MallocChecker.cpp revision ede875b794e8f35aa1432e61610ea6e84360b6d3 
1//=== MallocChecker.cpp - A malloc/free checker -------------------*- C++ -*--//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file defines malloc/free checker, which checks for potential memory
11// leaks, double free, and use-after-free problems.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "InterCheckerAPI.h"
17#include "clang/StaticAnalyzer/Core/Checker.h"
18#include "clang/StaticAnalyzer/Core/CheckerManager.h"
19#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
20#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
24#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
25#include "clang/Basic/SourceManager.h"
26#include "llvm/ADT/ImmutableMap.h"
27#include "llvm/ADT/SmallString.h"
28#include "llvm/ADT/STLExtras.h"
29#include <climits>
30
31using namespace clang;
32using namespace ento;
33
34namespace {
35
36class RefState {
37  enum Kind { // Reference to allocated memory.
38              Allocated,
39              // Reference to released/freed memory.
40              Released,
41              // Reference to escaped memory - no assumptions can be made of
42              // the state after the reference escapes.
43              Escaped,
44              // The responsibility for freeing resources has transfered from
45              // this reference. A relinquished symbol should not be freed.
46              Relinquished } K;
47  const Stmt *S;
48
49public:
50  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
51
52  bool isAllocated() const { return K == Allocated; }
53  bool isReleased() const { return K == Released; }
54  bool isRelinquished() const { return K == Relinquished; }
55
56  const Stmt *getStmt() const { return S; }
57
58  bool operator==(const RefState &X) const {
59    return K == X.K && S == X.S;
60  }
61
62  static RefState getAllocated(const Stmt *s) {
63    return RefState(Allocated, s);
64  }
65  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
66  static RefState getEscaped(const Stmt *s) { return RefState(Escaped, s); }
67  static RefState getRelinquished(const Stmt *s) {
68    return RefState(Relinquished, s);
69  }
70
71  void Profile(llvm::FoldingSetNodeID &ID) const {
72    ID.AddInteger(K);
73    ID.AddPointer(S);
74  }
75};
76
77struct ReallocPair {
78  SymbolRef ReallocatedSym;
79  bool IsFreeOnFailure;
80  ReallocPair(SymbolRef S, bool F) : ReallocatedSym(S), IsFreeOnFailure(F) {}
81  void Profile(llvm::FoldingSetNodeID &ID) const {
82    ID.AddInteger(IsFreeOnFailure);
83    ID.AddPointer(ReallocatedSym);
84  }
85  bool operator==(const ReallocPair &X) const {
86    return ReallocatedSym == X.ReallocatedSym &&
87           IsFreeOnFailure == X.IsFreeOnFailure;
88  }
89};
90
91typedef std::pair<const Stmt*, const MemRegion*> LeakInfo;
92
93class MallocChecker : public Checker<check::DeadSymbols,
94                                     check::EndPath,
95                                     check::PreStmt<ReturnStmt>,
96                                     check::PreStmt<CallExpr>,
97                                     check::PostStmt<CallExpr>,
98                                     check::PostStmt<BlockExpr>,
99                                     check::PreObjCMessage,
100                                     check::Location,
101                                     check::Bind,
102                                     eval::Assume,
103                                     check::RegionChanges>
104{
105  mutable OwningPtr<BugType> BT_DoubleFree;
106  mutable OwningPtr<BugType> BT_Leak;
107  mutable OwningPtr<BugType> BT_UseFree;
108  mutable OwningPtr<BugType> BT_BadFree;
109  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
110                         *II_valloc, *II_reallocf, *II_strndup, *II_strdup;
111
112public:
113  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
114                    II_valloc(0), II_reallocf(0), II_strndup(0), II_strdup(0) {}
115
116  /// In pessimistic mode, the checker assumes that it does not know which
117  /// functions might free the memory.
118  struct ChecksFilter {
119    DefaultBool CMallocPessimistic;
120    DefaultBool CMallocOptimistic;
121  };
122
123  ChecksFilter Filter;
124
125  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
126  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
127  void checkPreObjCMessage(const ObjCMethodCall &Call, CheckerContext &C) const;
128  void checkPostStmt(const BlockExpr *BE, CheckerContext &C) const;
129  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
130  void checkEndPath(CheckerContext &C) const;
131  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
132  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
133                            bool Assumption) const;
134  void checkLocation(SVal l, bool isLoad, const Stmt *S,
135                     CheckerContext &C) const;
136  void checkBind(SVal location, SVal val, const Stmt*S,
137                 CheckerContext &C) const;
138  ProgramStateRef
139  checkRegionChanges(ProgramStateRef state,
140                     const StoreManager::InvalidatedSymbols *invalidated,
141                     ArrayRef<const MemRegion *> ExplicitRegions,
142                     ArrayRef<const MemRegion *> Regions,
143                     const CallEvent *Call) const;
144  bool wantsRegionChangeUpdate(ProgramStateRef state) const {
145    return true;
146  }
147
148  void printState(raw_ostream &Out, ProgramStateRef State,
149                  const char *NL, const char *Sep) const;
150
151private:
152  void initIdentifierInfo(ASTContext &C) const;
153
154  /// Check if this is one of the functions which can allocate/reallocate memory
155  /// pointed to by one of its arguments.
156  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
157  bool isFreeFunction(const FunctionDecl *FD, ASTContext &C) const;
158  bool isAllocationFunction(const FunctionDecl *FD, ASTContext &C) const;
159
160  static ProgramStateRef MallocMemReturnsAttr(CheckerContext &C,
161                                              const CallExpr *CE,
162                                              const OwnershipAttr* Att);
163  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
164                                     const Expr *SizeEx, SVal Init,
165                                     ProgramStateRef state) {
166    return MallocMemAux(C, CE,
167                        state->getSVal(SizeEx, C.getLocationContext()),
168                        Init, state);
169  }
170
171  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
172                                     SVal SizeEx, SVal Init,
173                                     ProgramStateRef state);
174
175  /// Update the RefState to reflect the new memory allocation.
176  static ProgramStateRef MallocUpdateRefState(CheckerContext &C,
177                                              const CallExpr *CE,
178                                              ProgramStateRef state);
179
180  ProgramStateRef FreeMemAttr(CheckerContext &C, const CallExpr *CE,
181                              const OwnershipAttr* Att) const;
182  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
183                             ProgramStateRef state, unsigned Num,
184                             bool Hold) const;
185  ProgramStateRef FreeMemAux(CheckerContext &C, const Expr *Arg,
186                             const Expr *ParentExpr,
187                             ProgramStateRef state,
188                             bool Hold) const;
189
190  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
191                             bool FreesMemOnFailure) const;
192  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
193
194  ///\brief Check if the memory associated with this symbol was released.
195  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
196
197  bool checkEscape(SymbolRef Sym, const Stmt *S, CheckerContext &C) const;
198  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
199                         const Stmt *S = 0) const;
200
201  /// Check if the function is not known to us. So, for example, we could
202  /// conservatively assume it can free/reallocate it's pointer arguments.
203  bool doesNotFreeMemory(const CallEvent *Call,
204                         ProgramStateRef State) const;
205
206  static bool SummarizeValue(raw_ostream &os, SVal V);
207  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
208  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
209
210  /// Find the location of the allocation for Sym on the path leading to the
211  /// exploded node N.
212  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
213                             CheckerContext &C) const;
214
215  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
216
217  /// The bug visitor which allows us to print extra diagnostics along the
218  /// BugReport path. For example, showing the allocation site of the leaked
219  /// region.
220  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
221  protected:
222    enum NotificationMode {
223      Normal,
224      ReallocationFailed
225    };
226
227    // The allocated region symbol tracked by the main analysis.
228    SymbolRef Sym;
229
230    // The mode we are in, i.e. what kind of diagnostics will be emitted.
231    NotificationMode Mode;
232
233    // A symbol from when the primary region should have been reallocated.
234    SymbolRef FailedReallocSymbol;
235
236    bool IsLeak;
237
238  public:
239    MallocBugVisitor(SymbolRef S, bool isLeak = false)
240       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
241
242    virtual ~MallocBugVisitor() {}
243
244    void Profile(llvm::FoldingSetNodeID &ID) const {
245      static int X = 0;
246      ID.AddPointer(&X);
247      ID.AddPointer(Sym);
248    }
249
250    inline bool isAllocated(const RefState *S, const RefState *SPrev,
251                            const Stmt *Stmt) {
252      // Did not track -> allocated. Other state (released) -> allocated.
253      return (Stmt && isa<CallExpr>(Stmt) &&
254              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
255    }
256
257    inline bool isReleased(const RefState *S, const RefState *SPrev,
258                           const Stmt *Stmt) {
259      // Did not track -> released. Other state (allocated) -> released.
260      return (Stmt && isa<CallExpr>(Stmt) &&
261              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
262    }
263
264    inline bool isRelinquished(const RefState *S, const RefState *SPrev,
265                               const Stmt *Stmt) {
266      // Did not track -> relinquished. Other state (allocated) -> relinquished.
267      return (Stmt && (isa<CallExpr>(Stmt) || isa<ObjCMessageExpr>(Stmt) ||
268                                              isa<ObjCPropertyRefExpr>(Stmt)) &&
269              (S && S->isRelinquished()) &&
270              (!SPrev || !SPrev->isRelinquished()));
271    }
272
273    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
274                                     const Stmt *Stmt) {
275      // If the expression is not a call, and the state change is
276      // released -> allocated, it must be the realloc return value
277      // check. If we have to handle more cases here, it might be cleaner just
278      // to track this extra bit in the state itself.
279      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
280              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
281    }
282
283    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
284                                   const ExplodedNode *PrevN,
285                                   BugReporterContext &BRC,
286                                   BugReport &BR);
287
288    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
289                                    const ExplodedNode *EndPathNode,
290                                    BugReport &BR) {
291      if (!IsLeak)
292        return 0;
293
294      PathDiagnosticLocation L =
295        PathDiagnosticLocation::createEndOfPath(EndPathNode,
296                                                BRC.getSourceManager());
297      // Do not add the statement itself as a range in case of leak.
298      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
299    }
300
301  private:
302    class StackHintGeneratorForReallocationFailed
303        : public StackHintGeneratorForSymbol {
304    public:
305      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
306        : StackHintGeneratorForSymbol(S, M) {}
307
308      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
309        SmallString<200> buf;
310        llvm::raw_svector_ostream os(buf);
311
312        os << "Reallocation of ";
313        // Printed parameters start at 1, not 0.
314        printOrdinal(++ArgIndex, os);
315        os << " parameter failed";
316
317        return os.str();
318      }
319
320      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
321        return "Reallocation of returned value failed";
322      }
323    };
324  };
325};
326} // end anonymous namespace
327
328typedef llvm::ImmutableMap<SymbolRef, RefState> RegionStateTy;
329typedef llvm::ImmutableMap<SymbolRef, ReallocPair > ReallocMap;
330class RegionState {};
331class ReallocPairs {};
332namespace clang {
333namespace ento {
334  template <>
335  struct ProgramStateTrait<RegionState>
336    : public ProgramStatePartialTrait<RegionStateTy> {
337    static void *GDMIndex() { static int x; return &x; }
338  };
339
340  template <>
341  struct ProgramStateTrait<ReallocPairs>
342    : public ProgramStatePartialTrait<ReallocMap> {
343    static void *GDMIndex() { static int x; return &x; }
344  };
345}
346}
347
348namespace {
349class StopTrackingCallback : public SymbolVisitor {
350  ProgramStateRef state;
351public:
352  StopTrackingCallback(ProgramStateRef st) : state(st) {}
353  ProgramStateRef getState() const { return state; }
354
355  bool VisitSymbol(SymbolRef sym) {
356    state = state->remove<RegionState>(sym);
357    return true;
358  }
359};
360} // end anonymous namespace
361
362void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
363  if (II_malloc)
364    return;
365  II_malloc = &Ctx.Idents.get("malloc");
366  II_free = &Ctx.Idents.get("free");
367  II_realloc = &Ctx.Idents.get("realloc");
368  II_reallocf = &Ctx.Idents.get("reallocf");
369  II_calloc = &Ctx.Idents.get("calloc");
370  II_valloc = &Ctx.Idents.get("valloc");
371  II_strdup = &Ctx.Idents.get("strdup");
372  II_strndup = &Ctx.Idents.get("strndup");
373}
374
375bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
376  if (isFreeFunction(FD, C))
377    return true;
378
379  if (isAllocationFunction(FD, C))
380    return true;
381
382  return false;
383}
384
385bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
386                                         ASTContext &C) const {
387  if (!FD)
388    return false;
389
390  if (FD->getKind() == Decl::Function) {
391    IdentifierInfo *FunI = FD->getIdentifier();
392    initIdentifierInfo(C);
393
394    if (FunI == II_malloc || FunI == II_realloc ||
395        FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
396        FunI == II_strdup || FunI == II_strndup)
397      return true;
398  }
399
400  if (Filter.CMallocOptimistic && FD->hasAttrs())
401    for (specific_attr_iterator<OwnershipAttr>
402           i = FD->specific_attr_begin<OwnershipAttr>(),
403           e = FD->specific_attr_end<OwnershipAttr>();
404           i != e; ++i)
405      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
406        return true;
407  return false;
408}
409
410bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
411  if (!FD)
412    return false;
413
414  if (FD->getKind() == Decl::Function) {
415    IdentifierInfo *FunI = FD->getIdentifier();
416    initIdentifierInfo(C);
417
418    if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
419      return true;
420  }
421
422  if (Filter.CMallocOptimistic && FD->hasAttrs())
423    for (specific_attr_iterator<OwnershipAttr>
424           i = FD->specific_attr_begin<OwnershipAttr>(),
425           e = FD->specific_attr_end<OwnershipAttr>();
426           i != e; ++i)
427      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
428          (*i)->getOwnKind() == OwnershipAttr::Holds)
429        return true;
430  return false;
431}
432
433void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
434  const FunctionDecl *FD = C.getCalleeDecl(CE);
435  if (!FD)
436    return;
437
438  ProgramStateRef State = C.getState();
439
440  if (FD->getKind() == Decl::Function) {
441    initIdentifierInfo(C.getASTContext());
442    IdentifierInfo *FunI = FD->getIdentifier();
443
444    if (FunI == II_malloc || FunI == II_valloc) {
445      if (CE->getNumArgs() < 1)
446        return;
447      State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
448    } else if (FunI == II_realloc) {
449      State = ReallocMem(C, CE, false);
450    } else if (FunI == II_reallocf) {
451      State = ReallocMem(C, CE, true);
452    } else if (FunI == II_calloc) {
453      State = CallocMem(C, CE);
454    } else if (FunI == II_free) {
455      State = FreeMemAux(C, CE, State, 0, false);
456    } else if (FunI == II_strdup) {
457      State = MallocUpdateRefState(C, CE, State);
458    } else if (FunI == II_strndup) {
459      State = MallocUpdateRefState(C, CE, State);
460    }
461  }
462
463  if (Filter.CMallocOptimistic) {
464    // Check all the attributes, if there are any.
465    // There can be multiple of these attributes.
466    if (FD->hasAttrs())
467      for (specific_attr_iterator<OwnershipAttr>
468          i = FD->specific_attr_begin<OwnershipAttr>(),
469          e = FD->specific_attr_end<OwnershipAttr>();
470          i != e; ++i) {
471        switch ((*i)->getOwnKind()) {
472        case OwnershipAttr::Returns:
473          State = MallocMemReturnsAttr(C, CE, *i);
474          break;
475        case OwnershipAttr::Takes:
476        case OwnershipAttr::Holds:
477          State = FreeMemAttr(C, CE, *i);
478          break;
479        }
480      }
481  }
482  C.addTransition(State);
483}
484
485static bool isFreeWhenDoneSetToZero(const ObjCMethodCall &Call) {
486  Selector S = Call.getSelector();
487  for (unsigned i = 1; i < S.getNumArgs(); ++i)
488    if (S.getNameForSlot(i).equals("freeWhenDone"))
489      if (Call.getArgSVal(i).isConstant(0))
490        return true;
491
492  return false;
493}
494
495void MallocChecker::checkPreObjCMessage(const ObjCMethodCall &Call,
496                                        CheckerContext &C) const {
497  // If the first selector is dataWithBytesNoCopy, assume that the memory will
498  // be released with 'free' by the new object.
499  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
500  // Unless 'freeWhenDone' param set to 0.
501  // TODO: Check that the memory was allocated with malloc.
502  Selector S = Call.getSelector();
503  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
504       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
505       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
506      !isFreeWhenDoneSetToZero(Call)){
507    unsigned int argIdx  = 0;
508    C.addTransition(FreeMemAux(C, Call.getArgExpr(argIdx),
509                    Call.getOriginExpr(), C.getState(), true));
510  }
511}
512
513ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
514                                                    const CallExpr *CE,
515                                                    const OwnershipAttr* Att) {
516  if (Att->getModule() != "malloc")
517    return 0;
518
519  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
520  if (I != E) {
521    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
522  }
523  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
524}
525
526ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
527                                           const CallExpr *CE,
528                                           SVal Size, SVal Init,
529                                           ProgramStateRef state) {
530
531  // Bind the return value to the symbolic value from the heap region.
532  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
533  // side effects other than what we model here.
534  unsigned Count = C.getCurrentBlockCount();
535  SValBuilder &svalBuilder = C.getSValBuilder();
536  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
537  DefinedSVal RetVal =
538    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
539  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
540
541  // We expect the malloc functions to return a pointer.
542  if (!isa<Loc>(RetVal))
543    return 0;
544
545  // Fill the region with the initialization value.
546  state = state->bindDefault(RetVal, Init);
547
548  // Set the region's extent equal to the Size parameter.
549  const SymbolicRegion *R =
550      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
551  if (!R)
552    return 0;
553  if (isa<DefinedOrUnknownSVal>(Size)) {
554    SValBuilder &svalBuilder = C.getSValBuilder();
555    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
556    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
557    DefinedOrUnknownSVal extentMatchesSize =
558        svalBuilder.evalEQ(state, Extent, DefinedSize);
559
560    state = state->assume(extentMatchesSize, true);
561    assert(state);
562  }
563
564  return MallocUpdateRefState(C, CE, state);
565}
566
567ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
568                                                    const CallExpr *CE,
569                                                    ProgramStateRef state) {
570  // Get the return value.
571  SVal retVal = state->getSVal(CE, C.getLocationContext());
572
573  // We expect the malloc functions to return a pointer.
574  if (!isa<Loc>(retVal))
575    return 0;
576
577  SymbolRef Sym = retVal.getAsLocSymbol();
578  assert(Sym);
579
580  // Set the symbol's state to Allocated.
581  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
582
583}
584
585ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
586                                           const CallExpr *CE,
587                                           const OwnershipAttr* Att) const {
588  if (Att->getModule() != "malloc")
589    return 0;
590
591  ProgramStateRef State = C.getState();
592
593  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
594       I != E; ++I) {
595    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
596                               Att->getOwnKind() == OwnershipAttr::Holds);
597    if (StateI)
598      State = StateI;
599  }
600  return State;
601}
602
603ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
604                                          const CallExpr *CE,
605                                          ProgramStateRef state,
606                                          unsigned Num,
607                                          bool Hold) const {
608  if (CE->getNumArgs() < (Num + 1))
609    return 0;
610
611  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold);
612}
613
614ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
615                                          const Expr *ArgExpr,
616                                          const Expr *ParentExpr,
617                                          ProgramStateRef state,
618                                          bool Hold) const {
619
620  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
621  if (!isa<DefinedOrUnknownSVal>(ArgVal))
622    return 0;
623  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
624
625  // Check for null dereferences.
626  if (!isa<Loc>(location))
627    return 0;
628
629  // The explicit NULL case, no operation is performed.
630  ProgramStateRef notNullState, nullState;
631  llvm::tie(notNullState, nullState) = state->assume(location);
632  if (nullState && !notNullState)
633    return 0;
634
635  // Unknown values could easily be okay
636  // Undefined values are handled elsewhere
637  if (ArgVal.isUnknownOrUndef())
638    return 0;
639
640  const MemRegion *R = ArgVal.getAsRegion();
641
642  // Nonlocs can't be freed, of course.
643  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
644  if (!R) {
645    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
646    return 0;
647  }
648
649  R = R->StripCasts();
650
651  // Blocks might show up as heap data, but should not be free()d
652  if (isa<BlockDataRegion>(R)) {
653    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
654    return 0;
655  }
656
657  const MemSpaceRegion *MS = R->getMemorySpace();
658
659  // Parameters, locals, statics, and globals shouldn't be freed.
660  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
661    // FIXME: at the time this code was written, malloc() regions were
662    // represented by conjured symbols, which are all in UnknownSpaceRegion.
663    // This means that there isn't actually anything from HeapSpaceRegion
664    // that should be freed, even though we allow it here.
665    // Of course, free() can work on memory allocated outside the current
666    // function, so UnknownSpaceRegion is always a possibility.
667    // False negatives are better than false positives.
668
669    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
670    return 0;
671  }
672
673  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
674  // Various cases could lead to non-symbol values here.
675  // For now, ignore them.
676  if (!SR)
677    return 0;
678
679  SymbolRef Sym = SR->getSymbol();
680  const RefState *RS = state->get<RegionState>(Sym);
681
682  // Check double free.
683  if (RS && (RS->isReleased() || RS->isRelinquished())) {
684    if (ExplodedNode *N = C.generateSink()) {
685      if (!BT_DoubleFree)
686        BT_DoubleFree.reset(
687          new BugType("Double free", "Memory Error"));
688      BugReport *R = new BugReport(*BT_DoubleFree,
689        (RS->isReleased() ? "Attempt to free released memory" :
690                            "Attempt to free non-owned memory"), N);
691      R->addRange(ArgExpr->getSourceRange());
692      R->markInteresting(Sym);
693      R->addVisitor(new MallocBugVisitor(Sym));
694      C.EmitReport(R);
695    }
696    return 0;
697  }
698
699  // Normal free.
700  if (Hold)
701    return state->set<RegionState>(Sym, RefState::getRelinquished(ParentExpr));
702  return state->set<RegionState>(Sym, RefState::getReleased(ParentExpr));
703}
704
705bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
706  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
707    os << "an integer (" << IntVal->getValue() << ")";
708  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
709    os << "a constant address (" << ConstAddr->getValue() << ")";
710  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
711    os << "the address of the label '" << Label->getLabel()->getName() << "'";
712  else
713    return false;
714
715  return true;
716}
717
718bool MallocChecker::SummarizeRegion(raw_ostream &os,
719                                    const MemRegion *MR) {
720  switch (MR->getKind()) {
721  case MemRegion::FunctionTextRegionKind: {
722    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
723    if (FD)
724      os << "the address of the function '" << *FD << '\'';
725    else
726      os << "the address of a function";
727    return true;
728  }
729  case MemRegion::BlockTextRegionKind:
730    os << "block text";
731    return true;
732  case MemRegion::BlockDataRegionKind:
733    // FIXME: where the block came from?
734    os << "a block";
735    return true;
736  default: {
737    const MemSpaceRegion *MS = MR->getMemorySpace();
738
739    if (isa<StackLocalsSpaceRegion>(MS)) {
740      const VarRegion *VR = dyn_cast<VarRegion>(MR);
741      const VarDecl *VD;
742      if (VR)
743        VD = VR->getDecl();
744      else
745        VD = NULL;
746
747      if (VD)
748        os << "the address of the local variable '" << VD->getName() << "'";
749      else
750        os << "the address of a local stack variable";
751      return true;
752    }
753
754    if (isa<StackArgumentsSpaceRegion>(MS)) {
755      const VarRegion *VR = dyn_cast<VarRegion>(MR);
756      const VarDecl *VD;
757      if (VR)
758        VD = VR->getDecl();
759      else
760        VD = NULL;
761
762      if (VD)
763        os << "the address of the parameter '" << VD->getName() << "'";
764      else
765        os << "the address of a parameter";
766      return true;
767    }
768
769    if (isa<GlobalsSpaceRegion>(MS)) {
770      const VarRegion *VR = dyn_cast<VarRegion>(MR);
771      const VarDecl *VD;
772      if (VR)
773        VD = VR->getDecl();
774      else
775        VD = NULL;
776
777      if (VD) {
778        if (VD->isStaticLocal())
779          os << "the address of the static variable '" << VD->getName() << "'";
780        else
781          os << "the address of the global variable '" << VD->getName() << "'";
782      } else
783        os << "the address of a global variable";
784      return true;
785    }
786
787    return false;
788  }
789  }
790}
791
792void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
793                                  SourceRange range) const {
794  if (ExplodedNode *N = C.generateSink()) {
795    if (!BT_BadFree)
796      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
797
798    SmallString<100> buf;
799    llvm::raw_svector_ostream os(buf);
800
801    const MemRegion *MR = ArgVal.getAsRegion();
802    if (MR) {
803      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
804        MR = ER->getSuperRegion();
805
806      // Special case for alloca()
807      if (isa<AllocaRegion>(MR))
808        os << "Argument to free() was allocated by alloca(), not malloc()";
809      else {
810        os << "Argument to free() is ";
811        if (SummarizeRegion(os, MR))
812          os << ", which is not memory allocated by malloc()";
813        else
814          os << "not memory allocated by malloc()";
815      }
816    } else {
817      os << "Argument to free() is ";
818      if (SummarizeValue(os, ArgVal))
819        os << ", which is not memory allocated by malloc()";
820      else
821        os << "not memory allocated by malloc()";
822    }
823
824    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
825    R->markInteresting(MR);
826    R->addRange(range);
827    C.EmitReport(R);
828  }
829}
830
831ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
832                                          const CallExpr *CE,
833                                          bool FreesOnFail) const {
834  if (CE->getNumArgs() < 2)
835    return 0;
836
837  ProgramStateRef state = C.getState();
838  const Expr *arg0Expr = CE->getArg(0);
839  const LocationContext *LCtx = C.getLocationContext();
840  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
841  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
842    return 0;
843  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
844
845  SValBuilder &svalBuilder = C.getSValBuilder();
846
847  DefinedOrUnknownSVal PtrEQ =
848    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
849
850  // Get the size argument. If there is no size arg then give up.
851  const Expr *Arg1 = CE->getArg(1);
852  if (!Arg1)
853    return 0;
854
855  // Get the value of the size argument.
856  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
857  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
858    return 0;
859  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
860
861  // Compare the size argument to 0.
862  DefinedOrUnknownSVal SizeZero =
863    svalBuilder.evalEQ(state, Arg1Val,
864                       svalBuilder.makeIntValWithPtrWidth(0, false));
865
866  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
867  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
868  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
869  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
870  // We only assume exceptional states if they are definitely true; if the
871  // state is under-constrained, assume regular realloc behavior.
872  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
873  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
874
875  // If the ptr is NULL and the size is not 0, the call is equivalent to
876  // malloc(size).
877  if ( PrtIsNull && !SizeIsZero) {
878    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
879                                               UndefinedVal(), StatePtrIsNull);
880    return stateMalloc;
881  }
882
883  if (PrtIsNull && SizeIsZero)
884    return 0;
885
886  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
887  assert(!PrtIsNull);
888  SymbolRef FromPtr = arg0Val.getAsSymbol();
889  SVal RetVal = state->getSVal(CE, LCtx);
890  SymbolRef ToPtr = RetVal.getAsSymbol();
891  if (!FromPtr || !ToPtr)
892    return 0;
893
894  // If the size is 0, free the memory.
895  if (SizeIsZero)
896    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
897      // The semantics of the return value are:
898      // If size was equal to 0, either NULL or a pointer suitable to be passed
899      // to free() is returned. We just free the input pointer and do not add
900      // any constrains on the output pointer.
901      return stateFree;
902    }
903
904  // Default behavior.
905  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
906    // FIXME: We should copy the content of the original buffer.
907    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
908                                                UnknownVal(), stateFree);
909    if (!stateRealloc)
910      return 0;
911    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
912                                            ReallocPair(FromPtr, FreesOnFail));
913    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
914    return stateRealloc;
915  }
916  return 0;
917}
918
919ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
920  if (CE->getNumArgs() < 2)
921    return 0;
922
923  ProgramStateRef state = C.getState();
924  SValBuilder &svalBuilder = C.getSValBuilder();
925  const LocationContext *LCtx = C.getLocationContext();
926  SVal count = state->getSVal(CE->getArg(0), LCtx);
927  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
928  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
929                                        svalBuilder.getContext().getSizeType());
930  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
931
932  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
933}
934
935LeakInfo
936MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
937                                 CheckerContext &C) const {
938  const LocationContext *LeakContext = N->getLocationContext();
939  // Walk the ExplodedGraph backwards and find the first node that referred to
940  // the tracked symbol.
941  const ExplodedNode *AllocNode = N;
942  const MemRegion *ReferenceRegion = 0;
943
944  while (N) {
945    ProgramStateRef State = N->getState();
946    if (!State->get<RegionState>(Sym))
947      break;
948
949    // Find the most recent expression bound to the symbol in the current
950    // context.
951    if (!ReferenceRegion) {
952      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
953        SVal Val = State->getSVal(MR);
954        if (Val.getAsLocSymbol() == Sym)
955          ReferenceRegion = MR;
956      }
957    }
958
959    // Allocation node, is the last node in the current context in which the
960    // symbol was tracked.
961    if (N->getLocationContext() == LeakContext)
962      AllocNode = N;
963    N = N->pred_empty() ? NULL : *(N->pred_begin());
964  }
965
966  ProgramPoint P = AllocNode->getLocation();
967  const Stmt *AllocationStmt = 0;
968  if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&P))
969    AllocationStmt = Exit->getCalleeContext()->getCallSite();
970  else if (StmtPoint *SP = dyn_cast<StmtPoint>(&P))
971    AllocationStmt = SP->getStmt();
972
973  return LeakInfo(AllocationStmt, ReferenceRegion);
974}
975
976void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
977                               CheckerContext &C) const {
978  assert(N);
979  if (!BT_Leak) {
980    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
981    // Leaks should not be reported if they are post-dominated by a sink:
982    // (1) Sinks are higher importance bugs.
983    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
984    //     with __noreturn functions such as assert() or exit(). We choose not
985    //     to report leaks on such paths.
986    BT_Leak->setSuppressOnSink(true);
987  }
988
989  // Most bug reports are cached at the location where they occurred.
990  // With leaks, we want to unique them by the location where they were
991  // allocated, and only report a single path.
992  PathDiagnosticLocation LocUsedForUniqueing;
993  const Stmt *AllocStmt = 0;
994  const MemRegion *Region = 0;
995  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
996  if (AllocStmt)
997    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
998                            C.getSourceManager(), N->getLocationContext());
999
1000  SmallString<200> buf;
1001  llvm::raw_svector_ostream os(buf);
1002  os << "Memory is never released; potential leak";
1003  if (Region) {
1004    os << " of memory pointed to by '";
1005    Region->dumpPretty(os);
1006    os <<'\'';
1007  }
1008
1009  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
1010  R->markInteresting(Sym);
1011  R->addVisitor(new MallocBugVisitor(Sym, true));
1012  C.EmitReport(R);
1013}
1014
1015void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1016                                     CheckerContext &C) const
1017{
1018  if (!SymReaper.hasDeadSymbols())
1019    return;
1020
1021  ProgramStateRef state = C.getState();
1022  RegionStateTy RS = state->get<RegionState>();
1023  RegionStateTy::Factory &F = state->get_context<RegionState>();
1024
1025  bool generateReport = false;
1026  llvm::SmallVector<SymbolRef, 2> Errors;
1027  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1028    if (SymReaper.isDead(I->first)) {
1029      if (I->second.isAllocated()) {
1030        generateReport = true;
1031        Errors.push_back(I->first);
1032      }
1033      // Remove the dead symbol from the map.
1034      RS = F.remove(RS, I->first);
1035
1036    }
1037  }
1038
1039  // Cleanup the Realloc Pairs Map.
1040  ReallocMap RP = state->get<ReallocPairs>();
1041  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1042    if (SymReaper.isDead(I->first) ||
1043        SymReaper.isDead(I->second.ReallocatedSym)) {
1044      state = state->remove<ReallocPairs>(I->first);
1045    }
1046  }
1047
1048  // Generate leak node.
1049  static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1050  ExplodedNode *N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1051
1052  if (generateReport) {
1053    for (llvm::SmallVector<SymbolRef, 2>::iterator
1054         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1055      reportLeak(*I, N, C);
1056    }
1057  }
1058  C.addTransition(state->set<RegionState>(RS), N);
1059}
1060
1061void MallocChecker::checkEndPath(CheckerContext &C) const {
1062  ProgramStateRef state = C.getState();
1063  RegionStateTy M = state->get<RegionState>();
1064
1065  // If inside inlined call, skip it.
1066  if (C.getLocationContext()->getParent() != 0)
1067    return;
1068
1069  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1070    RefState RS = I->second;
1071    if (RS.isAllocated()) {
1072      ExplodedNode *N = C.addTransition(state);
1073      if (N)
1074        reportLeak(I->first, N, C);
1075    }
1076  }
1077}
1078
1079bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
1080                                CheckerContext &C) const {
1081  ProgramStateRef state = C.getState();
1082  const RefState *RS = state->get<RegionState>(Sym);
1083  if (!RS)
1084    return false;
1085
1086  if (RS->isAllocated()) {
1087    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
1088    C.addTransition(state);
1089    return true;
1090  }
1091  return false;
1092}
1093
1094void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1095  // We will check for double free in the post visit.
1096  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1097    return;
1098
1099  // Check use after free, when a freed pointer is passed to a call.
1100  ProgramStateRef State = C.getState();
1101  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1102                                    E = CE->arg_end(); I != E; ++I) {
1103    const Expr *A = *I;
1104    if (A->getType().getTypePtr()->isAnyPointerType()) {
1105      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1106      if (!Sym)
1107        continue;
1108      if (checkUseAfterFree(Sym, C, A))
1109        return;
1110    }
1111  }
1112}
1113
1114void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1115  const Expr *E = S->getRetValue();
1116  if (!E)
1117    return;
1118
1119  // Check if we are returning a symbol.
1120  SVal RetVal = C.getState()->getSVal(E, C.getLocationContext());
1121  SymbolRef Sym = RetVal.getAsSymbol();
1122  if (!Sym)
1123    // If we are returning a field of the allocated struct or an array element,
1124    // the callee could still free the memory.
1125    // TODO: This logic should be a part of generic symbol escape callback.
1126    if (const MemRegion *MR = RetVal.getAsRegion())
1127      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1128        if (const SymbolicRegion *BMR =
1129              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1130          Sym = BMR->getSymbol();
1131  if (!Sym)
1132    return;
1133
1134  // Check if we are returning freed memory.
1135  if (checkUseAfterFree(Sym, C, E))
1136    return;
1137
1138  // If this function body is not inlined, check if the symbol is escaping.
1139  if (C.getLocationContext()->getParent() == 0)
1140    checkEscape(Sym, E, C);
1141}
1142
1143// TODO: Blocks should be either inlined or should call invalidate regions
1144// upon invocation. After that's in place, special casing here will not be
1145// needed.
1146void MallocChecker::checkPostStmt(const BlockExpr *BE,
1147                                  CheckerContext &C) const {
1148
1149  // Scan the BlockDecRefExprs for any object the retain count checker
1150  // may be tracking.
1151  if (!BE->getBlockDecl()->hasCaptures())
1152    return;
1153
1154  ProgramStateRef state = C.getState();
1155  const BlockDataRegion *R =
1156    cast<BlockDataRegion>(state->getSVal(BE,
1157                                         C.getLocationContext()).getAsRegion());
1158
1159  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1160                                            E = R->referenced_vars_end();
1161
1162  if (I == E)
1163    return;
1164
1165  SmallVector<const MemRegion*, 10> Regions;
1166  const LocationContext *LC = C.getLocationContext();
1167  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1168
1169  for ( ; I != E; ++I) {
1170    const VarRegion *VR = *I;
1171    if (VR->getSuperRegion() == R) {
1172      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1173    }
1174    Regions.push_back(VR);
1175  }
1176
1177  state =
1178    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1179                                    Regions.data() + Regions.size()).getState();
1180  C.addTransition(state);
1181}
1182
1183bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1184  assert(Sym);
1185  const RefState *RS = C.getState()->get<RegionState>(Sym);
1186  return (RS && RS->isReleased());
1187}
1188
1189bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1190                                      const Stmt *S) const {
1191  if (isReleased(Sym, C)) {
1192    if (ExplodedNode *N = C.generateSink()) {
1193      if (!BT_UseFree)
1194        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1195
1196      BugReport *R = new BugReport(*BT_UseFree,
1197                                   "Use of memory after it is freed",N);
1198      if (S)
1199        R->addRange(S->getSourceRange());
1200      R->markInteresting(Sym);
1201      R->addVisitor(new MallocBugVisitor(Sym));
1202      C.EmitReport(R);
1203      return true;
1204    }
1205  }
1206  return false;
1207}
1208
1209// Check if the location is a freed symbolic region.
1210void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1211                                  CheckerContext &C) const {
1212  SymbolRef Sym = l.getLocSymbolInBase();
1213  if (Sym)
1214    checkUseAfterFree(Sym, C, S);
1215}
1216
1217//===----------------------------------------------------------------------===//
1218// Check various ways a symbol can be invalidated.
1219// TODO: This logic (the next 3 functions) is copied/similar to the
1220// RetainRelease checker. We might want to factor this out.
1221//===----------------------------------------------------------------------===//
1222
1223// Stop tracking symbols when a value escapes as a result of checkBind.
1224// A value escapes in three possible cases:
1225// (1) we are binding to something that is not a memory region.
1226// (2) we are binding to a memregion that does not have stack storage
1227// (3) we are binding to a memregion with stack storage that the store
1228//     does not understand.
1229void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1230                              CheckerContext &C) const {
1231  // Are we storing to something that causes the value to "escape"?
1232  bool escapes = true;
1233  ProgramStateRef state = C.getState();
1234
1235  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1236    escapes = !regionLoc->getRegion()->hasStackStorage();
1237
1238    if (!escapes) {
1239      // To test (3), generate a new state with the binding added.  If it is
1240      // the same state, then it escapes (since the store cannot represent
1241      // the binding).
1242      // Do this only if we know that the store is not supposed to generate the
1243      // same state.
1244      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1245      if (StoredVal != val)
1246        escapes = (state == (state->bindLoc(*regionLoc, val)));
1247    }
1248    if (!escapes) {
1249      // Case 4: We do not currently model what happens when a symbol is
1250      // assigned to a struct field, so be conservative here and let the symbol
1251      // go. TODO: This could definitely be improved upon.
1252      escapes = !isa<VarRegion>(regionLoc->getRegion());
1253    }
1254  }
1255
1256  // If our store can represent the binding and we aren't storing to something
1257  // that doesn't have local storage then just return and have the simulation
1258  // state continue as is.
1259  if (!escapes)
1260      return;
1261
1262  // Otherwise, find all symbols referenced by 'val' that we are tracking
1263  // and stop tracking them.
1264  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1265  C.addTransition(state);
1266}
1267
1268// If a symbolic region is assumed to NULL (or another constant), stop tracking
1269// it - assuming that allocation failed on this path.
1270ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1271                                              SVal Cond,
1272                                              bool Assumption) const {
1273  RegionStateTy RS = state->get<RegionState>();
1274  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1275    // If the symbol is assumed to NULL or another constant, this will
1276    // return an APSInt*.
1277    if (state->getSymVal(I.getKey()))
1278      state = state->remove<RegionState>(I.getKey());
1279  }
1280
1281  // Realloc returns 0 when reallocation fails, which means that we should
1282  // restore the state of the pointer being reallocated.
1283  ReallocMap RP = state->get<ReallocPairs>();
1284  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1285    // If the symbol is assumed to NULL or another constant, this will
1286    // return an APSInt*.
1287    if (state->getSymVal(I.getKey())) {
1288      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1289      const RefState *RS = state->get<RegionState>(ReallocSym);
1290      if (RS) {
1291        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1292          state = state->set<RegionState>(ReallocSym,
1293                             RefState::getAllocated(RS->getStmt()));
1294      }
1295      state = state->remove<ReallocPairs>(I.getKey());
1296    }
1297  }
1298
1299  return state;
1300}
1301
1302// Check if the function is known to us. So, for example, we could
1303// conservatively assume it can free/reallocate its pointer arguments.
1304// (We assume that the pointers cannot escape through calls to system
1305// functions not handled by this checker.)
1306bool MallocChecker::doesNotFreeMemory(const CallEvent *Call,
1307                                      ProgramStateRef State) const {
1308  assert(Call);
1309
1310  // For now, assume that any C++ call can free memory.
1311  // TODO: If we want to be more optimistic here, we'll need to make sure that
1312  // regions escape to C++ containers. They seem to do that even now, but for
1313  // mysterious reasons.
1314  if (!(isa<FunctionCall>(Call) || isa<ObjCMethodCall>(Call)))
1315    return false;
1316
1317  // Check Objective-C messages by selector name.
1318  if (const ObjCMethodCall *Msg = dyn_cast<ObjCMethodCall>(Call)) {
1319    // If it's not a framework call, or if it takes a callback, assume it
1320    // can free memory.
1321    if (!Call->isInSystemHeader() || Call->hasNonZeroCallbackArg())
1322      return false;
1323
1324    Selector S = Msg->getSelector();
1325
1326    // Whitelist the ObjC methods which do free memory.
1327    // - Anything containing 'freeWhenDone' param set to 1.
1328    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1329    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1330      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1331        if (Call->getArgSVal(i).isConstant(1))
1332          return false;
1333        else
1334          return true;
1335      }
1336    }
1337
1338    // If the first selector ends with NoCopy, assume that the ownership is
1339    // transferred as well.
1340    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1341    StringRef FirstSlot = S.getNameForSlot(0);
1342    if (FirstSlot.endswith("NoCopy"))
1343      return false;
1344
1345    // If the first selector starts with addPointer, insertPointer,
1346    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1347    // This is similar to C++ containers (vector); we still might want to check
1348    // that the pointers get freed by following the container itself.
1349    if (FirstSlot.startswith("addPointer") ||
1350        FirstSlot.startswith("insertPointer") ||
1351        FirstSlot.startswith("replacePointer")) {
1352      return false;
1353    }
1354
1355    // Otherwise, assume that the method does not free memory.
1356    // Most framework methods do not free memory.
1357    return true;
1358  }
1359
1360  // At this point the only thing left to handle is straight function calls.
1361  const FunctionDecl *FD = cast<FunctionCall>(Call)->getDecl();
1362  if (!FD)
1363    return false;
1364
1365  ASTContext &ASTC = State->getStateManager().getContext();
1366
1367  // If it's one of the allocation functions we can reason about, we model
1368  // its behavior explicitly.
1369  if (isMemFunction(FD, ASTC))
1370    return true;
1371
1372  // If it's not a system call, assume it frees memory.
1373  if (!Call->isInSystemHeader())
1374    return false;
1375
1376  // White list the system functions whose arguments escape.
1377  const IdentifierInfo *II = FD->getIdentifier();
1378  if (!II)
1379    return false;
1380  StringRef FName = II->getName();
1381
1382  // White list the 'XXXNoCopy' CoreFoundation functions.
1383  // We specifically check these before
1384  if (FName.endswith("NoCopy")) {
1385    // Look for the deallocator argument. We know that the memory ownership
1386    // is not transferred only if the deallocator argument is
1387    // 'kCFAllocatorNull'.
1388    for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1389      const Expr *ArgE = Call->getArgExpr(i)->IgnoreParenCasts();
1390      if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1391        StringRef DeallocatorName = DE->getFoundDecl()->getName();
1392        if (DeallocatorName == "kCFAllocatorNull")
1393          return true;
1394      }
1395    }
1396    return false;
1397  }
1398
1399  // Associating streams with malloced buffers. The pointer can escape if
1400  // 'closefn' is specified (and if that function does free memory),
1401  // but it will not if closefn is not specified.
1402  // Currently, we do not inspect the 'closefn' function (PR12101).
1403  if (FName == "funopen")
1404    if (Call->getNumArgs() >= 4 && Call->getArgSVal(4).isConstant(0))
1405      return true;
1406
1407  // Do not warn on pointers passed to 'setbuf' when used with std streams,
1408  // these leaks might be intentional when setting the buffer for stdio.
1409  // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1410  if (FName == "setbuf" || FName =="setbuffer" ||
1411      FName == "setlinebuf" || FName == "setvbuf") {
1412    if (Call->getNumArgs() >= 1) {
1413      const Expr *ArgE = Call->getArgExpr(0)->IgnoreParenCasts();
1414      if (const DeclRefExpr *ArgDRE = dyn_cast<DeclRefExpr>(ArgE))
1415        if (const VarDecl *D = dyn_cast<VarDecl>(ArgDRE->getDecl()))
1416          if (D->getCanonicalDecl()->getName().find("std") != StringRef::npos)
1417            return false;
1418    }
1419  }
1420
1421  // A bunch of other functions which either take ownership of a pointer or
1422  // wrap the result up in a struct or object, meaning it can be freed later.
1423  // (See RetainCountChecker.) Not all the parameters here are invalidated,
1424  // but the Malloc checker cannot differentiate between them. The right way
1425  // of doing this would be to implement a pointer escapes callback.
1426  if (FName == "CGBitmapContextCreate" ||
1427      FName == "CGBitmapContextCreateWithData" ||
1428      FName == "CVPixelBufferCreateWithBytes" ||
1429      FName == "CVPixelBufferCreateWithPlanarBytes" ||
1430      FName == "OSAtomicEnqueue") {
1431    return false;
1432  }
1433
1434  // Handle cases where we know a buffer's /address/ can escape.
1435  // Note that the above checks handle some special cases where we know that
1436  // even though the address escapes, it's still our responsibility to free the
1437  // buffer.
1438  if (Call->argumentsMayEscape())
1439    return false;
1440
1441  // Otherwise, assume that the function does not free memory.
1442  // Most system calls do not free the memory.
1443  return true;
1444}
1445
1446// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1447// escapes, when we are tracking p), do not track the symbol as we cannot reason
1448// about it anymore.
1449ProgramStateRef
1450MallocChecker::checkRegionChanges(ProgramStateRef State,
1451                            const StoreManager::InvalidatedSymbols *invalidated,
1452                                    ArrayRef<const MemRegion *> ExplicitRegions,
1453                                    ArrayRef<const MemRegion *> Regions,
1454                                    const CallEvent *Call) const {
1455  if (!invalidated || invalidated->empty())
1456    return State;
1457  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1458
1459  // If it's a call which might free or reallocate memory, we assume that all
1460  // regions (explicit and implicit) escaped.
1461
1462  // Otherwise, whitelist explicit pointers; we still can track them.
1463  if (!Call || doesNotFreeMemory(Call, State)) {
1464    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1465        E = ExplicitRegions.end(); I != E; ++I) {
1466      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1467        WhitelistedSymbols.insert(R->getSymbol());
1468    }
1469  }
1470
1471  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1472       E = invalidated->end(); I!=E; ++I) {
1473    SymbolRef sym = *I;
1474    if (WhitelistedSymbols.count(sym))
1475      continue;
1476    // The symbol escaped. Note, we assume that if the symbol is released,
1477    // passing it out will result in a use after free. We also keep tracking
1478    // relinquished symbols.
1479    if (const RefState *RS = State->get<RegionState>(sym)) {
1480      if (RS->isAllocated())
1481        State = State->set<RegionState>(sym,
1482                                        RefState::getEscaped(RS->getStmt()));
1483    }
1484  }
1485  return State;
1486}
1487
1488static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1489                                         ProgramStateRef prevState) {
1490  ReallocMap currMap = currState->get<ReallocPairs>();
1491  ReallocMap prevMap = prevState->get<ReallocPairs>();
1492
1493  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1494       I != E; ++I) {
1495    SymbolRef sym = I.getKey();
1496    if (!currMap.lookup(sym))
1497      return sym;
1498  }
1499
1500  return NULL;
1501}
1502
1503PathDiagnosticPiece *
1504MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1505                                           const ExplodedNode *PrevN,
1506                                           BugReporterContext &BRC,
1507                                           BugReport &BR) {
1508  ProgramStateRef state = N->getState();
1509  ProgramStateRef statePrev = PrevN->getState();
1510
1511  const RefState *RS = state->get<RegionState>(Sym);
1512  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1513  if (!RS)
1514    return 0;
1515
1516  const Stmt *S = 0;
1517  const char *Msg = 0;
1518  StackHintGeneratorForSymbol *StackHint = 0;
1519
1520  // Retrieve the associated statement.
1521  ProgramPoint ProgLoc = N->getLocation();
1522  if (StmtPoint *SP = dyn_cast<StmtPoint>(&ProgLoc))
1523    S = SP->getStmt();
1524  else if (CallExitEnd *Exit = dyn_cast<CallExitEnd>(&ProgLoc))
1525    S = Exit->getCalleeContext()->getCallSite();
1526  // If an assumption was made on a branch, it should be caught
1527  // here by looking at the state transition.
1528  else if (BlockEdge *Edge = dyn_cast<BlockEdge>(&ProgLoc)) {
1529    const CFGBlock *srcBlk = Edge->getSrc();
1530    S = srcBlk->getTerminator();
1531  }
1532  if (!S)
1533    return 0;
1534
1535  // FIXME: We will eventually need to handle non-statement-based events
1536  // (__attribute__((cleanup))).
1537
1538  // Find out if this is an interesting point and what is the kind.
1539  if (Mode == Normal) {
1540    if (isAllocated(RS, RSPrev, S)) {
1541      Msg = "Memory is allocated";
1542      StackHint = new StackHintGeneratorForSymbol(Sym,
1543                                                  "Returned allocated memory");
1544    } else if (isReleased(RS, RSPrev, S)) {
1545      Msg = "Memory is released";
1546      StackHint = new StackHintGeneratorForSymbol(Sym,
1547                                                  "Returned released memory");
1548    } else if (isRelinquished(RS, RSPrev, S)) {
1549      Msg = "Memory ownership is transfered";
1550      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1551    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1552      Mode = ReallocationFailed;
1553      Msg = "Reallocation failed";
1554      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1555                                                       "Reallocation failed");
1556
1557      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1558        // Is it possible to fail two reallocs WITHOUT testing in between?
1559        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1560          "We only support one failed realloc at a time.");
1561        BR.markInteresting(sym);
1562        FailedReallocSymbol = sym;
1563      }
1564    }
1565
1566  // We are in a special mode if a reallocation failed later in the path.
1567  } else if (Mode == ReallocationFailed) {
1568    assert(FailedReallocSymbol && "No symbol to look for.");
1569
1570    // Is this is the first appearance of the reallocated symbol?
1571    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1572      // If we ever hit this assert, that means BugReporter has decided to skip
1573      // node pairs or visit them out of order.
1574      assert(state->get<RegionState>(FailedReallocSymbol) &&
1575        "Missed the reallocation point");
1576
1577      // We're at the reallocation point.
1578      Msg = "Attempt to reallocate memory";
1579      StackHint = new StackHintGeneratorForSymbol(Sym,
1580                                                 "Returned reallocated memory");
1581      FailedReallocSymbol = NULL;
1582      Mode = Normal;
1583    }
1584  }
1585
1586  if (!Msg)
1587    return 0;
1588  assert(StackHint);
1589
1590  // Generate the extra diagnostic.
1591  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1592                             N->getLocationContext());
1593  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1594}
1595
1596void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1597                               const char *NL, const char *Sep) const {
1598
1599  RegionStateTy RS = State->get<RegionState>();
1600
1601  if (!RS.isEmpty())
1602    Out << "Has Malloc data" << NL;
1603}
1604
1605#define REGISTER_CHECKER(name) \
1606void ento::register##name(CheckerManager &mgr) {\
1607  registerCStringCheckerBasic(mgr); \
1608  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1609}
1610
1611REGISTER_CHECKER(MallocPessimistic)
1612REGISTER_CHECKER(MallocOptimistic)
1613