MallocChecker.cpp revision fe571608b925079227d053a459eca86f7408e5c6 
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 "clang/StaticAnalyzer/Core/Checker.h"
17#include "clang/StaticAnalyzer/Core/CheckerManager.h"
18#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
19#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
20#include "clang/StaticAnalyzer/Core/PathSensitive/ObjCMessage.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
24#include "clang/Basic/SourceManager.h"
25#include "llvm/ADT/ImmutableMap.h"
26#include "llvm/ADT/SmallString.h"
27#include "llvm/ADT/STLExtras.h"
28using namespace clang;
29using namespace ento;
30
31namespace {
32
33class RefState {
34  enum Kind { AllocateUnchecked, AllocateFailed, Released, Escaped,
35              Relinquished } K;
36  const Stmt *S;
37
38public:
39  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
40
41  bool isAllocated() const { return K == AllocateUnchecked; }
42  //bool isFailed() const { return K == AllocateFailed; }
43  bool isReleased() const { return K == Released; }
44  //bool isEscaped() const { return K == Escaped; }
45  //bool isRelinquished() const { return K == Relinquished; }
46  const Stmt *getStmt() const { return S; }
47
48  bool operator==(const RefState &X) const {
49    return K == X.K && S == X.S;
50  }
51
52  static RefState getAllocateUnchecked(const Stmt *s) {
53    return RefState(AllocateUnchecked, s);
54  }
55  static RefState getAllocateFailed() {
56    return RefState(AllocateFailed, 0);
57  }
58  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
59  static RefState getEscaped(const Stmt *s) { return RefState(Escaped, s); }
60  static RefState getRelinquished(const Stmt *s) {
61    return RefState(Relinquished, s);
62  }
63
64  void Profile(llvm::FoldingSetNodeID &ID) const {
65    ID.AddInteger(K);
66    ID.AddPointer(S);
67  }
68};
69
70struct ReallocPair {
71  SymbolRef ReallocatedSym;
72  bool IsFreeOnFailure;
73  ReallocPair(SymbolRef S, bool F) : ReallocatedSym(S), IsFreeOnFailure(F) {}
74  void Profile(llvm::FoldingSetNodeID &ID) const {
75    ID.AddInteger(IsFreeOnFailure);
76    ID.AddPointer(ReallocatedSym);
77  }
78  bool operator==(const ReallocPair &X) const {
79    return ReallocatedSym == X.ReallocatedSym &&
80           IsFreeOnFailure == X.IsFreeOnFailure;
81  }
82};
83
84class MallocChecker : public Checker<check::DeadSymbols,
85                                     check::EndPath,
86                                     check::PreStmt<ReturnStmt>,
87                                     check::PreStmt<CallExpr>,
88                                     check::PostStmt<CallExpr>,
89                                     check::Location,
90                                     check::Bind,
91                                     eval::Assume,
92                                     check::RegionChanges>
93{
94  mutable OwningPtr<BuiltinBug> BT_DoubleFree;
95  mutable OwningPtr<BuiltinBug> BT_Leak;
96  mutable OwningPtr<BuiltinBug> BT_UseFree;
97  mutable OwningPtr<BuiltinBug> BT_UseRelinquished;
98  mutable OwningPtr<BuiltinBug> 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 BuiltinBug("Double free",
523                         "Try to free a memory block that has been released"));
524      BugReport *R = new BugReport(*BT_DoubleFree,
525                                   BT_DoubleFree->getDescription(), N);
526      R->addRange(ArgExpr->getSourceRange());
527      R->addVisitor(new MallocBugVisitor(Sym));
528      C.EmitReport(R);
529    }
530    return 0;
531  }
532
533  // Normal free.
534  if (Hold)
535    return state->set<RegionState>(Sym, RefState::getRelinquished(CE));
536  return state->set<RegionState>(Sym, RefState::getReleased(CE));
537}
538
539bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
540  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
541    os << "an integer (" << IntVal->getValue() << ")";
542  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
543    os << "a constant address (" << ConstAddr->getValue() << ")";
544  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
545    os << "the address of the label '" << Label->getLabel()->getName() << "'";
546  else
547    return false;
548
549  return true;
550}
551
552bool MallocChecker::SummarizeRegion(raw_ostream &os,
553                                    const MemRegion *MR) {
554  switch (MR->getKind()) {
555  case MemRegion::FunctionTextRegionKind: {
556    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
557    if (FD)
558      os << "the address of the function '" << *FD << '\'';
559    else
560      os << "the address of a function";
561    return true;
562  }
563  case MemRegion::BlockTextRegionKind:
564    os << "block text";
565    return true;
566  case MemRegion::BlockDataRegionKind:
567    // FIXME: where the block came from?
568    os << "a block";
569    return true;
570  default: {
571    const MemSpaceRegion *MS = MR->getMemorySpace();
572
573    if (isa<StackLocalsSpaceRegion>(MS)) {
574      const VarRegion *VR = dyn_cast<VarRegion>(MR);
575      const VarDecl *VD;
576      if (VR)
577        VD = VR->getDecl();
578      else
579        VD = NULL;
580
581      if (VD)
582        os << "the address of the local variable '" << VD->getName() << "'";
583      else
584        os << "the address of a local stack variable";
585      return true;
586    }
587
588    if (isa<StackArgumentsSpaceRegion>(MS)) {
589      const VarRegion *VR = dyn_cast<VarRegion>(MR);
590      const VarDecl *VD;
591      if (VR)
592        VD = VR->getDecl();
593      else
594        VD = NULL;
595
596      if (VD)
597        os << "the address of the parameter '" << VD->getName() << "'";
598      else
599        os << "the address of a parameter";
600      return true;
601    }
602
603    if (isa<GlobalsSpaceRegion>(MS)) {
604      const VarRegion *VR = dyn_cast<VarRegion>(MR);
605      const VarDecl *VD;
606      if (VR)
607        VD = VR->getDecl();
608      else
609        VD = NULL;
610
611      if (VD) {
612        if (VD->isStaticLocal())
613          os << "the address of the static variable '" << VD->getName() << "'";
614        else
615          os << "the address of the global variable '" << VD->getName() << "'";
616      } else
617        os << "the address of a global variable";
618      return true;
619    }
620
621    return false;
622  }
623  }
624}
625
626void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
627                                  SourceRange range) const {
628  if (ExplodedNode *N = C.generateSink()) {
629    if (!BT_BadFree)
630      BT_BadFree.reset(new BuiltinBug("Bad free"));
631
632    SmallString<100> buf;
633    llvm::raw_svector_ostream os(buf);
634
635    const MemRegion *MR = ArgVal.getAsRegion();
636    if (MR) {
637      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
638        MR = ER->getSuperRegion();
639
640      // Special case for alloca()
641      if (isa<AllocaRegion>(MR))
642        os << "Argument to free() was allocated by alloca(), not malloc()";
643      else {
644        os << "Argument to free() is ";
645        if (SummarizeRegion(os, MR))
646          os << ", which is not memory allocated by malloc()";
647        else
648          os << "not memory allocated by malloc()";
649      }
650    } else {
651      os << "Argument to free() is ";
652      if (SummarizeValue(os, ArgVal))
653        os << ", which is not memory allocated by malloc()";
654      else
655        os << "not memory allocated by malloc()";
656    }
657
658    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
659    R->addRange(range);
660    C.EmitReport(R);
661  }
662}
663
664void MallocChecker::ReallocMem(CheckerContext &C, const CallExpr *CE,
665                               bool FreesOnFail) const {
666  ProgramStateRef state = C.getState();
667  const Expr *arg0Expr = CE->getArg(0);
668  const LocationContext *LCtx = C.getLocationContext();
669  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
670  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
671    return;
672  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
673
674  SValBuilder &svalBuilder = C.getSValBuilder();
675
676  DefinedOrUnknownSVal PtrEQ =
677    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
678
679  // Get the size argument. If there is no size arg then give up.
680  const Expr *Arg1 = CE->getArg(1);
681  if (!Arg1)
682    return;
683
684  // Get the value of the size argument.
685  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
686  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
687    return;
688  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
689
690  // Compare the size argument to 0.
691  DefinedOrUnknownSVal SizeZero =
692    svalBuilder.evalEQ(state, Arg1Val,
693                       svalBuilder.makeIntValWithPtrWidth(0, false));
694
695  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
696  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
697  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
698  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
699  // We only assume exceptional states if they are definitely true; if the
700  // state is under-constrained, assume regular realloc behavior.
701  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
702  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
703
704  // If the ptr is NULL and the size is not 0, the call is equivalent to
705  // malloc(size).
706  if ( PrtIsNull && !SizeIsZero) {
707    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
708                                               UndefinedVal(), StatePtrIsNull);
709    C.addTransition(stateMalloc);
710    return;
711  }
712
713  if (PrtIsNull && SizeIsZero)
714    return;
715
716  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
717  assert(!PrtIsNull);
718  SymbolRef FromPtr = arg0Val.getAsSymbol();
719  SVal RetVal = state->getSVal(CE, LCtx);
720  SymbolRef ToPtr = RetVal.getAsSymbol();
721  if (!FromPtr || !ToPtr)
722    return;
723
724  // If the size is 0, free the memory.
725  if (SizeIsZero)
726    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
727      // The semantics of the return value are:
728      // If size was equal to 0, either NULL or a pointer suitable to be passed
729      // to free() is returned.
730      stateFree = stateFree->set<ReallocPairs>(ToPtr,
731                                            ReallocPair(FromPtr, FreesOnFail));
732      C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
733      C.addTransition(stateFree);
734      return;
735    }
736
737  // Default behavior.
738  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
739    // FIXME: We should copy the content of the original buffer.
740    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
741                                                UnknownVal(), stateFree);
742    if (!stateRealloc)
743      return;
744    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
745                                            ReallocPair(FromPtr, FreesOnFail));
746    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
747    C.addTransition(stateRealloc);
748    return;
749  }
750}
751
752void MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE) {
753  ProgramStateRef state = C.getState();
754  SValBuilder &svalBuilder = C.getSValBuilder();
755  const LocationContext *LCtx = C.getLocationContext();
756  SVal count = state->getSVal(CE->getArg(0), LCtx);
757  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
758  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
759                                        svalBuilder.getContext().getSizeType());
760  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
761
762  C.addTransition(MallocMemAux(C, CE, TotalSize, zeroVal, state));
763}
764
765void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
766                               CheckerContext &C) const {
767  assert(N);
768  if (!BT_Leak) {
769    BT_Leak.reset(new BuiltinBug("Memory leak",
770        "Allocated memory never released. Potential memory leak."));
771    // Leaks should not be reported if they are post-dominated by a sink:
772    // (1) Sinks are higher importance bugs.
773    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
774    //     with __noreturn functions such as assert() or exit(). We choose not
775    //     to report leaks on such paths.
776    BT_Leak->setSuppressOnSink(true);
777  }
778
779  BugReport *R = new BugReport(*BT_Leak, BT_Leak->getDescription(), N);
780  R->addVisitor(new MallocBugVisitor(Sym));
781  C.EmitReport(R);
782}
783
784void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
785                                     CheckerContext &C) const
786{
787  if (!SymReaper.hasDeadSymbols())
788    return;
789
790  ProgramStateRef state = C.getState();
791  RegionStateTy RS = state->get<RegionState>();
792  RegionStateTy::Factory &F = state->get_context<RegionState>();
793
794  bool generateReport = false;
795  llvm::SmallVector<SymbolRef, 2> Errors;
796  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
797    if (SymReaper.isDead(I->first)) {
798      if (I->second.isAllocated()) {
799        generateReport = true;
800        Errors.push_back(I->first);
801      }
802      // Remove the dead symbol from the map.
803      RS = F.remove(RS, I->first);
804
805    }
806  }
807
808  // Cleanup the Realloc Pairs Map.
809  ReallocMap RP = state->get<ReallocPairs>();
810  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
811    if (SymReaper.isDead(I->first) ||
812        SymReaper.isDead(I->second.ReallocatedSym)) {
813      state = state->remove<ReallocPairs>(I->first);
814    }
815  }
816
817  ExplodedNode *N = C.addTransition(state->set<RegionState>(RS));
818
819  if (N && generateReport) {
820    for (llvm::SmallVector<SymbolRef, 2>::iterator
821         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
822      reportLeak(*I, N, C);
823    }
824  }
825}
826
827void MallocChecker::checkEndPath(CheckerContext &C) const {
828  ProgramStateRef state = C.getState();
829  RegionStateTy M = state->get<RegionState>();
830
831  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
832    RefState RS = I->second;
833    if (RS.isAllocated()) {
834      ExplodedNode *N = C.addTransition(state);
835      if (N)
836        reportLeak(I->first, N, C);
837    }
838  }
839}
840
841bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
842                                CheckerContext &C) const {
843  ProgramStateRef state = C.getState();
844  const RefState *RS = state->get<RegionState>(Sym);
845  if (!RS)
846    return false;
847
848  if (RS->isAllocated()) {
849    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
850    C.addTransition(state);
851    return true;
852  }
853  return false;
854}
855
856void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
857  if (isMemFunction(C.getCalleeDecl(CE), C.getASTContext()))
858    return;
859
860  // Check use after free, when a freed pointer is passed to a call.
861  ProgramStateRef State = C.getState();
862  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
863                                    E = CE->arg_end(); I != E; ++I) {
864    const Expr *A = *I;
865    if (A->getType().getTypePtr()->isAnyPointerType()) {
866      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
867      if (!Sym)
868        continue;
869      if (checkUseAfterFree(Sym, C, A))
870        return;
871    }
872  }
873}
874
875void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
876  const Expr *E = S->getRetValue();
877  if (!E)
878    return;
879
880  // Check if we are returning a symbol.
881  SymbolRef Sym = C.getState()->getSVal(E, C.getLocationContext()).getAsSymbol();
882  if (!Sym)
883    return;
884
885  // Check if we are returning freed memory.
886  if (checkUseAfterFree(Sym, C, E))
887    return;
888
889  // Check if the symbol is escaping.
890  checkEscape(Sym, E, C);
891}
892
893bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
894                                      const Stmt *S) const {
895  assert(Sym);
896  const RefState *RS = C.getState()->get<RegionState>(Sym);
897  if (RS && RS->isReleased()) {
898    if (ExplodedNode *N = C.generateSink()) {
899      if (!BT_UseFree)
900        BT_UseFree.reset(new BuiltinBug("Use of dynamically allocated memory "
901            "after it is freed."));
902
903      BugReport *R = new BugReport(*BT_UseFree, BT_UseFree->getDescription(),N);
904      if (S)
905        R->addRange(S->getSourceRange());
906      R->addVisitor(new MallocBugVisitor(Sym));
907      C.EmitReport(R);
908      return true;
909    }
910  }
911  return false;
912}
913
914// Check if the location is a freed symbolic region.
915void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
916                                  CheckerContext &C) const {
917  SymbolRef Sym = l.getLocSymbolInBase();
918  if (Sym)
919    checkUseAfterFree(Sym, C);
920}
921
922//===----------------------------------------------------------------------===//
923// Check various ways a symbol can be invalidated.
924// TODO: This logic (the next 3 functions) is copied/similar to the
925// RetainRelease checker. We might want to factor this out.
926//===----------------------------------------------------------------------===//
927
928// Stop tracking symbols when a value escapes as a result of checkBind.
929// A value escapes in three possible cases:
930// (1) we are binding to something that is not a memory region.
931// (2) we are binding to a memregion that does not have stack storage
932// (3) we are binding to a memregion with stack storage that the store
933//     does not understand.
934void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
935                              CheckerContext &C) const {
936  // Are we storing to something that causes the value to "escape"?
937  bool escapes = true;
938  ProgramStateRef state = C.getState();
939
940  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
941    escapes = !regionLoc->getRegion()->hasStackStorage();
942
943    if (!escapes) {
944      // To test (3), generate a new state with the binding added.  If it is
945      // the same state, then it escapes (since the store cannot represent
946      // the binding).
947      escapes = (state == (state->bindLoc(*regionLoc, val)));
948    }
949    if (!escapes) {
950      // Case 4: We do not currently model what happens when a symbol is
951      // assigned to a struct field, so be conservative here and let the symbol
952      // go. TODO: This could definitely be improved upon.
953      escapes = !isa<VarRegion>(regionLoc->getRegion());
954    }
955  }
956
957  // If our store can represent the binding and we aren't storing to something
958  // that doesn't have local storage then just return and have the simulation
959  // state continue as is.
960  if (!escapes)
961      return;
962
963  // Otherwise, find all symbols referenced by 'val' that we are tracking
964  // and stop tracking them.
965  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
966  C.addTransition(state);
967}
968
969// If a symbolic region is assumed to NULL (or another constant), stop tracking
970// it - assuming that allocation failed on this path.
971ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
972                                              SVal Cond,
973                                              bool Assumption) const {
974  RegionStateTy RS = state->get<RegionState>();
975  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
976    // If the symbol is assumed to NULL or another constant, this will
977    // return an APSInt*.
978    if (state->getSymVal(I.getKey()))
979      state = state->remove<RegionState>(I.getKey());
980  }
981
982  // Realloc returns 0 when reallocation fails, which means that we should
983  // restore the state of the pointer being reallocated.
984  ReallocMap RP = state->get<ReallocPairs>();
985  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
986    // If the symbol is assumed to NULL or another constant, this will
987    // return an APSInt*.
988    if (state->getSymVal(I.getKey())) {
989      SymbolRef ReallocSym = I.getData().ReallocatedSym;
990      const RefState *RS = state->get<RegionState>(ReallocSym);
991      if (RS) {
992        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
993          state = state->set<RegionState>(ReallocSym,
994                             RefState::getAllocateUnchecked(RS->getStmt()));
995      }
996      state = state->remove<ReallocPairs>(I.getKey());
997    }
998  }
999
1000  return state;
1001}
1002
1003// Check if the function is not known to us. So, for example, we could
1004// conservatively assume it can free/reallocate it's pointer arguments.
1005// (We assume that the pointers cannot escape through calls to system
1006// functions not handled by this checker.)
1007bool MallocChecker::hasUnknownBehavior(const FunctionDecl *FD,
1008                                       ProgramStateRef State) const {
1009  ASTContext &ASTC = State->getStateManager().getContext();
1010
1011  // If it's one of the allocation functions we can reason about, we model it's
1012  // behavior explicitly.
1013  if (isMemFunction(FD, ASTC)) {
1014    return false;
1015  }
1016
1017  // If it's a system call, we know it does not free the memory.
1018  SourceManager &SM = ASTC.getSourceManager();
1019  if (SM.isInSystemHeader(FD->getLocation())) {
1020    return false;
1021  }
1022
1023  // Otherwise, assume that the function can free memory.
1024  return true;
1025}
1026
1027// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1028// escapes, when we are tracking p), do not track the symbol as we cannot reason
1029// about it anymore.
1030ProgramStateRef
1031MallocChecker::checkRegionChanges(ProgramStateRef State,
1032                            const StoreManager::InvalidatedSymbols *invalidated,
1033                                    ArrayRef<const MemRegion *> ExplicitRegions,
1034                                    ArrayRef<const MemRegion *> Regions,
1035                                    const CallOrObjCMessage *Call) const {
1036  if (!invalidated)
1037    return State;
1038  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1039
1040  const FunctionDecl *FD = (Call ?
1041                            dyn_cast_or_null<FunctionDecl>(Call->getDecl()) :0);
1042
1043  // If it's a call which might free or reallocate memory, we assume that all
1044  // regions (explicit and implicit) escaped. Otherwise, whitelist explicit
1045  // pointers; we still can track them.
1046  if (!(FD && hasUnknownBehavior(FD, State))) {
1047    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1048        E = ExplicitRegions.end(); I != E; ++I) {
1049      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1050        WhitelistedSymbols.insert(R->getSymbol());
1051    }
1052  }
1053
1054  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1055       E = invalidated->end(); I!=E; ++I) {
1056    SymbolRef sym = *I;
1057    if (WhitelistedSymbols.count(sym))
1058      continue;
1059    // The symbol escaped.
1060    if (const RefState *RS = State->get<RegionState>(sym))
1061      State = State->set<RegionState>(sym, RefState::getEscaped(RS->getStmt()));
1062  }
1063  return State;
1064}
1065
1066PathDiagnosticPiece *
1067MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1068                                           const ExplodedNode *PrevN,
1069                                           BugReporterContext &BRC,
1070                                           BugReport &BR) {
1071  const RefState *RS = N->getState()->get<RegionState>(Sym);
1072  const RefState *RSPrev = PrevN->getState()->get<RegionState>(Sym);
1073  if (!RS && !RSPrev)
1074    return 0;
1075
1076  const Stmt *S = 0;
1077  const char *Msg = 0;
1078
1079  // Retrieve the associated statement.
1080  ProgramPoint ProgLoc = N->getLocation();
1081  if (isa<StmtPoint>(ProgLoc))
1082    S = cast<StmtPoint>(ProgLoc).getStmt();
1083  // If an assumption was made on a branch, it should be caught
1084  // here by looking at the state transition.
1085  if (isa<BlockEdge>(ProgLoc)) {
1086    const CFGBlock *srcBlk = cast<BlockEdge>(ProgLoc).getSrc();
1087    S = srcBlk->getTerminator();
1088  }
1089  if (!S)
1090    return 0;
1091
1092  // Find out if this is an interesting point and what is the kind.
1093  if (Mode == Normal) {
1094    if (isAllocated(RS, RSPrev, S))
1095      Msg = "Memory is allocated";
1096    else if (isReleased(RS, RSPrev, S))
1097      Msg = "Memory is released";
1098    else if (isReallocFailedCheck(RS, RSPrev, S)) {
1099      Mode = ReallocationFailed;
1100      Msg = "Reallocation failed";
1101    }
1102
1103  // We are in a special mode if a reallocation failed later in the path.
1104  } else if (Mode == ReallocationFailed) {
1105    // Generate a special diagnostic for the first realloc we find.
1106    if (!isAllocated(RS, RSPrev, S) && !isReleased(RS, RSPrev, S))
1107      return 0;
1108
1109    // Check that the name of the function is realloc.
1110    const CallExpr *CE = dyn_cast<CallExpr>(S);
1111    if (!CE)
1112      return 0;
1113    const FunctionDecl *funDecl = CE->getDirectCallee();
1114    if (!funDecl)
1115      return 0;
1116    StringRef FunName = funDecl->getName();
1117    if (!(FunName.equals("realloc") || FunName.equals("reallocf")))
1118      return 0;
1119    Msg = "Attempt to reallocate memory";
1120    Mode = Normal;
1121  }
1122
1123  if (!Msg)
1124    return 0;
1125
1126  // Generate the extra diagnostic.
1127  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1128                             N->getLocationContext());
1129  return new PathDiagnosticEventPiece(Pos, Msg);
1130}
1131
1132
1133#define REGISTER_CHECKER(name) \
1134void ento::register##name(CheckerManager &mgr) {\
1135  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1136}
1137
1138REGISTER_CHECKER(MallocPessimistic)
1139REGISTER_CHECKER(MallocOptimistic)
1140