CStringSyntaxChecker.cpp revision 55fc873017f10f6f566b182b70f6fc22aefa3464 
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
29using namespace clang;
30using namespace ento;
31
32namespace {
33class WalkAST: public StmtVisitor<WalkAST> {
34  BugReporter &BR;
35  AnalysisDeclContext* AC;
36
37  /// Check if two expressions refer to the same declaration.
38  inline bool sameDecl(const Expr *A1, const Expr *A2) {
39    if (const DeclRefExpr *D1 = dyn_cast<DeclRefExpr>(A1->IgnoreParenCasts()))
40      if (const DeclRefExpr *D2 = dyn_cast<DeclRefExpr>(A2->IgnoreParenCasts()))
41        return D1->getDecl() == D2->getDecl();
42    return false;
43  }
44
45  /// Check if the expression E is a sizeof(WithArg).
46  inline bool isSizeof(const Expr *E, const Expr *WithArg) {
47    if (const UnaryExprOrTypeTraitExpr *UE =
48    dyn_cast<UnaryExprOrTypeTraitExpr>(E))
49      if (UE->getKind() == UETT_SizeOf)
50        return sameDecl(UE->getArgumentExpr(), WithArg);
51    return false;
52  }
53
54  /// Check if the expression E is a strlen(WithArg).
55  inline bool isStrlen(const Expr *E, const Expr *WithArg) {
56    if (const CallExpr *CE = dyn_cast<CallExpr>(E)) {
57      const FunctionDecl *FD = CE->getDirectCallee();
58      if (!FD)
59        return false;
60      return (CheckerContext::isCLibraryFunction(FD, "strlen") &&
61              sameDecl(CE->getArg(0), WithArg));
62    }
63    return false;
64  }
65
66  /// Check if the expression is an integer literal with value 1.
67  inline bool isOne(const Expr *E) {
68    if (const IntegerLiteral *IL = dyn_cast<IntegerLiteral>(E))
69      return (IL->getValue().isIntN(1));
70    return false;
71  }
72
73  inline StringRef getPrintableName(const Expr *E) {
74    if (const DeclRefExpr *D = dyn_cast<DeclRefExpr>(E->IgnoreParenCasts()))
75      return D->getDecl()->getName();
76    return StringRef();
77  }
78
79  /// Identify erroneous patterns in the last argument to strncat - the number
80  /// of bytes to copy.
81  bool containsBadStrncatPattern(const CallExpr *CE);
82
83public:
84  WalkAST(BugReporter &br, AnalysisDeclContext* ac) :
85      BR(br), AC(ac) {
86  }
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));
103bool WalkAST::containsBadStrncatPattern(const CallExpr *CE) {
104  const Expr *DstArg = CE->getArg(0);
105  const Expr *SrcArg = CE->getArg(1);
106  const Expr *LenArg = CE->getArg(2);
107
108  // Identify wrong size expressions, which are commonly used instead.
109  if (const BinaryOperator *BE =
110              dyn_cast<BinaryOperator>(LenArg->IgnoreParenCasts())) {
111    // - sizeof(dst) - strlen(dst)
112    if (BE->getOpcode() == BO_Sub) {
113      const Expr *L = BE->getLHS();
114      const Expr *R = BE->getRHS();
115      if (isSizeof(L, DstArg) && isStrlen(R, DstArg))
116        return true;
117
118      // - sizeof(dst) - 1
119      if (isSizeof(L, DstArg) && isOne(R->IgnoreParenCasts()))
120        return true;
121    }
122  }
123  // - sizeof(dst)
124  if (isSizeof(LenArg, DstArg))
125    return true;
126
127  // - sizeof(src)
128  if (isSizeof(LenArg, SrcArg))
129    return true;
130  return false;
131}
132
133void WalkAST::VisitCallExpr(CallExpr *CE) {
134  const FunctionDecl *FD = CE->getDirectCallee();
135  if (!FD)
136    return;
137
138  if (CheckerContext::isCLibraryFunction(FD, "strncat")) {
139    if (containsBadStrncatPattern(CE)) {
140      const Expr *DstArg = CE->getArg(0);
141      const Expr *LenArg = CE->getArg(2);
142      SourceRange R = LenArg->getSourceRange();
143      PathDiagnosticLocation Loc =
144        PathDiagnosticLocation::createBegin(LenArg, BR.getSourceManager(), AC);
145
146      StringRef DstName = getPrintableName(DstArg);
147
148      SmallString<256> S;
149      llvm::raw_svector_ostream os(S);
150      os << "Potential buffer overflow. ";
151      if (!DstName.empty()) {
152        os << "Replace with 'sizeof(" << DstName << ") "
153              "- strlen(" << DstName <<") - 1'";
154        os << " or u";
155      } else
156        os << "U";
157      os << "se a safer 'strlcat' API";
158
159      BR.EmitBasicReport(FD, "Anti-pattern in the argument", "C String API",
160                         os.str(), Loc, &R, 1);
161    }
162  }
163
164  // Recurse and check children.
165  VisitChildren(CE);
166}
167
168void WalkAST::VisitChildren(Stmt *S) {
169  for (Stmt::child_iterator I = S->child_begin(), E = S->child_end(); I != E;
170      ++I)
171    if (Stmt *child = *I)
172      Visit(child);
173}
174
175namespace {
176class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
177public:
178
179  void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
180      BugReporter &BR) const {
181    WalkAST walker(BR, Mgr.getAnalysisDeclContext(D));
182    walker.Visit(D->getBody());
183  }
184};
185}
186
187void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
188  mgr.registerChecker<CStringSyntaxChecker>();
189}
190
191