blob: afb862cd6c9a9bcb7320010b9b14c2ae0b7a064c [file] [log] [blame]
Anna Zaksdf18c5a2011-11-16 19:58:13 +00001//== GenericTaintChecker.cpp ----------------------------------- -*- 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// This checker defines the attack surface for generic taint propagation.
11//
12// The taint information produced by it might be useful to other checkers. For
13// example, checkers should report errors which involve tainted data more
14// aggressively, even if the involved symbols are under constrained.
15//
16//===----------------------------------------------------------------------===//
17#include "ClangSACheckers.h"
18#include "clang/StaticAnalyzer/Core/Checker.h"
19#include "clang/StaticAnalyzer/Core/CheckerManager.h"
20#include "clang/StaticAnalyzer/Core/PathSensitive/CheckerContext.h"
Anna Zaks9ffbe242011-12-17 00:26:34 +000021#include "clang/StaticAnalyzer/Core/PathSensitive/ProgramStateTrait.h"
Anna Zaksdf18c5a2011-11-16 19:58:13 +000022#include "clang/StaticAnalyzer/Core/BugReporter/BugType.h"
Anna Zaks9b0c7492012-01-18 02:45:07 +000023#include "clang/Basic/Builtins.h"
Anna Zaks1fb826a2012-01-12 02:22:34 +000024#include <climits>
Anna Zaksdf18c5a2011-11-16 19:58:13 +000025
26using namespace clang;
27using namespace ento;
28
29namespace {
Anna Zaksefd69892011-12-14 00:56:18 +000030class GenericTaintChecker : public Checker< check::PostStmt<CallExpr>,
Anna Zaks9ffbe242011-12-17 00:26:34 +000031 check::PreStmt<CallExpr> > {
32public:
Anna Zaks8568ee72012-01-14 02:48:40 +000033 static void *getTag() { static int Tag; return &Tag; }
34
35 void checkPostStmt(const CallExpr *CE, CheckerContext &C) const;
36 void checkPostStmt(const DeclRefExpr *DRE, CheckerContext &C) const;
37
38 void checkPreStmt(const CallExpr *CE, CheckerContext &C) const;
Anna Zaksdf18c5a2011-11-16 19:58:13 +000039
Anna Zaks9ffbe242011-12-17 00:26:34 +000040private:
Anna Zaksb9ac30c2012-01-24 19:32:25 +000041 static const unsigned InvalidArgIndex = UINT_MAX;
42 /// Denotes the return vale.
43 static const unsigned ReturnValueIndex = UINT_MAX - 1;
Anna Zaks8568ee72012-01-14 02:48:40 +000044
Dylan Noblesmith6f42b622012-02-05 02:12:40 +000045 mutable OwningPtr<BugType> BT;
Anna Zaks9b0c7492012-01-18 02:45:07 +000046 inline void initBugType() const {
47 if (!BT)
Anna Zaks5fdadf42012-02-22 02:35:58 +000048 BT.reset(new BugType("Use of Untrusted Data", "Untrusted Data"));
Anna Zaks9b0c7492012-01-18 02:45:07 +000049 }
Anna Zaks8f4caf52011-11-18 02:26:36 +000050
Anna Zaks1fb826a2012-01-12 02:22:34 +000051 /// \brief Catch taint related bugs. Check if tainted data is passed to a
52 /// system call etc.
Anna Zaks9f03b622012-01-07 02:33:10 +000053 bool checkPre(const CallExpr *CE, CheckerContext &C) const;
54
Anna Zaks1fb826a2012-01-12 02:22:34 +000055 /// \brief Add taint sources on a pre-visit.
56 void addSourcesPre(const CallExpr *CE, CheckerContext &C) const;
57
58 /// \brief Propagate taint generated at pre-visit.
59 bool propagateFromPre(const CallExpr *CE, CheckerContext &C) const;
60
61 /// \brief Add taint sources on a post visit.
62 void addSourcesPost(const CallExpr *CE, CheckerContext &C) const;
63
Anna Zaksb9ac30c2012-01-24 19:32:25 +000064 /// Check if the region the expression evaluates to is the standard input,
65 /// and thus, is tainted.
66 static bool isStdin(const Expr *E, CheckerContext &C);
67
Anna Zaks1fb826a2012-01-12 02:22:34 +000068 /// \brief Given a pointer argument, get the symbol of the value it contains
Anna Zaks8f4caf52011-11-18 02:26:36 +000069 /// (points to).
Anna Zaks7cdfe292012-01-18 02:45:13 +000070 static SymbolRef getPointedToSymbol(CheckerContext &C, const Expr *Arg);
Anna Zaksdf18c5a2011-11-16 19:58:13 +000071
Anna Zaks9ffbe242011-12-17 00:26:34 +000072 /// Functions defining the attack surface.
Ted Kremenek8bef8232012-01-26 21:29:00 +000073 typedef ProgramStateRef (GenericTaintChecker::*FnCheck)(const CallExpr *,
Anna Zaks9ffbe242011-12-17 00:26:34 +000074 CheckerContext &C) const;
Ted Kremenek8bef8232012-01-26 21:29:00 +000075 ProgramStateRef postScanf(const CallExpr *CE, CheckerContext &C) const;
76 ProgramStateRef postSocket(const CallExpr *CE, CheckerContext &C) const;
77 ProgramStateRef postRetTaint(const CallExpr *CE, CheckerContext &C) const;
Anna Zaks9ffbe242011-12-17 00:26:34 +000078
79 /// Taint the scanned input if the file is tainted.
Ted Kremenek8bef8232012-01-26 21:29:00 +000080 ProgramStateRef preFscanf(const CallExpr *CE, CheckerContext &C) const;
Anna Zaksdf18c5a2011-11-16 19:58:13 +000081
Anna Zaks9f03b622012-01-07 02:33:10 +000082 /// Check for CWE-134: Uncontrolled Format String.
Anna Zaks8568ee72012-01-14 02:48:40 +000083 static const char MsgUncontrolledFormatString[];
Anna Zaks9f03b622012-01-07 02:33:10 +000084 bool checkUncontrolledFormatString(const CallExpr *CE,
85 CheckerContext &C) const;
86
Anna Zaks8568ee72012-01-14 02:48:40 +000087 /// Check for:
88 /// CERT/STR02-C. "Sanitize data passed to complex subsystems"
89 /// CWE-78, "Failure to Sanitize Data into an OS Command"
90 static const char MsgSanitizeSystemArgs[];
91 bool checkSystemCall(const CallExpr *CE, StringRef Name,
92 CheckerContext &C) const;
Anna Zaks9ffbe242011-12-17 00:26:34 +000093
Anna Zaks4e462212012-01-18 02:45:11 +000094 /// Check if tainted data is used as a buffer size ins strn.. functions,
95 /// and allocators.
96 static const char MsgTaintedBufferSize[];
97 bool checkTaintedBufferSize(const CallExpr *CE, const FunctionDecl *FDecl,
98 CheckerContext &C) const;
99
Anna Zaks8568ee72012-01-14 02:48:40 +0000100 /// Generate a report if the expression is tainted or points to tainted data.
101 bool generateReportIfTainted(const Expr *E, const char Msg[],
102 CheckerContext &C) const;
Anna Zaks022b3f42012-01-17 00:37:02 +0000103
104
105 typedef llvm::SmallVector<unsigned, 2> ArgVector;
Anna Zaks9ffbe242011-12-17 00:26:34 +0000106
Anna Zaks022b3f42012-01-17 00:37:02 +0000107 /// \brief A struct used to specify taint propagation rules for a function.
108 ///
109 /// If any of the possible taint source arguments is tainted, all of the
110 /// destination arguments should also be tainted. Use InvalidArgIndex in the
111 /// src list to specify that all of the arguments can introduce taint. Use
112 /// InvalidArgIndex in the dst arguments to signify that all the non-const
113 /// pointer and reference arguments might be tainted on return. If
114 /// ReturnValueIndex is added to the dst list, the return value will be
115 /// tainted.
116 struct TaintPropagationRule {
117 /// List of arguments which can be taint sources and should be checked.
118 ArgVector SrcArgs;
119 /// List of arguments which should be tainted on function return.
120 ArgVector DstArgs;
Anna Zaks9b0c7492012-01-18 02:45:07 +0000121 // TODO: Check if using other data structures would be more optimal.
Anna Zaks022b3f42012-01-17 00:37:02 +0000122
123 TaintPropagationRule() {}
124
Anna Zaks9b0c7492012-01-18 02:45:07 +0000125 TaintPropagationRule(unsigned SArg,
126 unsigned DArg, bool TaintRet = false) {
Anna Zaks022b3f42012-01-17 00:37:02 +0000127 SrcArgs.push_back(SArg);
128 DstArgs.push_back(DArg);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000129 if (TaintRet)
130 DstArgs.push_back(ReturnValueIndex);
Anna Zaks022b3f42012-01-17 00:37:02 +0000131 }
132
Anna Zaks9b0c7492012-01-18 02:45:07 +0000133 TaintPropagationRule(unsigned SArg1, unsigned SArg2,
134 unsigned DArg, bool TaintRet = false) {
135 SrcArgs.push_back(SArg1);
136 SrcArgs.push_back(SArg2);
137 DstArgs.push_back(DArg);
138 if (TaintRet)
139 DstArgs.push_back(ReturnValueIndex);
140 }
141
142 /// Get the propagation rule for a given function.
143 static TaintPropagationRule
144 getTaintPropagationRule(const FunctionDecl *FDecl,
145 StringRef Name,
146 CheckerContext &C);
147
Anna Zaks022b3f42012-01-17 00:37:02 +0000148 inline void addSrcArg(unsigned A) { SrcArgs.push_back(A); }
149 inline void addDstArg(unsigned A) { DstArgs.push_back(A); }
150
Anna Zaks9b0c7492012-01-18 02:45:07 +0000151 inline bool isNull() const { return SrcArgs.empty(); }
152
153 inline bool isDestinationArgument(unsigned ArgNum) const {
154 return (std::find(DstArgs.begin(),
155 DstArgs.end(), ArgNum) != DstArgs.end());
156 }
Anna Zaks022b3f42012-01-17 00:37:02 +0000157
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000158 static inline bool isTaintedOrPointsToTainted(const Expr *E,
Ted Kremenek8bef8232012-01-26 21:29:00 +0000159 ProgramStateRef State,
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000160 CheckerContext &C) {
161 return (State->isTainted(E, C.getLocationContext()) || isStdin(E, C) ||
162 (E->getType().getTypePtr()->isPointerType() &&
163 State->isTainted(getPointedToSymbol(C, E))));
164 }
165
Anna Zaks7cdfe292012-01-18 02:45:13 +0000166 /// \brief Pre-process a function which propagates taint according to the
167 /// taint rule.
Ted Kremenek8bef8232012-01-26 21:29:00 +0000168 ProgramStateRef process(const CallExpr *CE, CheckerContext &C) const;
Anna Zaks7cdfe292012-01-18 02:45:13 +0000169
170 };
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000171};
Anna Zaks9b0c7492012-01-18 02:45:07 +0000172
173const unsigned GenericTaintChecker::ReturnValueIndex;
174const unsigned GenericTaintChecker::InvalidArgIndex;
175
Anna Zaks8568ee72012-01-14 02:48:40 +0000176const char GenericTaintChecker::MsgUncontrolledFormatString[] =
Anna Zaks5fdadf42012-02-22 02:35:58 +0000177 "Untrusted data is used as a format string "
178 "(CWE-134: Uncontrolled Format String)";
Anna Zaks8568ee72012-01-14 02:48:40 +0000179
180const char GenericTaintChecker::MsgSanitizeSystemArgs[] =
Anna Zaks5fdadf42012-02-22 02:35:58 +0000181 "Untrusted data is passed to a system call "
Anna Zaks8568ee72012-01-14 02:48:40 +0000182 "(CERT/STR02-C. Sanitize data passed to complex subsystems)";
Anna Zaks4e462212012-01-18 02:45:11 +0000183
184const char GenericTaintChecker::MsgTaintedBufferSize[] =
Anna Zaks5fdadf42012-02-22 02:35:58 +0000185 "Untrusted data is used to specify the buffer size "
Anna Zaks4e462212012-01-18 02:45:11 +0000186 "(CERT/STR31-C. Guarantee that storage for strings has sufficient space for "
187 "character data and the null terminator)";
188
189} // end of anonymous namespace
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000190
Anna Zaks1fb826a2012-01-12 02:22:34 +0000191/// A set which is used to pass information from call pre-visit instruction
192/// to the call post-visit. The values are unsigned integers, which are either
193/// ReturnValueIndex, or indexes of the pointer/reference argument, which
194/// points to data, which should be tainted on return.
195namespace { struct TaintArgsOnPostVisit{}; }
196namespace clang { namespace ento {
197template<> struct ProgramStateTrait<TaintArgsOnPostVisit>
198 : public ProgramStatePartialTrait<llvm::ImmutableSet<unsigned> > {
199 static void *GDMIndex() { return GenericTaintChecker::getTag(); }
200};
201}}
Anna Zaks9ffbe242011-12-17 00:26:34 +0000202
Anna Zaks9b0c7492012-01-18 02:45:07 +0000203GenericTaintChecker::TaintPropagationRule
204GenericTaintChecker::TaintPropagationRule::getTaintPropagationRule(
205 const FunctionDecl *FDecl,
206 StringRef Name,
207 CheckerContext &C) {
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000208 // TODO: Currently, we might loose precision here: we always mark a return
209 // value as tainted even if it's just a pointer, pointing to tainted data.
210
Anna Zaks9b0c7492012-01-18 02:45:07 +0000211 // Check for exact name match for functions without builtin substitutes.
212 TaintPropagationRule Rule = llvm::StringSwitch<TaintPropagationRule>(Name)
213 .Case("atoi", TaintPropagationRule(0, ReturnValueIndex))
214 .Case("atol", TaintPropagationRule(0, ReturnValueIndex))
215 .Case("atoll", TaintPropagationRule(0, ReturnValueIndex))
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000216 .Case("getc", TaintPropagationRule(0, ReturnValueIndex))
217 .Case("fgetc", TaintPropagationRule(0, ReturnValueIndex))
218 .Case("getc_unlocked", TaintPropagationRule(0, ReturnValueIndex))
219 .Case("getw", TaintPropagationRule(0, ReturnValueIndex))
220 .Case("toupper", TaintPropagationRule(0, ReturnValueIndex))
221 .Case("tolower", TaintPropagationRule(0, ReturnValueIndex))
222 .Case("strchr", TaintPropagationRule(0, ReturnValueIndex))
223 .Case("strrchr", TaintPropagationRule(0, ReturnValueIndex))
Anna Zaks2bf8fd82012-01-20 00:11:19 +0000224 .Case("read", TaintPropagationRule(0, 2, 1, true))
225 .Case("pread", TaintPropagationRule(InvalidArgIndex, 1, true))
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000226 .Case("gets", TaintPropagationRule(InvalidArgIndex, 0, true))
227 .Case("fgets", TaintPropagationRule(2, 0, true))
228 .Case("getline", TaintPropagationRule(2, 0))
229 .Case("getdelim", TaintPropagationRule(3, 0))
230 .Case("fgetln", TaintPropagationRule(0, ReturnValueIndex))
Anna Zaks9b0c7492012-01-18 02:45:07 +0000231 .Default(TaintPropagationRule());
232
233 if (!Rule.isNull())
234 return Rule;
235
236 // Check if it's one of the memory setting/copying functions.
237 // This check is specialized but faster then calling isCLibraryFunction.
238 unsigned BId = 0;
239 if ( (BId = FDecl->getMemoryFunctionKind()) )
240 switch(BId) {
241 case Builtin::BImemcpy:
242 case Builtin::BImemmove:
243 case Builtin::BIstrncpy:
244 case Builtin::BIstrncat:
245 return TaintPropagationRule(1, 2, 0, true);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000246 case Builtin::BIstrlcpy:
247 case Builtin::BIstrlcat:
248 return TaintPropagationRule(1, 2, 0, false);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000249 case Builtin::BIstrndup:
250 return TaintPropagationRule(0, 1, ReturnValueIndex);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000251
252 default:
253 break;
254 };
255
256 // Process all other functions which could be defined as builtins.
257 if (Rule.isNull()) {
258 if (C.isCLibraryFunction(FDecl, "snprintf") ||
259 C.isCLibraryFunction(FDecl, "sprintf"))
260 return TaintPropagationRule(InvalidArgIndex, 0, true);
261 else if (C.isCLibraryFunction(FDecl, "strcpy") ||
262 C.isCLibraryFunction(FDecl, "stpcpy") ||
263 C.isCLibraryFunction(FDecl, "strcat"))
264 return TaintPropagationRule(1, 0, true);
265 else if (C.isCLibraryFunction(FDecl, "bcopy"))
266 return TaintPropagationRule(0, 2, 1, false);
267 else if (C.isCLibraryFunction(FDecl, "strdup") ||
268 C.isCLibraryFunction(FDecl, "strdupa"))
269 return TaintPropagationRule(0, ReturnValueIndex);
Anna Zaks4e462212012-01-18 02:45:11 +0000270 else if (C.isCLibraryFunction(FDecl, "wcsdup"))
271 return TaintPropagationRule(0, ReturnValueIndex);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000272 }
273
274 // Skipping the following functions, since they might be used for cleansing
275 // or smart memory copy:
Benjamin Kramer48d798c2012-06-02 10:20:41 +0000276 // - memccpy - copying until hitting a special character.
Anna Zaks9b0c7492012-01-18 02:45:07 +0000277
278 return TaintPropagationRule();
Anna Zaks8f4caf52011-11-18 02:26:36 +0000279}
280
Anna Zaks9ffbe242011-12-17 00:26:34 +0000281void GenericTaintChecker::checkPreStmt(const CallExpr *CE,
282 CheckerContext &C) const {
Anna Zaks9f03b622012-01-07 02:33:10 +0000283 // Check for errors first.
284 if (checkPre(CE, C))
285 return;
Anna Zaks9ffbe242011-12-17 00:26:34 +0000286
Anna Zaks9f03b622012-01-07 02:33:10 +0000287 // Add taint second.
Anna Zaks1fb826a2012-01-12 02:22:34 +0000288 addSourcesPre(CE, C);
Anna Zaks9f03b622012-01-07 02:33:10 +0000289}
290
291void GenericTaintChecker::checkPostStmt(const CallExpr *CE,
292 CheckerContext &C) const {
Anna Zaks1fb826a2012-01-12 02:22:34 +0000293 if (propagateFromPre(CE, C))
294 return;
295 addSourcesPost(CE, C);
Anna Zaks9f03b622012-01-07 02:33:10 +0000296}
297
Anna Zaks1fb826a2012-01-12 02:22:34 +0000298void GenericTaintChecker::addSourcesPre(const CallExpr *CE,
299 CheckerContext &C) const {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000300 ProgramStateRef State = 0;
Anna Zaks9b0c7492012-01-18 02:45:07 +0000301 const FunctionDecl *FDecl = C.getCalleeDecl(CE);
Jordan Rose5ef6e942012-07-10 23:13:01 +0000302 if (!FDecl || FDecl->getKind() != Decl::Function)
303 return;
304
Anna Zaks9b0c7492012-01-18 02:45:07 +0000305 StringRef Name = C.getCalleeName(FDecl);
Anna Zaks1fb826a2012-01-12 02:22:34 +0000306 if (Name.empty())
307 return;
Anna Zaks022b3f42012-01-17 00:37:02 +0000308
Anna Zaks9b0c7492012-01-18 02:45:07 +0000309 // First, try generating a propagation rule for this function.
310 TaintPropagationRule Rule =
311 TaintPropagationRule::getTaintPropagationRule(FDecl, Name, C);
Anna Zaks022b3f42012-01-17 00:37:02 +0000312 if (!Rule.isNull()) {
Anna Zaks7cdfe292012-01-18 02:45:13 +0000313 State = Rule.process(CE, C);
Anna Zaks022b3f42012-01-17 00:37:02 +0000314 if (!State)
315 return;
316 C.addTransition(State);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000317 return;
Anna Zaks022b3f42012-01-17 00:37:02 +0000318 }
319
Anna Zaks9b0c7492012-01-18 02:45:07 +0000320 // Otherwise, check if we have custom pre-processing implemented.
Anna Zaks9ffbe242011-12-17 00:26:34 +0000321 FnCheck evalFunction = llvm::StringSwitch<FnCheck>(Name)
Anna Zaks1fb826a2012-01-12 02:22:34 +0000322 .Case("fscanf", &GenericTaintChecker::preFscanf)
Anna Zaks9ffbe242011-12-17 00:26:34 +0000323 .Default(0);
Anna Zaks9ffbe242011-12-17 00:26:34 +0000324 // Check and evaluate the call.
325 if (evalFunction)
326 State = (this->*evalFunction)(CE, C);
327 if (!State)
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000328 return;
Anna Zaks9ffbe242011-12-17 00:26:34 +0000329 C.addTransition(State);
Anna Zaks9b0c7492012-01-18 02:45:07 +0000330
Anna Zaks9ffbe242011-12-17 00:26:34 +0000331}
332
Anna Zaks1fb826a2012-01-12 02:22:34 +0000333bool GenericTaintChecker::propagateFromPre(const CallExpr *CE,
334 CheckerContext &C) const {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000335 ProgramStateRef State = C.getState();
Anna Zaks1fb826a2012-01-12 02:22:34 +0000336
337 // Depending on what was tainted at pre-visit, we determined a set of
338 // arguments which should be tainted after the function returns. These are
339 // stored in the state as TaintArgsOnPostVisit set.
340 llvm::ImmutableSet<unsigned> TaintArgs = State->get<TaintArgsOnPostVisit>();
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000341 if (TaintArgs.isEmpty())
342 return false;
343
Anna Zaks1fb826a2012-01-12 02:22:34 +0000344 for (llvm::ImmutableSet<unsigned>::iterator
345 I = TaintArgs.begin(), E = TaintArgs.end(); I != E; ++I) {
346 unsigned ArgNum = *I;
347
348 // Special handling for the tainted return value.
349 if (ArgNum == ReturnValueIndex) {
350 State = State->addTaint(CE, C.getLocationContext());
351 continue;
352 }
353
354 // The arguments are pointer arguments. The data they are pointing at is
355 // tainted after the call.
Anna Zaks259052d2012-04-10 23:41:11 +0000356 if (CE->getNumArgs() < (ArgNum + 1))
357 return false;
Anna Zaks1fb826a2012-01-12 02:22:34 +0000358 const Expr* Arg = CE->getArg(ArgNum);
Anna Zaks7cdfe292012-01-18 02:45:13 +0000359 SymbolRef Sym = getPointedToSymbol(C, Arg);
Anna Zaks1fb826a2012-01-12 02:22:34 +0000360 if (Sym)
361 State = State->addTaint(Sym);
362 }
363
364 // Clear up the taint info from the state.
365 State = State->remove<TaintArgsOnPostVisit>();
366
367 if (State != C.getState()) {
368 C.addTransition(State);
369 return true;
370 }
371 return false;
372}
373
374void GenericTaintChecker::addSourcesPost(const CallExpr *CE,
375 CheckerContext &C) const {
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000376 // Define the attack surface.
377 // Set the evaluation function by switching on the callee name.
Jordan Rose5ef6e942012-07-10 23:13:01 +0000378 const FunctionDecl *FDecl = C.getCalleeDecl(CE);
379 if (!FDecl || FDecl->getKind() != Decl::Function)
380 return;
381
382 StringRef Name = C.getCalleeName(FDecl);
Anna Zaks1fb826a2012-01-12 02:22:34 +0000383 if (Name.empty())
384 return;
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000385 FnCheck evalFunction = llvm::StringSwitch<FnCheck>(Name)
Anna Zaks9ffbe242011-12-17 00:26:34 +0000386 .Case("scanf", &GenericTaintChecker::postScanf)
Anna Zaks1009ac72011-12-14 00:56:02 +0000387 // TODO: Add support for vfscanf & family.
Anna Zaks9ffbe242011-12-17 00:26:34 +0000388 .Case("getchar", &GenericTaintChecker::postRetTaint)
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000389 .Case("getchar_unlocked", &GenericTaintChecker::postRetTaint)
Anna Zaks9ffbe242011-12-17 00:26:34 +0000390 .Case("getenv", &GenericTaintChecker::postRetTaint)
391 .Case("fopen", &GenericTaintChecker::postRetTaint)
392 .Case("fdopen", &GenericTaintChecker::postRetTaint)
393 .Case("freopen", &GenericTaintChecker::postRetTaint)
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000394 .Case("getch", &GenericTaintChecker::postRetTaint)
395 .Case("wgetch", &GenericTaintChecker::postRetTaint)
Anna Zaks2bf8fd82012-01-20 00:11:19 +0000396 .Case("socket", &GenericTaintChecker::postSocket)
Anna Zaks1fb826a2012-01-12 02:22:34 +0000397 .Default(0);
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000398
399 // If the callee isn't defined, it is not of security concern.
400 // Check and evaluate the call.
Ted Kremenek8bef8232012-01-26 21:29:00 +0000401 ProgramStateRef State = 0;
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000402 if (evalFunction)
Anna Zaks9ffbe242011-12-17 00:26:34 +0000403 State = (this->*evalFunction)(CE, C);
404 if (!State)
405 return;
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000406
Anna Zaks9ffbe242011-12-17 00:26:34 +0000407 C.addTransition(State);
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000408}
Anna Zaks8f4caf52011-11-18 02:26:36 +0000409
Anna Zaks9f03b622012-01-07 02:33:10 +0000410bool GenericTaintChecker::checkPre(const CallExpr *CE, CheckerContext &C) const{
411
412 if (checkUncontrolledFormatString(CE, C))
413 return true;
414
Anna Zaks4e462212012-01-18 02:45:11 +0000415 const FunctionDecl *FDecl = C.getCalleeDecl(CE);
Jordan Rose5ef6e942012-07-10 23:13:01 +0000416 if (!FDecl || FDecl->getKind() != Decl::Function)
417 return false;
418
Anna Zaks4e462212012-01-18 02:45:11 +0000419 StringRef Name = C.getCalleeName(FDecl);
Anna Zaks8568ee72012-01-14 02:48:40 +0000420 if (Name.empty())
421 return false;
422
423 if (checkSystemCall(CE, Name, C))
424 return true;
425
Anna Zaks4e462212012-01-18 02:45:11 +0000426 if (checkTaintedBufferSize(CE, FDecl, C))
427 return true;
428
Anna Zaks9f03b622012-01-07 02:33:10 +0000429 return false;
430}
431
Anna Zaks8f4caf52011-11-18 02:26:36 +0000432SymbolRef GenericTaintChecker::getPointedToSymbol(CheckerContext &C,
Anna Zaks7cdfe292012-01-18 02:45:13 +0000433 const Expr* Arg) {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000434 ProgramStateRef State = C.getState();
Ted Kremenek5eca4822012-01-06 22:09:28 +0000435 SVal AddrVal = State->getSVal(Arg->IgnoreParens(), C.getLocationContext());
Anna Zaksd3d85482011-12-16 18:28:50 +0000436 if (AddrVal.isUnknownOrUndef())
Anna Zakse3d250e2011-12-11 18:43:40 +0000437 return 0;
438
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000439 Loc *AddrLoc = dyn_cast<Loc>(&AddrVal);
Anna Zaks7cdfe292012-01-18 02:45:13 +0000440 if (!AddrLoc)
Anna Zaks8f4caf52011-11-18 02:26:36 +0000441 return 0;
442
Anna Zaks71d29092012-01-13 00:56:51 +0000443 const PointerType *ArgTy =
444 dyn_cast<PointerType>(Arg->getType().getCanonicalType().getTypePtr());
Anna Zaks665b0022012-01-21 06:59:01 +0000445 SVal Val = State->getSVal(*AddrLoc,
446 ArgTy ? ArgTy->getPointeeType(): QualType());
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000447 return Val.getAsSymbol();
448}
449
Ted Kremenek8bef8232012-01-26 21:29:00 +0000450ProgramStateRef
Anna Zaks7cdfe292012-01-18 02:45:13 +0000451GenericTaintChecker::TaintPropagationRule::process(const CallExpr *CE,
452 CheckerContext &C) const {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000453 ProgramStateRef State = C.getState();
Anna Zaks022b3f42012-01-17 00:37:02 +0000454
455 // Check for taint in arguments.
456 bool IsTainted = false;
Anna Zaks7cdfe292012-01-18 02:45:13 +0000457 for (ArgVector::const_iterator I = SrcArgs.begin(),
458 E = SrcArgs.end(); I != E; ++I) {
Anna Zaks022b3f42012-01-17 00:37:02 +0000459 unsigned ArgNum = *I;
460
461 if (ArgNum == InvalidArgIndex) {
Anna Zaks9b0c7492012-01-18 02:45:07 +0000462 // Check if any of the arguments is tainted, but skip the
463 // destination arguments.
464 for (unsigned int i = 0; i < CE->getNumArgs(); ++i) {
Anna Zaks7cdfe292012-01-18 02:45:13 +0000465 if (isDestinationArgument(i))
Anna Zaks9b0c7492012-01-18 02:45:07 +0000466 continue;
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000467 if ((IsTainted = isTaintedOrPointsToTainted(CE->getArg(i), State, C)))
Anna Zaks022b3f42012-01-17 00:37:02 +0000468 break;
Anna Zaks9b0c7492012-01-18 02:45:07 +0000469 }
Anna Zaks022b3f42012-01-17 00:37:02 +0000470 break;
471 }
472
Anna Zaks259052d2012-04-10 23:41:11 +0000473 if (CE->getNumArgs() < (ArgNum + 1))
474 return State;
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000475 if ((IsTainted = isTaintedOrPointsToTainted(CE->getArg(ArgNum), State, C)))
Anna Zaks022b3f42012-01-17 00:37:02 +0000476 break;
477 }
478 if (!IsTainted)
479 return State;
480
481 // Mark the arguments which should be tainted after the function returns.
Anna Zaks7cdfe292012-01-18 02:45:13 +0000482 for (ArgVector::const_iterator I = DstArgs.begin(),
483 E = DstArgs.end(); I != E; ++I) {
Anna Zaks022b3f42012-01-17 00:37:02 +0000484 unsigned ArgNum = *I;
485
486 // Should we mark all arguments as tainted?
487 if (ArgNum == InvalidArgIndex) {
488 // For all pointer and references that were passed in:
489 // If they are not pointing to const data, mark data as tainted.
490 // TODO: So far we are just going one level down; ideally we'd need to
491 // recurse here.
492 for (unsigned int i = 0; i < CE->getNumArgs(); ++i) {
493 const Expr *Arg = CE->getArg(i);
494 // Process pointer argument.
495 const Type *ArgTy = Arg->getType().getTypePtr();
496 QualType PType = ArgTy->getPointeeType();
497 if ((!PType.isNull() && !PType.isConstQualified())
498 || (ArgTy->isReferenceType() && !Arg->getType().isConstQualified()))
499 State = State->add<TaintArgsOnPostVisit>(i);
500 }
501 continue;
502 }
503
504 // Should mark the return value?
505 if (ArgNum == ReturnValueIndex) {
506 State = State->add<TaintArgsOnPostVisit>(ReturnValueIndex);
507 continue;
508 }
509
510 // Mark the given argument.
511 assert(ArgNum < CE->getNumArgs());
512 State = State->add<TaintArgsOnPostVisit>(ArgNum);
513 }
514
515 return State;
516}
517
518
Anna Zaks1fb826a2012-01-12 02:22:34 +0000519// If argument 0 (file descriptor) is tainted, all arguments except for arg 0
520// and arg 1 should get taint.
Ted Kremenek8bef8232012-01-26 21:29:00 +0000521ProgramStateRef GenericTaintChecker::preFscanf(const CallExpr *CE,
Anna Zaks9ffbe242011-12-17 00:26:34 +0000522 CheckerContext &C) const {
523 assert(CE->getNumArgs() >= 2);
Ted Kremenek8bef8232012-01-26 21:29:00 +0000524 ProgramStateRef State = C.getState();
Anna Zaks9ffbe242011-12-17 00:26:34 +0000525
526 // Check is the file descriptor is tainted.
Ted Kremenek5eca4822012-01-06 22:09:28 +0000527 if (State->isTainted(CE->getArg(0), C.getLocationContext()) ||
Anna Zaks1fb826a2012-01-12 02:22:34 +0000528 isStdin(CE->getArg(0), C)) {
529 // All arguments except for the first two should get taint.
530 for (unsigned int i = 2; i < CE->getNumArgs(); ++i)
531 State = State->add<TaintArgsOnPostVisit>(i);
532 return State;
533 }
534
Anna Zaks9ffbe242011-12-17 00:26:34 +0000535 return 0;
536}
537
Anna Zaks2bf8fd82012-01-20 00:11:19 +0000538
539// If argument 0(protocol domain) is network, the return value should get taint.
Ted Kremenek8bef8232012-01-26 21:29:00 +0000540ProgramStateRef GenericTaintChecker::postSocket(const CallExpr *CE,
Anna Zaks259052d2012-04-10 23:41:11 +0000541 CheckerContext &C) const {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000542 ProgramStateRef State = C.getState();
Anna Zaks259052d2012-04-10 23:41:11 +0000543 if (CE->getNumArgs() < 3)
544 return State;
Anna Zaks2bf8fd82012-01-20 00:11:19 +0000545
546 SourceLocation DomLoc = CE->getArg(0)->getExprLoc();
547 StringRef DomName = C.getMacroNameOrSpelling(DomLoc);
548 // White list the internal communication protocols.
549 if (DomName.equals("AF_SYSTEM") || DomName.equals("AF_LOCAL") ||
550 DomName.equals("AF_UNIX") || DomName.equals("AF_RESERVED_36"))
551 return State;
552 State = State->addTaint(CE, C.getLocationContext());
553 return State;
554}
555
Ted Kremenek8bef8232012-01-26 21:29:00 +0000556ProgramStateRef GenericTaintChecker::postScanf(const CallExpr *CE,
Anna Zaks9ffbe242011-12-17 00:26:34 +0000557 CheckerContext &C) const {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000558 ProgramStateRef State = C.getState();
Anna Zaks259052d2012-04-10 23:41:11 +0000559 if (CE->getNumArgs() < 2)
560 return State;
561
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000562 // All arguments except for the very first one should get taint.
563 for (unsigned int i = 1; i < CE->getNumArgs(); ++i) {
564 // The arguments are pointer arguments. The data they are pointing at is
565 // tainted after the call.
566 const Expr* Arg = CE->getArg(i);
Anna Zaks7cdfe292012-01-18 02:45:13 +0000567 SymbolRef Sym = getPointedToSymbol(C, Arg);
Anna Zaks1009ac72011-12-14 00:56:02 +0000568 if (Sym)
569 State = State->addTaint(Sym);
570 }
Anna Zaks9ffbe242011-12-17 00:26:34 +0000571 return State;
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000572}
573
Ted Kremenek8bef8232012-01-26 21:29:00 +0000574ProgramStateRef GenericTaintChecker::postRetTaint(const CallExpr *CE,
Anna Zaks259052d2012-04-10 23:41:11 +0000575 CheckerContext &C) const {
Ted Kremenek5eca4822012-01-06 22:09:28 +0000576 return C.getState()->addTaint(CE, C.getLocationContext());
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000577}
578
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000579bool GenericTaintChecker::isStdin(const Expr *E, CheckerContext &C) {
Ted Kremenek8bef8232012-01-26 21:29:00 +0000580 ProgramStateRef State = C.getState();
Ted Kremenek5eca4822012-01-06 22:09:28 +0000581 SVal Val = State->getSVal(E, C.getLocationContext());
Anna Zaksefd69892011-12-14 00:56:18 +0000582
Anna Zaksd3d85482011-12-16 18:28:50 +0000583 // stdin is a pointer, so it would be a region.
584 const MemRegion *MemReg = Val.getAsRegion();
585
586 // The region should be symbolic, we do not know it's value.
587 const SymbolicRegion *SymReg = dyn_cast_or_null<SymbolicRegion>(MemReg);
588 if (!SymReg)
Anna Zaksefd69892011-12-14 00:56:18 +0000589 return false;
590
Anna Zaksd3d85482011-12-16 18:28:50 +0000591 // Get it's symbol and find the declaration region it's pointing to.
592 const SymbolRegionValue *Sm =dyn_cast<SymbolRegionValue>(SymReg->getSymbol());
593 if (!Sm)
594 return false;
595 const DeclRegion *DeclReg = dyn_cast_or_null<DeclRegion>(Sm->getRegion());
596 if (!DeclReg)
597 return false;
Anna Zaksefd69892011-12-14 00:56:18 +0000598
Anna Zaksd3d85482011-12-16 18:28:50 +0000599 // This region corresponds to a declaration, find out if it's a global/extern
600 // variable named stdin with the proper type.
601 if (const VarDecl *D = dyn_cast_or_null<VarDecl>(DeclReg->getDecl())) {
602 D = D->getCanonicalDecl();
603 if ((D->getName().find("stdin") != StringRef::npos) && D->isExternC())
604 if (const PointerType * PtrTy =
605 dyn_cast<PointerType>(D->getType().getTypePtr()))
606 if (PtrTy->getPointeeType() == C.getASTContext().getFILEType())
607 return true;
608 }
Anna Zaksefd69892011-12-14 00:56:18 +0000609 return false;
610}
611
Anna Zaks9f03b622012-01-07 02:33:10 +0000612static bool getPrintfFormatArgumentNum(const CallExpr *CE,
613 const CheckerContext &C,
614 unsigned int &ArgNum) {
615 // Find if the function contains a format string argument.
616 // Handles: fprintf, printf, sprintf, snprintf, vfprintf, vprintf, vsprintf,
617 // vsnprintf, syslog, custom annotated functions.
618 const FunctionDecl *FDecl = C.getCalleeDecl(CE);
619 if (!FDecl)
620 return false;
621 for (specific_attr_iterator<FormatAttr>
622 i = FDecl->specific_attr_begin<FormatAttr>(),
623 e = FDecl->specific_attr_end<FormatAttr>(); i != e ; ++i) {
624
625 const FormatAttr *Format = *i;
626 ArgNum = Format->getFormatIdx() - 1;
627 if ((Format->getType() == "printf") && CE->getNumArgs() > ArgNum)
628 return true;
629 }
630
631 // Or if a function is named setproctitle (this is a heuristic).
632 if (C.getCalleeName(CE).find("setproctitle") != StringRef::npos) {
633 ArgNum = 0;
634 return true;
635 }
636
637 return false;
638}
639
Anna Zaks8568ee72012-01-14 02:48:40 +0000640bool GenericTaintChecker::generateReportIfTainted(const Expr *E,
641 const char Msg[],
642 CheckerContext &C) const {
643 assert(E);
644
645 // Check for taint.
Ted Kremenek8bef8232012-01-26 21:29:00 +0000646 ProgramStateRef State = C.getState();
Anna Zaks8568ee72012-01-14 02:48:40 +0000647 if (!State->isTainted(getPointedToSymbol(C, E)) &&
648 !State->isTainted(E, C.getLocationContext()))
649 return false;
650
651 // Generate diagnostic.
652 if (ExplodedNode *N = C.addTransition()) {
653 initBugType();
654 BugReport *report = new BugReport(*BT, Msg, N);
655 report->addRange(E->getSourceRange());
656 C.EmitReport(report);
657 return true;
658 }
659 return false;
660}
661
Anna Zaks9f03b622012-01-07 02:33:10 +0000662bool GenericTaintChecker::checkUncontrolledFormatString(const CallExpr *CE,
663 CheckerContext &C) const{
664 // Check if the function contains a format string argument.
665 unsigned int ArgNum = 0;
666 if (!getPrintfFormatArgumentNum(CE, C, ArgNum))
667 return false;
668
669 // If either the format string content or the pointer itself are tainted, warn.
Anna Zaks8568ee72012-01-14 02:48:40 +0000670 if (generateReportIfTainted(CE->getArg(ArgNum),
671 MsgUncontrolledFormatString, C))
672 return true;
673 return false;
674}
675
676bool GenericTaintChecker::checkSystemCall(const CallExpr *CE,
677 StringRef Name,
678 CheckerContext &C) const {
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000679 // TODO: It might make sense to run this check on demand. In some cases,
680 // we should check if the environment has been cleansed here. We also might
681 // need to know if the user was reset before these calls(seteuid).
Anna Zaks8568ee72012-01-14 02:48:40 +0000682 unsigned ArgNum = llvm::StringSwitch<unsigned>(Name)
683 .Case("system", 0)
684 .Case("popen", 0)
Anna Zaks2bf8fd82012-01-20 00:11:19 +0000685 .Case("execl", 0)
686 .Case("execle", 0)
687 .Case("execlp", 0)
688 .Case("execv", 0)
689 .Case("execvp", 0)
690 .Case("execvP", 0)
Anna Zaksb9ac30c2012-01-24 19:32:25 +0000691 .Case("execve", 0)
692 .Case("dlopen", 0)
Anna Zaks8568ee72012-01-14 02:48:40 +0000693 .Default(UINT_MAX);
694
Anna Zaks259052d2012-04-10 23:41:11 +0000695 if (ArgNum == UINT_MAX || CE->getNumArgs() < (ArgNum + 1))
Anna Zaks8568ee72012-01-14 02:48:40 +0000696 return false;
697
698 if (generateReportIfTainted(CE->getArg(ArgNum),
699 MsgSanitizeSystemArgs, C))
700 return true;
701
Anna Zaks9f03b622012-01-07 02:33:10 +0000702 return false;
703}
704
Anna Zaks4e462212012-01-18 02:45:11 +0000705// TODO: Should this check be a part of the CString checker?
706// If yes, should taint be a global setting?
707bool GenericTaintChecker::checkTaintedBufferSize(const CallExpr *CE,
708 const FunctionDecl *FDecl,
709 CheckerContext &C) const {
710 // If the function has a buffer size argument, set ArgNum.
711 unsigned ArgNum = InvalidArgIndex;
712 unsigned BId = 0;
713 if ( (BId = FDecl->getMemoryFunctionKind()) )
714 switch(BId) {
715 case Builtin::BImemcpy:
716 case Builtin::BImemmove:
717 case Builtin::BIstrncpy:
718 ArgNum = 2;
719 break;
720 case Builtin::BIstrndup:
721 ArgNum = 1;
722 break;
723 default:
724 break;
725 };
726
727 if (ArgNum == InvalidArgIndex) {
728 if (C.isCLibraryFunction(FDecl, "malloc") ||
729 C.isCLibraryFunction(FDecl, "calloc") ||
730 C.isCLibraryFunction(FDecl, "alloca"))
731 ArgNum = 0;
732 else if (C.isCLibraryFunction(FDecl, "memccpy"))
733 ArgNum = 3;
734 else if (C.isCLibraryFunction(FDecl, "realloc"))
735 ArgNum = 1;
736 else if (C.isCLibraryFunction(FDecl, "bcopy"))
737 ArgNum = 2;
738 }
739
Anna Zaks259052d2012-04-10 23:41:11 +0000740 if (ArgNum != InvalidArgIndex && CE->getNumArgs() > ArgNum &&
Anna Zaks4e462212012-01-18 02:45:11 +0000741 generateReportIfTainted(CE->getArg(ArgNum), MsgTaintedBufferSize, C))
742 return true;
743
744 return false;
745}
746
Anna Zaksdf18c5a2011-11-16 19:58:13 +0000747void ento::registerGenericTaintChecker(CheckerManager &mgr) {
748 mgr.registerChecker<GenericTaintChecker>();
749}