunction(cast<FunctionDecl>(D), ASTC)) {
1266    return true;
1267  }
1268
1269  // If it's not a system call, assume it frees memory.
1270  SourceManager &SM = ASTC.getSourceManager();
1271  if (!SM.isInSystemHeader(D->getLocation()))
1272    return false;
1273
1274  // Process C/ObjC functions.
1275  if (const FunctionDecl *FD  = dyn_cast<FunctionDecl>(D)) {
1276    // White list the system functions whose arguments escape.
1277    const IdentifierInfo *II = FD->getIdentifier();
1278    if (!II)
1279      return true;
1280    StringRef FName = II->getName();
1281
1282    // White list thread local storage.
1283    if (FName.equals("pthread_setspecific"))
1284      return false;
1285
1286    // White list the 'XXXNoCopy' ObjC functions.
1287    if (FName.endswith("NoCopy")) {
1288      // Look for the deallocator argument. We know that the memory ownership
1289      // is not transferred only if the deallocator argument is
1290      // 'kCFAllocatorNull'.
1291      for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1292        const Expr *ArgE = Call->getArg(i)->IgnoreParenCasts();
1293        if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1294          StringRef DeallocatorName = DE->getFoundDecl()->getName();
1295          if (DeallocatorName == "kCFAllocatorNull")
1296            return true;
1297        }
1298      }
1299      return false;
1300    }
1301
1302    // PR12101
1303    // Many CoreFoundation and CoreGraphics might allow a tracked object
1304    // to escape.
1305    if (Call->isCFCGAllowingEscape(FName))
1306      return false;
1307
1308    // Associating streams with malloced buffers. The pointer can escape if
1309    // 'closefn' is specified (and if that function does free memory).
1310    // Currently, we do not inspect the 'closefn' function (PR12101).
1311    if (FName == "funopen")
1312      if (Call->getNumArgs() >= 4 && !Call->getArgSVal(4).isConstant(0))
1313        return false;
1314
1315    // Do not warn on pointers passed to 'setbuf' when used with std streams,
1316    // these leaks might be intentional when setting the buffer for stdio.
1317    // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1318    if (FName == "setbuf" || FName =="setbuffer" ||
1319        FName == "setlinebuf" || FName == "setvbuf") {
1320      if (Call->getNumArgs() >= 1)
1321        if (const DeclRefExpr *Arg =
1322              dyn_cast<DeclRefExpr>(Call->getArg(0)->IgnoreParenCasts()))
1323          if (const VarDecl *D = dyn_cast<VarDecl>(Arg->getDecl()))
1324              if (D->getCanonicalDecl()->getName().find("std")
1325                                                   != StringRef::npos)
1326                return false;
1327    }
1328
1329    // A bunch of other functions, which take ownership of a pointer (See retain
1330    // release checker). Not all the parameters here are invalidated, but the
1331    // Malloc checker cannot differentiate between them. The right way of doing
1332    // this would be to implement a pointer escapes callback.
1333    if (FName == "CVPixelBufferCreateWithBytes" ||
1334        FName == "CGBitmapContextCreateWithData" ||
1335        FName == "CVPixelBufferCreateWithPlanarBytes" ||
1336        FName == "OSAtomicEnqueue") {
1337      return false;
1338    }
1339
1340    // Whitelist NSXXInsertXX, for example NSMapInsertIfAbsent, since they can
1341    // be deallocated by NSMapRemove.
1342    if (FName.startswith("NS") && (FName.find("Insert") != StringRef::npos))
1343      return false;
1344
1345    // If the call has a callback as an argument, assume the memory
1346    // can be freed.
1347    if (Call->hasNonZeroCallbackArg())
1348      return false;
1349
1350    // Otherwise, assume that the function does not free memory.
1351    // Most system calls, do not free the memory.
1352    return true;
1353
1354  // Process ObjC functions.
1355  } else if (const ObjCMethodDecl * ObjCD = dyn_cast<ObjCMethodDecl>(D)) {
1356    Selector S = ObjCD->getSelector();
1357
1358    // White list the ObjC functions which do free memory.
1359    // - Anything containing 'freeWhenDone' param set to 1.
1360    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1361    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1362      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1363        if (Call->getArgSVal(i).isConstant(1))
1364          return false;
1365        else
1366          return true;
1367      }
1368    }
1369
1370    // If the first selector ends with NoCopy, assume that the ownership is
1371    // transferred as well.
1372    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1373    if (S.getNameForSlot(0).endswith("NoCopy")) {
1374      return false;
1375    }
1376
1377    // If the call has a callback as an argument, assume the memory
1378    // can be freed.
1379    if (Call->hasNonZeroCallbackArg())
1380      return false;
1381
1382    // Otherwise, assume that the function does not free memory.
1383    // Most system calls, do not free the memory.
1384    return true;
1385  }
1386
1387  // Otherwise, assume that the function can free memory.
1388  return false;
1389
1390}
1391
1392// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1393// escapes, when we are tracking p), do not track the symbol as we cannot reason
1394// about it anymore.
1395ProgramStateRef
1396MallocChecker::checkRegionChanges(ProgramStateRef State,
1397                            const StoreManager::InvalidatedSymbols *invalidated,
1398                                    ArrayRef<const MemRegion *> ExplicitRegions,
1399                                    ArrayRef<const MemRegion *> Regions,
1400                                    const CallOrObjCMessage *Call) const {
1401  if (!invalidated || invalidated->empty())
1402    return State;
1403  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1404
1405  // If it's a call which might free or reallocate memory, we assume that all
1406  // regions (explicit and implicit) escaped.
1407
1408  // Otherwise, whitelist explicit pointers; we still can track them.
1409  if (!Call || doesNotFreeMemory(Call, State)) {
1410    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1411        E = ExplicitRegions.end(); I != E; ++I) {
1412      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1413        WhitelistedSymbols.insert(R->getSymbol());
1414    }
1415  }
1416
1417  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1418       E = invalidated->end(); I!=E; ++I) {
1419    SymbolRef sym = *I;
1420    if (WhitelistedSymbols.count(sym))
1421      continue;
1422    // The symbol escaped.
1423    if (const RefState *RS = State->get<RegionState>(sym))
1424      State = State->set<RegionState>(sym, RefState::getEscaped(RS->getStmt()));
1425  }
1426  return State;
1427}
1428
1429static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1430                                         ProgramStateRef prevState) {
1431  ReallocMap currMap = currState->get<ReallocPairs>();
1432  ReallocMap prevMap = prevState->get<ReallocPairs>();
1433
1434  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1435       I != E; ++I) {
1436    SymbolRef sym = I.getKey();
1437    if (!currMap.lookup(sym))
1438      return sym;
1439  }
1440
1441  return NULL;
1442}
1443
1444PathDiagnosticPiece *
1445MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1446                                           const ExplodedNode *PrevN,
1447                                           BugReporterContext &BRC,
1448                                           BugReport &BR) {
1449  ProgramStateRef state = N->getState();
1450  ProgramStateRef statePrev = PrevN->getState();
1451
1452  const RefState *RS = state->get<RegionState>(Sym);
1453  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1454  if (!RS && !RSPrev)
1455    return 0;
1456
1457  const Stmt *S = 0;
1458  const char *Msg = 0;
1459  StackHintGeneratorForSymbol *StackHint = 0;
1460
1461  // Retrieve the associated statement.
1462  ProgramPoint ProgLoc = N->getLocation();
1463  if (isa<StmtPoint>(ProgLoc))
1464    S = cast<StmtPoint>(ProgLoc).getStmt();
1465  // If an assumption was made on a branch, it should be caught
1466  // here by looking at the state transition.
1467  if (isa<BlockEdge>(ProgLoc)) {
1468    const CFGBlock *srcBlk = cast<BlockEdge>(ProgLoc).getSrc();
1469    S = srcBlk->getTerminator();
1470  }
1471  if (!S)
1472    return 0;
1473
1474  // Find out if this is an interesting point and what is the kind.
1475  if (Mode == Normal) {
1476    if (isAllocated(RS, RSPrev, S)) {
1477      Msg = "Memory is allocated";
1478      StackHint = new StackHintGeneratorForSymbol(Sym,
1479                                                  "Returned allocated memory");
1480    } else if (isReleased(RS, RSPrev, S)) {
1481      Msg = "Memory is released";
1482      StackHint = new StackHintGeneratorForSymbol(Sym,
1483                                                  "Returned released memory");
1484    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1485      Mode = ReallocationFailed;
1486      Msg = "Reallocation failed";
1487      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1488                                                       "Reallocation failed");
1489
1490      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1491        // Is it possible to fail two reallocs WITHOUT testing in between?
1492        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1493          "We only support one failed realloc at a time.");
1494        BR.markInteresting(sym);
1495        FailedReallocSymbol = sym;
1496      }
1497    }
1498
1499  // We are in a special mode if a reallocation failed later in the path.
1500  } else if (Mode == ReallocationFailed) {
1501    assert(FailedReallocSymbol && "No symbol to look for.");
1502
1503    // Is this is the first appearance of the reallocated symbol?
1504    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1505      // If we ever hit this assert, that means BugReporter has decided to skip
1506      // node pairs or visit them out of order.
1507      assert(state->get<RegionState>(FailedReallocSymbol) &&
1508        "Missed the reallocation point");
1509
1510      // We're at the reallocation point.
1511      Msg = "Attempt to reallocate memory";
1512      StackHint = new StackHintGeneratorForSymbol(Sym,
1513                                                 "Returned reallocated memory");
1514      FailedReallocSymbol = NULL;
1515      Mode = Normal;
1516    }
1517  }
1518
1519  if (!Msg)
1520    return 0;
1521  assert(StackHint);
1522
1523  // Generate the extra diagnostic.
1524  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1525                             N->getLocationContext());
1526  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1527}
1528
1529void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1530                               const char *NL, const char *Sep) const {
1531
1532  RegionStateTy RS = State->get<RegionState>();
1533
1534  if (!RS.isEmpty())
1535    Out << "Has Malloc data" << NL;
1536}
1537
1538#define REGISTER_CHECKER(name) \
1539void ento::register##name(CheckerManager &mgr) {\
1540  registerCStringCheckerBasic(mgr); \
1541  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1542}
1543
1544REGISTER_CHECKER(MallocPessimistic)
1545REGISTER_CHECKER(MallocOptimistic)
1546