LLVMConventionsChecker.cpp revision 581deb3da481053c4993c7600f97acf7768caac5 
1//=== LLVMConventionsChecker.cpp - Check LLVM codebase conventions ---*- 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 LLVMConventionsChecker, a bunch of small little checks
11// for checking specific coding conventions in the LLVM/Clang codebase.
12//
13//===----------------------------------------------------------------------===//
14
15#include "ClangSACheckers.h"
16#include "clang/StaticAnalyzer/Core/Checker.h"
17#include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h"
18#include "clang/AST/DeclTemplate.h"
19#include "clang/AST/StmtVisitor.h"
20#include "llvm/ADT/SmallString.h"
21
22using namespace clang;
23using namespace ento;
24
25//===----------------------------------------------------------------------===//
26// Generic type checking routines.
27//===----------------------------------------------------------------------===//
28
29static bool IsLLVMStringRef(QualType T) {
30  const RecordType *RT = T->getAs<RecordType>();
31  if (!RT)
32    return false;
33
34  return StringRef(QualType(RT, 0).getAsString()) ==
35          "class StringRef";
36}
37
38/// Check whether the declaration is semantically inside the top-level
39/// namespace named by ns.
40static bool InNamespace(const Decl *D, StringRef NS) {
41  const NamespaceDecl *ND = dyn_cast<NamespaceDecl>(D->getDeclContext());
42  if (!ND)
43    return false;
44  const IdentifierInfo *II = ND->getIdentifier();
45  if (!II || !II->getName().equals(NS))
46    return false;
47  return isa<TranslationUnitDecl>(ND->getDeclContext());
48}
49
50static bool IsStdString(QualType T) {
51  if (const ElaboratedType *QT = T->getAs<ElaboratedType>())
52    T = QT->getNamedType();
53
54  const TypedefType *TT = T->getAs<TypedefType>();
55  if (!TT)
56    return false;
57
58  const TypedefNameDecl *TD = TT->getDecl();
59
60  if (!InNamespace(TD, "std"))
61    return false;
62
63  return TD->getName() == "string";
64}
65
66static bool IsClangType(const RecordDecl *RD) {
67  return RD->getName() == "Type" && InNamespace(RD, "clang");
68}
69
70static bool IsClangDecl(const RecordDecl *RD) {
71  return RD->getName() == "Decl" && InNamespace(RD, "clang");
72}
73
74static bool IsClangStmt(const RecordDecl *RD) {
75  return RD->getName() == "Stmt" && InNamespace(RD, "clang");
76}
77
78static bool IsClangAttr(const RecordDecl *RD) {
79  return RD->getName() == "Attr" && InNamespace(RD, "clang");
80}
81
82static bool IsStdVector(QualType T) {
83  const TemplateSpecializationType *TS = T->getAs<TemplateSpecializationType>();
84  if (!TS)
85    return false;
86
87  TemplateName TM = TS->getTemplateName();
88  TemplateDecl *TD = TM.getAsTemplateDecl();
89
90  if (!TD || !InNamespace(TD, "std"))
91    return false;
92
93  return TD->getName() == "vector";
94}
95
96static bool IsSmallVector(QualType T) {
97  const TemplateSpecializationType *TS = T->getAs<TemplateSpecializationType>();
98  if (!TS)
99    return false;
100
101  TemplateName TM = TS->getTemplateName();
102  TemplateDecl *TD = TM.getAsTemplateDecl();
103
104  if (!TD || !InNamespace(TD, "llvm"))
105    return false;
106
107  return TD->getName() == "SmallVector";
108}
109
110//===----------------------------------------------------------------------===//
111// CHECK: a StringRef should not be bound to a temporary std::string whose
112// lifetime is shorter than the StringRef's.
113//===----------------------------------------------------------------------===//
114
115namespace {
116class StringRefCheckerVisitor : public StmtVisitor<StringRefCheckerVisitor> {
117  BugReporter &BR;
118  const Decl *DeclWithIssue;
119public:
120  StringRefCheckerVisitor(const Decl *declWithIssue, BugReporter &br)
121    : BR(br), DeclWithIssue(declWithIssue) {}
122  void VisitChildren(Stmt *S) {
123    for (Stmt::child_iterator I = S->child_begin(), E = S->child_end() ;
124      I != E; ++I)
125      if (Stmt *child = *I)
126        Visit(child);
127  }
128  void VisitStmt(Stmt *S) { VisitChildren(S); }
129  void VisitDeclStmt(DeclStmt *DS);
130private:
131  void VisitVarDecl(VarDecl *VD);
132};
133} // end anonymous namespace
134
135static void CheckStringRefAssignedTemporary(const Decl *D, BugReporter &BR) {
136  StringRefCheckerVisitor walker(D, BR);
137  walker.Visit(D->getBody());
138}
139
140void StringRefCheckerVisitor::VisitDeclStmt(DeclStmt *S) {
141  VisitChildren(S);
142
143  for (DeclStmt::decl_iterator I = S->decl_begin(), E = S->decl_end();I!=E; ++I)
144    if (VarDecl *VD = dyn_cast<VarDecl>(*I))
145      VisitVarDecl(VD);
146}
147
148void StringRefCheckerVisitor::VisitVarDecl(VarDecl *VD) {
149  Expr *Init = VD->getInit();
150  if (!Init)
151    return;
152
153  // Pattern match for:
154  // StringRef x = call() (where call returns std::string)
155  if (!IsLLVMStringRef(VD->getType()))
156    return;
157  ExprWithCleanups *Ex1 = dyn_cast<ExprWithCleanups>(Init);
158  if (!Ex1)
159    return;
160  CXXConstructExpr *Ex2 = dyn_cast<CXXConstructExpr>(Ex1->getSubExpr());
161  if (!Ex2 || Ex2->getNumArgs() != 1)
162    return;
163  ImplicitCastExpr *Ex3 = dyn_cast<ImplicitCastExpr>(Ex2->getArg(0));
164  if (!Ex3)
165    return;
166  CXXConstructExpr *Ex4 = dyn_cast<CXXConstructExpr>(Ex3->getSubExpr());
167  if (!Ex4 || Ex4->getNumArgs() != 1)
168    return;
169  ImplicitCastExpr *Ex5 = dyn_cast<ImplicitCastExpr>(Ex4->getArg(0));
170  if (!Ex5)
171    return;
172  CXXBindTemporaryExpr *Ex6 = dyn_cast<CXXBindTemporaryExpr>(Ex5->getSubExpr());
173  if (!Ex6 || !IsStdString(Ex6->getType()))
174    return;
175
176  // Okay, badness!  Report an error.
177  const char *desc = "StringRef should not be bound to temporary "
178                     "std::string that it outlives";
179  PathDiagnosticLocation VDLoc =
180    PathDiagnosticLocation::createBegin(VD, BR.getSourceManager());
181  BR.EmitBasicReport(DeclWithIssue, desc, "LLVM Conventions", desc,
182                     VDLoc, Init->getSourceRange());
183}
184
185//===----------------------------------------------------------------------===//
186// CHECK: Clang AST nodes should not have fields that can allocate
187//   memory.
188//===----------------------------------------------------------------------===//
189
190static bool AllocatesMemory(QualType T) {
191  return IsStdVector(T) || IsStdString(T) || IsSmallVector(T);
192}
193
194// This type checking could be sped up via dynamic programming.
195static bool IsPartOfAST(const CXXRecordDecl *R) {
196  if (IsClangStmt(R) || IsClangType(R) || IsClangDecl(R) || IsClangAttr(R))
197    return true;
198
199  for (CXXRecordDecl::base_class_const_iterator I = R->bases_begin(),
200                                                E = R->bases_end(); I!=E; ++I) {
201    CXXBaseSpecifier BS = *I;
202    QualType T = BS.getType();
203    if (const RecordType *baseT = T->getAs<RecordType>()) {
204      CXXRecordDecl *baseD = cast<CXXRecordDecl>(baseT->getDecl());
205      if (IsPartOfAST(baseD))
206        return true;
207    }
208  }
209
210  return false;
211}
212
213namespace {
214class ASTFieldVisitor {
215  SmallVector<FieldDecl*, 10> FieldChain;
216  const CXXRecordDecl *Root;
217  BugReporter &BR;
218public:
219  ASTFieldVisitor(const CXXRecordDecl *root, BugReporter &br)
220    : Root(root), BR(br) {}
221
222  void Visit(FieldDecl *D);
223  void ReportError(QualType T);
224};
225} // end anonymous namespace
226
227static void CheckASTMemory(const CXXRecordDecl *R, BugReporter &BR) {
228  if (!IsPartOfAST(R))
229    return;
230
231  for (RecordDecl::field_iterator I = R->field_begin(), E = R->field_end();
232       I != E; ++I) {
233    ASTFieldVisitor walker(R, BR);
234    walker.Visit(*I);
235  }
236}
237
238void ASTFieldVisitor::Visit(FieldDecl *D) {
239  FieldChain.push_back(D);
240
241  QualType T = D->getType();
242
243  if (AllocatesMemory(T))
244    ReportError(T);
245
246  if (const RecordType *RT = T->getAs<RecordType>()) {
247    const RecordDecl *RD = RT->getDecl()->getDefinition();
248    for (RecordDecl::field_iterator I = RD->field_begin(), E = RD->field_end();
249         I != E; ++I)
250      Visit(*I);
251  }
252
253  FieldChain.pop_back();
254}
255
256void ASTFieldVisitor::ReportError(QualType T) {
257  SmallString<1024> buf;
258  llvm::raw_svector_ostream os(buf);
259
260  os << "AST class '" << Root->getName() << "' has a field '"
261     << FieldChain.front()->getName() << "' that allocates heap memory";
262  if (FieldChain.size() > 1) {
263    os << " via the following chain: ";
264    bool isFirst = true;
265    for (SmallVectorImpl<FieldDecl*>::iterator I=FieldChain.begin(),
266         E=FieldChain.end(); I!=E; ++I) {
267      if (!isFirst)
268        os << '.';
269      else
270        isFirst = false;
271      os << (*I)->getName();
272    }
273  }
274  os << " (type " << FieldChain.back()->getType().getAsString() << ")";
275  os.flush();
276
277  // Note that this will fire for every translation unit that uses this
278  // class.  This is suboptimal, but at least scan-build will merge
279  // duplicate HTML reports.  In the future we need a unified way of merging
280  // duplicate reports across translation units.  For C++ classes we cannot
281  // just report warnings when we see an out-of-line method definition for a
282  // class, as that heuristic doesn't always work (the complete definition of
283  // the class may be in the header file, for example).
284  PathDiagnosticLocation L = PathDiagnosticLocation::createBegin(
285                               FieldChain.front(), BR.getSourceManager());
286  BR.EmitBasicReport(Root, "AST node allocates heap memory", "LLVM Conventions",
287                     os.str(), L);
288}
289
290//===----------------------------------------------------------------------===//
291// LLVMConventionsChecker
292//===----------------------------------------------------------------------===//
293
294namespace {
295class LLVMConventionsChecker : public Checker<
296                                                check::ASTDecl<CXXRecordDecl>,
297                                                check::ASTCodeBody > {
298public:
299  void checkASTDecl(const CXXRecordDecl *R, AnalysisManager& mgr,
300                    BugReporter &BR) const {
301    if (R->isCompleteDefinition())
302      CheckASTMemory(R, BR);
303  }
304
305  void checkASTCodeBody(const Decl *D, AnalysisManager& mgr,
306                        BugReporter &BR) const {
307    CheckStringRefAssignedTemporary(D, BR);
308  }
309};
310}
311
312void ento::registerLLVMConventionsChecker(CheckerManager &mgr) {
313  mgr.registerChecker<LLVMConventionsChecker>();
314}
315