1*87b6ff09SAnna Zaks //== CStringSyntaxChecker.cpp - CoreFoundation containers API *- C++ -*-==// 2*87b6ff09SAnna Zaks // 3*87b6ff09SAnna Zaks // The LLVM Compiler Infrastructure 4*87b6ff09SAnna Zaks // 5*87b6ff09SAnna Zaks // This file is distributed under the University of Illinois Open Source 6*87b6ff09SAnna Zaks // License. See LICENSE.TXT for details. 7*87b6ff09SAnna Zaks // 8*87b6ff09SAnna Zaks //===----------------------------------------------------------------------===// 9*87b6ff09SAnna Zaks // 10*87b6ff09SAnna Zaks // An AST checker that looks for common pitfalls when using C string APIs. 11*87b6ff09SAnna Zaks // - Identifies erroneous patterns in the last argument to strncat - the number 12*87b6ff09SAnna Zaks // of bytes to copy. 13*87b6ff09SAnna Zaks // 14*87b6ff09SAnna Zaks //===----------------------------------------------------------------------===// 15*87b6ff09SAnna Zaks #include "ClangSACheckers.h" 16*87b6ff09SAnna Zaks #include "clang/Analysis/AnalysisContext.h" 17*87b6ff09SAnna Zaks #include "clang/AST/Expr.h" 18*87b6ff09SAnna Zaks #include "clang/AST/OperationKinds.h" 19*87b6ff09SAnna Zaks #include "clang/AST/StmtVisitor.h" 20*87b6ff09SAnna Zaks #include "clang/Basic/TargetInfo.h" 21*87b6ff09SAnna Zaks #include "clang/Basic/TypeTraits.h" 22*87b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/Checker.h" 23*87b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h" 24*87b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/AnalysisManager.h" 25*87b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h" 26*87b6ff09SAnna Zaks #include "llvm/Support/raw_ostream.h" 27*87b6ff09SAnna Zaks 28*87b6ff09SAnna Zaks using namespace clang; 29*87b6ff09SAnna Zaks using namespace ento; 30*87b6ff09SAnna Zaks 31*87b6ff09SAnna Zaks namespace { 32*87b6ff09SAnna Zaks class WalkAST: public StmtVisitor<WalkAST> { 33*87b6ff09SAnna Zaks BugReporter &BR; 34*87b6ff09SAnna Zaks AnalysisDeclContext* AC; 35*87b6ff09SAnna Zaks ASTContext &ASTC; 36*87b6ff09SAnna Zaks 37*87b6ff09SAnna Zaks /// Check if two expressions refer to the same declaration. 38*87b6ff09SAnna Zaks inline bool sameDecl(const Expr *A1, const Expr *A2) { 39*87b6ff09SAnna Zaks if (const DeclRefExpr *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts())) 40*87b6ff09SAnna Zaks if (const DeclRefExpr *D2 = dyn_cast<DeclRefExpr>(A2->IgnoreParenCasts())) 41*87b6ff09SAnna Zaks return D1->getDecl() == D2->getDecl(); 42*87b6ff09SAnna Zaks return false; 43*87b6ff09SAnna Zaks } 44*87b6ff09SAnna Zaks 45*87b6ff09SAnna Zaks /// Check if the expression E is a sizeof(WithArg). 46*87b6ff09SAnna Zaks inline bool isSizeof(const Expr *E, const Expr *WithArg) { 47*87b6ff09SAnna Zaks if (const UnaryExprOrTypeTraitExpr *UE = 48*87b6ff09SAnna Zaks dyn_cast<UnaryExprOrTypeTraitExpr>(E)) 49*87b6ff09SAnna Zaks if (UE->getKind() == UETT_SizeOf) 50*87b6ff09SAnna Zaks return sameDecl(UE->getArgumentExpr(), WithArg); 51*87b6ff09SAnna Zaks return false; 52*87b6ff09SAnna Zaks } 53*87b6ff09SAnna Zaks 54*87b6ff09SAnna Zaks /// Check if the expression E is a strlen(WithArg). 55*87b6ff09SAnna Zaks inline bool isStrlen(const Expr *E, const Expr *WithArg) { 56*87b6ff09SAnna Zaks if (const CallExpr *CE = dyn_cast<CallExpr>(E)) { 57*87b6ff09SAnna Zaks const FunctionDecl *FD = CE->getDirectCallee(); 58*87b6ff09SAnna Zaks if (!FD) 59*87b6ff09SAnna Zaks return false; 60*87b6ff09SAnna Zaks return (CheckerContext::isCLibraryFunction(FD, "strlen", ASTC) 61*87b6ff09SAnna Zaks && sameDecl(CE->getArg(0), WithArg)); 62*87b6ff09SAnna Zaks } 63*87b6ff09SAnna Zaks return false; 64*87b6ff09SAnna Zaks } 65*87b6ff09SAnna Zaks 66*87b6ff09SAnna Zaks /// Check if the expression is an integer literal with value 1. 67*87b6ff09SAnna Zaks inline bool isOne(const Expr *E) { 68*87b6ff09SAnna Zaks if (const IntegerLiteral *IL = dyn_cast<IntegerLiteral>(E)) 69*87b6ff09SAnna Zaks return (IL->getValue().isIntN(1)); 70*87b6ff09SAnna Zaks return false; 71*87b6ff09SAnna Zaks } 72*87b6ff09SAnna Zaks 73*87b6ff09SAnna Zaks inline StringRef getPrintableName(const Expr *E) { 74*87b6ff09SAnna Zaks if (const DeclRefExpr *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts())) 75*87b6ff09SAnna Zaks return D->getDecl()->getName(); 76*87b6ff09SAnna Zaks return StringRef(); 77*87b6ff09SAnna Zaks } 78*87b6ff09SAnna Zaks 79*87b6ff09SAnna Zaks /// Identify erroneous patterns in the last argument to strncat - the number 80*87b6ff09SAnna Zaks /// of bytes to copy. 81*87b6ff09SAnna Zaks bool containsBadStrncatPattern(const CallExpr *CE); 82*87b6ff09SAnna Zaks 83*87b6ff09SAnna Zaks public: 84*87b6ff09SAnna Zaks WalkAST(BugReporter &br, AnalysisDeclContext* ac) : 85*87b6ff09SAnna Zaks BR(br), AC(ac), ASTC(AC->getASTContext()) { 86*87b6ff09SAnna Zaks } 87*87b6ff09SAnna Zaks 88*87b6ff09SAnna Zaks // Statement visitor methods. 89*87b6ff09SAnna Zaks void VisitChildren(Stmt *S); 90*87b6ff09SAnna Zaks void VisitStmt(Stmt *S) { 91*87b6ff09SAnna Zaks VisitChildren(S); 92*87b6ff09SAnna Zaks } 93*87b6ff09SAnna Zaks void VisitCallExpr(CallExpr *CE); 94*87b6ff09SAnna Zaks }; 95*87b6ff09SAnna Zaks } // end anonymous namespace 96*87b6ff09SAnna Zaks 97*87b6ff09SAnna Zaks // The correct size argument should look like following: 98*87b6ff09SAnna Zaks // strncat(dst, src, sizeof(dst) - strlen(dest) - 1); 99*87b6ff09SAnna Zaks // We look for the following anti-patterns: 100*87b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst) - strlen(dst)); 101*87b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst) - 1); 102*87b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst)); 103*87b6ff09SAnna Zaks bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) { 104*87b6ff09SAnna Zaks const Expr *DstArg = CE->getArg(0); 105*87b6ff09SAnna Zaks const Expr *SrcArg = CE->getArg(1); 106*87b6ff09SAnna Zaks const Expr *LenArg = CE->getArg(2); 107*87b6ff09SAnna Zaks 108*87b6ff09SAnna Zaks // Identify wrong size expressions, which are commonly used instead. 109*87b6ff09SAnna Zaks if (const BinaryOperator *BE = 110*87b6ff09SAnna Zaks dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) { 111*87b6ff09SAnna Zaks // - sizeof(dst) - strlen(dst) 112*87b6ff09SAnna Zaks if (BE->getOpcode() == BO_Sub) { 113*87b6ff09SAnna Zaks const Expr *L = BE->getLHS(); 114*87b6ff09SAnna Zaks const Expr *R = BE->getRHS(); 115*87b6ff09SAnna Zaks if (isSizeof(L, DstArg) && isStrlen(R, DstArg)) 116*87b6ff09SAnna Zaks return true; 117*87b6ff09SAnna Zaks 118*87b6ff09SAnna Zaks // - sizeof(dst) - 1 119*87b6ff09SAnna Zaks if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts())) 120*87b6ff09SAnna Zaks return true; 121*87b6ff09SAnna Zaks } 122*87b6ff09SAnna Zaks } 123*87b6ff09SAnna Zaks // - sizeof(dst) 124*87b6ff09SAnna Zaks if (isSizeof(LenArg, DstArg)) 125*87b6ff09SAnna Zaks return true; 126*87b6ff09SAnna Zaks 127*87b6ff09SAnna Zaks // - sizeof(src) 128*87b6ff09SAnna Zaks if (isSizeof(LenArg, SrcArg)) 129*87b6ff09SAnna Zaks return true; 130*87b6ff09SAnna Zaks return false; 131*87b6ff09SAnna Zaks } 132*87b6ff09SAnna Zaks 133*87b6ff09SAnna Zaks void WalkAST::VisitCallExpr(CallExpr *CE) { 134*87b6ff09SAnna Zaks const FunctionDecl *FD = CE->getDirectCallee(); 135*87b6ff09SAnna Zaks if (!FD) 136*87b6ff09SAnna Zaks return; 137*87b6ff09SAnna Zaks 138*87b6ff09SAnna Zaks if (CheckerContext::isCLibraryFunction(FD, "strncat", ASTC)) { 139*87b6ff09SAnna Zaks if (containsBadStrncatPattern(CE)) { 140*87b6ff09SAnna Zaks const Expr *DstArg = CE->getArg(0); 141*87b6ff09SAnna Zaks const Expr *LenArg = CE->getArg(2); 142*87b6ff09SAnna Zaks SourceRange R = LenArg->getSourceRange(); 143*87b6ff09SAnna Zaks PathDiagnosticLocation Loc = 144*87b6ff09SAnna Zaks PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC); 145*87b6ff09SAnna Zaks 146*87b6ff09SAnna Zaks StringRef DstName = getPrintableName(DstArg); 147*87b6ff09SAnna Zaks 148*87b6ff09SAnna Zaks llvm::SmallString<256> S; 149*87b6ff09SAnna Zaks llvm::raw_svector_ostream os(S); 150*87b6ff09SAnna Zaks os << "Potential buffer overflow. "; 151*87b6ff09SAnna Zaks if (!DstName.empty()) { 152*87b6ff09SAnna Zaks os << "Replace with 'sizeof(" << DstName << ") " 153*87b6ff09SAnna Zaks "- strlen(" << DstName <<") - 1'"; 154*87b6ff09SAnna Zaks os << " or u"; 155*87b6ff09SAnna Zaks } else 156*87b6ff09SAnna Zaks os << "U"; 157*87b6ff09SAnna Zaks os << "se a safer 'strlcat' API"; 158*87b6ff09SAnna Zaks 159*87b6ff09SAnna Zaks BR.EmitBasicReport("Anti-pattern in the argument", "C String API", 160*87b6ff09SAnna Zaks os.str(), Loc, &R, 1); 161*87b6ff09SAnna Zaks } 162*87b6ff09SAnna Zaks } 163*87b6ff09SAnna Zaks 164*87b6ff09SAnna Zaks // Recurse and check children. 165*87b6ff09SAnna Zaks VisitChildren(CE); 166*87b6ff09SAnna Zaks } 167*87b6ff09SAnna Zaks 168*87b6ff09SAnna Zaks void WalkAST::VisitChildren(Stmt *S) { 169*87b6ff09SAnna Zaks for (Stmt::child_iterator I = S->child_begin(), E = S->child_end(); I != E; 170*87b6ff09SAnna Zaks ++I) 171*87b6ff09SAnna Zaks if (Stmt *child = *I) 172*87b6ff09SAnna Zaks Visit(child); 173*87b6ff09SAnna Zaks } 174*87b6ff09SAnna Zaks 175*87b6ff09SAnna Zaks namespace { 176*87b6ff09SAnna Zaks class CStringSyntaxChecker: public Checker<check::ASTCodeBody> { 177*87b6ff09SAnna Zaks public: 178*87b6ff09SAnna Zaks 179*87b6ff09SAnna Zaks void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr, 180*87b6ff09SAnna Zaks BugReporter &BR) const { 181*87b6ff09SAnna Zaks WalkAST walker(BR, Mgr.getAnalysisDeclContext(D)); 182*87b6ff09SAnna Zaks walker.Visit(D->getBody()); 183*87b6ff09SAnna Zaks } 184*87b6ff09SAnna Zaks }; 185*87b6ff09SAnna Zaks } 186*87b6ff09SAnna Zaks 187*87b6ff09SAnna Zaks void ento::registerCStringSyntaxChecker(CheckerManager &mgr) { 188*87b6ff09SAnna Zaks mgr.registerChecker<CStringSyntaxChecker>(); 189*87b6ff09SAnna Zaks } 190*87b6ff09SAnna Zaks 191