CallAndMessageChecker.cpp revision dc84cd5efdd3430efb22546b4ac656aa0540b210 
1//===--- CallAndMessageChecker.cpp ------------------------------*- 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 defines CallAndMessageChecker, a builtin checker that checks for various
11// errors of call and objc message expressions.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "clang/AST/ParentMap.h"
17#include "clang/Basic/TargetInfo.h"
18#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
19#include "clang/StaticAnalyzer/Core/Checker.h"
20#include "clang/StaticAnalyzer/Core/CheckerManager.h"
21#include "clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h"
22#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
23#include "llvm/ADT/SmallString.h"
24#include "llvm/Support/raw_ostream.h"
25
26using namespace clang;
27using namespace ento;
28
29namespace {
30class CallAndMessageChecker
31  : public Checker< check::PreStmt<CallExpr>, check::PreObjCMessage,
32                    check::PreCall > {
33  mutable OwningPtr<BugType> BT_call_null;
34  mutable OwningPtr<BugType> BT_call_undef;
35  mutable OwningPtr<BugType> BT_cxx_call_null;
36  mutable OwningPtr<BugType> BT_cxx_call_undef;
37  mutable OwningPtr<BugType> BT_call_arg;
38  mutable OwningPtr<BugType> BT_msg_undef;
39  mutable OwningPtr<BugType> BT_objc_prop_undef;
40  mutable OwningPtr<BugType> BT_objc_subscript_undef;
41  mutable OwningPtr<BugType> BT_msg_arg;
42  mutable OwningPtr<BugType> BT_msg_ret;
43public:
44
45  void checkPreStmt(const CallExpr *CE, CheckerContext &C) const;
46  void checkPreObjCMessage(const ObjCMethodCall &msg, CheckerContext &C) const;
47  void checkPreCall(const CallEvent &Call, CheckerContext &C) const;
48
49private:
50  static bool PreVisitProcessArg(CheckerContext &C, SVal V,
51                                 SourceRange argRange, const Expr *argEx,
52                                 bool IsFirstArgument, bool checkUninitFields,
53                                 const CallEvent &Call, OwningPtr<BugType> &BT);
54
55  static void emitBadCall(BugType *BT, CheckerContext &C, const Expr *BadE);
56  void emitNilReceiverBug(CheckerContext &C, const ObjCMethodCall &msg,
57                          ExplodedNode *N) const;
58
59  void HandleNilReceiver(CheckerContext &C,
60                         ProgramStateRef state,
61                         const ObjCMethodCall &msg) const;
62
63  static void LazyInit_BT(const char *desc, OwningPtr<BugType> &BT) {
64    if (!BT)
65      BT.reset(new BuiltinBug(desc));
66  }
67};
68} // end anonymous namespace
69
70void CallAndMessageChecker::emitBadCall(BugType *BT, CheckerContext &C,
71                                        const Expr *BadE) {
72  ExplodedNode *N = C.generateSink();
73  if (!N)
74    return;
75
76  BugReport *R = new BugReport(*BT, BT->getName(), N);
77  if (BadE) {
78    R->addRange(BadE->getSourceRange());
79    if (BadE->isGLValue())
80      BadE = bugreporter::getDerefExpr(BadE);
81    bugreporter::trackNullOrUndefValue(N, BadE, *R);
82  }
83  C.emitReport(R);
84}
85
86static StringRef describeUninitializedArgumentInCall(const CallEvent &Call,
87                                                     bool IsFirstArgument) {
88  switch (Call.getKind()) {
89  case CE_ObjCMessage: {
90    const ObjCMethodCall &Msg = cast<ObjCMethodCall>(Call);
91    switch (Msg.getMessageKind()) {
92    case OCM_Message:
93      return "Argument in message expression is an uninitialized value";
94    case OCM_PropertyAccess:
95      assert(Msg.isSetter() && "Getters have no args");
96      return "Argument for property setter is an uninitialized value";
97    case OCM_Subscript:
98      if (Msg.isSetter() && IsFirstArgument)
99        return "Argument for subscript setter is an uninitialized value";
100      return "Subscript index is an uninitialized value";
101    }
102    llvm_unreachable("Unknown message kind.");
103  }
104  case CE_Block:
105    return "Block call argument is an uninitialized value";
106  default:
107    return "Function call argument is an uninitialized value";
108  }
109}
110
111bool CallAndMessageChecker::PreVisitProcessArg(CheckerContext &C,
112                                               SVal V, SourceRange argRange,
113                                               const Expr *argEx,
114                                               bool IsFirstArgument,
115                                               bool checkUninitFields,
116                                               const CallEvent &Call,
117                                               OwningPtr<BugType> &BT) {
118  if (V.isUndef()) {
119    if (ExplodedNode *N = C.generateSink()) {
120      LazyInit_BT("Uninitialized argument value", BT);
121
122      // Generate a report for this bug.
123      StringRef Desc = describeUninitializedArgumentInCall(Call,
124                                                           IsFirstArgument);
125      BugReport *R = new BugReport(*BT, Desc, N);
126      R->addRange(argRange);
127      if (argEx)
128        bugreporter::trackNullOrUndefValue(N, argEx, *R);
129      C.emitReport(R);
130    }
131    return true;
132  }
133
134  if (!checkUninitFields)
135    return false;
136
137  if (Optional<nonloc::LazyCompoundVal> LV =
138          V.getAs<nonloc::LazyCompoundVal>()) {
139
140    class FindUninitializedField {
141    public:
142      SmallVector<const FieldDecl *, 10> FieldChain;
143    private:
144      StoreManager &StoreMgr;
145      MemRegionManager &MrMgr;
146      Store store;
147    public:
148      FindUninitializedField(StoreManager &storeMgr,
149                             MemRegionManager &mrMgr, Store s)
150      : StoreMgr(storeMgr), MrMgr(mrMgr), store(s) {}
151
152      bool Find(const TypedValueRegion *R) {
153        QualType T = R->getValueType();
154        if (const RecordType *RT = T->getAsStructureType()) {
155          const RecordDecl *RD = RT->getDecl()->getDefinition();
156          assert(RD && "Referred record has no definition");
157          for (RecordDecl::field_iterator I =
158               RD->field_begin(), E = RD->field_end(); I!=E; ++I) {
159            const FieldRegion *FR = MrMgr.getFieldRegion(*I, R);
160            FieldChain.push_back(*I);
161            T = I->getType();
162            if (T->getAsStructureType()) {
163              if (Find(FR))
164                return true;
165            }
166            else {
167              const SVal &V = StoreMgr.getBinding(store, loc::MemRegionVal(FR));
168              if (V.isUndef())
169                return true;
170            }
171            FieldChain.pop_back();
172          }
173        }
174
175        return false;
176      }
177    };
178
179    const LazyCompoundValData *D = LV->getCVData();
180    FindUninitializedField F(C.getState()->getStateManager().getStoreManager(),
181                             C.getSValBuilder().getRegionManager(),
182                             D->getStore());
183
184    if (F.Find(D->getRegion())) {
185      if (ExplodedNode *N = C.generateSink()) {
186        LazyInit_BT("Uninitialized argument value", BT);
187        SmallString<512> Str;
188        llvm::raw_svector_ostream os(Str);
189        os << "Passed-by-value struct argument contains uninitialized data";
190
191        if (F.FieldChain.size() == 1)
192          os << " (e.g., field: '" << *F.FieldChain[0] << "')";
193        else {
194          os << " (e.g., via the field chain: '";
195          bool first = true;
196          for (SmallVectorImpl<const FieldDecl *>::iterator
197               DI = F.FieldChain.begin(), DE = F.FieldChain.end(); DI!=DE;++DI){
198            if (first)
199              first = false;
200            else
201              os << '.';
202            os << **DI;
203          }
204          os << "')";
205        }
206
207        // Generate a report for this bug.
208        BugReport *R = new BugReport(*BT, os.str(), N);
209        R->addRange(argRange);
210
211        // FIXME: enhance track back for uninitialized value for arbitrary
212        // memregions
213        C.emitReport(R);
214      }
215      return true;
216    }
217  }
218
219  return false;
220}
221
222void CallAndMessageChecker::checkPreStmt(const CallExpr *CE,
223                                         CheckerContext &C) const{
224
225  const Expr *Callee = CE->getCallee()->IgnoreParens();
226  ProgramStateRef State = C.getState();
227  const LocationContext *LCtx = C.getLocationContext();
228  SVal L = State->getSVal(Callee, LCtx);
229
230  if (L.isUndef()) {
231    if (!BT_call_undef)
232      BT_call_undef.reset(new BuiltinBug("Called function pointer is an "
233                                         "uninitalized pointer value"));
234    emitBadCall(BT_call_undef.get(), C, Callee);
235    return;
236  }
237
238  ProgramStateRef StNonNull, StNull;
239  llvm::tie(StNonNull, StNull) =
240      State->assume(L.castAs<DefinedOrUnknownSVal>());
241
242  if (StNull && !StNonNull) {
243    if (!BT_call_null)
244      BT_call_null.reset(
245        new BuiltinBug("Called function pointer is null (null dereference)"));
246    emitBadCall(BT_call_null.get(), C, Callee);
247  }
248
249  C.addTransition(StNonNull);
250}
251
252void CallAndMessageChecker::checkPreCall(const CallEvent &Call,
253                                         CheckerContext &C) const {
254  ProgramStateRef State = C.getState();
255
256  // If this is a call to a C++ method, check if the callee is null or
257  // undefined.
258  if (const CXXInstanceCall *CC = dyn_cast<CXXInstanceCall>(&Call)) {
259    SVal V = CC->getCXXThisVal();
260    if (V.isUndef()) {
261      if (!BT_cxx_call_undef)
262        BT_cxx_call_undef.reset(new BuiltinBug("Called C++ object pointer is "
263                                               "uninitialized"));
264      emitBadCall(BT_cxx_call_undef.get(), C, CC->getCXXThisExpr());
265      return;
266    }
267
268    ProgramStateRef StNonNull, StNull;
269    llvm::tie(StNonNull, StNull) =
270        State->assume(V.castAs<DefinedOrUnknownSVal>());
271
272    if (StNull && !StNonNull) {
273      if (!BT_cxx_call_null)
274        BT_cxx_call_null.reset(new BuiltinBug("Called C++ object pointer "
275                                              "is null"));
276      emitBadCall(BT_cxx_call_null.get(), C, CC->getCXXThisExpr());
277      return;
278    }
279
280    State = StNonNull;
281  }
282
283  // Don't check for uninitialized field values in arguments if the
284  // caller has a body that is available and we have the chance to inline it.
285  // This is a hack, but is a reasonable compromise betweens sometimes warning
286  // and sometimes not depending on if we decide to inline a function.
287  const Decl *D = Call.getDecl();
288  const bool checkUninitFields =
289    !(C.getAnalysisManager().shouldInlineCall() && (D && D->getBody()));
290
291  OwningPtr<BugType> *BT;
292  if (isa<ObjCMethodCall>(Call))
293    BT = &BT_msg_arg;
294  else
295    BT = &BT_call_arg;
296
297  for (unsigned i = 0, e = Call.getNumArgs(); i != e; ++i)
298    if (PreVisitProcessArg(C, Call.getArgSVal(i), Call.getArgSourceRange(i),
299                           Call.getArgExpr(i), /*IsFirstArgument=*/i == 0,
300                           checkUninitFields, Call, *BT))
301      return;
302
303  // If we make it here, record our assumptions about the callee.
304  C.addTransition(State);
305}
306
307void CallAndMessageChecker::checkPreObjCMessage(const ObjCMethodCall &msg,
308                                                CheckerContext &C) const {
309  SVal recVal = msg.getReceiverSVal();
310  if (recVal.isUndef()) {
311    if (ExplodedNode *N = C.generateSink()) {
312      BugType *BT = 0;
313      switch (msg.getMessageKind()) {
314      case OCM_Message:
315        if (!BT_msg_undef)
316          BT_msg_undef.reset(new BuiltinBug("Receiver in message expression "
317                                            "is an uninitialized value"));
318        BT = BT_msg_undef.get();
319        break;
320      case OCM_PropertyAccess:
321        if (!BT_objc_prop_undef)
322          BT_objc_prop_undef.reset(new BuiltinBug("Property access on an "
323                                                  "uninitialized object "
324                                                  "pointer"));
325        BT = BT_objc_prop_undef.get();
326        break;
327      case OCM_Subscript:
328        if (!BT_objc_subscript_undef)
329          BT_objc_subscript_undef.reset(new BuiltinBug("Subscript access on an "
330                                                       "uninitialized object "
331                                                       "pointer"));
332        BT = BT_objc_subscript_undef.get();
333        break;
334      }
335      assert(BT && "Unknown message kind.");
336
337      BugReport *R = new BugReport(*BT, BT->getName(), N);
338      const ObjCMessageExpr *ME = msg.getOriginExpr();
339      R->addRange(ME->getReceiverRange());
340
341      // FIXME: getTrackNullOrUndefValueVisitor can't handle "super" yet.
342      if (const Expr *ReceiverE = ME->getInstanceReceiver())
343        bugreporter::trackNullOrUndefValue(N, ReceiverE, *R);
344      C.emitReport(R);
345    }
346    return;
347  } else {
348    // Bifurcate the state into nil and non-nil ones.
349    DefinedOrUnknownSVal receiverVal = recVal.castAs<DefinedOrUnknownSVal>();
350
351    ProgramStateRef state = C.getState();
352    ProgramStateRef notNilState, nilState;
353    llvm::tie(notNilState, nilState) = state->assume(receiverVal);
354
355    // Handle receiver must be nil.
356    if (nilState && !notNilState) {
357      HandleNilReceiver(C, state, msg);
358      return;
359    }
360  }
361}
362
363void CallAndMessageChecker::emitNilReceiverBug(CheckerContext &C,
364                                               const ObjCMethodCall &msg,
365                                               ExplodedNode *N) const {
366
367  if (!BT_msg_ret)
368    BT_msg_ret.reset(
369      new BuiltinBug("Receiver in message expression is "
370                     "'nil' and returns a garbage value"));
371
372  const ObjCMessageExpr *ME = msg.getOriginExpr();
373
374  SmallString<200> buf;
375  llvm::raw_svector_ostream os(buf);
376  os << "The receiver of message '" << ME->getSelector().getAsString()
377     << "' is nil and returns a value of type '";
378  msg.getResultType().print(os, C.getLangOpts());
379  os << "' that will be garbage";
380
381  BugReport *report = new BugReport(*BT_msg_ret, os.str(), N);
382  report->addRange(ME->getReceiverRange());
383  // FIXME: This won't track "self" in messages to super.
384  if (const Expr *receiver = ME->getInstanceReceiver()) {
385    bugreporter::trackNullOrUndefValue(N, receiver, *report);
386  }
387  C.emitReport(report);
388}
389
390static bool supportsNilWithFloatRet(const llvm::Triple &triple) {
391  return (triple.getVendor() == llvm::Triple::Apple &&
392          (triple.getOS() == llvm::Triple::IOS ||
393           !triple.isMacOSXVersionLT(10,5)));
394}
395
396void CallAndMessageChecker::HandleNilReceiver(CheckerContext &C,
397                                              ProgramStateRef state,
398                                              const ObjCMethodCall &Msg) const {
399  ASTContext &Ctx = C.getASTContext();
400
401  // Check the return type of the message expression.  A message to nil will
402  // return different values depending on the return type and the architecture.
403  QualType RetTy = Msg.getResultType();
404  CanQualType CanRetTy = Ctx.getCanonicalType(RetTy);
405  const LocationContext *LCtx = C.getLocationContext();
406
407  if (CanRetTy->isStructureOrClassType()) {
408    // Structure returns are safe since the compiler zeroes them out.
409    SVal V = C.getSValBuilder().makeZeroVal(RetTy);
410    C.addTransition(state->BindExpr(Msg.getOriginExpr(), LCtx, V));
411    return;
412  }
413
414  // Other cases: check if sizeof(return type) > sizeof(void*)
415  if (CanRetTy != Ctx.VoidTy && C.getLocationContext()->getParentMap()
416                                  .isConsumedExpr(Msg.getOriginExpr())) {
417    // Compute: sizeof(void *) and sizeof(return type)
418    const uint64_t voidPtrSize = Ctx.getTypeSize(Ctx.VoidPtrTy);
419    const uint64_t returnTypeSize = Ctx.getTypeSize(CanRetTy);
420
421    if (voidPtrSize < returnTypeSize &&
422        !(supportsNilWithFloatRet(Ctx.getTargetInfo().getTriple()) &&
423          (Ctx.FloatTy == CanRetTy ||
424           Ctx.DoubleTy == CanRetTy ||
425           Ctx.LongDoubleTy == CanRetTy ||
426           Ctx.LongLongTy == CanRetTy ||
427           Ctx.UnsignedLongLongTy == CanRetTy))) {
428      if (ExplodedNode *N = C.generateSink(state))
429        emitNilReceiverBug(C, Msg, N);
430      return;
431    }
432
433    // Handle the safe cases where the return value is 0 if the
434    // receiver is nil.
435    //
436    // FIXME: For now take the conservative approach that we only
437    // return null values if we *know* that the receiver is nil.
438    // This is because we can have surprises like:
439    //
440    //   ... = [[NSScreens screens] objectAtIndex:0];
441    //
442    // What can happen is that [... screens] could return nil, but
443    // it most likely isn't nil.  We should assume the semantics
444    // of this case unless we have *a lot* more knowledge.
445    //
446    SVal V = C.getSValBuilder().makeZeroVal(RetTy);
447    C.addTransition(state->BindExpr(Msg.getOriginExpr(), LCtx, V));
448    return;
449  }
450
451  C.addTransition(state);
452}
453
454void ento::registerCallAndMessageChecker(CheckerManager &mgr) {
455  mgr.registerChecker<CallAndMessageChecker>();
456}
457