MallocChecker.cpp revision ec8d420d4fa57fc6b5a5a2b1446742e976a7ba00 
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::PreObjCMessage,
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 checkPreObjCMessage(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) const;
197  ProgramStateRef FreeMemAux(CheckerContext &C, const Expr *Arg,
198                             const Expr *ParentExpr,
199                             ProgramStateRef state,
200                             bool Hold,
201                             bool &ReleasedAllocated) const;
202
203  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
204                             bool FreesMemOnFailure) const;
205  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
206
207  ///\brief Check if the memory associated with this symbol was released.
208  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
209
210  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
211                         const Stmt *S = 0) const;
212
213  /// Check if the function is not known to us. So, for example, we could
214  /// conservatively assume it can free/reallocate it's pointer arguments.
215  bool doesNotFreeMemory(const CallEvent *Call,
216                         ProgramStateRef State) const;
217
218  static bool SummarizeValue(raw_ostream &os, SVal V);
219  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
220  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
221
222  /// Find the location of the allocation for Sym on the path leading to the
223  /// exploded node N.
224  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
225                             CheckerContext &C) const;
226
227  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
228
229  /// The bug visitor which allows us to print extra diagnostics along the
230  /// BugReport path. For example, showing the allocation site of the leaked
231  /// region.
232  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
233  protected:
234    enum NotificationMode {
235      Normal,
236      ReallocationFailed
237    };
238
239    // The allocated region symbol tracked by the main analysis.
240    SymbolRef Sym;
241
242    // The mode we are in, i.e. what kind of diagnostics will be emitted.
243    NotificationMode Mode;
244
245    // A symbol from when the primary region should have been reallocated.
246    SymbolRef FailedReallocSymbol;
247
248    bool IsLeak;
249
250  public:
251    MallocBugVisitor(SymbolRef S, bool isLeak = false)
252       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
253
254    virtual ~MallocBugVisitor() {}
255
256    void Profile(llvm::FoldingSetNodeID &ID) const {
257      static int X = 0;
258      ID.AddPointer(&X);
259      ID.AddPointer(Sym);
260    }
261
262    inline bool isAllocated(const RefState *S, const RefState *SPrev,
263                            const Stmt *Stmt) {
264      // Did not track -> allocated. Other state (released) -> allocated.
265      return (Stmt && isa<CallExpr>(Stmt) &&
266              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
267    }
268
269    inline bool isReleased(const RefState *S, const RefState *SPrev,
270                           const Stmt *Stmt) {
271      // Did not track -> released. Other state (allocated) -> released.
272      return (Stmt && isa<CallExpr>(Stmt) &&
273              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
274    }
275
276    inline bool isRelinquished(const RefState *S, const RefState *SPrev,
277                               const Stmt *Stmt) {
278      // Did not track -> relinquished. Other state (allocated) -> relinquished.
279      return (Stmt && (isa<CallExpr>(Stmt) || isa<ObjCMessageExpr>(Stmt) ||
280                                              isa<ObjCPropertyRefExpr>(Stmt)) &&
281              (S && S->isRelinquished()) &&
282              (!SPrev || !SPrev->isRelinquished()));
283    }
284
285    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
286                                     const Stmt *Stmt) {
287      // If the expression is not a call, and the state change is
288      // released -> allocated, it must be the realloc return value
289      // check. If we have to handle more cases here, it might be cleaner just
290      // to track this extra bit in the state itself.
291      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
292              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
293    }
294
295    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
296                                   const ExplodedNode *PrevN,
297                                   BugReporterContext &BRC,
298                                   BugReport &BR);
299
300    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
301                                    const ExplodedNode *EndPathNode,
302                                    BugReport &BR) {
303      if (!IsLeak)
304        return 0;
305
306      PathDiagnosticLocation L =
307        PathDiagnosticLocation::createEndOfPath(EndPathNode,
308                                                BRC.getSourceManager());
309      // Do not add the statement itself as a range in case of leak.
310      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
311    }
312
313  private:
314    class StackHintGeneratorForReallocationFailed
315        : public StackHintGeneratorForSymbol {
316    public:
317      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
318        : StackHintGeneratorForSymbol(S, M) {}
319
320      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
321        // Printed parameters start at 1, not 0.
322        ++ArgIndex;
323
324        SmallString<200> buf;
325        llvm::raw_svector_ostream os(buf);
326
327        os << "Reallocation of " << ArgIndex << llvm::getOrdinalSuffix(ArgIndex)
328           << " parameter failed";
329
330        return os.str();
331      }
332
333      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
334        return "Reallocation of returned value failed";
335      }
336    };
337  };
338};
339} // end anonymous namespace
340
341typedef llvm::ImmutableMap<SymbolRef, RefState> RegionStateTy;
342typedef llvm::ImmutableMap<SymbolRef, ReallocPair > ReallocMap;
343class RegionState {};
344class ReallocPairs {};
345namespace clang {
346namespace ento {
347  template <>
348  struct ProgramStateTrait<RegionState>
349    : public ProgramStatePartialTrait<RegionStateTy> {
350    static void *GDMIndex() { static int x; return &x; }
351  };
352
353  template <>
354  struct ProgramStateTrait<ReallocPairs>
355    : public ProgramStatePartialTrait<ReallocMap> {
356    static void *GDMIndex() { static int x; return &x; }
357  };
358}
359}
360
361namespace {
362class StopTrackingCallback : public SymbolVisitor {
363  ProgramStateRef state;
364public:
365  StopTrackingCallback(ProgramStateRef st) : state(st) {}
366  ProgramStateRef getState() const { return state; }
367
368  bool VisitSymbol(SymbolRef sym) {
369    state = state->remove<RegionState>(sym);
370    return true;
371  }
372};
373} // end anonymous namespace
374
375void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
376  if (II_malloc)
377    return;
378  II_malloc = &Ctx.Idents.get("malloc");
379  II_free = &Ctx.Idents.get("free");
380  II_realloc = &Ctx.Idents.get("realloc");
381  II_reallocf = &Ctx.Idents.get("reallocf");
382  II_calloc = &Ctx.Idents.get("calloc");
383  II_valloc = &Ctx.Idents.get("valloc");
384  II_strdup = &Ctx.Idents.get("strdup");
385  II_strndup = &Ctx.Idents.get("strndup");
386}
387
388bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
389  if (isFreeFunction(FD, C))
390    return true;
391
392  if (isAllocationFunction(FD, C))
393    return true;
394
395  return false;
396}
397
398bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
399                                         ASTContext &C) const {
400  if (!FD)
401    return false;
402
403  if (FD->getKind() == Decl::Function) {
404    IdentifierInfo *FunI = FD->getIdentifier();
405    initIdentifierInfo(C);
406
407    if (FunI == II_malloc || FunI == II_realloc ||
408        FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
409        FunI == II_strdup || FunI == II_strndup)
410      return true;
411  }
412
413  if (Filter.CMallocOptimistic && FD->hasAttrs())
414    for (specific_attr_iterator<OwnershipAttr>
415           i = FD->specific_attr_begin<OwnershipAttr>(),
416           e = FD->specific_attr_end<OwnershipAttr>();
417           i != e; ++i)
418      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
419        return true;
420  return false;
421}
422
423bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
424  if (!FD)
425    return false;
426
427  if (FD->getKind() == Decl::Function) {
428    IdentifierInfo *FunI = FD->getIdentifier();
429    initIdentifierInfo(C);
430
431    if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
432      return true;
433  }
434
435  if (Filter.CMallocOptimistic && FD->hasAttrs())
436    for (specific_attr_iterator<OwnershipAttr>
437           i = FD->specific_attr_begin<OwnershipAttr>(),
438           e = FD->specific_attr_end<OwnershipAttr>();
439           i != e; ++i)
440      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
441          (*i)->getOwnKind() == OwnershipAttr::Holds)
442        return true;
443  return false;
444}
445
446void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
447  if (C.wasInlined)
448    return;
449
450  const FunctionDecl *FD = C.getCalleeDecl(CE);
451  if (!FD)
452    return;
453
454  ProgramStateRef State = C.getState();
455  bool ReleasedAllocatedMemory = false;
456
457  if (FD->getKind() == Decl::Function) {
458    initIdentifierInfo(C.getASTContext());
459    IdentifierInfo *FunI = FD->getIdentifier();
460
461    if (FunI == II_malloc || FunI == II_valloc) {
462      if (CE->getNumArgs() < 1)
463        return;
464      State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
465    } else if (FunI == II_realloc) {
466      State = ReallocMem(C, CE, false);
467    } else if (FunI == II_reallocf) {
468      State = ReallocMem(C, CE, true);
469    } else if (FunI == II_calloc) {
470      State = CallocMem(C, CE);
471    } else if (FunI == II_free) {
472      State = FreeMemAux(C, CE, State, 0, false, ReleasedAllocatedMemory);
473    } else if (FunI == II_strdup) {
474      State = MallocUpdateRefState(C, CE, State);
475    } else if (FunI == II_strndup) {
476      State = MallocUpdateRefState(C, CE, State);
477    }
478  }
479
480  if (Filter.CMallocOptimistic) {
481    // Check all the attributes, if there are any.
482    // There can be multiple of these attributes.
483    if (FD->hasAttrs())
484      for (specific_attr_iterator<OwnershipAttr>
485          i = FD->specific_attr_begin<OwnershipAttr>(),
486          e = FD->specific_attr_end<OwnershipAttr>();
487          i != e; ++i) {
488        switch ((*i)->getOwnKind()) {
489        case OwnershipAttr::Returns:
490          State = MallocMemReturnsAttr(C, CE, *i);
491          break;
492        case OwnershipAttr::Takes:
493        case OwnershipAttr::Holds:
494          State = FreeMemAttr(C, CE, *i);
495          break;
496        }
497      }
498  }
499  C.addTransition(State);
500}
501
502static bool isFreeWhenDoneSetToZero(const ObjCMethodCall &Call) {
503  Selector S = Call.getSelector();
504  for (unsigned i = 1; i < S.getNumArgs(); ++i)
505    if (S.getNameForSlot(i).equals("freeWhenDone"))
506      if (Call.getArgSVal(i).isConstant(0))
507        return true;
508
509  return false;
510}
511
512void MallocChecker::checkPreObjCMessage(const ObjCMethodCall &Call,
513                                        CheckerContext &C) const {
514  // If the first selector is dataWithBytesNoCopy, assume that the memory will
515  // be released with 'free' by the new object.
516  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
517  // Unless 'freeWhenDone' param set to 0.
518  // TODO: Check that the memory was allocated with malloc.
519  bool ReleasedAllocatedMemory = false;
520  Selector S = Call.getSelector();
521  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
522       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
523       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
524      !isFreeWhenDoneSetToZero(Call)){
525    unsigned int argIdx  = 0;
526    C.addTransition(FreeMemAux(C, Call.getArgExpr(argIdx),
527                    Call.getOriginExpr(), C.getState(), true,
528                    ReleasedAllocatedMemory));
529  }
530}
531
532ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
533                                                    const CallExpr *CE,
534                                                    const OwnershipAttr* Att) {
535  if (Att->getModule() != "malloc")
536    return 0;
537
538  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
539  if (I != E) {
540    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
541  }
542  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
543}
544
545ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
546                                           const CallExpr *CE,
547                                           SVal Size, SVal Init,
548                                           ProgramStateRef state) {
549
550  // Bind the return value to the symbolic value from the heap region.
551  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
552  // side effects other than what we model here.
553  unsigned Count = C.blockCount();
554  SValBuilder &svalBuilder = C.getSValBuilder();
555  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
556  DefinedSVal RetVal =
557    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
558  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
559
560  // We expect the malloc functions to return a pointer.
561  if (!isa<Loc>(RetVal))
562    return 0;
563
564  // Fill the region with the initialization value.
565  state = state->bindDefault(RetVal, Init);
566
567  // Set the region's extent equal to the Size parameter.
568  const SymbolicRegion *R =
569      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
570  if (!R)
571    return 0;
572  if (isa<DefinedOrUnknownSVal>(Size)) {
573    SValBuilder &svalBuilder = C.getSValBuilder();
574    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
575    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
576    DefinedOrUnknownSVal extentMatchesSize =
577        svalBuilder.evalEQ(state, Extent, DefinedSize);
578
579    state = state->assume(extentMatchesSize, true);
580    assert(state);
581  }
582
583  return MallocUpdateRefState(C, CE, state);
584}
585
586ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
587                                                    const CallExpr *CE,
588                                                    ProgramStateRef state) {
589  // Get the return value.
590  SVal retVal = state->getSVal(CE, C.getLocationContext());
591
592  // We expect the malloc functions to return a pointer.
593  if (!isa<Loc>(retVal))
594    return 0;
595
596  SymbolRef Sym = retVal.getAsLocSymbol();
597  assert(Sym);
598
599  // Set the symbol's state to Allocated.
600  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
601
602}
603
604ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
605                                           const CallExpr *CE,
606                                           const OwnershipAttr* Att) const {
607  if (Att->getModule() != "malloc")
608    return 0;
609
610  ProgramStateRef State = C.getState();
611  bool ReleasedAllocated = false;
612
613  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
614       I != E; ++I) {
615    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
616                               Att->getOwnKind() == OwnershipAttr::Holds,
617                               ReleasedAllocated);
618    if (StateI)
619      State = StateI;
620  }
621  return State;
622}
623
624ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
625                                          const CallExpr *CE,
626                                          ProgramStateRef state,
627                                          unsigned Num,
628                                          bool Hold,
629                                          bool &ReleasedAllocated) const {
630  if (CE->getNumArgs() < (Num + 1))
631    return 0;
632
633  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold, ReleasedAllocated);
634}
635
636ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
637                                          const Expr *ArgExpr,
638                                          const Expr *ParentExpr,
639                                          ProgramStateRef state,
640                                          bool Hold,
641                                          bool &ReleasedAllocated) const {
642
643  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
644  if (!isa<DefinedOrUnknownSVal>(ArgVal))
645    return 0;
646  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
647
648  // Check for null dereferences.
649  if (!isa<Loc>(location))
650    return 0;
651
652  // The explicit NULL case, no operation is performed.
653  ProgramStateRef notNullState, nullState;
654  llvm::tie(notNullState, nullState) = state->assume(location);
655  if (nullState && !notNullState)
656    return 0;
657
658  // Unknown values could easily be okay
659  // Undefined values are handled elsewhere
660  if (ArgVal.isUnknownOrUndef())
661    return 0;
662
663  const MemRegion *R = ArgVal.getAsRegion();
664
665  // Nonlocs can't be freed, of course.
666  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
667  if (!R) {
668    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
669    return 0;
670  }
671
672  R = R->StripCasts();
673
674  // Blocks might show up as heap data, but should not be free()d
675  if (isa<BlockDataRegion>(R)) {
676    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
677    return 0;
678  }
679
680  const MemSpaceRegion *MS = R->getMemorySpace();
681
682  // Parameters, locals, statics, and globals shouldn't be freed.
683  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
684    // FIXME: at the time this code was written, malloc() regions were
685    // represented by conjured symbols, which are all in UnknownSpaceRegion.
686    // This means that there isn't actually anything from HeapSpaceRegion
687    // that should be freed, even though we allow it here.
688    // Of course, free() can work on memory allocated outside the current
689    // function, so UnknownSpaceRegion is always a possibility.
690    // False negatives are better than false positives.
691
692    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
693    return 0;
694  }
695
696  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
697  // Various cases could lead to non-symbol values here.
698  // For now, ignore them.
699  if (!SR)
700    return 0;
701
702  SymbolRef Sym = SR->getSymbol();
703  const RefState *RS = state->get<RegionState>(Sym);
704
705  // Check double free.
706  if (RS && (RS->isReleased() || RS->isRelinquished())) {
707    if (ExplodedNode *N = C.generateSink()) {
708      if (!BT_DoubleFree)
709        BT_DoubleFree.reset(
710          new BugType("Double free", "Memory Error"));
711      BugReport *R = new BugReport(*BT_DoubleFree,
712        (RS->isReleased() ? "Attempt to free released memory" :
713                            "Attempt to free non-owned memory"), N);
714      R->addRange(ArgExpr->getSourceRange());
715      R->markInteresting(Sym);
716      R->addVisitor(new MallocBugVisitor(Sym));
717      C.EmitReport(R);
718    }
719    return 0;
720  }
721
722  ReleasedAllocated = (RS != 0);
723
724  // Normal free.
725  if (Hold)
726    return state->set<RegionState>(Sym, RefState::getRelinquished(ParentExpr));
727  return state->set<RegionState>(Sym, RefState::getReleased(ParentExpr));
728}
729
730bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
731  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
732    os << "an integer (" << IntVal->getValue() << ")";
733  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
734    os << "a constant address (" << ConstAddr->getValue() << ")";
735  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
736    os << "the address of the label '" << Label->getLabel()->getName() << "'";
737  else
738    return false;
739
740  return true;
741}
742
743bool MallocChecker::SummarizeRegion(raw_ostream &os,
744                                    const MemRegion *MR) {
745  switch (MR->getKind()) {
746  case MemRegion::FunctionTextRegionKind: {
747    const NamedDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
748    if (FD)
749      os << "the address of the function '" << *FD << '\'';
750    else
751      os << "the address of a function";
752    return true;
753  }
754  case MemRegion::BlockTextRegionKind:
755    os << "block text";
756    return true;
757  case MemRegion::BlockDataRegionKind:
758    // FIXME: where the block came from?
759    os << "a block";
760    return true;
761  default: {
762    const MemSpaceRegion *MS = MR->getMemorySpace();
763
764    if (isa<StackLocalsSpaceRegion>(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        os << "the address of the local variable '" << VD->getName() << "'";
774      else
775        os << "the address of a local stack variable";
776      return true;
777    }
778
779    if (isa<StackArgumentsSpaceRegion>(MS)) {
780      const VarRegion *VR = dyn_cast<VarRegion>(MR);
781      const VarDecl *VD;
782      if (VR)
783        VD = VR->getDecl();
784      else
785        VD = NULL;
786
787      if (VD)
788        os << "the address of the parameter '" << VD->getName() << "'";
789      else
790        os << "the address of a parameter";
791      return true;
792    }
793
794    if (isa<GlobalsSpaceRegion>(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        if (VD->isStaticLocal())
804          os << "the address of the static variable '" << VD->getName() << "'";
805        else
806          os << "the address of the global variable '" << VD->getName() << "'";
807      } else
808        os << "the address of a global variable";
809      return true;
810    }
811
812    return false;
813  }
814  }
815}
816
817void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
818                                  SourceRange range) const {
819  if (ExplodedNode *N = C.generateSink()) {
820    if (!BT_BadFree)
821      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
822
823    SmallString<100> buf;
824    llvm::raw_svector_ostream os(buf);
825
826    const MemRegion *MR = ArgVal.getAsRegion();
827    if (MR) {
828      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
829        MR = ER->getSuperRegion();
830
831      // Special case for alloca()
832      if (isa<AllocaRegion>(MR))
833        os << "Argument to free() was allocated by alloca(), not malloc()";
834      else {
835        os << "Argument to free() is ";
836        if (SummarizeRegion(os, MR))
837          os << ", which is not memory allocated by malloc()";
838        else
839          os << "not memory allocated by malloc()";
840      }
841    } else {
842      os << "Argument to free() is ";
843      if (SummarizeValue(os, ArgVal))
844        os << ", which is not memory allocated by malloc()";
845      else
846        os << "not memory allocated by malloc()";
847    }
848
849    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
850    R->markInteresting(MR);
851    R->addRange(range);
852    C.EmitReport(R);
853  }
854}
855
856ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
857                                          const CallExpr *CE,
858                                          bool FreesOnFail) const {
859  if (CE->getNumArgs() < 2)
860    return 0;
861
862  ProgramStateRef state = C.getState();
863  const Expr *arg0Expr = CE->getArg(0);
864  const LocationContext *LCtx = C.getLocationContext();
865  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
866  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
867    return 0;
868  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
869
870  SValBuilder &svalBuilder = C.getSValBuilder();
871
872  DefinedOrUnknownSVal PtrEQ =
873    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
874
875  // Get the size argument. If there is no size arg then give up.
876  const Expr *Arg1 = CE->getArg(1);
877  if (!Arg1)
878    return 0;
879
880  // Get the value of the size argument.
881  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
882  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
883    return 0;
884  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
885
886  // Compare the size argument to 0.
887  DefinedOrUnknownSVal SizeZero =
888    svalBuilder.evalEQ(state, Arg1Val,
889                       svalBuilder.makeIntValWithPtrWidth(0, false));
890
891  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
892  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
893  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
894  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
895  // We only assume exceptional states if they are definitely true; if the
896  // state is under-constrained, assume regular realloc behavior.
897  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
898  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
899
900  // If the ptr is NULL and the size is not 0, the call is equivalent to
901  // malloc(size).
902  if ( PrtIsNull && !SizeIsZero) {
903    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
904                                               UndefinedVal(), StatePtrIsNull);
905    return stateMalloc;
906  }
907
908  if (PrtIsNull && SizeIsZero)
909    return 0;
910
911  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
912  assert(!PrtIsNull);
913  SymbolRef FromPtr = arg0Val.getAsSymbol();
914  SVal RetVal = state->getSVal(CE, LCtx);
915  SymbolRef ToPtr = RetVal.getAsSymbol();
916  if (!FromPtr || !ToPtr)
917    return 0;
918
919  bool ReleasedAllocated = false;
920
921  // If the size is 0, free the memory.
922  if (SizeIsZero)
923    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero, 0,
924                                               false, ReleasedAllocated)){
925      // The semantics of the return value are:
926      // If size was equal to 0, either NULL or a pointer suitable to be passed
927      // to free() is returned. We just free the input pointer and do not add
928      // any constrains on the output pointer.
929      return stateFree;
930    }
931
932  // Default behavior.
933  if (ProgramStateRef stateFree =
934        FreeMemAux(C, CE, state, 0, false, ReleasedAllocated)) {
935
936    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
937                                                UnknownVal(), stateFree);
938    if (!stateRealloc)
939      return 0;
940
941    ReallocPairKind Kind = RPToBeFreedAfterFailure;
942    if (FreesOnFail)
943      Kind = RPIsFreeOnFailure;
944    else if (!ReleasedAllocated)
945      Kind = RPDoNotTrackAfterFailure;
946
947    // Record the info about the reallocated symbol so that we could properly
948    // process failed reallocation.
949    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
950                                                   ReallocPair(FromPtr, Kind));
951    // The reallocated symbol should stay alive for as long as the new symbol.
952    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
953    return stateRealloc;
954  }
955  return 0;
956}
957
958ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
959  if (CE->getNumArgs() < 2)
960    return 0;
961
962  ProgramStateRef state = C.getState();
963  SValBuilder &svalBuilder = C.getSValBuilder();
964  const LocationContext *LCtx = C.getLocationContext();
965  SVal count = state->getSVal(CE->getArg(0), LCtx);
966  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
967  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
968                                        svalBuilder.getContext().getSizeType());
969  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
970
971  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
972}
973
974LeakInfo
975MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
976                                 CheckerContext &C) const {
977  const LocationContext *LeakContext = N->getLocationContext();
978  // Walk the ExplodedGraph backwards and find the first node that referred to
979  // the tracked symbol.
980  const ExplodedNode *AllocNode = N;
981  const MemRegion *ReferenceRegion = 0;
982
983  while (N) {
984    ProgramStateRef State = N->getState();
985    if (!State->get<RegionState>(Sym))
986      break;
987
988    // Find the most recent expression bound to the symbol in the current
989    // context.
990    if (!ReferenceRegion) {
991      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
992        SVal Val = State->getSVal(MR);
993        if (Val.getAsLocSymbol() == Sym)
994          ReferenceRegion = MR;
995      }
996    }
997
998    // Allocation node, is the last node in the current context in which the
999    // symbol was tracked.
1000    if (N->getLocationContext() == LeakContext)
1001      AllocNode = N;
1002    N = N->pred_empty() ? NULL : *(N->pred_begin());
1003  }
1004
1005  ProgramPoint P = AllocNode->getLocation();
1006  const Stmt *AllocationStmt = 0;
1007  if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&P))
1008    AllocationStmt = Exit->getCalleeContext()->getCallSite();
1009  else if (StmtPoint *SP = dyn_cast<StmtPoint>(&P))
1010    AllocationStmt = SP->getStmt();
1011
1012  return LeakInfo(AllocationStmt, ReferenceRegion);
1013}
1014
1015void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
1016                               CheckerContext &C) const {
1017  assert(N);
1018  if (!BT_Leak) {
1019    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
1020    // Leaks should not be reported if they are post-dominated by a sink:
1021    // (1) Sinks are higher importance bugs.
1022    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
1023    //     with __noreturn functions such as assert() or exit(). We choose not
1024    //     to report leaks on such paths.
1025    BT_Leak->setSuppressOnSink(true);
1026  }
1027
1028  // Most bug reports are cached at the location where they occurred.
1029  // With leaks, we want to unique them by the location where they were
1030  // allocated, and only report a single path.
1031  PathDiagnosticLocation LocUsedForUniqueing;
1032  const Stmt *AllocStmt = 0;
1033  const MemRegion *Region = 0;
1034  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
1035  if (AllocStmt)
1036    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
1037                            C.getSourceManager(), N->getLocationContext());
1038
1039  SmallString<200> buf;
1040  llvm::raw_svector_ostream os(buf);
1041  os << "Memory is never released; potential leak";
1042  if (Region && Region->canPrintPretty()) {
1043    os << " of memory pointed to by '";
1044    Region->printPretty(os);
1045    os << '\'';
1046  }
1047
1048  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
1049  R->markInteresting(Sym);
1050  R->addVisitor(new MallocBugVisitor(Sym, true));
1051  C.EmitReport(R);
1052}
1053
1054void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1055                                     CheckerContext &C) const
1056{
1057  if (!SymReaper.hasDeadSymbols())
1058    return;
1059
1060  ProgramStateRef state = C.getState();
1061  RegionStateTy RS = state->get<RegionState>();
1062  RegionStateTy::Factory &F = state->get_context<RegionState>();
1063
1064  llvm::SmallVector<SymbolRef, 2> Errors;
1065  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1066    if (SymReaper.isDead(I->first)) {
1067      if (I->second.isAllocated())
1068        Errors.push_back(I->first);
1069      // Remove the dead symbol from the map.
1070      RS = F.remove(RS, I->first);
1071
1072    }
1073  }
1074
1075  // Cleanup the Realloc Pairs Map.
1076  ReallocMap RP = state->get<ReallocPairs>();
1077  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1078    if (SymReaper.isDead(I->first) ||
1079        SymReaper.isDead(I->second.ReallocatedSym)) {
1080      state = state->remove<ReallocPairs>(I->first);
1081    }
1082  }
1083
1084  // Generate leak node.
1085  ExplodedNode *N = C.getPredecessor();
1086  if (!Errors.empty()) {
1087    static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1088    N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1089    for (llvm::SmallVector<SymbolRef, 2>::iterator
1090        I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1091      reportLeak(*I, N, C);
1092    }
1093  }
1094
1095  C.addTransition(state->set<RegionState>(RS), N);
1096}
1097
1098void MallocChecker::checkEndPath(CheckerContext &C) const {
1099  ProgramStateRef state = C.getState();
1100  RegionStateTy M = state->get<RegionState>();
1101
1102  // If inside inlined call, skip it.
1103  if (C.getLocationContext()->getParent() != 0)
1104    return;
1105
1106  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1107    RefState RS = I->second;
1108    if (RS.isAllocated()) {
1109      ExplodedNode *N = C.addTransition(state);
1110      if (N)
1111        reportLeak(I->first, N, C);
1112    }
1113  }
1114}
1115
1116void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1117  // We will check for double free in the post visit.
1118  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1119    return;
1120
1121  // Check use after free, when a freed pointer is passed to a call.
1122  ProgramStateRef State = C.getState();
1123  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1124                                    E = CE->arg_end(); I != E; ++I) {
1125    const Expr *A = *I;
1126    if (A->getType().getTypePtr()->isAnyPointerType()) {
1127      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1128      if (!Sym)
1129        continue;
1130      if (checkUseAfterFree(Sym, C, A))
1131        return;
1132    }
1133  }
1134}
1135
1136void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1137  const Expr *E = S->getRetValue();
1138  if (!E)
1139    return;
1140
1141  // Check if we are returning a symbol.
1142  ProgramStateRef State = C.getState();
1143  SVal RetVal = State->getSVal(E, C.getLocationContext());
1144  SymbolRef Sym = RetVal.getAsSymbol();
1145  if (!Sym)
1146    // If we are returning a field of the allocated struct or an array element,
1147    // the callee could still free the memory.
1148    // TODO: This logic should be a part of generic symbol escape callback.
1149    if (const MemRegion *MR = RetVal.getAsRegion())
1150      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1151        if (const SymbolicRegion *BMR =
1152              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1153          Sym = BMR->getSymbol();
1154
1155  // Check if we are returning freed memory.
1156  if (Sym)
1157    if (checkUseAfterFree(Sym, C, E))
1158      return;
1159
1160  // If this function body is not inlined, stop tracking any returned symbols.
1161  if (C.getLocationContext()->getParent() == 0) {
1162    State =
1163      State->scanReachableSymbols<StopTrackingCallback>(RetVal).getState();
1164    C.addTransition(State);
1165  }
1166}
1167
1168// TODO: Blocks should be either inlined or should call invalidate regions
1169// upon invocation. After that's in place, special casing here will not be
1170// needed.
1171void MallocChecker::checkPostStmt(const BlockExpr *BE,
1172                                  CheckerContext &C) const {
1173
1174  // Scan the BlockDecRefExprs for any object the retain count checker
1175  // may be tracking.
1176  if (!BE->getBlockDecl()->hasCaptures())
1177    return;
1178
1179  ProgramStateRef state = C.getState();
1180  const BlockDataRegion *R =
1181    cast<BlockDataRegion>(state->getSVal(BE,
1182                                         C.getLocationContext()).getAsRegion());
1183
1184  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1185                                            E = R->referenced_vars_end();
1186
1187  if (I == E)
1188    return;
1189
1190  SmallVector<const MemRegion*, 10> Regions;
1191  const LocationContext *LC = C.getLocationContext();
1192  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1193
1194  for ( ; I != E; ++I) {
1195    const VarRegion *VR = *I;
1196    if (VR->getSuperRegion() == R) {
1197      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1198    }
1199    Regions.push_back(VR);
1200  }
1201
1202  state =
1203    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1204                                    Regions.data() + Regions.size()).getState();
1205  C.addTransition(state);
1206}
1207
1208bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1209  assert(Sym);
1210  const RefState *RS = C.getState()->get<RegionState>(Sym);
1211  return (RS && RS->isReleased());
1212}
1213
1214bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1215                                      const Stmt *S) const {
1216  if (isReleased(Sym, C)) {
1217    if (ExplodedNode *N = C.generateSink()) {
1218      if (!BT_UseFree)
1219        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1220
1221      BugReport *R = new BugReport(*BT_UseFree,
1222                                   "Use of memory after it is freed",N);
1223      if (S)
1224        R->addRange(S->getSourceRange());
1225      R->markInteresting(Sym);
1226      R->addVisitor(new MallocBugVisitor(Sym));
1227      C.EmitReport(R);
1228      return true;
1229    }
1230  }
1231  return false;
1232}
1233
1234// Check if the location is a freed symbolic region.
1235void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1236                                  CheckerContext &C) const {
1237  SymbolRef Sym = l.getLocSymbolInBase();
1238  if (Sym)
1239    checkUseAfterFree(Sym, C, S);
1240}
1241
1242//===----------------------------------------------------------------------===//
1243// Check various ways a symbol can be invalidated.
1244// TODO: This logic (the next 3 functions) is copied/similar to the
1245// RetainRelease checker. We might want to factor this out.
1246//===----------------------------------------------------------------------===//
1247
1248// Stop tracking symbols when a value escapes as a result of checkBind.
1249// A value escapes in three possible cases:
1250// (1) we are binding to something that is not a memory region.
1251// (2) we are binding to a memregion that does not have stack storage
1252// (3) we are binding to a memregion with stack storage that the store
1253//     does not understand.
1254void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1255                              CheckerContext &C) const {
1256  // Are we storing to something that causes the value to "escape"?
1257  bool escapes = true;
1258  ProgramStateRef state = C.getState();
1259
1260  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1261    escapes = !regionLoc->getRegion()->hasStackStorage();
1262
1263    if (!escapes) {
1264      // To test (3), generate a new state with the binding added.  If it is
1265      // the same state, then it escapes (since the store cannot represent
1266      // the binding).
1267      // Do this only if we know that the store is not supposed to generate the
1268      // same state.
1269      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1270      if (StoredVal != val)
1271        escapes = (state == (state->bindLoc(*regionLoc, val)));
1272    }
1273  }
1274
1275  // If our store can represent the binding and we aren't storing to something
1276  // that doesn't have local storage then just return and have the simulation
1277  // state continue as is.
1278  if (!escapes)
1279      return;
1280
1281  // Otherwise, find all symbols referenced by 'val' that we are tracking
1282  // and stop tracking them.
1283  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1284  C.addTransition(state);
1285}
1286
1287// If a symbolic region is assumed to NULL (or another constant), stop tracking
1288// it - assuming that allocation failed on this path.
1289ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1290                                              SVal Cond,
1291                                              bool Assumption) const {
1292  RegionStateTy RS = state->get<RegionState>();
1293  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1294    // If the symbol is assumed to be NULL, remove it from consideration.
1295    ConstraintManager &CMgr = state->getConstraintManager();
1296    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1297    if (AllocFailed.isConstrainedTrue())
1298      state = state->remove<RegionState>(I.getKey());
1299  }
1300
1301  // Realloc returns 0 when reallocation fails, which means that we should
1302  // restore the state of the pointer being reallocated.
1303  ReallocMap RP = state->get<ReallocPairs>();
1304  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1305    // If the symbol is assumed to be NULL, remove it from consideration.
1306    ConstraintManager &CMgr = state->getConstraintManager();
1307    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1308    if (AllocFailed.isConstrainedTrue())
1309      continue;
1310
1311    SymbolRef ReallocSym = I.getData().ReallocatedSym;
1312    if (const RefState *RS = state->get<RegionState>(ReallocSym)) {
1313      if (RS->isReleased()) {
1314        if (I.getData().Kind == RPToBeFreedAfterFailure)
1315          state = state->set<RegionState>(ReallocSym,
1316              RefState::getAllocated(RS->getStmt()));
1317        else if (I.getData().Kind == RPDoNotTrackAfterFailure)
1318          state = state->remove<RegionState>(ReallocSym);
1319        else
1320          assert(I.getData().Kind == RPIsFreeOnFailure);
1321      }
1322    }
1323    state = state->remove<ReallocPairs>(I.getKey());
1324  }
1325
1326  return state;
1327}
1328
1329// Check if the function is known to us. So, for example, we could
1330// conservatively assume it can free/reallocate its pointer arguments.
1331// (We assume that the pointers cannot escape through calls to system
1332// functions not handled by this checker.)
1333bool MallocChecker::doesNotFreeMemory(const CallEvent *Call,
1334                                      ProgramStateRef State) const {
1335  assert(Call);
1336
1337  // For now, assume that any C++ call can free memory.
1338  // TODO: If we want to be more optimistic here, we'll need to make sure that
1339  // regions escape to C++ containers. They seem to do that even now, but for
1340  // mysterious reasons.
1341  if (!(isa<FunctionCall>(Call) || isa<ObjCMethodCall>(Call)))
1342    return false;
1343
1344  // Check Objective-C messages by selector name.
1345  if (const ObjCMethodCall *Msg = dyn_cast<ObjCMethodCall>(Call)) {
1346    // If it's not a framework call, or if it takes a callback, assume it
1347    // can free memory.
1348    if (!Call->isInSystemHeader() || Call->hasNonZeroCallbackArg())
1349      return false;
1350
1351    Selector S = Msg->getSelector();
1352
1353    // Whitelist the ObjC methods which do free memory.
1354    // - Anything containing 'freeWhenDone' param set to 1.
1355    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1356    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1357      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1358        if (Call->getArgSVal(i).isConstant(1))
1359          return false;
1360        else
1361          return true;
1362      }
1363    }
1364
1365    // If the first selector ends with NoCopy, assume that the ownership is
1366    // transferred as well.
1367    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1368    StringRef FirstSlot = S.getNameForSlot(0);
1369    if (FirstSlot.endswith("NoCopy"))
1370      return false;
1371
1372    // If the first selector starts with addPointer, insertPointer,
1373    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1374    // This is similar to C++ containers (vector); we still might want to check
1375    // that the pointers get freed by following the container itself.
1376    if (FirstSlot.startswith("addPointer") ||
1377        FirstSlot.startswith("insertPointer") ||
1378        FirstSlot.startswith("replacePointer")) {
1379      return false;
1380    }
1381
1382    // Otherwise, assume that the method does not free memory.
1383    // Most framework methods do not free memory.
1384    return true;
1385  }
1386
1387  // At this point the only thing left to handle is straight function calls.
1388  const FunctionDecl *FD = cast<FunctionCall>(Call)->getDecl();
1389  if (!FD)
1390    return false;
1391
1392  ASTContext &ASTC = State->getStateManager().getContext();
1393
1394  // If it's one of the allocation functions we can reason about, we model
1395  // its behavior explicitly.
1396  if (isMemFunction(FD, ASTC))
1397    return true;
1398
1399  // If it's not a system call, assume it frees memory.
1400  if (!Call->isInSystemHeader())
1401    return false;
1402
1403  // White list the system functions whose arguments escape.
1404  const IdentifierInfo *II = FD->getIdentifier();
1405  if (!II)
1406    return false;
1407  StringRef FName = II->getName();
1408
1409  // White list the 'XXXNoCopy' CoreFoundation functions.
1410  // We specifically check these before
1411  if (FName.endswith("NoCopy")) {
1412    // Look for the deallocator argument. We know that the memory ownership
1413    // is not transferred only if the deallocator argument is
1414    // 'kCFAllocatorNull'.
1415    for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1416      const Expr *ArgE = Call->getArgExpr(i)->IgnoreParenCasts();
1417      if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1418        StringRef DeallocatorName = DE->getFoundDecl()->getName();
1419        if (DeallocatorName == "kCFAllocatorNull")
1420          return true;
1421      }
1422    }
1423    return false;
1424  }
1425
1426  // Associating streams with malloced buffers. The pointer can escape if
1427  // 'closefn' is specified (and if that function does free memory),
1428  // but it will not if closefn is not specified.
1429  // Currently, we do not inspect the 'closefn' function (PR12101).
1430  if (FName == "funopen")
1431    if (Call->getNumArgs() >= 4 && Call->getArgSVal(4).isConstant(0))
1432      return true;
1433
1434  // Do not warn on pointers passed to 'setbuf' when used with std streams,
1435  // these leaks might be intentional when setting the buffer for stdio.
1436  // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1437  if (FName == "setbuf" || FName =="setbuffer" ||
1438      FName == "setlinebuf" || FName == "setvbuf") {
1439    if (Call->getNumArgs() >= 1) {
1440      const Expr *ArgE = Call->getArgExpr(0)->IgnoreParenCasts();
1441      if (const DeclRefExpr *ArgDRE = dyn_cast<DeclRefExpr>(ArgE))
1442        if (const VarDecl *D = dyn_cast<VarDecl>(ArgDRE->getDecl()))
1443          if (D->getCanonicalDecl()->getName().find("std") != StringRef::npos)
1444            return false;
1445    }
1446  }
1447
1448  // A bunch of other functions which either take ownership of a pointer or
1449  // wrap the result up in a struct or object, meaning it can be freed later.
1450  // (See RetainCountChecker.) Not all the parameters here are invalidated,
1451  // but the Malloc checker cannot differentiate between them. The right way
1452  // of doing this would be to implement a pointer escapes callback.
1453  if (FName == "CGBitmapContextCreate" ||
1454      FName == "CGBitmapContextCreateWithData" ||
1455      FName == "CVPixelBufferCreateWithBytes" ||
1456      FName == "CVPixelBufferCreateWithPlanarBytes" ||
1457      FName == "OSAtomicEnqueue") {
1458    return false;
1459  }
1460
1461  // Handle cases where we know a buffer's /address/ can escape.
1462  // Note that the above checks handle some special cases where we know that
1463  // even though the address escapes, it's still our responsibility to free the
1464  // buffer.
1465  if (Call->argumentsMayEscape())
1466    return false;
1467
1468  // Otherwise, assume that the function does not free memory.
1469  // Most system calls do not free the memory.
1470  return true;
1471}
1472
1473// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1474// escapes, when we are tracking p), do not track the symbol as we cannot reason
1475// about it anymore.
1476ProgramStateRef
1477MallocChecker::checkRegionChanges(ProgramStateRef State,
1478                            const StoreManager::InvalidatedSymbols *invalidated,
1479                                    ArrayRef<const MemRegion *> ExplicitRegions,
1480                                    ArrayRef<const MemRegion *> Regions,
1481                                    const CallEvent *Call) const {
1482  if (!invalidated || invalidated->empty())
1483    return State;
1484  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1485
1486  // If it's a call which might free or reallocate memory, we assume that all
1487  // regions (explicit and implicit) escaped.
1488
1489  // Otherwise, whitelist explicit pointers; we still can track them.
1490  if (!Call || doesNotFreeMemory(Call, State)) {
1491    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1492        E = ExplicitRegions.end(); I != E; ++I) {
1493      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1494        WhitelistedSymbols.insert(R->getSymbol());
1495    }
1496  }
1497
1498  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1499       E = invalidated->end(); I!=E; ++I) {
1500    SymbolRef sym = *I;
1501    if (WhitelistedSymbols.count(sym))
1502      continue;
1503    // The symbol escaped. Note, we assume that if the symbol is released,
1504    // passing it out will result in a use after free. We also keep tracking
1505    // relinquished symbols.
1506    if (const RefState *RS = State->get<RegionState>(sym)) {
1507      if (RS->isAllocated())
1508        State = State->remove<RegionState>(sym);
1509    }
1510  }
1511  return State;
1512}
1513
1514static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1515                                         ProgramStateRef prevState) {
1516  ReallocMap currMap = currState->get<ReallocPairs>();
1517  ReallocMap prevMap = prevState->get<ReallocPairs>();
1518
1519  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1520       I != E; ++I) {
1521    SymbolRef sym = I.getKey();
1522    if (!currMap.lookup(sym))
1523      return sym;
1524  }
1525
1526  return NULL;
1527}
1528
1529PathDiagnosticPiece *
1530MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1531                                           const ExplodedNode *PrevN,
1532                                           BugReporterContext &BRC,
1533                                           BugReport &BR) {
1534  ProgramStateRef state = N->getState();
1535  ProgramStateRef statePrev = PrevN->getState();
1536
1537  const RefState *RS = state->get<RegionState>(Sym);
1538  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1539  if (!RS)
1540    return 0;
1541
1542  const Stmt *S = 0;
1543  const char *Msg = 0;
1544  StackHintGeneratorForSymbol *StackHint = 0;
1545
1546  // Retrieve the associated statement.
1547  ProgramPoint ProgLoc = N->getLocation();
1548  if (StmtPoint *SP = dyn_cast<StmtPoint>(&ProgLoc))
1549    S = SP->getStmt();
1550  else if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&ProgLoc))
1551    S = Exit->getCalleeContext()->getCallSite();
1552  // If an assumption was made on a branch, it should be caught
1553  // here by looking at the state transition.
1554  else if (BlockEdge *Edge = dyn_cast<BlockEdge>(&ProgLoc)) {
1555    const CFGBlock *srcBlk = Edge->getSrc();
1556    S = srcBlk->getTerminator();
1557  }
1558  if (!S)
1559    return 0;
1560
1561  // FIXME: We will eventually need to handle non-statement-based events
1562  // (__attribute__((cleanup))).
1563
1564  // Find out if this is an interesting point and what is the kind.
1565  if (Mode == Normal) {
1566    if (isAllocated(RS, RSPrev, S)) {
1567      Msg = "Memory is allocated";
1568      StackHint = new StackHintGeneratorForSymbol(Sym,
1569                                                  "Returned allocated memory");
1570    } else if (isReleased(RS, RSPrev, S)) {
1571      Msg = "Memory is released";
1572      StackHint = new StackHintGeneratorForSymbol(Sym,
1573                                                  "Returned released memory");
1574    } else if (isRelinquished(RS, RSPrev, S)) {
1575      Msg = "Memory ownership is transfered";
1576      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1577    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1578      Mode = ReallocationFailed;
1579      Msg = "Reallocation failed";
1580      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1581                                                       "Reallocation failed");
1582
1583      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1584        // Is it possible to fail two reallocs WITHOUT testing in between?
1585        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1586          "We only support one failed realloc at a time.");
1587        BR.markInteresting(sym);
1588        FailedReallocSymbol = sym;
1589      }
1590    }
1591
1592  // We are in a special mode if a reallocation failed later in the path.
1593  } else if (Mode == ReallocationFailed) {
1594    assert(FailedReallocSymbol && "No symbol to look for.");
1595
1596    // Is this is the first appearance of the reallocated symbol?
1597    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1598      // We're at the reallocation point.
1599      Msg = "Attempt to reallocate memory";
1600      StackHint = new StackHintGeneratorForSymbol(Sym,
1601                                                 "Returned reallocated memory");
1602      FailedReallocSymbol = NULL;
1603      Mode = Normal;
1604    }
1605  }
1606
1607  if (!Msg)
1608    return 0;
1609  assert(StackHint);
1610
1611  // Generate the extra diagnostic.
1612  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1613                             N->getLocationContext());
1614  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1615}
1616
1617void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1618                               const char *NL, const char *Sep) const {
1619
1620  RegionStateTy RS = State->get<RegionState>();
1621
1622  if (!RS.isEmpty())
1623    Out << "Has Malloc data" << NL;
1624}
1625
1626#define REGISTER_CHECKER(name) \
1627void ento::register##name(CheckerManager &mgr) {\
1628  registerCStringCheckerBasic(mgr); \
1629  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1630}
1631
1632REGISTER_CHECKER(MallocPessimistic)
1633REGISTER_CHECKER(MallocOptimistic)
1634