xref: /llvm-project/clang/lib/StaticAnalyzer/Checkers/CStringSyntaxChecker.cpp (revision 5a10f08b52eb8f6a9bedccd712026c9baae529df)
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/Analysis/AnalysisContext.h"
1787b6ff09SAnna Zaks #include "clang/AST/Expr.h"
1887b6ff09SAnna Zaks #include "clang/AST/OperationKinds.h"
1987b6ff09SAnna Zaks #include "clang/AST/StmtVisitor.h"
2087b6ff09SAnna Zaks #include "clang/Basic/TargetInfo.h"
2187b6ff09SAnna Zaks #include "clang/Basic/TypeTraits.h"
2287b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/Checker.h"
2387b6ff09SAnna Zaks #include "clang/StaticAnalyzer/Core/BugReporter/BugReporter.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> {
3487b6ff09SAnna Zaks   BugReporter &BR;
3587b6ff09SAnna Zaks   AnalysisDeclContext* AC;
3687b6ff09SAnna Zaks   ASTContext &ASTC;
3787b6ff09SAnna Zaks 
3887b6ff09SAnna Zaks   /// Check if two expressions refer to the same declaration.
3987b6ff09SAnna Zaks   inline bool sameDecl(const Expr *A1, const Expr *A2) {
4087b6ff09SAnna Zaks     if (const DeclRefExpr *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts()))
4187b6ff09SAnna Zaks       if (const DeclRefExpr *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).
4787b6ff09SAnna Zaks   inline bool isSizeof(const Expr *E, const Expr *WithArg) {
4887b6ff09SAnna Zaks     if (const UnaryExprOrTypeTraitExpr *UE =
4987b6ff09SAnna Zaks     dyn_cast<UnaryExprOrTypeTraitExpr>(E))
5087b6ff09SAnna Zaks       if (UE->getKind() == UETT_SizeOf)
5187b6ff09SAnna Zaks         return sameDecl(UE->getArgumentExpr(), WithArg);
5287b6ff09SAnna Zaks     return false;
5387b6ff09SAnna Zaks   }
5487b6ff09SAnna Zaks 
5587b6ff09SAnna Zaks   /// Check if the expression E is a strlen(WithArg).
5687b6ff09SAnna Zaks   inline bool isStrlen(const Expr *E, const Expr *WithArg) {
5787b6ff09SAnna Zaks     if (const CallExpr *CE = dyn_cast<CallExpr>(E)) {
5887b6ff09SAnna Zaks       const FunctionDecl *FD = CE->getDirectCallee();
5987b6ff09SAnna Zaks       if (!FD)
6087b6ff09SAnna Zaks         return false;
6187b6ff09SAnna Zaks       return (CheckerContext::isCLibraryFunction(FD, "strlen", ASTC)
6287b6ff09SAnna Zaks           && sameDecl(CE->getArg(0), WithArg));
6387b6ff09SAnna Zaks     }
6487b6ff09SAnna Zaks     return false;
6587b6ff09SAnna Zaks   }
6687b6ff09SAnna Zaks 
6787b6ff09SAnna Zaks   /// Check if the expression is an integer literal with value 1.
6887b6ff09SAnna Zaks   inline bool isOne(const Expr *E) {
6987b6ff09SAnna Zaks     if (const IntegerLiteral *IL = dyn_cast<IntegerLiteral>(E))
7087b6ff09SAnna Zaks       return (IL->getValue().isIntN(1));
7187b6ff09SAnna Zaks     return false;
7287b6ff09SAnna Zaks   }
7387b6ff09SAnna Zaks 
7487b6ff09SAnna Zaks   inline StringRef getPrintableName(const Expr *E) {
7587b6ff09SAnna Zaks     if (const DeclRefExpr *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts()))
7687b6ff09SAnna Zaks       return D->getDecl()->getName();
7787b6ff09SAnna Zaks     return StringRef();
7887b6ff09SAnna Zaks   }
7987b6ff09SAnna Zaks 
8087b6ff09SAnna Zaks   /// Identify erroneous patterns in the last argument to strncat - the number
8187b6ff09SAnna Zaks   /// of bytes to copy.
8287b6ff09SAnna Zaks   bool containsBadStrncatPattern(const CallExpr *CE);
8387b6ff09SAnna Zaks 
8487b6ff09SAnna Zaks public:
8587b6ff09SAnna Zaks   WalkAST(BugReporter &br, AnalysisDeclContext* ac) :
8687b6ff09SAnna Zaks       BR(br), AC(ac), ASTC(AC->getASTContext()) {
8787b6ff09SAnna Zaks   }
8887b6ff09SAnna Zaks 
8987b6ff09SAnna Zaks   // Statement visitor methods.
9087b6ff09SAnna Zaks   void VisitChildren(Stmt *S);
9187b6ff09SAnna Zaks   void VisitStmt(Stmt *S) {
9287b6ff09SAnna Zaks     VisitChildren(S);
9387b6ff09SAnna Zaks   }
9487b6ff09SAnna Zaks   void VisitCallExpr(CallExpr *CE);
9587b6ff09SAnna Zaks };
9687b6ff09SAnna Zaks } // end anonymous namespace
9787b6ff09SAnna Zaks 
9887b6ff09SAnna Zaks // The correct size argument should look like following:
9987b6ff09SAnna Zaks //   strncat(dst, src, sizeof(dst) - strlen(dest) - 1);
10087b6ff09SAnna Zaks // We look for the following anti-patterns:
10187b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - strlen(dst));
10287b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst) - 1);
10387b6ff09SAnna Zaks //   - strncat(dst, src, sizeof(dst));
10487b6ff09SAnna Zaks bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) {
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.
11087b6ff09SAnna Zaks   if (const BinaryOperator *BE =
11187b6ff09SAnna Zaks               dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) {
11287b6ff09SAnna Zaks     // - sizeof(dst) - strlen(dst)
11387b6ff09SAnna Zaks     if (BE->getOpcode() == BO_Sub) {
11487b6ff09SAnna Zaks       const Expr *L = BE->getLHS();
11587b6ff09SAnna Zaks       const Expr *R = BE->getRHS();
11687b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isStrlen(R, DstArg))
11787b6ff09SAnna Zaks         return true;
11887b6ff09SAnna Zaks 
11987b6ff09SAnna Zaks       // - sizeof(dst) - 1
12087b6ff09SAnna Zaks       if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts()))
12187b6ff09SAnna Zaks         return true;
12287b6ff09SAnna Zaks     }
12387b6ff09SAnna Zaks   }
12487b6ff09SAnna Zaks   // - sizeof(dst)
12587b6ff09SAnna Zaks   if (isSizeof(LenArg, DstArg))
12687b6ff09SAnna Zaks     return true;
12787b6ff09SAnna Zaks 
12887b6ff09SAnna Zaks   // - sizeof(src)
12987b6ff09SAnna Zaks   if (isSizeof(LenArg, SrcArg))
13087b6ff09SAnna Zaks     return true;
13187b6ff09SAnna Zaks   return false;
13287b6ff09SAnna Zaks }
13387b6ff09SAnna Zaks 
13487b6ff09SAnna Zaks void WalkAST::VisitCallExpr(CallExpr *CE) {
13587b6ff09SAnna Zaks   const FunctionDecl *FD = CE->getDirectCallee();
13687b6ff09SAnna Zaks   if (!FD)
13787b6ff09SAnna Zaks     return;
13887b6ff09SAnna Zaks 
13987b6ff09SAnna Zaks   if (CheckerContext::isCLibraryFunction(FD, "strncat", ASTC)) {
14087b6ff09SAnna Zaks     if (containsBadStrncatPattern(CE)) {
14187b6ff09SAnna Zaks       const Expr *DstArg = CE->getArg(0);
14287b6ff09SAnna Zaks       const Expr *LenArg = CE->getArg(2);
14387b6ff09SAnna Zaks       SourceRange R = LenArg->getSourceRange();
14487b6ff09SAnna Zaks       PathDiagnosticLocation Loc =
14587b6ff09SAnna Zaks         PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
14687b6ff09SAnna Zaks 
14787b6ff09SAnna Zaks       StringRef DstName = getPrintableName(DstArg);
14887b6ff09SAnna Zaks 
1492c1dd271SDylan Noblesmith       SmallString<256> S;
15087b6ff09SAnna Zaks       llvm::raw_svector_ostream os(S);
15187b6ff09SAnna Zaks       os << "Potential buffer overflow. ";
15287b6ff09SAnna Zaks       if (!DstName.empty()) {
15387b6ff09SAnna Zaks         os << "Replace with 'sizeof(" << DstName << ") "
15487b6ff09SAnna Zaks               "- strlen(" << DstName <<") - 1'";
15587b6ff09SAnna Zaks         os << " or u";
15687b6ff09SAnna Zaks       } else
15787b6ff09SAnna Zaks         os << "U";
15887b6ff09SAnna Zaks       os << "se a safer 'strlcat' API";
15987b6ff09SAnna Zaks 
160*5a10f08bSTed Kremenek       BR.EmitBasicReport(FD, "Anti-pattern in the argument", "C String API",
16187b6ff09SAnna Zaks                          os.str(), Loc, &R, 1);
16287b6ff09SAnna Zaks     }
16387b6ff09SAnna Zaks   }
16487b6ff09SAnna Zaks 
16587b6ff09SAnna Zaks   // Recurse and check children.
16687b6ff09SAnna Zaks   VisitChildren(CE);
16787b6ff09SAnna Zaks }
16887b6ff09SAnna Zaks 
16987b6ff09SAnna Zaks void WalkAST::VisitChildren(Stmt *S) {
17087b6ff09SAnna Zaks   for (Stmt::child_iterator I = S->child_begin(), E = S->child_end(); I != E;
17187b6ff09SAnna Zaks       ++I)
17287b6ff09SAnna Zaks     if (Stmt *child = *I)
17387b6ff09SAnna Zaks       Visit(child);
17487b6ff09SAnna Zaks }
17587b6ff09SAnna Zaks 
17687b6ff09SAnna Zaks namespace {
17787b6ff09SAnna Zaks class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
17887b6ff09SAnna Zaks public:
17987b6ff09SAnna Zaks 
18087b6ff09SAnna Zaks   void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
18187b6ff09SAnna Zaks       BugReporter &BR) const {
18287b6ff09SAnna Zaks     WalkAST walker(BR, Mgr.getAnalysisDeclContext(D));
18387b6ff09SAnna Zaks     walker.Visit(D->getBody());
18487b6ff09SAnna Zaks   }
18587b6ff09SAnna Zaks };
18687b6ff09SAnna Zaks }
18787b6ff09SAnna Zaks 
18887b6ff09SAnna Zaks void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
18987b6ff09SAnna Zaks   mgr.registerChecker<CStringSyntaxChecker>();
19087b6ff09SAnna Zaks }
19187b6ff09SAnna Zaks 
192