187b6ff09SAnna Zaks //== CStringSyntaxChecker.cpp - CoreFoundation containers API *- C++ -*-==// 287b6ff09SAnna Zaks // 387b6ff09SAnna Zaks // The LLVM Compiler Infrastructure 487b6ff09SAnna Zaks // 587b6ff09SAnna Zaks // This file is distributed under the University of Illinois Open Source 687b6ff09SAnna Zaks // License. See LICENSE.TXT for details. 787b6ff09SAnna Zaks // 887b6ff09SAnna Zaks //===----------------------------------------------------------------------===// 987b6ff09SAnna Zaks // 1087b6ff09SAnna Zaks // An AST checker that looks for common pitfalls when using C string APIs. 1187b6ff09SAnna Zaks // - Identifies erroneous patterns in the last argument to strncat - the number 1287b6ff09SAnna Zaks // of bytes to copy. 1387b6ff09SAnna Zaks // 1487b6ff09SAnna Zaks //===----------------------------------------------------------------------===// 1587b6ff09SAnna Zaks #include "ClangSACheckers.h" 1687b6ff09SAnna Zaks #include "clang/AST/Expr.h" 1787b6ff09SAnna Zaks #include "clang/AST/OperationKinds.h" 1887b6ff09SAnna Zaks #include "clang/AST/StmtVisitor.h" 19*50657f6bSGeorge Karpenkov #include "clang/Analysis/AnalysisDeclContext.h" 2087b6ff09SAnna Zaks #include "clang/Basic/TargetInfo.h" 2187b6ff09SAnna Zaks #include "clang/Basic/TypeTraits.h" 2287b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h" 233a02247dSChandler Carruth #include "clang/StaticAnalyzer/Core/Checker.h" 2487b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/AnalysisManager.h" 2587b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h" 264903802fSBenjamin Kramer #include "llvm/ADT/SmallString.h" 2787b6ff09SAnna Zaks #include "llvm/Support/raw_ostream.h" 2887b6ff09SAnna Zaks 2987b6ff09SAnna Zaks using namespace clang; 3087b6ff09SAnna Zaks using namespace ento; 3187b6ff09SAnna Zaks 3287b6ff09SAnna Zaks namespace { 3387b6ff09SAnna Zaks class WalkAST: public StmtVisitor<WalkAST> { 344aca9b1cSAlexander Kornienko const CheckerBase *Checker; 3587b6ff09SAnna Zaks BugReporter &BR; 3687b6ff09SAnna Zaks AnalysisDeclContext* AC; 3787b6ff09SAnna Zaks 3887b6ff09SAnna Zaks /// Check if two expressions refer to the same declaration. 393b008536SGabor Horvath bool sameDecl(const Expr *A1, const Expr *A2) { 403b008536SGabor Horvath if (const auto *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts())) 413b008536SGabor Horvath if (const auto *D2 = dyn_cast<DeclRefExpr>(A2->IgnoreParenCasts())) 4287b6ff09SAnna Zaks return D1->getDecl() == D2->getDecl(); 4387b6ff09SAnna Zaks return false; 4487b6ff09SAnna Zaks } 4587b6ff09SAnna Zaks 4687b6ff09SAnna Zaks /// Check if the expression E is a sizeof(WithArg). 473b008536SGabor Horvath bool isSizeof(const Expr *E, const Expr *WithArg) { 483b008536SGabor Horvath if (const auto *UE = dyn_cast<UnaryExprOrTypeTraitExpr>(E)) 493b008536SGabor Horvath if (UE->getKind() == UETT_SizeOf && !UE->isArgumentType()) 5087b6ff09SAnna Zaks return sameDecl(UE->getArgumentExpr(), WithArg); 5187b6ff09SAnna Zaks return false; 5287b6ff09SAnna Zaks } 5387b6ff09SAnna Zaks 5487b6ff09SAnna Zaks /// Check if the expression E is a strlen(WithArg). 553b008536SGabor Horvath bool isStrlen(const Expr *E, const Expr *WithArg) { 563b008536SGabor Horvath if (const auto *CE = dyn_cast<CallExpr>(E)) { 5787b6ff09SAnna Zaks const FunctionDecl *FD = CE->getDirectCallee(); 5887b6ff09SAnna Zaks if (!FD) 5987b6ff09SAnna Zaks return false; 600da67479SJordan Rose return (CheckerContext::isCLibraryFunction(FD, "strlen") && 610da67479SJordan Rose sameDecl(CE->getArg(0), WithArg)); 6287b6ff09SAnna Zaks } 6387b6ff09SAnna Zaks return false; 6487b6ff09SAnna Zaks } 6587b6ff09SAnna Zaks 6687b6ff09SAnna Zaks /// Check if the expression is an integer literal with value 1. 673b008536SGabor Horvath bool isOne(const Expr *E) { 683b008536SGabor Horvath if (const auto *IL = dyn_cast<IntegerLiteral>(E)) 6987b6ff09SAnna Zaks return (IL->getValue().isIntN(1)); 7087b6ff09SAnna Zaks return false; 7187b6ff09SAnna Zaks } 7287b6ff09SAnna Zaks 733b008536SGabor Horvath StringRef getPrintableName(const Expr *E) { 743b008536SGabor Horvath if (const auto *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts())) 7587b6ff09SAnna Zaks return D->getDecl()->getName(); 7687b6ff09SAnna Zaks return StringRef(); 7787b6ff09SAnna Zaks } 7887b6ff09SAnna Zaks 7987b6ff09SAnna Zaks /// Identify erroneous patterns in the last argument to strncat - the number 8087b6ff09SAnna Zaks /// of bytes to copy. 8187b6ff09SAnna Zaks bool containsBadStrncatPattern(const CallExpr *CE); 8287b6ff09SAnna Zaks 8387b6ff09SAnna Zaks public: 843b008536SGabor Horvath WalkAST(const CheckerBase *Checker, BugReporter &BR, AnalysisDeclContext *AC) 853b008536SGabor Horvath : Checker(Checker), BR(BR), AC(AC) {} 8687b6ff09SAnna Zaks 8787b6ff09SAnna Zaks // Statement visitor methods. 8887b6ff09SAnna Zaks void VisitChildren(Stmt *S); 8987b6ff09SAnna Zaks void VisitStmt(Stmt *S) { 9087b6ff09SAnna Zaks VisitChildren(S); 9187b6ff09SAnna Zaks } 9287b6ff09SAnna Zaks void VisitCallExpr(CallExpr *CE); 9387b6ff09SAnna Zaks }; 9487b6ff09SAnna Zaks } // end anonymous namespace 9587b6ff09SAnna Zaks 9687b6ff09SAnna Zaks // The correct size argument should look like following: 9787b6ff09SAnna Zaks // strncat(dst, src, sizeof(dst) - strlen(dest) - 1); 9887b6ff09SAnna Zaks // We look for the following anti-patterns: 9987b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst) - strlen(dst)); 10087b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst) - 1); 10187b6ff09SAnna Zaks // - strncat(dst, src, sizeof(dst)); 10287b6ff09SAnna Zaks bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) { 10360d98befSAnna Zaks if (CE->getNumArgs() != 3) 10460d98befSAnna Zaks return false; 10587b6ff09SAnna Zaks const Expr *DstArg = CE->getArg(0); 10687b6ff09SAnna Zaks const Expr *SrcArg = CE->getArg(1); 10787b6ff09SAnna Zaks const Expr *LenArg = CE->getArg(2); 10887b6ff09SAnna Zaks 10987b6ff09SAnna Zaks // Identify wrong size expressions, which are commonly used instead. 1103b008536SGabor Horvath if (const auto *BE = dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) { 11187b6ff09SAnna Zaks // - sizeof(dst) - strlen(dst) 11287b6ff09SAnna Zaks if (BE->getOpcode() == BO_Sub) { 11387b6ff09SAnna Zaks const Expr *L = BE->getLHS(); 11487b6ff09SAnna Zaks const Expr *R = BE->getRHS(); 11587b6ff09SAnna Zaks if (isSizeof(L, DstArg) && isStrlen(R, DstArg)) 11687b6ff09SAnna Zaks return true; 11787b6ff09SAnna Zaks 11887b6ff09SAnna Zaks // - sizeof(dst) - 1 11987b6ff09SAnna Zaks if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts())) 12087b6ff09SAnna Zaks return true; 12187b6ff09SAnna Zaks } 12287b6ff09SAnna Zaks } 12387b6ff09SAnna Zaks // - sizeof(dst) 12487b6ff09SAnna Zaks if (isSizeof(LenArg, DstArg)) 12587b6ff09SAnna Zaks return true; 12687b6ff09SAnna Zaks 12787b6ff09SAnna Zaks // - sizeof(src) 12887b6ff09SAnna Zaks if (isSizeof(LenArg, SrcArg)) 12987b6ff09SAnna Zaks return true; 13087b6ff09SAnna Zaks return false; 13187b6ff09SAnna Zaks } 13287b6ff09SAnna Zaks 13387b6ff09SAnna Zaks void WalkAST::VisitCallExpr(CallExpr *CE) { 13487b6ff09SAnna Zaks const FunctionDecl *FD = CE->getDirectCallee(); 13587b6ff09SAnna Zaks if (!FD) 13687b6ff09SAnna Zaks return; 13787b6ff09SAnna Zaks 1380da67479SJordan Rose if (CheckerContext::isCLibraryFunction(FD, "strncat")) { 13987b6ff09SAnna Zaks if (containsBadStrncatPattern(CE)) { 14087b6ff09SAnna Zaks const Expr *DstArg = CE->getArg(0); 14187b6ff09SAnna Zaks const Expr *LenArg = CE->getArg(2); 14287b6ff09SAnna Zaks PathDiagnosticLocation Loc = 14387b6ff09SAnna Zaks PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC); 14487b6ff09SAnna Zaks 14587b6ff09SAnna Zaks StringRef DstName = getPrintableName(DstArg); 14687b6ff09SAnna Zaks 1472c1dd271SDylan Noblesmith SmallString<256> S; 14887b6ff09SAnna Zaks llvm::raw_svector_ostream os(S); 14987b6ff09SAnna Zaks os << "Potential buffer overflow. "; 15087b6ff09SAnna Zaks if (!DstName.empty()) { 15187b6ff09SAnna Zaks os << "Replace with 'sizeof(" << DstName << ") " 15287b6ff09SAnna Zaks "- strlen(" << DstName <<") - 1'"; 15387b6ff09SAnna Zaks os << " or u"; 15487b6ff09SAnna Zaks } else 15587b6ff09SAnna Zaks os << "U"; 15687b6ff09SAnna Zaks os << "se a safer 'strlcat' API"; 15787b6ff09SAnna Zaks 1584aca9b1cSAlexander Kornienko BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument", 1594aca9b1cSAlexander Kornienko "C String API", os.str(), Loc, 1604aca9b1cSAlexander Kornienko LenArg->getSourceRange()); 16187b6ff09SAnna Zaks } 16287b6ff09SAnna Zaks } 16387b6ff09SAnna Zaks 16487b6ff09SAnna Zaks // Recurse and check children. 16587b6ff09SAnna Zaks VisitChildren(CE); 16687b6ff09SAnna Zaks } 16787b6ff09SAnna Zaks 16887b6ff09SAnna Zaks void WalkAST::VisitChildren(Stmt *S) { 169973431b2SBenjamin Kramer for (Stmt *Child : S->children()) 170973431b2SBenjamin Kramer if (Child) 171973431b2SBenjamin Kramer Visit(Child); 17287b6ff09SAnna Zaks } 17387b6ff09SAnna Zaks 17487b6ff09SAnna Zaks namespace { 17587b6ff09SAnna Zaks class CStringSyntaxChecker: public Checker<check::ASTCodeBody> { 17687b6ff09SAnna Zaks public: 17787b6ff09SAnna Zaks 17887b6ff09SAnna Zaks void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr, 17987b6ff09SAnna Zaks BugReporter &BR) const { 1804aca9b1cSAlexander Kornienko WalkAST walker(this, BR, Mgr.getAnalysisDeclContext(D)); 18187b6ff09SAnna Zaks walker.Visit(D->getBody()); 18287b6ff09SAnna Zaks } 18387b6ff09SAnna Zaks }; 184ab9db510SAlexander Kornienko } 18587b6ff09SAnna Zaks 18687b6ff09SAnna Zaks void ento::registerCStringSyntaxChecker(CheckerManager &mgr) { 18787b6ff09SAnna Zaks mgr.registerChecker<CStringSyntaxChecker>(); 18887b6ff09SAnna Zaks } 18987b6ff09SAnna Zaks 190