MallocChecker.cpp revision 7186dc63094d3ba24e57e16a66a226d21448dd4f 
1//=== MallocChecker.cpp - A malloc/free checker -------------------*- C++ -*--//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file defines malloc/free checker, which checks for potential memory
11// leaks, double free, and use-after-free problems.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "InterCheckerAPI.h"
17#include "clang/StaticAnalyzer/Core/Checker.h"
18#include "clang/StaticAnalyzer/Core/CheckerManager.h"
19#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
20#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/ObjCMessage.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h"
23#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
24#include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
25#include "clang/Basic/SourceManager.h"
26#include "llvm/ADT/ImmutableMap.h"
27#include "llvm/ADT/SmallString.h"
28#include "llvm/ADT/STLExtras.h"
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 ObjCMessage &Msg, 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 CallOrObjCMessage *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 CallOrObjCMessage *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  IdentifierInfo *FunI = FD->getIdentifier();
391  if (!FunI)
392    return false;
393
394  initIdentifierInfo(C);
395
396  if (FunI == II_malloc || FunI == II_realloc ||
397      FunI == II_reallocf || FunI == II_calloc || FunI == II_valloc ||
398      FunI == II_strdup || FunI == II_strndup)
399    return true;
400
401  if (Filter.CMallocOptimistic && FD->hasAttrs())
402    for (specific_attr_iterator<OwnershipAttr>
403           i = FD->specific_attr_begin<OwnershipAttr>(),
404           e = FD->specific_attr_end<OwnershipAttr>();
405           i != e; ++i)
406      if ((*i)->getOwnKind() == OwnershipAttr::Returns)
407        return true;
408  return false;
409}
410
411bool MallocChecker::isFreeFunction(const FunctionDecl *FD, ASTContext &C) const {
412  if (!FD)
413    return false;
414
415  IdentifierInfo *FunI = FD->getIdentifier();
416  if (!FunI)
417    return false;
418
419  initIdentifierInfo(C);
420
421  if (FunI == II_free || FunI == II_realloc || FunI == II_reallocf)
422    return true;
423
424  if (Filter.CMallocOptimistic && FD->hasAttrs())
425    for (specific_attr_iterator<OwnershipAttr>
426           i = FD->specific_attr_begin<OwnershipAttr>(),
427           e = FD->specific_attr_end<OwnershipAttr>();
428           i != e; ++i)
429      if ((*i)->getOwnKind() == OwnershipAttr::Takes ||
430          (*i)->getOwnKind() == OwnershipAttr::Holds)
431        return true;
432  return false;
433}
434
435void MallocChecker::checkPostStmt(const CallExpr *CE, CheckerContext &C) const {
436  const FunctionDecl *FD = C.getCalleeDecl(CE);
437  if (!FD)
438    return;
439
440  initIdentifierInfo(C.getASTContext());
441  IdentifierInfo *FunI = FD->getIdentifier();
442  if (!FunI)
443    return;
444
445  ProgramStateRef State = C.getState();
446  if (FunI == II_malloc || FunI == II_valloc) {
447    if (CE->getNumArgs() < 1)
448      return;
449    State = MallocMemAux(C, CE, CE->getArg(0), UndefinedVal(), State);
450  } else if (FunI == II_realloc) {
451    State = ReallocMem(C, CE, false);
452  } else if (FunI == II_reallocf) {
453    State = ReallocMem(C, CE, true);
454  } else if (FunI == II_calloc) {
455    State = CallocMem(C, CE);
456  } else if (FunI == II_free) {
457    State = FreeMemAux(C, CE, C.getState(), 0, false);
458  } else if (FunI == II_strdup) {
459    State = MallocUpdateRefState(C, CE, State);
460  } else if (FunI == II_strndup) {
461    State = MallocUpdateRefState(C, CE, State);
462  } else if (Filter.CMallocOptimistic) {
463    // Check all the attributes, if there are any.
464    // There can be multiple of these attributes.
465    if (FD->hasAttrs())
466      for (specific_attr_iterator<OwnershipAttr>
467          i = FD->specific_attr_begin<OwnershipAttr>(),
468          e = FD->specific_attr_end<OwnershipAttr>();
469          i != e; ++i) {
470        switch ((*i)->getOwnKind()) {
471        case OwnershipAttr::Returns:
472          State = MallocMemReturnsAttr(C, CE, *i);
473          break;
474        case OwnershipAttr::Takes:
475        case OwnershipAttr::Holds:
476          State = FreeMemAttr(C, CE, *i);
477          break;
478        }
479      }
480  }
481  C.addTransition(State);
482}
483
484static bool isFreeWhenDoneSetToZero(CallOrObjCMessage Call, Selector &S) {
485  for (unsigned i = 1; i < S.getNumArgs(); ++i)
486    if (S.getNameForSlot(i).equals("freeWhenDone"))
487      if (Call.getArgSVal(i).isConstant(0))
488        return true;
489
490  return false;
491}
492
493void MallocChecker::checkPreObjCMessage(const ObjCMessage &Msg,
494                                         CheckerContext &C) const {
495  const ObjCMethodDecl *MD = Msg.getMethodDecl();
496  if (!MD)
497    return;
498
499  CallOrObjCMessage Call(Msg, C.getState(), C.getLocationContext());
500  Selector S = Msg.getSelector();
501
502  // If the first selector is dataWithBytesNoCopy, assume that the memory will
503  // be released with 'free' by the new object.
504  // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
505  // Unless 'freeWhenDone' param set to 0.
506  // TODO: Check that the memory was allocated with malloc.
507  if ((S.getNameForSlot(0) == "dataWithBytesNoCopy" ||
508       S.getNameForSlot(0) == "initWithBytesNoCopy" ||
509       S.getNameForSlot(0) == "initWithCharactersNoCopy") &&
510      !isFreeWhenDoneSetToZero(Call, S)){
511    unsigned int argIdx  = 0;
512    C.addTransition(FreeMemAux(C, Call.getArg(argIdx),
513                    Msg.getMessageExpr(), C.getState(), true));
514  }
515}
516
517ProgramStateRef MallocChecker::MallocMemReturnsAttr(CheckerContext &C,
518                                                    const CallExpr *CE,
519                                                    const OwnershipAttr* Att) {
520  if (Att->getModule() != "malloc")
521    return 0;
522
523  OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
524  if (I != E) {
525    return MallocMemAux(C, CE, CE->getArg(*I), UndefinedVal(), C.getState());
526  }
527  return MallocMemAux(C, CE, UnknownVal(), UndefinedVal(), C.getState());
528}
529
530ProgramStateRef MallocChecker::MallocMemAux(CheckerContext &C,
531                                           const CallExpr *CE,
532                                           SVal Size, SVal Init,
533                                           ProgramStateRef state) {
534
535  // Bind the return value to the symbolic value from the heap region.
536  // TODO: We could rewrite post visit to eval call; 'malloc' does not have
537  // side effects other than what we model here.
538  unsigned Count = C.getCurrentBlockCount();
539  SValBuilder &svalBuilder = C.getSValBuilder();
540  const LocationContext *LCtx = C.getPredecessor()->getLocationContext();
541  DefinedSVal RetVal =
542    cast<DefinedSVal>(svalBuilder.getConjuredHeapSymbolVal(CE, LCtx, Count));
543  state = state->BindExpr(CE, C.getLocationContext(), RetVal);
544
545  // We expect the malloc functions to return a pointer.
546  if (!isa<Loc>(RetVal))
547    return 0;
548
549  // Fill the region with the initialization value.
550  state = state->bindDefault(RetVal, Init);
551
552  // Set the region's extent equal to the Size parameter.
553  const SymbolicRegion *R =
554      dyn_cast_or_null<SymbolicRegion>(RetVal.getAsRegion());
555  if (!R)
556    return 0;
557  if (isa<DefinedOrUnknownSVal>(Size)) {
558    SValBuilder &svalBuilder = C.getSValBuilder();
559    DefinedOrUnknownSVal Extent = R->getExtent(svalBuilder);
560    DefinedOrUnknownSVal DefinedSize = cast<DefinedOrUnknownSVal>(Size);
561    DefinedOrUnknownSVal extentMatchesSize =
562        svalBuilder.evalEQ(state, Extent, DefinedSize);
563
564    state = state->assume(extentMatchesSize, true);
565    assert(state);
566  }
567
568  return MallocUpdateRefState(C, CE, state);
569}
570
571ProgramStateRef MallocChecker::MallocUpdateRefState(CheckerContext &C,
572                                                    const CallExpr *CE,
573                                                    ProgramStateRef state) {
574  // Get the return value.
575  SVal retVal = state->getSVal(CE, C.getLocationContext());
576
577  // We expect the malloc functions to return a pointer.
578  if (!isa<Loc>(retVal))
579    return 0;
580
581  SymbolRef Sym = retVal.getAsLocSymbol();
582  assert(Sym);
583
584  // Set the symbol's state to Allocated.
585  return state->set<RegionState>(Sym, RefState::getAllocated(CE));
586
587}
588
589ProgramStateRef MallocChecker::FreeMemAttr(CheckerContext &C,
590                                           const CallExpr *CE,
591                                           const OwnershipAttr* Att) const {
592  if (Att->getModule() != "malloc")
593    return 0;
594
595  ProgramStateRef State = C.getState();
596
597  for (OwnershipAttr::args_iterator I = Att->args_begin(), E = Att->args_end();
598       I != E; ++I) {
599    ProgramStateRef StateI = FreeMemAux(C, CE, State, *I,
600                               Att->getOwnKind() == OwnershipAttr::Holds);
601    if (StateI)
602      State = StateI;
603  }
604  return State;
605}
606
607ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
608                                          const CallExpr *CE,
609                                          ProgramStateRef state,
610                                          unsigned Num,
611                                          bool Hold) const {
612  if (CE->getNumArgs() < (Num + 1))
613    return 0;
614
615  return FreeMemAux(C, CE->getArg(Num), CE, state, Hold);
616}
617
618ProgramStateRef MallocChecker::FreeMemAux(CheckerContext &C,
619                                          const Expr *ArgExpr,
620                                          const Expr *ParentExpr,
621                                          ProgramStateRef state,
622                                          bool Hold) const {
623
624  SVal ArgVal = state->getSVal(ArgExpr, C.getLocationContext());
625  if (!isa<DefinedOrUnknownSVal>(ArgVal))
626    return 0;
627  DefinedOrUnknownSVal location = cast<DefinedOrUnknownSVal>(ArgVal);
628
629  // Check for null dereferences.
630  if (!isa<Loc>(location))
631    return 0;
632
633  // The explicit NULL case, no operation is performed.
634  ProgramStateRef notNullState, nullState;
635  llvm::tie(notNullState, nullState) = state->assume(location);
636  if (nullState && !notNullState)
637    return 0;
638
639  // Unknown values could easily be okay
640  // Undefined values are handled elsewhere
641  if (ArgVal.isUnknownOrUndef())
642    return 0;
643
644  const MemRegion *R = ArgVal.getAsRegion();
645
646  // Nonlocs can't be freed, of course.
647  // Non-region locations (labels and fixed addresses) also shouldn't be freed.
648  if (!R) {
649    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
650    return 0;
651  }
652
653  R = R->StripCasts();
654
655  // Blocks might show up as heap data, but should not be free()d
656  if (isa<BlockDataRegion>(R)) {
657    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
658    return 0;
659  }
660
661  const MemSpaceRegion *MS = R->getMemorySpace();
662
663  // Parameters, locals, statics, and globals shouldn't be freed.
664  if (!(isa<UnknownSpaceRegion>(MS) || isa<HeapSpaceRegion>(MS))) {
665    // FIXME: at the time this code was written, malloc() regions were
666    // represented by conjured symbols, which are all in UnknownSpaceRegion.
667    // This means that there isn't actually anything from HeapSpaceRegion
668    // that should be freed, even though we allow it here.
669    // Of course, free() can work on memory allocated outside the current
670    // function, so UnknownSpaceRegion is always a possibility.
671    // False negatives are better than false positives.
672
673    ReportBadFree(C, ArgVal, ArgExpr->getSourceRange());
674    return 0;
675  }
676
677  const SymbolicRegion *SR = dyn_cast<SymbolicRegion>(R);
678  // Various cases could lead to non-symbol values here.
679  // For now, ignore them.
680  if (!SR)
681    return 0;
682
683  SymbolRef Sym = SR->getSymbol();
684  const RefState *RS = state->get<RegionState>(Sym);
685
686  // If the symbol has not been tracked, return. This is possible when free() is
687  // called on a pointer that does not get its pointee directly from malloc().
688  // Full support of this requires inter-procedural analysis.
689  if (!RS)
690    return 0;
691
692  // Check double free.
693  if (RS->isReleased() || RS->isRelinquished()) {
694    if (ExplodedNode *N = C.generateSink()) {
695      if (!BT_DoubleFree)
696        BT_DoubleFree.reset(
697          new BugType("Double free", "Memory Error"));
698      BugReport *R = new BugReport(*BT_DoubleFree,
699        (RS->isReleased() ? "Attempt to free released memory" :
700                            "Attempt to free non-owned memory"), N);
701      R->addRange(ArgExpr->getSourceRange());
702      R->markInteresting(Sym);
703      R->addVisitor(new MallocBugVisitor(Sym));
704      C.EmitReport(R);
705    }
706    return 0;
707  }
708
709  // Normal free.
710  if (Hold)
711    return state->set<RegionState>(Sym, RefState::getRelinquished(ParentExpr));
712  return state->set<RegionState>(Sym, RefState::getReleased(ParentExpr));
713}
714
715bool MallocChecker::SummarizeValue(raw_ostream &os, SVal V) {
716  if (nonloc::ConcreteInt *IntVal = dyn_cast<nonloc::ConcreteInt>(&V))
717    os << "an integer (" << IntVal->getValue() << ")";
718  else if (loc::ConcreteInt *ConstAddr = dyn_cast<loc::ConcreteInt>(&V))
719    os << "a constant address (" << ConstAddr->getValue() << ")";
720  else if (loc::GotoLabel *Label = dyn_cast<loc::GotoLabel>(&V))
721    os << "the address of the label '" << Label->getLabel()->getName() << "'";
722  else
723    return false;
724
725  return true;
726}
727
728bool MallocChecker::SummarizeRegion(raw_ostream &os,
729                                    const MemRegion *MR) {
730  switch (MR->getKind()) {
731  case MemRegion::FunctionTextRegionKind: {
732    const FunctionDecl *FD = cast<FunctionTextRegion>(MR)->getDecl();
733    if (FD)
734      os << "the address of the function '" << *FD << '\'';
735    else
736      os << "the address of a function";
737    return true;
738  }
739  case MemRegion::BlockTextRegionKind:
740    os << "block text";
741    return true;
742  case MemRegion::BlockDataRegionKind:
743    // FIXME: where the block came from?
744    os << "a block";
745    return true;
746  default: {
747    const MemSpaceRegion *MS = MR->getMemorySpace();
748
749    if (isa<StackLocalsSpaceRegion>(MS)) {
750      const VarRegion *VR = dyn_cast<VarRegion>(MR);
751      const VarDecl *VD;
752      if (VR)
753        VD = VR->getDecl();
754      else
755        VD = NULL;
756
757      if (VD)
758        os << "the address of the local variable '" << VD->getName() << "'";
759      else
760        os << "the address of a local stack variable";
761      return true;
762    }
763
764    if (isa<StackArgumentsSpaceRegion>(MS)) {
765      const VarRegion *VR = dyn_cast<VarRegion>(MR);
766      const VarDecl *VD;
767      if (VR)
768        VD = VR->getDecl();
769      else
770        VD = NULL;
771
772      if (VD)
773        os << "the address of the parameter '" << VD->getName() << "'";
774      else
775        os << "the address of a parameter";
776      return true;
777    }
778
779    if (isa<GlobalsSpaceRegion>(MS)) {
780      const VarRegion *VR = dyn_cast<VarRegion>(MR);
781      const VarDecl *VD;
782      if (VR)
783        VD = VR->getDecl();
784      else
785        VD = NULL;
786
787      if (VD) {
788        if (VD->isStaticLocal())
789          os << "the address of the static variable '" << VD->getName() << "'";
790        else
791          os << "the address of the global variable '" << VD->getName() << "'";
792      } else
793        os << "the address of a global variable";
794      return true;
795    }
796
797    return false;
798  }
799  }
800}
801
802void MallocChecker::ReportBadFree(CheckerContext &C, SVal ArgVal,
803                                  SourceRange range) const {
804  if (ExplodedNode *N = C.generateSink()) {
805    if (!BT_BadFree)
806      BT_BadFree.reset(new BugType("Bad free", "Memory Error"));
807
808    SmallString<100> buf;
809    llvm::raw_svector_ostream os(buf);
810
811    const MemRegion *MR = ArgVal.getAsRegion();
812    if (MR) {
813      while (const ElementRegion *ER = dyn_cast<ElementRegion>(MR))
814        MR = ER->getSuperRegion();
815
816      // Special case for alloca()
817      if (isa<AllocaRegion>(MR))
818        os << "Argument to free() was allocated by alloca(), not malloc()";
819      else {
820        os << "Argument to free() is ";
821        if (SummarizeRegion(os, MR))
822          os << ", which is not memory allocated by malloc()";
823        else
824          os << "not memory allocated by malloc()";
825      }
826    } else {
827      os << "Argument to free() is ";
828      if (SummarizeValue(os, ArgVal))
829        os << ", which is not memory allocated by malloc()";
830      else
831        os << "not memory allocated by malloc()";
832    }
833
834    BugReport *R = new BugReport(*BT_BadFree, os.str(), N);
835    R->markInteresting(MR);
836    R->addRange(range);
837    C.EmitReport(R);
838  }
839}
840
841ProgramStateRef MallocChecker::ReallocMem(CheckerContext &C,
842                                          const CallExpr *CE,
843                                          bool FreesOnFail) const {
844  if (CE->getNumArgs() < 2)
845    return 0;
846
847  ProgramStateRef state = C.getState();
848  const Expr *arg0Expr = CE->getArg(0);
849  const LocationContext *LCtx = C.getLocationContext();
850  SVal Arg0Val = state->getSVal(arg0Expr, LCtx);
851  if (!isa<DefinedOrUnknownSVal>(Arg0Val))
852    return 0;
853  DefinedOrUnknownSVal arg0Val = cast<DefinedOrUnknownSVal>(Arg0Val);
854
855  SValBuilder &svalBuilder = C.getSValBuilder();
856
857  DefinedOrUnknownSVal PtrEQ =
858    svalBuilder.evalEQ(state, arg0Val, svalBuilder.makeNull());
859
860  // Get the size argument. If there is no size arg then give up.
861  const Expr *Arg1 = CE->getArg(1);
862  if (!Arg1)
863    return 0;
864
865  // Get the value of the size argument.
866  SVal Arg1ValG = state->getSVal(Arg1, LCtx);
867  if (!isa<DefinedOrUnknownSVal>(Arg1ValG))
868    return 0;
869  DefinedOrUnknownSVal Arg1Val = cast<DefinedOrUnknownSVal>(Arg1ValG);
870
871  // Compare the size argument to 0.
872  DefinedOrUnknownSVal SizeZero =
873    svalBuilder.evalEQ(state, Arg1Val,
874                       svalBuilder.makeIntValWithPtrWidth(0, false));
875
876  ProgramStateRef StatePtrIsNull, StatePtrNotNull;
877  llvm::tie(StatePtrIsNull, StatePtrNotNull) = state->assume(PtrEQ);
878  ProgramStateRef StateSizeIsZero, StateSizeNotZero;
879  llvm::tie(StateSizeIsZero, StateSizeNotZero) = state->assume(SizeZero);
880  // We only assume exceptional states if they are definitely true; if the
881  // state is under-constrained, assume regular realloc behavior.
882  bool PrtIsNull = StatePtrIsNull && !StatePtrNotNull;
883  bool SizeIsZero = StateSizeIsZero && !StateSizeNotZero;
884
885  // If the ptr is NULL and the size is not 0, the call is equivalent to
886  // malloc(size).
887  if ( PrtIsNull && !SizeIsZero) {
888    ProgramStateRef stateMalloc = MallocMemAux(C, CE, CE->getArg(1),
889                                               UndefinedVal(), StatePtrIsNull);
890    return stateMalloc;
891  }
892
893  if (PrtIsNull && SizeIsZero)
894    return 0;
895
896  // Get the from and to pointer symbols as in toPtr = realloc(fromPtr, size).
897  assert(!PrtIsNull);
898  SymbolRef FromPtr = arg0Val.getAsSymbol();
899  SVal RetVal = state->getSVal(CE, LCtx);
900  SymbolRef ToPtr = RetVal.getAsSymbol();
901  if (!FromPtr || !ToPtr)
902    return 0;
903
904  // If the size is 0, free the memory.
905  if (SizeIsZero)
906    if (ProgramStateRef stateFree = FreeMemAux(C, CE, StateSizeIsZero,0,false)){
907      // The semantics of the return value are:
908      // If size was equal to 0, either NULL or a pointer suitable to be passed
909      // to free() is returned.
910      stateFree = stateFree->set<ReallocPairs>(ToPtr,
911                                            ReallocPair(FromPtr, FreesOnFail));
912      C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
913      return stateFree;
914    }
915
916  // Default behavior.
917  if (ProgramStateRef stateFree = FreeMemAux(C, CE, state, 0, false)) {
918    // FIXME: We should copy the content of the original buffer.
919    ProgramStateRef stateRealloc = MallocMemAux(C, CE, CE->getArg(1),
920                                                UnknownVal(), stateFree);
921    if (!stateRealloc)
922      return 0;
923    stateRealloc = stateRealloc->set<ReallocPairs>(ToPtr,
924                                            ReallocPair(FromPtr, FreesOnFail));
925    C.getSymbolManager().addSymbolDependency(ToPtr, FromPtr);
926    return stateRealloc;
927  }
928  return 0;
929}
930
931ProgramStateRef MallocChecker::CallocMem(CheckerContext &C, const CallExpr *CE){
932  if (CE->getNumArgs() < 2)
933    return 0;
934
935  ProgramStateRef state = C.getState();
936  SValBuilder &svalBuilder = C.getSValBuilder();
937  const LocationContext *LCtx = C.getLocationContext();
938  SVal count = state->getSVal(CE->getArg(0), LCtx);
939  SVal elementSize = state->getSVal(CE->getArg(1), LCtx);
940  SVal TotalSize = svalBuilder.evalBinOp(state, BO_Mul, count, elementSize,
941                                        svalBuilder.getContext().getSizeType());
942  SVal zeroVal = svalBuilder.makeZeroVal(svalBuilder.getContext().CharTy);
943
944  return MallocMemAux(C, CE, TotalSize, zeroVal, state);
945}
946
947LeakInfo
948MallocChecker::getAllocationSite(const ExplodedNode *N, SymbolRef Sym,
949                                 CheckerContext &C) const {
950  const LocationContext *LeakContext = N->getLocationContext();
951  // Walk the ExplodedGraph backwards and find the first node that referred to
952  // the tracked symbol.
953  const ExplodedNode *AllocNode = N;
954  const MemRegion *ReferenceRegion = 0;
955
956  while (N) {
957    ProgramStateRef State = N->getState();
958    if (!State->get<RegionState>(Sym))
959      break;
960
961    // Find the most recent expression bound to the symbol in the current
962    // context.
963    if (!ReferenceRegion) {
964      if (const MemRegion *MR = C.getLocationRegionIfPostStore(N)) {
965        SVal Val = State->getSVal(MR);
966        if (Val.getAsLocSymbol() == Sym)
967          ReferenceRegion = MR;
968      }
969    }
970
971    // Allocation node, is the last node in the current context in which the
972    // symbol was tracked.
973    if (N->getLocationContext() == LeakContext)
974      AllocNode = N;
975    N = N->pred_empty() ? NULL : *(N->pred_begin());
976  }
977
978  ProgramPoint P = AllocNode->getLocation();
979  const Stmt *AllocationStmt = 0;
980  if (isa<StmtPoint>(P))
981    AllocationStmt = cast<StmtPoint>(P).getStmt();
982
983  return LeakInfo(AllocationStmt, ReferenceRegion);
984}
985
986void MallocChecker::reportLeak(SymbolRef Sym, ExplodedNode *N,
987                               CheckerContext &C) const {
988  assert(N);
989  if (!BT_Leak) {
990    BT_Leak.reset(new BugType("Memory leak", "Memory Error"));
991    // Leaks should not be reported if they are post-dominated by a sink:
992    // (1) Sinks are higher importance bugs.
993    // (2) NoReturnFunctionChecker uses sink nodes to represent paths ending
994    //     with __noreturn functions such as assert() or exit(). We choose not
995    //     to report leaks on such paths.
996    BT_Leak->setSuppressOnSink(true);
997  }
998
999  // Most bug reports are cached at the location where they occurred.
1000  // With leaks, we want to unique them by the location where they were
1001  // allocated, and only report a single path.
1002  PathDiagnosticLocation LocUsedForUniqueing;
1003  const Stmt *AllocStmt = 0;
1004  const MemRegion *Region = 0;
1005  llvm::tie(AllocStmt, Region) = getAllocationSite(N, Sym, C);
1006  if (AllocStmt)
1007    LocUsedForUniqueing = PathDiagnosticLocation::createBegin(AllocStmt,
1008                            C.getSourceManager(), N->getLocationContext());
1009
1010  SmallString<200> buf;
1011  llvm::raw_svector_ostream os(buf);
1012  os << "Memory is never released; potential leak";
1013  if (Region) {
1014    os << " of memory pointed to by '";
1015    Region->dumpPretty(os);
1016    os <<'\'';
1017  }
1018
1019  BugReport *R = new BugReport(*BT_Leak, os.str(), N, LocUsedForUniqueing);
1020  R->markInteresting(Sym);
1021  R->addVisitor(new MallocBugVisitor(Sym, true));
1022  C.EmitReport(R);
1023}
1024
1025void MallocChecker::checkDeadSymbols(SymbolReaper &SymReaper,
1026                                     CheckerContext &C) const
1027{
1028  if (!SymReaper.hasDeadSymbols())
1029    return;
1030
1031  ProgramStateRef state = C.getState();
1032  RegionStateTy RS = state->get<RegionState>();
1033  RegionStateTy::Factory &F = state->get_context<RegionState>();
1034
1035  bool generateReport = false;
1036  llvm::SmallVector<SymbolRef, 2> Errors;
1037  for (RegionStateTy::iterator I = RS.begin(), E = RS.end(); I != E; ++I) {
1038    if (SymReaper.isDead(I->first)) {
1039      if (I->second.isAllocated()) {
1040        generateReport = true;
1041        Errors.push_back(I->first);
1042      }
1043      // Remove the dead symbol from the map.
1044      RS = F.remove(RS, I->first);
1045
1046    }
1047  }
1048
1049  // Cleanup the Realloc Pairs Map.
1050  ReallocMap RP = state->get<ReallocPairs>();
1051  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1052    if (SymReaper.isDead(I->first) ||
1053        SymReaper.isDead(I->second.ReallocatedSym)) {
1054      state = state->remove<ReallocPairs>(I->first);
1055    }
1056  }
1057
1058  // Generate leak node.
1059  static SimpleProgramPointTag Tag("MallocChecker : DeadSymbolsLeak");
1060  ExplodedNode *N = C.addTransition(C.getState(), C.getPredecessor(), &Tag);
1061
1062  if (generateReport) {
1063    for (llvm::SmallVector<SymbolRef, 2>::iterator
1064         I = Errors.begin(), E = Errors.end(); I != E; ++I) {
1065      reportLeak(*I, N, C);
1066    }
1067  }
1068  C.addTransition(state->set<RegionState>(RS), N);
1069}
1070
1071void MallocChecker::checkEndPath(CheckerContext &C) const {
1072  ProgramStateRef state = C.getState();
1073  RegionStateTy M = state->get<RegionState>();
1074
1075  // If inside inlined call, skip it.
1076  if (C.getLocationContext()->getParent() != 0)
1077    return;
1078
1079  for (RegionStateTy::iterator I = M.begin(), E = M.end(); I != E; ++I) {
1080    RefState RS = I->second;
1081    if (RS.isAllocated()) {
1082      ExplodedNode *N = C.addTransition(state);
1083      if (N)
1084        reportLeak(I->first, N, C);
1085    }
1086  }
1087}
1088
1089bool MallocChecker::checkEscape(SymbolRef Sym, const Stmt *S,
1090                                CheckerContext &C) const {
1091  ProgramStateRef state = C.getState();
1092  const RefState *RS = state->get<RegionState>(Sym);
1093  if (!RS)
1094    return false;
1095
1096  if (RS->isAllocated()) {
1097    state = state->set<RegionState>(Sym, RefState::getEscaped(S));
1098    C.addTransition(state);
1099    return true;
1100  }
1101  return false;
1102}
1103
1104void MallocChecker::checkPreStmt(const CallExpr *CE, CheckerContext &C) const {
1105  // We will check for double free in the post visit.
1106  if (isFreeFunction(C.getCalleeDecl(CE), C.getASTContext()))
1107    return;
1108
1109  // Check use after free, when a freed pointer is passed to a call.
1110  ProgramStateRef State = C.getState();
1111  for (CallExpr::const_arg_iterator I = CE->arg_begin(),
1112                                    E = CE->arg_end(); I != E; ++I) {
1113    const Expr *A = *I;
1114    if (A->getType().getTypePtr()->isAnyPointerType()) {
1115      SymbolRef Sym = State->getSVal(A, C.getLocationContext()).getAsSymbol();
1116      if (!Sym)
1117        continue;
1118      if (checkUseAfterFree(Sym, C, A))
1119        return;
1120    }
1121  }
1122}
1123
1124void MallocChecker::checkPreStmt(const ReturnStmt *S, CheckerContext &C) const {
1125  const Expr *E = S->getRetValue();
1126  if (!E)
1127    return;
1128
1129  // Check if we are returning a symbol.
1130  SVal RetVal = C.getState()->getSVal(E, C.getLocationContext());
1131  SymbolRef Sym = RetVal.getAsSymbol();
1132  if (!Sym)
1133    // If we are returning a field of the allocated struct or an array element,
1134    // the callee could still free the memory.
1135    // TODO: This logic should be a part of generic symbol escape callback.
1136    if (const MemRegion *MR = RetVal.getAsRegion())
1137      if (isa<FieldRegion>(MR) || isa<ElementRegion>(MR))
1138        if (const SymbolicRegion *BMR =
1139              dyn_cast<SymbolicRegion>(MR->getBaseRegion()))
1140          Sym = BMR->getSymbol();
1141  if (!Sym)
1142    return;
1143
1144  // Check if we are returning freed memory.
1145  if (checkUseAfterFree(Sym, C, E))
1146    return;
1147
1148  // If this function body is not inlined, check if the symbol is escaping.
1149  if (C.getLocationContext()->getParent() == 0)
1150    checkEscape(Sym, E, C);
1151}
1152
1153// TODO: Blocks should be either inlined or should call invalidate regions
1154// upon invocation. After that's in place, special casing here will not be
1155// needed.
1156void MallocChecker::checkPostStmt(const BlockExpr *BE,
1157                                  CheckerContext &C) const {
1158
1159  // Scan the BlockDecRefExprs for any object the retain count checker
1160  // may be tracking.
1161  if (!BE->getBlockDecl()->hasCaptures())
1162    return;
1163
1164  ProgramStateRef state = C.getState();
1165  const BlockDataRegion *R =
1166    cast<BlockDataRegion>(state->getSVal(BE,
1167                                         C.getLocationContext()).getAsRegion());
1168
1169  BlockDataRegion::referenced_vars_iterator I = R->referenced_vars_begin(),
1170                                            E = R->referenced_vars_end();
1171
1172  if (I == E)
1173    return;
1174
1175  SmallVector<const MemRegion*, 10> Regions;
1176  const LocationContext *LC = C.getLocationContext();
1177  MemRegionManager &MemMgr = C.getSValBuilder().getRegionManager();
1178
1179  for ( ; I != E; ++I) {
1180    const VarRegion *VR = *I;
1181    if (VR->getSuperRegion() == R) {
1182      VR = MemMgr.getVarRegion(VR->getDecl(), LC);
1183    }
1184    Regions.push_back(VR);
1185  }
1186
1187  state =
1188    state->scanReachableSymbols<StopTrackingCallback>(Regions.data(),
1189                                    Regions.data() + Regions.size()).getState();
1190  C.addTransition(state);
1191}
1192
1193bool MallocChecker::isReleased(SymbolRef Sym, CheckerContext &C) const {
1194  assert(Sym);
1195  const RefState *RS = C.getState()->get<RegionState>(Sym);
1196  return (RS && RS->isReleased());
1197}
1198
1199bool MallocChecker::checkUseAfterFree(SymbolRef Sym, CheckerContext &C,
1200                                      const Stmt *S) const {
1201  if (isReleased(Sym, C)) {
1202    if (ExplodedNode *N = C.generateSink()) {
1203      if (!BT_UseFree)
1204        BT_UseFree.reset(new BugType("Use-after-free", "Memory Error"));
1205
1206      BugReport *R = new BugReport(*BT_UseFree,
1207                                   "Use of memory after it is freed",N);
1208      if (S)
1209        R->addRange(S->getSourceRange());
1210      R->markInteresting(Sym);
1211      R->addVisitor(new MallocBugVisitor(Sym));
1212      C.EmitReport(R);
1213      return true;
1214    }
1215  }
1216  return false;
1217}
1218
1219// Check if the location is a freed symbolic region.
1220void MallocChecker::checkLocation(SVal l, bool isLoad, const Stmt *S,
1221                                  CheckerContext &C) const {
1222  SymbolRef Sym = l.getLocSymbolInBase();
1223  if (Sym)
1224    checkUseAfterFree(Sym, C, S);
1225}
1226
1227//===----------------------------------------------------------------------===//
1228// Check various ways a symbol can be invalidated.
1229// TODO: This logic (the next 3 functions) is copied/similar to the
1230// RetainRelease checker. We might want to factor this out.
1231//===----------------------------------------------------------------------===//
1232
1233// Stop tracking symbols when a value escapes as a result of checkBind.
1234// A value escapes in three possible cases:
1235// (1) we are binding to something that is not a memory region.
1236// (2) we are binding to a memregion that does not have stack storage
1237// (3) we are binding to a memregion with stack storage that the store
1238//     does not understand.
1239void MallocChecker::checkBind(SVal loc, SVal val, const Stmt *S,
1240                              CheckerContext &C) const {
1241  // Are we storing to something that causes the value to "escape"?
1242  bool escapes = true;
1243  ProgramStateRef state = C.getState();
1244
1245  if (loc::MemRegionVal *regionLoc = dyn_cast<loc::MemRegionVal>(&loc)) {
1246    escapes = !regionLoc->getRegion()->hasStackStorage();
1247
1248    if (!escapes) {
1249      // To test (3), generate a new state with the binding added.  If it is
1250      // the same state, then it escapes (since the store cannot represent
1251      // the binding).
1252      // Do this only if we know that the store is not supposed to generate the
1253      // same state.
1254      SVal StoredVal = state->getSVal(regionLoc->getRegion());
1255      if (StoredVal != val)
1256        escapes = (state == (state->bindLoc(*regionLoc, val)));
1257    }
1258    if (!escapes) {
1259      // Case 4: We do not currently model what happens when a symbol is
1260      // assigned to a struct field, so be conservative here and let the symbol
1261      // go. TODO: This could definitely be improved upon.
1262      escapes = !isa<VarRegion>(regionLoc->getRegion());
1263    }
1264  }
1265
1266  // If our store can represent the binding and we aren't storing to something
1267  // that doesn't have local storage then just return and have the simulation
1268  // state continue as is.
1269  if (!escapes)
1270      return;
1271
1272  // Otherwise, find all symbols referenced by 'val' that we are tracking
1273  // and stop tracking them.
1274  state = state->scanReachableSymbols<StopTrackingCallback>(val).getState();
1275  C.addTransition(state);
1276}
1277
1278// If a symbolic region is assumed to NULL (or another constant), stop tracking
1279// it - assuming that allocation failed on this path.
1280ProgramStateRef MallocChecker::evalAssume(ProgramStateRef state,
1281                                              SVal Cond,
1282                                              bool Assumption) const {
1283  RegionStateTy RS = state->get<RegionState>();
1284  for (RegionStateTy::iterator I = RS.begin(), E = RS.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      state = state->remove<RegionState>(I.getKey());
1289  }
1290
1291  // Realloc returns 0 when reallocation fails, which means that we should
1292  // restore the state of the pointer being reallocated.
1293  ReallocMap RP = state->get<ReallocPairs>();
1294  for (ReallocMap::iterator I = RP.begin(), E = RP.end(); I != E; ++I) {
1295    // If the symbol is assumed to NULL or another constant, this will
1296    // return an APSInt*.
1297    if (state->getSymVal(I.getKey())) {
1298      SymbolRef ReallocSym = I.getData().ReallocatedSym;
1299      const RefState *RS = state->get<RegionState>(ReallocSym);
1300      if (RS) {
1301        if (RS->isReleased() && ! I.getData().IsFreeOnFailure)
1302          state = state->set<RegionState>(ReallocSym,
1303                             RefState::getAllocated(RS->getStmt()));
1304      }
1305      state = state->remove<ReallocPairs>(I.getKey());
1306    }
1307  }
1308
1309  return state;
1310}
1311
1312// Check if the function is known to us. So, for example, we could
1313// conservatively assume it can free/reallocate it's pointer arguments.
1314// (We assume that the pointers cannot escape through calls to system
1315// functions not handled by this checker.)
1316bool MallocChecker::doesNotFreeMemory(const CallOrObjCMessage *Call,
1317                                      ProgramStateRef State) const {
1318  if (!Call)
1319    return false;
1320
1321  // For now, assume that any C++ call can free memory.
1322  // TODO: If we want to be more optimistic here, we'll need to make sure that
1323  // regions escape to C++ containers. They seem to do that even now, but for
1324  // mysterious reasons.
1325  if (Call->isCXXCall())
1326    return false;
1327
1328  const Decl *D = Call->getDecl();
1329  if (!D)
1330    return false;
1331
1332  ASTContext &ASTC = State->getStateManager().getContext();
1333
1334  // If it's one of the allocation functions we can reason about, we model
1335  // its behavior explicitly.
1336  if (isa<FunctionDecl>(D) && isMemFunction(cast<FunctionDecl>(D), ASTC)) {
1337    return true;
1338  }
1339
1340  // If it's not a system call, assume it frees memory.
1341  SourceManager &SM = ASTC.getSourceManager();
1342  if (!SM.isInSystemHeader(D->getLocation()))
1343    return false;
1344
1345  // Process C/ObjC functions.
1346  if (const FunctionDecl *FD  = dyn_cast<FunctionDecl>(D)) {
1347    // White list the system functions whose arguments escape.
1348    const IdentifierInfo *II = FD->getIdentifier();
1349    if (!II)
1350      return true;
1351    StringRef FName = II->getName();
1352
1353    // White list thread local storage.
1354    if (FName.equals("pthread_setspecific"))
1355      return false;
1356
1357    // White list xpc connection context.
1358    // TODO: Ensure that the deallocation actually happens, need to reason
1359    // about "xpc_connection_set_finalizer_f".
1360    if (FName.equals("xpc_connection_set_context"))
1361      return false;
1362
1363    // White list the 'XXXNoCopy' ObjC functions.
1364    if (FName.endswith("NoCopy")) {
1365      // Look for the deallocator argument. We know that the memory ownership
1366      // is not transferred only if the deallocator argument is
1367      // 'kCFAllocatorNull'.
1368      for (unsigned i = 1; i < Call->getNumArgs(); ++i) {
1369        const Expr *ArgE = Call->getArg(i)->IgnoreParenCasts();
1370        if (const DeclRefExpr *DE = dyn_cast<DeclRefExpr>(ArgE)) {
1371          StringRef DeallocatorName = DE->getFoundDecl()->getName();
1372          if (DeallocatorName == "kCFAllocatorNull")
1373            return true;
1374        }
1375      }
1376      return false;
1377    }
1378
1379    // PR12101
1380    // Many CoreFoundation and CoreGraphics might allow a tracked object
1381    // to escape.
1382    if (Call->isCFCGAllowingEscape(FName))
1383      return false;
1384
1385    // Associating streams with malloced buffers. The pointer can escape if
1386    // 'closefn' is specified (and if that function does free memory).
1387    // Currently, we do not inspect the 'closefn' function (PR12101).
1388    if (FName == "funopen")
1389      if (Call->getNumArgs() >= 4 && !Call->getArgSVal(4).isConstant(0))
1390        return false;
1391
1392    // Do not warn on pointers passed to 'setbuf' when used with std streams,
1393    // these leaks might be intentional when setting the buffer for stdio.
1394    // http://stackoverflow.com/questions/2671151/who-frees-setvbuf-buffer
1395    if (FName == "setbuf" || FName =="setbuffer" ||
1396        FName == "setlinebuf" || FName == "setvbuf") {
1397      if (Call->getNumArgs() >= 1)
1398        if (const DeclRefExpr *Arg =
1399              dyn_cast<DeclRefExpr>(Call->getArg(0)->IgnoreParenCasts()))
1400          if (const VarDecl *D = dyn_cast<VarDecl>(Arg->getDecl()))
1401              if (D->getCanonicalDecl()->getName().find("std")
1402                                                   != StringRef::npos)
1403                return false;
1404    }
1405
1406    // A bunch of other functions which either take ownership of a pointer or
1407    // wrap the result up in a struct or object, meaning it can be freed later.
1408    // (See RetainCountChecker.) Not all the parameters here are invalidated,
1409    // but the Malloc checker cannot differentiate between them. The right way
1410    // of doing this would be to implement a pointer escapes callback.
1411    if (FName == "CGBitmapContextCreate" ||
1412        FName == "CGBitmapContextCreateWithData" ||
1413        FName == "CVPixelBufferCreateWithBytes" ||
1414        FName == "CVPixelBufferCreateWithPlanarBytes" ||
1415        FName == "OSAtomicEnqueue") {
1416      return false;
1417    }
1418
1419    // Whitelist NSXXInsertXX, for example NSMapInsertIfAbsent, since they can
1420    // be deallocated by NSMapRemove.
1421    if (FName.startswith("NS") && (FName.find("Insert") != StringRef::npos))
1422      return false;
1423
1424    // If the call has a callback as an argument, assume the memory
1425    // can be freed.
1426    if (Call->hasNonZeroCallbackArg())
1427      return false;
1428
1429    // Otherwise, assume that the function does not free memory.
1430    // Most system calls, do not free the memory.
1431    return true;
1432
1433  // Process ObjC functions.
1434  } else if (const ObjCMethodDecl * ObjCD = dyn_cast<ObjCMethodDecl>(D)) {
1435    Selector S = ObjCD->getSelector();
1436
1437    // White list the ObjC functions which do free memory.
1438    // - Anything containing 'freeWhenDone' param set to 1.
1439    //   Ex: dataWithBytesNoCopy:length:freeWhenDone.
1440    for (unsigned i = 1; i < S.getNumArgs(); ++i) {
1441      if (S.getNameForSlot(i).equals("freeWhenDone")) {
1442        if (Call->getArgSVal(i).isConstant(1))
1443          return false;
1444        else
1445          return true;
1446      }
1447    }
1448
1449    // If the first selector ends with NoCopy, assume that the ownership is
1450    // transferred as well.
1451    // Ex:  [NSData dataWithBytesNoCopy:bytes length:10];
1452    if (S.getNameForSlot(0).endswith("NoCopy")) {
1453      return false;
1454    }
1455
1456    // If the first selector starts with addPointer, insertPointer,
1457    // or replacePointer, assume we are dealing with NSPointerArray or similar.
1458    // This is similar to C++ containers (vector); we still might want to check
1459    // that the pointers get freed, by following the container itself.
1460    if (S.getNameForSlot(0).startswith("addPointer") ||
1461        S.getNameForSlot(0).startswith("insertPointer") ||
1462        S.getNameForSlot(0).startswith("replacePointer")) {
1463      return false;
1464    }
1465
1466    // If the call has a callback as an argument, assume the memory
1467    // can be freed.
1468    if (Call->hasNonZeroCallbackArg())
1469      return false;
1470
1471    // Otherwise, assume that the function does not free memory.
1472    // Most system calls, do not free the memory.
1473    return true;
1474  }
1475
1476  // Otherwise, assume that the function can free memory.
1477  return false;
1478
1479}
1480
1481// If the symbol we are tracking is invalidated, but not explicitly (ex: the &p
1482// escapes, when we are tracking p), do not track the symbol as we cannot reason
1483// about it anymore.
1484ProgramStateRef
1485MallocChecker::checkRegionChanges(ProgramStateRef State,
1486                            const StoreManager::InvalidatedSymbols *invalidated,
1487                                    ArrayRef<const MemRegion *> ExplicitRegions,
1488                                    ArrayRef<const MemRegion *> Regions,
1489                                    const CallOrObjCMessage *Call) const {
1490  if (!invalidated || invalidated->empty())
1491    return State;
1492  llvm::SmallPtrSet<SymbolRef, 8> WhitelistedSymbols;
1493
1494  // If it's a call which might free or reallocate memory, we assume that all
1495  // regions (explicit and implicit) escaped.
1496
1497  // Otherwise, whitelist explicit pointers; we still can track them.
1498  if (!Call || doesNotFreeMemory(Call, State)) {
1499    for (ArrayRef<const MemRegion *>::iterator I = ExplicitRegions.begin(),
1500        E = ExplicitRegions.end(); I != E; ++I) {
1501      if (const SymbolicRegion *R = (*I)->StripCasts()->getAs<SymbolicRegion>())
1502        WhitelistedSymbols.insert(R->getSymbol());
1503    }
1504  }
1505
1506  for (StoreManager::InvalidatedSymbols::const_iterator I=invalidated->begin(),
1507       E = invalidated->end(); I!=E; ++I) {
1508    SymbolRef sym = *I;
1509    if (WhitelistedSymbols.count(sym))
1510      continue;
1511    // The symbol escaped. Note, we assume that if the symbol is released,
1512    // passing it out will result in a use after free. We also keep tracking
1513    // relinquished symbols.
1514    if (const RefState *RS = State->get<RegionState>(sym)) {
1515      if (RS->isAllocated())
1516        State = State->set<RegionState>(sym,
1517                                        RefState::getEscaped(RS->getStmt()));
1518    }
1519  }
1520  return State;
1521}
1522
1523static SymbolRef findFailedReallocSymbol(ProgramStateRef currState,
1524                                         ProgramStateRef prevState) {
1525  ReallocMap currMap = currState->get<ReallocPairs>();
1526  ReallocMap prevMap = prevState->get<ReallocPairs>();
1527
1528  for (ReallocMap::iterator I = prevMap.begin(), E = prevMap.end();
1529       I != E; ++I) {
1530    SymbolRef sym = I.getKey();
1531    if (!currMap.lookup(sym))
1532      return sym;
1533  }
1534
1535  return NULL;
1536}
1537
1538PathDiagnosticPiece *
1539MallocChecker::MallocBugVisitor::VisitNode(const ExplodedNode *N,
1540                                           const ExplodedNode *PrevN,
1541                                           BugReporterContext &BRC,
1542                                           BugReport &BR) {
1543  ProgramStateRef state = N->getState();
1544  ProgramStateRef statePrev = PrevN->getState();
1545
1546  const RefState *RS = state->get<RegionState>(Sym);
1547  const RefState *RSPrev = statePrev->get<RegionState>(Sym);
1548  if (!RS && !RSPrev)
1549    return 0;
1550
1551  const Stmt *S = 0;
1552  const char *Msg = 0;
1553  StackHintGeneratorForSymbol *StackHint = 0;
1554
1555  // Retrieve the associated statement.
1556  ProgramPoint ProgLoc = N->getLocation();
1557  if (isa<StmtPoint>(ProgLoc))
1558    S = cast<StmtPoint>(ProgLoc).getStmt();
1559  // If an assumption was made on a branch, it should be caught
1560  // here by looking at the state transition.
1561  if (isa<BlockEdge>(ProgLoc)) {
1562    const CFGBlock *srcBlk = cast<BlockEdge>(ProgLoc).getSrc();
1563    S = srcBlk->getTerminator();
1564  }
1565  if (!S)
1566    return 0;
1567
1568  // Find out if this is an interesting point and what is the kind.
1569  if (Mode == Normal) {
1570    if (isAllocated(RS, RSPrev, S)) {
1571      Msg = "Memory is allocated";
1572      StackHint = new StackHintGeneratorForSymbol(Sym,
1573                                                  "Returned allocated memory");
1574    } else if (isReleased(RS, RSPrev, S)) {
1575      Msg = "Memory is released";
1576      StackHint = new StackHintGeneratorForSymbol(Sym,
1577                                                  "Returned released memory");
1578    } else if (isRelinquished(RS, RSPrev, S)) {
1579      Msg = "Memory ownership is transfered";
1580      StackHint = new StackHintGeneratorForSymbol(Sym, "");
1581    } else if (isReallocFailedCheck(RS, RSPrev, S)) {
1582      Mode = ReallocationFailed;
1583      Msg = "Reallocation failed";
1584      StackHint = new StackHintGeneratorForReallocationFailed(Sym,
1585                                                       "Reallocation failed");
1586
1587      if (SymbolRef sym = findFailedReallocSymbol(state, statePrev)) {
1588        // Is it possible to fail two reallocs WITHOUT testing in between?
1589        assert((!FailedReallocSymbol || FailedReallocSymbol == sym) &&
1590          "We only support one failed realloc at a time.");
1591        BR.markInteresting(sym);
1592        FailedReallocSymbol = sym;
1593      }
1594    }
1595
1596  // We are in a special mode if a reallocation failed later in the path.
1597  } else if (Mode == ReallocationFailed) {
1598    assert(FailedReallocSymbol && "No symbol to look for.");
1599
1600    // Is this is the first appearance of the reallocated symbol?
1601    if (!statePrev->get<RegionState>(FailedReallocSymbol)) {
1602      // If we ever hit this assert, that means BugReporter has decided to skip
1603      // node pairs or visit them out of order.
1604      assert(state->get<RegionState>(FailedReallocSymbol) &&
1605        "Missed the reallocation point");
1606
1607      // We're at the reallocation point.
1608      Msg = "Attempt to reallocate memory";
1609      StackHint = new StackHintGeneratorForSymbol(Sym,
1610                                                 "Returned reallocated memory");
1611      FailedReallocSymbol = NULL;
1612      Mode = Normal;
1613    }
1614  }
1615
1616  if (!Msg)
1617    return 0;
1618  assert(StackHint);
1619
1620  // Generate the extra diagnostic.
1621  PathDiagnosticLocation Pos(S, BRC.getSourceManager(),
1622                             N->getLocationContext());
1623  return new PathDiagnosticEventPiece(Pos, Msg, true, StackHint);
1624}
1625
1626void MallocChecker::printState(raw_ostream &Out, ProgramStateRef State,
1627                               const char *NL, const char *Sep) const {
1628
1629  RegionStateTy RS = State->get<RegionState>();
1630
1631  if (!RS.isEmpty())
1632    Out << "Has Malloc data" << NL;
1633}
1634
1635#define REGISTER_CHECKER(name) \
1636void ento::register##name(CheckerManager &mgr) {\
1637  registerCStringCheckerBasic(mgr); \
1638  mgr.registerChecker<MallocChecker>()->Filter.C##name = true;\
1639}
1640
1641REGISTER_CHECKER(MallocPessimistic)
1642REGISTER_CHECKER(MallocOptimistic)
1643