MallocChecker.cpp revision d9ab7bbbd43639247c9a82c8e8dbfaa617f4e266 
1//=== MallocChecker.cpp - A malloc/free checker -------------------*- C++ -*--//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file defines malloc/free checker, which checks for potential memory
11// leaks, double free, and use-after-free problems.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "InterCheckerAPI.h"
17#include "clang/StaticAnalyzer/Core/Checker.h"
18#include "clang/StaticAnalyzer/Core/CheckerManager.h"
19#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
20#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/ObjCMessage.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
24#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
25#include "clang/Basic/SourceManager.h"
26#include "llvm/ADT/ImmutableMap.h"
27#include "llvm/ADT/SmallString.h"
28#include "llvm/ADT/STLExtras.h"
29using namespace clang;
30using namespace ento;
31
32namespace {
33
34class RefState {
35  enum Kind { AllocateUnchecked, AllocateFailed, Released, Escaped,
36              Relinquished } K;
37  const Stmt *S;
38
39public:
40  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
41
42  bool isAllocated() const { return K == AllocateUnchecked; }
43  //bool isFailed() const { return K == AllocateFailed; }
44  bool isReleased() const { return K == Released; }
45  //bool isEscaped() const { return K == Escaped; }
46  //bool isRelinquished() const { return K == Relinquished; }
47  const Stmt *getStmt() const { return S; }
48
49  bool operator==(const RefState &X) const {
50    return K == X.K && S == X.S;
51  }
52
53  static RefState getAllocateUnchecked(const Stmt *s) {
54    return RefState(AllocateUnchecked, s);
55  }
56  static RefState getAllocateFailed() {
57    return RefState(AllocateFailed, 0);
58  }
59  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
60  static RefState getEscaped(const Stmt *s) { return RefState(Escaped, s); }
61  static RefState getRelinquished(const Stmt *s) {
62    return RefState(Relinquished, s);
63  }
64
65  void Profile(llvm::FoldingSetNodeID &ID) const {
66    ID.AddInteger(K);
67    ID.AddPointer(S);
68  }
69};
70
71struct ReallocPair {
72  SymbolRef ReallocatedSym;
73  bool IsFreeOnFailure;
74  ReallocPair(SymbolRef S, bool F) : ReallocatedSym(S), IsFreeOnFailure(F) {}
75  void Profile(llvm::FoldingSetNodeID &ID) const {
76    ID.AddInteger(IsFreeOnFailure);
77    ID.AddPointer(ReallocatedSym);
78  }
79  bool operator==(const ReallocPair &X) const {
80    return ReallocatedSym == X.ReallocatedSym &&
81           IsFreeOnFailure == X.IsFreeOnFailure;
82  }
83};
84
85class MallocChecker : public Checker<check::DeadSymbols,
86                                     check::EndPath,
87                                     check::PreStmt<ReturnStmt>,
88                                     check::PreStmt<CallExpr>,
89                                     check::PostStmt<CallExpr>,
90                                     check::Location,
91                                     check::Bind,
92                                     eval::Assume,
93                                     check::RegionChanges>
94{
95  mutable OwningPtr<BugType> BT_DoubleFree;
96  mutable OwningPtr<BugType> BT_Leak;
97  mutable OwningPtr<BugType> BT_UseFree;
98  mutable OwningPtr<BugType> BT_BadFree;
99  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
100                         *II_valloc, *II_reallocf;
101
102public:
103  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
104                    II_valloc(0), II_reallocf(0) {}
105
106  /// In pessimistic mode, the checker assumes that it does not know which
107  /// functions might free the memory.
108  struct ChecksFilter {
109    DefaultBool CMallocPessimistic;
110    DefaultBool CMallocOptimistic;
111  };
112
113  ChecksFilter Filter;
114
115  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
116  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
117  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
118  void checkEndPath(CheckerContext &C) const;
119  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
120  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
121                            bool Assumption) const;
122  void checkLocation(SVal l, bool isLoad, const Stmt *S,
123                     CheckerContext &C) const;
124  void checkBind(SVal location, SVal val, const Stmt*S,
125                 CheckerContext &C) const;
126  ProgramStateRef
127  checkRegionChanges(ProgramStateRef state,
128                     const StoreManager::InvalidatedSymbols *invalidated,
129                     ArrayRef<const MemRegion *> ExplicitRegions,
130                     ArrayRef<const MemRegion *> Regions,
131                     const CallOrObjCMessage *Call) const;
132  bool wantsRegionChangeUpdate(ProgramStateRef state) const {
133    return true;
134  }
135
136private:
137  void initIdentifierInfo(ASTContext &C) const;
138
139  /// Check if this is one of the functions which can allocate/reallocate memory
140  /// pointed to by one of its arguments.
141  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
142
143  static void MallocMem(CheckerContext &C, const CallExpr *CE);
144  static void MallocMemReturnsAttr(CheckerContext &C, const CallExpr *CE,
145                                   const OwnershipAttr* Att);
146  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
147                                     const Expr *SizeEx, SVal Init,
148                                     ProgramStateRef state) {
149    return MallocMemAux(C, CE,
150                        state->getSVal(SizeEx, C.getLocationContext()),
151                        Init, state);
152  }
153  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
154                                     SVal SizeEx, SVal Init,
155                                     ProgramStateRef state);
156
157  void FreeMem(CheckerContext &C, const CallExpr *CE) const;
158  void FreeMemAttr(CheckerContext &C, const CallExpr *CE,
159                   const OwnershipAttr* Att) const;
160  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
161                                 ProgramStateRef state, unsigned Num,
162                                 bool Hold) const;
163
164  void ReallocMem(CheckerContext &C, const CallExpr *CE,
165                  bool FreesMemOnFailure) const;
166  static void CallocMem(CheckerContext &C, const CallExpr *CE);
167
168  bool checkEscape(SymbolRef Sym, const Stmt *S, CheckerContext &C) const;
169  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
170                         const Stmt *S = 0) const;
171
172  /// Check if the function is not known to us. So, for example, we could
173  /// conservatively assume it can free/reallocate it's pointer arguments.
174  bool hasUnknownBehavior(const FunctionDecl *FD, ProgramStateRef State) const;
175
176  static bool SummarizeValue(raw_ostream &os, SVal V);
177  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
178  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
179
180  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
181
182  /// The bug visitor which allows us to print extra diagnostics along the
183  /// BugReport path. For example, showing the allocation site of the leaked
184  /// region.
185  class MallocBugVisitor : public BugReporterVisitor {
186  protected:
187    enum NotificationMode {
188      Normal,
189      Complete,
190      ReallocationFailed
191    };
192
193    // The allocated region symbol tracked by the main analysis.
194    SymbolRef Sym;
195    NotificationMode Mode;
196
197  public:
198    MallocBugVisitor(SymbolRef S) : Sym(S), Mode(Normal) {}
199    virtual ~MallocBugVisitor() {}
200
201    void Profile(llvm::FoldingSetNodeID &ID) const {
202      static int X = 0;
203      ID.AddPointer(&X);
204      ID.AddPointer(Sym);
205    }
206
207    inline bool isAllocated(const RefState *S, const RefState *SPrev,
208                            const Stmt *Stmt) {
209      // Did not track -> allocated. Other state (released) -> allocated.
210      return (Stmt && isa<CallExpr>(Stmt) &&
211              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
212    }
213
214    inline bool isReleased(const RefState *S, const RefState *SPrev,
215                           const Stmt *Stmt) {
216      // Did not track -> released. Other state (allocated) -> released.
217      return (Stmt && isa<CallExpr>(Stmt) &&
218              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
219    }
220
221    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
222                                     const Stmt *Stmt) {
223      // If the expression is not a call, and the state change is
224      // released -> allocated, it must be the realloc return value
225      // check. If we have to handle more cases here, it might be cleaner just
226      // to track this extra bit in the state itself.
227      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
228              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
229    }
230
231    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
232                                   const ExplodedNode *PrevN,
233                                   BugReporterContext &BRC,
234                                   BugReport &BR);
235  };
236};
237} // end anonymous namespace
238
239typedef llvm::ImmutableMap<SymbolRef, RefState> RegionStateTy;
240typedef llvm::ImmutableMap<SymbolRef, ReallocPair > ReallocMap;
241class RegionState {};
242class ReallocPairs {};
243namespace clang {
244namespace ento {
245  template <>
246  struct ProgramStateTrait<RegionState>
247    : public ProgramStatePartialTrait<RegionStateTy> {
248    static void *GDMIndex() { static int x; return &x; }
249  };
250
251  template <>
252  struct ProgramStateTrait<ReallocPairs>
253    : public ProgramStatePartialTrait<ReallocMap> {
254    static void *GDMIndex() { static int x; return &x; }
255  };
256}
257}
258
259namespace {
260class StopTrackingCallback : public SymbolVisitor {
261  ProgramStateRef state;
262public:
263  StopTrackingCallback(ProgramStateRef st) : state(st) {}
264  ProgramStateRef getState() const { return state; }
265
266  bool VisitSymbol(SymbolRef sym) {
267    state = state->remove<RegionState>(sym);
268    return true;
269  }
270};
271} // end anonymous namespace
272
273void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
274  if (!II_malloc)
275    II_malloc = &Ctx.Idents.get("malloc");
276  if (!II_free)
277    II_free = &Ctx.Idents.get("free");
278  if (!II_realloc)
279    II_realloc = &Ctx.Idents.get("realloc");
280  if (!II_reallocf)
281    II_reallocf = &Ctx.Idents.get("reallocf");
282  if (!II_calloc)
283    II_calloc = &Ctx.Idents.get("calloc");
284  if (!II_valloc)
285    II_valloc = &Ctx.Idents.get("valloc");
286}
287
288bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
289  if (!FD)
290    return false;
291  IdentifierInfo *FunI = FD->getIdentifier();
292  if (!FunI)
293    return false;
294
295  initIdentifierInfo(C);
296
297  // TODO: Add more here : ex: reallocf!
298  if (FunI == II_malloc || FunI == II_free || FunI == II_realloc ||
299      FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc)
300    return true;
301
302  if (Filter.CMallocOptimistic && FD->hasAttrs() &&
303      FD->specific_attr_begin<OwnershipAttr>() !=
304          FD->specific_attr_end<OwnershipAttr>())
305    return true;
306
307
308  return false;
309}
310
311void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
312  const FunctionDecl *FD = C.getCalleeDecl(CE);
313  if (!FD)
314    return;
315
316  initIdentifierInfo(C.getASTContext());
317  IdentifierInfo *FunI = FD->getIdentifier();
318  if (!FunI)
319    return;
320
321  if (FunI == II_malloc || FunI == II_valloc) {
322    MallocMem(C, CE);
323    return;
324  } else if (FunI == II_realloc) {
325    ReallocMem(C, CE, false);
326    return;
327  } else if (FunI == II_reallocf) {
328    ReallocMem(C, CE, true);
329    return;
330  } else if (FunI == II_calloc) {
331    CallocMem(C, CE);
332    return;
333  }else if (FunI == II_free) {
334    FreeMem(C, CE);
335    return;
336  }
337
338  if (Filter.CMallocOptimistic)
339  // Check all the attributes, if there are any.
340  // There can be multiple of these attributes.
341  if (FD->hasAttrs()) {
342    for (specific_attr_iterator<OwnershipAttr>
343                  i = FD->specific_attr_begin<OwnershipAttr>(),
344                  e = FD->specific_attr_end<OwnershipAttr>();
345         i != e; ++i) {
346      switch ((*i)->getOwnKind()) {
347      case OwnershipAttr::Returns: {
348        MallocMemReturnsAttr(C, CE, *i);
349        return;
350      }
351      case OwnershipAttr::Takes:
352      case OwnershipAttr::Holds: {
353        FreeMemAttr(C, CE, *i);
354        return;
355      }
356      }
357    }
358  }
359}
360
361void MallocChecker::MallocMem(CheckerContext &C, const CallExpr *CE) {
362  ProgramStateRef state = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(),
363                                      C.getState());
364  C.addTransition(state);
365}
366
367void MallocChecker::MallocMemReturnsAttr(CheckerContext &C, const CallExpr *CE,
368                                         const OwnershipAttr* Att) {
369  if (Att->getModule() != "malloc")
370    return;
371
372  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
373  if (I != E) {
374    ProgramStateRef state =
375        MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
376    C.addTransition(state);
377    return;
378  }
379  ProgramStateRef state = MallocMemAux(C, CE, UnknownVal(), UndefinedVal(),
380                                        C.getState());
381  C.addTransition(state);
382}
383
384ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
385                                           const CallExpr *CE,
386                                           SVal Size, SVal Init,
387                                           ProgramStateRef state) {
388  SValBuilder &svalBuilder = C.getSValBuilder();
389
390  // Get the return value.
391  SVal retVal = state->getSVal(CE, C.getLocationContext());
392
393  // We expect the malloc functions to return a pointer.
394  if (!isa<Loc>(retVal))
395    return 0;
396
397  // Fill the region with the initialization value.
398  state = state->bindDefault(retVal, Init);
399
400  // Set the region's extent equal to the Size parameter.
401  const SymbolicRegion *R =
402      dyn_cast_or_null<SymbolicRegion>(retVal.getAsRegion());
403  if (!R || !isa<DefinedOrUnknownSVal>(Size))
404    return 0;
405
406  DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
407  DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
408  DefinedOrUnknownSVal extentMatchesSize =
409    svalBuilder.evalEQ(state, Extent, DefinedSize);
410
411  state = state->assume(extentMatchesSize, true);
412  assert(state);
413
414  SymbolRef Sym = retVal.getAsLocSymbol();
415  assert(Sym);
416
417  // Set the symbol's state to Allocated.
418  return state->set<RegionState>(Sym, RefState::getAllocateUnchecked(CE));
419}
420
421void MallocChecker::FreeMem(CheckerContext &C, const CallExpr *CE) const {
422  ProgramStateRef state = FreeMemAux(C, CE, C.getState(), 0, false);
423
424  if (state)
425    C.addTransition(state);
426}
427
428void MallocChecker::FreeMemAttr(CheckerContext &C, const CallExpr *CE,
429                                const OwnershipAttr* Att) const {
430  if (Att->getModule() != "malloc")
431    return;
432
433  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
434       I != E; ++I) {
435    ProgramStateRef state =
436      FreeMemAux(C, CE, C.getState(), *I,
437                 Att->getOwnKind() == OwnershipAttr::Holds);
438    if (state)
439      C.addTransition(state);
440  }
441}
442
443ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
444                                          const CallExpr *CE,
445                                          ProgramStateRef state,
446                                          unsigned Num,
447                                          bool Hold) const {
448  const Expr *ArgExpr = CE->getArg(Num);
449  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
450  if (!isa<DefinedOrUnknownSVal>(ArgVal))
451    return 0;
452  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
453
454  // Check for null dereferences.
455  if (!isa<Loc>(location))
456    return 0;
457
458  // The explicit NULL case, no operation is performed.
459  ProgramStateRef notNullState, nullState;
460  llvm::tie(notNullState, nullState) = state->assume(location);
461  if (nullState && !notNullState)
462    return 0;
463
464  // Unknown values could easily be okay
465  // Undefined values are handled elsewhere
466  if (ArgVal.isUnknownOrUndef())
467    return 0;
468
469  const MemRegion *R = ArgVal.getAsRegion();
470
471  // Nonlocs can't be freed, of course.
472  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
473  if (!R) {
474    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
475    return 0;
476  }
477
478  R = R->StripCasts();
479
480  // Blocks might show up as heap data, but should not be free()d
481  if (isa<BlockDataRegion>(R)) {
482    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
483    return 0;
484  }
485
486  const MemSpaceRegion *MS = R->getMemorySpace();
487
488  // Parameters, locals, statics, and globals shouldn't be freed.
489  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
490    // FIXME: at the time this code was written, malloc() regions were
491    // represented by conjured symbols, which are all in UnknownSpaceRegion.
492    // This means that there isn't actually anything from HeapSpaceRegion
493    // that should be freed, even though we allow it here.
494    // Of course, free() can work on memory allocated outside the current
495    // function, so UnknownSpaceRegion is always a possibility.
496    // False negatives are better than false positives.
497
498    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
499    return 0;
500  }
501
502  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
503  // Various cases could lead to non-symbol values here.
504  // For now, ignore them.
505  if (!SR)
506    return 0;
507
508  SymbolRef Sym = SR->getSymbol();
509  const RefState *RS = state->get<RegionState>(Sym);
510
511  // If the symbol has not been tracked, return. This is possible when free() is
512  // called on a pointer that does not get its pointee directly from malloc().
513  // Full support of this requires inter-procedural analysis.
514  if (!RS)
515    return 0;
516
517  // Check double free.
518  if (RS->isReleased()) {
519    if (ExplodedNode *N = C.generateSink()) {
520      if (!BT_DoubleFree)
521        BT_DoubleFree.reset(
522          new BugType("Double free", "Memory Error"));
523      BugReport *R = new BugReport(*BT_DoubleFree,
524                        "Attempt to free released memory", N);
525      R->addRange(ArgExpr->getSourceRange());
526      R->addVisitor(new MallocBugVisitor(Sym));
527      C.EmitReport(R);
528    }
529    return 0;
530  }
531
532  // Normal free.
533  if (Hold)
534    return state->set<RegionState>(Sym, RefState::getRelinquished(CE));
535  return state->set<RegionState>(Sym, RefState::getReleased(CE));
536}
537
538bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
539  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
540    os << "an integer (" << IntVal->getValue() << ")";
541  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
542    os << "a constant address (" << ConstAddr->getValue() << ")";
543  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
544    os << "the address of the label '" << Label->getLabel()->getName() << "'";
545  else
546    return false;
547
548  return true;
549}
550
551bool MallocChecker::SummarizeRegion(raw_ostream &os,
552                                    const MemRegion *MR) {
553  switch (MR->getKind()) {
554  case MemRegion::FunctionTextRegionKind: {
555    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
556    if (FD)
557      os << "the address of the function '" << *FD << '\'';
558    else
559      os << "the address of a function";
560    return true;
561  }
562  case MemRegion::BlockTextRegionKind:
563    os << "block text";
564    return true;
565  case MemRegion::BlockDataRegionKind:
566    // FIXME: where the block came from?
567    os << "a block";
568    return true;
569  default: {
570    const MemSpaceRegion *MS = MR->getMemorySpace();
571
572    if (isa<StackLocalsSpaceRegion>(MS)) {
573      const VarRegion *VR = dyn_cast<VarRegion>(MR);
574      const VarDecl *VD;
575      if (VR)
576        VD = VR->getDecl();
577      else
578        VD = NULL;
579
580      if (VD)
581        os << "the address of the local variable '" << VD->getName() << "'";
582      else
583        os << "the address of a local stack variable";
584      return true;
585    }
586
587    if (isa<StackArgumentsSpaceRegion>(MS)) {
588      const VarRegion *VR = dyn_cast<VarRegion>(MR);
589      const VarDecl *VD;
590      if (VR)
591        VD = VR->getDecl();
592      else
593        VD = NULL;
594
595      if (VD)
596        os << "the address of the parameter '" << VD->getName() << "'";
597      else
598        os << "the address of a parameter";
599      return true;
600    }
601
602    if (isa<GlobalsSpaceRegion>(MS)) {
603      const VarRegion *VR = dyn_cast<VarRegion>(MR);
604      const VarDecl *VD;
605      if (VR)
606        VD = VR->getDecl();
607      else
608        VD = NULL;
609
610      if (VD) {
611        if (VD->isStaticLocal())
612          os << "the address of the static variable '" << VD->getName() << "'";
613        else
614          os << "the address of the global variable '" << VD->getName() << "'";
615      } else
616        os << "the address of a global variable";
617      return true;
618    }
619
620    return false;
621  }
622  }
623}
624
625void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
626                                  SourceRange range) const {
627  if (ExplodedNode *N = C.generateSink()) {
628    if (!BT_BadFree)
629      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
630
631    SmallString<100> buf;
632    llvm::raw_svector_ostream os(buf);
633
634    const MemRegion *MR = ArgVal.getAsRegion();
635    if (MR) {
636      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
637        MR = ER->getSuperRegion();
638
639      // Special case for alloca()
640      if (isa<AllocaRegion>(MR))
641        os << "Argument to free() was allocated by alloca(), not malloc()";
642      else {
643        os << "Argument to free() is ";
644        if (SummarizeRegion(os, MR))
645          os << ", which is not memory allocated by malloc()";
646        else
647          os << "not memory allocated by malloc()";
648      }
649    } else {
650      os << "Argument to free() is ";
651      if (SummarizeValue(os, ArgVal))
652        os << ", which is not memory allocated by malloc()";
653      else
654        os << "not memory allocated by malloc()";
655    }
656
657    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
658    R->addRange(range);
659    C.EmitReport(R);
660  }
661}
662
663void MallocChecker::ReallocMem(CheckerContext &C, const CallExpr *CE,
664                               bool FreesOnFail) const {
665  ProgramStateRef state = C.getState();
666  const Expr *arg0Expr = CE->getArg(0);
667  const LocationContext *LCtx = C.getLocationContext();
668  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
669  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
670    return;
671  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
672
673  SValBuilder &svalBuilder = C.getSValBuilder();
674
675  DefinedOrUnknownSVal PtrEQ =
676    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
677
678  // Get the size argument. If there is no size arg then give up.
679  const Expr *Arg1 = CE->getArg(1);
680  if (!Arg1)
681    return;
682
683  // Get the value of the size argument.
684  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
685  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
686    return;
687  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
688
689  // Compare the size argument to 0.
690  DefinedOrUnknownSVal SizeZero =
691    svalBuilder.evalEQ(state, Arg1Val,
692                       svalBuilder.makeIntValWithPtrWidth(0, false));
693
694  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
695  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
696  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
697  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
698  // We only assume exceptional states if they are definitely true; if the
699  // state is under-constrained, assume regular realloc behavior.
700  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
701  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
702
703  // If the ptr is NULL and the size is not 0, the call is equivalent to
704  // malloc(size).
705  if ( PrtIsNull && !SizeIsZero) {
706    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
707                                               UndefinedVal(), StatePtrIsNull);
708    C.addTransition(stateMalloc);
709    return;
710  }
711
712  if (PrtIsNull && SizeIsZero)
713    return;
714
715  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
716  assert(!PrtIsNull);
717  SymbolRef FromPtr = arg0Val.getAsSymbol();
718  SVal RetVal = state->getSVal(CE, LCtx);
719  SymbolRef ToPtr = RetVal.getAsSymbol();
720  if (!FromPtr || !ToPtr)
721    return;
722
723  // If the size is 0, free the memory.
724  if (SizeIsZero)
725    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
726      // The semantics of the return value are:
727      // If size was equal to 0, either NULL or a pointer suitable to be passed
728      // to free() is returned.
729      stateFree = stateFree->set<ReallocPairs>(ToPtr,
730                                            ReallocPair(FromPtr, FreesOnFail));
731      C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
732      C.addTransition(stateFree);
733      return;
734    }
735
736  // Default behavior.
737  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
738    // FIXME: We should copy the content of the original buffer.
739    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
740                                                UnknownVal(), stateFree);
741    if (!stateRealloc)
742      return;
743    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
744                                            ReallocPair(FromPtr, FreesOnFail));
745    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
746    C.addTransition(stateRealloc);
747    return;
748  }
749}
750
751void MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE) {
752  ProgramStateRef state = C.getState();
753  SValBuilder &svalBuilder = C.getSValBuilder();
754  const LocationContext *LCtx = C.getLocationContext();
755  SVal count = state->getSVal(CE->getArg(0), LCtx);
756  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
757  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
758                                        svalBuilder.getContext().getSizeType());
759  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
760
761  C.addTransition(MallocMemAux(C, CE, TotalSize, zeroVal, state));
762}
763
764void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
765                               CheckerContext &C) const {
766  assert(N);
767  if (!BT_Leak) {
768    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
769    // Leaks should not be reported if they are post-dominated by a sink:
770    // (1) Sinks are higher importance bugs.
771    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
772    //     with __noreturn functions such as assert() or exit(). We choose not
773    //     to report leaks on such paths.
774    BT_Leak->setSuppressOnSink(true);
775  }
776
777  BugReport *R = new BugReport(*BT_Leak,
778                  "Memory is never released; potential memory leak", N);
779  R->addVisitor(new MallocBugVisitor(Sym));
780  C.EmitReport(R);
781}
782
783void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
784                                     CheckerContext &C) const
785{
786  if (!SymReaper.hasDeadSymbols())
787    return;
788
789  ProgramStateRef state = C.getState();
790  RegionStateTy RS = state->get<RegionState>();
791  RegionStateTy::Factory &F = state->get_context<RegionState>();
792
793  bool generateReport = false;
794  llvm::SmallVector<SymbolRef, 2> Errors;
795  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
796    if (SymReaper.isDead(I->first)) {
797      if (I->second.isAllocated()) {
798        generateReport = true;
799        Errors.push_back(I->first);
800      }
801      // Remove the dead symbol from the map.
802      RS = F.remove(RS, I->first);
803
804    }
805  }
806
807  // Cleanup the Realloc Pairs Map.
808  ReallocMap RP = state->get<ReallocPairs>();
809  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
810    if (SymReaper.isDead(I->first) ||
811        SymReaper.isDead(I->second.ReallocatedSym)) {
812      state = state->remove<ReallocPairs>(I->first);
813    }
814  }
815
816  ExplodedNode *N = C.addTransition(state->set<RegionState>(RS));
817
818  if (N && generateReport) {
819    for (llvm::SmallVector<SymbolRef, 2>::iterator
820         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
821      reportLeak(*I, N, C);
822    }
823  }
824}
825
826void MallocChecker::checkEndPath(CheckerContext &C) const {
827  ProgramStateRef state = C.getState();
828  RegionStateTy M = state->get<RegionState>();
829
830  // If inside inlined call, skip it.
831  if (C.getLocationContext()->getParent() != 0)
832    return;
833
834  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
835    RefState RS = I->second;
836    if (RS.isAllocated()) {
837      ExplodedNode *N = C.addTransition(state);
838      if (N)
839        reportLeak(I->first, N, C);
840    }
841  }
842}
843
844bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
845                                CheckerContext &C) const {
846  ProgramStateRef state = C.getState();
847  const RefState *RS = state->get<RegionState>(Sym);
848  if (!RS)
849    return false;
850
851  if (RS->isAllocated()) {
852    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
853    C.addTransition(state);
854    return true;
855  }
856  return false;
857}
858
859void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
860  if (isMemFunction(C.getCalleeDecl(CE), C.getASTContext()))
861    return;
862
863  // Check use after free, when a freed pointer is passed to a call.
864  ProgramStateRef State = C.getState();
865  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
866                                    E = CE->arg_end(); I != E; ++I) {
867    const Expr *A = *I;
868    if (A->getType().getTypePtr()->isAnyPointerType()) {
869      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
870      if (!Sym)
871        continue;
872      if (checkUseAfterFree(Sym, C, A))
873        return;
874    }
875  }
876}
877
878void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
879  const Expr *E = S->getRetValue();
880  if (!E)
881    return;
882
883  // Check if we are returning a symbol.
884  SVal RetVal = C.getState()->getSVal(E, C.getLocationContext());
885  SymbolRef Sym = RetVal.getAsSymbol();
886  if (!Sym)
887    // If we are returning a field of the allocated struct or an array element,
888    // the callee could still free the memory.
889    // TODO: This logic should be a part of generic symbol escape callback.
890    if (const MemRegion *MR = RetVal.getAsRegion())
891      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
892        if (const SymbolicRegion *BMR =
893              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
894          Sym = BMR->getSymbol();
895  if (!Sym)
896    return;
897
898  // Check if we are returning freed memory.
899  if (checkUseAfterFree(Sym, C, E))
900    return;
901
902  // If this function body is not inlined, check if the symbol is escaping.
903  if (C.getLocationContext()->getParent() == 0)
904    checkEscape(Sym, E, C);
905}
906
907bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
908                                      const Stmt *S) const {
909  assert(Sym);
910  const RefState *RS = C.getState()->get<RegionState>(Sym);
911  if (RS && RS->isReleased()) {
912    if (ExplodedNode *N = C.generateSink()) {
913      if (!BT_UseFree)
914        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
915
916      BugReport *R = new BugReport(*BT_UseFree,
917                                   "Use of memory after it is freed",N);
918      if (S)
919        R->addRange(S->getSourceRange());
920      R->addVisitor(new MallocBugVisitor(Sym));
921      C.EmitReport(R);
922      return true;
923    }
924  }
925  return false;
926}
927
928// Check if the location is a freed symbolic region.
929void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
930                                  CheckerContext &C) const {
931  SymbolRef Sym = l.getLocSymbolInBase();
932  if (Sym)
933    checkUseAfterFree(Sym, C);
934}
935
936//===----------------------------------------------------------------------===//
937// Check various ways a symbol can be invalidated.
938// TODO: This logic (the next 3 functions) is copied/similar to the
939// RetainRelease checker. We might want to factor this out.
940//===----------------------------------------------------------------------===//
941
942// Stop tracking symbols when a value escapes as a result of checkBind.
943// A value escapes in three possible cases:
944// (1) we are binding to something that is not a memory region.
945// (2) we are binding to a memregion that does not have stack storage
946// (3) we are binding to a memregion with stack storage that the store
947//     does not understand.
948void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
949                              CheckerContext &C) const {
950  // Are we storing to something that causes the value to "escape"?
951  bool escapes = true;
952  ProgramStateRef state = C.getState();
953
954  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
955    escapes = !regionLoc->getRegion()->hasStackStorage();
956
957    if (!escapes) {
958      // To test (3), generate a new state with the binding added.  If it is
959      // the same state, then it escapes (since the store cannot represent
960      // the binding).
961      escapes = (state == (state->bindLoc(*regionLoc, val)));
962    }
963    if (!escapes) {
964      // Case 4: We do not currently model what happens when a symbol is
965      // assigned to a struct field, so be conservative here and let the symbol
966      // go. TODO: This could definitely be improved upon.
967      escapes = !isa<VarRegion>(regionLoc->getRegion());
968    }
969  }
970
971  // If our store can represent the binding and we aren't storing to something
972  // that doesn't have local storage then just return and have the simulation
973  // state continue as is.
974  if (!escapes)
975      return;
976
977  // Otherwise, find all symbols referenced by 'val' that we are tracking
978  // and stop tracking them.
979  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
980  C.addTransition(state);
981}
982
983// If a symbolic region is assumed to NULL (or another constant), stop tracking
984// it - assuming that allocation failed on this path.
985ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
986                                              SVal Cond,
987                                              bool Assumption) const {
988  RegionStateTy RS = state->get<RegionState>();
989  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
990    // If the symbol is assumed to NULL or another constant, this will
991    // return an APSInt*.
992    if (state->getSymVal(I.getKey()))
993      state = state->remove<RegionState>(I.getKey());
994  }
995
996  // Realloc returns 0 when reallocation fails, which means that we should
997  // restore the state of the pointer being reallocated.
998  ReallocMap RP = state->get<ReallocPairs>();
999  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1000    // If the symbol is assumed to NULL or another constant, this will
1001    // return an APSInt*.
1002    if (state->getSymVal(I.getKey())) {
1003      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1004      const RefState *RS = state->get<RegionState>(ReallocSym);
1005      if (RS) {
1006        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1007          state = state->set<RegionState>(ReallocSym,
1008                             RefState::getAllocateUnchecked(RS->getStmt()));
1009      }
1010      state = state->remove<ReallocPairs>(I.getKey());
1011    }
1012  }
1013
1014  return state;
1015}
1016
1017// Check if the function is not known to us. So, for example, we could
1018// conservatively assume it can free/reallocate it's pointer arguments.
1019// (We assume that the pointers cannot escape through calls to system
1020// functions not handled by this checker.)
1021bool MallocChecker::hasUnknownBehavior(const FunctionDecl *FD,
1022                                       ProgramStateRef State) const {
1023  ASTContext &ASTC = State->getStateManager().getContext();
1024
1025  // If it's one of the allocation functions we can reason about, we model it's
1026  // behavior explicitly.
1027  if (isMemFunction(FD, ASTC)) {
1028    return false;
1029  }
1030
1031  // If it's a system call, we know it does not free the memory.
1032  SourceManager &SM = ASTC.getSourceManager();
1033  if (SM.isInSystemHeader(FD->getLocation())) {
1034    return false;
1035  }
1036
1037  // Otherwise, assume that the function can free memory.
1038  return true;
1039}
1040
1041// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1042// escapes, when we are tracking p), do not track the symbol as we cannot reason
1043// about it anymore.
1044ProgramStateRef
1045MallocChecker::checkRegionChanges(ProgramStateRef State,
1046                            const StoreManager::InvalidatedSymbols *invalidated,
1047                                    ArrayRef<const MemRegion *> ExplicitRegions,
1048                                    ArrayRef<const MemRegion *> Regions,
1049                                    const CallOrObjCMessage *Call) const {
1050  if (!invalidated)
1051    return State;
1052  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1053
1054  const FunctionDecl *FD = (Call ?
1055                            dyn_cast_or_null<FunctionDecl>(Call->getDecl()) :0);
1056
1057  // If it's a call which might free or reallocate memory, we assume that all
1058  // regions (explicit and implicit) escaped. Otherwise, whitelist explicit
1059  // pointers; we still can track them.
1060  if (!(FD && hasUnknownBehavior(FD, State))) {
1061    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1062        E = ExplicitRegions.end(); I != E; ++I) {
1063      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1064        WhitelistedSymbols.insert(R->getSymbol());
1065    }
1066  }
1067
1068  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1069       E = invalidated->end(); I!=E; ++I) {
1070    SymbolRef sym = *I;
1071    if (WhitelistedSymbols.count(sym))
1072      continue;
1073    // The symbol escaped.
1074    if (const RefState *RS = State->get<RegionState>(sym))
1075      State = State->set<RegionState>(sym, RefState::getEscaped(RS->getStmt()));
1076  }
1077  return State;
1078}
1079
1080PathDiagnosticPiece *
1081MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1082                                           const ExplodedNode *PrevN,
1083                                           BugReporterContext &BRC,
1084                                           BugReport &BR) {
1085  const RefState *RS = N->getState()->get<RegionState>(Sym);
1086  const RefState *RSPrev = PrevN->getState()->get<RegionState>(Sym);
1087  if (!RS && !RSPrev)
1088    return 0;
1089
1090  const Stmt *S = 0;
1091  const char *Msg = 0;
1092
1093  // Retrieve the associated statement.
1094  ProgramPoint ProgLoc = N->getLocation();
1095  if (isa<StmtPoint>(ProgLoc))
1096    S = cast<StmtPoint>(ProgLoc).getStmt();
1097  // If an assumption was made on a branch, it should be caught
1098  // here by looking at the state transition.
1099  if (isa<BlockEdge>(ProgLoc)) {
1100    const CFGBlock *srcBlk = cast<BlockEdge>(ProgLoc).getSrc();
1101    S = srcBlk->getTerminator();
1102  }
1103  if (!S)
1104    return 0;
1105
1106  // Find out if this is an interesting point and what is the kind.
1107  if (Mode == Normal) {
1108    if (isAllocated(RS, RSPrev, S))
1109      Msg = "Memory is allocated";
1110    else if (isReleased(RS, RSPrev, S))
1111      Msg = "Memory is released";
1112    else if (isReallocFailedCheck(RS, RSPrev, S)) {
1113      Mode = ReallocationFailed;
1114      Msg = "Reallocation failed";
1115    }
1116
1117  // We are in a special mode if a reallocation failed later in the path.
1118  } else if (Mode == ReallocationFailed) {
1119    // Generate a special diagnostic for the first realloc we find.
1120    if (!isAllocated(RS, RSPrev, S) && !isReleased(RS, RSPrev, S))
1121      return 0;
1122
1123    // Check that the name of the function is realloc.
1124    const CallExpr *CE = dyn_cast<CallExpr>(S);
1125    if (!CE)
1126      return 0;
1127    const FunctionDecl *funDecl = CE->getDirectCallee();
1128    if (!funDecl)
1129      return 0;
1130    StringRef FunName = funDecl->getName();
1131    if (!(FunName.equals("realloc") || FunName.equals("reallocf")))
1132      return 0;
1133    Msg = "Attempt to reallocate memory";
1134    Mode = Normal;
1135  }
1136
1137  if (!Msg)
1138    return 0;
1139
1140  // Generate the extra diagnostic.
1141  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1142                             N->getLocationContext());
1143  return new PathDiagnosticEventPiece(Pos, Msg);
1144}
1145
1146
1147#define REGISTER_CHECKER(name) \
1148void ento::register##name(CheckerManager &mgr) {\
1149  registerCStringCheckerBasic(mgr); \
1150  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1151}
1152
1153REGISTER_CHECKER(MallocPessimistic)
1154REGISTER_CHECKER(MallocOptimistic)
1155