LLVMConventionsChecker.cpp revision c367a876d0abcf32cb443712ce2709a0491be00b 
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/CheckerV2.h"
17#include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h"
18#include "clang/AST/DeclTemplate.h"
19#include "clang/AST/StmtVisitor.h"
20#include <string>
21#include "llvm/ADT/StringRef.h"
22
23using namespace clang;
24using namespace ento;
25
26//===----------------------------------------------------------------------===//
27// Generic type checking routines.
28//===----------------------------------------------------------------------===//
29
30static bool IsLLVMStringRef(QualType T) {
31  const RecordType *RT = T->getAs<RecordType>();
32  if (!RT)
33    return false;
34
35  return llvm::StringRef(QualType(RT, 0).getAsString()) ==
36          "class llvm::StringRef";
37}
38
39/// Check whether the declaration is semantically inside the top-level
40/// namespace named by ns.
41static bool InNamespace(const Decl *D, llvm::StringRef NS) {
42  const NamespaceDecl *ND = dyn_cast<NamespaceDecl>(D->getDeclContext());
43  if (!ND)
44    return false;
45  const IdentifierInfo *II = ND->getIdentifier();
46  if (!II || !II->getName().equals(NS))
47    return false;
48  return isa<TranslationUnitDecl>(ND->getDeclContext());
49}
50
51static bool IsStdString(QualType T) {
52  if (const ElaboratedType *QT = T->getAs<ElaboratedType>())
53    T = QT->getNamedType();
54
55  const TypedefType *TT = T->getAs<TypedefType>();
56  if (!TT)
57    return false;
58
59  const TypedefDecl *TD = TT->getDecl();
60
61  if (!InNamespace(TD, "std"))
62    return false;
63
64  return TD->getName() == "string";
65}
66
67static bool IsClangType(const RecordDecl *RD) {
68  return RD->getName() == "Type" && InNamespace(RD, "clang");
69}
70
71static bool IsClangDecl(const RecordDecl *RD) {
72  return RD->getName() == "Decl" && InNamespace(RD, "clang");
73}
74
75static bool IsClangStmt(const RecordDecl *RD) {
76  return RD->getName() == "Stmt" && InNamespace(RD, "clang");
77}
78
79static bool IsClangAttr(const RecordDecl *RD) {
80  return RD->getName() == "Attr" && InNamespace(RD, "clang");
81}
82
83static bool IsStdVector(QualType T) {
84  const TemplateSpecializationType *TS = T->getAs<TemplateSpecializationType>();
85  if (!TS)
86    return false;
87
88  TemplateName TM = TS->getTemplateName();
89  TemplateDecl *TD = TM.getAsTemplateDecl();
90
91  if (!TD || !InNamespace(TD, "std"))
92    return false;
93
94  return TD->getName() == "vector";
95}
96
97static bool IsSmallVector(QualType T) {
98  const TemplateSpecializationType *TS = T->getAs<TemplateSpecializationType>();
99  if (!TS)
100    return false;
101
102  TemplateName TM = TS->getTemplateName();
103  TemplateDecl *TD = TM.getAsTemplateDecl();
104
105  if (!TD || !InNamespace(TD, "llvm"))
106    return false;
107
108  return TD->getName() == "SmallVector";
109}
110
111//===----------------------------------------------------------------------===//
112// CHECK: a llvm::StringRef should not be bound to a temporary std::string whose
113// lifetime is shorter than the StringRef's.
114//===----------------------------------------------------------------------===//
115
116namespace {
117class StringRefCheckerVisitor : public StmtVisitor<StringRefCheckerVisitor> {
118  BugReporter &BR;
119public:
120  StringRefCheckerVisitor(BugReporter &br) : BR(br) {}
121  void VisitChildren(Stmt *S) {
122    for (Stmt::child_iterator I = S->child_begin(), E = S->child_end() ;
123      I != E; ++I)
124      if (Stmt *child = *I)
125        Visit(child);
126  }
127  void VisitStmt(Stmt *S) { VisitChildren(S); }
128  void VisitDeclStmt(DeclStmt *DS);
129private:
130  void VisitVarDecl(VarDecl *VD);
131};
132} // end anonymous namespace
133
134static void CheckStringRefAssignedTemporary(const Decl *D, BugReporter &BR) {
135  StringRefCheckerVisitor walker(BR);
136  walker.Visit(D->getBody());
137}
138
139void StringRefCheckerVisitor::VisitDeclStmt(DeclStmt *S) {
140  VisitChildren(S);
141
142  for (DeclStmt::decl_iterator I = S->decl_begin(), E = S->decl_end();I!=E; ++I)
143    if (VarDecl *VD = dyn_cast<VarDecl>(*I))
144      VisitVarDecl(VD);
145}
146
147void StringRefCheckerVisitor::VisitVarDecl(VarDecl *VD) {
148  Expr *Init = VD->getInit();
149  if (!Init)
150    return;
151
152  // Pattern match for:
153  // llvm::StringRef x = call() (where call returns std::string)
154  if (!IsLLVMStringRef(VD->getType()))
155    return;
156  ExprWithCleanups *Ex1 = dyn_cast<ExprWithCleanups>(Init);
157  if (!Ex1)
158    return;
159  CXXConstructExpr *Ex2 = dyn_cast<CXXConstructExpr>(Ex1->getSubExpr());
160  if (!Ex2 || Ex2->getNumArgs() != 1)
161    return;
162  ImplicitCastExpr *Ex3 = dyn_cast<ImplicitCastExpr>(Ex2->getArg(0));
163  if (!Ex3)
164    return;
165  CXXConstructExpr *Ex4 = dyn_cast<CXXConstructExpr>(Ex3->getSubExpr());
166  if (!Ex4 || Ex4->getNumArgs() != 1)
167    return;
168  ImplicitCastExpr *Ex5 = dyn_cast<ImplicitCastExpr>(Ex4->getArg(0));
169  if (!Ex5)
170    return;
171  CXXBindTemporaryExpr *Ex6 = dyn_cast<CXXBindTemporaryExpr>(Ex5->getSubExpr());
172  if (!Ex6 || !IsStdString(Ex6->getType()))
173    return;
174
175  // Okay, badness!  Report an error.
176  const char *desc = "StringRef should not be bound to temporary "
177                     "std::string that it outlives";
178
179  BR.EmitBasicReport(desc, "LLVM Conventions", desc,
180                     VD->getLocStart(), 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  llvm::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  llvm::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 (llvm::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  BR.EmitBasicReport("AST node allocates heap memory", "LLVM Conventions",
283                     os.str(), FieldChain.front()->getLocStart());
284}
285
286//===----------------------------------------------------------------------===//
287// LLVMConventionsChecker
288//===----------------------------------------------------------------------===//
289
290namespace {
291class LLVMConventionsChecker : public CheckerV2<
292                                                check::ASTDecl<CXXRecordDecl>,
293                                                check::ASTCodeBody > {
294public:
295  void checkASTDecl(const CXXRecordDecl *R, AnalysisManager& mgr,
296                    BugReporter &BR) const {
297    if (R->isDefinition())
298      CheckASTMemory(R, BR);
299  }
300
301  void checkASTCodeBody(const Decl *D, AnalysisManager& mgr,
302                        BugReporter &BR) const {
303    CheckStringRefAssignedTemporary(D, BR);
304  }
305};
306}
307
308void ento::registerLLVMConventionsChecker(CheckerManager &mgr) {
309  mgr.registerChecker<LLVMConventionsChecker>();
310}
311