CStringSyntaxChecker.cpp revision 31b71f3097a338315a144067dde5b160c4e44fc9 
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  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
135void 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, "Anti-pattern in the argument", "C String API",
161                         os.str(), Loc, LenArg->getSourceRange());
162    }
163  }
164
165  // Recurse and check children.
166  VisitChildren(CE);
167}
168
169void WalkAST::VisitChildren(Stmt *S) {
170  for (Stmt::child_iterator I = S->child_begin(), E = S->child_end(); I != E;
171      ++I)
172    if (Stmt *child = *I)
173      Visit(child);
174}
175
176namespace {
177class CStringSyntaxChecker: public Checker<check::ASTCodeBody> {
178public:
179
180  void checkASTCodeBody(const Decl *D, AnalysisManager& Mgr,
181      BugReporter &BR) const {
182    WalkAST walker(BR, Mgr.getAnalysisDeclContext(D));
183    walker.Visit(D->getBody());
184  }
185};
186}
187
188void ento::registerCStringSyntaxChecker(CheckerManager &mgr) {
189  mgr.registerChecker<CStringSyntaxChecker>();
190}
191
192