MallocChecker.cpp revision 66c486f275531df6362b3511fc3af6563561801b 
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 <climits>
30
31using namespace clang;
32using namespace ento;
33
34namespace {
35
36class RefState {
37  enum Kind { // Reference to allocated memory.
38              Allocated,
39              // Reference to released/freed memory.
40              Released,
41              // The responsibility for freeing resources has transfered from
42              // this reference. A relinquished symbol should not be freed.
43              Relinquished } K;
44  const Stmt *S;
45
46public:
47  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
48
49  bool isAllocated() const { return K == Allocated; }
50  bool isReleased() const { return K == Released; }
51  bool isRelinquished() const { return K == Relinquished; }
52
53  const Stmt *getStmt() const { return S; }
54
55  bool operator==(const RefState &X) const {
56    return K == X.K && S == X.S;
57  }
58
59  static RefState getAllocated(const Stmt *s) {
60    return RefState(Allocated, s);
61  }
62  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
63  static RefState getRelinquished(const Stmt *s) {
64    return RefState(Relinquished, s);
65  }
66
67  void Profile(llvm::FoldingSetNodeID &ID) const {
68    ID.AddInteger(K);
69    ID.AddPointer(S);
70  }
71};
72
73struct ReallocPair {
74  SymbolRef ReallocatedSym;
75  bool IsFreeOnFailure;
76  ReallocPair(SymbolRef S, bool F) : ReallocatedSym(S), IsFreeOnFailure(F) {}
77  void Profile(llvm::FoldingSetNodeID &ID) const {
78    ID.AddInteger(IsFreeOnFailure);
79    ID.AddPointer(ReallocatedSym);
80  }
81  bool operator==(const ReallocPair &X) const {
82    return ReallocatedSym == X.ReallocatedSym &&
83           IsFreeOnFailure == X.IsFreeOnFailure;
84  }
85};
86
87typedef std::pair<const Stmt*, const MemRegion*> LeakInfo;
88
89class MallocChecker : public Checker<check::DeadSymbols,
90                                     check::EndPath,
91                                     check::PreStmt<ReturnStmt>,
92                                     check::PreStmt<CallExpr>,
93                                     check::PostStmt<CallExpr>,
94                                     check::PostStmt<BlockExpr>,
95                                     check::PreObjCMessage,
96                                     check::Location,
97                                     check::Bind,
98                                     eval::Assume,
99                                     check::RegionChanges>
100{
101  mutable OwningPtr<BugType> BT_DoubleFree;
102  mutable OwningPtr<BugType> BT_Leak;
103  mutable OwningPtr<BugType> BT_UseFree;
104  mutable OwningPtr<BugType> BT_BadFree;
105  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
106                         *II_valloc, *II_reallocf, *II_strndup, *II_strdup;
107
108public:
109  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
110                    II_valloc(0), II_reallocf(0), II_strndup(0), II_strdup(0) {}
111
112  /// In pessimistic mode, the checker assumes that it does not know which
113  /// functions might free the memory.
114  struct ChecksFilter {
115    DefaultBool CMallocPessimistic;
116    DefaultBool CMallocOptimistic;
117  };
118
119  ChecksFilter Filter;
120
121  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
122  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
123  void checkPreObjCMessage(const ObjCMethodCall &Call, CheckerContext &C) const;
124  void checkPostStmt(const BlockExpr *BE, CheckerContext &C) const;
125  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
126  void checkEndPath(CheckerContext &C) const;
127  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
128  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
129                            bool Assumption) const;
130  void checkLocation(SVal l, bool isLoad, const Stmt *S,
131                     CheckerContext &C) const;
132  void checkBind(SVal location, SVal val, const Stmt*S,
133                 CheckerContext &C) const;
134  ProgramStateRef
135  checkRegionChanges(ProgramStateRef state,
136                     const StoreManager::InvalidatedSymbols *invalidated,
137                     ArrayRef<const MemRegion *> ExplicitRegions,
138                     ArrayRef<const MemRegion *> Regions,
139                     const CallEvent *Call) const;
140  bool wantsRegionChangeUpdate(ProgramStateRef state) const {
141    return true;
142  }
143
144  void printState(raw_ostream &Out, ProgramStateRef State,
145                  const char *NL, const char *Sep) const;
146
147private:
148  void initIdentifierInfo(ASTContext &C) const;
149
150  /// Check if this is one of the functions which can allocate/reallocate memory
151  /// pointed to by one of its arguments.
152  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
153  bool isFreeFunction(const FunctionDecl *FD, ASTContext &C) const;
154  bool isAllocationFunction(const FunctionDecl *FD, ASTContext &C) const;
155
156  static ProgramStateRef MallocMemReturnsAttr(CheckerContext &C,
157                                              const CallExpr *CE,
158                                              const OwnershipAttr* Att);
159  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
160                                     const Expr *SizeEx, SVal Init,
161                                     ProgramStateRef state) {
162    return MallocMemAux(C, CE,
163                        state->getSVal(SizeEx, C.getLocationContext()),
164                        Init, state);
165  }
166
167  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
168                                     SVal SizeEx, SVal Init,
169                                     ProgramStateRef state);
170
171  /// Update the RefState to reflect the new memory allocation.
172  static ProgramStateRef MallocUpdateRefState(CheckerContext &C,
173                                              const CallExpr *CE,
174                                              ProgramStateRef state);
175
176  ProgramStateRef FreeMemAttr(CheckerContext &C, const CallExpr *CE,
177                              const OwnershipAttr* Att) const;
178  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
179                             ProgramStateRef state, unsigned Num,
180                             bool Hold) const;
181  ProgramStateRef FreeMemAux(CheckerContext &C, const Expr *Arg,
182                             const Expr *ParentExpr,
183                             ProgramStateRef state,
184                             bool Hold) const;
185
186  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
187                             bool FreesMemOnFailure) const;
188  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
189
190  ///\brief Check if the memory associated with this symbol was released.
191  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
192
193  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
194                         const Stmt *S = 0) const;
195
196  /// Check if the function is not known to us. So, for example, we could
197  /// conservatively assume it can free/reallocate it's pointer arguments.
198  bool doesNotFreeMemory(const CallEvent *Call,
199                         ProgramStateRef State) const;
200
201  static bool SummarizeValue(raw_ostream &os, SVal V);
202  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
203  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
204
205  /// Find the location of the allocation for Sym on the path leading to the
206  /// exploded node N.
207  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
208                             CheckerContext &C) const;
209
210  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
211
212  /// The bug visitor which allows us to print extra diagnostics along the
213  /// BugReport path. For example, showing the allocation site of the leaked
214  /// region.
215  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
216  protected:
217    enum NotificationMode {
218      Normal,
219      ReallocationFailed
220    };
221
222    // The allocated region symbol tracked by the main analysis.
223    SymbolRef Sym;
224
225    // The mode we are in, i.e. what kind of diagnostics will be emitted.
226    NotificationMode Mode;
227
228    // A symbol from when the primary region should have been reallocated.
229    SymbolRef FailedReallocSymbol;
230
231    bool IsLeak;
232
233  public:
234    MallocBugVisitor(SymbolRef S, bool isLeak = false)
235       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
236
237    virtual ~MallocBugVisitor() {}
238
239    void Profile(llvm::FoldingSetNodeID &ID) const {
240      static int X = 0;
241      ID.AddPointer(&X);
242      ID.AddPointer(Sym);
243    }
244
245    inline bool isAllocated(const RefState *S, const RefState *SPrev,
246                            const Stmt *Stmt) {
247      // Did not track -> allocated. Other state (released) -> allocated.
248      return (Stmt && isa<CallExpr>(Stmt) &&
249              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
250    }
251
252    inline bool isReleased(const RefState *S, const RefState *SPrev,
253                           const Stmt *Stmt) {
254      // Did not track -> released. Other state (allocated) -> released.
255      return (Stmt && isa<CallExpr>(Stmt) &&
256              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
257    }
258
259    inline bool isRelinquished(const RefState *S, const RefState *SPrev,
260                               const Stmt *Stmt) {
261      // Did not track -> relinquished. Other state (allocated) -> relinquished.
262      return (Stmt && (isa<CallExpr>(Stmt) || isa<ObjCMessageExpr>(Stmt) ||
263                                              isa<ObjCPropertyRefExpr>(Stmt)) &&
264              (S && S->isRelinquished()) &&
265              (!SPrev || !SPrev->isRelinquished()));
266    }
267
268    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
269                                     const Stmt *Stmt) {
270      // If the expression is not a call, and the state change is
271      // released -> allocated, it must be the realloc return value
272      // check. If we have to handle more cases here, it might be cleaner just
273      // to track this extra bit in the state itself.
274      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
275              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
276    }
277
278    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
279                                   const ExplodedNode *PrevN,
280                                   BugReporterContext &BRC,
281                                   BugReport &BR);
282
283    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
284                                    const ExplodedNode *EndPathNode,
285                                    BugReport &BR) {
286      if (!IsLeak)
287        return 0;
288
289      PathDiagnosticLocation L =
290        PathDiagnosticLocation::createEndOfPath(EndPathNode,
291                                                BRC.getSourceManager());
292      // Do not add the statement itself as a range in case of leak.
293      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
294    }
295
296  private:
297    class StackHintGeneratorForReallocationFailed
298        : public StackHintGeneratorForSymbol {
299    public:
300      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
301        : StackHintGeneratorForSymbol(S, M) {}
302
303      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
304        SmallString<200> buf;
305        llvm::raw_svector_ostream os(buf);
306
307        os << "Reallocation of ";
308        // Printed parameters start at 1, not 0.
309        printOrdinal(++ArgIndex, os);
310        os << " parameter failed";
311
312        return os.str();
313      }
314
315      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
316        return "Reallocation of returned value failed";
317      }
318    };
319  };
320};
321} // end anonymous namespace
322
323typedef llvm::ImmutableMap<SymbolRef, RefState> RegionStateTy;
324typedef llvm::ImmutableMap<SymbolRef, ReallocPair > ReallocMap;
325class RegionState {};
326class ReallocPairs {};
327namespace clang {
328namespace ento {
329  template <>
330  struct ProgramStateTrait<RegionState>
331    : public ProgramStatePartialTrait<RegionStateTy> {
332    static void *GDMIndex() { static int x; return &x; }
333  };
334
335  template <>
336  struct ProgramStateTrait<ReallocPairs>
337    : public ProgramStatePartialTrait<ReallocMap> {
338    static void *GDMIndex() { static int x; return &x; }
339  };
340}
341}
342
343namespace {
344class StopTrackingCallback : public SymbolVisitor {
345  ProgramStateRef state;
346public:
347  StopTrackingCallback(ProgramStateRef st) : state(st) {}
348  ProgramStateRef getState() const { return state; }
349
350  bool VisitSymbol(SymbolRef sym) {
351    state = state->remove<RegionState>(sym);
352    return true;
353  }
354};
355} // end anonymous namespace
356
357void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
358  if (II_malloc)
359    return;
360  II_malloc = &Ctx.Idents.get("malloc");
361  II_free = &Ctx.Idents.get("free");
362  II_realloc = &Ctx.Idents.get("realloc");
363  II_reallocf = &Ctx.Idents.get("reallocf");
364  II_calloc = &Ctx.Idents.get("calloc");
365  II_valloc = &Ctx.Idents.get("valloc");
366  II_strdup = &Ctx.Idents.get("strdup");
367  II_strndup = &Ctx.Idents.get("strndup");
368}
369
370bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
371  if (isFreeFunction(FD, C))
372    return true;
373
374  if (isAllocationFunction(FD, C))
375    return true;
376
377  return false;
378}
379
380bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
381                                         ASTContext &C) const {
382  if (!FD)
383    return false;
384
385  if (FD->getKind() == Decl::Function) {
386    IdentifierInfo *FunI = FD->getIdentifier();
387    initIdentifierInfo(C);
388
389    if (FunI == II_malloc || FunI == II_realloc ||
390        FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
391        FunI == II_strdup || FunI == II_strndup)
392      return true;
393  }
394
395  if (Filter.CMallocOptimistic && FD->hasAttrs())
396    for (specific_attr_iterator<OwnershipAttr>
397           i = FD->specific_attr_begin<OwnershipAttr>(),
398           e = FD->specific_attr_end<OwnershipAttr>();
399           i != e; ++i)
400      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
401        return true;
402  return false;
403}
404
405bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
406  if (!FD)
407    return false;
408
409  if (FD->getKind() == Decl::Function) {
410    IdentifierInfo *FunI = FD->getIdentifier();
411    initIdentifierInfo(C);
412
413    if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
414      return true;
415  }
416
417  if (Filter.CMallocOptimistic && FD->hasAttrs())
418    for (specific_attr_iterator<OwnershipAttr>
419           i = FD->specific_attr_begin<OwnershipAttr>(),
420           e = FD->specific_attr_end<OwnershipAttr>();
421           i != e; ++i)
422      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
423          (*i)->getOwnKind() == OwnershipAttr::Holds)
424        return true;
425  return false;
426}
427
428void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
429  const FunctionDecl *FD = C.getCalleeDecl(CE);
430  if (!FD)
431    return;
432
433  ProgramStateRef State = C.getState();
434
435  if (FD->getKind() == Decl::Function) {
436    initIdentifierInfo(C.getASTContext());
437    IdentifierInfo *FunI = FD->getIdentifier();
438
439    if (FunI == II_malloc || FunI == II_valloc) {
440      if (CE->getNumArgs() < 1)
441        return;
442      State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
443    } else if (FunI == II_realloc) {
444      State = ReallocMem(C, CE, false);
445    } else if (FunI == II_reallocf) {
446      State = ReallocMem(C, CE, true);
447    } else if (FunI == II_calloc) {
448      State = CallocMem(C, CE);
449    } else if (FunI == II_free) {
450      State = FreeMemAux(C, CE, State, 0, false);
451    } else if (FunI == II_strdup) {
452      State = MallocUpdateRefState(C, CE, State);
453    } else if (FunI == II_strndup) {
454      State = MallocUpdateRefState(C, CE, State);
455    }
456  }
457
458  if (Filter.CMallocOptimistic) {
459    // Check all the attributes, if there are any.
460    // There can be multiple of these attributes.
461    if (FD->hasAttrs())
462      for (specific_attr_iterator<OwnershipAttr>
463          i = FD->specific_attr_begin<OwnershipAttr>(),
464          e = FD->specific_attr_end<OwnershipAttr>();
465          i != e; ++i) {
466        switch ((*i)->getOwnKind()) {
467        case OwnershipAttr::Returns:
468          State = MallocMemReturnsAttr(C, CE, *i);
469          break;
470        case OwnershipAttr::Takes:
471        case OwnershipAttr::Holds:
472          State = FreeMemAttr(C, CE, *i);
473          break;
474        }
475      }
476  }
477  C.addTransition(State);
478}
479
480static bool isFreeWhenDoneSetToZero(const ObjCMethodCall &Call) {
481  Selector S = Call.getSelector();
482  for (unsigned i = 1; i < S.getNumArgs(); ++i)
483    if (S.getNameForSlot(i).equals("freeWhenDone"))
484      if (Call.getArgSVal(i).isConstant(0))
485        return true;
486
487  return false;
488}
489
490void MallocChecker::checkPreObjCMessage(const ObjCMethodCall &Call,
491                                        CheckerContext &C) const {
492  // If the first selector is dataWithBytesNoCopy, assume that the memory will
493  // be released with 'free' by the new object.
494  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
495  // Unless 'freeWhenDone' param set to 0.
496  // TODO: Check that the memory was allocated with malloc.
497  Selector S = Call.getSelector();
498  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
499       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
500       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
501      !isFreeWhenDoneSetToZero(Call)){
502    unsigned int argIdx  = 0;
503    C.addTransition(FreeMemAux(C, Call.getArgExpr(argIdx),
504                    Call.getOriginExpr(), C.getState(), true));
505  }
506}
507
508ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
509                                                    const CallExpr *CE,
510                                                    const OwnershipAttr* Att) {
511  if (Att->getModule() != "malloc")
512    return 0;
513
514  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
515  if (I != E) {
516    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
517  }
518  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
519}
520
521ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
522                                           const CallExpr *CE,
523                                           SVal Size, SVal Init,
524                                           ProgramStateRef state) {
525
526  // Bind the return value to the symbolic value from the heap region.
527  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
528  // side effects other than what we model here.
529  unsigned Count = C.blockCount();
530  SValBuilder &svalBuilder = C.getSValBuilder();
531  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
532  DefinedSVal RetVal =
533    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
534  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
535
536  // We expect the malloc functions to return a pointer.
537  if (!isa<Loc>(RetVal))
538    return 0;
539
540  // Fill the region with the initialization value.
541  state = state->bindDefault(RetVal, Init);
542
543  // Set the region's extent equal to the Size parameter.
544  const SymbolicRegion *R =
545      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
546  if (!R)
547    return 0;
548  if (isa<DefinedOrUnknownSVal>(Size)) {
549    SValBuilder &svalBuilder = C.getSValBuilder();
550    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
551    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
552    DefinedOrUnknownSVal extentMatchesSize =
553        svalBuilder.evalEQ(state, Extent, DefinedSize);
554
555    state = state->assume(extentMatchesSize, true);
556    assert(state);
557  }
558
559  return MallocUpdateRefState(C, CE, state);
560}
561
562ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
563                                                    const CallExpr *CE,
564                                                    ProgramStateRef state) {
565  // Get the return value.
566  SVal retVal = state->getSVal(CE, C.getLocationContext());
567
568  // We expect the malloc functions to return a pointer.
569  if (!isa<Loc>(retVal))
570    return 0;
571
572  SymbolRef Sym = retVal.getAsLocSymbol();
573  assert(Sym);
574
575  // Set the symbol's state to Allocated.
576  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
577
578}
579
580ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
581                                           const CallExpr *CE,
582                                           const OwnershipAttr* Att) const {
583  if (Att->getModule() != "malloc")
584    return 0;
585
586  ProgramStateRef State = C.getState();
587
588  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
589       I != E; ++I) {
590    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
591                               Att->getOwnKind() == OwnershipAttr::Holds);
592    if (StateI)
593      State = StateI;
594  }
595  return State;
596}
597
598ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
599                                          const CallExpr *CE,
600                                          ProgramStateRef state,
601                                          unsigned Num,
602                                          bool Hold) const {
603  if (CE->getNumArgs() < (Num + 1))
604    return 0;
605
606  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold);
607}
608
609ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
610                                          const Expr *ArgExpr,
611                                          const Expr *ParentExpr,
612                                          ProgramStateRef state,
613                                          bool Hold) const {
614
615  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
616  if (!isa<DefinedOrUnknownSVal>(ArgVal))
617    return 0;
618  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
619
620  // Check for null dereferences.
621  if (!isa<Loc>(location))
622    return 0;
623
624  // The explicit NULL case, no operation is performed.
625  ProgramStateRef notNullState, nullState;
626  llvm::tie(notNullState, nullState) = state->assume(location);
627  if (nullState && !notNullState)
628    return 0;
629
630  // Unknown values could easily be okay
631  // Undefined values are handled elsewhere
632  if (ArgVal.isUnknownOrUndef())
633    return 0;
634
635  const MemRegion *R = ArgVal.getAsRegion();
636
637  // Nonlocs can't be freed, of course.
638  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
639  if (!R) {
640    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
641    return 0;
642  }
643
644  R = R->StripCasts();
645
646  // Blocks might show up as heap data, but should not be free()d
647  if (isa<BlockDataRegion>(R)) {
648    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
649    return 0;
650  }
651
652  const MemSpaceRegion *MS = R->getMemorySpace();
653
654  // Parameters, locals, statics, and globals shouldn't be freed.
655  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
656    // FIXME: at the time this code was written, malloc() regions were
657    // represented by conjured symbols, which are all in UnknownSpaceRegion.
658    // This means that there isn't actually anything from HeapSpaceRegion
659    // that should be freed, even though we allow it here.
660    // Of course, free() can work on memory allocated outside the current
661    // function, so UnknownSpaceRegion is always a possibility.
662    // False negatives are better than false positives.
663
664    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
665    return 0;
666  }
667
668  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
669  // Various cases could lead to non-symbol values here.
670  // For now, ignore them.
671  if (!SR)
672    return 0;
673
674  SymbolRef Sym = SR->getSymbol();
675  const RefState *RS = state->get<RegionState>(Sym);
676
677  // Check double free.
678  if (RS && (RS->isReleased() || RS->isRelinquished())) {
679    if (ExplodedNode *N = C.generateSink()) {
680      if (!BT_DoubleFree)
681        BT_DoubleFree.reset(
682          new BugType("Double free", "Memory Error"));
683      BugReport *R = new BugReport(*BT_DoubleFree,
684        (RS->isReleased() ? "Attempt to free released memory" :
685                            "Attempt to free non-owned memory"), N);
686      R->addRange(ArgExpr->getSourceRange());
687      R->markInteresting(Sym);
688      R->addVisitor(new MallocBugVisitor(Sym));
689      C.EmitReport(R);
690    }
691    return 0;
692  }
693
694  // Normal free.
695  if (Hold)
696    return state->set<RegionState>(Sym, RefState::getRelinquished(ParentExpr));
697  return state->set<RegionState>(Sym, RefState::getReleased(ParentExpr));
698}
699
700bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
701  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
702    os << "an integer (" << IntVal->getValue() << ")";
703  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
704    os << "a constant address (" << ConstAddr->getValue() << ")";
705  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
706    os << "the address of the label '" << Label->getLabel()->getName() << "'";
707  else
708    return false;
709
710  return true;
711}
712
713bool MallocChecker::SummarizeRegion(raw_ostream &os,
714                                    const MemRegion *MR) {
715  switch (MR->getKind()) {
716  case MemRegion::FunctionTextRegionKind: {
717    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
718    if (FD)
719      os << "the address of the function '" << *FD << '\'';
720    else
721      os << "the address of a function";
722    return true;
723  }
724  case MemRegion::BlockTextRegionKind:
725    os << "block text";
726    return true;
727  case MemRegion::BlockDataRegionKind:
728    // FIXME: where the block came from?
729    os << "a block";
730    return true;
731  default: {
732    const MemSpaceRegion *MS = MR->getMemorySpace();
733
734    if (isa<StackLocalsSpaceRegion>(MS)) {
735      const VarRegion *VR = dyn_cast<VarRegion>(MR);
736      const VarDecl *VD;
737      if (VR)
738        VD = VR->getDecl();
739      else
740        VD = NULL;
741
742      if (VD)
743        os << "the address of the local variable '" << VD->getName() << "'";
744      else
745        os << "the address of a local stack variable";
746      return true;
747    }
748
749    if (isa<StackArgumentsSpaceRegion>(MS)) {
750      const VarRegion *VR = dyn_cast<VarRegion>(MR);
751      const VarDecl *VD;
752      if (VR)
753        VD = VR->getDecl();
754      else
755        VD = NULL;
756
757      if (VD)
758        os << "the address of the parameter '" << VD->getName() << "'";
759      else
760        os << "the address of a parameter";
761      return true;
762    }
763
764    if (isa<GlobalsSpaceRegion>(MS)) {
765      const VarRegion *VR = dyn_cast<VarRegion>(MR);
766      const VarDecl *VD;
767      if (VR)
768        VD = VR->getDecl();
769      else
770        VD = NULL;
771
772      if (VD) {
773        if (VD->isStaticLocal())
774          os << "the address of the static variable '" << VD->getName() << "'";
775        else
776          os << "the address of the global variable '" << VD->getName() << "'";
777      } else
778        os << "the address of a global variable";
779      return true;
780    }
781
782    return false;
783  }
784  }
785}
786
787void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
788                                  SourceRange range) const {
789  if (ExplodedNode *N = C.generateSink()) {
790    if (!BT_BadFree)
791      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
792
793    SmallString<100> buf;
794    llvm::raw_svector_ostream os(buf);
795
796    const MemRegion *MR = ArgVal.getAsRegion();
797    if (MR) {
798      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
799        MR = ER->getSuperRegion();
800
801      // Special case for alloca()
802      if (isa<AllocaRegion>(MR))
803        os << "Argument to free() was allocated by alloca(), not malloc()";
804      else {
805        os << "Argument to free() is ";
806        if (SummarizeRegion(os, MR))
807          os << ", which is not memory allocated by malloc()";
808        else
809          os << "not memory allocated by malloc()";
810      }
811    } else {
812      os << "Argument to free() is ";
813      if (SummarizeValue(os, ArgVal))
814        os << ", which is not memory allocated by malloc()";
815      else
816        os << "not memory allocated by malloc()";
817    }
818
819    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
820    R->markInteresting(MR);
821    R->addRange(range);
822    C.EmitReport(R);
823  }
824}
825
826ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
827                                          const CallExpr *CE,
828                                          bool FreesOnFail) const {
829  if (CE->getNumArgs() < 2)
830    return 0;
831
832  ProgramStateRef state = C.getState();
833  const Expr *arg0Expr = CE->getArg(0);
834  const LocationContext *LCtx = C.getLocationContext();
835  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
836  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
837    return 0;
838  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
839
840  SValBuilder &svalBuilder = C.getSValBuilder();
841
842  DefinedOrUnknownSVal PtrEQ =
843    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
844
845  // Get the size argument. If there is no size arg then give up.
846  const Expr *Arg1 = CE->getArg(1);
847  if (!Arg1)
848    return 0;
849
850  // Get the value of the size argument.
851  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
852  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
853    return 0;
854  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
855
856  // Compare the size argument to 0.
857  DefinedOrUnknownSVal SizeZero =
858    svalBuilder.evalEQ(state, Arg1Val,
859                       svalBuilder.makeIntValWithPtrWidth(0, false));
860
861  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
862  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
863  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
864  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
865  // We only assume exceptional states if they are definitely true; if the
866  // state is under-constrained, assume regular realloc behavior.
867  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
868  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
869
870  // If the ptr is NULL and the size is not 0, the call is equivalent to
871  // malloc(size).
872  if ( PrtIsNull && !SizeIsZero) {
873    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
874                                               UndefinedVal(), StatePtrIsNull);
875    return stateMalloc;
876  }
877
878  if (PrtIsNull && SizeIsZero)
879    return 0;
880
881  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
882  assert(!PrtIsNull);
883  SymbolRef FromPtr = arg0Val.getAsSymbol();
884  SVal RetVal = state->getSVal(CE, LCtx);
885  SymbolRef ToPtr = RetVal.getAsSymbol();
886  if (!FromPtr || !ToPtr)
887    return 0;
888
889  // If the size is 0, free the memory.
890  if (SizeIsZero)
891    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
892      // The semantics of the return value are:
893      // If size was equal to 0, either NULL or a pointer suitable to be passed
894      // to free() is returned. We just free the input pointer and do not add
895      // any constrains on the output pointer.
896      return stateFree;
897    }
898
899  // Default behavior.
900  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
901    // FIXME: We should copy the content of the original buffer.
902    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
903                                                UnknownVal(), stateFree);
904    if (!stateRealloc)
905      return 0;
906    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
907                                            ReallocPair(FromPtr, FreesOnFail));
908    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
909    return stateRealloc;
910  }
911  return 0;
912}
913
914ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
915  if (CE->getNumArgs() < 2)
916    return 0;
917
918  ProgramStateRef state = C.getState();
919  SValBuilder &svalBuilder = C.getSValBuilder();
920  const LocationContext *LCtx = C.getLocationContext();
921  SVal count = state->getSVal(CE->getArg(0), LCtx);
922  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
923  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
924                                        svalBuilder.getContext().getSizeType());
925  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
926
927  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
928}
929
930LeakInfo
931MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
932                                 CheckerContext &C) const {
933  const LocationContext *LeakContext = N->getLocationContext();
934  // Walk the ExplodedGraph backwards and find the first node that referred to
935  // the tracked symbol.
936  const ExplodedNode *AllocNode = N;
937  const MemRegion *ReferenceRegion = 0;
938
939  while (N) {
940    ProgramStateRef State = N->getState();
941    if (!State->get<RegionState>(Sym))
942      break;
943
944    // Find the most recent expression bound to the symbol in the current
945    // context.
946    if (!ReferenceRegion) {
947      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
948        SVal Val = State->getSVal(MR);
949        if (Val.getAsLocSymbol() == Sym)
950          ReferenceRegion = MR;
951      }
952    }
953
954    // Allocation node, is the last node in the current context in which the
955    // symbol was tracked.
956    if (N->getLocationContext() == LeakContext)
957      AllocNode = N;
958    N = N->pred_empty() ? NULL : *(N->pred_begin());
959  }
960
961  ProgramPoint P = AllocNode->getLocation();
962  const Stmt *AllocationStmt = 0;
963  if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&P))
964    AllocationStmt = Exit->getCalleeContext()->getCallSite();
965  else if (StmtPoint *SP = dyn_cast<StmtPoint>(&P))
966    AllocationStmt = SP->getStmt();
967
968  return LeakInfo(AllocationStmt, ReferenceRegion);
969}
970
971void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
972                               CheckerContext &C) const {
973  assert(N);
974  if (!BT_Leak) {
975    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
976    // Leaks should not be reported if they are post-dominated by a sink:
977    // (1) Sinks are higher importance bugs.
978    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
979    //     with __noreturn functions such as assert() or exit(). We choose not
980    //     to report leaks on such paths.
981    BT_Leak->setSuppressOnSink(true);
982  }
983
984  // Most bug reports are cached at the location where they occurred.
985  // With leaks, we want to unique them by the location where they were
986  // allocated, and only report a single path.
987  PathDiagnosticLocation LocUsedForUniqueing;
988  const Stmt *AllocStmt = 0;
989  const MemRegion *Region = 0;
990  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
991  if (AllocStmt)
992    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
993                            C.getSourceManager(), N->getLocationContext());
994
995  SmallString<200> buf;
996  llvm::raw_svector_ostream os(buf);
997  os << "Memory is never released; potential leak";
998  if (Region && Region->canPrintPretty()) {
999    os << " of memory pointed to by '";
1000    Region->printPretty(os);
1001    os << '\'';
1002  }
1003
1004  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
1005  R->markInteresting(Sym);
1006  R->addVisitor(new MallocBugVisitor(Sym, true));
1007  C.EmitReport(R);
1008}
1009
1010void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1011                                     CheckerContext &C) const
1012{
1013  if (!SymReaper.hasDeadSymbols())
1014    return;
1015
1016  ProgramStateRef state = C.getState();
1017  RegionStateTy RS = state->get<RegionState>();
1018  RegionStateTy::Factory &F = state->get_context<RegionState>();
1019
1020  bool generateReport = false;
1021  llvm::SmallVector<SymbolRef, 2> Errors;
1022  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1023    if (SymReaper.isDead(I->first)) {
1024      if (I->second.isAllocated()) {
1025        generateReport = true;
1026        Errors.push_back(I->first);
1027      }
1028      // Remove the dead symbol from the map.
1029      RS = F.remove(RS, I->first);
1030
1031    }
1032  }
1033
1034  // Cleanup the Realloc Pairs Map.
1035  ReallocMap RP = state->get<ReallocPairs>();
1036  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1037    if (SymReaper.isDead(I->first) ||
1038        SymReaper.isDead(I->second.ReallocatedSym)) {
1039      state = state->remove<ReallocPairs>(I->first);
1040    }
1041  }
1042
1043  // Generate leak node.
1044  static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1045  ExplodedNode *N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1046
1047  if (generateReport) {
1048    for (llvm::SmallVector<SymbolRef, 2>::iterator
1049         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1050      reportLeak(*I, N, C);
1051    }
1052  }
1053  C.addTransition(state->set<RegionState>(RS), N);
1054}
1055
1056void MallocChecker::checkEndPath(CheckerContext &C) const {
1057  ProgramStateRef state = C.getState();
1058  RegionStateTy M = state->get<RegionState>();
1059
1060  // If inside inlined call, skip it.
1061  if (C.getLocationContext()->getParent() != 0)
1062    return;
1063
1064  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1065    RefState RS = I->second;
1066    if (RS.isAllocated()) {
1067      ExplodedNode *N = C.addTransition(state);
1068      if (N)
1069        reportLeak(I->first, N, C);
1070    }
1071  }
1072}
1073
1074void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1075  // We will check for double free in the post visit.
1076  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1077    return;
1078
1079  // Check use after free, when a freed pointer is passed to a call.
1080  ProgramStateRef State = C.getState();
1081  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1082                                    E = CE->arg_end(); I != E; ++I) {
1083    const Expr *A = *I;
1084    if (A->getType().getTypePtr()->isAnyPointerType()) {
1085      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1086      if (!Sym)
1087        continue;
1088      if (checkUseAfterFree(Sym, C, A))
1089        return;
1090    }
1091  }
1092}
1093
1094void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1095  const Expr *E = S->getRetValue();
1096  if (!E)
1097    return;
1098
1099  // Check if we are returning a symbol.
1100  ProgramStateRef State = C.getState();
1101  SVal RetVal = State->getSVal(E, C.getLocationContext());
1102  SymbolRef Sym = RetVal.getAsSymbol();
1103  if (!Sym)
1104    // If we are returning a field of the allocated struct or an array element,
1105    // the callee could still free the memory.
1106    // TODO: This logic should be a part of generic symbol escape callback.
1107    if (const MemRegion *MR = RetVal.getAsRegion())
1108      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1109        if (const SymbolicRegion *BMR =
1110              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1111          Sym = BMR->getSymbol();
1112
1113  // Check if we are returning freed memory.
1114  if (Sym)
1115    if (checkUseAfterFree(Sym, C, E))
1116      return;
1117
1118  // If this function body is not inlined, stop tracking any returned symbols.
1119  if (C.getLocationContext()->getParent() == 0) {
1120    State =
1121      State->scanReachableSymbols<StopTrackingCallback>(RetVal).getState();
1122    C.addTransition(State);
1123  }
1124}
1125
1126// TODO: Blocks should be either inlined or should call invalidate regions
1127// upon invocation. After that's in place, special casing here will not be
1128// needed.
1129void MallocChecker::checkPostStmt(const BlockExpr *BE,
1130                                  CheckerContext &C) const {
1131
1132  // Scan the BlockDecRefExprs for any object the retain count checker
1133  // may be tracking.
1134  if (!BE->getBlockDecl()->hasCaptures())
1135    return;
1136
1137  ProgramStateRef state = C.getState();
1138  const BlockDataRegion *R =
1139    cast<BlockDataRegion>(state->getSVal(BE,
1140                                         C.getLocationContext()).getAsRegion());
1141
1142  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1143                                            E = R->referenced_vars_end();
1144
1145  if (I == E)
1146    return;
1147
1148  SmallVector<const MemRegion*, 10> Regions;
1149  const LocationContext *LC = C.getLocationContext();
1150  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1151
1152  for ( ; I != E; ++I) {
1153    const VarRegion *VR = *I;
1154    if (VR->getSuperRegion() == R) {
1155      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1156    }
1157    Regions.push_back(VR);
1158  }
1159
1160  state =
1161    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1162                                    Regions.data() + Regions.size()).getState();
1163  C.addTransition(state);
1164}
1165
1166bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1167  assert(Sym);
1168  const RefState *RS = C.getState()->get<RegionState>(Sym);
1169  return (RS && RS->isReleased());
1170}
1171
1172bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1173                                      const Stmt *S) const {
1174  if (isReleased(Sym, C)) {
1175    if (ExplodedNode *N = C.generateSink()) {
1176      if (!BT_UseFree)
1177        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1178
1179      BugReport *R = new BugReport(*BT_UseFree,
1180                                   "Use of memory after it is freed",N);
1181      if (S)
1182        R->addRange(S->getSourceRange());
1183      R->markInteresting(Sym);
1184      R->addVisitor(new MallocBugVisitor(Sym));
1185      C.EmitReport(R);
1186      return true;
1187    }
1188  }
1189  return false;
1190}
1191
1192// Check if the location is a freed symbolic region.
1193void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1194                                  CheckerContext &C) const {
1195  SymbolRef Sym = l.getLocSymbolInBase();
1196  if (Sym)
1197    checkUseAfterFree(Sym, C, S);
1198}
1199
1200//===----------------------------------------------------------------------===//
1201// Check various ways a symbol can be invalidated.
1202// TODO: This logic (the next 3 functions) is copied/similar to the
1203// RetainRelease checker. We might want to factor this out.
1204//===----------------------------------------------------------------------===//
1205
1206// Stop tracking symbols when a value escapes as a result of checkBind.
1207// A value escapes in three possible cases:
1208// (1) we are binding to something that is not a memory region.
1209// (2) we are binding to a memregion that does not have stack storage
1210// (3) we are binding to a memregion with stack storage that the store
1211//     does not understand.
1212void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1213                              CheckerContext &C) const {
1214  // Are we storing to something that causes the value to "escape"?
1215  bool escapes = true;
1216  ProgramStateRef state = C.getState();
1217
1218  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1219    escapes = !regionLoc->getRegion()->hasStackStorage();
1220
1221    if (!escapes) {
1222      // To test (3), generate a new state with the binding added.  If it is
1223      // the same state, then it escapes (since the store cannot represent
1224      // the binding).
1225      // Do this only if we know that the store is not supposed to generate the
1226      // same state.
1227      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1228      if (StoredVal != val)
1229        escapes = (state == (state->bindLoc(*regionLoc, val)));
1230    }
1231  }
1232
1233  // If our store can represent the binding and we aren't storing to something
1234  // that doesn't have local storage then just return and have the simulation
1235  // state continue as is.
1236  if (!escapes)
1237      return;
1238
1239  // Otherwise, find all symbols referenced by 'val' that we are tracking
1240  // and stop tracking them.
1241  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1242  C.addTransition(state);
1243}
1244
1245// If a symbolic region is assumed to NULL (or another constant), stop tracking
1246// it - assuming that allocation failed on this path.
1247ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1248                                              SVal Cond,
1249                                              bool Assumption) const {
1250  RegionStateTy RS = state->get<RegionState>();
1251  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1252    // If the symbol is assumed to NULL or another constant, this will
1253    // return an APSInt*.
1254    if (state->getSymVal(I.getKey()))
1255      state = state->remove<RegionState>(I.getKey());
1256  }
1257
1258  // Realloc returns 0 when reallocation fails, which means that we should
1259  // restore the state of the pointer being reallocated.
1260  ReallocMap RP = state->get<ReallocPairs>();
1261  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1262    // If the symbol is assumed to NULL or another constant, this will
1263    // return an APSInt*.
1264    if (state->getSymVal(I.getKey())) {
1265      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1266      const RefState *RS = state->get<RegionState>(ReallocSym);
1267      if (RS) {
1268        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1269          state = state->set<RegionState>(ReallocSym,
1270                             RefState::getAllocated(RS->getStmt()));
1271      }
1272      state = state->remove<ReallocPairs>(I.getKey());
1273    }
1274  }
1275
1276  return state;
1277}
1278
1279// Check if the function is known to us. So, for example, we could
1280// conservatively assume it can free/reallocate its pointer arguments.
1281// (We assume that the pointers cannot escape through calls to system
1282// functions not handled by this checker.)
1283bool MallocChecker::doesNotFreeMemory(const CallEvent *Call,
1284                                      ProgramStateRef State) const {
1285  assert(Call);
1286
1287  // For now, assume that any C++ call can free memory.
1288  // TODO: If we want to be more optimistic here, we'll need to make sure that
1289  // regions escape to C++ containers. They seem to do that even now, but for
1290  // mysterious reasons.
1291  if (!(isa<FunctionCall>(Call) || isa<ObjCMethodCall>(Call)))
1292    return false;
1293
1294  // Check Objective-C messages by selector name.
1295  if (const ObjCMethodCall *Msg = dyn_cast<ObjCMethodCall>(Call)) {
1296    // If it's not a framework call, or if it takes a callback, assume it
1297    // can free memory.
1298    if (!Call->isInSystemHeader() || Call->hasNonZeroCallbackArg())
1299      return false;
1300
1301    Selector S = Msg->getSelector();
1302
1303    // Whitelist the ObjC methods which do free memory.
1304    // - Anything containing 'freeWhenDone' param set to 1.
1305    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1306    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1307      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1308        if (Call->getArgSVal(i).isConstant(1))
1309          return false;
1310        else
1311          return true;
1312      }
1313    }
1314
1315    // If the first selector ends with NoCopy, assume that the ownership is
1316    // transferred as well.
1317    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1318    StringRef FirstSlot = S.getNameForSlot(0);
1319    if (FirstSlot.endswith("NoCopy"))
1320      return false;
1321
1322    // If the first selector starts with addPointer, insertPointer,
1323    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1324    // This is similar to C++ containers (vector); we still might want to check
1325    // that the pointers get freed by following the container itself.
1326    if (FirstSlot.startswith("addPointer") ||
1327        FirstSlot.startswith("insertPointer") ||
1328        FirstSlot.startswith("replacePointer")) {
1329      return false;
1330    }
1331
1332    // Otherwise, assume that the method does not free memory.
1333    // Most framework methods do not free memory.
1334    return true;
1335  }
1336
1337  // At this point the only thing left to handle is straight function calls.
1338  const FunctionDecl *FD = cast<FunctionCall>(Call)->getDecl();
1339  if (!FD)
1340    return false;
1341
1342  ASTContext &ASTC = State->getStateManager().getContext();
1343
1344  // If it's one of the allocation functions we can reason about, we model
1345  // its behavior explicitly.
1346  if (isMemFunction(FD, ASTC))
1347    return true;
1348
1349  // If it's not a system call, assume it frees memory.
1350  if (!Call->isInSystemHeader())
1351    return false;
1352
1353  // White list the system functions whose arguments escape.
1354  const IdentifierInfo *II = FD->getIdentifier();
1355  if (!II)
1356    return false;
1357  StringRef FName = II->getName();
1358
1359  // White list the 'XXXNoCopy' CoreFoundation functions.
1360  // We specifically check these before
1361  if (FName.endswith("NoCopy")) {
1362    // Look for the deallocator argument. We know that the memory ownership
1363    // is not transferred only if the deallocator argument is
1364    // 'kCFAllocatorNull'.
1365    for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1366      const Expr *ArgE = Call->getArgExpr(i)->IgnoreParenCasts();
1367      if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1368        StringRef DeallocatorName = DE->getFoundDecl()->getName();
1369        if (DeallocatorName == "kCFAllocatorNull")
1370          return true;
1371      }
1372    }
1373    return false;
1374  }
1375
1376  // Associating streams with malloced buffers. The pointer can escape if
1377  // 'closefn' is specified (and if that function does free memory),
1378  // but it will not if closefn is not specified.
1379  // Currently, we do not inspect the 'closefn' function (PR12101).
1380  if (FName == "funopen")
1381    if (Call->getNumArgs() >= 4 && Call->getArgSVal(4).isConstant(0))
1382      return true;
1383
1384  // Do not warn on pointers passed to 'setbuf' when used with std streams,
1385  // these leaks might be intentional when setting the buffer for stdio.
1386  // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1387  if (FName == "setbuf" || FName =="setbuffer" ||
1388      FName == "setlinebuf" || FName == "setvbuf") {
1389    if (Call->getNumArgs() >= 1) {
1390      const Expr *ArgE = Call->getArgExpr(0)->IgnoreParenCasts();
1391      if (const DeclRefExpr *ArgDRE = dyn_cast<DeclRefExpr>(ArgE))
1392        if (const VarDecl *D = dyn_cast<VarDecl>(ArgDRE->getDecl()))
1393          if (D->getCanonicalDecl()->getName().find("std") != StringRef::npos)
1394            return false;
1395    }
1396  }
1397
1398  // A bunch of other functions which either take ownership of a pointer or
1399  // wrap the result up in a struct or object, meaning it can be freed later.
1400  // (See RetainCountChecker.) Not all the parameters here are invalidated,
1401  // but the Malloc checker cannot differentiate between them. The right way
1402  // of doing this would be to implement a pointer escapes callback.
1403  if (FName == "CGBitmapContextCreate" ||
1404      FName == "CGBitmapContextCreateWithData" ||
1405      FName == "CVPixelBufferCreateWithBytes" ||
1406      FName == "CVPixelBufferCreateWithPlanarBytes" ||
1407      FName == "OSAtomicEnqueue") {
1408    return false;
1409  }
1410
1411  // Handle cases where we know a buffer's /address/ can escape.
1412  // Note that the above checks handle some special cases where we know that
1413  // even though the address escapes, it's still our responsibility to free the
1414  // buffer.
1415  if (Call->argumentsMayEscape())
1416    return false;
1417
1418  // Otherwise, assume that the function does not free memory.
1419  // Most system calls do not free the memory.
1420  return true;
1421}
1422
1423// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1424// escapes, when we are tracking p), do not track the symbol as we cannot reason
1425// about it anymore.
1426ProgramStateRef
1427MallocChecker::checkRegionChanges(ProgramStateRef State,
1428                            const StoreManager::InvalidatedSymbols *invalidated,
1429                                    ArrayRef<const MemRegion *> ExplicitRegions,
1430                                    ArrayRef<const MemRegion *> Regions,
1431                                    const CallEvent *Call) const {
1432  if (!invalidated || invalidated->empty())
1433    return State;
1434  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1435
1436  // If it's a call which might free or reallocate memory, we assume that all
1437  // regions (explicit and implicit) escaped.
1438
1439  // Otherwise, whitelist explicit pointers; we still can track them.
1440  if (!Call || doesNotFreeMemory(Call, State)) {
1441    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1442        E = ExplicitRegions.end(); I != E; ++I) {
1443      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1444        WhitelistedSymbols.insert(R->getSymbol());
1445    }
1446  }
1447
1448  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1449       E = invalidated->end(); I!=E; ++I) {
1450    SymbolRef sym = *I;
1451    if (WhitelistedSymbols.count(sym))
1452      continue;
1453    // The symbol escaped. Note, we assume that if the symbol is released,
1454    // passing it out will result in a use after free. We also keep tracking
1455    // relinquished symbols.
1456    if (const RefState *RS = State->get<RegionState>(sym)) {
1457      if (RS->isAllocated())
1458        State = State->remove<RegionState>(sym);
1459    }
1460  }
1461  return State;
1462}
1463
1464static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1465                                         ProgramStateRef prevState) {
1466  ReallocMap currMap = currState->get<ReallocPairs>();
1467  ReallocMap prevMap = prevState->get<ReallocPairs>();
1468
1469  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1470       I != E; ++I) {
1471    SymbolRef sym = I.getKey();
1472    if (!currMap.lookup(sym))
1473      return sym;
1474  }
1475
1476  return NULL;
1477}
1478
1479PathDiagnosticPiece *
1480MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1481                                           const ExplodedNode *PrevN,
1482                                           BugReporterContext &BRC,
1483                                           BugReport &BR) {
1484  ProgramStateRef state = N->getState();
1485  ProgramStateRef statePrev = PrevN->getState();
1486
1487  const RefState *RS = state->get<RegionState>(Sym);
1488  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1489  if (!RS)
1490    return 0;
1491
1492  const Stmt *S = 0;
1493  const char *Msg = 0;
1494  StackHintGeneratorForSymbol *StackHint = 0;
1495
1496  // Retrieve the associated statement.
1497  ProgramPoint ProgLoc = N->getLocation();
1498  if (StmtPoint *SP = dyn_cast<StmtPoint>(&ProgLoc))
1499    S = SP->getStmt();
1500  else if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&ProgLoc))
1501    S = Exit->getCalleeContext()->getCallSite();
1502  // If an assumption was made on a branch, it should be caught
1503  // here by looking at the state transition.
1504  else if (BlockEdge *Edge = dyn_cast<BlockEdge>(&ProgLoc)) {
1505    const CFGBlock *srcBlk = Edge->getSrc();
1506    S = srcBlk->getTerminator();
1507  }
1508  if (!S)
1509    return 0;
1510
1511  // FIXME: We will eventually need to handle non-statement-based events
1512  // (__attribute__((cleanup))).
1513
1514  // Find out if this is an interesting point and what is the kind.
1515  if (Mode == Normal) {
1516    if (isAllocated(RS, RSPrev, S)) {
1517      Msg = "Memory is allocated";
1518      StackHint = new StackHintGeneratorForSymbol(Sym,
1519                                                  "Returned allocated memory");
1520    } else if (isReleased(RS, RSPrev, S)) {
1521      Msg = "Memory is released";
1522      StackHint = new StackHintGeneratorForSymbol(Sym,
1523                                                  "Returned released memory");
1524    } else if (isRelinquished(RS, RSPrev, S)) {
1525      Msg = "Memory ownership is transfered";
1526      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1527    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1528      Mode = ReallocationFailed;
1529      Msg = "Reallocation failed";
1530      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1531                                                       "Reallocation failed");
1532
1533      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1534        // Is it possible to fail two reallocs WITHOUT testing in between?
1535        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1536          "We only support one failed realloc at a time.");
1537        BR.markInteresting(sym);
1538        FailedReallocSymbol = sym;
1539      }
1540    }
1541
1542  // We are in a special mode if a reallocation failed later in the path.
1543  } else if (Mode == ReallocationFailed) {
1544    assert(FailedReallocSymbol && "No symbol to look for.");
1545
1546    // Is this is the first appearance of the reallocated symbol?
1547    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1548      // We're at the reallocation point.
1549      Msg = "Attempt to reallocate memory";
1550      StackHint = new StackHintGeneratorForSymbol(Sym,
1551                                                 "Returned reallocated memory");
1552      FailedReallocSymbol = NULL;
1553      Mode = Normal;
1554    }
1555  }
1556
1557  if (!Msg)
1558    return 0;
1559  assert(StackHint);
1560
1561  // Generate the extra diagnostic.
1562  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1563                             N->getLocationContext());
1564  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1565}
1566
1567void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1568                               const char *NL, const char *Sep) const {
1569
1570  RegionStateTy RS = State->get<RegionState>();
1571
1572  if (!RS.isEmpty())
1573    Out << "Has Malloc data" << NL;
1574}
1575
1576#define REGISTER_CHECKER(name) \
1577void ento::register##name(CheckerManager &mgr) {\
1578  registerCStringCheckerBasic(mgr); \
1579  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1580}
1581
1582REGISTER_CHECKER(MallocPessimistic)
1583REGISTER_CHECKER(MallocOptimistic)
1584