MallocChecker.cpp revision 2ccecfaa4852c134191d4075d94e09399ab46fea 
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/CallEvent.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 "llvm/ADT/StringExtras.h"
30#include <climits>
31
32using namespace clang;
33using namespace ento;
34
35namespace {
36
37class RefState {
38  enum Kind { // Reference to allocated memory.
39              Allocated,
40              // Reference to released/freed memory.
41              Released,
42              // The responsibility for freeing resources has transfered from
43              // this reference. A relinquished symbol should not be freed.
44              Relinquished } K;
45  const Stmt *S;
46
47public:
48  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
49
50  bool isAllocated() const { return K == Allocated; }
51  bool isReleased() const { return K == Released; }
52  bool isRelinquished() const { return K == Relinquished; }
53
54  const Stmt *getStmt() const { return S; }
55
56  bool operator==(const RefState &X) const {
57    return K == X.K && S == X.S;
58  }
59
60  static RefState getAllocated(const Stmt *s) {
61    return RefState(Allocated, s);
62  }
63  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
64  static RefState getRelinquished(const Stmt *s) {
65    return RefState(Relinquished, s);
66  }
67
68  void Profile(llvm::FoldingSetNodeID &ID) const {
69    ID.AddInteger(K);
70    ID.AddPointer(S);
71  }
72};
73
74enum ReallocPairKind {
75  RPToBeFreedAfterFailure,
76  // The symbol has been freed when reallocation failed.
77  RPIsFreeOnFailure,
78  // The symbol does not need to be freed after reallocation fails.
79  RPDoNotTrackAfterFailure
80};
81
82/// \class ReallocPair
83/// \brief Stores information about the symbol being reallocated by a call to
84/// 'realloc' to allow modeling failed reallocation later in the path.
85struct ReallocPair {
86  // \brief The symbol which realloc reallocated.
87  SymbolRef ReallocatedSym;
88  ReallocPairKind Kind;
89
90  ReallocPair(SymbolRef S, ReallocPairKind K) :
91    ReallocatedSym(S), Kind(K) {}
92  void Profile(llvm::FoldingSetNodeID &ID) const {
93    ID.AddInteger(Kind);
94    ID.AddPointer(ReallocatedSym);
95  }
96  bool operator==(const ReallocPair &X) const {
97    return ReallocatedSym == X.ReallocatedSym &&
98           Kind == X.Kind;
99  }
100};
101
102typedef std::pair<const Stmt*, const MemRegion*> LeakInfo;
103
104class MallocChecker : public Checker<check::DeadSymbols,
105                                     check::EndPath,
106                                     check::PreStmt<ReturnStmt>,
107                                     check::PreStmt<CallExpr>,
108                                     check::PostStmt<CallExpr>,
109                                     check::PostStmt<BlockExpr>,
110                                     check::PostObjCMessage,
111                                     check::Location,
112                                     check::Bind,
113                                     eval::Assume,
114                                     check::RegionChanges>
115{
116  mutable OwningPtr<BugType> BT_DoubleFree;
117  mutable OwningPtr<BugType> BT_Leak;
118  mutable OwningPtr<BugType> BT_UseFree;
119  mutable OwningPtr<BugType> BT_BadFree;
120  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
121                         *II_valloc, *II_reallocf, *II_strndup, *II_strdup;
122
123public:
124  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
125                    II_valloc(0), II_reallocf(0), II_strndup(0), II_strdup(0) {}
126
127  /// In pessimistic mode, the checker assumes that it does not know which
128  /// functions might free the memory.
129  struct ChecksFilter {
130    DefaultBool CMallocPessimistic;
131    DefaultBool CMallocOptimistic;
132  };
133
134  ChecksFilter Filter;
135
136  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
137  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
138  void checkPostObjCMessage(const ObjCMethodCall &Call, CheckerContext &C) const;
139  void checkPostStmt(const BlockExpr *BE, CheckerContext &C) const;
140  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
141  void checkEndPath(CheckerContext &C) const;
142  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
143  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
144                            bool Assumption) const;
145  void checkLocation(SVal l, bool isLoad, const Stmt *S,
146                     CheckerContext &C) const;
147  void checkBind(SVal location, SVal val, const Stmt*S,
148                 CheckerContext &C) const;
149  ProgramStateRef
150  checkRegionChanges(ProgramStateRef state,
151                     const StoreManager::InvalidatedSymbols *invalidated,
152                     ArrayRef<const MemRegion *> ExplicitRegions,
153                     ArrayRef<const MemRegion *> Regions,
154                     const CallEvent *Call) const;
155  bool wantsRegionChangeUpdate(ProgramStateRef state) const {
156    return true;
157  }
158
159  void printState(raw_ostream &Out, ProgramStateRef State,
160                  const char *NL, const char *Sep) const;
161
162private:
163  void initIdentifierInfo(ASTContext &C) const;
164
165  /// Check if this is one of the functions which can allocate/reallocate memory
166  /// pointed to by one of its arguments.
167  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
168  bool isFreeFunction(const FunctionDecl *FD, ASTContext &C) const;
169  bool isAllocationFunction(const FunctionDecl *FD, ASTContext &C) const;
170
171  static ProgramStateRef MallocMemReturnsAttr(CheckerContext &C,
172                                              const CallExpr *CE,
173                                              const OwnershipAttr* Att);
174  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
175                                     const Expr *SizeEx, SVal Init,
176                                     ProgramStateRef state) {
177    return MallocMemAux(C, CE,
178                        state->getSVal(SizeEx, C.getLocationContext()),
179                        Init, state);
180  }
181
182  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
183                                     SVal SizeEx, SVal Init,
184                                     ProgramStateRef state);
185
186  /// Update the RefState to reflect the new memory allocation.
187  static ProgramStateRef MallocUpdateRefState(CheckerContext &C,
188                                              const CallExpr *CE,
189                                              ProgramStateRef state);
190
191  ProgramStateRef FreeMemAttr(CheckerContext &C, const CallExpr *CE,
192                              const OwnershipAttr* Att) const;
193  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
194                             ProgramStateRef state, unsigned Num,
195                             bool Hold,
196                             bool &ReleasedAllocated,
197                             bool ReturnsNullOnFailure = false) const;
198  ProgramStateRef FreeMemAux(CheckerContext &C, const Expr *Arg,
199                             const Expr *ParentExpr,
200                             ProgramStateRef State,
201                             bool Hold,
202                             bool &ReleasedAllocated,
203                             bool ReturnsNullOnFailure = false) const;
204
205  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
206                             bool FreesMemOnFailure) const;
207  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
208
209  ///\brief Check if the memory associated with this symbol was released.
210  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
211
212  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
213                         const Stmt *S = 0) const;
214
215  /// Check if the function is not known to us. So, for example, we could
216  /// conservatively assume it can free/reallocate it's pointer arguments.
217  bool doesNotFreeMemory(const CallEvent *Call,
218                         ProgramStateRef State) const;
219
220  static bool SummarizeValue(raw_ostream &os, SVal V);
221  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
222  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
223
224  /// Find the location of the allocation for Sym on the path leading to the
225  /// exploded node N.
226  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
227                             CheckerContext &C) const;
228
229  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
230
231  /// The bug visitor which allows us to print extra diagnostics along the
232  /// BugReport path. For example, showing the allocation site of the leaked
233  /// region.
234  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
235  protected:
236    enum NotificationMode {
237      Normal,
238      ReallocationFailed
239    };
240
241    // The allocated region symbol tracked by the main analysis.
242    SymbolRef Sym;
243
244    // The mode we are in, i.e. what kind of diagnostics will be emitted.
245    NotificationMode Mode;
246
247    // A symbol from when the primary region should have been reallocated.
248    SymbolRef FailedReallocSymbol;
249
250    bool IsLeak;
251
252  public:
253    MallocBugVisitor(SymbolRef S, bool isLeak = false)
254       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
255
256    virtual ~MallocBugVisitor() {}
257
258    void Profile(llvm::FoldingSetNodeID &ID) const {
259      static int X = 0;
260      ID.AddPointer(&X);
261      ID.AddPointer(Sym);
262    }
263
264    inline bool isAllocated(const RefState *S, const RefState *SPrev,
265                            const Stmt *Stmt) {
266      // Did not track -> allocated. Other state (released) -> allocated.
267      return (Stmt && isa<CallExpr>(Stmt) &&
268              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
269    }
270
271    inline bool isReleased(const RefState *S, const RefState *SPrev,
272                           const Stmt *Stmt) {
273      // Did not track -> released. Other state (allocated) -> released.
274      return (Stmt && isa<CallExpr>(Stmt) &&
275              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
276    }
277
278    inline bool isRelinquished(const RefState *S, const RefState *SPrev,
279                               const Stmt *Stmt) {
280      // Did not track -> relinquished. Other state (allocated) -> relinquished.
281      return (Stmt && (isa<CallExpr>(Stmt) || isa<ObjCMessageExpr>(Stmt) ||
282                                              isa<ObjCPropertyRefExpr>(Stmt)) &&
283              (S && S->isRelinquished()) &&
284              (!SPrev || !SPrev->isRelinquished()));
285    }
286
287    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
288                                     const Stmt *Stmt) {
289      // If the expression is not a call, and the state change is
290      // released -> allocated, it must be the realloc return value
291      // check. If we have to handle more cases here, it might be cleaner just
292      // to track this extra bit in the state itself.
293      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
294              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
295    }
296
297    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
298                                   const ExplodedNode *PrevN,
299                                   BugReporterContext &BRC,
300                                   BugReport &BR);
301
302    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
303                                    const ExplodedNode *EndPathNode,
304                                    BugReport &BR) {
305      if (!IsLeak)
306        return 0;
307
308      PathDiagnosticLocation L =
309        PathDiagnosticLocation::createEndOfPath(EndPathNode,
310                                                BRC.getSourceManager());
311      // Do not add the statement itself as a range in case of leak.
312      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
313    }
314
315  private:
316    class StackHintGeneratorForReallocationFailed
317        : public StackHintGeneratorForSymbol {
318    public:
319      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
320        : StackHintGeneratorForSymbol(S, M) {}
321
322      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
323        // Printed parameters start at 1, not 0.
324        ++ArgIndex;
325
326        SmallString<200> buf;
327        llvm::raw_svector_ostream os(buf);
328
329        os << "Reallocation of " << ArgIndex << llvm::getOrdinalSuffix(ArgIndex)
330           << " parameter failed";
331
332        return os.str();
333      }
334
335      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
336        return "Reallocation of returned value failed";
337      }
338    };
339  };
340};
341} // end anonymous namespace
342
343REGISTER_MAP_WITH_PROGRAMSTATE(RegionState, SymbolRef, RefState)
344REGISTER_MAP_WITH_PROGRAMSTATE(ReallocPairs, SymbolRef, ReallocPair)
345
346// A map from the freed symbol to the symbol representing the return value of
347// the free function.
348REGISTER_MAP_WITH_PROGRAMSTATE(FreeReturnValue, SymbolRef, SymbolRef)
349
350namespace {
351class StopTrackingCallback : public SymbolVisitor {
352  ProgramStateRef state;
353public:
354  StopTrackingCallback(ProgramStateRef st) : state(st) {}
355  ProgramStateRef getState() const { return state; }
356
357  bool VisitSymbol(SymbolRef sym) {
358    state = state->remove<RegionState>(sym);
359    return true;
360  }
361};
362} // end anonymous namespace
363
364void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
365  if (II_malloc)
366    return;
367  II_malloc = &Ctx.Idents.get("malloc");
368  II_free = &Ctx.Idents.get("free");
369  II_realloc = &Ctx.Idents.get("realloc");
370  II_reallocf = &Ctx.Idents.get("reallocf");
371  II_calloc = &Ctx.Idents.get("calloc");
372  II_valloc = &Ctx.Idents.get("valloc");
373  II_strdup = &Ctx.Idents.get("strdup");
374  II_strndup = &Ctx.Idents.get("strndup");
375}
376
377bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
378  if (isFreeFunction(FD, C))
379    return true;
380
381  if (isAllocationFunction(FD, C))
382    return true;
383
384  return false;
385}
386
387bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
388                                         ASTContext &C) const {
389  if (!FD)
390    return false;
391
392  if (FD->getKind() == Decl::Function) {
393    IdentifierInfo *FunI = FD->getIdentifier();
394    initIdentifierInfo(C);
395
396    if (FunI == II_malloc || FunI == II_realloc ||
397        FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
398        FunI == II_strdup || FunI == II_strndup)
399      return true;
400  }
401
402  if (Filter.CMallocOptimistic && FD->hasAttrs())
403    for (specific_attr_iterator<OwnershipAttr>
404           i = FD->specific_attr_begin<OwnershipAttr>(),
405           e = FD->specific_attr_end<OwnershipAttr>();
406           i != e; ++i)
407      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
408        return true;
409  return false;
410}
411
412bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
413  if (!FD)
414    return false;
415
416  if (FD->getKind() == Decl::Function) {
417    IdentifierInfo *FunI = FD->getIdentifier();
418    initIdentifierInfo(C);
419
420    if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
421      return true;
422  }
423
424  if (Filter.CMallocOptimistic && FD->hasAttrs())
425    for (specific_attr_iterator<OwnershipAttr>
426           i = FD->specific_attr_begin<OwnershipAttr>(),
427           e = FD->specific_attr_end<OwnershipAttr>();
428           i != e; ++i)
429      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
430          (*i)->getOwnKind() == OwnershipAttr::Holds)
431        return true;
432  return false;
433}
434
435void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
436  if (C.wasInlined)
437    return;
438
439  const FunctionDecl *FD = C.getCalleeDecl(CE);
440  if (!FD)
441    return;
442
443  ProgramStateRef State = C.getState();
444  bool ReleasedAllocatedMemory = false;
445
446  if (FD->getKind() == Decl::Function) {
447    initIdentifierInfo(C.getASTContext());
448    IdentifierInfo *FunI = FD->getIdentifier();
449
450    if (FunI == II_malloc || FunI == II_valloc) {
451      if (CE->getNumArgs() < 1)
452        return;
453      State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
454    } else if (FunI == II_realloc) {
455      State = ReallocMem(C, CE, false);
456    } else if (FunI == II_reallocf) {
457      State = ReallocMem(C, CE, true);
458    } else if (FunI == II_calloc) {
459      State = CallocMem(C, CE);
460    } else if (FunI == II_free) {
461      State = FreeMemAux(C, CE, State, 0, false, ReleasedAllocatedMemory);
462    } else if (FunI == II_strdup) {
463      State = MallocUpdateRefState(C, CE, State);
464    } else if (FunI == II_strndup) {
465      State = MallocUpdateRefState(C, CE, State);
466    }
467  }
468
469  if (Filter.CMallocOptimistic) {
470    // Check all the attributes, if there are any.
471    // There can be multiple of these attributes.
472    if (FD->hasAttrs())
473      for (specific_attr_iterator<OwnershipAttr>
474          i = FD->specific_attr_begin<OwnershipAttr>(),
475          e = FD->specific_attr_end<OwnershipAttr>();
476          i != e; ++i) {
477        switch ((*i)->getOwnKind()) {
478        case OwnershipAttr::Returns:
479          State = MallocMemReturnsAttr(C, CE, *i);
480          break;
481        case OwnershipAttr::Takes:
482        case OwnershipAttr::Holds:
483          State = FreeMemAttr(C, CE, *i);
484          break;
485        }
486      }
487  }
488  C.addTransition(State);
489}
490
491static bool isFreeWhenDoneSetToZero(const ObjCMethodCall &Call) {
492  Selector S = Call.getSelector();
493  for (unsigned i = 1; i < S.getNumArgs(); ++i)
494    if (S.getNameForSlot(i).equals("freeWhenDone"))
495      if (Call.getArgSVal(i).isConstant(0))
496        return true;
497
498  return false;
499}
500
501void MallocChecker::checkPostObjCMessage(const ObjCMethodCall &Call,
502                                         CheckerContext &C) const {
503  // If the first selector is dataWithBytesNoCopy, assume that the memory will
504  // be released with 'free' by the new object.
505  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
506  // Unless 'freeWhenDone' param set to 0.
507  // TODO: Check that the memory was allocated with malloc.
508  bool ReleasedAllocatedMemory = false;
509  Selector S = Call.getSelector();
510  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
511       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
512       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
513      !isFreeWhenDoneSetToZero(Call)){
514    unsigned int argIdx  = 0;
515    ProgramStateRef State = FreeMemAux(C, Call.getArgExpr(argIdx),
516                                       Call.getOriginExpr(), C.getState(), true,
517                                       ReleasedAllocatedMemory,
518                                       /* RetNullOnFailure*/ true);
519
520    C.addTransition(State);
521  }
522}
523
524ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
525                                                    const CallExpr *CE,
526                                                    const OwnershipAttr* Att) {
527  if (Att->getModule() != "malloc")
528    return 0;
529
530  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
531  if (I != E) {
532    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
533  }
534  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
535}
536
537ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
538                                           const CallExpr *CE,
539                                           SVal Size, SVal Init,
540                                           ProgramStateRef state) {
541
542  // Bind the return value to the symbolic value from the heap region.
543  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
544  // side effects other than what we model here.
545  unsigned Count = C.blockCount();
546  SValBuilder &svalBuilder = C.getSValBuilder();
547  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
548  DefinedSVal RetVal =
549    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
550  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
551
552  // We expect the malloc functions to return a pointer.
553  if (!isa<Loc>(RetVal))
554    return 0;
555
556  // Fill the region with the initialization value.
557  state = state->bindDefault(RetVal, Init);
558
559  // Set the region's extent equal to the Size parameter.
560  const SymbolicRegion *R =
561      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
562  if (!R)
563    return 0;
564  if (isa<DefinedOrUnknownSVal>(Size)) {
565    SValBuilder &svalBuilder = C.getSValBuilder();
566    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
567    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
568    DefinedOrUnknownSVal extentMatchesSize =
569        svalBuilder.evalEQ(state, Extent, DefinedSize);
570
571    state = state->assume(extentMatchesSize, true);
572    assert(state);
573  }
574
575  return MallocUpdateRefState(C, CE, state);
576}
577
578ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
579                                                    const CallExpr *CE,
580                                                    ProgramStateRef state) {
581  // Get the return value.
582  SVal retVal = state->getSVal(CE, C.getLocationContext());
583
584  // We expect the malloc functions to return a pointer.
585  if (!isa<Loc>(retVal))
586    return 0;
587
588  SymbolRef Sym = retVal.getAsLocSymbol();
589  assert(Sym);
590
591  // Set the symbol's state to Allocated.
592  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
593
594}
595
596ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
597                                           const CallExpr *CE,
598                                           const OwnershipAttr* Att) const {
599  if (Att->getModule() != "malloc")
600    return 0;
601
602  ProgramStateRef State = C.getState();
603  bool ReleasedAllocated = false;
604
605  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
606       I != E; ++I) {
607    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
608                               Att->getOwnKind() == OwnershipAttr::Holds,
609                               ReleasedAllocated);
610    if (StateI)
611      State = StateI;
612  }
613  return State;
614}
615
616ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
617                                          const CallExpr *CE,
618                                          ProgramStateRef state,
619                                          unsigned Num,
620                                          bool Hold,
621                                          bool &ReleasedAllocated,
622                                          bool ReturnsNullOnFailure) const {
623  if (CE->getNumArgs() < (Num + 1))
624    return 0;
625
626  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold,
627                    ReleasedAllocated, ReturnsNullOnFailure);
628}
629
630/// Checks if the previous call to free on the given symbol failed - if free
631/// failed, returns true. Also, returns the corresponding return value symbol.
632bool didPreviousFreeFail(ProgramStateRef State,
633                         SymbolRef Sym, SymbolRef &RetStatusSymbol) {
634  const SymbolRef *Ret = State->get<FreeReturnValue>(Sym);
635  if (Ret) {
636    assert(*Ret && "We should not store the null return symbol");
637    ConstraintManager &CMgr = State->getConstraintManager();
638    ConditionTruthVal FreeFailed = CMgr.isNull(State, *Ret);
639    RetStatusSymbol = *Ret;
640    return FreeFailed.isConstrainedTrue();
641  }
642  return false;
643}
644
645ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
646                                          const Expr *ArgExpr,
647                                          const Expr *ParentExpr,
648                                          ProgramStateRef State,
649                                          bool Hold,
650                                          bool &ReleasedAllocated,
651                                          bool ReturnsNullOnFailure) const {
652
653  SVal ArgVal = State->getSVal(ArgExpr, C.getLocationContext());
654  if (!isa<DefinedOrUnknownSVal>(ArgVal))
655    return 0;
656  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
657
658  // Check for null dereferences.
659  if (!isa<Loc>(location))
660    return 0;
661
662  // The explicit NULL case, no operation is performed.
663  ProgramStateRef notNullState, nullState;
664  llvm::tie(notNullState, nullState) = State->assume(location);
665  if (nullState && !notNullState)
666    return 0;
667
668  // Unknown values could easily be okay
669  // Undefined values are handled elsewhere
670  if (ArgVal.isUnknownOrUndef())
671    return 0;
672
673  const MemRegion *R = ArgVal.getAsRegion();
674
675  // Nonlocs can't be freed, of course.
676  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
677  if (!R) {
678    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
679    return 0;
680  }
681
682  R = R->StripCasts();
683
684  // Blocks might show up as heap data, but should not be free()d
685  if (isa<BlockDataRegion>(R)) {
686    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
687    return 0;
688  }
689
690  const MemSpaceRegion *MS = R->getMemorySpace();
691
692  // Parameters, locals, statics, and globals shouldn't be freed.
693  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
694    // FIXME: at the time this code was written, malloc() regions were
695    // represented by conjured symbols, which are all in UnknownSpaceRegion.
696    // This means that there isn't actually anything from HeapSpaceRegion
697    // that should be freed, even though we allow it here.
698    // Of course, free() can work on memory allocated outside the current
699    // function, so UnknownSpaceRegion is always a possibility.
700    // False negatives are better than false positives.
701
702    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
703    return 0;
704  }
705
706  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
707  // Various cases could lead to non-symbol values here.
708  // For now, ignore them.
709  if (!SR)
710    return 0;
711
712  SymbolRef Sym = SR->getSymbol();
713  const RefState *RS = State->get<RegionState>(Sym);
714  SymbolRef PreviousRetStatusSymbol = 0;
715
716  // Check double free.
717  if (RS &&
718      (RS->isReleased() || RS->isRelinquished()) &&
719      !didPreviousFreeFail(State, Sym, PreviousRetStatusSymbol)) {
720
721    if (ExplodedNode *N = C.generateSink()) {
722      if (!BT_DoubleFree)
723        BT_DoubleFree.reset(
724          new BugType("Double free", "Memory Error"));
725      BugReport *R = new BugReport(*BT_DoubleFree,
726        (RS->isReleased() ? "Attempt to free released memory" :
727                            "Attempt to free non-owned memory"), N);
728      R->addRange(ArgExpr->getSourceRange());
729      R->markInteresting(Sym);
730      if (PreviousRetStatusSymbol)
731        R->markInteresting(PreviousRetStatusSymbol);
732      R->addVisitor(new MallocBugVisitor(Sym));
733      C.emitReport(R);
734    }
735    return 0;
736  }
737
738  ReleasedAllocated = (RS != 0);
739
740  // Clean out the info on previous call to free return info.
741  State = State->remove<FreeReturnValue>(Sym);
742
743  // Keep track of the return value. If it is NULL, we will know that free
744  // failed.
745  if (ReturnsNullOnFailure) {
746    SVal RetVal = C.getSVal(ParentExpr);
747    SymbolRef RetStatusSymbol = RetVal.getAsSymbol();
748    if (RetStatusSymbol) {
749      C.getSymbolManager().addSymbolDependency(Sym, RetStatusSymbol);
750      State = State->set<FreeReturnValue>(Sym, RetStatusSymbol);
751    }
752  }
753
754  // Normal free.
755  if (Hold)
756    return State->set<RegionState>(Sym, RefState::getRelinquished(ParentExpr));
757  return State->set<RegionState>(Sym, RefState::getReleased(ParentExpr));
758}
759
760bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
761  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
762    os << "an integer (" << IntVal->getValue() << ")";
763  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
764    os << "a constant address (" << ConstAddr->getValue() << ")";
765  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
766    os << "the address of the label '" << Label->getLabel()->getName() << "'";
767  else
768    return false;
769
770  return true;
771}
772
773bool MallocChecker::SummarizeRegion(raw_ostream &os,
774                                    const MemRegion *MR) {
775  switch (MR->getKind()) {
776  case MemRegion::FunctionTextRegionKind: {
777    const NamedDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
778    if (FD)
779      os << "the address of the function '" << *FD << '\'';
780    else
781      os << "the address of a function";
782    return true;
783  }
784  case MemRegion::BlockTextRegionKind:
785    os << "block text";
786    return true;
787  case MemRegion::BlockDataRegionKind:
788    // FIXME: where the block came from?
789    os << "a block";
790    return true;
791  default: {
792    const MemSpaceRegion *MS = MR->getMemorySpace();
793
794    if (isa<StackLocalsSpaceRegion>(MS)) {
795      const VarRegion *VR = dyn_cast<VarRegion>(MR);
796      const VarDecl *VD;
797      if (VR)
798        VD = VR->getDecl();
799      else
800        VD = NULL;
801
802      if (VD)
803        os << "the address of the local variable '" << VD->getName() << "'";
804      else
805        os << "the address of a local stack variable";
806      return true;
807    }
808
809    if (isa<StackArgumentsSpaceRegion>(MS)) {
810      const VarRegion *VR = dyn_cast<VarRegion>(MR);
811      const VarDecl *VD;
812      if (VR)
813        VD = VR->getDecl();
814      else
815        VD = NULL;
816
817      if (VD)
818        os << "the address of the parameter '" << VD->getName() << "'";
819      else
820        os << "the address of a parameter";
821      return true;
822    }
823
824    if (isa<GlobalsSpaceRegion>(MS)) {
825      const VarRegion *VR = dyn_cast<VarRegion>(MR);
826      const VarDecl *VD;
827      if (VR)
828        VD = VR->getDecl();
829      else
830        VD = NULL;
831
832      if (VD) {
833        if (VD->isStaticLocal())
834          os << "the address of the static variable '" << VD->getName() << "'";
835        else
836          os << "the address of the global variable '" << VD->getName() << "'";
837      } else
838        os << "the address of a global variable";
839      return true;
840    }
841
842    return false;
843  }
844  }
845}
846
847void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
848                                  SourceRange range) const {
849  if (ExplodedNode *N = C.generateSink()) {
850    if (!BT_BadFree)
851      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
852
853    SmallString<100> buf;
854    llvm::raw_svector_ostream os(buf);
855
856    const MemRegion *MR = ArgVal.getAsRegion();
857    if (MR) {
858      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
859        MR = ER->getSuperRegion();
860
861      // Special case for alloca()
862      if (isa<AllocaRegion>(MR))
863        os << "Argument to free() was allocated by alloca(), not malloc()";
864      else {
865        os << "Argument to free() is ";
866        if (SummarizeRegion(os, MR))
867          os << ", which is not memory allocated by malloc()";
868        else
869          os << "not memory allocated by malloc()";
870      }
871    } else {
872      os << "Argument to free() is ";
873      if (SummarizeValue(os, ArgVal))
874        os << ", which is not memory allocated by malloc()";
875      else
876        os << "not memory allocated by malloc()";
877    }
878
879    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
880    R->markInteresting(MR);
881    R->addRange(range);
882    C.emitReport(R);
883  }
884}
885
886ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
887                                          const CallExpr *CE,
888                                          bool FreesOnFail) const {
889  if (CE->getNumArgs() < 2)
890    return 0;
891
892  ProgramStateRef state = C.getState();
893  const Expr *arg0Expr = CE->getArg(0);
894  const LocationContext *LCtx = C.getLocationContext();
895  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
896  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
897    return 0;
898  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
899
900  SValBuilder &svalBuilder = C.getSValBuilder();
901
902  DefinedOrUnknownSVal PtrEQ =
903    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
904
905  // Get the size argument. If there is no size arg then give up.
906  const Expr *Arg1 = CE->getArg(1);
907  if (!Arg1)
908    return 0;
909
910  // Get the value of the size argument.
911  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
912  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
913    return 0;
914  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
915
916  // Compare the size argument to 0.
917  DefinedOrUnknownSVal SizeZero =
918    svalBuilder.evalEQ(state, Arg1Val,
919                       svalBuilder.makeIntValWithPtrWidth(0, false));
920
921  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
922  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
923  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
924  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
925  // We only assume exceptional states if they are definitely true; if the
926  // state is under-constrained, assume regular realloc behavior.
927  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
928  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
929
930  // If the ptr is NULL and the size is not 0, the call is equivalent to
931  // malloc(size).
932  if ( PrtIsNull && !SizeIsZero) {
933    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
934                                               UndefinedVal(), StatePtrIsNull);
935    return stateMalloc;
936  }
937
938  if (PrtIsNull && SizeIsZero)
939    return 0;
940
941  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
942  assert(!PrtIsNull);
943  SymbolRef FromPtr = arg0Val.getAsSymbol();
944  SVal RetVal = state->getSVal(CE, LCtx);
945  SymbolRef ToPtr = RetVal.getAsSymbol();
946  if (!FromPtr || !ToPtr)
947    return 0;
948
949  bool ReleasedAllocated = false;
950
951  // If the size is 0, free the memory.
952  if (SizeIsZero)
953    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero, 0,
954                                               false, ReleasedAllocated)){
955      // The semantics of the return value are:
956      // If size was equal to 0, either NULL or a pointer suitable to be passed
957      // to free() is returned. We just free the input pointer and do not add
958      // any constrains on the output pointer.
959      return stateFree;
960    }
961
962  // Default behavior.
963  if (ProgramStateRef stateFree =
964        FreeMemAux(C, CE, state, 0, false, ReleasedAllocated)) {
965
966    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
967                                                UnknownVal(), stateFree);
968    if (!stateRealloc)
969      return 0;
970
971    ReallocPairKind Kind = RPToBeFreedAfterFailure;
972    if (FreesOnFail)
973      Kind = RPIsFreeOnFailure;
974    else if (!ReleasedAllocated)
975      Kind = RPDoNotTrackAfterFailure;
976
977    // Record the info about the reallocated symbol so that we could properly
978    // process failed reallocation.
979    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
980                                                   ReallocPair(FromPtr, Kind));
981    // The reallocated symbol should stay alive for as long as the new symbol.
982    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
983    return stateRealloc;
984  }
985  return 0;
986}
987
988ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
989  if (CE->getNumArgs() < 2)
990    return 0;
991
992  ProgramStateRef state = C.getState();
993  SValBuilder &svalBuilder = C.getSValBuilder();
994  const LocationContext *LCtx = C.getLocationContext();
995  SVal count = state->getSVal(CE->getArg(0), LCtx);
996  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
997  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
998                                        svalBuilder.getContext().getSizeType());
999  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
1000
1001  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
1002}
1003
1004LeakInfo
1005MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
1006                                 CheckerContext &C) const {
1007  const LocationContext *LeakContext = N->getLocationContext();
1008  // Walk the ExplodedGraph backwards and find the first node that referred to
1009  // the tracked symbol.
1010  const ExplodedNode *AllocNode = N;
1011  const MemRegion *ReferenceRegion = 0;
1012
1013  while (N) {
1014    ProgramStateRef State = N->getState();
1015    if (!State->get<RegionState>(Sym))
1016      break;
1017
1018    // Find the most recent expression bound to the symbol in the current
1019    // context.
1020    if (!ReferenceRegion) {
1021      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
1022        SVal Val = State->getSVal(MR);
1023        if (Val.getAsLocSymbol() == Sym)
1024          ReferenceRegion = MR;
1025      }
1026    }
1027
1028    // Allocation node, is the last node in the current context in which the
1029    // symbol was tracked.
1030    if (N->getLocationContext() == LeakContext)
1031      AllocNode = N;
1032    N = N->pred_empty() ? NULL : *(N->pred_begin());
1033  }
1034
1035  ProgramPoint P = AllocNode->getLocation();
1036  const Stmt *AllocationStmt = 0;
1037  if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&P))
1038    AllocationStmt = Exit->getCalleeContext()->getCallSite();
1039  else if (StmtPoint *SP = dyn_cast<StmtPoint>(&P))
1040    AllocationStmt = SP->getStmt();
1041
1042  return LeakInfo(AllocationStmt, ReferenceRegion);
1043}
1044
1045void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
1046                               CheckerContext &C) const {
1047  assert(N);
1048  if (!BT_Leak) {
1049    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
1050    // Leaks should not be reported if they are post-dominated by a sink:
1051    // (1) Sinks are higher importance bugs.
1052    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
1053    //     with __noreturn functions such as assert() or exit(). We choose not
1054    //     to report leaks on such paths.
1055    BT_Leak->setSuppressOnSink(true);
1056  }
1057
1058  // Most bug reports are cached at the location where they occurred.
1059  // With leaks, we want to unique them by the location where they were
1060  // allocated, and only report a single path.
1061  PathDiagnosticLocation LocUsedForUniqueing;
1062  const Stmt *AllocStmt = 0;
1063  const MemRegion *Region = 0;
1064  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
1065  if (AllocStmt)
1066    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
1067                            C.getSourceManager(), N->getLocationContext());
1068
1069  SmallString<200> buf;
1070  llvm::raw_svector_ostream os(buf);
1071  os << "Memory is never released; potential leak";
1072  if (Region && Region->canPrintPretty()) {
1073    os << " of memory pointed to by '";
1074    Region->printPretty(os);
1075    os << '\'';
1076  }
1077
1078  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
1079  R->markInteresting(Sym);
1080  R->addVisitor(new MallocBugVisitor(Sym, true));
1081  C.emitReport(R);
1082}
1083
1084void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1085                                     CheckerContext &C) const
1086{
1087  if (!SymReaper.hasDeadSymbols())
1088    return;
1089
1090  ProgramStateRef state = C.getState();
1091  RegionStateTy RS = state->get<RegionState>();
1092  RegionStateTy::Factory &F = state->get_context<RegionState>();
1093
1094  llvm::SmallVector<SymbolRef, 2> Errors;
1095  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1096    if (SymReaper.isDead(I->first)) {
1097      if (I->second.isAllocated())
1098        Errors.push_back(I->first);
1099      // Remove the dead symbol from the map.
1100      RS = F.remove(RS, I->first);
1101
1102    }
1103  }
1104
1105  // Cleanup the Realloc Pairs Map.
1106  ReallocPairsTy RP = state->get<ReallocPairs>();
1107  for (ReallocPairsTy::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1108    if (SymReaper.isDead(I->first) ||
1109        SymReaper.isDead(I->second.ReallocatedSym)) {
1110      state = state->remove<ReallocPairs>(I->first);
1111    }
1112  }
1113
1114  // Cleanup the FreeReturnValue Map.
1115  FreeReturnValueTy FR = state->get<FreeReturnValue>();
1116  for (FreeReturnValueTy::iterator I = FR.begin(), E = FR.end(); I != E; ++I) {
1117    if (SymReaper.isDead(I->first) ||
1118        SymReaper.isDead(I->second)) {
1119      state = state->remove<FreeReturnValue>(I->first);
1120    }
1121  }
1122
1123  // Generate leak node.
1124  ExplodedNode *N = C.getPredecessor();
1125  if (!Errors.empty()) {
1126    static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1127    N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1128    for (llvm::SmallVector<SymbolRef, 2>::iterator
1129        I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1130      reportLeak(*I, N, C);
1131    }
1132  }
1133
1134  C.addTransition(state->set<RegionState>(RS), N);
1135}
1136
1137void MallocChecker::checkEndPath(CheckerContext &C) const {
1138  ProgramStateRef state = C.getState();
1139  RegionStateTy M = state->get<RegionState>();
1140
1141  // If inside inlined call, skip it.
1142  if (C.getLocationContext()->getParent() != 0)
1143    return;
1144
1145  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1146    RefState RS = I->second;
1147    if (RS.isAllocated()) {
1148      ExplodedNode *N = C.addTransition(state);
1149      if (N)
1150        reportLeak(I->first, N, C);
1151    }
1152  }
1153}
1154
1155void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1156  // We will check for double free in the post visit.
1157  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1158    return;
1159
1160  // Check use after free, when a freed pointer is passed to a call.
1161  ProgramStateRef State = C.getState();
1162  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1163                                    E = CE->arg_end(); I != E; ++I) {
1164    const Expr *A = *I;
1165    if (A->getType().getTypePtr()->isAnyPointerType()) {
1166      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1167      if (!Sym)
1168        continue;
1169      if (checkUseAfterFree(Sym, C, A))
1170        return;
1171    }
1172  }
1173}
1174
1175void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1176  const Expr *E = S->getRetValue();
1177  if (!E)
1178    return;
1179
1180  // Check if we are returning a symbol.
1181  ProgramStateRef State = C.getState();
1182  SVal RetVal = State->getSVal(E, C.getLocationContext());
1183  SymbolRef Sym = RetVal.getAsSymbol();
1184  if (!Sym)
1185    // If we are returning a field of the allocated struct or an array element,
1186    // the callee could still free the memory.
1187    // TODO: This logic should be a part of generic symbol escape callback.
1188    if (const MemRegion *MR = RetVal.getAsRegion())
1189      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1190        if (const SymbolicRegion *BMR =
1191              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1192          Sym = BMR->getSymbol();
1193
1194  // Check if we are returning freed memory.
1195  if (Sym)
1196    if (checkUseAfterFree(Sym, C, E))
1197      return;
1198
1199  // If this function body is not inlined, stop tracking any returned symbols.
1200  if (C.getLocationContext()->getParent() == 0) {
1201    State =
1202      State->scanReachableSymbols<StopTrackingCallback>(RetVal).getState();
1203    C.addTransition(State);
1204  }
1205}
1206
1207// TODO: Blocks should be either inlined or should call invalidate regions
1208// upon invocation. After that's in place, special casing here will not be
1209// needed.
1210void MallocChecker::checkPostStmt(const BlockExpr *BE,
1211                                  CheckerContext &C) const {
1212
1213  // Scan the BlockDecRefExprs for any object the retain count checker
1214  // may be tracking.
1215  if (!BE->getBlockDecl()->hasCaptures())
1216    return;
1217
1218  ProgramStateRef state = C.getState();
1219  const BlockDataRegion *R =
1220    cast<BlockDataRegion>(state->getSVal(BE,
1221                                         C.getLocationContext()).getAsRegion());
1222
1223  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1224                                            E = R->referenced_vars_end();
1225
1226  if (I == E)
1227    return;
1228
1229  SmallVector<const MemRegion*, 10> Regions;
1230  const LocationContext *LC = C.getLocationContext();
1231  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1232
1233  for ( ; I != E; ++I) {
1234    const VarRegion *VR = *I;
1235    if (VR->getSuperRegion() == R) {
1236      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1237    }
1238    Regions.push_back(VR);
1239  }
1240
1241  state =
1242    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1243                                    Regions.data() + Regions.size()).getState();
1244  C.addTransition(state);
1245}
1246
1247bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1248  assert(Sym);
1249  const RefState *RS = C.getState()->get<RegionState>(Sym);
1250  return (RS && RS->isReleased());
1251}
1252
1253bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1254                                      const Stmt *S) const {
1255  if (isReleased(Sym, C)) {
1256    if (ExplodedNode *N = C.generateSink()) {
1257      if (!BT_UseFree)
1258        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1259
1260      BugReport *R = new BugReport(*BT_UseFree,
1261                                   "Use of memory after it is freed",N);
1262      if (S)
1263        R->addRange(S->getSourceRange());
1264      R->markInteresting(Sym);
1265      R->addVisitor(new MallocBugVisitor(Sym));
1266      C.emitReport(R);
1267      return true;
1268    }
1269  }
1270  return false;
1271}
1272
1273// Check if the location is a freed symbolic region.
1274void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1275                                  CheckerContext &C) const {
1276  SymbolRef Sym = l.getLocSymbolInBase();
1277  if (Sym)
1278    checkUseAfterFree(Sym, C, S);
1279}
1280
1281//===----------------------------------------------------------------------===//
1282// Check various ways a symbol can be invalidated.
1283// TODO: This logic (the next 3 functions) is copied/similar to the
1284// RetainRelease checker. We might want to factor this out.
1285//===----------------------------------------------------------------------===//
1286
1287// Stop tracking symbols when a value escapes as a result of checkBind.
1288// A value escapes in three possible cases:
1289// (1) we are binding to something that is not a memory region.
1290// (2) we are binding to a memregion that does not have stack storage
1291// (3) we are binding to a memregion with stack storage that the store
1292//     does not understand.
1293void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1294                              CheckerContext &C) const {
1295  // Are we storing to something that causes the value to "escape"?
1296  bool escapes = true;
1297  ProgramStateRef state = C.getState();
1298
1299  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1300    escapes = !regionLoc->getRegion()->hasStackStorage();
1301
1302    if (!escapes) {
1303      // To test (3), generate a new state with the binding added.  If it is
1304      // the same state, then it escapes (since the store cannot represent
1305      // the binding).
1306      // Do this only if we know that the store is not supposed to generate the
1307      // same state.
1308      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1309      if (StoredVal != val)
1310        escapes = (state == (state->bindLoc(*regionLoc, val)));
1311    }
1312  }
1313
1314  // If our store can represent the binding and we aren't storing to something
1315  // that doesn't have local storage then just return and have the simulation
1316  // state continue as is.
1317  if (!escapes)
1318      return;
1319
1320  // Otherwise, find all symbols referenced by 'val' that we are tracking
1321  // and stop tracking them.
1322  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1323  C.addTransition(state);
1324}
1325
1326// If a symbolic region is assumed to NULL (or another constant), stop tracking
1327// it - assuming that allocation failed on this path.
1328ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1329                                              SVal Cond,
1330                                              bool Assumption) const {
1331  RegionStateTy RS = state->get<RegionState>();
1332  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1333    // If the symbol is assumed to be NULL, remove it from consideration.
1334    ConstraintManager &CMgr = state->getConstraintManager();
1335    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1336    if (AllocFailed.isConstrainedTrue())
1337      state = state->remove<RegionState>(I.getKey());
1338  }
1339
1340  // Realloc returns 0 when reallocation fails, which means that we should
1341  // restore the state of the pointer being reallocated.
1342  ReallocPairsTy RP = state->get<ReallocPairs>();
1343  for (ReallocPairsTy::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1344    // If the symbol is assumed to be NULL, remove it from consideration.
1345    ConstraintManager &CMgr = state->getConstraintManager();
1346    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1347    if (!AllocFailed.isConstrainedTrue())
1348      continue;
1349
1350    SymbolRef ReallocSym = I.getData().ReallocatedSym;
1351    if (const RefState *RS = state->get<RegionState>(ReallocSym)) {
1352      if (RS->isReleased()) {
1353        if (I.getData().Kind == RPToBeFreedAfterFailure)
1354          state = state->set<RegionState>(ReallocSym,
1355              RefState::getAllocated(RS->getStmt()));
1356        else if (I.getData().Kind == RPDoNotTrackAfterFailure)
1357          state = state->remove<RegionState>(ReallocSym);
1358        else
1359          assert(I.getData().Kind == RPIsFreeOnFailure);
1360      }
1361    }
1362    state = state->remove<ReallocPairs>(I.getKey());
1363  }
1364
1365  return state;
1366}
1367
1368// Check if the function is known to us. So, for example, we could
1369// conservatively assume it can free/reallocate its pointer arguments.
1370// (We assume that the pointers cannot escape through calls to system
1371// functions not handled by this checker.)
1372bool MallocChecker::doesNotFreeMemory(const CallEvent *Call,
1373                                      ProgramStateRef State) const {
1374  assert(Call);
1375
1376  // For now, assume that any C++ call can free memory.
1377  // TODO: If we want to be more optimistic here, we'll need to make sure that
1378  // regions escape to C++ containers. They seem to do that even now, but for
1379  // mysterious reasons.
1380  if (!(isa<FunctionCall>(Call) || isa<ObjCMethodCall>(Call)))
1381    return false;
1382
1383  // Check Objective-C messages by selector name.
1384  if (const ObjCMethodCall *Msg = dyn_cast<ObjCMethodCall>(Call)) {
1385    // If it's not a framework call, or if it takes a callback, assume it
1386    // can free memory.
1387    if (!Call->isInSystemHeader() || Call->hasNonZeroCallbackArg())
1388      return false;
1389
1390    Selector S = Msg->getSelector();
1391
1392    // Whitelist the ObjC methods which do free memory.
1393    // - Anything containing 'freeWhenDone' param set to 1.
1394    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1395    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1396      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1397        if (Call->getArgSVal(i).isConstant(1))
1398          return false;
1399        else
1400          return true;
1401      }
1402    }
1403
1404    // If the first selector ends with NoCopy, assume that the ownership is
1405    // transferred as well.
1406    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1407    StringRef FirstSlot = S.getNameForSlot(0);
1408    if (FirstSlot.endswith("NoCopy"))
1409      return false;
1410
1411    // If the first selector starts with addPointer, insertPointer,
1412    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1413    // This is similar to C++ containers (vector); we still might want to check
1414    // that the pointers get freed by following the container itself.
1415    if (FirstSlot.startswith("addPointer") ||
1416        FirstSlot.startswith("insertPointer") ||
1417        FirstSlot.startswith("replacePointer")) {
1418      return false;
1419    }
1420
1421    // Otherwise, assume that the method does not free memory.
1422    // Most framework methods do not free memory.
1423    return true;
1424  }
1425
1426  // At this point the only thing left to handle is straight function calls.
1427  const FunctionDecl *FD = cast<FunctionCall>(Call)->getDecl();
1428  if (!FD)
1429    return false;
1430
1431  ASTContext &ASTC = State->getStateManager().getContext();
1432
1433  // If it's one of the allocation functions we can reason about, we model
1434  // its behavior explicitly.
1435  if (isMemFunction(FD, ASTC))
1436    return true;
1437
1438  // If it's not a system call, assume it frees memory.
1439  if (!Call->isInSystemHeader())
1440    return false;
1441
1442  // White list the system functions whose arguments escape.
1443  const IdentifierInfo *II = FD->getIdentifier();
1444  if (!II)
1445    return false;
1446  StringRef FName = II->getName();
1447
1448  // White list the 'XXXNoCopy' CoreFoundation functions.
1449  // We specifically check these before
1450  if (FName.endswith("NoCopy")) {
1451    // Look for the deallocator argument. We know that the memory ownership
1452    // is not transferred only if the deallocator argument is
1453    // 'kCFAllocatorNull'.
1454    for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1455      const Expr *ArgE = Call->getArgExpr(i)->IgnoreParenCasts();
1456      if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1457        StringRef DeallocatorName = DE->getFoundDecl()->getName();
1458        if (DeallocatorName == "kCFAllocatorNull")
1459          return true;
1460      }
1461    }
1462    return false;
1463  }
1464
1465  // Associating streams with malloced buffers. The pointer can escape if
1466  // 'closefn' is specified (and if that function does free memory),
1467  // but it will not if closefn is not specified.
1468  // Currently, we do not inspect the 'closefn' function (PR12101).
1469  if (FName == "funopen")
1470    if (Call->getNumArgs() >= 4 && Call->getArgSVal(4).isConstant(0))
1471      return true;
1472
1473  // Do not warn on pointers passed to 'setbuf' when used with std streams,
1474  // these leaks might be intentional when setting the buffer for stdio.
1475  // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1476  if (FName == "setbuf" || FName =="setbuffer" ||
1477      FName == "setlinebuf" || FName == "setvbuf") {
1478    if (Call->getNumArgs() >= 1) {
1479      const Expr *ArgE = Call->getArgExpr(0)->IgnoreParenCasts();
1480      if (const DeclRefExpr *ArgDRE = dyn_cast<DeclRefExpr>(ArgE))
1481        if (const VarDecl *D = dyn_cast<VarDecl>(ArgDRE->getDecl()))
1482          if (D->getCanonicalDecl()->getName().find("std") != StringRef::npos)
1483            return false;
1484    }
1485  }
1486
1487  // A bunch of other functions which either take ownership of a pointer or
1488  // wrap the result up in a struct or object, meaning it can be freed later.
1489  // (See RetainCountChecker.) Not all the parameters here are invalidated,
1490  // but the Malloc checker cannot differentiate between them. The right way
1491  // of doing this would be to implement a pointer escapes callback.
1492  if (FName == "CGBitmapContextCreate" ||
1493      FName == "CGBitmapContextCreateWithData" ||
1494      FName == "CVPixelBufferCreateWithBytes" ||
1495      FName == "CVPixelBufferCreateWithPlanarBytes" ||
1496      FName == "OSAtomicEnqueue") {
1497    return false;
1498  }
1499
1500  // Handle cases where we know a buffer's /address/ can escape.
1501  // Note that the above checks handle some special cases where we know that
1502  // even though the address escapes, it's still our responsibility to free the
1503  // buffer.
1504  if (Call->argumentsMayEscape())
1505    return false;
1506
1507  // Otherwise, assume that the function does not free memory.
1508  // Most system calls do not free the memory.
1509  return true;
1510}
1511
1512// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1513// escapes, when we are tracking p), do not track the symbol as we cannot reason
1514// about it anymore.
1515ProgramStateRef
1516MallocChecker::checkRegionChanges(ProgramStateRef State,
1517                            const StoreManager::InvalidatedSymbols *invalidated,
1518                                    ArrayRef<const MemRegion *> ExplicitRegions,
1519                                    ArrayRef<const MemRegion *> Regions,
1520                                    const CallEvent *Call) const {
1521  if (!invalidated || invalidated->empty())
1522    return State;
1523  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1524
1525  // If it's a call which might free or reallocate memory, we assume that all
1526  // regions (explicit and implicit) escaped.
1527
1528  // Otherwise, whitelist explicit pointers; we still can track them.
1529  if (!Call || doesNotFreeMemory(Call, State)) {
1530    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1531        E = ExplicitRegions.end(); I != E; ++I) {
1532      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1533        WhitelistedSymbols.insert(R->getSymbol());
1534    }
1535  }
1536
1537  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1538       E = invalidated->end(); I!=E; ++I) {
1539    SymbolRef sym = *I;
1540    if (WhitelistedSymbols.count(sym))
1541      continue;
1542    // The symbol escaped. Note, we assume that if the symbol is released,
1543    // passing it out will result in a use after free. We also keep tracking
1544    // relinquished symbols.
1545    if (const RefState *RS = State->get<RegionState>(sym)) {
1546      if (RS->isAllocated())
1547        State = State->remove<RegionState>(sym);
1548    }
1549  }
1550  return State;
1551}
1552
1553static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1554                                         ProgramStateRef prevState) {
1555  ReallocPairsTy currMap = currState->get<ReallocPairs>();
1556  ReallocPairsTy prevMap = prevState->get<ReallocPairs>();
1557
1558  for (ReallocPairsTy::iterator I = prevMap.begin(), E = prevMap.end();
1559       I != E; ++I) {
1560    SymbolRef sym = I.getKey();
1561    if (!currMap.lookup(sym))
1562      return sym;
1563  }
1564
1565  return NULL;
1566}
1567
1568PathDiagnosticPiece *
1569MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1570                                           const ExplodedNode *PrevN,
1571                                           BugReporterContext &BRC,
1572                                           BugReport &BR) {
1573  ProgramStateRef state = N->getState();
1574  ProgramStateRef statePrev = PrevN->getState();
1575
1576  const RefState *RS = state->get<RegionState>(Sym);
1577  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1578  if (!RS)
1579    return 0;
1580
1581  const Stmt *S = 0;
1582  const char *Msg = 0;
1583  StackHintGeneratorForSymbol *StackHint = 0;
1584
1585  // Retrieve the associated statement.
1586  ProgramPoint ProgLoc = N->getLocation();
1587  if (StmtPoint *SP = dyn_cast<StmtPoint>(&ProgLoc))
1588    S = SP->getStmt();
1589  else if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&ProgLoc))
1590    S = Exit->getCalleeContext()->getCallSite();
1591  // If an assumption was made on a branch, it should be caught
1592  // here by looking at the state transition.
1593  else if (BlockEdge *Edge = dyn_cast<BlockEdge>(&ProgLoc)) {
1594    const CFGBlock *srcBlk = Edge->getSrc();
1595    S = srcBlk->getTerminator();
1596  }
1597  if (!S)
1598    return 0;
1599
1600  // FIXME: We will eventually need to handle non-statement-based events
1601  // (__attribute__((cleanup))).
1602
1603  // Find out if this is an interesting point and what is the kind.
1604  if (Mode == Normal) {
1605    if (isAllocated(RS, RSPrev, S)) {
1606      Msg = "Memory is allocated";
1607      StackHint = new StackHintGeneratorForSymbol(Sym,
1608                                                  "Returned allocated memory");
1609    } else if (isReleased(RS, RSPrev, S)) {
1610      Msg = "Memory is released";
1611      StackHint = new StackHintGeneratorForSymbol(Sym,
1612                                                  "Returned released memory");
1613    } else if (isRelinquished(RS, RSPrev, S)) {
1614      Msg = "Memory ownership is transfered";
1615      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1616    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1617      Mode = ReallocationFailed;
1618      Msg = "Reallocation failed";
1619      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1620                                                       "Reallocation failed");
1621
1622      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1623        // Is it possible to fail two reallocs WITHOUT testing in between?
1624        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1625          "We only support one failed realloc at a time.");
1626        BR.markInteresting(sym);
1627        FailedReallocSymbol = sym;
1628      }
1629    }
1630
1631  // We are in a special mode if a reallocation failed later in the path.
1632  } else if (Mode == ReallocationFailed) {
1633    assert(FailedReallocSymbol && "No symbol to look for.");
1634
1635    // Is this is the first appearance of the reallocated symbol?
1636    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1637      // We're at the reallocation point.
1638      Msg = "Attempt to reallocate memory";
1639      StackHint = new StackHintGeneratorForSymbol(Sym,
1640                                                 "Returned reallocated memory");
1641      FailedReallocSymbol = NULL;
1642      Mode = Normal;
1643    }
1644  }
1645
1646  if (!Msg)
1647    return 0;
1648  assert(StackHint);
1649
1650  // Generate the extra diagnostic.
1651  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1652                             N->getLocationContext());
1653  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1654}
1655
1656void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1657                               const char *NL, const char *Sep) const {
1658
1659  RegionStateTy RS = State->get<RegionState>();
1660
1661  if (!RS.isEmpty())
1662    Out << "Has Malloc data" << NL;
1663}
1664
1665#define REGISTER_CHECKER(name) \
1666void ento::register##name(CheckerManager &mgr) {\
1667  registerCStringCheckerBasic(mgr); \
1668  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1669}
1670
1671REGISTER_CHECKER(MallocPessimistic)
1672REGISTER_CHECKER(MallocOptimistic)
1673