MallocChecker.cpp revision 7a95de68c093991047ed8d339479ccad51b88663 
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/AST/Attr.h"
18#include "clang/Basic/SourceManager.h"
19#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
20#include "clang/StaticAnalyzer/Core/Checker.h"
21#include "clang/StaticAnalyzer/Core/CheckerManager.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
24#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
25#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
26#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
27#include "llvm/ADT/ImmutableMap.h"
28#include "llvm/ADT/STLExtras.h"
29#include "llvm/ADT/SmallString.h"
30#include "llvm/ADT/StringExtras.h"
31#include <climits>
32
33using namespace clang;
34using namespace ento;
35
36namespace {
37
38class RefState {
39  enum Kind { // Reference to allocated memory.
40              Allocated,
41              // Reference to released/freed memory.
42              Released,
43              // The responsibility for freeing resources has transfered from
44              // this reference. A relinquished symbol should not be freed.
45              Relinquished } K;
46  const Stmt *S;
47
48public:
49  RefState(Kind k, const Stmt *s) : K(k), S(s) {}
50
51  bool isAllocated() const { return K == Allocated; }
52  bool isReleased() const { return K == Released; }
53  bool isRelinquished() const { return K == Relinquished; }
54
55  const Stmt *getStmt() const { return S; }
56
57  bool operator==(const RefState &X) const {
58    return K == X.K && S == X.S;
59  }
60
61  static RefState getAllocated(const Stmt *s) {
62    return RefState(Allocated, s);
63  }
64  static RefState getReleased(const Stmt *s) { return RefState(Released, s); }
65  static RefState getRelinquished(const Stmt *s) {
66    return RefState(Relinquished, s);
67  }
68
69  void Profile(llvm::FoldingSetNodeID &ID) const {
70    ID.AddInteger(K);
71    ID.AddPointer(S);
72  }
73
74  void dump(raw_ostream &OS) const {
75    static const char *Table[] = {
76      "Allocated",
77      "Released",
78      "Relinquished"
79    };
80    OS << Table[(unsigned) K];
81  }
82
83  LLVM_ATTRIBUTE_USED void dump() const {
84    dump(llvm::errs());
85  }
86};
87
88enum ReallocPairKind {
89  RPToBeFreedAfterFailure,
90  // The symbol has been freed when reallocation failed.
91  RPIsFreeOnFailure,
92  // The symbol does not need to be freed after reallocation fails.
93  RPDoNotTrackAfterFailure
94};
95
96/// \class ReallocPair
97/// \brief Stores information about the symbol being reallocated by a call to
98/// 'realloc' to allow modeling failed reallocation later in the path.
99struct ReallocPair {
100  // \brief The symbol which realloc reallocated.
101  SymbolRef ReallocatedSym;
102  ReallocPairKind Kind;
103
104  ReallocPair(SymbolRef S, ReallocPairKind K) :
105    ReallocatedSym(S), Kind(K) {}
106  void Profile(llvm::FoldingSetNodeID &ID) const {
107    ID.AddInteger(Kind);
108    ID.AddPointer(ReallocatedSym);
109  }
110  bool operator==(const ReallocPair &X) const {
111    return ReallocatedSym == X.ReallocatedSym &&
112           Kind == X.Kind;
113  }
114};
115
116typedef std::pair<const ExplodedNode*, const MemRegion*> LeakInfo;
117
118class MallocChecker : public Checker<check::DeadSymbols,
119                                     check::PointerEscape,
120                                     check::PreStmt<ReturnStmt>,
121                                     check::PreStmt<CallExpr>,
122                                     check::PostStmt<CallExpr>,
123                                     check::PostStmt<BlockExpr>,
124                                     check::PostObjCMessage,
125                                     check::Location,
126                                     eval::Assume>
127{
128  mutable OwningPtr<BugType> BT_DoubleFree;
129  mutable OwningPtr<BugType> BT_Leak;
130  mutable OwningPtr<BugType> BT_UseFree;
131  mutable OwningPtr<BugType> BT_BadFree;
132  mutable OwningPtr<BugType> BT_OffsetFree;
133  mutable IdentifierInfo *II_malloc, *II_free, *II_realloc, *II_calloc,
134                         *II_valloc, *II_reallocf, *II_strndup, *II_strdup;
135
136public:
137  MallocChecker() : II_malloc(0), II_free(0), II_realloc(0), II_calloc(0),
138                    II_valloc(0), II_reallocf(0), II_strndup(0), II_strdup(0) {}
139
140  /// In pessimistic mode, the checker assumes that it does not know which
141  /// functions might free the memory.
142  struct ChecksFilter {
143    DefaultBool CMallocPessimistic;
144    DefaultBool CMallocOptimistic;
145  };
146
147  ChecksFilter Filter;
148
149  void checkPreStmt(const CallExpr *S, CheckerContext &C) const;
150  void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
151  void checkPostObjCMessage(const ObjCMethodCall &Call, CheckerContext &C) const;
152  void checkPostStmt(const BlockExpr *BE, CheckerContext &C) const;
153  void checkDeadSymbols(SymbolReaper &SymReaper, CheckerContext &C) const;
154  void checkPreStmt(const ReturnStmt *S, CheckerContext &C) const;
155  ProgramStateRef evalAssume(ProgramStateRef state, SVal Cond,
156                            bool Assumption) const;
157  void checkLocation(SVal l, bool isLoad, const Stmt *S,
158                     CheckerContext &C) const;
159
160  ProgramStateRef checkPointerEscape(ProgramStateRef State,
161                                    const InvalidatedSymbols &Escaped,
162                                    const CallEvent *Call,
163                                    PointerEscapeKind Kind) const;
164
165  void printState(raw_ostream &Out, ProgramStateRef State,
166                  const char *NL, const char *Sep) const;
167
168private:
169  void initIdentifierInfo(ASTContext &C) const;
170
171  /// Check if this is one of the functions which can allocate/reallocate memory
172  /// pointed to by one of its arguments.
173  bool isMemFunction(const FunctionDecl *FD, ASTContext &C) const;
174  bool isFreeFunction(const FunctionDecl *FD, ASTContext &C) const;
175  bool isAllocationFunction(const FunctionDecl *FD, ASTContext &C) const;
176
177  static ProgramStateRef MallocMemReturnsAttr(CheckerContext &C,
178                                              const CallExpr *CE,
179                                              const OwnershipAttr* Att);
180  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
181                                     const Expr *SizeEx, SVal Init,
182                                     ProgramStateRef state) {
183    return MallocMemAux(C, CE,
184                        state->getSVal(SizeEx, C.getLocationContext()),
185                        Init, state);
186  }
187
188  static ProgramStateRef MallocMemAux(CheckerContext &C, const CallExpr *CE,
189                                     SVal SizeEx, SVal Init,
190                                     ProgramStateRef state);
191
192  /// Update the RefState to reflect the new memory allocation.
193  static ProgramStateRef MallocUpdateRefState(CheckerContext &C,
194                                              const CallExpr *CE,
195                                              ProgramStateRef state);
196
197  ProgramStateRef FreeMemAttr(CheckerContext &C, const CallExpr *CE,
198                              const OwnershipAttr* Att) const;
199  ProgramStateRef FreeMemAux(CheckerContext &C, const CallExpr *CE,
200                             ProgramStateRef state, unsigned Num,
201                             bool Hold,
202                             bool &ReleasedAllocated,
203                             bool ReturnsNullOnFailure = false) const;
204  ProgramStateRef FreeMemAux(CheckerContext &C, const Expr *Arg,
205                             const Expr *ParentExpr,
206                             ProgramStateRef State,
207                             bool Hold,
208                             bool &ReleasedAllocated,
209                             bool ReturnsNullOnFailure = false) const;
210
211  ProgramStateRef ReallocMem(CheckerContext &C, const CallExpr *CE,
212                             bool FreesMemOnFailure) const;
213  static ProgramStateRef CallocMem(CheckerContext &C, const CallExpr *CE);
214
215  ///\brief Check if the memory associated with this symbol was released.
216  bool isReleased(SymbolRef Sym, CheckerContext &C) const;
217
218  bool checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
219                         const Stmt *S = 0) const;
220
221  /// Check if the function is not known to us. So, for example, we could
222  /// conservatively assume it can free/reallocate it's pointer arguments.
223  bool doesNotFreeMemory(const CallEvent *Call,
224                         ProgramStateRef State) const;
225
226  static bool SummarizeValue(raw_ostream &os, SVal V);
227  static bool SummarizeRegion(raw_ostream &os, const MemRegion *MR);
228  void ReportBadFree(CheckerContext &C, SVal ArgVal, SourceRange range) const;
229  void ReportOffsetFree(CheckerContext &C, SVal ArgVal, SourceRange Range)const;
230
231  /// Find the location of the allocation for Sym on the path leading to the
232  /// exploded node N.
233  LeakInfo getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
234                             CheckerContext &C) const;
235
236  void reportLeak(SymbolRef Sym, ExplodedNode *N, CheckerContext &C) const;
237
238  /// The bug visitor which allows us to print extra diagnostics along the
239  /// BugReport path. For example, showing the allocation site of the leaked
240  /// region.
241  class MallocBugVisitor : public BugReporterVisitorImpl<MallocBugVisitor> {
242  protected:
243    enum NotificationMode {
244      Normal,
245      ReallocationFailed
246    };
247
248    // The allocated region symbol tracked by the main analysis.
249    SymbolRef Sym;
250
251    // The mode we are in, i.e. what kind of diagnostics will be emitted.
252    NotificationMode Mode;
253
254    // A symbol from when the primary region should have been reallocated.
255    SymbolRef FailedReallocSymbol;
256
257    bool IsLeak;
258
259  public:
260    MallocBugVisitor(SymbolRef S, bool isLeak = false)
261       : Sym(S), Mode(Normal), FailedReallocSymbol(0), IsLeak(isLeak) {}
262
263    virtual ~MallocBugVisitor() {}
264
265    void Profile(llvm::FoldingSetNodeID &ID) const {
266      static int X = 0;
267      ID.AddPointer(&X);
268      ID.AddPointer(Sym);
269    }
270
271    inline bool isAllocated(const RefState *S, const RefState *SPrev,
272                            const Stmt *Stmt) {
273      // Did not track -> allocated. Other state (released) -> allocated.
274      return (Stmt && isa<CallExpr>(Stmt) &&
275              (S && S->isAllocated()) && (!SPrev || !SPrev->isAllocated()));
276    }
277
278    inline bool isReleased(const RefState *S, const RefState *SPrev,
279                           const Stmt *Stmt) {
280      // Did not track -> released. Other state (allocated) -> released.
281      return (Stmt && isa<CallExpr>(Stmt) &&
282              (S && S->isReleased()) && (!SPrev || !SPrev->isReleased()));
283    }
284
285    inline bool isRelinquished(const RefState *S, const RefState *SPrev,
286                               const Stmt *Stmt) {
287      // Did not track -> relinquished. Other state (allocated) -> relinquished.
288      return (Stmt && (isa<CallExpr>(Stmt) || isa<ObjCMessageExpr>(Stmt) ||
289                                              isa<ObjCPropertyRefExpr>(Stmt)) &&
290              (S && S->isRelinquished()) &&
291              (!SPrev || !SPrev->isRelinquished()));
292    }
293
294    inline bool isReallocFailedCheck(const RefState *S, const RefState *SPrev,
295                                     const Stmt *Stmt) {
296      // If the expression is not a call, and the state change is
297      // released -> allocated, it must be the realloc return value
298      // check. If we have to handle more cases here, it might be cleaner just
299      // to track this extra bit in the state itself.
300      return ((!Stmt || !isa<CallExpr>(Stmt)) &&
301              (S && S->isAllocated()) && (SPrev && !SPrev->isAllocated()));
302    }
303
304    PathDiagnosticPiece *VisitNode(const ExplodedNode *N,
305                                   const ExplodedNode *PrevN,
306                                   BugReporterContext &BRC,
307                                   BugReport &BR);
308
309    PathDiagnosticPiece* getEndPath(BugReporterContext &BRC,
310                                    const ExplodedNode *EndPathNode,
311                                    BugReport &BR) {
312      if (!IsLeak)
313        return 0;
314
315      PathDiagnosticLocation L =
316        PathDiagnosticLocation::createEndOfPath(EndPathNode,
317                                                BRC.getSourceManager());
318      // Do not add the statement itself as a range in case of leak.
319      return new PathDiagnosticEventPiece(L, BR.getDescription(), false);
320    }
321
322  private:
323    class StackHintGeneratorForReallocationFailed
324        : public StackHintGeneratorForSymbol {
325    public:
326      StackHintGeneratorForReallocationFailed(SymbolRef S, StringRef M)
327        : StackHintGeneratorForSymbol(S, M) {}
328
329      virtual std::string getMessageForArg(const Expr *ArgE, unsigned ArgIndex) {
330        // Printed parameters start at 1, not 0.
331        ++ArgIndex;
332
333        SmallString<200> buf;
334        llvm::raw_svector_ostream os(buf);
335
336        os << "Reallocation of " << ArgIndex << llvm::getOrdinalSuffix(ArgIndex)
337           << " parameter failed";
338
339        return os.str();
340      }
341
342      virtual std::string getMessageForReturn(const CallExpr *CallExpr) {
343        return "Reallocation of returned value failed";
344      }
345    };
346  };
347};
348} // end anonymous namespace
349
350REGISTER_MAP_WITH_PROGRAMSTATE(RegionState, SymbolRef, RefState)
351REGISTER_MAP_WITH_PROGRAMSTATE(ReallocPairs, SymbolRef, ReallocPair)
352
353// A map from the freed symbol to the symbol representing the return value of
354// the free function.
355REGISTER_MAP_WITH_PROGRAMSTATE(FreeReturnValue, SymbolRef, SymbolRef)
356
357namespace {
358class StopTrackingCallback : public SymbolVisitor {
359  ProgramStateRef state;
360public:
361  StopTrackingCallback(ProgramStateRef st) : state(st) {}
362  ProgramStateRef getState() const { return state; }
363
364  bool VisitSymbol(SymbolRef sym) {
365    state = state->remove<RegionState>(sym);
366    return true;
367  }
368};
369} // end anonymous namespace
370
371void MallocChecker::initIdentifierInfo(ASTContext &Ctx) const {
372  if (II_malloc)
373    return;
374  II_malloc = &Ctx.Idents.get("malloc");
375  II_free = &Ctx.Idents.get("free");
376  II_realloc = &Ctx.Idents.get("realloc");
377  II_reallocf = &Ctx.Idents.get("reallocf");
378  II_calloc = &Ctx.Idents.get("calloc");
379  II_valloc = &Ctx.Idents.get("valloc");
380  II_strdup = &Ctx.Idents.get("strdup");
381  II_strndup = &Ctx.Idents.get("strndup");
382}
383
384bool MallocChecker::isMemFunction(const FunctionDecl *FD, ASTContext &C) const {
385  if (isFreeFunction(FD, C))
386    return true;
387
388  if (isAllocationFunction(FD, C))
389    return true;
390
391  return false;
392}
393
394bool MallocChecker::isAllocationFunction(const FunctionDecl *FD,
395                                         ASTContext &C) const {
396  if (!FD)
397    return false;
398
399  if (FD->getKind() == Decl::Function) {
400    IdentifierInfo *FunI = FD->getIdentifier();
401    initIdentifierInfo(C);
402
403    if (FunI == II_malloc || FunI == II_realloc ||
404        FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
405        FunI == II_strdup || FunI == II_strndup)
406      return true;
407  }
408
409  if (Filter.CMallocOptimistic && FD->hasAttrs())
410    for (specific_attr_iterator<OwnershipAttr>
411           i = FD->specific_attr_begin<OwnershipAttr>(),
412           e = FD->specific_attr_end<OwnershipAttr>();
413           i != e; ++i)
414      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
415        return true;
416  return false;
417}
418
419bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
420  if (!FD)
421    return false;
422
423  if (FD->getKind() == Decl::Function) {
424    IdentifierInfo *FunI = FD->getIdentifier();
425    initIdentifierInfo(C);
426
427    if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
428      return true;
429  }
430
431  if (Filter.CMallocOptimistic && FD->hasAttrs())
432    for (specific_attr_iterator<OwnershipAttr>
433           i = FD->specific_attr_begin<OwnershipAttr>(),
434           e = FD->specific_attr_end<OwnershipAttr>();
435           i != e; ++i)
436      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
437          (*i)->getOwnKind() == OwnershipAttr::Holds)
438        return true;
439  return false;
440}
441
442void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
443  if (C.wasInlined)
444    return;
445
446  const FunctionDecl *FD = C.getCalleeDecl(CE);
447  if (!FD)
448    return;
449
450  ProgramStateRef State = C.getState();
451  bool ReleasedAllocatedMemory = false;
452
453  if (FD->getKind() == Decl::Function) {
454    initIdentifierInfo(C.getASTContext());
455    IdentifierInfo *FunI = FD->getIdentifier();
456
457    if (FunI == II_malloc || FunI == II_valloc) {
458      if (CE->getNumArgs() < 1)
459        return;
460      State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
461    } else if (FunI == II_realloc) {
462      State = ReallocMem(C, CE, false);
463    } else if (FunI == II_reallocf) {
464      State = ReallocMem(C, CE, true);
465    } else if (FunI == II_calloc) {
466      State = CallocMem(C, CE);
467    } else if (FunI == II_free) {
468      State = FreeMemAux(C, CE, State, 0, false, ReleasedAllocatedMemory);
469    } else if (FunI == II_strdup) {
470      State = MallocUpdateRefState(C, CE, State);
471    } else if (FunI == II_strndup) {
472      State = MallocUpdateRefState(C, CE, State);
473    }
474  }
475
476  if (Filter.CMallocOptimistic) {
477    // Check all the attributes, if there are any.
478    // There can be multiple of these attributes.
479    if (FD->hasAttrs())
480      for (specific_attr_iterator<OwnershipAttr>
481          i = FD->specific_attr_begin<OwnershipAttr>(),
482          e = FD->specific_attr_end<OwnershipAttr>();
483          i != e; ++i) {
484        switch ((*i)->getOwnKind()) {
485        case OwnershipAttr::Returns:
486          State = MallocMemReturnsAttr(C, CE, *i);
487          break;
488        case OwnershipAttr::Takes:
489        case OwnershipAttr::Holds:
490          State = FreeMemAttr(C, CE, *i);
491          break;
492        }
493      }
494  }
495  C.addTransition(State);
496}
497
498static bool isFreeWhenDoneSetToZero(const ObjCMethodCall &Call) {
499  Selector S = Call.getSelector();
500  for (unsigned i = 1; i < S.getNumArgs(); ++i)
501    if (S.getNameForSlot(i).equals("freeWhenDone"))
502      if (Call.getArgSVal(i).isConstant(0))
503        return true;
504
505  return false;
506}
507
508void MallocChecker::checkPostObjCMessage(const ObjCMethodCall &Call,
509                                         CheckerContext &C) const {
510  if (C.wasInlined)
511    return;
512
513  // If the first selector is dataWithBytesNoCopy, assume that the memory will
514  // be released with 'free' by the new object.
515  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
516  // Unless 'freeWhenDone' param set to 0.
517  // TODO: Check that the memory was allocated with malloc.
518  bool ReleasedAllocatedMemory = false;
519  Selector S = Call.getSelector();
520  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
521       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
522       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
523      !isFreeWhenDoneSetToZero(Call)){
524    unsigned int argIdx  = 0;
525    ProgramStateRef State = FreeMemAux(C, Call.getArgExpr(argIdx),
526                                       Call.getOriginExpr(), C.getState(), true,
527                                       ReleasedAllocatedMemory,
528                                       /* RetNullOnFailure*/ true);
529
530    C.addTransition(State);
531  }
532}
533
534ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
535                                                    const CallExpr *CE,
536                                                    const OwnershipAttr* Att) {
537  if (Att->getModule() != "malloc")
538    return 0;
539
540  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
541  if (I != E) {
542    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
543  }
544  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
545}
546
547ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
548                                           const CallExpr *CE,
549                                           SVal Size, SVal Init,
550                                           ProgramStateRef state) {
551
552  // Bind the return value to the symbolic value from the heap region.
553  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
554  // side effects other than what we model here.
555  unsigned Count = C.blockCount();
556  SValBuilder &svalBuilder = C.getSValBuilder();
557  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
558  DefinedSVal RetVal = svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count)
559      .castAs<DefinedSVal>();
560  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
561
562  // We expect the malloc functions to return a pointer.
563  if (!RetVal.getAs<Loc>())
564    return 0;
565
566  // Fill the region with the initialization value.
567  state = state->bindDefault(RetVal, Init);
568
569  // Set the region's extent equal to the Size parameter.
570  const SymbolicRegion *R =
571      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
572  if (!R)
573    return 0;
574  if (Optional<DefinedOrUnknownSVal> DefinedSize =
575          Size.getAs<DefinedOrUnknownSVal>()) {
576    SValBuilder &svalBuilder = C.getSValBuilder();
577    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
578    DefinedOrUnknownSVal extentMatchesSize =
579        svalBuilder.evalEQ(state, Extent, *DefinedSize);
580
581    state = state->assume(extentMatchesSize, true);
582    assert(state);
583  }
584
585  return MallocUpdateRefState(C, CE, state);
586}
587
588ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
589                                                    const CallExpr *CE,
590                                                    ProgramStateRef state) {
591  // Get the return value.
592  SVal retVal = state->getSVal(CE, C.getLocationContext());
593
594  // We expect the malloc functions to return a pointer.
595  if (!retVal.getAs<Loc>())
596    return 0;
597
598  SymbolRef Sym = retVal.getAsLocSymbol();
599  assert(Sym);
600
601  // Set the symbol's state to Allocated.
602  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
603
604}
605
606ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
607                                           const CallExpr *CE,
608                                           const OwnershipAttr* Att) const {
609  if (Att->getModule() != "malloc")
610    return 0;
611
612  ProgramStateRef State = C.getState();
613  bool ReleasedAllocated = false;
614
615  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
616       I != E; ++I) {
617    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
618                               Att->getOwnKind() == OwnershipAttr::Holds,
619                               ReleasedAllocated);
620    if (StateI)
621      State = StateI;
622  }
623  return State;
624}
625
626ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
627                                          const CallExpr *CE,
628                                          ProgramStateRef state,
629                                          unsigned Num,
630                                          bool Hold,
631                                          bool &ReleasedAllocated,
632                                          bool ReturnsNullOnFailure) const {
633  if (CE->getNumArgs() < (Num + 1))
634    return 0;
635
636  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold,
637                    ReleasedAllocated, ReturnsNullOnFailure);
638}
639
640/// Checks if the previous call to free on the given symbol failed - if free
641/// failed, returns true. Also, returns the corresponding return value symbol.
642static bool didPreviousFreeFail(ProgramStateRef State,
643                                SymbolRef Sym, SymbolRef &RetStatusSymbol) {
644  const SymbolRef *Ret = State->get<FreeReturnValue>(Sym);
645  if (Ret) {
646    assert(*Ret && "We should not store the null return symbol");
647    ConstraintManager &CMgr = State->getConstraintManager();
648    ConditionTruthVal FreeFailed = CMgr.isNull(State, *Ret);
649    RetStatusSymbol = *Ret;
650    return FreeFailed.isConstrainedTrue();
651  }
652  return false;
653}
654
655ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
656                                          const Expr *ArgExpr,
657                                          const Expr *ParentExpr,
658                                          ProgramStateRef State,
659                                          bool Hold,
660                                          bool &ReleasedAllocated,
661                                          bool ReturnsNullOnFailure) const {
662
663  SVal ArgVal = State->getSVal(ArgExpr, C.getLocationContext());
664  if (!ArgVal.getAs<DefinedOrUnknownSVal>())
665    return 0;
666  DefinedOrUnknownSVal location = ArgVal.castAs<DefinedOrUnknownSVal>();
667
668  // Check for null dereferences.
669  if (!location.getAs<Loc>())
670    return 0;
671
672  // The explicit NULL case, no operation is performed.
673  ProgramStateRef notNullState, nullState;
674  llvm::tie(notNullState, nullState) = State->assume(location);
675  if (nullState && !notNullState)
676    return 0;
677
678  // Unknown values could easily be okay
679  // Undefined values are handled elsewhere
680  if (ArgVal.isUnknownOrUndef())
681    return 0;
682
683  const MemRegion *R = ArgVal.getAsRegion();
684
685  // Nonlocs can't be freed, of course.
686  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
687  if (!R) {
688    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
689    return 0;
690  }
691
692  R = R->StripCasts();
693
694  // Blocks might show up as heap data, but should not be free()d
695  if (isa<BlockDataRegion>(R)) {
696    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
697    return 0;
698  }
699
700  const MemSpaceRegion *MS = R->getMemorySpace();
701
702  // Parameters, locals, statics, and globals shouldn't be freed.
703  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
704    // FIXME: at the time this code was written, malloc() regions were
705    // represented by conjured symbols, which are all in UnknownSpaceRegion.
706    // This means that there isn't actually anything from HeapSpaceRegion
707    // that should be freed, even though we allow it here.
708    // Of course, free() can work on memory allocated outside the current
709    // function, so UnknownSpaceRegion is always a possibility.
710    // False negatives are better than false positives.
711
712    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
713    return 0;
714  }
715
716  const SymbolicRegion *SrBase = dyn_cast<SymbolicRegion>(R->getBaseRegion());
717  // Various cases could lead to non-symbol values here.
718  // For now, ignore them.
719  if (!SrBase)
720    return 0;
721
722  SymbolRef SymBase = SrBase->getSymbol();
723  const RefState *RsBase = State->get<RegionState>(SymBase);
724  SymbolRef PreviousRetStatusSymbol = 0;
725
726  // Check double free.
727  if (RsBase &&
728      (RsBase->isReleased() || RsBase->isRelinquished()) &&
729      !didPreviousFreeFail(State, SymBase, PreviousRetStatusSymbol)) {
730
731    if (ExplodedNode *N = C.generateSink()) {
732      if (!BT_DoubleFree)
733        BT_DoubleFree.reset(
734          new BugType("Double free", "Memory Error"));
735      BugReport *R = new BugReport(*BT_DoubleFree,
736        (RsBase->isReleased() ? "Attempt to free released memory"
737                              : "Attempt to free non-owned memory"),
738        N);
739      R->addRange(ArgExpr->getSourceRange());
740      R->markInteresting(SymBase);
741      if (PreviousRetStatusSymbol)
742        R->markInteresting(PreviousRetStatusSymbol);
743      R->addVisitor(new MallocBugVisitor(SymBase));
744      C.emitReport(R);
745    }
746    return 0;
747  }
748
749  // Check if the memory location being freed is the actual location
750  // allocated, or an offset.
751  RegionOffset Offset = R->getAsOffset();
752  if (RsBase && RsBase->isAllocated() &&
753      Offset.isValid() &&
754      !Offset.hasSymbolicOffset() &&
755      Offset.getOffset() != 0) {
756    ReportOffsetFree(C, ArgVal, ArgExpr->getSourceRange());
757    return 0;
758  }
759
760  ReleasedAllocated = (RsBase != 0);
761
762  // Clean out the info on previous call to free return info.
763  State = State->remove<FreeReturnValue>(SymBase);
764
765  // Keep track of the return value. If it is NULL, we will know that free
766  // failed.
767  if (ReturnsNullOnFailure) {
768    SVal RetVal = C.getSVal(ParentExpr);
769    SymbolRef RetStatusSymbol = RetVal.getAsSymbol();
770    if (RetStatusSymbol) {
771      C.getSymbolManager().addSymbolDependency(SymBase, RetStatusSymbol);
772      State = State->set<FreeReturnValue>(SymBase, RetStatusSymbol);
773    }
774  }
775
776  // Normal free.
777  if (Hold) {
778    return State->set<RegionState>(SymBase,
779                                   RefState::getRelinquished(ParentExpr));
780  }
781  return State->set<RegionState>(SymBase, RefState::getReleased(ParentExpr));
782}
783
784bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
785  if (Optional<nonloc::ConcreteInt> IntVal = V.getAs<nonloc::ConcreteInt>())
786    os << "an integer (" << IntVal->getValue() << ")";
787  else if (Optional<loc::ConcreteInt> ConstAddr = V.getAs<loc::ConcreteInt>())
788    os << "a constant address (" << ConstAddr->getValue() << ")";
789  else if (Optional<loc::GotoLabel> Label = V.getAs<loc::GotoLabel>())
790    os << "the address of the label '" << Label->getLabel()->getName() << "'";
791  else
792    return false;
793
794  return true;
795}
796
797bool MallocChecker::SummarizeRegion(raw_ostream &os,
798                                    const MemRegion *MR) {
799  switch (MR->getKind()) {
800  case MemRegion::FunctionTextRegionKind: {
801    const NamedDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
802    if (FD)
803      os << "the address of the function '" << *FD << '\'';
804    else
805      os << "the address of a function";
806    return true;
807  }
808  case MemRegion::BlockTextRegionKind:
809    os << "block text";
810    return true;
811  case MemRegion::BlockDataRegionKind:
812    // FIXME: where the block came from?
813    os << "a block";
814    return true;
815  default: {
816    const MemSpaceRegion *MS = MR->getMemorySpace();
817
818    if (isa<StackLocalsSpaceRegion>(MS)) {
819      const VarRegion *VR = dyn_cast<VarRegion>(MR);
820      const VarDecl *VD;
821      if (VR)
822        VD = VR->getDecl();
823      else
824        VD = NULL;
825
826      if (VD)
827        os << "the address of the local variable '" << VD->getName() << "'";
828      else
829        os << "the address of a local stack variable";
830      return true;
831    }
832
833    if (isa<StackArgumentsSpaceRegion>(MS)) {
834      const VarRegion *VR = dyn_cast<VarRegion>(MR);
835      const VarDecl *VD;
836      if (VR)
837        VD = VR->getDecl();
838      else
839        VD = NULL;
840
841      if (VD)
842        os << "the address of the parameter '" << VD->getName() << "'";
843      else
844        os << "the address of a parameter";
845      return true;
846    }
847
848    if (isa<GlobalsSpaceRegion>(MS)) {
849      const VarRegion *VR = dyn_cast<VarRegion>(MR);
850      const VarDecl *VD;
851      if (VR)
852        VD = VR->getDecl();
853      else
854        VD = NULL;
855
856      if (VD) {
857        if (VD->isStaticLocal())
858          os << "the address of the static variable '" << VD->getName() << "'";
859        else
860          os << "the address of the global variable '" << VD->getName() << "'";
861      } else
862        os << "the address of a global variable";
863      return true;
864    }
865
866    return false;
867  }
868  }
869}
870
871void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
872                                  SourceRange range) const {
873  if (ExplodedNode *N = C.generateSink()) {
874    if (!BT_BadFree)
875      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
876
877    SmallString<100> buf;
878    llvm::raw_svector_ostream os(buf);
879
880    const MemRegion *MR = ArgVal.getAsRegion();
881    if (MR) {
882      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
883        MR = ER->getSuperRegion();
884
885      // Special case for alloca()
886      if (isa<AllocaRegion>(MR))
887        os << "Argument to free() was allocated by alloca(), not malloc()";
888      else {
889        os << "Argument to free() is ";
890        if (SummarizeRegion(os, MR))
891          os << ", which is not memory allocated by malloc()";
892        else
893          os << "not memory allocated by malloc()";
894      }
895    } else {
896      os << "Argument to free() is ";
897      if (SummarizeValue(os, ArgVal))
898        os << ", which is not memory allocated by malloc()";
899      else
900        os << "not memory allocated by malloc()";
901    }
902
903    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
904    R->markInteresting(MR);
905    R->addRange(range);
906    C.emitReport(R);
907  }
908}
909
910void MallocChecker::ReportOffsetFree(CheckerContext &C, SVal ArgVal,
911                                     SourceRange Range) const {
912  ExplodedNode *N = C.generateSink();
913  if (N == NULL)
914    return;
915
916  if (!BT_OffsetFree)
917    BT_OffsetFree.reset(new BugType("Offset free", "Memory Error"));
918
919  SmallString<100> buf;
920  llvm::raw_svector_ostream os(buf);
921
922  const MemRegion *MR = ArgVal.getAsRegion();
923  assert(MR && "Only MemRegion based symbols can have offset free errors");
924
925  RegionOffset Offset = MR->getAsOffset();
926  assert((Offset.isValid() &&
927          !Offset.hasSymbolicOffset() &&
928          Offset.getOffset() != 0) &&
929         "Only symbols with a valid offset can have offset free errors");
930
931  int offsetBytes = Offset.getOffset() / C.getASTContext().getCharWidth();
932
933  os << "Argument to free() is offset by "
934     << offsetBytes
935     << " "
936     << ((abs(offsetBytes) > 1) ? "bytes" : "byte")
937     << " from the start of memory allocated by malloc()";
938
939  BugReport *R = new BugReport(*BT_OffsetFree, os.str(), N);
940  R->markInteresting(MR->getBaseRegion());
941  R->addRange(Range);
942  C.emitReport(R);
943}
944
945ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
946                                          const CallExpr *CE,
947                                          bool FreesOnFail) const {
948  if (CE->getNumArgs() < 2)
949    return 0;
950
951  ProgramStateRef state = C.getState();
952  const Expr *arg0Expr = CE->getArg(0);
953  const LocationContext *LCtx = C.getLocationContext();
954  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
955  if (!Arg0Val.getAs<DefinedOrUnknownSVal>())
956    return 0;
957  DefinedOrUnknownSVal arg0Val = Arg0Val.castAs<DefinedOrUnknownSVal>();
958
959  SValBuilder &svalBuilder = C.getSValBuilder();
960
961  DefinedOrUnknownSVal PtrEQ =
962    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
963
964  // Get the size argument. If there is no size arg then give up.
965  const Expr *Arg1 = CE->getArg(1);
966  if (!Arg1)
967    return 0;
968
969  // Get the value of the size argument.
970  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
971  if (!Arg1ValG.getAs<DefinedOrUnknownSVal>())
972    return 0;
973  DefinedOrUnknownSVal Arg1Val = Arg1ValG.castAs<DefinedOrUnknownSVal>();
974
975  // Compare the size argument to 0.
976  DefinedOrUnknownSVal SizeZero =
977    svalBuilder.evalEQ(state, Arg1Val,
978                       svalBuilder.makeIntValWithPtrWidth(0, false));
979
980  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
981  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
982  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
983  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
984  // We only assume exceptional states if they are definitely true; if the
985  // state is under-constrained, assume regular realloc behavior.
986  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
987  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
988
989  // If the ptr is NULL and the size is not 0, the call is equivalent to
990  // malloc(size).
991  if ( PrtIsNull && !SizeIsZero) {
992    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
993                                               UndefinedVal(), StatePtrIsNull);
994    return stateMalloc;
995  }
996
997  if (PrtIsNull && SizeIsZero)
998    return 0;
999
1000  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
1001  assert(!PrtIsNull);
1002  SymbolRef FromPtr = arg0Val.getAsSymbol();
1003  SVal RetVal = state->getSVal(CE, LCtx);
1004  SymbolRef ToPtr = RetVal.getAsSymbol();
1005  if (!FromPtr || !ToPtr)
1006    return 0;
1007
1008  bool ReleasedAllocated = false;
1009
1010  // If the size is 0, free the memory.
1011  if (SizeIsZero)
1012    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero, 0,
1013                                               false, ReleasedAllocated)){
1014      // The semantics of the return value are:
1015      // If size was equal to 0, either NULL or a pointer suitable to be passed
1016      // to free() is returned. We just free the input pointer and do not add
1017      // any constrains on the output pointer.
1018      return stateFree;
1019    }
1020
1021  // Default behavior.
1022  if (ProgramStateRef stateFree =
1023        FreeMemAux(C, CE, state, 0, false, ReleasedAllocated)) {
1024
1025    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
1026                                                UnknownVal(), stateFree);
1027    if (!stateRealloc)
1028      return 0;
1029
1030    ReallocPairKind Kind = RPToBeFreedAfterFailure;
1031    if (FreesOnFail)
1032      Kind = RPIsFreeOnFailure;
1033    else if (!ReleasedAllocated)
1034      Kind = RPDoNotTrackAfterFailure;
1035
1036    // Record the info about the reallocated symbol so that we could properly
1037    // process failed reallocation.
1038    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
1039                                                   ReallocPair(FromPtr, Kind));
1040    // The reallocated symbol should stay alive for as long as the new symbol.
1041    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
1042    return stateRealloc;
1043  }
1044  return 0;
1045}
1046
1047ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
1048  if (CE->getNumArgs() < 2)
1049    return 0;
1050
1051  ProgramStateRef state = C.getState();
1052  SValBuilder &svalBuilder = C.getSValBuilder();
1053  const LocationContext *LCtx = C.getLocationContext();
1054  SVal count = state->getSVal(CE->getArg(0), LCtx);
1055  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
1056  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
1057                                        svalBuilder.getContext().getSizeType());
1058  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
1059
1060  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
1061}
1062
1063LeakInfo
1064MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
1065                                 CheckerContext &C) const {
1066  const LocationContext *LeakContext = N->getLocationContext();
1067  // Walk the ExplodedGraph backwards and find the first node that referred to
1068  // the tracked symbol.
1069  const ExplodedNode *AllocNode = N;
1070  const MemRegion *ReferenceRegion = 0;
1071
1072  while (N) {
1073    ProgramStateRef State = N->getState();
1074    if (!State->get<RegionState>(Sym))
1075      break;
1076
1077    // Find the most recent expression bound to the symbol in the current
1078    // context.
1079    if (!ReferenceRegion) {
1080      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
1081        SVal Val = State->getSVal(MR);
1082        if (Val.getAsLocSymbol() == Sym)
1083          ReferenceRegion = MR;
1084      }
1085    }
1086
1087    // Allocation node, is the last node in the current context in which the
1088    // symbol was tracked.
1089    if (N->getLocationContext() == LeakContext)
1090      AllocNode = N;
1091    N = N->pred_empty() ? NULL : *(N->pred_begin());
1092  }
1093
1094  return LeakInfo(AllocNode, ReferenceRegion);
1095}
1096
1097void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
1098                               CheckerContext &C) const {
1099  assert(N);
1100  if (!BT_Leak) {
1101    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
1102    // Leaks should not be reported if they are post-dominated by a sink:
1103    // (1) Sinks are higher importance bugs.
1104    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
1105    //     with __noreturn functions such as assert() or exit(). We choose not
1106    //     to report leaks on such paths.
1107    BT_Leak->setSuppressOnSink(true);
1108  }
1109
1110  // Most bug reports are cached at the location where they occurred.
1111  // With leaks, we want to unique them by the location where they were
1112  // allocated, and only report a single path.
1113  PathDiagnosticLocation LocUsedForUniqueing;
1114  const ExplodedNode *AllocNode = 0;
1115  const MemRegion *Region = 0;
1116  llvm::tie(AllocNode, Region) = getAllocationSite(N, Sym, C);
1117
1118  ProgramPoint P = AllocNode->getLocation();
1119  const Stmt *AllocationStmt = 0;
1120  if (Optional<CallExitEnd> Exit = P.getAs<CallExitEnd>())
1121    AllocationStmt = Exit->getCalleeContext()->getCallSite();
1122  else if (Optional<StmtPoint> SP = P.getAs<StmtPoint>())
1123    AllocationStmt = SP->getStmt();
1124  if (AllocationStmt)
1125    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocationStmt,
1126                                              C.getSourceManager(),
1127                                              AllocNode->getLocationContext());
1128
1129  SmallString<200> buf;
1130  llvm::raw_svector_ostream os(buf);
1131  os << "Memory is never released; potential leak";
1132  if (Region && Region->canPrintPretty()) {
1133    os << " of memory pointed to by '";
1134    Region->printPretty(os);
1135    os << '\'';
1136  }
1137
1138  BugReport *R = new BugReport(*BT_Leak, os.str(), N,
1139                               LocUsedForUniqueing,
1140                               AllocNode->getLocationContext()->getDecl());
1141  R->markInteresting(Sym);
1142  R->addVisitor(new MallocBugVisitor(Sym, true));
1143  C.emitReport(R);
1144}
1145
1146void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1147                                     CheckerContext &C) const
1148{
1149  if (!SymReaper.hasDeadSymbols())
1150    return;
1151
1152  ProgramStateRef state = C.getState();
1153  RegionStateTy RS = state->get<RegionState>();
1154  RegionStateTy::Factory &F = state->get_context<RegionState>();
1155
1156  SmallVector<SymbolRef, 2> Errors;
1157  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1158    if (SymReaper.isDead(I->first)) {
1159      if (I->second.isAllocated())
1160        Errors.push_back(I->first);
1161      // Remove the dead symbol from the map.
1162      RS = F.remove(RS, I->first);
1163
1164    }
1165  }
1166
1167  // Cleanup the Realloc Pairs Map.
1168  ReallocPairsTy RP = state->get<ReallocPairs>();
1169  for (ReallocPairsTy::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1170    if (SymReaper.isDead(I->first) ||
1171        SymReaper.isDead(I->second.ReallocatedSym)) {
1172      state = state->remove<ReallocPairs>(I->first);
1173    }
1174  }
1175
1176  // Cleanup the FreeReturnValue Map.
1177  FreeReturnValueTy FR = state->get<FreeReturnValue>();
1178  for (FreeReturnValueTy::iterator I = FR.begin(), E = FR.end(); I != E; ++I) {
1179    if (SymReaper.isDead(I->first) ||
1180        SymReaper.isDead(I->second)) {
1181      state = state->remove<FreeReturnValue>(I->first);
1182    }
1183  }
1184
1185  // Generate leak node.
1186  ExplodedNode *N = C.getPredecessor();
1187  if (!Errors.empty()) {
1188    static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1189    N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1190    for (SmallVector<SymbolRef, 2>::iterator
1191        I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1192      reportLeak(*I, N, C);
1193    }
1194  }
1195
1196  C.addTransition(state->set<RegionState>(RS), N);
1197}
1198
1199void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1200  // We will check for double free in the post visit.
1201  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1202    return;
1203
1204  // Check use after free, when a freed pointer is passed to a call.
1205  ProgramStateRef State = C.getState();
1206  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1207                                    E = CE->arg_end(); I != E; ++I) {
1208    const Expr *A = *I;
1209    if (A->getType().getTypePtr()->isAnyPointerType()) {
1210      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1211      if (!Sym)
1212        continue;
1213      if (checkUseAfterFree(Sym, C, A))
1214        return;
1215    }
1216  }
1217}
1218
1219void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1220  const Expr *E = S->getRetValue();
1221  if (!E)
1222    return;
1223
1224  // Check if we are returning a symbol.
1225  ProgramStateRef State = C.getState();
1226  SVal RetVal = State->getSVal(E, C.getLocationContext());
1227  SymbolRef Sym = RetVal.getAsSymbol();
1228  if (!Sym)
1229    // If we are returning a field of the allocated struct or an array element,
1230    // the callee could still free the memory.
1231    // TODO: This logic should be a part of generic symbol escape callback.
1232    if (const MemRegion *MR = RetVal.getAsRegion())
1233      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1234        if (const SymbolicRegion *BMR =
1235              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1236          Sym = BMR->getSymbol();
1237
1238  // Check if we are returning freed memory.
1239  if (Sym)
1240    checkUseAfterFree(Sym, C, E);
1241}
1242
1243// TODO: Blocks should be either inlined or should call invalidate regions
1244// upon invocation. After that's in place, special casing here will not be
1245// needed.
1246void MallocChecker::checkPostStmt(const BlockExpr *BE,
1247                                  CheckerContext &C) const {
1248
1249  // Scan the BlockDecRefExprs for any object the retain count checker
1250  // may be tracking.
1251  if (!BE->getBlockDecl()->hasCaptures())
1252    return;
1253
1254  ProgramStateRef state = C.getState();
1255  const BlockDataRegion *R =
1256    cast<BlockDataRegion>(state->getSVal(BE,
1257                                         C.getLocationContext()).getAsRegion());
1258
1259  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1260                                            E = R->referenced_vars_end();
1261
1262  if (I == E)
1263    return;
1264
1265  SmallVector<const MemRegion*, 10> Regions;
1266  const LocationContext *LC = C.getLocationContext();
1267  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1268
1269  for ( ; I != E; ++I) {
1270    const VarRegion *VR = I.getCapturedRegion();
1271    if (VR->getSuperRegion() == R) {
1272      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1273    }
1274    Regions.push_back(VR);
1275  }
1276
1277  state =
1278    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1279                                    Regions.data() + Regions.size()).getState();
1280  C.addTransition(state);
1281}
1282
1283bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1284  assert(Sym);
1285  const RefState *RS = C.getState()->get<RegionState>(Sym);
1286  return (RS && RS->isReleased());
1287}
1288
1289bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1290                                      const Stmt *S) const {
1291  if (isReleased(Sym, C)) {
1292    if (ExplodedNode *N = C.generateSink()) {
1293      if (!BT_UseFree)
1294        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1295
1296      BugReport *R = new BugReport(*BT_UseFree,
1297                                   "Use of memory after it is freed",N);
1298      if (S)
1299        R->addRange(S->getSourceRange());
1300      R->markInteresting(Sym);
1301      R->addVisitor(new MallocBugVisitor(Sym));
1302      C.emitReport(R);
1303      return true;
1304    }
1305  }
1306  return false;
1307}
1308
1309// Check if the location is a freed symbolic region.
1310void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1311                                  CheckerContext &C) const {
1312  SymbolRef Sym = l.getLocSymbolInBase();
1313  if (Sym)
1314    checkUseAfterFree(Sym, C, S);
1315}
1316
1317// If a symbolic region is assumed to NULL (or another constant), stop tracking
1318// it - assuming that allocation failed on this path.
1319ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1320                                              SVal Cond,
1321                                              bool Assumption) const {
1322  RegionStateTy RS = state->get<RegionState>();
1323  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1324    // If the symbol is assumed to be NULL, remove it from consideration.
1325    ConstraintManager &CMgr = state->getConstraintManager();
1326    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1327    if (AllocFailed.isConstrainedTrue())
1328      state = state->remove<RegionState>(I.getKey());
1329  }
1330
1331  // Realloc returns 0 when reallocation fails, which means that we should
1332  // restore the state of the pointer being reallocated.
1333  ReallocPairsTy RP = state->get<ReallocPairs>();
1334  for (ReallocPairsTy::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1335    // If the symbol is assumed to be NULL, remove it from consideration.
1336    ConstraintManager &CMgr = state->getConstraintManager();
1337    ConditionTruthVal AllocFailed = CMgr.isNull(state, I.getKey());
1338    if (!AllocFailed.isConstrainedTrue())
1339      continue;
1340
1341    SymbolRef ReallocSym = I.getData().ReallocatedSym;
1342    if (const RefState *RS = state->get<RegionState>(ReallocSym)) {
1343      if (RS->isReleased()) {
1344        if (I.getData().Kind == RPToBeFreedAfterFailure)
1345          state = state->set<RegionState>(ReallocSym,
1346              RefState::getAllocated(RS->getStmt()));
1347        else if (I.getData().Kind == RPDoNotTrackAfterFailure)
1348          state = state->remove<RegionState>(ReallocSym);
1349        else
1350          assert(I.getData().Kind == RPIsFreeOnFailure);
1351      }
1352    }
1353    state = state->remove<ReallocPairs>(I.getKey());
1354  }
1355
1356  return state;
1357}
1358
1359// Check if the function is known to us. So, for example, we could
1360// conservatively assume it can free/reallocate its pointer arguments.
1361// (We assume that the pointers cannot escape through calls to system
1362// functions not handled by this checker.)
1363bool MallocChecker::doesNotFreeMemory(const CallEvent *Call,
1364                                      ProgramStateRef State) const {
1365  assert(Call);
1366
1367  // For now, assume that any C++ call can free memory.
1368  // TODO: If we want to be more optimistic here, we'll need to make sure that
1369  // regions escape to C++ containers. They seem to do that even now, but for
1370  // mysterious reasons.
1371  if (!(isa<FunctionCall>(Call) || isa<ObjCMethodCall>(Call)))
1372    return false;
1373
1374  // Check Objective-C messages by selector name.
1375  if (const ObjCMethodCall *Msg = dyn_cast<ObjCMethodCall>(Call)) {
1376    // If it's not a framework call, or if it takes a callback, assume it
1377    // can free memory.
1378    if (!Call->isInSystemHeader() || Call->hasNonZeroCallbackArg())
1379      return false;
1380
1381    Selector S = Msg->getSelector();
1382
1383    // Whitelist the ObjC methods which do free memory.
1384    // - Anything containing 'freeWhenDone' param set to 1.
1385    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1386    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1387      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1388        if (Call->getArgSVal(i).isConstant(1))
1389          return false;
1390        else
1391          return true;
1392      }
1393    }
1394
1395    // If the first selector ends with NoCopy, assume that the ownership is
1396    // transferred as well.
1397    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1398    StringRef FirstSlot = S.getNameForSlot(0);
1399    if (FirstSlot.endswith("NoCopy"))
1400      return false;
1401
1402    // If the first selector starts with addPointer, insertPointer,
1403    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1404    // This is similar to C++ containers (vector); we still might want to check
1405    // that the pointers get freed by following the container itself.
1406    if (FirstSlot.startswith("addPointer") ||
1407        FirstSlot.startswith("insertPointer") ||
1408        FirstSlot.startswith("replacePointer")) {
1409      return false;
1410    }
1411
1412    // Otherwise, assume that the method does not free memory.
1413    // Most framework methods do not free memory.
1414    return true;
1415  }
1416
1417  // At this point the only thing left to handle is straight function calls.
1418  const FunctionDecl *FD = cast<FunctionCall>(Call)->getDecl();
1419  if (!FD)
1420    return false;
1421
1422  ASTContext &ASTC = State->getStateManager().getContext();
1423
1424  // If it's one of the allocation functions we can reason about, we model
1425  // its behavior explicitly.
1426  if (isMemFunction(FD, ASTC))
1427    return true;
1428
1429  // If it's not a system call, assume it frees memory.
1430  if (!Call->isInSystemHeader())
1431    return false;
1432
1433  // White list the system functions whose arguments escape.
1434  const IdentifierInfo *II = FD->getIdentifier();
1435  if (!II)
1436    return false;
1437  StringRef FName = II->getName();
1438
1439  // White list the 'XXXNoCopy' CoreFoundation functions.
1440  // We specifically check these before
1441  if (FName.endswith("NoCopy")) {
1442    // Look for the deallocator argument. We know that the memory ownership
1443    // is not transferred only if the deallocator argument is
1444    // 'kCFAllocatorNull'.
1445    for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1446      const Expr *ArgE = Call->getArgExpr(i)->IgnoreParenCasts();
1447      if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1448        StringRef DeallocatorName = DE->getFoundDecl()->getName();
1449        if (DeallocatorName == "kCFAllocatorNull")
1450          return true;
1451      }
1452    }
1453    return false;
1454  }
1455
1456  // Associating streams with malloced buffers. The pointer can escape if
1457  // 'closefn' is specified (and if that function does free memory),
1458  // but it will not if closefn is not specified.
1459  // Currently, we do not inspect the 'closefn' function (PR12101).
1460  if (FName == "funopen")
1461    if (Call->getNumArgs() >= 4 && Call->getArgSVal(4).isConstant(0))
1462      return true;
1463
1464  // Do not warn on pointers passed to 'setbuf' when used with std streams,
1465  // these leaks might be intentional when setting the buffer for stdio.
1466  // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1467  if (FName == "setbuf" || FName =="setbuffer" ||
1468      FName == "setlinebuf" || FName == "setvbuf") {
1469    if (Call->getNumArgs() >= 1) {
1470      const Expr *ArgE = Call->getArgExpr(0)->IgnoreParenCasts();
1471      if (const DeclRefExpr *ArgDRE = dyn_cast<DeclRefExpr>(ArgE))
1472        if (const VarDecl *D = dyn_cast<VarDecl>(ArgDRE->getDecl()))
1473          if (D->getCanonicalDecl()->getName().find("std") != StringRef::npos)
1474            return false;
1475    }
1476  }
1477
1478  // A bunch of other functions which either take ownership of a pointer or
1479  // wrap the result up in a struct or object, meaning it can be freed later.
1480  // (See RetainCountChecker.) Not all the parameters here are invalidated,
1481  // but the Malloc checker cannot differentiate between them. The right way
1482  // of doing this would be to implement a pointer escapes callback.
1483  if (FName == "CGBitmapContextCreate" ||
1484      FName == "CGBitmapContextCreateWithData" ||
1485      FName == "CVPixelBufferCreateWithBytes" ||
1486      FName == "CVPixelBufferCreateWithPlanarBytes" ||
1487      FName == "OSAtomicEnqueue") {
1488    return false;
1489  }
1490
1491  // Handle cases where we know a buffer's /address/ can escape.
1492  // Note that the above checks handle some special cases where we know that
1493  // even though the address escapes, it's still our responsibility to free the
1494  // buffer.
1495  if (Call->argumentsMayEscape())
1496    return false;
1497
1498  // Otherwise, assume that the function does not free memory.
1499  // Most system calls do not free the memory.
1500  return true;
1501}
1502
1503ProgramStateRef MallocChecker::checkPointerEscape(ProgramStateRef State,
1504                                             const InvalidatedSymbols &Escaped,
1505                                             const CallEvent *Call,
1506                                             PointerEscapeKind Kind) const {
1507  // If we know that the call does not free memory, keep tracking the top
1508  // level arguments.
1509  if ((Kind == PSK_DirectEscapeOnCall ||
1510       Kind == PSK_IndirectEscapeOnCall) &&
1511      doesNotFreeMemory(Call, State)) {
1512    return State;
1513  }
1514
1515  for (InvalidatedSymbols::const_iterator I = Escaped.begin(),
1516                                          E = Escaped.end();
1517                                          I != E; ++I) {
1518    SymbolRef sym = *I;
1519
1520    if (const RefState *RS = State->get<RegionState>(sym)) {
1521      if (RS->isAllocated())
1522        State = State->remove<RegionState>(sym);
1523    }
1524  }
1525  return State;
1526}
1527
1528static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1529                                         ProgramStateRef prevState) {
1530  ReallocPairsTy currMap = currState->get<ReallocPairs>();
1531  ReallocPairsTy prevMap = prevState->get<ReallocPairs>();
1532
1533  for (ReallocPairsTy::iterator I = prevMap.begin(), E = prevMap.end();
1534       I != E; ++I) {
1535    SymbolRef sym = I.getKey();
1536    if (!currMap.lookup(sym))
1537      return sym;
1538  }
1539
1540  return NULL;
1541}
1542
1543PathDiagnosticPiece *
1544MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1545                                           const ExplodedNode *PrevN,
1546                                           BugReporterContext &BRC,
1547                                           BugReport &BR) {
1548  ProgramStateRef state = N->getState();
1549  ProgramStateRef statePrev = PrevN->getState();
1550
1551  const RefState *RS = state->get<RegionState>(Sym);
1552  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1553  if (!RS)
1554    return 0;
1555
1556  const Stmt *S = 0;
1557  const char *Msg = 0;
1558  StackHintGeneratorForSymbol *StackHint = 0;
1559
1560  // Retrieve the associated statement.
1561  ProgramPoint ProgLoc = N->getLocation();
1562  if (Optional<StmtPoint> SP = ProgLoc.getAs<StmtPoint>()) {
1563    S = SP->getStmt();
1564  } else if (Optional<CallExitEnd> Exit = ProgLoc.getAs<CallExitEnd>()) {
1565    S = Exit->getCalleeContext()->getCallSite();
1566  } else if (Optional<BlockEdge> Edge = ProgLoc.getAs<BlockEdge>()) {
1567    // If an assumption was made on a branch, it should be caught
1568    // here by looking at the state transition.
1569    S = Edge->getSrc()->getTerminator();
1570  }
1571
1572  if (!S)
1573    return 0;
1574
1575  // FIXME: We will eventually need to handle non-statement-based events
1576  // (__attribute__((cleanup))).
1577
1578  // Find out if this is an interesting point and what is the kind.
1579  if (Mode == Normal) {
1580    if (isAllocated(RS, RSPrev, S)) {
1581      Msg = "Memory is allocated";
1582      StackHint = new StackHintGeneratorForSymbol(Sym,
1583                                                  "Returned allocated memory");
1584    } else if (isReleased(RS, RSPrev, S)) {
1585      Msg = "Memory is released";
1586      StackHint = new StackHintGeneratorForSymbol(Sym,
1587                                                  "Returned released memory");
1588    } else if (isRelinquished(RS, RSPrev, S)) {
1589      Msg = "Memory ownership is transfered";
1590      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1591    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1592      Mode = ReallocationFailed;
1593      Msg = "Reallocation failed";
1594      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1595                                                       "Reallocation failed");
1596
1597      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1598        // Is it possible to fail two reallocs WITHOUT testing in between?
1599        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1600          "We only support one failed realloc at a time.");
1601        BR.markInteresting(sym);
1602        FailedReallocSymbol = sym;
1603      }
1604    }
1605
1606  // We are in a special mode if a reallocation failed later in the path.
1607  } else if (Mode == ReallocationFailed) {
1608    assert(FailedReallocSymbol && "No symbol to look for.");
1609
1610    // Is this is the first appearance of the reallocated symbol?
1611    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1612      // We're at the reallocation point.
1613      Msg = "Attempt to reallocate memory";
1614      StackHint = new StackHintGeneratorForSymbol(Sym,
1615                                                 "Returned reallocated memory");
1616      FailedReallocSymbol = NULL;
1617      Mode = Normal;
1618    }
1619  }
1620
1621  if (!Msg)
1622    return 0;
1623  assert(StackHint);
1624
1625  // Generate the extra diagnostic.
1626  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1627                             N->getLocationContext());
1628  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1629}
1630
1631void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1632                               const char *NL, const char *Sep) const {
1633
1634  RegionStateTy RS = State->get<RegionState>();
1635
1636  if (!RS.isEmpty()) {
1637    Out << Sep << "MallocChecker:" << NL;
1638    for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1639      I.getKey()->dumpToStream(Out);
1640      Out << " : ";
1641      I.getData().dump(Out);
1642      Out << NL;
1643    }
1644  }
1645}
1646
1647#define REGISTER_CHECKER(name) \
1648void ento::register##name(CheckerManager &mgr) {\
1649  registerCStringCheckerBasic(mgr); \
1650  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1651}
1652
1653REGISTER_CHECKER(MallocPessimistic)
1654REGISTER_CHECKER(MallocOptimistic)
1655