Home | History | Annotate | Download | only in Checkers 
      1 //== CStringSyntaxChecker.cpp - CoreFoundation containers API *- 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 // An AST checker that looks for common pitfalls when using C string APIs.
     11 //  - Identifies erroneous patterns in the last argument to strncat - the number
     12 //    of bytes to copy.
     13 //
     14 //===----------------------------------------------------------------------===//
     15 #include "ClangSACheckers.h"
     16 #include "clang/AST/Expr.h"
     17 #include "clang/AST/OperationKinds.h"
     18 #include "clang/AST/StmtVisitor.h"
     19 #include "clang/Analysis/AnalysisContext.h"
     20 #include "clang/Basic/TargetInfo.h"
     21 #include "clang/Basic/TypeTraits.h"
     22 #include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h"
     23 #include "clang/StaticAnalyzer/Core/Checker.h"
     24 #include "clang/StaticAnalyzer/Core/PathSensitive/AnalysisManager.h"
     25 #include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
     26 #include "llvm/ADT/SmallString.h"
     27 #include "llvm/Support/raw_ostream.h"
     28 
     29 using namespace clang;
     30 using namespace ento;
     31 
     32 namespace {
     33 class WalkAST: public StmtVisitor<WalkAST> {
     34   const CheckerBase *Checker;
     35   BugReporter &BR;
     36   AnalysisDeclContext* AC;
     37 
     38   /// Check if two expressions refer to the same declaration.
     39   inline bool sameDecl(const Expr *A1, const Expr *A2) {
     40     if (const DeclRefExpr *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts()))
     41       if (const DeclRefExpr *D2 = dyn_cast<DeclRefExpr>(A2->IgnoreParenCasts()))
     42         return D1->getDecl() == D2->getDecl();
     43     return false;
     44   }
     45 
     46   /// Check if the expression E is a sizeof(WithArg).
     47   inline bool isSizeof(const Expr *E, const Expr *WithArg) {
     48     if (const UnaryExprOrTypeTraitExpr *UE =
     49     dyn_cast<UnaryExprOrTypeTraitExpr>(E))
     50       if (UE->getKind() == UETT_SizeOf)
     51         return sameDecl(UE->getArgumentExpr(), WithArg);
     52     return false;
     53   }
     54 
     55   /// Check if the expression E is a strlen(WithArg).
     56   inline bool isStrlen(const Expr *E, const Expr *WithArg) {
     57     if (const CallExpr *CE = dyn_cast<CallExpr>(E)) {
     58       const FunctionDecl *FD = CE->getDirectCallee();
     59       if (!FD)
     60         return false;
     61       return (CheckerContext::isCLibraryFunction(FD, "strlen") &&
     62               sameDecl(CE->getArg(0), WithArg));
     63     }
     64     return false;
     65   }
     66 
     67   /// Check if the expression is an integer literal with value 1.
     68   inline bool isOne(const Expr *E) {
     69     if (const IntegerLiteral *IL = dyn_cast<IntegerLiteral>(E))
     70       return (IL->getValue().isIntN(1));
     71     return false;
     72   }
     73 
     74   inline StringRef getPrintableName(const Expr *E) {
     75     if (const DeclRefExpr *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts()))
     76       return D->getDecl()->getName();
     77     return StringRef();
     78   }
     79 
     80   /// Identify erroneous patterns in the last argument to strncat - the number
     81   /// of bytes to copy.
     82   bool containsBadStrncatPattern(const CallExpr *CE);
     83 
     84 public:
     85   WalkAST(const CheckerBase *checker, BugReporter &br, AnalysisDeclContext *ac)
     86       : Checker(checker), BR(br), AC(ac) {}
     87 
     88   // Statement visitor methods.
     89   void VisitChildren(Stmt *S);
     90   void VisitStmt(Stmt *S) {
     91     VisitChildren(S);
     92   }
     93   void VisitCallExpr(CallExpr *CE);
     94 };
     95 } // end anonymous namespace
     96 
     97 // The correct size argument should look like following:
     98 //   strncat(dst, src, sizeof(dst) - strlen(dest) - 1);
     99 // We look for the following anti-patterns:
    100 //   - strncat(dst, src, sizeof(dst) - strlen(dst));
    101 //   - strncat(dst, src, sizeof(dst) - 1);
    102 //   - strncat(dst, src, sizeof(dst));
    103 bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) {
    104   if (CE->getNumArgs() != 3)
    105     return false;
    106   const Expr *DstArg = CE->getArg(0);
    107   const Expr *SrcArg = CE->getArg(1);
    108   const Expr *LenArg = CE->getArg(2);
    109 
    110   // Identify wrong size expressions, which are commonly used instead.
    111   if (const BinaryOperator *BE =
    112               dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) {
    113     // - sizeof(dst) - strlen(dst)
    114     if (BE->getOpcode() == BO_Sub) {
    115       const Expr *L = BE->getLHS();
    116       const Expr *R = BE->getRHS();
    117       if (isSizeof(L, DstArg) && isStrlen(R, DstArg))
    118         return true;
    119 
    120       // - sizeof(dst) - 1
    121       if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts()))
    122         return true;
    123     }
    124   }
    125   // - sizeof(dst)
    126   if (isSizeof(LenArg, DstArg))
    127     return true;
    128 
    129   // - sizeof(src)
    130   if (isSizeof(LenArg, SrcArg))
    131     return true;
    132   return false;
    133 }
    134 
    135 void WalkAST::VisitCallExpr(CallExpr *CE) {
    136   const FunctionDecl *FD = CE->getDirectCallee();
    137   if (!FD)
    138     return;
    139 
    140   if (CheckerContext::isCLibraryFunction(FD, "strncat")) {
    141     if (containsBadStrncatPattern(CE)) {
    142       const Expr *DstArg = CE->getArg(0);
    143       const Expr *LenArg = CE->getArg(2);
    144       PathDiagnosticLocation Loc =
    145         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
    146 
    147       StringRef DstName = getPrintableName(DstArg);
    148 
    149       SmallString<256> S;
    150       llvm::raw_svector_ostream os(S);
    151       os << "Potential buffer overflow. ";
    152       if (!DstName.empty()) {
    153         os << "Replace with 'sizeof(" << DstName << ") "
    154               "- strlen(" << DstName <<") - 1'";
    155         os << " or u";
    156       } else
    157         os << "U";
    158       os << "se a safer 'strlcat' API";
    159 
    160       BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument",
    161                          "C String API", os.str(), Loc,
    162                          LenArg->getSourceRange());
    163     }
    164   }
    165 
    166   // Recurse and check children.
    167   VisitChildren(CE);
    168 }
    169 
    170 void WalkAST::VisitChildren(Stmt *S) {
    171   for (Stmt *Child : S->children())
    172     if (Child)
    173       Visit(Child);
    174 }
    175 
    176 namespace {
    177 class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
    178 public:
    179 
    180   void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
    181       BugReporter &BR) const {
    182     WalkAST walker(this, BR, Mgr.getAnalysisDeclContext(D));
    183     walker.Visit(D->getBody());
    184   }
    185 };
    186 }
    187 
    188 void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
    189   mgr.registerChecker<CStringSyntaxChecker>();
    190 }
    191 
    192