LLVMConventionsChecker.cpp revision f7ccbad5d9949e7ddd1cbef43d482553b811e026 
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;
118public:
119  StringRefCheckerVisitor(BugReporter &br) : BR(br) {}
120  void VisitChildren(Stmt *S) {
121    for (Stmt::child_iterator I = S->child_begin(), E = S->child_end() ;
122      I != E; ++I)
123      if (Stmt *child = *I)
124        Visit(child);
125  }
126  void VisitStmt(Stmt *S) { VisitChildren(S); }
127  void VisitDeclStmt(DeclStmt *DS);
128private:
129  void VisitVarDecl(VarDecl *VD);
130};
131} // end anonymous namespace
132
133static void CheckStringRefAssignedTemporary(const Decl *D, BugReporter &BR) {
134  StringRefCheckerVisitor walker(BR);
135  walker.Visit(D->getBody());
136}
137
138void StringRefCheckerVisitor::VisitDeclStmt(DeclStmt *S) {
139  VisitChildren(S);
140
141  for (DeclStmt::decl_iterator I = S->decl_begin(), E = S->decl_end();I!=E; ++I)
142    if (VarDecl *VD = dyn_cast<VarDecl>(*I))
143      VisitVarDecl(VD);
144}
145
146void StringRefCheckerVisitor::VisitVarDecl(VarDecl *VD) {
147  Expr *Init = VD->getInit();
148  if (!Init)
149    return;
150
151  // Pattern match for:
152  // StringRef x = call() (where call returns std::string)
153  if (!IsLLVMStringRef(VD->getType()))
154    return;
155  ExprWithCleanups *Ex1 = dyn_cast<ExprWithCleanups>(Init);
156  if (!Ex1)
157    return;
158  CXXConstructExpr *Ex2 = dyn_cast<CXXConstructExpr>(Ex1->getSubExpr());
159  if (!Ex2 || Ex2->getNumArgs() != 1)
160    return;
161  ImplicitCastExpr *Ex3 = dyn_cast<ImplicitCastExpr>(Ex2->getArg(0));
162  if (!Ex3)
163    return;
164  CXXConstructExpr *Ex4 = dyn_cast<CXXConstructExpr>(Ex3->getSubExpr());
165  if (!Ex4 || Ex4->getNumArgs() != 1)
166    return;
167  ImplicitCastExpr *Ex5 = dyn_cast<ImplicitCastExpr>(Ex4->getArg(0));
168  if (!Ex5)
169    return;
170  CXXBindTemporaryExpr *Ex6 = dyn_cast<CXXBindTemporaryExpr>(Ex5->getSubExpr());
171  if (!Ex6 || !IsStdString(Ex6->getType()))
172    return;
173
174  // Okay, badness!  Report an error.
175  const char *desc = "StringRef should not be bound to temporary "
176                     "std::string that it outlives";
177  PathDiagnosticLocation VDLoc =
178    PathDiagnosticLocation::createBegin(VD, BR.getSourceManager());
179  BR.EmitBasicReport(desc, "LLVM Conventions", desc,
180                     VDLoc, Init->getSourceRange());
181}
182
183//===----------------------------------------------------------------------===//
184// CHECK: Clang AST nodes should not have fields that can allocate
185//   memory.
186//===----------------------------------------------------------------------===//
187
188static bool AllocatesMemory(QualType T) {
189  return IsStdVector(T) || IsStdString(T) || IsSmallVector(T);
190}
191
192// This type checking could be sped up via dynamic programming.
193static bool IsPartOfAST(const CXXRecordDecl *R) {
194  if (IsClangStmt(R) || IsClangType(R) || IsClangDecl(R) || IsClangAttr(R))
195    return true;
196
197  for (CXXRecordDecl::base_class_const_iterator I = R->bases_begin(),
198                                                E = R->bases_end(); I!=E; ++I) {
199    CXXBaseSpecifier BS = *I;
200    QualType T = BS.getType();
201    if (const RecordType *baseT = T->getAs<RecordType>()) {
202      CXXRecordDecl *baseD = cast<CXXRecordDecl>(baseT->getDecl());
203      if (IsPartOfAST(baseD))
204        return true;
205    }
206  }
207
208  return false;
209}
210
211namespace {
212class ASTFieldVisitor {
213  SmallVector<FieldDecl*, 10> FieldChain;
214  const CXXRecordDecl *Root;
215  BugReporter &BR;
216public:
217  ASTFieldVisitor(const CXXRecordDecl *root, BugReporter &br)
218    : Root(root), BR(br) {}
219
220  void Visit(FieldDecl *D);
221  void ReportError(QualType T);
222};
223} // end anonymous namespace
224
225static void CheckASTMemory(const CXXRecordDecl *R, BugReporter &BR) {
226  if (!IsPartOfAST(R))
227    return;
228
229  for (RecordDecl::field_iterator I = R->field_begin(), E = R->field_end();
230       I != E; ++I) {
231    ASTFieldVisitor walker(R, BR);
232    walker.Visit(*I);
233  }
234}
235
236void ASTFieldVisitor::Visit(FieldDecl *D) {
237  FieldChain.push_back(D);
238
239  QualType T = D->getType();
240
241  if (AllocatesMemory(T))
242    ReportError(T);
243
244  if (const RecordType *RT = T->getAs<RecordType>()) {
245    const RecordDecl *RD = RT->getDecl()->getDefinition();
246    for (RecordDecl::field_iterator I = RD->field_begin(), E = RD->field_end();
247         I != E; ++I)
248      Visit(*I);
249  }
250
251  FieldChain.pop_back();
252}
253
254void ASTFieldVisitor::ReportError(QualType T) {
255  SmallString<1024> buf;
256  llvm::raw_svector_ostream os(buf);
257
258  os << "AST class '" << Root->getName() << "' has a field '"
259     << FieldChain.front()->getName() << "' that allocates heap memory";
260  if (FieldChain.size() > 1) {
261    os << " via the following chain: ";
262    bool isFirst = true;
263    for (SmallVectorImpl<FieldDecl*>::iterator I=FieldChain.begin(),
264         E=FieldChain.end(); I!=E; ++I) {
265      if (!isFirst)
266        os << '.';
267      else
268        isFirst = false;
269      os << (*I)->getName();
270    }
271  }
272  os << " (type " << FieldChain.back()->getType().getAsString() << ")";
273  os.flush();
274
275  // Note that this will fire for every translation unit that uses this
276  // class.  This is suboptimal, but at least scan-build will merge
277  // duplicate HTML reports.  In the future we need a unified way of merging
278  // duplicate reports across translation units.  For C++ classes we cannot
279  // just report warnings when we see an out-of-line method definition for a
280  // class, as that heuristic doesn't always work (the complete definition of
281  // the class may be in the header file, for example).
282  PathDiagnosticLocation L = PathDiagnosticLocation::createBegin(
283                               FieldChain.front(), BR.getSourceManager());
284  BR.EmitBasicReport("AST node allocates heap memory", "LLVM Conventions",
285                     os.str(), L);
286}
287
288//===----------------------------------------------------------------------===//
289// LLVMConventionsChecker
290//===----------------------------------------------------------------------===//
291
292namespace {
293class LLVMConventionsChecker : public Checker<
294                                                check::ASTDecl<CXXRecordDecl>,
295                                                check::ASTCodeBody > {
296public:
297  void checkASTDecl(const CXXRecordDecl *R, AnalysisManager& mgr,
298                    BugReporter &BR) const {
299    if (R->isCompleteDefinition())
300      CheckASTMemory(R, BR);
301  }
302
303  void checkASTCodeBody(const Decl *D, AnalysisManager& mgr,
304                        BugReporter &BR) const {
305    CheckStringRefAssignedTemporary(D, BR);
306  }
307};
308}
309
310void ento::registerLLVMConventionsChecker(CheckerManager &mgr) {
311  mgr.registerChecker<LLVMConventionsChecker>();
312}
313