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/AST/DeclTemplate.h" 17#include "clang/AST/StmtVisitor.h" 18#include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h" 19#include "clang/StaticAnalyzer/Core/Checker.h" 20#include "llvm/ADT/SmallString.h" 21#include "llvm/Support/raw_ostream.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 StringRef(QualType(RT, 0).getAsString()) == 36 "class 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, 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 TypedefNameDecl *TD = TT->getDecl(); 60 61 if (!TD->isInStdNamespace()) 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 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 const Decl *DeclWithIssue; 119 BugReporter &BR; 120 const CheckerBase *Checker; 121 122public: 123 StringRefCheckerVisitor(const Decl *declWithIssue, BugReporter &br, 124 const CheckerBase *checker) 125 : DeclWithIssue(declWithIssue), BR(br), Checker(checker) {} 126 void VisitChildren(Stmt *S) { 127 for (Stmt::child_iterator I = S->child_begin(), E = S->child_end() ; 128 I != E; ++I) 129 if (Stmt *child = *I) 130 Visit(child); 131 } 132 void VisitStmt(Stmt *S) { VisitChildren(S); } 133 void VisitDeclStmt(DeclStmt *DS); 134private: 135 void VisitVarDecl(VarDecl *VD); 136}; 137} // end anonymous namespace 138 139static void CheckStringRefAssignedTemporary(const Decl *D, BugReporter &BR, 140 const CheckerBase *Checker) { 141 StringRefCheckerVisitor walker(D, BR, Checker); 142 walker.Visit(D->getBody()); 143} 144 145void StringRefCheckerVisitor::VisitDeclStmt(DeclStmt *S) { 146 VisitChildren(S); 147 148 for (auto *I : S->decls()) 149 if (VarDecl *VD = dyn_cast<VarDecl>(I)) 150 VisitVarDecl(VD); 151} 152 153void StringRefCheckerVisitor::VisitVarDecl(VarDecl *VD) { 154 Expr *Init = VD->getInit(); 155 if (!Init) 156 return; 157 158 // Pattern match for: 159 // StringRef x = call() (where call returns std::string) 160 if (!IsLLVMStringRef(VD->getType())) 161 return; 162 ExprWithCleanups *Ex1 = dyn_cast<ExprWithCleanups>(Init); 163 if (!Ex1) 164 return; 165 CXXConstructExpr *Ex2 = dyn_cast<CXXConstructExpr>(Ex1->getSubExpr()); 166 if (!Ex2 || Ex2->getNumArgs() != 1) 167 return; 168 ImplicitCastExpr *Ex3 = dyn_cast<ImplicitCastExpr>(Ex2->getArg(0)); 169 if (!Ex3) 170 return; 171 CXXConstructExpr *Ex4 = dyn_cast<CXXConstructExpr>(Ex3->getSubExpr()); 172 if (!Ex4 || Ex4->getNumArgs() != 1) 173 return; 174 ImplicitCastExpr *Ex5 = dyn_cast<ImplicitCastExpr>(Ex4->getArg(0)); 175 if (!Ex5) 176 return; 177 CXXBindTemporaryExpr *Ex6 = dyn_cast<CXXBindTemporaryExpr>(Ex5->getSubExpr()); 178 if (!Ex6 || !IsStdString(Ex6->getType())) 179 return; 180 181 // Okay, badness! Report an error. 182 const char *desc = "StringRef should not be bound to temporary " 183 "std::string that it outlives"; 184 PathDiagnosticLocation VDLoc = 185 PathDiagnosticLocation::createBegin(VD, BR.getSourceManager()); 186 BR.EmitBasicReport(DeclWithIssue, Checker, desc, "LLVM Conventions", desc, 187 VDLoc, Init->getSourceRange()); 188} 189 190//===----------------------------------------------------------------------===// 191// CHECK: Clang AST nodes should not have fields that can allocate 192// memory. 193//===----------------------------------------------------------------------===// 194 195static bool AllocatesMemory(QualType T) { 196 return IsStdVector(T) || IsStdString(T) || IsSmallVector(T); 197} 198 199// This type checking could be sped up via dynamic programming. 200static bool IsPartOfAST(const CXXRecordDecl *R) { 201 if (IsClangStmt(R) || IsClangType(R) || IsClangDecl(R) || IsClangAttr(R)) 202 return true; 203 204 for (const auto &BS : R->bases()) { 205 QualType T = BS.getType(); 206 if (const RecordType *baseT = T->getAs<RecordType>()) { 207 CXXRecordDecl *baseD = cast<CXXRecordDecl>(baseT->getDecl()); 208 if (IsPartOfAST(baseD)) 209 return true; 210 } 211 } 212 213 return false; 214} 215 216namespace { 217class ASTFieldVisitor { 218 SmallVector<FieldDecl*, 10> FieldChain; 219 const CXXRecordDecl *Root; 220 BugReporter &BR; 221 const CheckerBase *Checker; 222 223public: 224 ASTFieldVisitor(const CXXRecordDecl *root, BugReporter &br, 225 const CheckerBase *checker) 226 : Root(root), BR(br), Checker(checker) {} 227 228 void Visit(FieldDecl *D); 229 void ReportError(QualType T); 230}; 231} // end anonymous namespace 232 233static void CheckASTMemory(const CXXRecordDecl *R, BugReporter &BR, 234 const CheckerBase *Checker) { 235 if (!IsPartOfAST(R)) 236 return; 237 238 for (auto *I : R->fields()) { 239 ASTFieldVisitor walker(R, BR, Checker); 240 walker.Visit(I); 241 } 242} 243 244void ASTFieldVisitor::Visit(FieldDecl *D) { 245 FieldChain.push_back(D); 246 247 QualType T = D->getType(); 248 249 if (AllocatesMemory(T)) 250 ReportError(T); 251 252 if (const RecordType *RT = T->getAs<RecordType>()) { 253 const RecordDecl *RD = RT->getDecl()->getDefinition(); 254 for (auto *I : RD->fields()) 255 Visit(I); 256 } 257 258 FieldChain.pop_back(); 259} 260 261void ASTFieldVisitor::ReportError(QualType T) { 262 SmallString<1024> buf; 263 llvm::raw_svector_ostream os(buf); 264 265 os << "AST class '" << Root->getName() << "' has a field '" 266 << FieldChain.front()->getName() << "' that allocates heap memory"; 267 if (FieldChain.size() > 1) { 268 os << " via the following chain: "; 269 bool isFirst = true; 270 for (SmallVectorImpl<FieldDecl*>::iterator I=FieldChain.begin(), 271 E=FieldChain.end(); I!=E; ++I) { 272 if (!isFirst) 273 os << '.'; 274 else 275 isFirst = false; 276 os << (*I)->getName(); 277 } 278 } 279 os << " (type " << FieldChain.back()->getType().getAsString() << ")"; 280 os.flush(); 281 282 // Note that this will fire for every translation unit that uses this 283 // class. This is suboptimal, but at least scan-build will merge 284 // duplicate HTML reports. In the future we need a unified way of merging 285 // duplicate reports across translation units. For C++ classes we cannot 286 // just report warnings when we see an out-of-line method definition for a 287 // class, as that heuristic doesn't always work (the complete definition of 288 // the class may be in the header file, for example). 289 PathDiagnosticLocation L = PathDiagnosticLocation::createBegin( 290 FieldChain.front(), BR.getSourceManager()); 291 BR.EmitBasicReport(Root, Checker, "AST node allocates heap memory", 292 "LLVM Conventions", os.str(), L); 293} 294 295//===----------------------------------------------------------------------===// 296// LLVMConventionsChecker 297//===----------------------------------------------------------------------===// 298 299namespace { 300class LLVMConventionsChecker : public Checker< 301 check::ASTDecl<CXXRecordDecl>, 302 check::ASTCodeBody > { 303public: 304 void checkASTDecl(const CXXRecordDecl *R, AnalysisManager& mgr, 305 BugReporter &BR) const { 306 if (R->isCompleteDefinition()) 307 CheckASTMemory(R, BR, this); 308 } 309 310 void checkASTCodeBody(const Decl *D, AnalysisManager& mgr, 311 BugReporter &BR) const { 312 CheckStringRefAssignedTemporary(D, BR, this); 313 } 314}; 315} 316 317void ento::registerLLVMConventionsChecker(CheckerManager &mgr) { 318 mgr.registerChecker<LLVMConventionsChecker>(); 319} 320