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"
1950657f6bSGeorge 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 
838e75de21SDavid Carlier   /// Identify erroneous patterns in the last argument to strlcpy - the number
848e75de21SDavid Carlier   /// of bytes to copy.
858e75de21SDavid Carlier   /// The bad pattern checked is when the size is known
868e75de21SDavid Carlier   /// to be larger than the destination can handle.
878e75de21SDavid Carlier   ///   char dst[2];
888e75de21SDavid Carlier   ///   size_t cpy = 4;
898e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", sizeof("abcd") - 1);
908e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", 4);
918e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", cpy);
928e75de21SDavid Carlier   bool containsBadStrlcpyPattern(const CallExpr *CE);
938e75de21SDavid Carlier 
9487b6ff09SAnna Zaks public:
953b008536SGabor Horvath   WalkAST(const CheckerBase *Checker, BugReporter &BR, AnalysisDeclContext *AC)
963b008536SGabor Horvath       : Checker(Checker), BR(BR), AC(AC) {}
9787b6ff09SAnna Zaks 
9887b6ff09SAnna Zaks   // Statement visitor methods.
9987b6ff09SAnna Zaks   void VisitChildren(Stmt *S);
10087b6ff09SAnna Zaks   void VisitStmt(Stmt *S) {
10187b6ff09SAnna Zaks     VisitChildren(S);
10287b6ff09SAnna Zaks   }
10387b6ff09SAnna Zaks   void VisitCallExpr(CallExpr *CE);
10487b6ff09SAnna Zaks };
10587b6ff09SAnna Zaks } // end anonymous namespace
10687b6ff09SAnna Zaks 
10787b6ff09SAnna Zaks // The correct size argument should look like following:
10887b6ff09SAnna Zaks //   strncat(dst, src, sizeof(dst) - strlen(dest) - 1);
10987b6ff09SAnna Zaks // We look for the following anti-patterns:
11087b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - strlen(dst));
11187b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - 1);
11287b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst));
11387b6ff09SAnna Zaks bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) {
11460d98befSAnna Zaks   if (CE->getNumArgs() != 3)
11560d98befSAnna Zaks     return false;
11687b6ff09SAnna Zaks   const Expr *DstArg = CE->getArg(0);
11787b6ff09SAnna Zaks   const Expr *SrcArg = CE->getArg(1);
11887b6ff09SAnna Zaks   const Expr *LenArg = CE->getArg(2);
11987b6ff09SAnna Zaks 
12087b6ff09SAnna Zaks   // Identify wrong size expressions, which are commonly used instead.
1213b008536SGabor Horvath   if (const auto *BE = dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) {
12287b6ff09SAnna Zaks     // - sizeof(dst) - strlen(dst)
12387b6ff09SAnna Zaks     if (BE->getOpcode() == BO_Sub) {
12487b6ff09SAnna Zaks       const Expr *L = BE->getLHS();
12587b6ff09SAnna Zaks       const Expr *R = BE->getRHS();
12687b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isStrlen(R, DstArg))
12787b6ff09SAnna Zaks         return true;
12887b6ff09SAnna Zaks 
12987b6ff09SAnna Zaks       // - sizeof(dst) - 1
13087b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts()))
13187b6ff09SAnna Zaks         return true;
13287b6ff09SAnna Zaks     }
13387b6ff09SAnna Zaks   }
13487b6ff09SAnna Zaks   // - sizeof(dst)
13587b6ff09SAnna Zaks   if (isSizeof(LenArg, DstArg))
13687b6ff09SAnna Zaks     return true;
13787b6ff09SAnna Zaks 
13887b6ff09SAnna Zaks   // - sizeof(src)
13987b6ff09SAnna Zaks   if (isSizeof(LenArg, SrcArg))
14087b6ff09SAnna Zaks     return true;
14187b6ff09SAnna Zaks   return false;
14287b6ff09SAnna Zaks }
14387b6ff09SAnna Zaks 
1448e75de21SDavid Carlier bool WalkAST::containsBadStrlcpyPattern(const CallExpr *CE) {
1458e75de21SDavid Carlier   if (CE->getNumArgs() != 3)
1468e75de21SDavid Carlier     return false;
1478e75de21SDavid Carlier   const Expr *DstArg = CE->getArg(0);
1488e75de21SDavid Carlier   const Expr *LenArg = CE->getArg(2);
1498e75de21SDavid Carlier 
150*c30cedfcSDavid Carlier   const auto *DstArgDecl = dyn_cast<DeclRefExpr>(DstArg->IgnoreParenImpCasts());
1518e75de21SDavid Carlier   const auto *LenArgDecl = dyn_cast<DeclRefExpr>(LenArg->IgnoreParenLValueCasts());
1528e75de21SDavid Carlier   // - size_t dstlen = sizeof(dst)
1538e75de21SDavid Carlier   if (LenArgDecl) {
1548e75de21SDavid Carlier     const auto *LenArgVal = dyn_cast<VarDecl>(LenArgDecl->getDecl());
1558e75de21SDavid Carlier     if (LenArgVal->getInit())
1568e75de21SDavid Carlier       LenArg = LenArgVal->getInit();
1578e75de21SDavid Carlier   }
1588e75de21SDavid Carlier 
1598e75de21SDavid Carlier   // - integral value
1608e75de21SDavid Carlier   // We try to figure out if the last argument is possibly longer
1618e75de21SDavid Carlier   // than the destination can possibly handle if its size can be defined
162*c30cedfcSDavid Carlier   if (const auto *IL = dyn_cast<IntegerLiteral>(LenArg->IgnoreParenImpCasts())) {
1638e75de21SDavid Carlier     uint64_t ILRawVal = IL->getValue().getZExtValue();
164*c30cedfcSDavid Carlier     if (DstArgDecl) {
1658e75de21SDavid Carlier       if (const auto *Buffer = dyn_cast<ConstantArrayType>(DstArgDecl->getType())) {
1668e75de21SDavid Carlier         ASTContext &C = BR.getContext();
167*c30cedfcSDavid Carlier         uint64_t BufferLen = C.getTypeSize(Buffer) / 8;
1688e75de21SDavid Carlier         if (BufferLen < ILRawVal)
1698e75de21SDavid Carlier           return true;
1708e75de21SDavid Carlier       }
1718e75de21SDavid Carlier     }
172*c30cedfcSDavid Carlier   }
1738e75de21SDavid Carlier 
1748e75de21SDavid Carlier   return false;
1758e75de21SDavid Carlier }
1768e75de21SDavid Carlier 
17787b6ff09SAnna Zaks void WalkAST::VisitCallExpr(CallExpr *CE) {
17887b6ff09SAnna Zaks   const FunctionDecl *FD = CE->getDirectCallee();
17987b6ff09SAnna Zaks   if (!FD)
18087b6ff09SAnna Zaks     return;
18187b6ff09SAnna Zaks 
1820da67479SJordan Rose   if (CheckerContext::isCLibraryFunction(FD, "strncat")) {
18387b6ff09SAnna Zaks     if (containsBadStrncatPattern(CE)) {
18487b6ff09SAnna Zaks       const Expr *DstArg = CE->getArg(0);
18587b6ff09SAnna Zaks       const Expr *LenArg = CE->getArg(2);
18687b6ff09SAnna Zaks       PathDiagnosticLocation Loc =
18787b6ff09SAnna Zaks         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
18887b6ff09SAnna Zaks 
18987b6ff09SAnna Zaks       StringRef DstName = getPrintableName(DstArg);
19087b6ff09SAnna Zaks 
1912c1dd271SDylan Noblesmith       SmallString<256> S;
19287b6ff09SAnna Zaks       llvm::raw_svector_ostream os(S);
19387b6ff09SAnna Zaks       os << "Potential buffer overflow. ";
19487b6ff09SAnna Zaks       if (!DstName.empty()) {
19587b6ff09SAnna Zaks         os << "Replace with 'sizeof(" << DstName << ") "
19687b6ff09SAnna Zaks               "- strlen(" << DstName <<") - 1'";
19787b6ff09SAnna Zaks         os << " or u";
19887b6ff09SAnna Zaks       } else
19987b6ff09SAnna Zaks         os << "U";
20087b6ff09SAnna Zaks       os << "se a safer 'strlcat' API";
20187b6ff09SAnna Zaks 
2024aca9b1cSAlexander Kornienko       BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument",
2034aca9b1cSAlexander Kornienko                          "C String API", os.str(), Loc,
2044aca9b1cSAlexander Kornienko                          LenArg->getSourceRange());
20587b6ff09SAnna Zaks     }
2068e75de21SDavid Carlier   } else if (CheckerContext::isCLibraryFunction(FD, "strlcpy")) {
2078e75de21SDavid Carlier     if (containsBadStrlcpyPattern(CE)) {
2088e75de21SDavid Carlier       const Expr *DstArg = CE->getArg(0);
2098e75de21SDavid Carlier       const Expr *LenArg = CE->getArg(2);
2108e75de21SDavid Carlier       PathDiagnosticLocation Loc =
2118e75de21SDavid Carlier         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
2128e75de21SDavid Carlier 
2138e75de21SDavid Carlier       StringRef DstName = getPrintableName(DstArg);
2148e75de21SDavid Carlier 
2158e75de21SDavid Carlier       SmallString<256> S;
2168e75de21SDavid Carlier       llvm::raw_svector_ostream os(S);
2178e75de21SDavid Carlier       os << "The third argument is larger than the size of the input buffer. ";
2188e75de21SDavid Carlier       if (!DstName.empty())
2198e75de21SDavid Carlier         os << "Replace with the value 'sizeof(" << DstName << ")` or lower";
2208e75de21SDavid Carlier 
2218e75de21SDavid Carlier       BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument",
2228e75de21SDavid Carlier                          "C String API", os.str(), Loc,
2238e75de21SDavid Carlier                          LenArg->getSourceRange());
2248e75de21SDavid Carlier     }
22587b6ff09SAnna Zaks   }
22687b6ff09SAnna Zaks 
22787b6ff09SAnna Zaks   // Recurse and check children.
22887b6ff09SAnna Zaks   VisitChildren(CE);
22987b6ff09SAnna Zaks }
23087b6ff09SAnna Zaks 
23187b6ff09SAnna Zaks void WalkAST::VisitChildren(Stmt *S) {
232973431b2SBenjamin Kramer   for (Stmt *Child : S->children())
233973431b2SBenjamin Kramer     if (Child)
234973431b2SBenjamin Kramer       Visit(Child);
23587b6ff09SAnna Zaks }
23687b6ff09SAnna Zaks 
23787b6ff09SAnna Zaks namespace {
23887b6ff09SAnna Zaks class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
23987b6ff09SAnna Zaks public:
24087b6ff09SAnna Zaks 
24187b6ff09SAnna Zaks   void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
24287b6ff09SAnna Zaks       BugReporter &BR) const {
2434aca9b1cSAlexander Kornienko     WalkAST walker(this, BR, Mgr.getAnalysisDeclContext(D));
24487b6ff09SAnna Zaks     walker.Visit(D->getBody());
24587b6ff09SAnna Zaks   }
24687b6ff09SAnna Zaks };
247ab9db510SAlexander Kornienko }
24887b6ff09SAnna Zaks 
24987b6ff09SAnna Zaks void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
25087b6ff09SAnna Zaks   mgr.registerChecker<CStringSyntaxChecker>();
25187b6ff09SAnna Zaks }
25287b6ff09SAnna Zaks 
253