xref: /llvm-project/clang/lib/StaticAnalyzer/Checkers/CStringSyntaxChecker.cpp (revision 2946cd701067404b99c39fb29dc9c74bd7193eb3)
187b6ff09SAnna Zaks //== CStringSyntaxChecker.cpp - CoreFoundation containers API *- C++ -*-==//
287b6ff09SAnna Zaks //
3*2946cd70SChandler Carruth // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4*2946cd70SChandler Carruth // See https://llvm.org/LICENSE.txt for license information.
5*2946cd70SChandler Carruth // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
687b6ff09SAnna Zaks //
787b6ff09SAnna Zaks //===----------------------------------------------------------------------===//
887b6ff09SAnna Zaks //
987b6ff09SAnna Zaks // An AST checker that looks for common pitfalls when using C string APIs.
1087b6ff09SAnna Zaks //  - Identifies erroneous patterns in the last argument to strncat - the number
1187b6ff09SAnna Zaks //    of bytes to copy.
1287b6ff09SAnna Zaks //
1387b6ff09SAnna Zaks //===----------------------------------------------------------------------===//
1476a21502SKristof Umann #include "clang/StaticAnalyzer/Checkers/BuiltinCheckerRegistration.h"
1587b6ff09SAnna Zaks #include "clang/AST/Expr.h"
1687b6ff09SAnna Zaks #include "clang/AST/OperationKinds.h"
1787b6ff09SAnna Zaks #include "clang/AST/StmtVisitor.h"
1850657f6bSGeorge Karpenkov #include "clang/Analysis/AnalysisDeclContext.h"
1987b6ff09SAnna Zaks #include "clang/Basic/TargetInfo.h"
2087b6ff09SAnna Zaks #include "clang/Basic/TypeTraits.h"
2187b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.h"
223a02247dSChandler Carruth #include "clang/StaticAnalyzer/Core/Checker.h"
2387b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/AnalysisManager.h"
2487b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
254903802fSBenjamin Kramer #include "llvm/ADT/SmallString.h"
2687b6ff09SAnna Zaks #include "llvm/Support/raw_ostream.h"
2787b6ff09SAnna Zaks 
2887b6ff09SAnna Zaks using namespace clang;
2987b6ff09SAnna Zaks using namespace ento;
3087b6ff09SAnna Zaks 
3187b6ff09SAnna Zaks namespace {
3287b6ff09SAnna Zaks class WalkAST: public StmtVisitor<WalkAST> {
334aca9b1cSAlexander Kornienko   const CheckerBase *Checker;
3487b6ff09SAnna Zaks   BugReporter &BR;
3587b6ff09SAnna Zaks   AnalysisDeclContext* AC;
3687b6ff09SAnna Zaks 
3787b6ff09SAnna Zaks   /// Check if two expressions refer to the same declaration.
383b008536SGabor Horvath   bool sameDecl(const Expr *A1, const Expr *A2) {
393b008536SGabor Horvath     if (const auto *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts()))
403b008536SGabor Horvath       if (const auto *D2 = dyn_cast<DeclRefExpr>(A2->IgnoreParenCasts()))
4187b6ff09SAnna Zaks         return D1->getDecl() == D2->getDecl();
4287b6ff09SAnna Zaks     return false;
4387b6ff09SAnna Zaks   }
4487b6ff09SAnna Zaks 
4587b6ff09SAnna Zaks   /// Check if the expression E is a sizeof(WithArg).
463b008536SGabor Horvath   bool isSizeof(const Expr *E, const Expr *WithArg) {
473b008536SGabor Horvath     if (const auto *UE = dyn_cast<UnaryExprOrTypeTraitExpr>(E))
483b008536SGabor Horvath       if (UE->getKind() == UETT_SizeOf && !UE->isArgumentType())
4987b6ff09SAnna Zaks         return sameDecl(UE->getArgumentExpr(), WithArg);
5087b6ff09SAnna Zaks     return false;
5187b6ff09SAnna Zaks   }
5287b6ff09SAnna Zaks 
5387b6ff09SAnna Zaks   /// Check if the expression E is a strlen(WithArg).
543b008536SGabor Horvath   bool isStrlen(const Expr *E, const Expr *WithArg) {
553b008536SGabor Horvath     if (const auto *CE = dyn_cast<CallExpr>(E)) {
5687b6ff09SAnna Zaks       const FunctionDecl *FD = CE->getDirectCallee();
5787b6ff09SAnna Zaks       if (!FD)
5887b6ff09SAnna Zaks         return false;
590da67479SJordan Rose       return (CheckerContext::isCLibraryFunction(FD, "strlen") &&
600da67479SJordan Rose               sameDecl(CE->getArg(0), WithArg));
6187b6ff09SAnna Zaks     }
6287b6ff09SAnna Zaks     return false;
6387b6ff09SAnna Zaks   }
6487b6ff09SAnna Zaks 
6587b6ff09SAnna Zaks   /// Check if the expression is an integer literal with value 1.
663b008536SGabor Horvath   bool isOne(const Expr *E) {
673b008536SGabor Horvath     if (const auto *IL = dyn_cast<IntegerLiteral>(E))
6887b6ff09SAnna Zaks       return (IL->getValue().isIntN(1));
6987b6ff09SAnna Zaks     return false;
7087b6ff09SAnna Zaks   }
7187b6ff09SAnna Zaks 
723b008536SGabor Horvath   StringRef getPrintableName(const Expr *E) {
733b008536SGabor Horvath     if (const auto *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts()))
7487b6ff09SAnna Zaks       return D->getDecl()->getName();
7587b6ff09SAnna Zaks     return StringRef();
7687b6ff09SAnna Zaks   }
7787b6ff09SAnna Zaks 
7887b6ff09SAnna Zaks   /// Identify erroneous patterns in the last argument to strncat - the number
7987b6ff09SAnna Zaks   /// of bytes to copy.
8087b6ff09SAnna Zaks   bool containsBadStrncatPattern(const CallExpr *CE);
8187b6ff09SAnna Zaks 
828e75de21SDavid Carlier   /// Identify erroneous patterns in the last argument to strlcpy - the number
838e75de21SDavid Carlier   /// of bytes to copy.
848e75de21SDavid Carlier   /// The bad pattern checked is when the size is known
858e75de21SDavid Carlier   /// to be larger than the destination can handle.
868e75de21SDavid Carlier   ///   char dst[2];
878e75de21SDavid Carlier   ///   size_t cpy = 4;
888e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", sizeof("abcd") - 1);
898e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", 4);
902ea81639SDavid Carlier   ///   strlcpy(dst + 3, "abcd", 2);
918e75de21SDavid Carlier   ///   strlcpy(dst, "abcd", cpy);
9275cb0dd5SDavid Carlier   /// Identify erroneous patterns in the last argument to strlcat - the number
9375cb0dd5SDavid Carlier   /// of bytes to copy.
9475cb0dd5SDavid Carlier   /// The bad pattern checked is when the last argument is basically
9575cb0dd5SDavid Carlier   /// pointing to the destination buffer size or argument larger or
9675cb0dd5SDavid Carlier   /// equal to.
9775cb0dd5SDavid Carlier   ///   char dst[2];
9875cb0dd5SDavid Carlier   ///   strlcat(dst, src2, sizeof(dst));
9975cb0dd5SDavid Carlier   ///   strlcat(dst, src2, 2);
10075cb0dd5SDavid Carlier   ///   strlcat(dst, src2, 10);
10175cb0dd5SDavid Carlier   bool containsBadStrlcpyStrlcatPattern(const CallExpr *CE);
1028e75de21SDavid Carlier 
10387b6ff09SAnna Zaks public:
1043b008536SGabor Horvath   WalkAST(const CheckerBase *Checker, BugReporter &BR, AnalysisDeclContext *AC)
1053b008536SGabor Horvath       : Checker(Checker), BR(BR), AC(AC) {}
10687b6ff09SAnna Zaks 
10787b6ff09SAnna Zaks   // Statement visitor methods.
10887b6ff09SAnna Zaks   void VisitChildren(Stmt *S);
10987b6ff09SAnna Zaks   void VisitStmt(Stmt *S) {
11087b6ff09SAnna Zaks     VisitChildren(S);
11187b6ff09SAnna Zaks   }
11287b6ff09SAnna Zaks   void VisitCallExpr(CallExpr *CE);
11387b6ff09SAnna Zaks };
11487b6ff09SAnna Zaks } // end anonymous namespace
11587b6ff09SAnna Zaks 
11687b6ff09SAnna Zaks // The correct size argument should look like following:
11787b6ff09SAnna Zaks //   strncat(dst, src, sizeof(dst) - strlen(dest) - 1);
11887b6ff09SAnna Zaks // We look for the following anti-patterns:
11987b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - strlen(dst));
12087b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - 1);
12187b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst));
12287b6ff09SAnna Zaks bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) {
12360d98befSAnna Zaks   if (CE->getNumArgs() != 3)
12460d98befSAnna Zaks     return false;
12587b6ff09SAnna Zaks   const Expr *DstArg = CE->getArg(0);
12687b6ff09SAnna Zaks   const Expr *SrcArg = CE->getArg(1);
12787b6ff09SAnna Zaks   const Expr *LenArg = CE->getArg(2);
12887b6ff09SAnna Zaks 
12987b6ff09SAnna Zaks   // Identify wrong size expressions, which are commonly used instead.
1303b008536SGabor Horvath   if (const auto *BE = dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) {
13187b6ff09SAnna Zaks     // - sizeof(dst) - strlen(dst)
13287b6ff09SAnna Zaks     if (BE->getOpcode() == BO_Sub) {
13387b6ff09SAnna Zaks       const Expr *L = BE->getLHS();
13487b6ff09SAnna Zaks       const Expr *R = BE->getRHS();
13587b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isStrlen(R, DstArg))
13687b6ff09SAnna Zaks         return true;
13787b6ff09SAnna Zaks 
13887b6ff09SAnna Zaks       // - sizeof(dst) - 1
13987b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts()))
14087b6ff09SAnna Zaks         return true;
14187b6ff09SAnna Zaks     }
14287b6ff09SAnna Zaks   }
14387b6ff09SAnna Zaks   // - sizeof(dst)
14487b6ff09SAnna Zaks   if (isSizeof(LenArg, DstArg))
14587b6ff09SAnna Zaks     return true;
14687b6ff09SAnna Zaks 
14787b6ff09SAnna Zaks   // - sizeof(src)
14887b6ff09SAnna Zaks   if (isSizeof(LenArg, SrcArg))
14987b6ff09SAnna Zaks     return true;
15087b6ff09SAnna Zaks   return false;
15187b6ff09SAnna Zaks }
15287b6ff09SAnna Zaks 
15375cb0dd5SDavid Carlier bool WalkAST::containsBadStrlcpyStrlcatPattern(const CallExpr *CE) {
1548e75de21SDavid Carlier   if (CE->getNumArgs() != 3)
1558e75de21SDavid Carlier     return false;
15675cb0dd5SDavid Carlier   const FunctionDecl *FD = CE->getDirectCallee();
15775cb0dd5SDavid Carlier   bool Append = CheckerContext::isCLibraryFunction(FD, "strlcat");
1588e75de21SDavid Carlier   const Expr *DstArg = CE->getArg(0);
1598e75de21SDavid Carlier   const Expr *LenArg = CE->getArg(2);
1608e75de21SDavid Carlier 
161c30cedfcSDavid Carlier   const auto *DstArgDecl = dyn_cast<DeclRefExpr>(DstArg->IgnoreParenImpCasts());
1628e75de21SDavid Carlier   const auto *LenArgDecl = dyn_cast<DeclRefExpr>(LenArg->IgnoreParenLValueCasts());
1632ea81639SDavid Carlier   uint64_t DstOff = 0;
16475cb0dd5SDavid Carlier   if (isSizeof(LenArg, DstArg))
16575cb0dd5SDavid Carlier     return false;
1668e75de21SDavid Carlier   // - size_t dstlen = sizeof(dst)
1678e75de21SDavid Carlier   if (LenArgDecl) {
1688e75de21SDavid Carlier     const auto *LenArgVal = dyn_cast<VarDecl>(LenArgDecl->getDecl());
1698e75de21SDavid Carlier     if (LenArgVal->getInit())
1708e75de21SDavid Carlier       LenArg = LenArgVal->getInit();
1718e75de21SDavid Carlier   }
1728e75de21SDavid Carlier 
1738e75de21SDavid Carlier   // - integral value
1748e75de21SDavid Carlier   // We try to figure out if the last argument is possibly longer
1752ea81639SDavid Carlier   // than the destination can possibly handle if its size can be defined.
176c30cedfcSDavid Carlier   if (const auto *IL = dyn_cast<IntegerLiteral>(LenArg->IgnoreParenImpCasts())) {
1778e75de21SDavid Carlier     uint64_t ILRawVal = IL->getValue().getZExtValue();
1782ea81639SDavid Carlier 
1792ea81639SDavid Carlier     // Case when there is pointer arithmetic on the destination buffer
1802ea81639SDavid Carlier     // especially when we offset from the base decreasing the
1812ea81639SDavid Carlier     // buffer length accordingly.
1822ea81639SDavid Carlier     if (!DstArgDecl) {
1832ea81639SDavid Carlier       if (const auto *BE = dyn_cast<BinaryOperator>(DstArg->IgnoreParenImpCasts())) {
1842ea81639SDavid Carlier         DstArgDecl = dyn_cast<DeclRefExpr>(BE->getLHS()->IgnoreParenImpCasts());
1852ea81639SDavid Carlier         if (BE->getOpcode() == BO_Add) {
1862ea81639SDavid Carlier           if ((IL = dyn_cast<IntegerLiteral>(BE->getRHS()->IgnoreParenImpCasts()))) {
1872ea81639SDavid Carlier             DstOff = IL->getValue().getZExtValue();
1882ea81639SDavid Carlier           }
1892ea81639SDavid Carlier         }
1902ea81639SDavid Carlier       }
1912ea81639SDavid Carlier     }
192c30cedfcSDavid Carlier     if (DstArgDecl) {
1938e75de21SDavid Carlier       if (const auto *Buffer = dyn_cast<ConstantArrayType>(DstArgDecl->getType())) {
1948e75de21SDavid Carlier         ASTContext &C = BR.getContext();
195c30cedfcSDavid Carlier         uint64_t BufferLen = C.getTypeSize(Buffer) / 8;
19675cb0dd5SDavid Carlier         auto RemainingBufferLen = BufferLen - DstOff;
19775cb0dd5SDavid Carlier         if (Append) {
19875cb0dd5SDavid Carlier           if (RemainingBufferLen <= ILRawVal)
1998e75de21SDavid Carlier             return true;
20075cb0dd5SDavid Carlier         } else {
20175cb0dd5SDavid Carlier           if (RemainingBufferLen < ILRawVal)
20275cb0dd5SDavid Carlier             return true;
20375cb0dd5SDavid Carlier         }
2048e75de21SDavid Carlier       }
2058e75de21SDavid Carlier     }
206c30cedfcSDavid Carlier   }
2078e75de21SDavid Carlier 
2088e75de21SDavid Carlier   return false;
2098e75de21SDavid Carlier }
2108e75de21SDavid Carlier 
21187b6ff09SAnna Zaks void WalkAST::VisitCallExpr(CallExpr *CE) {
21287b6ff09SAnna Zaks   const FunctionDecl *FD = CE->getDirectCallee();
21387b6ff09SAnna Zaks   if (!FD)
21487b6ff09SAnna Zaks     return;
21587b6ff09SAnna Zaks 
2160da67479SJordan Rose   if (CheckerContext::isCLibraryFunction(FD, "strncat")) {
21787b6ff09SAnna Zaks     if (containsBadStrncatPattern(CE)) {
21887b6ff09SAnna Zaks       const Expr *DstArg = CE->getArg(0);
21987b6ff09SAnna Zaks       const Expr *LenArg = CE->getArg(2);
22087b6ff09SAnna Zaks       PathDiagnosticLocation Loc =
22187b6ff09SAnna Zaks         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
22287b6ff09SAnna Zaks 
22387b6ff09SAnna Zaks       StringRef DstName = getPrintableName(DstArg);
22487b6ff09SAnna Zaks 
2252c1dd271SDylan Noblesmith       SmallString<256> S;
22687b6ff09SAnna Zaks       llvm::raw_svector_ostream os(S);
22787b6ff09SAnna Zaks       os << "Potential buffer overflow. ";
22887b6ff09SAnna Zaks       if (!DstName.empty()) {
22987b6ff09SAnna Zaks         os << "Replace with 'sizeof(" << DstName << ") "
23087b6ff09SAnna Zaks               "- strlen(" << DstName <<") - 1'";
23187b6ff09SAnna Zaks         os << " or u";
23287b6ff09SAnna Zaks       } else
23387b6ff09SAnna Zaks         os << "U";
23487b6ff09SAnna Zaks       os << "se a safer 'strlcat' API";
23587b6ff09SAnna Zaks 
2364aca9b1cSAlexander Kornienko       BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument",
2374aca9b1cSAlexander Kornienko                          "C String API", os.str(), Loc,
2384aca9b1cSAlexander Kornienko                          LenArg->getSourceRange());
23987b6ff09SAnna Zaks     }
24075cb0dd5SDavid Carlier   } else if (CheckerContext::isCLibraryFunction(FD, "strlcpy") ||
24175cb0dd5SDavid Carlier              CheckerContext::isCLibraryFunction(FD, "strlcat")) {
24275cb0dd5SDavid Carlier     if (containsBadStrlcpyStrlcatPattern(CE)) {
2438e75de21SDavid Carlier       const Expr *DstArg = CE->getArg(0);
2448e75de21SDavid Carlier       const Expr *LenArg = CE->getArg(2);
2458e75de21SDavid Carlier       PathDiagnosticLocation Loc =
2468e75de21SDavid Carlier         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
2478e75de21SDavid Carlier 
2488e75de21SDavid Carlier       StringRef DstName = getPrintableName(DstArg);
2498e75de21SDavid Carlier 
2508e75de21SDavid Carlier       SmallString<256> S;
2518e75de21SDavid Carlier       llvm::raw_svector_ostream os(S);
25275cb0dd5SDavid Carlier       os << "The third argument allows to potentially copy more bytes than it should. ";
25375cb0dd5SDavid Carlier       os << "Replace with the value ";
2548e75de21SDavid Carlier       if (!DstName.empty())
25575cb0dd5SDavid Carlier           os << "sizeof(" << DstName << ")";
25675cb0dd5SDavid Carlier       else
25775cb0dd5SDavid Carlier           os << "sizeof(<destination buffer>)";
25875cb0dd5SDavid Carlier       os << " or lower";
2598e75de21SDavid Carlier 
2608e75de21SDavid Carlier       BR.EmitBasicReport(FD, Checker, "Anti-pattern in the argument",
2618e75de21SDavid Carlier               "C String API", os.str(), Loc,
2628e75de21SDavid Carlier               LenArg->getSourceRange());
2638e75de21SDavid Carlier     }
26487b6ff09SAnna Zaks   }
26587b6ff09SAnna Zaks 
26687b6ff09SAnna Zaks   // Recurse and check children.
26787b6ff09SAnna Zaks   VisitChildren(CE);
26887b6ff09SAnna Zaks }
26987b6ff09SAnna Zaks 
27087b6ff09SAnna Zaks void WalkAST::VisitChildren(Stmt *S) {
271973431b2SBenjamin Kramer   for (Stmt *Child : S->children())
272973431b2SBenjamin Kramer     if (Child)
273973431b2SBenjamin Kramer       Visit(Child);
27487b6ff09SAnna Zaks }
27587b6ff09SAnna Zaks 
27687b6ff09SAnna Zaks namespace {
27787b6ff09SAnna Zaks class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
27887b6ff09SAnna Zaks public:
27987b6ff09SAnna Zaks 
28087b6ff09SAnna Zaks   void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
28187b6ff09SAnna Zaks       BugReporter &BR) const {
2824aca9b1cSAlexander Kornienko     WalkAST walker(this, BR, Mgr.getAnalysisDeclContext(D));
28387b6ff09SAnna Zaks     walker.Visit(D->getBody());
28487b6ff09SAnna Zaks   }
28587b6ff09SAnna Zaks };
286ab9db510SAlexander Kornienko }
28787b6ff09SAnna Zaks 
28887b6ff09SAnna Zaks void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
28987b6ff09SAnna Zaks   mgr.registerChecker<CStringSyntaxChecker>();
29087b6ff09SAnna Zaks }
29187b6ff09SAnna Zaks 
292